Optimisation of extremely low energy residential buildings

Transcription

KATHOLIEKE UNIVERSITEIT LEUVEN
FACULTEIT INGENIEURSWETENSCHAPPEN
DEPARTEMENT BURGERLIJKE BOUWKUNDE
AFDELING BOUWFYSICA
Kasteelpark Arenberg 40
B-3001 Leuven
OPTIMISATION OF EXTREMELY LOW ENERGY
RESIDENTIAL BUILDINGS
Promotor:
Proefschrift voorgedragen tot
Prof.dr.ir. H. Hens
het behalen van het doctoraat
in de ingenieurswetenschappen
door
Griet VERBEECK
Mei 2007
KATHOLIEKE UNIVERSITEIT LEUVEN
FACULTEIT INGENIEURSWETENSCHAPPEN
DEPARTEMENT BURGERLIJKE BOUWKUNDE
AFDELING BOUWFYSICA
Kasteelpark Arenberg 40
B-3001 Leuven
OPTIMISATION OF EXTREMELY LOW ENERGY
RESIDENTIAL BUILDINGS
Examencommissie:
Proefschrift voorgedragen tot
Prof. Dr. Ir. A. Bultheel, voorzitter
het behalen van het doctoraat
Prof. Dr. Ir. H. Hens, promotor
in de ingenieurswetenschappen
Prof. Dr. Ir. W. D’haeseleer
Prof. Dr. S. Proost
door
Prof. Dr. Ir. D. Roose
Prof. Dr.-Ing. G. Hauser
Griet VERBEECK
Prof. Dr. Ir. R. Zmeureanu
UDC 502.55:697.12:728.1
Mei 2007
©
Katholieke Universiteit Leuven – Faculteit Ingenieurswetenschappen
Kasteelpark Arenberg, B-3001 Leuven
Alle rechten voorbehouden. Niets uit deze uitgave mag worden vermenigvuldigd en/of
openbaar gemaakt door middel van druk, fotokopie, microfilm, elektronisch of op welke
andere wijze ook zonder voorafgaandelijke schriftelijke toestemming van de uitgever.
All rights reserved. No part of the publication may be reproduced in any form by print,
photoprint, microfilm or any other means without written permission from the publisher.
Wettelijk depot D/2007/7515/35
ISBN 978-90-5682-801-1
VOORWOORD
Bij het begin van mijn doctoraat, was ik moeder van een zoontje dat bijna naar de peuterklas
mocht. Nu, vier jaar later, kan die zoon al lezen en rekenen en ben ik twee dochtertjes én
een doctoraat rijker. Het is een boeiende, maar ook zware tijd geweest. Soms waren de
kinderen een welkome afwisseling tijdens het doctoreren, soms was het onderzoek een
welkome afwisseling tijdens de zorg voor kroost en huishouden. Even een berekening
opstarten tussen het geven van een vieruurtje en het vervangen van een vuile luier, Monte
Carlo simulaties laten lopen met een slapende dochter op schoot, wetenschappelijke
literatuur lezen in afwachting dat de schoolbel gaat, nog even een machine was insteken
tijdens het wachten op rekenresultaten, kleine kindjes die vol bewondering kijken naar
kleurige grafiekjes die automatisch veranderen op het computerscherm, … Zoals de dochter
van professor en onderzoekster Christine Van Broeckhoven het ooit verwoordde in Titaantjes
op Radio 1: “in de voormiddag deed mijn moeder aan onderzoek en op de middag dweilde
ze nog snel even de keukenvloer, om dan weer terug te keren naar haar labo.” Zonder mij
wetenschappelijk met haar te willen meten, ging er toch een golf van herkenning door mij
heen.
Toch stond ik er zeker niet alleen voor. Mijn ‘nieuwe man’ Stijn is, ondanks zijn eigen
hectische job, altijd een ongelooflijke steun geweest, zowel moreel als in de zorg voor ons
gezin. Mijn schoonmoeder heeft ontelbare keren de rit Herk-de-Stad – Holsbeek gemaakt om
zieke kinderen op te vangen of kinderen van school te halen, zodat ik nog enkele uurtjes
langer kon werken. Ook mijn ouders hebben vaak hun drukke agenda opzij gezet om in te
springen voor noodgevallen.
Naast deze praktische en morele steun was er uiteraard ook de wetenschappelijke
ondersteuning van mijn promotor professor Hens. Van bij mijn start op het Laboratorium
Bouwfysica in 1991 heeft hij getracht mij te overtuigen om te doctoreren. Maar door mijn zeer
brede interesses en de nood aan directe maatschappelijke relevantie in mijn werk, heeft het
meer dan 10 jaar geduurd alvorens ik een onderwerp vond dat mij voldoende boeide om er
vier jaar van mijn leven aan te spenderen. In al die jaren heb ik echter zeer veel opgestoken
van professor Hens over energie, comfort en duurzaamheid in gebouwen. Bovendien heeft
hij mij steeds veel vrijheid en zelfstandigheid gelaten bij het uitvoeren van mijn onderzoek.
Iets wat ik altijd erg heb gewaardeerd. Ook was er de financiële ondersteuning van het IWT
via het GBOU-project dat mij de mogelijkheid gaf om vier jaar aan mijn onderzoek te
besteden.
i
Ik heb in al die jaren op het Labo Bouwfysica veel collega’s zien komen en gaan en met
verschillende van hen een bureau gedeeld. Steeds opnieuw waren het leuke, boeiende en
warme contacten. Een speciaal woordje van dank voor Staf, waarmee ik zowel professioneel
als familiaal vaak op dezelfde golflengte zat, voor mijn ‘jonge’ collega’s Jeroen, Veerle en
Leen die zeer aangename en enthousiaste co-researchers waren in het GBOU-project en
voor de ploeg die mij ‘logistiek’ ondersteunde: Patricia die altijd met de glimlach hielp bij mijn
computer-onkunde, Wim die elke morgen heerlijke koffie maakte, Paul die zijn jarenlange
loopervaring deelde met mij als beginnende loopster en Beatrice en Greet die ieder op hun
beurt mijn eindeloze reeks contracten hielpen in orde maken.
Voor mij is het een mooie tijd geweest. Veel dank hiervoor. Ik zal jullie missen.
Januari 2007
ii
ABSTRACT
The objective of the research is to establish a methodology to optimise the combinations of
passive (building envelope) and active (installations) measures that lead to (extremely) low
energy and low pollution residential buildings. Energy consumption and energy savings,
environmental impact and costs are considered simultaneously without neglecting the
boundary conditions for thermal comfort and indoor air quality and the legal requirements for
the energy performance of buildings. The methodology combines advanced evolutionary
multi-objective optimisation techniques, life cycle inventory and cost-benefit assessment and
is applied to buildings as a whole. Starting from non insulated reference dwellings, the
optimisation process is executed in two steps. Firstly, the net energy demand is minimised by
optimisation of the building envelope. In the second step, the focus is shifted towards the
most effective installation techniques to meet the very low energy demand. The methodology
has been used to derive cost-effective energy saving investments in new and retrofitted
buildings for the Brussels urban context and to develop concepts for extremely low energy
and low pollution dwellings for the Flemish Region. Also the economical optimal combination
of energy saving measures and a hierarchy of energy saving investments have been derived.
Keywords: energy consumption, costs, emissions, residential buildings, evolutionary multiobjective optimisation, life cycle assessment
iii
KORTE INHOUD
Doel van het onderzoek is het ontwikkelen van een methodologie die de optimale
combinaties van passieve (gebouwschil) en actieve (installaties) maatregelen bepaalt voor
(extreem) lage energie- en pollutiewoningen. Zowel energieverbruik, energiebesparing als
milieu-impact en kosten worden in rekening gebracht binnen de randvoorwaarden voor
thermisch comfort, binnenluchtkwaliteit en wettelijke vereisten voor de energieprestatie van
gebouwen. De optimalisatiemethodologie combineert geavanceerde evolutionaire multiobjectieve optimalisatietechnieken, levenscyclusanalyse en kosten-batenanalyse en past dit
toe op het niveau van gebouwen als geheel. Vertrekkend van niet geïsoleerde
referentiewoningen verloopt het optimalisatieproces in twee fasen. Eerst wordt de netto
energievraag geminimaliseerd door optimalisatie van de gebouwschil. In een tweede fase
worden dan de meest optimale installatietechnieken gezocht die aan deze zeer lage
energievraag kunnen voldoen. De ontwikkelde methodologie is gebruikt om de rendabiliteit
van energiebesparende investeringen in nieuwe en bestaande Brusselse gebouwen te
evalueren en om optimale concepten voor extreem lage energie- en pollutiewoningen te
ontwikkelen voor Vlaanderen. Ook is met deze methodologie de economisch meest optimale
combinatie van bouwkundige en installatietechnische maatregelen afgeleid, alsook een
logische hiërarchie van energiebesparende investeringen.
Trefwoorden: energieverbruik, kosten, emissies, woningen, evolutionaire multi-objectieve
optimalisatie, levenscyclusanalyse
iv
CONTENTS
VOORWOORD.........................................................................................................................i
ABSTRACT............................................................................................................................iii
KORTE INHOUD....................................................................................................................iv
CONTENTS.............................................................................................................................v
LIST OF ACRONYMS AND SYMBOLS ................................................................................ix
INTRODUCTION.....................................................................................................................1
PART ONE: LITERATURE REVIEW......................................................................................3
CHAPTER 1.
Evolutionary multi objective optimisation...........................................3
1.1 Introduction..............................................................................................................3
1.2 Multi-variable optimisation .......................................................................................5
1.2.1. Representation and population........................................................................5
1.2.2. Evaluation function and fitness assignment.....................................................6
1.2.3. Genetic operators ............................................................................................7
1.2.4. Convergence of a genetic algorithm ..............................................................10
1.3 Multi-objective optimisation ...................................................................................13
1.3.1
Search and decision making..........................................................................13
1.3.2
Concept of Pareto optimality..........................................................................14
1.3.3
Pareto-based fitness assignment and selection ............................................16
1.4 Constraint handling................................................................................................17
1.4.1
Introduction ....................................................................................................17
1.4.2
Penalty functions ...........................................................................................18
1.5 Building-related case studies in the literature ........................................................19
1.6 Conclusions ...........................................................................................................19
CHAPTER 2.
Life cycle analysis ...............................................................................23
2.1 History and background.........................................................................................23
2.2 Methods for environmental assessment ................................................................23
2.2.1
Life cycle analysis..........................................................................................24
2.2.2
Process-chain analysis ..................................................................................25
2.2.3
Input-output analysis......................................................................................26
2.2.4
Life cycle costing ...........................................................................................26
2.2.5
Product-related environmental assessment tools ..........................................27
2.2.6
Building-related environmental assessment tools .........................................28
2.3 LCA methodology ..................................................................................................29
2.3.1
Introduction ....................................................................................................29
2.3.2
Framework.....................................................................................................30
2.3.3
Inventory analysis ..........................................................................................32
2.3.4
Impact assessment........................................................................................34
2.3.5
Interpretation..................................................................................................36
2.3.6
Conclusions ...................................................................................................36
2.4 Information sources for LCA ..................................................................................36
v
Organisations and networks ..........................................................................36
2.4.1
2.4.2
Data and databases.......................................................................................37
2.5 LCA and the building sector ..................................................................................42
2.5.1
General ..........................................................................................................42
2.5.2
From material to building ...............................................................................44
2.6 Conclusions ...........................................................................................................46
CHAPTER 3.
Cost assessment .................................................................................47
3.1 Introduction............................................................................................................47
3.2 Cost-benefit analysis .............................................................................................48
3.2.1. General ..........................................................................................................48
3.2.2. Total and net present value ...........................................................................49
3.2.3. Discount rate and price evolution ..................................................................52
3.3 Rebound effect ......................................................................................................54
3.3.1. General ..........................................................................................................54
3.3.2. Rebound effect applied to energy saving measures in buildings...................55
3.4 Decision making on energy saving investments....................................................57
3.4.1. Objective........................................................................................................57
3.4.2. Facts and figures on energy related behaviour in Flanders...........................58
3.4.3. Decision making models................................................................................59
3.5 Conclusions ...........................................................................................................63
PART TWO: GLOBAL METHODOLOGY ............................................................................65
CHAPTER 4.
Model for optimisation ........................................................................65
4.1 Introduction............................................................................................................65
4.2 Genetic algorithm and coupling with Pareto concept.............................................65
4.2.1. Object and parameters for optimisation.........................................................65
4.2.2. Representation and boundary conditions ......................................................66
4.2.3. Cost functions, fitness functions and penalty functions .................................68
4.2.4. Genetic operators ..........................................................................................71
4.2.5. Outline of the programme ..............................................................................71
4.3 Validation and evaluation of the optimisation programme .....................................73
4.3.1. Validation with steady state energy simulation programme (EPB) ................73
4.3.2. Control of the evolution of the Pareto front ....................................................76
4.3.3. Conclusions on the optimisation programme.................................................77
CHAPTER 5.
Model for life cycle inventory .............................................................79
5.1 Introduction............................................................................................................79
5.2 Goal and scope of the LCI .....................................................................................79
5.2.1. Goal of the LCI...............................................................................................79
5.2.2. Scope of the LCI ............................................................................................80
5.3 Life cycle inventory data ........................................................................................81
5.3.1. Comparison of databases..............................................................................81
5.3.2. ecoinvent2000 database (Frischknecht and Jungbluth 2003) .......................83
5.3.3. Extraction of data from ecoinvent2000 ..........................................................87
5.4 Life cycle inventory model for buildings .................................................................90
5.4.1. Material and product models .........................................................................90
5.4.2. Transport model.............................................................................................91
5.4.3. Building model ...............................................................................................92
5.5 Uncertainty and sensitivity analysis .......................................................................96
5.5.1. Sensitivity analysis for the transport model ...................................................96
5.5.2. Contribution analysis for the building model ..................................................98
5.5.3. Perturbation analysis for the building model ................................................100
5.5.4. Uncertainty analysis with Monte Carlo simulations......................................102
vi
5.5.5.
Conclusions on the sensitivity and uncertainty analysis of the LCI model...112
CHAPTER 6.
Model for economic evaluation ........................................................113
6.1 Introduction..........................................................................................................113
6.2 Economic evaluation criteria................................................................................113
6.2.1. Description of the criteria .............................................................................113
6.2.2. Criteria for optimisation................................................................................117
6.3 Cost database .....................................................................................................120
6.3.1. Cost data for the building envelope .............................................................120
6.3.2. Cost data for components of the installation................................................120
6.3.3. Energy prices and price evolution................................................................120
6.4 Calculation models ..............................................................................................121
6.4.1. Building components ...................................................................................121
6.4.2. Components of the installation ....................................................................122
6.4.3. Integration of secondary cost effects ...........................................................122
6.5 Discussion of different assumptions ....................................................................123
6.5.1. Uncertainty on cost data ..............................................................................123
6.5.2. Uncertainty on energy price evolutions........................................................123
6.5.3. Impact of the utilisation period .....................................................................124
6.5.4. Integration of rebound effect........................................................................125
6.5.5. Integration of the residual value...................................................................128
PART THREE: APPLICATIONS ........................................................................................135
CHAPTER 7.
BIM-project .........................................................................................135
7.1 Introduction..........................................................................................................135
7.2 Implementation of energy saving measures ........................................................136
7.2.1. Reference buildings .....................................................................................136
7.2.2. Measures on the building envelope .............................................................137
7.2.3. System-related measures............................................................................139
7.2.4. Building simulation programme ...................................................................143
7.3 Life cycle inventory ..............................................................................................144
7.4 Cost assumptions ................................................................................................144
7.5 Results.................................................................................................................144
7.5.1. Energy saving measures on the building envelope .....................................144
7.5.2. Energy saving measures on the overall building .........................................148
7.5.3. Extra embodied energy due to energy saving measures ............................151
7.6 Discussion ...........................................................................................................152
7.6.1. Impact of the scenarios with varying economic parameters ........................152
7.6.2. New construction versus retrofit ..................................................................154
7.6.3. Position of alternative technologies against optimal solutions .....................156
7.7 Conclusions .........................................................................................................158
CHAPTER 8.
EL²EP-project .....................................................................................161
8.1 Introduction..........................................................................................................161
8.2 Implementation of energy saving measures ........................................................162
8.2.1. Reference buildings .....................................................................................162
8.2.2. Measures on the building envelope .............................................................162
8.2.3. System-related measures............................................................................164
8.2.4. Building simulation programme ...................................................................164
8.2.5. Boundary conditions ....................................................................................165
8.3 Life cycle inventory ..............................................................................................165
8.4 Cost assessment .................................................................................................165
8.5 Optimisation.........................................................................................................166
8.6 Results.................................................................................................................166
vii
8.6.1. Optimisation of the building envelope measures .........................................166
8.6.2. Globally optimised concepts for extremely low energy dwellings ................173
8.6.3. Embodied energy versus energy savings ....................................................174
8.7 Discussion ...........................................................................................................175
8.7.1. Strengths and weaknesses of the optimisation model.................................176
8.7.2. Impact of the constructional type .................................................................177
8.7.3. Impact of the compactness..........................................................................179
8.7.4. Impact of energy carrier and heat production system..................................181
8.7.5. Constructional cost versus installation cost .................................................184
8.7.6. Impact of price evolutions and discount rate ...............................................186
8.7.7. Position of extremely low energy dwellings to the economic optimum ........190
8.7.8. Position of existing concepts for extremely low energy dwellings ...............192
8.8 Conclusions .........................................................................................................198
CHAPTER 9.
9.1
9.2
CONCLUSIONS AND FURTHER RESEARCH ..................................201
Main results and conclusions ..............................................................................201
Perspectives for future research..........................................................................205
SAMENVATTING................................................................................................................207
ANNEX A: Relation between research database and ecoinvent database ..................225
ANNEX B: Extracted flows from the ecoinvent database..............................................230
ANNEX C: Models for life cycle inventory of building products...................................232
ANNEX D: Assumptions for the transport model...........................................................236
ANNEX E: Structure of the cost database ......................................................................240
ANNEX F: Reference buildings of the BIM- project .......................................................241
ANNEX G: Reference buildings of the EL²EP- project...................................................248
ANNEX H: Details on zero energy houses ......................................................................259
REFERENCES....................................................................................................................261
CURRICULUM VITAE ........................................................................................................271
viii
LIST OF ACRONYMS AND SYMBOLS
BIM
Brussels Institute for the Management of the Environment
CED
Cumulative energy demand
CHP
Cogeneration of heat and power
CV
Coefficient of variance
EEC
End energy consumption
EL²EP
Extremely low energy and low pollution
EPB
Energy Performance of Buildings (in Dutch)
EPBD
European Performance of Buildings Directive
EPBT
Energy payback time
EPS
Expanded polystyrene
GA
Genetic algorithms
GWP
Global warming potential
INV
Initial investment
IOA
Input-output analysis
IRR
Internal rate of return
LCA
Life cycle analysis or assessment
LCC
Life cycle costing
LCI
Life cycle inventory
LCIA
Life cycle impact assessment
MC
Monte Carlo
MW
Mineral wool
NBN
Norme Belge – Belgische Norm
NHD
Net heat demand
NMVOC
Non-methane volatile organic compounds
NPV
Net present value
NRE
Non renewable embodied energy
PCA
Process-chain analysis
PE
Primary energy consumption
PUR
Polyurethane
PV
Photovoltaic
RER
Europe
TPV
Total present value
TRNSYS
Transient Systems Simulation Program
U
Heat transmission coefficient [W/m²K]
WFC
Wood frame construction
WTE
Weighted temperature exceedings
XPS
Extruded polystyrene
ix
INTRODUCTION
INTRODUCTION
Background
The main driving force of this research is the need for sustainable development in the
residential sector in order to reduce the use of natural resources, the emission of greenhouse
gasses and the depletion of fossil fuels supplies. The term 'sustainability' has been defined in
many ways yet, but nowadays the definition of the UN Brundland commission Report of 1987
(WCED 1987) is the most commonly used. In that formulation, sustainable development is
coined as "development that meets the needs of the present without compromising the ability
of future generations to meet their own needs“. From this definition, it is clear that
sustainable development can only be accomplished if many conditions are fulfilled. This
certainly counts for the building sector, where the design, construction, utilisation and
demolition of buildings have an enormous environmental impact. In Flanders, building
utilisation is responsible for 30% of the annual end use of energy (Van Steertegem 2001).
For other developed countries, similar percentages are found. Energy use usually means
pollution due to emission of CO2, NOx, SO2 , dust a.o. In the residential sector in Flanders
only, 13.2 megatons of CO2 are emitted yearly due to energy use, including the emissions
from the electricity production for residential use (Van Steertegem 2001).
In the existing building stock, around 240 kWh/m²a of end energy is used for heating,
domestic hot water, electrical appliances and lighting. The main part is dedicated to heating
(ca. 180 kWh/m²a). Since the nineties, legislative initiatives have been taken to improve the
energy efficiency in the residential sector (Flemish Insulation and Ventilation Decree in
1991), but with very poor results as shown in SENVIVV (1998). This insulation standard, if
observed, would have reduced the final energy use for heating to 120 kWh/m²a. Buildings
that meet the minimum requirements of the energy performance regulation (EPB 2005),
brought into effect since January 2006, only do slightly better. Strengthening of these
requirements is expected for the near future, but this regulation is mainly limited to new
buildings.
Many national and international initiatives have been launched up to now to improve energy
efficiency
and
sustainability
in
buildings
(EPBD
2002,
IEA-ECBCS,
EU
Frame
Programmes,…). One of the current trends is to strongly focus on the integration of
renewable energy sources in buildings, often implemented through demonstration projects or
architectural
competitions
(EU-projects:
CEPHEUS,
SYNPACK,
EU
competition:
INSOLPLAN,...). However, the basic arguments for choosing one or another technology
often lack scientific foundation or they are only valid in a limited context. For instance, the
interactions between systems or the second order effects of energy saving measures, which
1
can have a non-negligible impact on the environmental performance of the building as a
whole, are often not taken into account.
What is lacking in the current trends is an underlying global methodology that allows, on a
scientific basis, the development of residential buildings that are globally optimised from the
point of view of energy, costs and ecology. The development and implementation of this
methodology is the subject of this work. The basic principle is a well-founded evaluation of
the environmental impact during the whole life cycle of the building and its installations,
through coupling of life cycle assessment and cost-benefit evaluation with advanced
optimisation techniques. This has resulted in the establishment of an economical optimal
combination of energy saving measures and a hierarchy of energy saving investments and in
an assessment of the strengths and weaknesses of concepts for extremely low energy
buildings.
Aim and methodology
The objective of this research has been a methodology to optimise building concepts for
energy, costs and emissions. This has been done in the frame of a research project for the
Flemish government (IWT-GBOU 020212), but it has already been applied and will be
applied in the future, within other policy supporting studies on the economic viability of
energy saving measures in buildings.
The optimisation process mainly consists of two subsequent steps:
The first step focuses on the optimisation of the net energy demand through optimisation of
energy saving measures on the building envelope. In the second step, the focus is shifted
towards the most appropriate technologies to meet this very low energy demand in an
optimal way. This includes systems for distribution, emission, production, control and storage
of heat and systems for local electricity production.
Questions to be answered are: What are the most optimal combinations of passive (building
envelope) and active (installations) components that lead to (extremely) low energy
dwellings? What is the lower limit that can be achieved in an economic, energetic and
ecological useful way, taking into account the large number of parameters and uncertainties
that play a role in the life cycle of a building?
The methodology, here presented, aims at answering these questions. However, as for all
research, also this research has its limits. Much attention has been paid to energy, costs and
emissions, but aspects, such as water use, urban planning, mobility, waste treatment, land
use, etc. have not been taken into account. Consumer choices and occupants’ behaviour
have a large impact on building construction and building use and have therefore been
incorporated to some extent. However, it only covers a fraction of the complexity of real life
behaviour of consumers. So, this work intends to be a small piece of the puzzle of
2
INTRODUCTION
sustainable buildings, but, to finally end up with a sustainable building stock, there still is a
long way to go.
Outline of the work
The first part of this dissertation contains an extensive literature review on the three main
pillars of the developed methodology, being evolutionary multi-objective optimisation (chapter
1), life cycle assessment (chapter 2) and cost evaluation (chapter 3). Each chapter starts
with background information, followed by a general state of the art and definitions to finally
end with an analysis of the application of the considered technique within the context of the
building sector.
Part two contains the body of this research and presents the developed methodology in three
chapters. Chapter 4 explains the model for optimisation, with a description of the different
assumptions for the developed genetic algorithm and the way the Pareto concept is
integrated in order to optimise for multiple objectives simultaneously. The outline of the
optimisation programme is presented, together with a validation and evaluation of the
programme. In chapter 5, the model for life cycle inventory is presented. Firstly, the goal and
scope of the life cycle inventory is described, followed by a presentation of the life cycle
inventory database. Finally, the life cycle inventory model for buildings is explained and
evaluated by an uncertainty analysis. Chapter 6 presents the model for cost assessment.
Firstly, the cost evaluation criteria are described, followed by a presentation of the cost
database and the calculation models for building components, installation components and
the integration of second order effects. Finally, several assumptions are discussed, such as
the uncertainty on cost data and energy price evolutions and the impact of the utilisation
period, the rebound effect and the residual value of energy saving buildings.
In part three, the developed methodology is applied to two projects. In chapter 7, the BIMproject is described. This project has been executed by order of the Brussels Institute for the
Management of the Environment (BIM) and is an application of the methodology to the
Brussels building stock. In this project, a hierarchy of energy saving measures is derived and
the economic viability of energy saving investments for new and retrofitted buildings is
determined. Chapter 8 presents the main results of the IWT-GBOU-EL²EP-project, in the
frame of which the methodology has been developed. The focus in this chapter is put on
concepts for extremely low energy houses and the comparison with existing concepts, such
as passive houses and zero energy houses.
Finally, in chapter 9, conclusions are drawn and paths for further research are presented.
3
EVOLUTIONARY MULTI-OBJECTIVE OPTIMISATION
PART ONE: LITERATURE REVIEW
CHAPTER 1.
1.1
Evolutionary multi objective optimisation
Introduction
Due to the complexity of real-world problems, almost every engineering domain is confronted
with optimisation problems in which a large number of parameters, variables, objectives and
constraints are involved. When solving such a problem, the best solution among others is
searched in the space of all feasible solutions, called the search space. Each feasible
solution can be marked by its value or fitness for the problem. The better a solution fits for a
problem, the higher will be its fitness. Looking for the best solution is then equal to looking for
a minimum or maximum in the search space. However, the search can be very complicated
and time-consuming (Obitko 1998). Several search and optimisation techniques dealing with
multiple variables and objectives are presented in the literature. Van Veldhuizen (1999)
classifies them into three categories: enumerative, deterministic and stochastic. The
enumerative strategies are the simplest, as they evaluate each possible solution. This is only
practicable in small search spaces. As search spaces become large, these techniques are
inefficient and even infeasible. Deterministic strategies attempt to limit the search space by
incorporating problem domain knowledge in the search process, thus trying to find
acceptable solutions in acceptable time. Hill-climbing and greedy algorithms are examples of
this type of strategies (Van Veldhuizen 1999).
Many engineering problems, however, are non-linear, cannot be expressed through
analytical functions and have to deal with the discontinuity and non separability of variables
and constraints. The optimisation of extremely low energy dwellings is an example of such a
complex optimisation problem:
-
The energy consumption of dwellings can roughly be estimated through a simple
analytical function such as the degree-days method, but these methods are very
simplified steady state models. They do not allow the incorporation of multizonal
temperature profiles, ventilation scenarios or solar shading. Neither can the transient
response of the building and its installations on ventilation and solar insolation or the
evaluation of summer comfort be taken into account. Therefore, dynamic building
simulation programmes are needed that are able to simulate the behaviour of the
building and its installations on one hour or smaller time steps.
-
The majority of the variables that determine the energy performance of buildings,
such as insulation thicknesses, glazing types and window frames, boilers, radiators,
3
CHAPTER 1
solar collectors, PV-modules, etc. are discontinuous, as their magnitude, dimensions
or thermal quality depends on what is available on the market.
-
Furthermore, several variables are not acting independently from each other or may
influence the final result in two opposite directions: e.g. by increasing the glass area
in a façade, both the solar gains and the heat losses will increase, but which impact
will dominate, will depend on many parameters, such as the magnitude of the glass
area, the U-value of both the windows and the opaque part of the façade, the g-value
of the glazing, the presence and characteristics of solar shading, the thermal capacity
of the room, the time constant of thermostatic valves, if present, etc. These are all
aspects that are difficult to incorporate in deterministic optimisation techniques.
-
Finally, most classical optimisation techniques aim at finding the single optimal
solution for the optimisation problem. This approach implicitly includes a priori
weighting of the objectives by the decision maker. Without further specifications on
weighting factors, a weighting factor of 1 for all objectives is assumed. To evaluate
other weighting factors, more optimisation runs are needed within these techniques.
However, the determination of the trade off between the different objectives (energy,
emissions and costs) is exactly one of the main points of interest within the search for
optimised extremely low energy dwellings, as it creates the opportunity to postpone
the decision process and adapt the weighting factors after the optimisation process
according to the interest and preference of the decision maker. This is further
explained in section 1.3.
As enumerative and deterministic strategies are not really suitable for such optimisation
problems, stochastic search and optimisation strategies are developed as alternative
approaches. They require a function that assigns fitness to possible solutions and a coding
mechanism that translates between the problem domain and the algorithm domain. They
generally provide good solutions to a wide range of optimisation problems, although most
cannot guarantee the optimal solution (Van Veldhuizen 1999). Examples are simulated
annealing, Monte Carlo methods and evolutionary computation.
Evolutionary computation is a generic term for stochastic search methods that is founded in
the 70s and for which the interest has been growing rapidly in the last decade. They are
adaptive optimisation methods, inspired by the genetic processes of biological organisms.
Over many generations, natural populations evolve according to the principles of natural
selection and 'survival of the fittest'. By mimicking this process, evolutionary computation is
able to evolve solutions to real world problems (Beasley et al. 1993).
A comparative study of the suitability of all kinds of optimisation techniques never has been
within the scope of the research. Rather was one of the goals the evaluation of the
4
opportunities of evolutionary computation for building related optimisation problems as up to
now the application within this domain is quite limited. Further details on the application of
this optimisation technique for building related issues are given in section 1.5, but first the
concept of evolutionary optimisation techniques, more specifically of genetic algorithms, the
technique applied within this research, is presented and discussed in the next sections.
1.2
Multi-variable optimisation
One of the techniques embodied by evolutionary computation is the optimisation method of
genetic algorithms (GA). The term has been introduced by J.H. Holland in the 70s (Holland
1975). In the last few years, the growing interest on this technique is reflected in an
increasing amount of software and literature devoted to this subject (Coello 1996). Generally,
an optimisation run corresponds to a search through the space of possible solutions. To
avoid that the optimisation process ends at local optima, such a search requires a balance
between exploiting the best solutions and exploring the search space (Michalewicz 1999).
Hill climbing is an example of an optimisation technique that exploits the best solution for
possible improvement, but without exploring the search space. Random search is a typical
example of an optimisation technique that explores the search space, but ignores the
exploitations of promising regions in the space. Genetic algorithms are a class of search
methods that balance between exploration of the search space for new, interesting genetic
material and exploitation of the genetic information of the best solutions (Michalewicz 1999).
The next section describes more in detail the basic concepts and terminology of multivariable optimisation by genetic algorithms. It is mainly limited to a description of what
genetic algorithms are, how they can be implemented and how they perform. Further
explanation on why genetic algorithms work as they work (cfr. Schema theorem and Building
Blocks Hypothesis) is beyond the scope of this research, but is clearly explained in chapter 3
of Michalewicz (1999).
1.2.1.
Representation and population
Genetic algorithms use a vocabulary borrowed from natural genetics. A possible solution to a
multi-variable optimisation problem is called an individual. It needs to be encoded to be used
in the genetic optimisation process. This encoded representation is called a chromosome or
a string. Each chromosome is made of units or genes, arranged linearly. Each gene is
located at a certain place in the chromosome and represents one variable or parameter of
the individual. All genes together form a chromosome that represents unequivocally a
possible solution of the problem. Applied to a building, a gene could represent one aspect of
5
CHAPTER 1
the building, such as construction type, insulation thickness, glazing type, etc. All genes
together form the chromosome and should unequivocally determine the building.
The algorithm starts from a random set of individuals, forming the initial population. The
number of individuals in a population is called the population size. For a building optimisation
problem, this initial population could consist of a randomly created set of variants, e.g. 100
variants of the same building of which each variant contains a randomly chosen glazing type,
a randomly chosen insulation thickness in the roof, a randomly chosen heating installation,
etc. Solutions from a population are selected to create a new and normally improved
population. To achieve this, solutions that act as parents, are selected according to their
fitness. This means the more suitable a solution is for the problem, the higher its probability
to be selected for reproduction (Obitko 1998, Dasgupta et al. 1997). If the building
optimisation problem would concern minimisation of the energy consumption, the fitness of
each variant of the building would depend on its energy consumption. Therefore, the energy
consumption of each variant within the population should be calculated and the lower the
energy consumption of a variant, the higher the probability that this variant would be selected
as ‘parent’ to create new variants. This application to buildings is only an example to illustrate
the technique of genetic algorithms. The real assumptions adopted within this research are
explained in detail in chapter 4.
1.2.2.
Evaluation function and fitness assignment
In order to determine their suitability for the problem, all individuals in a population need to be
evaluated. This requires both an evaluation or objective function and a fitness function. The
objective function is part of the problem domain and defines the condition for optimality.
Applied to buildings, this evaluation tool could range from a simple analytical function (e.g.
for calculating the U-value of a wall) to a complete simulation programme (e.g. for calculating
the primary energy consumption of a whole building).
The fitness function is part of the algorithm domain and measures how well a particular
solution satisfies the condition set by the objective function (Van Veldhuizen 1999). Based on
the results of the evaluation, the fitness function assigns a corresponding value to each
chromosome, thus rating them in terms of their fitness for the problem, taking into account
the other individuals in the population (Michalewicz 1999, Fonseca 1995). There are
essentially two types of fitness assignment strategies: scaling and ranking (Fonseca 1995).
For scaling, fitness is computed as a function of the evaluation results, ensuring nonnegative fitness for all individuals and giving the best individual in the population a controlled
advantage over the others. Ranking is performed by sorting the individuals in the population
according to the evaluation results and consequently assigning fitness values to individuals
6
according to their position or rank in the population. Applied to minimisation of the energy
consumption of a building, the lower the energy consumption of a variant, the higher will be
its fitness. With scaling, this fitness will be inversely proportional to the difference between
the energy consumption of this variant and the lowest energy consumption within the
population. With ranking, on the contrary, all variants within the population will be sorted
according to their energy consumption and the fitness will be assigned based on their rank in
the population, independently of the magnitude of their energy consumption or of the
difference with the best variant. Scaling is a more traditional approach, but with scaling, an
individual that is much stronger than the others, may be assigned a very large fitness and
through selection rapidly dominate the population. Ranking, on the contrary, eliminates any
sensitivity to the scale in which the problem is formulated. Since the best individual in the
population is always assigned the same fitness, would-be super individuals can never
reproduce excessively (Fonseca 1995). Choosing the appropriate fitness assignment
strategy is important, as the fitness values form the main basis for the genetic operators that
establish the actual genetic optimisation process.
1.2.3.
Genetic operators
Just as in nature, genetic operators operate on the solutions of an existing population trying
to generate a new population with offspring that have higher fitness than their parents (Van
Veldhuizen 1999). The major operators are selection, mutation, recombination and
reinsertion. Many variations on these basic operators are described in the literature (Fonseca
1995, Coello 1996, Michalewicz 1999, Zitzler 1999). Here, the main principles and the most
important variants of the operators are presented.
Selection
Selection is the process of choosing individuals from the old population to participate as
parents in the creation of offspring for the new population. The selection occurs
proportionally to the fitness of the individuals in the old population and is usually performed
stochastically (Fonseca 1995). The most frequently applied selection methods are stochastic
universal sampling (Baker 1987), roulette wheel selection (Goldberg 1989) and tournament
selection (Hancock 1994). In tournament selection, some number of individuals (usually two)
competes for selection. This competition or tournament is repeated population-size number
of times (Michalewicz 1999). Roulette wheel selection and stochastic universal sampling can
both easily be visualised by a spinning roulette wheel with slots proportional in width to the
fitness of the individuals in the population (see figures 1.1 and 1.2 by Fonseca (1995)).
7
CHAPTER 1
Figure 1.1: roulette wheel selection
(Fonseca 1995)
Figure 1.2: stochastic universal
sampling (Fonseca 1995)
In roulette wheel selection, a sequence of independent selection trials is performed and in
each trial, one individual is selected, with the probability for selection remaining constant for
each individual and equal to its normalised fitness. In stochastic universal sampling, only one
trial is performed, but with multiple, equally spaced pointers. When selecting a single
individual, roulette wheel selection and stochastic universal sampling are equivalent. When
selecting multiple individuals, however, stochastic universal sampling guarantees a more
robust selection procedure.
Selection on its own can never improve the population. It would only produce more of the
same. To change and improve the genetic material of the population, modifications on the
genetic material of the selected parents are necessary. Two main categories of modificatory
genetic operators can be defined: mutation and recombination. Figure 1.3 illustrates the
process of recombination and mutation for two string chromosomes that might represent two
variants of the same building.
Mutation
Mutation causes a change in an individual chromosome according to some probabilistic rule.
Usually, only one element or a small part of the chromosome is changed, causing offspring
that inherit most of the genetic information of the parent (Fonseca 1995).
Recombination
Recombination causes the exchange of genetic information between two or more individuals.
This process creates offspring that inherit genetic material of multiple individuals.
Mutation and recombination are both genetic operators that behave according to probabilistic
rules, but they have a distinguished role each in the process of exploitation and exploration
of the search space. By recombining individuals, the available genetic information in the old
8
population is further exploited in the new population. This means that knowledge found at
previously visited points is used to help find better points (Beasley 1993). This way,
recombination of the genetic material of two or more parents into a single individual can
significantly accelerate the search process (Fonseca 1995). Recombination is usually
performed with high probability. Mutation, on the other hand, creates offspring with new
genetic material, thus exploring new and unknown areas in the search space. To avoid that
mutation would reduce the genetic evolution process to a random search process, mutation
is usually performed with a low probability.
The implementation and performance of recombination and mutation strongly depends on
the type of problem. A great deal of research has been put into identifying the best genetic
operators for different classes of problems (Fonseca 1995). For individuals that are
represented by string chromosomes, a typical recombination operator is single-point
crossover. Hereby two individuals exchange a part of their chromosomes to create offspring.
The crossover point is selected at random. Other commonly used recombination operators
are double-point crossover or uniform crossover, in which two or several uniformly distributed
crossover points are selected at random.
Figure 1.3: Example of the process of recombination and mutation for two string
chromosomes that might represent two variants of the same building (Verbeeck 2003)
9
CHAPTER 1
1.2.4.
Convergence of a genetic algorithm
Any efficient optimisation algorithm must find a trade-off between exploration and exploitation
of the search space. Holland (1975) showed that genetic algorithms combine exploration and
exploitation both at the same time in an optimal way. However, this is only theoretically true.
In practice, there are inevitably problems because some simplifying assumptions of Holland
(1975) are not always satisfied (Beasley 1993):

The population size is never infinite

The fitness function does not always accurately reflect the feasibility of a solution for
the problem

There can be a significant interaction between elements in a chromosome
Population diversity
Because the population size is never infinite in practice, it is important to control the diversity
in the population over the optimisation process. Due to the finiteness of the population size,
the performance of a genetic algorithm will always be subject to stochastic errors. Just as in
nature, this can cause genetic drift, thus creating a premature loss of diversity in the
population resulting in the convergence to a sub-optimal solution (Booker 1987). Genetic drift
also means that even without any selective pressure (i.e. all chromosomes in a population
have the same fitness) the algorithm will still converge to a certain solution, simply because
of the accumulation of stochastic errors (Beasley 1993). The smaller the population, the
more vulnerable it is to selection errors and genetic drift. The population size can be
extended, but this has an impact on the computational cost. Higher mutation rates can
prevent genetic drift to a certain extent, but if the mutation rate is too high, the search
becomes purely random. Different alternatives are presented in the literature to reduce the
risk for genetic drift: introduction of random immigrants (Grefenstette 1992), crowding
(Goldberg 1989), fitness sharing (Goldberg and Richardson 1987), mating restriction (Deb
and Goldberg 1989). In the technique of fitness sharing, which is the most frequently used,
the fitness of similar individuals is reduced as they have to share the same resources with
each other. Individuals that are different from each other retain their original fitness. As a
consequence of fitness sharing, the replication of already abundant individuals is
discouraged and the population will cluster around different local optima in the search space.
Such clusters are called niches and represent favourable regions in the search space
(Fonseca 1995). A consequence of a population subdivided in niches is that recombination
from individuals in such different groups often creates unfit offspring outside the favourable
regions. To avoid this, a mating restriction scheme can be added to the genetic algorithm.
The main difficulty, however, of both fitness sharing and mating restriction is to decide how
similar individuals should be to begin to decrease each other’s fitness. A measure of distance
10
between individuals must be defined, but this needs assumptions on the location and
probable number of local optima in the search space and these are generally unknown (Deb
and Goldberg 1989).
Feasible and infeasible solutions
In genetic algorithms and more generally in all evolutionary computation methods, the fitness
function serves as the only link between the problem domain and the algorithm domain. The
fitness function rates individuals in the population: better individuals have better chances for
survival and reproduction. It is therefore essential to define an evaluation function that
characterises the problem in a perfect way and to couple it with an appropriate fitness
function that rates the solutions. The problem of how evaluating individuals in the population
is far from trivial. The search space as well as each population contains feasible and
infeasible solutions, but at the end of the optimisation process, a feasible optimum should be
found. Therefore, finding a proper evaluation measure for feasible and infeasible solutions is
of great importance, as it directly influences the success or failure of the algorithm
(Michalewicz 1999). If constraints have to be taken into account when optimising an
objective, defining a good evaluation function and fitness function is even more important, as
the constraints will increase the number of infeasible solutions. Many important questions
can be addressed: how to compare feasible solutions? What to do with infeasible solutions:
eliminate them, repair them to feasible solutions or penalise them? Several heuristic methods
to handle constraints emerge from the literature, most of them quite recently. Michalewicz
(1999) discusses these methods in a more systematic way.
For most optimisation problems that have an objective to optimise without extra constraints,
the evaluation function for feasible solutions is given. In the case of problems where the
objective is combined with constraints, infeasible solutions can be treated by reducing their
fitness by adding a penalty or a cost for repairing. Another possibility is to use different
weights for the fitness of feasible and infeasible solutions to scale their relative importance.
Both methods, however, allow infeasible individuals to be ‘better’ (with a higher fitness) than
feasible individuals, thus possibly resulting in convergence to an infeasible solution. An
additional weakness is their problem dependence. Selecting the appropriate penalty or the
weights might be as difficult as solving the original problem. Nevertheless, Michalewicz and
Xiao (1995) and Powell and Skolnick (1993) reported good results with this kind of methods
under the assumption that any feasible solution was better than any infeasible one.
Establishing a relationship between the fitness functions for feasible and infeasible solutions
is one of the most challenging problems to resolve while applying a genetic algorithm to a
particular problem (Michalewicz 1999). Several methods of constraint handling are further
presented and discussed in section 1.4 on multi-objective optimisation.
11
CHAPTER 1
Establishing a chromosome
Traditionally, genetic algorithms work with binary strings. Also the theoretical foundations of
genetic algorithms, being the Schema Theorem and the Building Blocks Hypothesis, rely on
binary string representation. In the binary implementation, each element of a chromosome is
coded using the same number of bits. However, the binary representation has some
drawbacks when applied to multidimensional, high-precision numerical problems as it can
create extremely large search spaces in which genetic algorithms perform poorly
(Michalewicz 1999). To move the genetic algorithm closer to the problem space, real-coded
or floating point representation is used more frequently lately. Such a move forces, but also
allows, the genetic operators to be more problem-specific by utilising specific characteristics
of the real space. For instance, two points close to each other in real-value representation
space are probably also close to each other in the problem space (e.g. a building with 10cm
roof insulation and the same building with 11cm roof insulation will have very similar energy
consumption, if all other parameters are the same), which is not generally true in a binary
representation (Michalewicz 1999). Real-value representation creates much room for
creativity when mapping from problem domain to algorithm domain, but careful consideration
is necessary as improper representations may have detrimental effects on the performance
of the genetic algorithm. Although there are little heuristics for representation and there is no
unique combination guaranteeing good performance, choosing wisely may well result in
more effective and efficient implementations (Van Veldhuizen 1999). One of the few, but
important rules, is to choose a representation with little or no interaction between the
elements of the chromosome. Epistasis or strong interaction between elements of a
chromosome creates a situation in which the contribution to fitness of one element depends
on the values of other elements. This has a strong impact on the formation of building blocks.
Without going into details, the role of building blocks in genetic algorithms can be compared
with the function of simple blocks of wood that a child uses to create fortresses (Goldberg
1989). With a high degree of interaction, building blocks cannot be formed and consequently,
the problem will be deceptive (Michalewicz 1999). Detailed explanation on the Schema
theorem and the Building Block Hypothesis can be found in chapter 3 of Michalewicz (1999).
The above sections have been concentrating on the techniques for optimising problems with
a large number of variables and parameters. Special attention was given to genetic
algorithms. However, most engineering problems involve not only a large number of
variables and parameters, but also require the simultaneous optimisation of multiple and
often competing objectives, thus resulting in a multi-objective non-linear optimisation
problem. In the following sections, different aspects of multi-objective optimisation are
presented and current approaches for search and decision making are discussed.
12
1.3
Multi-objective optimisation
In single-objective optimisation problems, one optimal solution can be defined. Multiobjective optimisation, on the contrary, mostly leads to a set of alternative solutions
representing the optimal trade-off between the different objectives. The following sections
present two key concepts of multi-objective optimisation: the processes of search and
decision making and the concept of Pareto optimality and its opportunities for multi-objective
optimisation.
1.3.1
Search and decision making
In multi-objective optimisation, two distinct processes can be identified, being the search
process and the decision making process (Zitzler 1999). The search process refers to the
difficulties of exploring large and complex search spaces for which exact optimisation
methods are unsuitable. Decision making refers to the analysis of the optimal set of solutions
with the aim of finally selecting the best compromise solution (Zitzler 1999, Van Veldhuizen
1999). In single-objective optimisation problems, the search process unequivocally leads to
the single optimal solution, thus making the decision making process of selecting the most
suitable solution redundant. In multi-objective optimisation problems, however, search and
decision making are distinguished processes. Depending on how these processes are
combined, a classification of multi-objective optimisation methods can be established (Zitzler
1999):

Decision making before search

Search before decision making

Decision making during search
Decision making before search
In the first method, the multiple objectives are combined into one single objective. This way,
preference information from the decision maker is implicitly included in the search process.
The advantage of this method lies in the fact that the classical single-objective optimisation
methods can be applied without modifications. On the other hand, it also requires profound
domain knowledge which is usually not available (Zitzler 1999). The weighted sum approach
is a typical example of decision making before search. With this technique, multiple
objectives are aggregated into a single objective by summing the different objectives
according to a different weight. However, the outcome might strongly depend on the selected
weights.
13
CHAPTER 1
Search before decision making
Search before decision making overcomes this drawback, as before and during optimisation,
no preference information is incorporated in the search process. The result is a set of optimal
solutions from which the decision maker makes the final choice. However, this method has
the disadvantage that, by excluding the incorporation of preference information during
search, the complexity of the search space cannot be reduced. Another problem of this
method might be the visualisation and presentation of the set of optimal solutions for more
than three objectives (Zitzler 1999, Pohlheim 1999).
Decision making during search
When decision making and search are interacting during the optimisation, preference
information is incorporated in the search process. Each optimisation step results in a set of
alternative trade-offs, on the basis of which the decision maker can incorporate further
preference information or guide the search. By integrating search and decision making in the
optimisation process, the advantages of both are united.
Analysing the literature on multi-objective optimisation highlights the popularity of the method
of search before decision making (Van Veldhuizen 1999). In real-world problems, it seems
reasonable that decision makers are tended to first perform a search for possible solutions.
Making a decision after search is probably less expensive in the long run than making
decisions without the additional knowledge gained through initial or interactive search (Van
Veldhuizen 1999). A key concept for decision making is the concept of Pareto optimality.
This concept is presented in the next section.
1.3.2
Concept of Pareto optimality
While genetic algorithms are very well suited for dealing with the optimisation of multiple
variables, the concept of Pareto optimality is very useful when simultaneously optimising
multiple objectives (Fonseca et al. 1995, Coello 1999, Gens 2001). A Pareto-based approach
treats all objectives equally during optimisation and tries to deduce the optimal trade off
between the objectives by determining the non-dominated solutions. This way, Pareto-based
concepts are often part of ‘search before decision making’ methods.
In single-objective optimisation, one objective function is to be optimised (minimised or
maximised) and the set of feasible solutions is totally ordered according to the objective
function: if a and b are two feasible solutions for the objective function f, either f(a) ≥ f(b) or
f(a) ≤ (f(b). The goal is to find the solution (or solutions) that give(s) the minimum or
maximum value of f.
14
When several objectives are involved, however, the set of feasible solutions is not totally
ordered, but partially ordered (Pareto 1896, Zitzler 1999). This is illustrated in figure 1.4 for a
maximisation problem for two objectives f1 (e.g. energy savings) and f2 (e.g. cost savings).
Solution B is better than solution C as it provides higher energy savings at higher cost
savings. Solution C performs better than solution D, as despite equal cost savings, C
achieves higher energy savings than D. However, when comparing B and E, neither can be
said superior. Although solution E saves more money, it provides lower energy savings than
solution B. In the terminology of Pareto optimality, solution B dominates solutions C and D,
whereas solution E is indifferent to B. Solution A on the other hand is not dominated by any
other solution and thus called non-dominated or Pareto-optimal.
Figure 1.4: Illustrative example of Pareto optimality in the objective space (left) and the
relations between solutions (right) (Zitzler 1999)
Generally, a solution for a multi-objective optimisation problem is called non-dominated or
Pareto optimal, if there is no other feasible solution that improves one objective without
causing simultaneously deterioration in at least one other objective. In contrast to singleobjective optimisation, where there is a single optimal solution, the final goal of multiobjective optimisation is to find the set of Pareto-optimal solutions, being the set of optimal
trade-offs. Illustrated in figure 1.4 on the left hand side, the final goal of the optimisation
process is to find as many solutions as possible that lie on the dotted line, being the Paretooptimal front. None of the solutions on this front can be considered as better than the others,
unless preference information is included (Zitzler 1999). For two objectives, as in figure 1.4,
the set of optimal trade-offs results in a curve of non-dominated solutions (Pareto-curve), for
three objectives in a surface of non-dominated solutions (Pareto-surface). Pareto
optimisation for more than three objectives is possible, but visualisation is not evident
(Pohlheim 1999).
15
CHAPTER 1
Traditional single-objective approaches, such as the weighting method, are capable of
generating the Pareto-optimal set. By aggregating the different objectives with different
weights into a single objective and systematically varying the weights, the Pareto-optimal set
can be established. However, this approach requires several optimisation runs to achieve a
set of solutions that approximates the Pareto-optimal set (Zitzler 1999). As these runs are
performed independently from each other, synergies can usually not be exploited.
In the last few years, the interest in the integration of the Pareto concept in evolutionary
optimisation techniques has been growing rapidly, as these techniques can generate multiple
alternative trade-offs in a single optimisation run. This growing interest is reflected in an
increasing amount of software and literature devoted to this subject. Of particular interest is
the EMOO website on Evolutionary Multi-Objective Optimisation, which aims at collecting
most
publications
on
the
subject
in
a
publicly
accessible
database
(http://www.lania.mx/~ccoello/EMOO/).
1.3.3
Pareto-based fitness assignment and selection
Genetic algorithms have the potential of finding multiple Pareto-optimal solutions in a single
optimisation run, but for many complex applications, it is not evident to generate the entire
set of Pareto-optimal solutions. Therefore, the optimisation goal may be reformulated as to
find a non-dominated front that is as close as possible to the Pareto-optimal front with
solutions that are well distributed and with a maximised spread of the solutions such that for
each objective a wide range of values should be covered by the non-dominated solutions
(Zitzler 1999). General techniques to control population diversity and common methods for
selection and fitness assignment are discussed earlier. In order to guide the search towards
the Pareto-optimal set, the concept of Pareto optimality should be incorporated in the
selection and fitness assignment.
The idea of calculating an individual’s fitness on the basis of Pareto dominance was first
suggested by Goldberg (1989). He presented an iterative ranking procedure that has been
taken up and modified by numerous researchers, resulting in several Pareto-based fitness
assignment schemes (Fonseca and Fleming 1993, Horn, Nafpliotis and Goldberg 1994).
Generally, a solution is given a rank depending on the number of solutions by which it is
dominated and fitness is assigned by interpolating from the best to the worst rank, eventually
followed by averaging between individuals with the same rank. This fitness is the basis for
selection. Fonseca and Fleming (1993) apply stochastic universal sampling, whereas Horn,
Nafpliotis and Goldberg (1994) apply tournament selection. Both combine the Pareto ranking
with fitness sharing and niche techniques in order to spread the population along the Pareto
optimal trade-off surface. Fonseca and Fleming (1995) noted that, when many competing
16
objectives are involved, pure Pareto-based evolutionary algorithms might perform badly due
to the large dimensionality and the size of the Pareto trade-off surface. Nevertheless, Paretobased techniques appear to be quite popular in combination with evolutionary algorithms and
well performing for two or three objectives (Zitzler 1999).
1.4
Constraint handling
1.4.1
Introduction
When solving real-world optimisation problems, there are not only the objectives to be
optimised, but also constraints to be satisfied. The presence of constraints significantly
affects the performance of any optimisation algorithm (Michalewicz 1996). Due to the
constraints, the search space as well as each population contains feasible and infeasible
solutions. At the end of the optimisation process, however, a feasible optimum should be
found. Smith and Coit (1995) discuss the main characteristics of constraints, being the
number of constraints in a problem, their metric, criticality and difficulty. The number of
constraints can vary as in some multi-objective problems, objectives are reformulated as
constraints or vice versa. Anyhow, the difficulty of satisfying constraints will generally
increase with the number of constraints. The metric is an aspect used to assess the violation
of the constraint as a distance to satisfaction. It can be continuous or discrete. The criticality
is related to the satisfaction of a constraint in absolute terms. Depending on whether the
criticality is hard or rather soft, small violations will or will not be accepted for the final
solution if the solution is superior to others for all other objectives and constraints (Smith and
Coit 1995). Finally, the difficulty can be characterised by the size of the feasible region
compared to the whole search space. Although the difficulty to satisfy a constraint is hard to
assess a priori, an estimation can be made in two ways: by testing how easily a solution that
violates the constraint can be changed into a solution that does not violate the constraint or
by determining the probability of violating the constraint during search (Smith and Coit 1995).
Several methods for handling constraints in multi-objective optimisation problems are
reported in the literature (Richardson et al. 1989, Smith and Coit 1995). The general way of
dealing with constraints is by penalising infeasible solutions. Other techniques that are
applied in genetic algorithms are more or less problem-dependent. The use of special
genetic operators or encoders that maintain the feasibility of the solutions is possible in a few
cases. In other cases, infeasible solutions can be repaired into feasible ones. Repair
mechanisms can be efficient and effective, if infeasible solutions can easily be modified into
feasible solutions without much change of the structure of the parent solutions (Smith and
17
CHAPTER 1
Coit 1995). However, some repair mechanisms introduce systematic bias into the search.
Another method of handling constraints is by rejection of infeasible solutions (‘death
penalty’). This is only efficient if constraint violations occur relatively infrequently during
evolution. Constraints can also be treated as extra objectives. This might affect the
performance of the evolutionary algorithm, as convergence can become problematic if too
many competing objectives are involved. Michalewicz discusses several constraint handling
techniques more in detail in Michalewicz et al. (1996) and Michalewicz (1999). In view of the
methodology developed in this research, only the technique of penalty functions is further
discussed.
1.4.2
Penalty functions
Penalising infeasible solutions is the most common approach in genetic algorithms. The
technique consists mainly of reducing the fitness of the infeasible solution by adding a
penalty to the fitness value. The major question is then how to design the penalty function.
Several researchers studied on rules for the design of penalty functions, resulting in some
hypotheses. Richardson et al. (1989) state that:

“penalties which are a function of the distance from feasibility are better performers
than those which are merely functions of the number of violated constraints,

for a problem having few constraints, and few full solutions, penalties which are solely
functions of the number of violated constraints are not likely to find solutions,

good penalty functions can be constructed from two quantities: the expected
completion cost (i.e. expected cost to repair the solution into a feasible one) and the
maximum completion cost”
and Siedlecki and Sklanski (1989) state that:

“the genetic algorithm with a variable penalty coefficient outperforms the fixed penalty
factor algorithm”
Smith and Coit (1995) give a detailed overview of the state of the art of penalty functions for
genetic algorithms. The simplest method is to apply a constant penalty to the solutions that
violate constraints in any way. A variation for multiple constraints can consist of adding a
metric, based on the number of violated constraints. A more sophisticated and more effective
penalty is to apply a distance metric for each constraint and to add a penalty that depends on
the distance to feasibility. However, in some problems it might be necessary for a good
performance of the search process that exploration of the infeasible region is allowed, yet still
requiring the final solution to be feasible. This is possible if a dynamic term is incorporated in
the penalty function that increases the severity of the penalty for a given violation as the
18
search progresses. This way, highly infeasible solutions are allowed and can be exploited
early in the search, while the search still moves towards the feasible region when the
evolution progresses. Although incorporating both the distance to feasibility and the length of
the search into the penalty function has been proven to be generally effective, some authors
propose further improvements by making the penalty function adaptive to the success or
failure of the search. Aim of adaptive penalty functions is to guide the search towards
attractive regions or away from unattractive regions, based on what has already been
observed. Smith and Coit (1995) finally present future research directions for penalty
functions, being the development of completely adaptive penalty functions and the
combination of multiple constraints into one appropriate penalty function.
1.5
Building-related case studies in the literature
Up to now, the application of evolutionary multi-objective optimisation techniques for buildingrelated issues is quite limited. Most applications concern the optimisation of installation
components. Asiedu et al. (2000) used genetic algorithms for the design of a HVAC duct
system, Wang and Jin (2000) developed an optimal control system for VAV (variable air
volume) air-conditioning systems based on genetic algorithms, Chow et al. (2002) did the
same for an absorption chiller system, Fleming and Purshouse (2002) give a survey of how
evolutionary algorithms are applied to adaptive control systems, whereas Kalogirou (2004)
used genetic algorithms in combination with neural networks to optimise solar systems.
The use of evolutionary algorithms to develop low energy building concepts or to evaluate
energy consumption in buildings has been even more limited. Coley and Schukat (2002)
combined genetic algorithms and thermal simulation to identify low energy building designs,
Ozturk et al. (2004(1), 2004(2)) used genetic algorithms to estimate future projections of
energy use in the residential-commercial building sector in Turkey, whereas Wang et al.
(2005(1), 2005(2)) applied multi-objective genetic algorithms for green building design
optimisation. However, there still is great potential for the combination of evolutionary
algorithms and multi-objective optimisation techniques, in view of the large number of
parameters, variables and objectives that interfere in the search for optimised sustainable
building concepts.
1.6
Conclusions
In general, the literature on evolutionary multi-objective optimisation methods can be
subdivided in two main domains: the literature that focuses on specific theoretical aspects of
evolutionary algorithms (Booker 1987, Coello 1996, 2000 and 2002, Deb and Goldberg 1989,
19
CHAPTER 1
Fonseca and Fleming 1993 and 1995, Grefenstette 1992, Goldberg and Richardson 1987,
Hancock 1994, Horn et al 1994, Michalewicz 1996 and 1999, Smith and Coit 1995, Van
Veldhuizen 1998 and 1999, …) and the literature that reports on the application of
evolutionary algorithms to a specific real world optimisation problem (Asiedu et al 2000,
Coley and Schukat 2002, Erickson et al 2002, Fleming and Purshouse 2002, Lotov 2001,
Ozturk et al 2004(1) and 2004(2), Vrugt et al 2003, Wang and Jin 2000, Wang et al 2005(1)
and 2005(2), Wright et al 2002, …). In the first group, evolutionary algorithms are themselves
the main goal of research and most emphasis is put on the improvement of their
performance, intending to create algorithms that perform well for a wide range of optimisation
problems. In the latter, the technique of evolutionary algorithms is mainly a tool to solve a
specific real world problem. This way, existing algorithms are modified and adapted to the
particular problem by adding problem-specific knowledge to the algorithm. They generally
have no intention of creating an algorithm that is applicable to more than one particular
problem. Such non-standard algorithms enjoy a significant popularity in the evolutionary
computation community. Very often, these methods, extended by the problem-specific
knowledge, outperform other classical evolutionary methods as well as other standard
techniques (Michalewicz 1999). Neither of the evolutionary techniques seems to be perfect
or robust across the whole range of real-world problems. Only the whole family of
evolutionary algorithms has this property of robustness. But the main key to successful
applications lies in heuristic methods that are mixed skillfully with evolutionary techniques
(Michalewicz 1999).
The research reported in this dissertation belongs to the last group, as genetic algorithms
have not been the main subject of the research, but have been used as a tool for
optimisation of extremely low energy buildings. In this view, a rather pragmatic strategy has
been applied:
1. The standard technique of genetic algorithms has been adopted as a starting point.
2. A problem-specific representation has been chosen that easily could be used for both
the genetic process and the calculation of energy consumption, life cycle inventory
and costs.
3. The parameters for the genetic operators have been analysed through a parameter
study in order to determine the most appropriate ones for the problem: population
size, generation gap, cross rate, crossover type, mutation rate, ….
4. The Pareto-concept has been adopted to deal with the multiple objectives of the
problem, by application of fitness assignment through ranking based on a Paretoscore. It is inspired by the approaches presented in the literature, but with the attempt
to find a good balance between the performance of the genetic algorithm and the
20
complexity of the fitness assignment process. This way, a simple Pareto-based
fitness ranking scheme is applied without fitness sharing or mating restriction, as this
scheme proved to perform well for the particular problem.
5. Constraints are incorporated through a penalty function, thus creating a final ranking
score that is the sum of the Pareto-score and a penalty, if one or more constraints are
violated. Tests have been executed to determine the magnitude of the penalty and
the proportion between penalty and violation for this specific problem.
The strategy that is schematically presented here will be further explained and assumptions
will be justified in chapter 4.
21
LIFE CYCLE ANALYSIS
CHAPTER 2.
2.1
Life cycle analysis
History and background
Together with the awakening environmental consciousness and the first reports on the
exhaustibility of our natural resources (Meadows et al. 1972), industrial products and
materials became the subject of environmental assessment. The first interest in life cycle
aspects of products and materials arose in the late sixties and early seventies and focused
mainly on consumption of energy and raw materials and to some extent, waste disposal. The
first multicriteria study was done by H. Teastley in 1969 for Coca Cola with the objectives to
choose between glass or plastic for product bottling, between internal or external bottle
production and between recycling or one-way for the chosen bottle. Contrary to all
expectations, the study revealed the plastic bottle as the best choice. With these first results,
the first discussions appeared on the validity of product comparisons (Ecobalance 2006). In
the late seventies and the eighties, more initiatives arose, such as the foundation in 1979 of
SETAC (Society of Environmental Toxicology and Chemistry) starting from the need for a
forum for interdisciplinary communication on environmental issues, and EMPA starting in
Switzerland with research on life cycle aspects of packaging materials (Astrup Jensen et al.
1997). From the early nineties on, the environmental impact of products and materials
became a more common interest and the domain of life cycle analysis and environmental
impact assessment grew rapidly. This was reflected in the foundation of industrial and
scientific LCA networks, such as SPOLD, LCANET, CHAINET, … and in the rise of
international LCA-oriented journals, such as the International Journal of LCA, Journal of
Cleaner Production, and Waste Management & Research with a rapidly increasing number
of publications on all kinds of LCA studies (Astrup Jensen et al. 1997).
This chapter first gives a short overview of several methods for environmental assessment.
Then, it focuses specifically to the methodology of life cycle analysis. In section 2.3, the LCA
methodology in general is explained, whereas section 2.4 discusses the important issue of
information sources for LCA. Finally in section 2.5, the focus is narrowed towards LCA and
the building sector. Strengths, weaknesses and challenges of the LCA methodology for
buildings are discussed there.
2.2
Methods for environmental assessment
The history of life cycle analysis clearly shows the important role played by the industry in the
development of LCA. Since the early nineties, a number of research programmes has been
conducted, most of them as an initiative of industrial federations or societies. Astrup Jensen
23
CHAPTER 2
et al. (1997) give a non exhaustive overview of research programmes in Scandinavia, the
Netherlands, Germany and the US. These programmes resulted in a range of environmental
tools to assist product developers and other decision makers with the assessment of
environmental impacts from products and processes. Some of these tools have continuously
being updated and are widely applied nowadays, such as the Eco-Indicator (NL) (Pré 2007).
Most of these methods are based on some kind of life cycle assessment, ranging from a
conceptual life cycle thinking over simplified LCA using generic data and focusing on the
most important environmental aspects, to detailed, but expensive and time-consuming LCA.
Anyhow, methods for environmental assessment most often focus on product development,
very few on whole systems, such as a building. The sections below give a short description
of most common environmental assessment approaches.
2.2.1
Life cycle analysis
Life cycle analysis or life cycle assessment (LCA) is a term that covers a wide range of life
cycle assessment approaches, mostly based on the International Standard ISO 14000 series
that provide a consensus framework for LCA. However, since a detailed LCA is expensive
and time-consuming, it is judged not to correspond always to the possible benefits of the
results for some applications. These limitations are particularly acute within contexts where a
rapid decision is required or where a rough first overview of the system’s impacts is needed
(Rebitzer et al. 2004). Therefore, in many applications, simplified LCA’s and LCA thinking
(also called streamlined LCA) are employed. The LCA framework, as presented in the ISO
14040 Standard, will be discussed in detail further on, but here some short definitions are
given in order to have a reference for comparison with the other simplified approaches
presented below.
After defining the goal and scope of a life cycle study, thus determining the system
boundaries, life cycle inventory or LCI is the first step in the life cycle assessment process. It
is a methodology for estimating the consumption of resources and the quantities of waste
flows and emissions caused by or attributable to a product’s life cycle (Rebitzer et al. 2004).
A product or product system is typically composed of unit processes, each representing one
or more activities, such as production processes, transport or retail. For each unit process,
data are gathered on the input of natural resources and on the output of emissions, waste
flows and other environmental exchanges (Rebitzer et al. 2004). All unit processes are linked
through intermediate product flows, thus creating a model for the product system. Outcome
of the LCI is an inventory of all input flows, being the natural resources required for the
24
LIFE CYCLE ANALYSIS
product system, and all output flows, being the emissions and waste generated by the
product system.
Life cycle impact assessment
In a next step, different input and/or output flows are weighted and combined into
environmental indicators. These indicators are used in the life cycle impact assessment to
evaluate a product’s life cycle for a number of impact categories, such as climate change,
stratospheric ozone depletion, tropospheric ozone creation (smog), eutrophication,
acidification, toxicological stress on human health and ecosystems, the depletion of
resources, water use, land use, noise and others (Rebitzer et al. 2004). Advantage of
indicators is the ease of use and the apparent simplicity of the results for interpretation.
However, as for all weighting and aggregation issues, the choice of the different weighting
factors is crucial to provide reliable results, as thoughtless weighting may strongly distort the
comparison between product alternatives.
Life cycle interpretation
Life cycle interpretation is an essential step in the life cycle assessment process, but can
occur at every stage. When comparing two product alternatives, an interpretation purely
based on LCI can be conclusive, but in some cases, a practitioner may also want to compare
across impact categories. The latter however, creates the need for information on the priority
of impact categories. Resolving such issues clearly demands extra information, not only from
natural sciences, but even more from social science and economics (Rebitzer et al. 2004).
2.2.2
Process-chain analysis
Process-chain analysis (PCA) is at first view strongly comparable with life cycle inventory.
However, as in a traditional LCA the inventory analysis is the most time-consuming phase,
the approach of PCA attempts to simplify the process-oriented modelling (Rebitzer et al.
2004). Where LCI mostly considers a wide range of input and output flows based on detail
process information, PCA mainly focuses on energy and greenhouse gas emissions, using
generic data for different materials. Starting from a product, PCA converts the product into its
composing materials and associates the materials with the corresponding amounts of energy
used and GHG emissions emitted during all underlying production steps (Voorspools et al.
2000). In a final step, the energy and emissions related to the production from materials into
the final product are added. To reduce the time for process modelling, PCA’s are based on
average data for all basic composing materials, such as steel, glass, plastics, concrete,….
This way, however, this method neglects the fact that one basic material can result from
25
CHAPTER 2
different production processes, with different input and output flows depending on the final
destination of the material (Voorspools et al. 2000). Therefore, the strength of the approach
is at the same time one of its weaknesses. Another drawback is that processes, such as
engineering and services, cannot be expressed as an amount of material used and thus, are
not taken into account.
2.2.3
Input-output analysis
While process chain analysis decomposes a product or system into its composing materials,
the input/output analysis (IOA) converts a product into its economic components. Each
component is attributed to an economic sector (metallurgy, electrotechnics, chemistry,
services, …). For each economic sector an average product is calculated and characterised
by an amount of energy needed and GHGs emitted (Voorspools et al. 2000). Every input can
also be expressed easily as a monetary value. The life cycle of a product is then interpreted
as a sequence of economic activities, each characterised by an amount of energy and
emissions. Sector-related cost and energy input data are based on the national economic
input/output tables and the national energy balance. In contrast to process-chain analysis,
input/output analysis can take all processes into account that can be linked to an economic
sector and thus, does not exclude processes such as engineering and services.
Nevertheless, the main weakness of the IO approach is the use of an average product as a
characterisation of a whole sector. Furthermore, the number of economic sectors is very
limited (64 in Belgium, not all relevant for each product), thus, risking to have components
that can be attributed to more than one sector. Due to the characteristics of the input/output
approach and the large uncertainties on the energy and cost data, this method may be useful
for a first evaluation of the orders of magnitude of the environmental impact and economic
cost of certain products, but might be too rough for detailed comparison of system
alternatives. Voorspools et al. (2000) combined a PCA and an IOA to calculate the energy
content and indirect greenhouse gas emissions embedded in ‘emission-free’ power plants,
such as nuclear plants, hydro power plants, fotovoltaic systems, etc.
2.2.4
Life cycle costing
Although originally life cycle costing or LCC was not developed in an environmental context,
there is quite some ambiguity about the concept in relation to environmental issues.
Therefore, LCC is briefly discussed here. While the aim of LCA is to assess the total
environmental impact throughout the life cycle of a product or a system, LCA does not
consider economic or societal aspects. Life cycle costing, on the other hand, originally was a
26
LIFE CYCLE ANALYSIS
type of investment calculus used to rank different investment alternatives (Gluch and
Baumann 2004), but it has evolved towards a technique for comparative cost assessments
over a specific period of time. It takes into account all relevant economic factors, both in
terms of initial capital costs and future operational costs and is specifically developed for
building-related contexts (TG4 2003). However, the expansion of the system boundaries to
include the product’s lifetime does not mean automatically that all environmental costs are
included. From the early nineties on, modified environmental accounting tools have been
developed, attempting to include environmental impacts as costs into the corporate
accounting systems (Gluch and Baumann 2004). Life cycle costing and other economic
assessment tools are discussed in detail in chapter 3.
2.2.5
Product-related environmental assessment tools
Historically, environmental assessment of products and materials originated from the
production industry, thus resulting in several environmental assessment tools for sustainable
product development. Scope of this dissertation is the sustainability of low energy buildings
as a whole. However, products and materials as such also contribute to this overall
sustainability. Therefore, in an attempt to give a more complete overview of the current
environmental assessment tools, one of the most important product-related environmental
strategies is briefly presented here, namely Ecodesign or Design for Environment (DfE). In
the literature, both terms appear next to each other, but they are in fact two names for one
concept. DfE is mostly used in USA and Canada, Ecodesign mostly in Europe.
Ecodesign (Duflou and De Wulf 2006)
Ecodesign is an application that uses LCA in the process of product development. Aim of all
ecodesign strategies is to reduce the environmental impact of products, not only during the
production stage, but also during the utilisation stage and the end-of-life stage. The
assessment of the environmental impact from cradle to grave is done by LCA.
For the production phase, the main strategies for ecodesign consist of (1) choosing
production processes with a lower environmental impact, (2) using materials with a lower
environmental impact and (3) reduction of material use, also called ‘dematerialisation of the
design’.
For the utilisation phase, the main strategies consist of (1) minimising the environmental
impact of the logistics by reducing volume, weight, transport distances and/or packaging
materials, (2) extension of the lifetime of the product and (3) optimisation of the efficiency
(energy, emissions, maintenance,…) during utilisation. Care must be taken, however, with
the secondary effects of some decisions due to the mutual dependency of some product
27
CHAPTER 2
characteristics. For instance, in case of a refrigerator, the extension of the lifetime might have
a negative impact on the energy efficiency, as the thermal quality of the insulation degrades
in time. Furthermore, life extension prevents renewal of the refrigerator stock, whereby the
newest generation usually consumes considerably less energy.
To reduce the environmental impact of the end phase, ecodesign focuses on product design
that allows (1) reuse of the product or its components or (2) recycling through selective
dismantling of the product, through destructive separation or through regeneration of the raw
materials.
2.2.6
Building-related environmental assessment tools
LCA is an important tool for evaluating the environmental impact of products or product
systems, as it allows quantification of the global environmental impact. Traditionally, LCA is
mostly concerned with components and product design and hardly considers large systems,
such as buildings, as a whole. However, in the literature and in building practice, several
environmental assessment tools can be found, ranging from simple checklists over building
rating schemes to detailed LCA modelling tools. Anon. (2001(1)), Anon. (2001(2)) and
REGENER (1997) give an extensive overview of a large number of building related life cycle
assessment tools. Some are limited to material and product level, others consider building
parts or buildings as a whole. For some, the use is limited to one or two countries, others are
used worldwide by hundreds of multinationals, consultants, research institutes and
universities. It is beyond the scope of this research to discuss and evaluate all these tools.
One particular research activity on building-related life cycle analysis is reported here, as it
appears to be the main basis for the overviews of Anon. (2001(1)) and REGENER (1997):
the IEA Annex 31 on ‘Assessing the Energy-Related Environmental Impacts of Buildings’.
IEA Annex 31
Annex 31 was executed in the period of 1996-1999. It examined how tools and assessment
methods could be developed and used to improve the energy-related impact of buildings on
interior, local and global environments. The research executed in the project was not limited
to the analysis of existing assessment tools, but also contained an in-depth analysis of
different aspects of environmental impact of buildings. It considered the environmental loads,
effects and impacts, as well as how to influence the decision-making process for
environmental issues or how to assess the adaptability of buildings. All reports are freely
available on www.annex31.org .
28
LIFE CYCLE ANALYSIS
As appears clearly from the IEA Annex 31 research, buildings have specific properties in
comparison with ‘traditional’ industrial products (very long lifetime, large impact of the
utilisation phase, one-of-a-kind character, …), and thus, application of LCA to buildings
requires specific attention. An in-depth discussion of these specific requirements and how
particular case studies from the literature on the environmental impact of buildings dealt with
it is presented in section 2.5, after a more detailed description of the LCA methodology and
of the availability and quality of LCA information.
2.3
LCA methodology
2.3.1
Introduction
The rapidly growing interest in sustainable product development in the nineties was
accompanied by a growing need for methods and tools to quantify and compare the
environmental impacts of products. At the same time, there was a need for standardisation in
order to create a common framework for LCA studies by which LCA results from different
studies could be compared and interpreted in a standardised way. This resulted in the
development of the International Standards of the ISO 14000 series which are nowadays
accepted as a consensus framework for LCA:
-
International Standard ISO 14040 (1997) on principles and framework
-
International Standard ISO 14041 (1998) on goal and scope definition and inventory
analysis
-
International Standard ISO 14042 (2000) on life cycle impact assessment
-
International Standard ISO 14043(2000) on life cycle interpretation
Every product or system has a life, starting with the design of the product, followed by
resource extraction, production, use and finally end-of-life activities. All processes in a
product’s or system’s life result in environmental impacts due to resource consumption,
emissions into the natural environment and other environmental exchanges. In LCA, the
design phase is usually excluded, since this phase itself is often assumed not to contribute
significantly to the environmental impact of the product. However, the design strongly
predetermines the behaviour of the production in the subsequent phases, and thus,
decisions taken in this phase might have a significant impact on the generation of
environmental loads in the other life cycle stages (Rebitzer et al. 2004). Therefore, if LCA is
used for the improvement of the sustainability of products, it should be carried out as early in
the design process as possible. The same can be applied to the design or improvement of a
process within the life cycle of a product.
29
CHAPTER 2
The sections below present the key features of an LCA as defined in the ISO 14000
standards. As these standards are developed as a general framework and applied in a first
place to industrial products, they will be reported as such below. But as the ISO 14040
(1997) mentions: ‘There is no single method for conducting LCA studies. Organisations
should have flexibility to implement LCA practically as established in this International
Standard, based upon the specific application and the requirements of the user.’ Therefore,
further on, the requirements and consequences of applying this framework to the
environmental impact of buildings are discussed.
2.3.2
Framework
The ISO Standards on life cycle assessment originated from the ‘Code of Practice’
developed by the Society of Environmental Toxicology and Chemistry (SETAC). This code
originally distinguished four methodological components within LCA: goal and scope
definition, life cycle inventory analysis, life cycle impact assessment and life cycle
improvement assessment. In ISO 14040 (1997), life cycle improvement assessment is no
longer regarded as a phase on its own, but rather as having an influence throughout the
whole LCA methodology. In addition, life cycle interpretation has been introduced. This
phase interacts with all other phases in the LCA procedure, as illustrated in figure 2.1.
The International Standard recognises that LCA is still at an early stage of development.
Some phases, such as impact assessment, are still in relative infancy. Therefore, it is
important that the results of LCA be interpreted and applied appropriately (ISO 14040 1997).
Life cycle assessment framework
Goal and scope
definition
Direct applications:
-
Inventory analysis
Interpretation
Impact assessment
Figure 2.1: Phases of an LCA (ISO 14040 1997)
30
-
Product development
and improvement
Strategic planning
Public policy making
Marketing
Other
LIFE CYCLE ANALYSIS
Goal and scope
The first step in an LCA is defining the goal and scope of the LCA study. They provide a
description of the product system to be analysed. The goal definition focuses on the intended
application, the reason why the study is performed and the intended audience, whereas the
scope definition describes the system, or systems in case of comparative studies, in terms of
system boundaries, functional unit, allocation procedures, data requirements, assumptions
and limitations (ISO 14040 1997). The functional unit is the essential basis for comparing
goods and services. Usually, the functional unit is not just a quantity of material, but more a
service that the product system(s) should provide (Rebitzer et al. 2004). The scope definition
may require modifications during the LCA process to meet the original goal. Even the goal
definition may be revised due to unexpected limitations or constraints. This way, an LCA
study must be considered as an iterative technique (ISO 14041 1998).
As said earlier, the functional unit is a quantitative description of the service that the
investigated system should provide. In this sense, when comparing product system
alternatives, care must be taken to avoid choosing a too narrow product perspective. This
could result in differences in the functions that the alternatives should provide. This can be
avoided by choosing a broader function-based perspective, based on the needs fulfilled by
the products rather than on the physical products themselves (e.g. ‘lighting’ or ‘cooling of
food’ instead of ‘lamps’ or ‘refrigerators’) (Rebitzer et al. 2004). Applied to buildings, care
must be taken to not limit the functional unit to a building material or building component, but
to consider the overall performance of the building as a whole (functionality, thermal,
acoustical and visual comfort, safety, etc.) according to the IEA Annex 32 performance
matrix for buildings (Hendriks and Hens 2000).
System boundaries and product system modelling
After defining the functional unit, the product system has to be modelled before an inventory
of environmental exchanges of the functional unit can be established. The model of the
product system is typically a static simulation model, composed of unit processes, each
representing one or more activities. The system boundaries define which unit processes will
be included in the system(s) to be modelled. For each unit process, data are recorded on the
input of natural resources, the emissions, waste flows and other environmental exchanges.
The exchanges are typically assumed to be linearly related to one of the product flows of the
unit process. All unit processes are linked through intermediate product flows, what makes
the typical process system model linear with respect to the quantity of function it provides
(Rebitzer et al. 2004).
In practice, lack of time, data or resources will necessarily lead to decisions on which unit
processes will be modelled and to which level of detail these processes will be studied.
31
CHAPTER 2
Obviously, processes should be included that are deemed to contribute significantly to the
studied product and its function. In the same way, decisions need to be made on the inputs
from the environment and releases to the environment to be evaluated and the level of detail
of this evaluation (ISO 14041 1998). The choices and assumptions made during system
modelling, especially with respect to the system boundaries and what processes and
releases to include, are often decisive for the results of an LCA study, although it is difficult to
deduce the exact influence of the choices made on the results (Rebitzer et al. 2004). As for
goal and scope definition, also the system boundaries may need modification or refinement
during the LCA process on the basis of results of preliminary work (ISO 14041 1998).
In principle, all life cycle stages should be incorporated from extraction to manufacturing over
distribution and transportation to utilisation and maintenance and, finally, disposal of wastes
and products and recovery of used products. If not, this should clearly and explicitly be
mentioned in the description of the system boundaries.
2.3.3
Inventory analysis
Data collection
Life cycle inventory is a methodology for estimating the consumption of resources and the
quantities of waste flows and emissions caused by or attributable to a product’s life cycle.
This means that for each process of the product system, a data set is needed. This way, the
inventory process itself is mainly concerned with data collection and calculation procedures.
However, consumption of resources as well as generation of waste and emissions occurs at
multiple sites and regions of the world, at different times and over different time periods and
as different fractions of the total consumption or emission at any one site (Rebitzer et al.
2004). This makes data collection and compilation often the most work- and time-consuming
steps in an LCA. Requirements for data quality and data collection for LCI are discussed
further in the section on information sources.
Allocation
In general, life cycle inventory analysis relies on being able to link different unit processes to
each other by simple material or energy flows. However, in practice, very few industrial
processes yield a single product. In fact, most processes yield more than one product and at
the same time, they recycle intermediate or discarded products as raw materials (ISO 14041
1998). This raises the question how flows should be partitioned and distributed among the
multiple products. This has been one of the most controversial issues in the development of
LCA. The ISO 14041 (1998) proposes a stepwise procedure:
32
LIFE CYCLE ANALYSIS
1. In the first place, allocation should be avoided, wherever possible, by dividing the unit
process in one or more sub processes or by expanding the product system to include
the additional functions.
2. Where allocation cannot be avoided, partitioning of inputs and outputs should reflect
the underlying physical relationships.
3. Where physical relationships alone cannot serve as a basis for allocation, the
allocation should be based on other relationships between products and functions,
e.g. in proportion to the economic value of the products.
These allocation principles also apply to reuse and recycling. However, additional elaboration
is required, because reuse and recycling may imply that flows associated with processes for
extraction and processing of raw materials on the one hand, and final disposal of waste on
the other hand, are to be shared by more than one product system. Besides, reuse and
recycling may change the inherent properties of materials. Furthermore, care should be
taken in defining the system boundaries concerning recovery processes. Figure 2.2 outlines
some conceptual procedures as proposed in ISO 14041 (1998). Distinction is made between
the technical description of the product system and the allocation procedure for recycling.
Technical description of a product system
Material from a product
system is recycled in the
same production system
Allocation procedure for recycling
Closed
loop
Closed
loop
Material from a product
system is recycled in a
different production
system
Material is recycled
without changes to
inherent properties
Open
loop
Open
loop
Material is recycled with
changes to inherent
properties
Figure 2.2: Distinction between technical description of a product system and allocation
procedures for recycling (ISO 14041 1998)
The Technical Report ISO/TR 14049: 2000E provides examples of application of allocation
procedures both for multiple output issues and recycling issues.
33
CHAPTER 2
2.3.4
Impact assessment
Life cycle impact assessment (LCIA) is a methodology that provides the factors for
calculating indicators of the potential impact contributions from wastes, emissions and
resource use, attributable to a product system. LCIA, as described in the ISO 14042 (2000),
consists of both mandatory and optional elements. Mandatory is (Pennington et al. 2004):
1. Selection of the impact categories of interest, the indicators for each category and the
underlying models
2. Classification, being assignment of the inventory data from the LCI to the chosen
impact categories
3. Characterisation
through
calculation
of
impact
category
indicators
using
characterisation factors
Optional elements are (Pennington et al. 2004):
4. Normalisation: calculation of the indicator results relative to reference values
5. Grouping and/or weighting of the results
6. Data quality analysis: this analysis is in fact mandatory in comparative assertions
according to ISO 14042 (2000), but up to now it receives little attention in current
practice
Selection of impact categories
Indicators express the contributions from waste, emissions and resource use in terms of
impact categories. Impact categories include climate change, stratospheric ozone depletion,
acidification,
aquatic
and
terrestrial
eutrophication,
human
toxicological
effects,
ecotoxicological effects, photo-oxidant formation (smog), biotic and abiotic resources, land
use and other physical interventions such as odour and noise.
Characterisation
Indicators for each impact category can be calculated from the inventory data of the product
system using generic characterisation factors resulting from characterisation models. As
emission inventory data are expressed in terms of mass (e.g. kg) released in the
environment per functional unit, the characterisation factors linearly express the contribution
to an impact category of a unit mass (1 kg) of an emission to the environment.
The scope of this research is mainly limited to life cycle inventory. The only impact category
implemented is climate change and the characterisation factor used is the Global Warming
Potential. The relative contributions of different gases to climate change are commonly
compared in terms of carbon dioxide equivalents using Global Warming Potentials (GWPs).
Three different units are generally used: GWP20, GWP100 and GWP500. A gas with a GWP100
34
LIFE CYCLE ANALYSIS
of 20 implies that 1 kg of this gas has the same cumulative climate change effect as 20 kg of
carbon dioxide during a 100 year time period (Pennington et al. 2004).
For the different impact categories, different characterisation models can be considered for
the determination of characterisation factors. In principle, a characterisation model should be
chosen by an LCA practitioner when executing an LCIA. However, in practice, this choice is
mostly considered implicitly by choosing directly characterisation factors. Further discussion
on the key issues for modelling, normalisation, grouping and weighting in LCIA can be found
in Pennington et al. (2004), together with an overview of available models for common
impact categories. This is not further discussed here, as it is beyond the scope of this
research.
Table 2.1: Global warming potentials (GWP) given in kg CO2 eq./kg gas (Astrup Jensen et
al. 1997)
35
CHAPTER 2
2.3.5
Interpretation
As illustrated in figure 2.1, life cycle interpretation is an essential step in the life cycle
assessment process, but interacts with every stage. Related to goal and scope definition, the
interpretation of preliminary results may lead to modifications of the definitions of goal, scope
and/ or system boundaries. Analysis of purely LCI results can be conclusive, as a few data in
a life cycle inventory can often dominate, making the results readily interpretable. However,
an LCA practitioner may also want to compare across impact categories to set priorities or to
resolve trade-offs between product alternatives. To some extent, this can be achieved using
natural science approaches, but limitations remain, as this optional step demands expert
judgment, analysis from different viewpoints and optional application of socioeconomic
techniques. Techniques of normalisation, grouping and/or weighting may be helpful in the
process of interpretation (Pennington et al. 2004):
2.3.6
Conclusions
LCA is a tool for comparing goods and services in an environmental context and for
identifying opportunities for improvement of the environmental impact. The International
Standard recognises that LCA is still at an early stage of development, especially the impact
assessment process. As with any form of impact assessment tool, the science behind LCIA
is continually evolving. Current research includes improving the modelling of environmental
mechanisms, identification of appropriate temporal and spatial differentiation, quantification
of the overall uncertainties and further development of the techniques for comparison across
impact categories (Pennington et al. 2004).
Furthermore, as most standards, these ISO standards remain quite theoretically and do not
provide detailed methodological guidance. Rebitzer et al. (2004) and Pennington et al. (2004)
give an in-depth review of the different ISO life cycle assessment standards. Further
comprehensive and detailed guidelines can be found in Consoli et al. (1993), Guinée et al.
(2002), Hauschild and Wenzel (1998), Wenzel et al. (1997) and others.
2.4
Information sources for LCA
2.4.1
Organisations and networks
Together with the growing interest in life cycle assessment and preceding the development
of the ISO LCA standards, several LCA-related organisations and networks arose from the
need for interdisciplinary communication among environmental scientists and others
interested in environmental issues. With its foundation in 1979, the Society of Environmental
36
LIFE CYCLE ANALYSIS
Toxicology and Chemistry (SETAC) is one of the oldest and its network covers nearly the
whole world. Because of its multidisciplinary approach, the scope of science of SETAC is
often broader in concept and application than that of other LCA-related societies (SETAC
2006).
Another organisation that played an important role in the development of LCA is SPOLD, the
Society for the Promotion of Life Cycle Assessment Development, funded in 1992 as an
organisation of industries interested in accelerating the development of LCA as an accepted
management tool for restructuring company policies towards sustainable development. Its
most important achievement is the development of the SPOLD format for life cycle inventory
data, which has meanwhile found wide acceptance. Further detail is given in the section
below. SPOLD closely co-operated with other organisations, such as SETAC-Europe, on
data availability, data quality and data formats. Since 2002, part of its work is continued
under the auspices of UNEP (United Nations Environmental Programme).
Important European networks for LCA were LCANET (March 1996 - June 1997) and later
CHAINET (December 1997 - December 1999). They both were concerted actions in the EU
Environment and Climate Programme. The main objectives of LCANET were to be a platform
for discussion on LCA research among European research institutes, companies and the
European Commission. In the meantime, there was an immense supply of tools and
concepts providing environmental information, but most developments of tools took place
independently from each other and often without communication with the demand side for
environmental information. Therefore, CHAINET as the successor of LCANET, attempted to
link the different communities concerned with LCA and to establish a toolbox with guidelines
for linking demands for environmental information with supply of environmental information.
2.4.2
Data and databases
Data collection, data quality and data structure
When executing a life cycle inventory, apart from the modelling of the product system, a
large number of data sets is needed covering all processes of the product system. These
data sets are each a compilation of inputs and outputs related to the function or product
generated by the process they belong to. Collection of data is often the most time- and workconsuming step in LCA. However, most product systems contain process types that are
common to nearly all studies, such as energy supply, transport, waste treatment services
and production of commodity chemicals and materials. Because of the global markets, many
of these processes are even similar or identical (Rebitzer et al. 2004). Other processes have
a more local, regional or national character. Both, if data are to be collected specifically for a
given project or for the purpose of creating a database, a number of difficulties may arise:
37
CHAPTER 2
most LCA practitioners have little or no previous knowledge of all the processes for which
data are collected, methodological choices have an impact on the data e.g. the partitioning of
overall inputs and outputs between different functions if a process is providing more than one
product or function, measurement points may be lacking, etc.
At the same time, in the nineties, the LCA community realised that not only data, but also
data documentation is crucial. Data documentation or description of data quality is important
to understand the reliability of the study results and to properly interpret the outcome of the
study. Important parameters for data quality are (ISO 14041 1998):
-
Time-related coverage: the age of data and the length of time over which data are
collected
-
Geographical coverage: the geographical areas for which the data for unit processes
are valid (local, regional, national, continental, global)
-
Technological coverage: the technology mix on the basis of which data are collected
(weighted average of the actual process mix, best available technology, worst
operating unit,…)
Additionally, also the level of detail is important for the data quality. This is related to
parameters of precision, completeness, representativeness, consistency and reproducibility
(ISO 14041 1998).
Up to the nineties, meta-information on temporal, geographical or technical validity of LCI
data was rarely provided. SPOLD initiated the development of a data documentation format
which facilitates extensive documentation of LCI data for processes and services (Weidema
1999). Parallel, the SPINE data reporting and exchange format was developed. SPINE
(Sustainable Product Information Network for the Environment) is a data model and a
database format for LCA that has been developed within the Swedish competence centre
CPM (Centre for Environmental Assessment of Product and Material System) (Carlson and
Tillman 1998). In 2001, ISO published a technical specification on data documentation format
for life cycle inventory data (Rebitzer et al. 2004). More details on the structure of the format
and the specifications of the meta-information can be found in ISO 14048 (2001). Two
database systems (ecoinvent and SPINE, discussed in the section below) try to follow these
specifications. Also a recent US national LCI database (NREL 2002) aims at ISO 14048
compliance (Rebitzer et al. 2004).
Uncertainty and variability
The existence of uncertainties and variations in the LCI data are a difficulty for LCA, as they
represent a limitation for clear interpretation of the LCA results. There are different sources of
uncertainty and variability. Variability is understood as inherent variations in the real world,
38
LIFE CYCLE ANALYSIS
while uncertainty mainly results from inaccurate measurements, lack of data, model
assumptions, etc. (Sonneman et al. 2003). Huijbregts (1998) gives a short classification of
uncertainty and variability:
-
Parameter uncertainty due to imprecise measurements, incomplete or outdated
measurements or simply lack of measurement data
-
Model uncertainty for the model describing the product system and its unit processes.
Care must be taken when performing a parameter uncertainty analysis, as large
model uncertainties might distort the results of the uncertainty analysis.
-
Uncertainties due to choices, inherent to LCA, such as the choice of the functional
unit or the allocation procedure
-
Spatial variability due to the variations of processes in the real world, depending on
their location
-
Temporal variability due to variations in time, occurring frequently for emissions and
technical process characteristics
-
Variability between sources and objects due to inherent differences in inputs and
emissions of comparable processes in a product system
Up to now, reports on uncertainty analysis or studies on the influence of data quality to the
final results are rare to find. Illustrative for this gap is the fact that the IEA Annex 31 report on
sensitivity and uncertainty is the shortest of all, only containing 4 pages with rather general
information (Annex31 2004). In the literature, mostly discussions on the dimensions of
uncertainty analysis in LCA (Norris ND, NREL-LCI, Heijungs 1996), proposals for
approaches for partial domains of the uncertainty problem of LCI and LCA (von Bahr and
Steen 2004) or proposals for quality assessment for LCA (van den Berg et al. 1999) are
found. Sonneman et al. (2003) gives a short overview of the existing methods that have been
proposed during recent years. Distinction can be made between qualitative and quantitative
assessment. Qualitative assessment attempts to describe the data by means of a
characterisation of their quality, whereas quantitative assessment aims at quantifying all the
inherent uncertainties and variations in an LCA. The limitation of the latter, however, lies in
the fact that it is hardly possible to analyse all types of uncertainty. In the methods presented
by Sonneman et al. (2003), distinction is made between methods applicable for uncertainty
analysis of life cycle inventories and methods applicable to uncertainties of LCA. Sonneman
et al. (2003) and Maurice et al. (2000) both report on case studies where a combination of
qualitative assessment by use of data quality indicators and quantitative assessment by use
of stochastic models (Monte Carlo) is applied to the life cycle inventory of two types of
electricity production plants. The qualitative assessment is used to determine the most
39
CHAPTER 2
relevant factors in the LCI data. A sensitivity analysis then, by means of a Monte Carlo
simulation, shows which parameters have most impact on the final results.
Important in this context is also the work of Heijungs et al. that present five numerical
approaches, elaborated to support the life cycle interpretation phase (Heijungs and Kleijn
2000, Heijungs et al. 2005). The methods proposed are:
-
contribution analysis
-
perturbation analysis
-
uncertainty analysis
-
comparative analysis
-
discernibility analysis.
They are implemented in the educational free software tool Chain Management by Life Cycle
Assessment (CMLCA) and used for an analysis of the often-used ecoinvent’96 database
(Heijungs et al. 2005). As some of these approaches are applied to assess the quality of the
results from this PhD research, a short definition of all five approaches is given here. Indepth discussion of the methods and the results for the ecoinvent’96 database can be found
in Heijungs et al. (2005):
-
A contribution analysis decomposes the aggregated results of an inventory analysis
in a way that for each inventory item of a system, the contribution of the different unit
processes for this inventory item can be traced back.
-
A perturbation analysis identifies the sensitive parameters, being the input
parameters of which a small change induces a large change in the selected results.
The factor that couples a small change in input to a change in output is referred to as
the multiplier. Multipliers larger than 1 or smaller than -1 indicate sensitive
parameters, while multipliers close to 0 indicate insensitive parameters. The
advantage of a perturbation analysis is that it allows the researcher to study inherent
sensitivities, even for variables for which no uncertainty indication is known.
-
An uncertainty analysis aims at a systematic study of the propagation of input
uncertainties into output uncertainties. There are two basic ways of running an
uncertainty analysis: by random sampling, such as a Monte Carlo simulation, and by
analytical formulas for error propagation. Apart from Heijungs et al. (2005), also
Maurice et al. (2000) applied both methods in a case study.
-
A comparative analysis is nothing more than a systematic and simultaneously
established list of the LCA results for different product alternatives.
-
A discernibility analysis combines a comparative analysis and an uncertainty
analysis. It is based on comparing product alternatives for a large number of Monte
Carlo runs.
40
LIFE CYCLE ANALYSIS
In the research here reported, the LCI results are analysed and interpreted by means of a
contribution analysis, a perturbation analysis and an uncertainty analysis based on Monte
Carlo simulations. Further details can be found in chapter 5.
Databases for LCI
When considering databases for LCA, distinction can be made between public database
initiatives (free or commercially available) and industry database initiatives (Rebitzer et al.
2004). Over the years, several publicly funded initiatives for developing databases of LCI
data have been undertaken. Many of these databases provide data on the level of life cycle
inventory results for commonly used goods and services. Some databases, such as the
Swedish SPINE and the Swiss “Ökoinventare von Energiesystemen” and its successor
ecoinvent2000 (Frischknecht and Rebitzer 2005) also offer data per technological process
(Rebitzer et al. 2004). In addition, several national-level database development activities in
Japan, USA, Canada, Germany, Australia as well as some international coordination projects
are going on. At the same time, many industry sectors provide data to be used in LCAs.
Parallel with the increasing interest for LCA applications, the demand for high quality,
transparent and consistent LCA data rises as well. Only a few publicly available LCI
databases fulfil these criteria, although not all up to date (Frischknecht and Rebitzer 2005). In
the literature and on the internet, several overviews of existing databases can be found
(Anon. 2001(2), Astrup Jensen et al. 1997, Doka 2004, REGENER 1997, Sidoroff 2004,
SimaPro 2006, UWME 2006). One of the most detailed, complete and coherent public
databases at the moment is the ecoinvent2000 database, developed by the Swiss Centre for
Life Cycle Inventories. The database accommodates more than 2500 background processes
often required in LCA case studies. As most data in ecoinvent2000 are representative for
Europe, this database is selected as the main information source for LCI data within this PhD
research. A short description of the ecoinvent2000 database is given here, based on
Frischknecht and Rebitzer (2005).
Ecoinvent2000 database
In the last decade, several different LCA databases have been developed in Switzerland
within the ETH Zürich and other Swiss Federal institutions. Firstly, the ‘Ökoinventare von
Energiesystemen’ was established in 1996 by ETH Zürich and PSI. It was a life cycle
inventory database including data on current energy supply systems, transport and waste
treatment services, and material supply. In Switzerland at the same time, EMPA built up
experience in compiling LCI's for building materials, components and construction, as well as
consumer goods such as detergents, paper, information technologies and packaging. Under
the lead of EMPA and several other Swiss LCA-institutes, the Swiss Centre for Life Cycle
41
CHAPTER 2
Inventories was founded and the project ecoinvent2000 launched. This resulted in a
harmonised, revised and updated Swiss national LCI database. The database comprises LCI
data from the energy, transport, building materials, chemicals, paper and pulp, waste
treatment and agricultural sectors. The data reflect the production and supply situation in the
year 2000, based on the Swiss and European demand patterns. The major applicability is in
the European context, but selected data sets, mainly on oil or natural gas production, have a
broader international application. The database contains LCI data as well as results for
impact categories based on several commonly used life cycle impact assessment methods,
such as Eco-Indicator’99 (Goedkoop and Spriensma 1999(1), 1999(2)), IMPACT 2002+
(IMPACT 2000+), Ecological Scarcity (Brand et al. 1998) or the CML characterisation
scheme 2001 (Guinée et al. 2002). Uncertainty information is available for unit process data
as well as for LCI results. The database is easily accessible via the internet. The data
exchange format is evolved from the SPOLD data exchange format and corresponds to the
international technical specification ISO/TS 14048. With the help of the EcoSpold free
software, inventory data can easily be uploaded in EXCEL. Several commercially available
LCA software tools, such as Emis, PEMS, Regis, SimaPro, TEAM and Umberto, are tailored
for implementation of ecoinvent data. More details on the structure of the database, the data
quality and the calculation routines implemented in the ecoinvent database system can be
found in section 5.3, in Frischknecht and Rebitzer (2005) and on the ecoinvent website
(www.ecoinvent.ch).
2.5
LCA and the building sector
2.5.1
General
In LCA, products are modelled as a system in order to assess its material and energy
balance (Chevalier and Le Téno 1996). The traditional LCA methodology, however, has
historically grown from the packaging industry and the chemical sector and has mainly been
developed and used for the assessment of industrial products. This makes the application in
totally differently structured sectors, such as the building sector, not evident at all. Firstly, the
end products of the building sector, being buildings and constructions, are neither bulk
material nor serial products. This way, the results from an LCA study of a particular building
cannot be extrapolated as such to another building and the one-of-a-kind character of
buildings makes comparisons difficult (REGENER 1997).
Furthermore, some basic and often implicit hypotheses of the LCA methodology do not cope
with the characteristics of products of the building sector. One of the basic assumptions
made in an LCA is time stability, meaning that the product system is considered as a time
42
LIFE CYCLE ANALYSIS
stable system. This implies that when a product reaches the end of its service life, the LCA
assumes that the resulting waste will be treated as it used to be at the beginning of its
service life (Chevalier and Le Téno 1996). Due to the very long lifetime of building products
(mostly more than 25 years) and buildings (80 years and more), hypotheses on processes of
the end phase such as waste treatment and recycling procedures, will result in highly
uncertain and even unrealistic results. At the same time, before reaching its end of life, most
buildings have been undergoing several refurbishments or renovations often resulting in
thorough modifications of the building. As for each single building, multiple and very different
refurbishment scenarios can be outlined, inclusion of refurbishment and of the end phase in
the global LCA of buildings will inevitably be based on highly uncertain assumptions,
resulting once again in highly uncertain results. Extension of the system boundaries to the
whole lifetime of the building can therefore be considered as debatable.
Another specific property of buildings is the impact of the utilisation phase that is in general
much larger than the impact of the other phases, especially when considering energy use
and greenhouse gas emissions. Furthermore, the energy use is highly dependent on correct
workmanship, on correct maintenance and operation of the building and on the behaviour of
the occupants of the building. Therefore, assumptions on these aspects should be clearly
documented, as they might have a large impact of the final LCI and LCA results.
Figure 2.3: Process-phase model for buildings (REGENER 1997)
43
CHAPTER 2
The process-phase model proposed in REGENER (1997) illustrates the complexity of the
structure of an LCI for buildings (figure 2.3). In this context, Erlandsson and Borg (2003)
propose sequential life cycle thinking for buildings, thus treating the different phases
separately in a life cycle inventory. Depending on the actual boundary conditions, it is then
possible to add up the sufficient life cycle phases, corresponding to the goal and scope
definition (Erlandsson and Borg 2003).
2.5.2
From material to building
Probably because of the complexity of the course of life of a building, researchers in the past
often opted for building materials, building products or building components as subject for
LCA research, also in comparative studies. In the literature, a large number of reports on
case studies on the environmental impact of building materials and products (Börjesson and
Gustavsson 2000, Chevalier and Le Téno 1996, Jönsson 2000, Lenzen and Treloar 2002,
Venkatarama and Jagadish 2003) or building components (Chevalier et al. 2002, Citherlet et
al. 2000, Erlandsson et al. 1997, Fossdal 1996, Weir and Muneer 1998) can be found.
However, by limiting the functional unit of LCA to a building material or building component, a
number of functions or boundary conditions, that should be provided by buildings, such as
thermal comfort in winter and summer, indoor air quality, etc., cannot be taken into account.
This way, decisions based on isolated LCA for materials or components might lead to
unexpected secondary effects when the materials or components are applied in buildings
without taking into account their interaction with other components and their impact on the
performance of the building as a whole. This does not imply that it is not possible to improve
materials and components. On the contrary, existing LCA techniques apply as well to the
production of building materials as to any kind of other material (REGENER 1997).
Recently, research has been executed on the application and adaptation of the LCA
framework to buildings as a whole, especially within the EU-REGENER project and the IEA
Annex 31. In analogy with the proposal of Erlandsson and Borg (2003) to treat the different
life cycle phases separately, the EU REGENER project has established a basic framework
composed of partial models for the different phases and processes (REGENER 1997).
Sensitivity studies from the REGENER project showed the importance of a good design, but
also of appropriate occupants’ behaviour in order to achieve good environmental
performance (Peuportier et al. 1997). The developed methodology is also applied to the
comparative evaluation of single family houses in France, being the present construction
standard in France, a solar house and a wood framed house (Peuportier 2001).
On several aspects overlapping with the EU REGENER project due to collaboration, is the
earlier discussed research executed in the IEA Annex 31 on Energy-Related Environmental
44
LIFE CYCLE ANALYSIS
Impact of Buildings. Aim of Annex 31 was not so much the development of an LCA
methodology for buildings, but more to provide information on how to improve the energyrelated environmental impact of buildings. This way, Annex 31 was more an application of
the developed methods, as it focused on how assessment tools and methods might improve
the energy related impact of buildings (Hobday 2005).
In the literature, more reports can be found on LCA studies of particular buildings in Sweden
(Adalberth 1997(1), Adalberth 1997(2), Erlandsson and Borg 2003, Thormark 2002),
Germany (Binz et al. 2000), USA (Blanchard and Reppe 1998, Scheuer et al. 2003) and
Australia and New Zealand (Anon. 2001(1), Anon. 2001(2), Mithraratne and Vale 2004). Due
to the one-of-a-kind character of these buildings and the building tradition that may quite
differ from country to country, it is not evident to draw general conclusions from these case
studies. Nevertheless, in analogy to the results of the REGENER project and to earlier
research on the energy performance of buildings, all case studies emphasise the importance
of the operational phase, when comparing the environmental impact of the different life cycle
phases of a building. Scheuer et al. (2003) reports that the operational phase of a new
university building in Michigan accounted for 83% of the inventoried environmental burdens.
He concludes that the optimisation of the energy performance during the utilisation phase of
the building still should be the primary emphasis for design, until it is evident that there is a
significant shift in distribution of life cycle burdens. Same conclusion was drawn in Adalberth
(1997(2)) where the analysis of three dwellings, built in 1991 and 1992 with an energy use
for space heating and ventilation between 64 and 83 kWh/(m²
floor area,a),
revealed that 85% of
the total energy use was required during the utilisation phase. Similar conclusions are found
in Peuportier (2001) and Blanchard and Reppe (1998). High performance or very low energy
houses might show significantly lower contributions of the operational phase. Thormark
(2002) reports on the most energy efficient apartment housing in Sweden (in 2002, overall
energy needed for operation estimated at 45 kWh/m²a), having an embodied energy
accounting for 45% of the total energy need over a life span of 50 years. By paying attention
to the recycling potential, however, 35 to 40% of the embodied energy could be recovered
through recycling.
Another quite general conclusion is the significant contribution of materials and products with
a high replacement rate such as finishes, appliances, furniture, carpets, etc. Not only the
relatively short life of these products (Mithraratne and Vale 2004), but also the fact that
several of these products are found to have high material production energy intensities
(Scheuer et al. 2003) are responsible for the considerable contribution to the impact of the
operational phase. These materials and products, however, are not taken into account in this
research.
45
CHAPTER 2
2.6
Conclusions
Due to the growing environmental awareness, life cycle assessment has become an
indispensable part of sustainable development. However, the uncertainty still remains high.
As the overall methodology focuses on energy, emissions and cost, only primary energy use,
GWP and emissions of CO2, NOx, SOx, NMVOC’s and particles will be taken into account.
Since the life span of a building exceeds the utilisation period of one generation, resulting in
large uncertainties on modifications and destination of the building, the adopted time scale
within this research is the utilisation phase by one generation plus the phases upstream
(extraction, production, transport, construction). No assumptions will be made on the end
phase.
However, in contrast to much research, the functional unit will not be limited to building
materials or building components, but buildings as a whole will be considered in order to take
into account the overall performance of the buildings and to be sure that buildings with a
same level of performance are compared.
46
COST ASSESSMENT
CHAPTER 3.
3.1
Cost assessment
Introduction
An important principle in cost assessment is that the definition of a cost depends on who the
decision maker is (Proost and Van Regemorter 2000). If the decision maker is a household,
only the cost that is effectively encountered by the household is relevant. This cost includes
taxes, market prices and subsidies (Proost and Van Regemorter 2000). It corresponds to the
micro-economic approach, as it only considers the economic behaviour of a small economic
entity, here a household. If the decision maker is a government, the macro-economic
approach should be adopted, that takes into account the transfers of money and changes in
industrial production, unemployment, price levels, inflation, etc. on the level of the (supra)
national economy. Thus, the impact of the decision on the society as a whole should be
considered.
When assessing the economic impact of energy saving measures in buildings, both the
micro- and macro-economic approach can be adopted, and they are strongly related.
In a macro-economic analysis, the focus is put on the costs and benefits of an energy policy
trying to enhance investments in energy saving measures, taking into account the expected
and unexpected transfers and changes for the economy as a whole that are induced by this
policy. Such an energy policy normally includes both regulatory and voluntary policy
instruments (Lee and Yik 2004). One of the recent regulatory instruments for buildings in
Flanders is the Flemish Energy Performance Regulation, an implementation of the European
Performance of Buildings Directive (EPBD). It is a mandatory code to control the primary
energy use in dwellings and utility buildings (EPB 2005). This is combined with voluntary
incentives, such as tax reduction for roof insulation, high performance glazing, replacement
of old boilers, a.o. and subsidies from local communities and distribution network operators
for insulation measures, replacement of single glazing, installation of solar collectors, a.o. For
the evaluation of the cost-efficiency of such a policy, a wide range of costs and benefits need
to be considered. There is not only the private cost for the consumer, being the investment
cost and the financial savings, which has an impact on his purchasing power. There are also
the tax losses for the government, from the direct tax reduction for the investments to the
indirect tax reduction due to lower energy consumption in the enhanced buildings. Besides of
that, externalities related to the environment have to be incorporated, not only the costs for
restoration, control and prevention of environmental effects of buildings, but also the
reduction of these costs due to the energy saving measures. Furthermore, changes in
industrial production, unemployment and price levels in the building sector are highly
probable to be induced by this policy and should therefore taken into account. This makes
47
CHAPTER 3
the macro-economic evaluation of energy saving measures in buildings very complex.
Several studies have been executed to evaluate the macro-economic impact of measures to
reduce the greenhouse gas emissions, on a general level (Sanne 2000, Proost and Van
Regemorter 2000, Bossier et al. 2002, Moons 2003, Rousseau and Proost 2004) and
specifically for the building sector (Jakob 2006), the transport sector (Proost et al. 2002,
Knockaert and Proost 2005), the forestry sector (Garcia-Quijano et al. 2005) etc.
Nevertheless, when it comes to energy saving investments in buildings, the private building
owner is the final decision maker. In his decision making process, he will not incorporate
items, such as changes in social costs or tax redistribution. He will only try to trade off the
initial investments and the financial savings resulting from the energy saving measures and
relate them to the financial breathing space at the moment the investments need to be made.
Therefore within the context of this research, the choice has been made for a cost-benefit
analysis on a micro-economic scale, in order to evaluate the real financial cost for the
building owner over the period he uses the building. The technique and assumptions of a
cost-benefit analysis as well as references from the literature are described in the next
section. Section 3.3 focuses on the rebound effect, a well-known behavioural phenomenon
by which enhanced energy efficiency is countered with adaptation in behaviour and spending
patterns, resulting in lower energy savings than theoretically estimated. In the last section of
this chapter, some theories about decision making on energy saving measures are
discussed in order to give a start in answering the question why in practice, only a few
private building owners effectively decide to invest in energy saving measures, despite of all
incentives and economic viability.
3.2
Cost-benefit analysis
3.2.1. General
Cost-benefit analysis is a technique developed to evaluate investments related to projects. It
can be applied to any decision that involves a reallocation of resources and helps decision
makers to choose among several alternative projects (Moons 2003). Generally, cost-benefit
analysis does not only consider the real financial costs of the projects, but is also applied to
evaluate impacts other than financial costs, by valuating these impacts in monetary terms.
So, in order to complete a full cost-benefit analysis, six steps need to be considered (Moons
2003):
1. definition of the project and the relevant population;
2. identification of the relevant impacts;
3. valuation of relevant costs and benefits in monetary terms;
48
COST ASSESSMENT
4. aggregation of benefits and costs over time by discounting;
5. comparison of total discounted benefits with total discounted costs, to produce the
net present value (NPV);
6. sensitivity analysis on important parameters such as discount rate, project life
span and cost and benefit estimates.
Depending on the extent of impacts that are incorporated in the analysis, cost-benefit
analysis can serve as an alternative for other environmental decision-making tools, such as
multi-criteria analysis or environmental impact analysis (LCA).
However, the technique of cost-benefit analysis can also be applied in a narrower context as
a purely financial evaluation tool for energy saving measures done by an individual
consumer, without considering their environmental impact. In this research, the
environmental impact is already assessed through a detailed life cycle inventory. There is no
need for further valuation of environmental impacts in monetary terms. But there still remains
the need for an economic evaluation of the energy saving measures. Therefore, a costbenefit analysis is applied on the level of the private building owner, not as a single decisioncriterion, but incorporated as the economic objective in a multi-objective life cycle
optimisation.
3.2.2. Total and net present value
One of the important issues in a cost-benefit analysis is the discounting of costs and benefits.
Discounting reflects the time value of money, meaning that expenses and savings in the
future are not valuated as high as present values (Tommerup and Svendsen 2006). The time
value of money, expressed as a discount rate, depends on inflation, cost of capital,
investment opportunities and personal consumption preferences (Gluch and Baumann
2004). Being ‘a’ the discount rate, the present value PV of a future expense FV at year n
from now can be calculated as follows:
PV =
FV
(1 + a ) n
[3.1]
Equation [3.1] shows clearly that with a discount rate of 0%, the present value equals the
future value, meaning that timing does not matter. With a discount rate > 0%, the higher the
discount rate, the more importance is given to what happens in the near-present and the less
importance to what happens in the distant future (Gluch and Baumann 2004). By
discounting, incoming and outgoing payments from different times become comparable and
can be aggregated into a total present value or a net present value. In this research, a total
49
CHAPTER 3
present value (TPV) considers all present and future costs as absolute values, whereas a net
present value (NPV) considers costs and benefits relative to a reference case.
Assuming that investments in energy saving measures in buildings occur at present and also
in the future due to replacements, and that these investments result in annual energy cost
savings that are constant during the lifetime or until larger refurbishment is necessary, the
total present value (in €) and the net present value (in €) can be calculated as follows:
TPV = I 0 +
∑
I j (1 + rI ) j
j = x, y, z
⎡
NPV = − ⎢ I 0 +
⎢⎣
∑
(1 + a) j
j = x, y , z
n
+∑
I j (1 + rI ) j
(1 + a ) j
i =1
n
K E (1 + rE ) i
K M (1 + rM ) i
+
− R0
∑
(1 + a) i
(1 + a) i
i =1
n
+∑
i =1
K M (1 + rM ) i ⎤ n ∆K E (1 + rE ) i
+ R0
⎥+∑
(1 + a) i ⎥⎦ i =1 (1 + a ) i
[3.2]
[3.3]
with:
I0
Ij
KE
∆KE
KM
n
R0
rI
rE
rM
a
the initial investment [€]
the investment for replacement j at time x, y or z [€]
annual energy cost [€]
annual energy cost saving according to a reference case [€]
annual maintenance cost [€]
considered time period for evaluation [year]
residual value of the building at time n
change of the investment cost above inflation
change of energy cost above inflation
change of maintenance cost above inflation
discount rate or real interest rate, corrected for inflation
Other related financial methods to assess energy-saving measures concern the internal rate
of return, the simple payback time and the dynamic payback time.
-
The internal rate of return (IRR) is strongly related to the concept of net present
value, as it is defined as the discount rate that results in a net present value of zero. It
is the real interest rate for which the investment will turn out viable within the
considered period. If the assumed discount rate is higher than the internal rate of
return, this means that from an economic point of view it is more interesting to invest
differently, as the expected return from the investment will be higher. The IRR is
calculated iteratively from equation [3.3], with NPV=0 and a=IRR. However, care
must be taken with the IRR, as it does not take into account the amount of capital that
is used.
50
COST ASSESSMENT
-
The simple payback time (SPBT in years) is based on the initial investment cost and
the annual energy cost savings. It is calculated without taking into account inflation or
discounting of costs and savings:
SPBT =
I0
∆K E
[3.4]
Although the SPBT neglects the time value of money, it is a fairly good tool for
comparing different energy-saving measures with a short lifespan (less than 5 years),
but it is less well suited as a basis for decisions that have consequences running 50
to 100 years into the future (Tommerup and Svendsen 2006).
-
To counter this drawback, the dynamic payback time (DPBT in years) is defined,
similar to the simple payback time, but taking into account discounting of the benefits.
The DPBT can be calculated from the following equation:
DPBT
∑
t =0
∆K E
= I0
(1 + a ) t
[3.5]
In the literature, several reports are found in which the total or net present value is used to
assess the cost-effectiveness of energy-saving measures in buildings. Tommerup and
Svendsen (2006) used the net present value to assess the potential for energy savings in the
Danish residential building stock, whereas Jakob (2006) applied it to the Swiss residential
sector. Verbeeck and Hens (2002, 2005) calculated the trade off between the TPV and the
primary energy consumption over a period of 30 years for energy saving measures in
retrofitted dwellings in the Flemish Region of Belgium and deduced from the results a logical
hierarchy for energy saving investments. De Coninck and Verbeeck (2005) did a similar
study for both dwellings and office buildings in the Brussels Capital Region of Belgium. Not
only the total and net present value were calculated, but also the internal rate of return and
the simple and dynamic payback time, for both new built and retrofitted buildings. Results of
this study are discussed more in detail in chapter 7 of this dissertation. Alanne et al. (2006)
applied the technique of net present value to analyse the financial viability of cogeneration in
single-family dwellings in two locations in Canada. Most studies included a sensitivity
analysis with respect to at least energy prices and discount rate (Verbeeck and Hens 2002,
De Coninck and Verbeeck 2005, Alanne et al. 2006, Tommerup and Svendsen 2006).
51
CHAPTER 3
A concept that is strongly related to cost discounting, but is specifically applied within the
building sector, is the concept of life cycle costing (LCC). LCC is defined in ISO 15686 (2000)
(on service life planning of buildings and construction assets) as
‘the total cost of a building or its parts throughout its life, including the costs of
planning, design, acquisition, operations, maintenance and disposal, less any
residual value. [It] is a technique which enables comparative cost assessments to be
made over a specified period of time, taking into account all relevant economic
factors both in terms of initial capital cost and future operational costs. In particular, it
is an economic assessment considering all projected relevant cost flows over a
period of analysis expressed in monetary value.’
It is, however, a somewhat confusing term, as within the literature, this term is used for two
distinguished approaches. Wang et al. (2005) applied life cycle costing as defined above,
thus summing the initial construction cost and the discounted annual operating cost for a
building with a 40-year life expectancy. Gluch and Baumann (2004) on the other hand,
analyse the usefulness of the life cycle costing approach as a tool for environmental
decision-making, by attempting to include also environmental impacts as costs into the life
cycle cost. A comparable approach to integrate economic costs and environmental impact
assessment is found in the report of Task Group 4 (TG4 2003) of the Working Group
Sustainable Construction of the EU Commission. This way, the approach of Gluch and
Baumann (2004) and TG4 (2003) approximates more to the broad definition of cost-benefit
analysis as defined in Moons (2003), whereas the approach of Wang et al. (2005) is a
narrower interpretation of cost-benefit analysis that comes closer to the approach adopted in
this research.
3.2.3. Discount rate and price evolution
An important aspect of discounting costs is the choice of the discount rate, as it has a large
impact on the final result (Moons 2003). The discount rate expresses the point of view of
society and, in case of energy saving investments in buildings, it determines whether money
should be invested in buildings rather than in other economical sectors. When considering
the case of the private builder or building owner who wants to invest money in energy saving
measures, two scenarios may occur:
1) The private building owner has the money available. By investing it in energy saving
measures, he will loose part of his purchasing power and will not be able to invest it
alternatively, thus being deprived of a direct investment return. So, the discount rate
52
COST ASSESSMENT
to be applied is the expected real interest rate, being the expected nominal interest
rate minus the expected inflation rate. However, the energy saving measures will
result in an annual lower energy cost, thus partly increasing his purchasing power
and his capital for investment. So the investment can be considered as cost-effective
over a certain period, if the discounted savings equal at least the deprivation of
purchasing power and direct investment return over that period.
2) The private building owner has to borrow the money for the investments. In this case,
the discount rate to be applied is the real interest rate of the loan. The investment is
cost-effective, if the discounted savings cover at least the overall cost of the loan,
being the capital plus interests minus eventual fiscal tax reduction.
As generally the difference between the interest rate on risk-free investments, such as
government bonds, and the interest rate on loans is low, the same interest rate can be
adopted for both cases.
However, long term projections on interest rate, inflation rate and energy prices are
connected with huge uncertainties (Tommerup and Svendsen 2006). Therefore, sensitivity
analyses are indispensable in order to investigate the impact of the assumptions for real
interest rate and energy prices on the final results. Tommerup and Svendsen (2006)
calculated the energy saving potential for the Danish building stock over a period of 30 years
and considered two levels for the real interest rate: a real interest rate of 2.5% and of 0%.
The latter is by them called ‘a sustainable level’, because of the importance that is given to
the future energy savings. For the energy prices, two scenarios were considered: one
corresponding to the present Danish energy prices (0.08 €/kWh) and one corresponding to
double the price. Jakob (2006) applied for the Swiss context over a period of 30 years a real
interest rate of 3.5% for private building owners and 5% for institutional investors. For the
energy price, only one fixed price of 0.055 CHF/kWh (0.036 €/kWh) was assumed, but two
scenarios for price increase due to a CO2 tax were considered: a CO2 tax of 100 CHF/t CO2
(66 €/t CO2) and one of 210 CHF/t CO2 (139 €/t CO2). For the viability analysis of a
cogeneration system, Alanne et al. (2006) adopted the current energy prices from Ottawa
and Vancouver as reference prices and a change of 15%, based on the evolution of the
energy price in 2002. This assumption was estimated by Alanne et al. (2005) as rather
conservative, compared to the long term price evolution. Real interest rates of 3% and 10%
were considered. For the payback period, both short (5 and 10 years) and longer (15 and 20
years) periods were considered. Individual consumers mostly prefer relatively short payback
time, whereas longer payback times need to be considered when introducing new,
sustainable energy technologies (Alanne et al. 2005). Verbeeck and Hens (2002, 2005)
assumed an utilisation phase of 30 years, a net discount rate of 5% and an annual net
53
CHAPTER 3
increase of the energy prices of 2%. Despite the high uncertainty of these assumptions, the
trends that were found did not change significantly when other assumptions on the discount
rate and price changes were used. In the study for the Brussels Capital Region, De Coninck
and Verbeeck (2005) considered three different discount rates, depending on the type of
investor: a discount rate of 4.5% was adopted for individual investors, 4% for the public
sector and 6.5% for professional real estate investors. For the energy prices, the current
energy prices were adopted, together with three scenarios for the evolution of the energy
prices: a low (0%), medium (± 2%) and high (3-4%) scenario, with differentiation between oil,
gas and electricity prices. The utilisation period to evaluate the investments was set at 40
years.
As a remark it should be mentioned that fiscal cost depreciation of building investments is not
considered within the context of this research. An individual consumer cannot deduct the
cost of the investment and enjoy tax benefit, in contrast to companies who can depreciate
the cost of an asset over its useful life. Since all cost assumptions are made from the point of
view of the private builder or building owner, depreciation of investments does not need to be
taken into account. Furthermore, no subsidies or fiscal reduction for energy saving measures
is taken into account. This is important for the interpretation of results, as due to subsidies or
fiscal incentives, energy saving investment that are not economically viable (negative NPV
for a discount rate of 4%) from the point of view of society, might turn out to be economically
viable for the private builder (equivalent to a positive NPV for a discount rate of 2%), whereas
from the point of view of society, it would have been better to invest this money elsewhere.
3.3
Rebound effect
3.3.1. General
During the last decades, considerable effort has been expended by engineers and
economists to identify technically feasible and economically efficient opportunities to save
energy in the industry, transport sector and households (Ramesohl 1999). However, the
benefits of these opportunities often evoke a behavioural response by the consumers that
results in lower energy savings than should be expected by a purely engineering approach.
The financial profit from energy savings essentially can be considered as equivalent to a
reduction of the price for energy services (Berkhout et al. 2000). A price reduction has not
only a direct price effect, but also indirect effects. Assuming a profit maximising consumer
buying e.g. a more energy efficient car, the direct price effect will lead to more use of the
transport services by the car (more kilometres), since the use of the car per km costs less.
54
COST ASSESSMENT
So, part of the energy conservation due to the improved technology will vanish. In addition,
the reduced energy bill enlarges the purchasing power of the consumer, thus leading to more
expenditure on other commodities, among which also commodities that require energy use
(Berkhout et al. 2000). There might even be a third effect, on a macro level, as the shift of
spending patterns of households and companies can lead to a shift in sales patterns of the
production sector (Birol and Keppler 2000). All these effects have consequences for the
energy demand and are denoted as the rebound effect. In the literature, the magnitude of
this effect is strongly under debate. While some argue that it should be small in most cases,
others suggest that it could even exceed the initial energy reduction, the so-called backfire
effect (Jaccard and Bataille 2000). As the controversy over the rebound effect appeared
everywhere, the scientific journal ‘Energy Policy’ dedicated a special issue to the rebound
effect (Energy Policy 2000). The question was not whether such an effect exists, but rather
how much the effect appears, how rapidly, in which sectors and in what manifestations
(Schipper 2000). A detailed presentation and discussion of all scientific opinions on the
rebound effect is beyond the scope of this research, but can be found in Energy Policy
(2000). In the next section, one of the effects of rebound, the direct price effect due to energy
saving measures in buildings, such as insulation and high efficient heating systems, is
discussed more in detail.
3.3.2. Rebound effect applied to energy saving measures in buildings
Systems for heating and domestic hot water production, fittings for lighting and electrical
domestic appliances, they all need energy to deliver their energy services, being thermal
comfort, visual comfort, hot water, etc. Domestic energy services, such as thermal comfort,
however, are in many ways different from other goods and services, as they are not available
as such on the ordinary supply/demand market (Cuijpers 1996). Moreover, the building
occupant is not only consumer of energy services, but also producer, as in case of thermal
comfort e.g., he combines the input of energy with choices for heating systems, insulation,
etc. to produce comfort. This makes the price of these energy services endogenous and
varying from household to household. Depending on the choices for energy carrier, heating
system, insulation thickness, etc., the price of an additional unit of comfort, the so-called
shadow cost of comfort, will differ amongst dwellings and amongst households (Verdonck
and Hens 1998). The more efficiently energy is transformed into the demanded energy
service, the lower the shadow cost of comfort. However, with a profit maximising consumer,
this reduced energy bill due to improved efficiency will result in a higher demand of energy
services, in the same way an energy price reduction would do. So, application of energy
saving measures in buildings can be expected to boost the demand for thermal comfort,
55
CHAPTER 3
resulting in lower energy conservation than theoretically expected. This is shown in figure 3.1
(Haas and Biermayr 2000), where the initial situation (E0, S0) represents the comfort level S0
and energy consumption E0 in a badly insulated dwelling (efficiency η0). When improving the
energy performance of the building to an efficiency level η1, the theoretical energy
consumption will be E1th. However, due to the energy cost reduction, the profit or comfort
maximising consumer will increase its comfort level up to S1, resulting in a actual higher
energy consumption E1pr than theoretically calculated. This way, the actual savings will be
lower than the calculated ones, but the comfort level will be higher than in the initial case.
This illustrates the rebound effect of energy saving measures in buildings.
Figure 3.1 Relationship between energy consumption, efficiency improvement, comfort
levels and the rebound effect (Haas and Biermayr 2000)
In the literature, several reports on the rebound effect in energy efficient buildings can be
found, both theoretical analyses and reports on empirical evidence. Cuijpers (1996)
developed a micro-economic behavioural model to calculate the rebound effect of an
improved insulation level for buildings and applied it to an estimation of residential heat
demand and heat production, based on statistical data for Belgian household expenses. This
resulted in an estimation for the rebound effect of 31%. Verdonck (1999) adopted the same
behavioural model to analyse the difference in energy use between electrically heated and
gas or fuel heated dwellings and the impact of improved insulation on it. The reference case
for both types of dwellings was an insulation level K70 (Umean≈ 0.70 W/m²K). Based on
empirical data for energy consumption and calculations of the shadow costs for comfort, the
rebound effect appeared to be very limited for dwellings on gas or fuel, whereas for
electrically heated dwellings, a rebound effect of 30% was found. This difference could be
56
COST ASSESSMENT
explained from the difference in demand curve between the electrically heated dwellings and
the dwellings heated with gas or fuel. Due to the already high comfort level in the reference
case of the latter, an energy price reduction or an improvement of the energy efficiency only
has a minor effect on the demand for comfort or economically spoken, we are operating in
the inelastic part of the demand curve. In case of the electrically heated dwellings, the much
higher electricity price effectively curbs the energy consumption, resulting in a lower comfort
level. Therefore, a price reduction or an improvement of the efficiency will effectively result in
a higher comfort level, with a higher energy use than theoretically expected. Haas and
Biermayr (2000) analysed the rebound effect for space heating in Austria. Different
approaches were applied on empirical data, all providing evidence of a rebound effect
between 20 and 30%. Milne and Boardman (2000) reported on the effect of energy efficiency
improvements in low-income houses in Great Britain. They found that about 30% of the
benefit was taken as a temperature increase and the rest as an energy saving. Greening et
al. (2000) provide a review from the literature on empirical evidence for the size of the
rebound effect in buildings, concluding that 10 to 30% of the energy consumption savings is
taken back by price effects, substitution or income effects. That the largest rebound effect
can be expected when retrofitting old energy-devouring houses is also stated in Greening et
al. (2000). They dedicate, however, part of the size of the rebound effect to the underlying
assumptions regarding consumer behaviour. Many conclusions on energy conservation are
based on theoretical calculations of the energy consumption before and after installation of
energy saving measures, assuming the same behaviour or the same comfort level in both
cases. This leads, however, most of the time to an overestimation of the energy consumption
before improvement as most of these dwellings originally have very low thermal comfort.
Henly et al. (1988) dedicate 25% of the rebound effect to lack of proper benchmarking.
3.4
Decision making on energy saving investments
3.4.1. Objective
A cost-benefit analysis as described above is considered an interesting economic tool to
support the decision-making process on investments, such as investments in energy saving
measures in buildings. Witness hereof is the large amount of papers in the literature that
report results from a cost-benefit analysis of energy saving measures in buildings (Verbeeck
and Hens 2002, Gluch and Baumann 2004, Wang et al. 2005, Alanne et al. 2005, Verbeeck
and Hens 2005, Jakob 2006, Tommerup and Svendsen 2006,…). However, as the literature
on the rebound effect already shows, the evaluation of the potential of energy saving
measures cannot just rely on a purely engineering-economics approach. The every day
57
CHAPTER 3
practice of Flemish building owners and occupants reveals a lack of interest and actions in
energy conservation (Enquête 1998, 2001, 2003, 2005) and non-compliance with existing
regulations (SENVIVV 1998). So, there are other conditions and driving forces beside cost
minimisation that determine the real up-take and implementation of energy efficient
technologies and practices by the building users.
Although the process of decision making, final choice making and effective implementation of
energy saving measures is in fact beyond the scope of this research, this section has the
intention to frame the research more broadly in order to show that what is done here can only
be considered as a first step, i.e. a deduction of important boundary conditions for the
development of (extremely) low energy dwellings on an engineering basis. In order to
communicate these results to individual building users and implement them in an energy
policy in the most effective way, this assessment should be enlarged by a coupling with
investigations of behavioural aspects of the uptake of energy efficiency measures. This way,
this section tends to give an onset to that coupling by first presenting some facts and figures
on energy related behaviour of Flemish households, followed by some of the –sometimes
conflicting- theories and approaches found in the literature on this subject.
3.4.2. Facts and figures on energy related behaviour in Flanders
Several studies on the knowledge, attitude, actions and intentions of Flemish households
concerning their own energy consumption reveal poor results when focusing on the
implementation of knowledge and attitude into real energy saving behaviour and actions.
The SENVIVV study (1998) analysed the compliance of new built dwellings with the
regulation on thermal insulation and ventilation, introduced in 1992. The analysis of a sample
of 200 dwellings built after September 1993 revealed that the average insulation level was
K70 (Umean ≈ 0.70 W/m²K) or higher, thus not complying at all with the legal upper limit K55
(Umean ≈ 0.55 W/m²K). One of the conclusions was that the number of buildings with a better
insulation quality increases when more severe requirements are imposed, but that on the
average, no improvement can be notified.
Energy surveys in 1998, 2001, 2003 and 2005, commissioned by the Flemish government,
were executed with a varying sample of 1000 Flemish households in order to sound them
about their attitude, knowledge, actions and intentions related to their own energy
consumption and the energy policy of the Flemish government (Enquête 1998, 2001, 2003,
2005). Although 93% of the surveyed Flemish people describe energy conservation as
‘important’ to ‘very important’, only 60% perceives themselves as economically acting with
regard to energy. Mentioned reasons to act economically are mainly of financial kind,
whereas the most important reasons for not acting economically are laziness and lack of
58
COST ASSESSMENT
consciousness (‘I hardly bear it in mind’). Other mentioned reasons are fear of loosing
comfort, high investment costs, lack of information, limitedness of energy costs, disbelief of
the effective impact of the measures... The subsequent surveys show that the knowledge
about energy saving measures and the Flemish energy policy has been increasing slightly
over the years. However, when investigating the actual behaviour and actions, the surveys
also reveal that the increased knowledge and positive attitude rather rarely are turned into
real economical behaviour and energy saving actions. Furthermore, the energy consumption
of the households still increases, mainly dedicated by the households to an increase of the
number of electrical appliances per household. The Flemish building occupant/owner
declares himself willing to reduce his energy consumption, as long as it does not demand too
much effort of any kind (time, money, paperwork …). Asked for their intentions related to
energy saving measures in the near future, the purchase of energy efficient appliances, low
energy light bulbs and low flow showerheads appears to be the most popular measures.
Investments in building-related energy saving measures are less popular. These facts and
figures show that, despite the promotional campaigns of the Flemish government, the
regulations on insulation level and since January 2006, on the overall energy performance of
buildings, and despite the financial incentives through subsidies and tax reduction, energy
conservation still is not a real issue for the majority of the Flemish households.
3.4.3. Decision making models
The failure of the current energy policies, not only in Belgium, but also in other countries
(Hinchy et al. 1991, Clinch and Healey 2000, Dewick and Miozzo 2002, van Rooijen and van
Wees 2006,) inspires scientists from a wide range of research domains to investigate and
declare the reasons for this failure and to propose ways to improve the efficiency of energy
policy measures. Most scientists start from the conclusion that traditionally, economic
behaviour is represented by a notion of decision making based on neoclassic economic
theory. From this perspective, energy related behaviour is perceived as an economic
optimisation (mostly cost minimisation) undertaken by the economic agent (Ramesohl 1999).
This optimisation takes place in a market environment, and is considered being a function of
input variables which are exogenously given, such as energy prices, technology costs, legal
constraints, etc. Most scientists realise and accept the limits of this neoclassic theory, but the
ways to adapt this theory differ quite strongly.
A first direction of thought dedicates the deviation from real life economic behaviour to
reasons of market barriers and imperfections, such as distorted energy prices, incomplete
information, legal dis-incentives, etc. (Rousseau and Proost 2004). Others argue that the
incomplete exploitation of apparently cost effective energy saving opportunities can be
59
CHAPTER 3
explained by hidden costs and risk factors that are not sufficiently considered. Ramesohl
(1999) gives an overview of papers that subscribe to these viewpoints. A theory that is linked
to this direction of thought is the modern economic theory of investment under uncertainty, to
which Dixit and Pindyck (1994) are major contributors (Hubbard 1994). They argue that the
failures of the neoclassic economic theory can be explained by the fact that it neglects three
important characteristics of investment decisions: the partial or complete irreversibility of
most investments, the uncertainty over the future benefits from the investment and the
leeway the decision maker has for timing his investment. The interaction of these three
characteristics has important implications that are not taken into account in the traditional
theory of investment. Traditional courses in corporate finance and capital budgeting strongly
rely on the basic principle of positive net present value as measure for a good investment.
However, a firm or a person with an opportunity to invest, might hold an option, in analogy
with the options on the financial market, and wait for better information. When a person
decides for an irreversible investment, he kills his option and gives up the possibility of
waiting for new information that might influence the desirability or timing of the investment.
This way, this lost option value is an opportunity cost that should be included in the cost of
the investment (Dixit and Pindyck 1994). This opportunity cost can be large and is highly
sensitive to the uncertainty of the future outcome of the investment. Applied to energy saving
investments in buildings, the characteristics of irreversibility of the investment, uncertainty
over future benefits and ability to delay the decision are certainly valid and might explain the
hesitation of building owners to invest in energy efficiency, even when solutions with positive
net present value are presented. In case of renovation, the ability of delaying the investment
decision will have an even larger impact than for new buildings.
A second direction of thought is developed within the domain of behavioural economics.
They attempt to increase the explanatory power of economics by providing it with
psychological foundations. Camerer and Loewenstein (2002) give an overview of past,
present and future directions within behavioural economics. The discussed papers do not
reject the neoclassical theory of utility maximisation, equilibrium and efficiency, but modify
some of the assumptions in the standard theory in order to result in greater psychological
realism. Examples are loss-aversion, framing, fairness and willingness to pay (Kahneman
and Tversky 2000):
-
Loss-aversion is the disparity between the strong aversion to losses relative to a
reference and the weaker desire to gains of equivalent magnitude. Or with other
words, a loss of X€ is more aversive than a gain of X€ is attractive. Loss-aversion is
considered more realistic than the standard continuous, concave, utility function over
wealth. It is related to the typical behaviour of risk seeking when considering losses
60
COST ASSESSMENT
(rejecting a sure loss in favour of a gamble with higher or equal expected losses) and
risk aversion when considering gains (preferring a sure gain over a gamble with
higher or equal expected gains). Figure 3.2 presents a qualitative view of both the
standard value function (dashed line) and the value function representing behaviour
of loss-aversion, risk seeking in losses and risk aversion in gains (straight line).
value
losses
gains
Figure 3.2 Risk aversion in gains (concave curve), risk seeking in losses (convex
curve) and loss aversion (curve is steeper for losses than for gains) (Kahneman and
Tversky 2000)
Application of these behavioural patterns on energy savings in buildings could partly
explain why building users are difficult to persuade for investments in energy saving
measures, even if they are cost-effective on the long term. The investment itself can
be considered as a high loss with a certainty of 100%, resulting in very small gains of
which the magnitude remains uncertain as the annual effective savings on energy
cost will highly depend on the real effectiveness of the measures and on the evolution
of the energy prices. From the point of view of loss aversion, risk aversion in gains
and risk seeking in losses, this situation is less attractive than the case where no
energy saving investments are made. In the latter, the building users prefer to keep
their money (rejecting a sure high loss) and pay higher annual energy costs, since
these energy costs are perceived as only small losses compared to the investment
and as costs distributed over the years. Furthermore, building owners probably prefer
to gamble that the energy prices will not increase as already predicted for years by
‘environmental pessimists’, thus gambling for lower future losses than the high certain
loss related to the investment. Besides, the money that is not invested in energy
saving measures can be invested in risk-free investments with a fixed rate of return,
thus resulting in a sure, but maybe lower gain than the expected gains from energy
savings.
61
CHAPTER 3
-
Framing intends to counter the principle of invariance in rational choice. Invariance
requires that two versions of a choice problem that are recognised as equivalent
when shown together, should elicit the same preference even when shown
separately. Kahneman and Tversky (2000), however, state that the preferential
outcome of a choice problem may depend on the framing or the description of the
problem. An outcome can e.g. be described as a gain or loss relative to a reference
or as absolute asset incorporating the initial wealth. Applied to energy saving
investments in buildings, it can be expected that presenting the cost-effectiveness of
an investment as a maximised net present value (solution with the highest gains
compared to the losses) will be more convincing than presenting the same solution as
a solution with minimised total present value (both the initial situation and the
improved situation are described only in terms of costs or losses, not in terms of
gains).
-
Other concepts considered to evaluate behavioural response on environmental policy
measures are the concepts of fairness and willingness to pay. Without going too
much into detail, both fairness and willingness to pay are described within
behavioural economics as concepts that tend to influence the acceptability of
economic outcomes of a policy, especially when the economic actor experiences
uncertainty due to insufficient knowledge of some aspects surrounding the policy
(Jorgensen et al. 2006). Psychological and sociological studies mainly use interviews
and diaries to evaluate fairness and willingness to pay in naturally occurring situations
within
households,
whereas
economic
studies
frequently
use
manipulated
experimental situations or constructed scenarios (Banfi et al. 2006). Both directions
within behavioural economics, however, tend to conclude that the judgment of
fairness and the willingness to pay may strongly depend on the framing of a choice
problem, whether the choice is valuated as a loss or as a gain (Kahnemann and
Tversky 2000, Antonides and Kroft 2005).
Finally, a third direction of thought is developed within the domain of sociology and mainly
focuses on concepts such as learning processes and motivational strategies to enhance the
energy-consciousness and to increase the adoption of energy-saving technology. A study on
the determinants for energy consumption of households (STEM 2004), commissioned by the
Flemish government, tried to improve the insights in the driving forces of households
concerning their energy-related behaviour. Through the combination of a theoretical
behavioural model, in depth discussions with five different socio-cultural and sociodemographic focus groups and panel discussions with experts, the study tried to reveal the
underlying driving forces for energy-related behaviour. Aim of the study was to provide the
62
COST ASSESSMENT
policy makers with well-founded guidelines in order to realise more effective energy-related
campaigns, policy measures and energy services. Most of the proposals from this study are
situated in the domain of cognitive-motivational strategies, such as incorporation of energyconscious behaviour in television programmes, information adapted to the individual, use of
social networks for the promotion of energy-conscious behaviour, etc. aiming at a policy
approach that is differentiated for the different socio-cultural and socio-demographic focus
groups. Other examples are found in the literature. Darby (2006) presents the findings of a
survey of residents of an English village that had won an energy-conscious village
competition to illustrate how individual and social learning can contribute to increase
awareness and actions related to energy conservation. Vermeulen and Hovens (2006)
developed a conceptual framework that integrates different partial explanations for the
diffusion of energy-saving innovations in industry and the built environment. One of their
conclusions was that for the Dutch context, the adoption of mature innovations is mainly
based on routine procedures, whereas the adoption of young innovations is more based on
project-specific considerations. These differences should be taken into account when
designing energy policy measures.
3.5
Conclusions
Cost-assessment is an essential part of the evaluation of energy saving measures in
buildings. Within the micro-economic approach, adopted in this research, a cost-benefit
analysis is the most appropriate economic evaluation tool. However, care must be taken
when interpreting the results from such cost-benefit analysis, as phenomena, such as the
rebound effect, may strongly influence the effective cost savings. Furthermore, the building
concepts and boundary conditions determined in this research can only be considered as a
first step. For the implementation of these results in energy policy and the adoption of the
concepts in real life, further research needs to be done, as the adopted economic evaluation
tool cannot be considered as the ultimate determinant in the decision-making process, but
only as rough model for the complex real life behaviour of consumers.
63
MODEL FOR OPTIMISATION
PART TWO: GLOBAL METHODOLOGY
CHAPTER 4.
4.1
Model for optimisation
Introduction
The main goal of this research is to create a global methodology for developing extremely
low energy and pollution dwellings. The methodology consists of three main pillars, each of
them presented in the following chapters.
The first pillar is represented by the optimisation strategy. As discussed in chapter 1, multiobjective optimisation with genetic algorithms has mainly been applied as the tool for
optimisation of extremely low energy buildings, rather than being the main subject of the
research. The developed optimisation strategy has schematically been presented at the end
of chapter 1 and will be explained in detail in this chapter.
The two other pillars consist of the development of a life cycle inventory model and of a cost
evaluation model data for buildings as a whole. The life cycle inventory model is presented in
chapter 5, the cost evaluation model in chapter 6. The models for energy simulation of
buildings that are adopted within this research depend on the application and are presented
in part three.
4.2
Genetic algorithm and coupling with Pareto concept
4.2.1. Object and parameters for optimisation
The objects for optimisation are residential buildings. As the methodology is developed for
the Belgian building practice, representative reference dwellings are designed following the
statistical average of the Belgian residential sector (Verbeeck and Hens 2002). For these
dwellings, the geometry is fixed for the non insulated version. This way, the configuration and
dimensions of the rooms inside are not objects for optimisation. Only the glass area per room
can vary in order to include the impact on the net heat demand and on summer comfort.
Depending on the application of the methodology, different reference buildings are designed.
Description of the buildings can be found in chapters 7 and 8 where the application of the
methodology in two different projects is presented.
The parameters for optimisation are related to the energy saving measures. For all projects,
these measures are applied to both the building envelope and the heating system, but the
optimisation is performed in two steps. In the first step, only envelope-related energy saving
measures are considered, such as better insulation, better glazing, glass area, sun shading,
air tightness and natural ventilation scenarios. In the second step, installation related
65
CHAPTER 4
variables are optimised in order to meet the needs of the optimised building concepts from
the first step. The variables for the second phase concern space heating systems, ventilation
systems, systems for domestic hot water and renewable energy systems.
The methodology could be adapted easily to optimise all energy saving measures
simultaneously, but this option is not applied for several reasons. Firstly, the computational
time for energy simulations strongly differs between simulations of the building envelope and
simulation of installations systems, especially when dynamic building simulations are
performed. A one-hour time step is a common time step for calculating the annual net heat
demand of a building, resulting in short computational time (< 1 minute per simulation run).
However, to incorporate the dynamic response of installation systems, much smaller time
steps are needed, 5 minutes or less, resulting in large computational time for calculations of
the total annual energy consumption (> 5 hours per simulation run). Secondly, the
dimensions of the installation depend on the thermal quality of the building and should be
determined separately for each building variant. This is very time-consuming and not feasible
to integrate in the optimisation process. However, the most important reason for splitting the
optimisation in two steps lies in the difference in life span between the building envelope and
the installation components. After all, the building envelope can be considered as the
hardware of the building, due to its long life span in comparison with most components of the
heating system. The heating system then corresponds to the midware of the building.
Because of the long lasting impact of envelope-related decisions on the energy performance
and on the dimensions of the heating system, it is preferable to first optimise the building
envelope, i.e. minimise the net heat demand in a cost-effective and ecologically optimal way.
Subsequently, the most appropriate heating systems for this low net heat demand can be
identified. This way, the logical hierarchy of energy saving measures that has already been
deduced in earlier work, is respected (Verbeeck and Hens 2005). This hierarchy is presented
and discussed in detail in chapters 7 and 8.
4.2.2. Representation and boundary conditions
For the optimisation process, the standard technique of genetic algorithms is adopted as a
starting point. A problem-specific representation is established that suits as input for both the
genetic process and the calculation of energy consumption, life cycle inventory and costs.
Each potential building design is represented unequivocally as a set of parameters, joined
together in a string of values. This string characterises the chromosome, in which each value
represents one variable. In combination with the fixed geometry per reference building, the
chromosome defines unequivocally and completely the building design. For the first
optimisation step, the chromosome contains minimum 14 genes and maximum
66
(18+2*number of windows) genes, depending whether flat and/or sloped roofs and/or attic
roofs are present and whether the glass area varies. The chromosome of a terraced house
with partly a flat roof and partly a sloped roof, with an unheated attic and 12 windows
contains 42 genes and may appear as follows:
number
1
2
Sloped
roof
pitches
3
4
value
17
0
19
Flat
roof
Element
Window
area 1
17
18
89365
Window
area 2
19
20
0
35287
3
…
…
1
…
Attic
roof
Façade
Window
characteristics
Floor
5
6
7
8
9 10 11 12 13 14 15 16
18
0
25
1
1
Window
area 11
37
38
20276
8
2
Window
area 12
39
40
0
60379
65 11
3
2
89
Air
tightness
41
Summer
ventilation
42
2
1
0
For roofs, (attic) floors and façade, the first gene indicates the insulation thickness in cm, the
second gene the insulation material, the third gene for the façade reflects the constructional
type, not only of the façade (cavity wall, outer insulation or wood frame façade), but also of
the whole building (massive inner walls and floors or complete wood frame construction).
For the window characteristics, the first gene reflects the glass type, the second the frame
type, the third the type of glass spacer, the fourth the type of sun shading and the fifth the
degree of opacity of the sun shading. In case the glass area varies, the first gene per window
area indicates the amount of variation in cm² and the second gene whether the variation is
an increase or decrease of the glass area. All gene values can vary randomly between
predetermined minimum and maximum values. For the elements of insulation material, glass
type, frame type, type of spacer, type of sun shading, air tightness and ventilation scenario,
the element is related to a list of possible choices and each value corresponds to one
particular choice.
Similarly, the chromosome for the installation is designed to define the installation systems
unequivocally. For the installation, a chromosome with 10 genes is established:
Heating
Energy
Pro-
Element
carrier
duction
number
1
value
2
Solar collector
PV system
Emis-
Venti-
Hot
Surface
Surface
Control
sion
lation
water
Present
[m²]
Present
[m²]
2
3
4
5
6
7
8
9
10
1
1
3
2
3
0
6
1
10
For most genes of the installation chromosome, the gene value represents one particular
choice from a predefined list. The provided choices in the different lists may vary from project
67
CHAPTER 4
to project: e.g. in the project presented in chapter 7, central and local heating systems are
considered, whereas in the project presented in chapter 8, only central heating systems are
available in the predefined list.
4.2.3. Cost functions, fitness functions and penalty functions
Cost functions
The fitness of a building solution is evaluated through its performance for an energy criterion,
an economic criterion and an ecological criterion and for the constraints. Depending on the
application or the application step, different evaluation criteria may serve as cost function.
The energy criterion is always calculated with a building simulation programme, steady state
(EPB) or dynamic (TRNSYS). The net heat demand is selected as energy criterion for cases
where no assumptions are made yet on the installation system (this is the case in the first
optimisation step of the project of chapter 8). For other cases, the overall primary energy
consumption over the utilisation phase of the building and its installations is set as energy
criterion in order to compare systems with different energy carrier on an equivalent basis.
The economic criterion is a financial cost, calculated with a cost database, based on the
amount of material used. Secondary costs for foundations or windows due to e.g. large
insulation thicknesses can be incorporated by adapting the amount of material used. Result
may be the initial investment cost, the total or net present value over the utilisation phase of
the building, the static payback time,…
The ecological criterion relies on a life cycle inventory of the building and is calculated with a
life cycle inventory database, based on the amount of material used. It is possible to
calculate the energy content, non renewable embodied energy, global warming potential or
emissions, such as CO2, SO2, NOx, etc. for the building as a whole or for the extra energy
saving measures compared to a reference.
The constraints incorporated in the optimisation process concern the insulation level and the
summer comfort:
-
the maximum insulation level is set at K45 (Umean = 0.45 W/m²K), the legal insulation
standard in Flanders since January 1st 2006 (EPB Besluit 2005)
-
the limit of summer comfort is set at 130 weighted temperature exceeding hours
(WTE-hours) per room, calculated according to the Dutch method of weighted
temperature exceeding hours to assess summer comfort in office buildings
(ISSO/SBR 1994). Applied to dwellings, this method assumes a comfort temperature
of ca. 27°C in summer and assesses summer comfort by weighting the exceeding of
this comfort temperature per hour with a weighting factor that is proportional to the
exceeding. If the limit of 130 WTE-hours is respected per room, it can be expected
68
that the comfort level in each room remains within the zone of acceptable comfort,
according to the comfort theory of Fanger (1972).
Constraints, such as indoor air quality, winter comfort and visual comfort, are integrated
independently from the optimisation, by always designing the ventilation system according to
the ruling ventilation standard NBN D50-001 (1991), by imposing temperature profiles per
room to be satisfied and by respecting a minimum glass area per room proportional to the
floor area. Constraints for maximum U-values or inhomogeneity of the insulation level are not
incorporated in the optimisation process, but controlled after the optimisation process in
order to limit the complexity of the penalty functions (see next section).
Fitness functions and penalty functions
Before running the programme, the particular energy, economic and ecological criteria need
to be specified. In a first evaluation step, the criteria are calculated for all chromosomes of
the population, together with the constraints. Both the results for the criteria and the
constraints serve as input for the fitness function to calculate the ranking of the
chromosomes in the population. Since the quality of a solution should be reflected by its
ranking, fitness assignment is applied through ranking based on a Pareto-score. It is inspired
by the approaches presented in the literature, but with the attempt to find a good balance
between the performance of the genetic algorithm and the complexity of the fitness
assignment process. This way, a simple Pareto-based fitness ranking scheme is applied
without fitness sharing or mating restriction, as this scheme performed well for the particular
problem. Constraints, such as summer comfort and the insulation standard, are handled
through a penalty function. Different penalty functions have been tested on standard GA test
functions (Gens 2001) in order to identify answers to the following questions:
-
Concerning the Pareto score: Should the population be subdivided in two
subpopulations (meeting ↔ not meeting the constraints) before calculating the Pareto
score?
-
Concerning the penalty: Should the penalty be proportional to the exceeding of the
maximum value and should there be a limit to the magnitude of the penalty?
The empirical tests established the following conclusions, in good agreement with results
from the literature (Richardson et al. 1989, Siedlecki and Sklanski 1989):
-
Dividing the population in subpopulations of ‘good’ (meeting the constraints) and ‘bad’
(not meeting the constraints) chromosomes is preferable, else
69
CHAPTER 4
o
there is a risk that only ‘bad’ chromosomes have the best ranking before
penalisation and in this case, no non dominated solutions will be produced
after penalisation
o
after penalisation, there risks to be no distinction between ‘good’ and ‘bad’
chromosomes, while the constraints are important conditions to be satisfied.
-
The severity of the penalty should be proportional to the distance to feasibility in order
to create a final ranking that reflects in the best way the quality of the solution.
-
The magnitude of the penalty should be in proportion with the range of Pareto scores
in order to avoid that after recomposing the two subpopulations, the range of ranking
scores is too extended and causes disturbance of the stochastic selection process.
So the following final score is established:
Final score = Pareto score (per subpopulation) + penalty_c1 + penalty_c2 [4.1]
The fitness assignment occurs by the following algorithm:
1. Firstly, the population is divided in two subpopulations: one with chromosomes that
satisfy the conditions set by the constraints and one with chromosomes who do not.
2. For each solution in the subpopulations, the Pareto score is calculated for the 3
objectives (energy, costs and ecology) and equals the number of variants in the
subpopulation that dominate the solution. Non-dominated solutions have score 0.
3. In the subpopulation of ‘bad’ chromosomes, the violation of the constraints is
incorporated through a penalty function. For each constraint that is violated, a penalty
is added, proportional to the distance to feasibility for the constraint:
-
For the insulation level, the penalty equals the violation, i.e. the difference
between the real insulation level and the maximum allowed insulation level K45.
As the insulation level of non insulated buildings lies around K150-K200 and the
maximum Pareto score equals the size of the subpopulation, the penalty has the
same order of magnitude as the Pareto score and no further adaptation is
necessary.
penalty_c1 = max [0,(Kreal – 45)]
-
[4.2]
The constraint for summer comfort is treated in the same way, but with adaptation
of the order of magnitude, since very high numbers of WTE-hours can be
established in case of unacceptable summer comfort. As the temperature
exceeding is weighted proportional to its exceeding, results up to 10,000 WTE-
70
hours and more are possible. Therefore, the difference between the real number
of WTE-hours and the limit of 130 WTE-hours is divided by 100 to achieve a
penalty with the same order of magnitude as penalty_c1 and as the Pareto score.
penalty_c2 = max [0,(WTEreal – 130)/100]
[4.3]
4. In a last step the subpopulations are recomposed into one population and fitness
values are assigned to the entire population by linear ranking of the final scores,
achieved through equation [4.1]. These fitness values ensure the basis for the genetic
process of selection, recombination, mutation and reinsertion.
4.2.4. Genetic operators
In order to compose the genetic algorithm, several genetic operators have to be defined,
such as population size, selection method, recombination type, etc. The parameters to be
determined for some of these operators are analysed through a parameter study in order to
define the most appropriate for the problem: population size, generation gap, crossrate,
crossover type, mutation rate, ….
The selection method applied in the methodology is the stochastic universal sampling
method with the possibility for a generation gap. The most optimal percentage for the
generation gap is derived from the parameter analysis. For the crossover operator, the
single-point crossover is selected. For mutation, the discrete mutation operator of the GA
Toolbox is applied (GA Toolbox 1994). This includes the definition of BaseVec, a vector that
defines the basis for the individual elements of the chromosome in order to avoid the creation
of illegal chromosomes after mutation. Since it is infeasible due to the nature of the
optimisation problem to define an error margin, a maximum number of generations is
established as stop criterion. Details of the parameter analysis are presented in section 4.3
on validation.
4.2.5. Outline of the programme
The programme for the global methodology, including the genetic algorithm programme is
written in MATLAB. Advantage of MATLAB is the easy coupling with other programmes,
such as building simulation programmes or spreadsheet programmes with databases for
costs and life cycle inventory. The genetic algorithm programme relies on the ‘Genetic
Algorithm Toolbox for MATLAB ®’. This GA toolbox, containing basic operators for genetic
algorithms has been developed at the Department of Automatic Control and Systems
71
CHAPTER 4
Engineering of The University of Sheffield, UK. The toolbox has been written with the support
of a UK SERC grant and the final version (v1.2) has been completed in 1994. The GA
Toolbox is freely available for download (GA Toolbox 1994).
Figure 4.1 shows the scheme of the programme for the overall methodology
Figure 4.1: Outline of the programme for the global methodology
72
4.3
Validation and evaluation of the optimisation programme
4.3.1. Validation with steady state energy simulation programme (EPB)
Results for validation
In earlier work on energy saving renovation of residential buildings (Verbeeck and Hens
2002), all possible combinations of energy saving measures with insulation and glazing have
been calculated for 5 different dwellings (around 4000 combinations per dwelling).
Calculations have been executed with the steady state Energy Performance for Buildings
(EPB) programme and the results have then been sorted in EXCEL to obtain the most
optimal solution per dwelling. To validate the developed GA-optimisation methodology, the
methodology has been applied to the same 5 dwellings considering the same energy saving
measures and using the building simulation programme, cost and life cycle inventory
database from that earlier work. The advantage of validating with these results is the short
calculation time of the EPB programme (1 sec per building variant) and the fact that the
optimal solution to be found per dwelling is known a priori. The validation has shown that the
developed GA-optimisation programme leads for each of the 5 dwellings to the same
optimum as found in the earlier work.
Parameter analysis
For one of the 5 dwellings, a parameter analysis has been executed with the total present
value over 30 years as the cost function. Several parameters have been analysed stepwise.
When considering a subsequent parameter, the value for the previous parameter has been
selected based on the results of the parameter analysis. The following parameters have
been studied with a preliminary version of the methodology:
-
Population size: 20, 50, 80 and 100 individuals
-
Generation gap: 0%, 5%,10%, 15% and 20% (in case of a generation gap of x%,
the x% best solutions are transferred to the next generation without alterations)
-
Crossover rate: 90%, 85%, 80% and 70% (chance for recombination after
selection of 2 parents)
-
Crossover type: single-point (xovsp: parent-chromosomes are cut at 1 point for
crossover), double-point (xovdp: parent-chromosomes are cut at 2 points for
crossover)
-
Mutation rate: 10%, 15% and 20% (chance for mutation of an offspring after
recombination)
The stop criterion is set at a maximum of 50 generations. For each parameter variation, 10
runs of the programme have been performed, recording each time the result for the optimum
73
CHAPTER 4
and the number of generations necessary to achieve the optimum. Based on the 10 runs, an
average run and a best run have been derived, as well as the percentage of runs that
resulted in the correct optimum and the average number of generations needed to achieve
the best result. Table 4.1 presents the results of the parameter analysis and the settings for
the distinguished parameters.
Due to the simplicity of the optimisation problem, the optimum (cost of 26,540.90€) has been
found in all 10 runs. With a population of at least 80 individuals and a generation gap of at
least 5%, the optimum was achieved in 100% of all runs. In a next step, the number of
generations needed to achieve the optimum has been analysed as well as the number of
calculation times. The differences are very small, but the best or fastest result has been
achieved with a population size of 80, double-point crossover with a cross rate of 70%, a
mutation rate of 10% and a generation gap of 20%.
The validation and parameter analysis has been performed with a preliminary version of the
optimisation methodology that differs on several aspects from the final version. The first
difference lies in the building simulation programme that is used. In the preliminary version,
the steady state EPB programme has been used, whereas in the final version, the dynamic
building simulation programme TRNSYS has been applied that allows a more detailed
simulation of installation components. Also the databases for costs and life cycle inventory
are more elaborated in the final version than in the preliminary one. As in the final version
more parameters and variables are taken into account than in the preliminary one, the
number of genes in a chromosome is higher: from 14 genes in the preliminary to 42 genes in
the final version. The most important difference, however, is the number of objectives to be
optimised. The preliminary version is a single-objective optimisation methodology, as it
considers only one objective to be optimised, here the total present value. The final version,
on the contrary, is a multi-objective optimisation methodology, as it optimises simultaneously
multiple objectives by using the concept of Pareto ranking. However, the basic structure of
the chromosome and of the global optimisation programme has not been modified, when
evolving from the preliminary to the final version. Therefore, most decisions on the genetic
operators in the final version rely on the above parameter analysis. The performance of the
concept of Pareto ranking is evaluated separately through an analysis of the evolution of the
Pareto front based on TRNSYS calculations. Results are discussed in the next section.
74
Parameter analysis
Value
Value
Value
% runs
# generations
# calculation
(cost function = total
parameter
average
best run
that
needed to
times
achieve
achieve
minimum
minimum
present value)
Population size
run
20
26,569.3
26,540.9
50%
33.0
19500
50
26,549.0
26,540.9
90%
30.0
48750
80
26,540.9
26,540.9
100%
21.5
78000
100
26,540.9
26,540.9
100%
20.6
97500
0%
26,578.3
26,540.9
20%
5%
26,540.9
26,540.9
100%
21.5
78000
10%
26,540.9
26,540.9
100%
18.4
76000
15%
26,540.9
26,540.9
100%
18.4
74000
20%
26,540.9
26,540.9
100%
19.6
72000
90%
26,540.9
26,540.9
100%
85%
26,540.9
26,540.9
100%
20.1
72000
80%
26,540.9
26,540.9
100%
20.0
72000
70%
26,540.9
26,540.9
100%
20.4
72000
26,540.9
26,540.9
100%
22.7
72000
26,540.9
26,540.9
100%
18.7
72000
10%
26,540.9
26,540.9
100%
16.6
74000
15%
26,540.9
26,540.9
100%
23.5
74000
20%
26,540.9
26,540.9
100%
27.2
74000
Ggap = 5%
crossrate = 90%
(xovsp)
mut = 10%
# gen = 50
Generation gap
80000
Popsize = 80
crossrate = 90%
(xovsp)
mut = 10%
# gen = 50
Crossrate
72000
Popsize = 80
Ggap = 20%
Mut = 10%
# gen = 50
Crossover type
Same as previous
xovdp
70%
xovdp
80%
Mutation rate
Pop size = 80
Ggap = 15%
Crossrate = 80%
(xovdp)
#gen = 50
Table 4.1: Parameter analysis for the GA optimisation programme
75
CHAPTER 4
4.3.2. Control of the evolution of the Pareto front
During the optimisation process, Pareto optima are determined within each generation and
stored in a Pareto archive. At the end of the process, all optima in the archive are mutually
compared to retain only the globally non dominated solutions. To control the evolution of the
Pareto front, the optimisation process of the semi-detached house with cavity wall of the
EL²EP-project (see chapter 8) has been analysed. The chromosome of this house contains
40 genes, the population size is fixed to 100 chromosomes and an evolution process of 60
generations is assumed. Figure 4.2 presents the number of Pareto optima within each
generation and the number of Pareto optima per generation that stand the final assessment.
Evolving from one generation to another, the genetic process slowly improves the solutions
of the population. This results not only in an increasing number of Pareto optima within one
generation, but more importantly in the development of an increasing number of globally
optimal solutions. This effect is clearly noticeable from the 20th generation on. The
optimisation run finally resulted in some 90 different Pareto optimal solutions for the semidetached house with cavity wall, containing solutions with an insulation level between K17
and K45 (Umean = 0.20 to 0.52 W/m²K) and WTE-hours between 0 and 127. The same
analysis has been executed for the other four buildings from the EL²EP-project with similar
results for the evolution of the Pareto front.
Pareto optima over optimisation process
number of Pareto-optima
35
30
25
20
15
10
5
58
55
52
49
46
43
40
37
34
31
28
25
22
19
16
13
10
7
4
1
0
generations
within 1 generation
over total optimisation process
Figure 4.2: Semi-detached house with cavity wall of the EL²EP-project: evolution of the
Pareto optima over the optimisation process: optima within 1 generation (grey bars) and
optima over the whole optimisation process (black bars)
76
4.3.3. Conclusions on the optimisation programme
Through the optimisation process, a group of Pareto optima is obtained that represents
particular versions of the building concept, optimised for the three predefined objectives
(energy, costs and emissions). As a consequence of the concept of Pareto optima, this
series of optima contains building variants that perform averagely for the 3 objectives, but
also variants that perform well for one objective, but less for the other two. For further
decision making on the different optimal building variants, preferences need to be defined for
the different objectives. Also an in-depth analysis of the final solutions is necessary to filter
possibly infeasible solutions because of illogical combinations of energy saving measures
(e.g. 2 cm roof insulation combined with 30cm façade insulation). Further discussion of the
strengths and weaknesses of the programme is presented in chapter 8, based on the results
of the EL²EP-project.
77
MODEL FOR LIFE CYCLE INVENTORY
CHAPTER 5.
5.1
Model for life cycle inventory
Introduction
As explained earlier, the global methodology is a tool to develop extremely low energy
buildings that are optimised over their life cycle for criteria on energy, cost and ecology.
Therefore, energy and emission data from all life cycle phases need to be incorporated in the
optimisation process. This requires the establishment of a life cycle inventory (LCI) database
and a life cycle inventory model that can be integrated in the optimisation model. This
chapter first outlines the goal and scope of the LCI executed in this research, followed by a
discussion of some aspects of the applied inventory data. Subsequently, the partial inventory
models as well as the overall building inventory model are presented. Finally, this chapter
concludes with a discussion on the results of an uncertainty and sensitivity analysis on the
inventory model.
5.2
Goal and scope of the LCI
5.2.1. Goal of the LCI
Primarily, the goal of the LCI is to establish a LCI database that can be incorporated in the
optimisation process. Therefore, an inventory of energy flows and emissions for all phases in
the life cycle of the building is executed. Furthermore, apart from providing input data, one of
the underlying goals of the LCI is to analyse also the relation between the energy savings
realised with extremely low energy building concepts and the embodied energy needed for
the creation of these concepts. After all, the creation of extremely low energy buildings
demands an extra input of materials and products in comparison with common buildings.
However, the balance between embodied energy and energy savings should always remain
positive over the life cycle of the building. Otherwise, the final goal of developing sustainable
buildings would be overshot.
The life cycle analysis is mainly limited to an inventory of energy flows and emissions. No
impact indicators are calculated, except for the global warming potential. The energy flows
from all phases and the corresponding emissions can be easily summed up to create energyrelated and ecology-related optimisation criteria, such as the total primary energy
consumption and the total global warming potential.
79
CHAPTER 5
5.2.2. Scope of the LCI
Functional unit
As the optimisation process aims at developing building concepts that are globally optimised
and at the same time satisfy the boundary conditions for thermal comfort, visual comfort,
indoor air quality, etc., according to the overall performance matrix of IEA Annex 32
(Hendriks and Hens 2000), the LCI does not focus on materials or building components, but
considers the building as a whole. For the same reason, the functional unit for the LCI is not
limited to the energy saving measures only, but contains the whole building. Each reference
dwelling can be considered as a different functional unit that has to yield a comfortable living
environment in winter and summer for a family of one to four persons. As explained earlier,
the geometry of the dwellings is fixed, except for the glass area. Since for each dwelling, the
non insulated version is defined as the reference situation, it is possible to assess the impact
of the energy saving measures from the overall LCI point of view. The same way, also the
impact of constructional options, such as massive versus wood frame construction, that go
beyond energy saving options, can be analysed through the life cycle analysis. The collection
of inventory data is also very suitable for comparing the impact of the different life cycle
phases on energy consumption and emissions.
System boundaries
As explained in section 2.5, some of the basic assumptions of traditional LCI do not cope
with the characteristics of buildings. Due to the very long lifetime of buildings (80 years and
more), hypotheses on processes of the end phase, such as waste treatment and recycling
procedures, might result in highly uncertain and even unrealistic results. At the same time,
before reaching its end of life, most buildings undergo several refurbishments or renovations
often resulting in thorough modifications of the building. Therefore, this research does not
concentrate on the whole life span of the building, but considers only the impact of one
generation. This results in the scenario that the building is designed and constructed and
then, used and maintained by one generation during 30 to 40 years. No assumptions are
made on the destination of the building after passing to the next generation. The building
may directly be demolished and replaced by a new building, or it may undergo one or more
small or thorough refurbishments before reaching its end of life. Thus, the phases of
extraction, production, transport, utilisation and replacement are taken into account, whereas
the end phase of the building is not considered.
80
5.3
Life cycle inventory data
5.3.1. Comparison of databases
In the search for qualitative and representative input data for the LCI of Belgian buildings,
some existing databases have been compared, such as the freely available databases
GEMIS (Germany GEMIS 4.14 2002) and LISA (Australia LISA 3.0 2000) and the
commercial databases ECOINVENT (Switzerland 2003) and SPINE (Sweden SPINE@CPM
2000). From the commercial databases, all data documentation is freely available, but the
inventory data themselves are protected by a commercial license. Already at first glance,
large differences between the databases appear. Each database is set up following its own
methodology, using its own units, incorporating more or less processes and revealing more
or less details on the evaluation of a certain material or product. Also the number of materials
and in- and outflows strongly differs between the databases. Table 5.1 presents a short
comparison between the largest available databases ECOINVENT (ecoinvent v1.0 version
2003), GEMIS (GEMIS 4.14 version 2002) and LISA (LISA 3.0 version 2000) with relation to
the available data. The difference in degree of detail between the databases is clear.
When further comparing the databases in detail with relation to their data for the same
building materials, also a large variation between the data of the different databases
appears. This might be caused by differences in scope and system boundaries, processes to
be in- or excluded, applied calculation methods, etc. Anyhow, extraction of partial data from
the databases to mutually compare them appeared to be very difficult. Therefore, considering
the scope of this research, the choice has been made to apply only one database in order to
have consistent data for all materials. Since the ecoinvent2000 database is the most
extensive and most complete database at the moment, with representative data for Western
Europe, including Belgium, and is frequently updated, this database has been selected as
the basis for all LCI data within this research.
81
CHAPTER 5
ECOINVENT
GEMIS
LISA
≠ units
x
kWh/kg
x
GJ
x
Details per energy carrier
x
x
Air polluters total
Air polluters detailed
Overall global warming potential
Details per greenhouse gas
kg
x
x
x
x
g/kg
x
x
x
x
Resources
Waste
Land use
Water use
Heat waste
x
x
x
x
x
x
x
x
Extraction
Production
Transport
Waste treatment
x
x
x
x
Energy consumption
Total
Pollution
ton
x
x
x
Others
x
Processes included
Data for building materials
Metals
Plastics
Concrete, mortar, sand, gravel
Limestone, lime, gypsum
Bricks, roof tiles
Mineral wool
Insulation foams
Glass
Wood and derived products
Finishing materials
Data for composed components
Insulation measures
Heat pump, solar collectors, PVcells
Domestic appliances
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Others
Heat production
x
Electricity production net
x
x
Electricity production renewable
x
x
Combined heat and power
x
Table 5.1: Comparison between ecoinvent v1.0, GEMIS 4.14 and LISA 3.0
82
5.3.2. ecoinvent2000 database (Frischknecht and Jungbluth 2003)
General
The ecoinvent database is a large network-based database, consisting of a central database,
local databases, calculation routines, the data exchange format EcoSpold, software for
translation between MS EXCEL and XML and administration and query tools. The central
database contains LCI data on energy systems, transport systems, waste treatment systems,
chemicals, building materials, etc. as well as LCIA methods, such as the Swiss Ecological
Scarcity 1997, Eco-indicator 99 or the CML characterisation scheme 2001. The database is
accessible via the internet. In order to achieve the intended harmonisation of over 2500
datasets in the database, quality guidelines for the investigation of life cycle inventories have
been set up to be followed by the institutes and analysts involved in the ecoinvent project.
The ecoinvent studies follow the LCA method according to the ISO standards on LCA (ISO
14040 – 14048). The focus is on the compilation of life cycle inventories, predominantly for
basic commodities, but the data also contain impact assessment results, according to
already developed LCIA methods. No new methods have been developed, except for the
cumulative energy demand (see 5.3.3) and no particular LCIA method is favoured.
Scope of ecoinvent2000
The selection of products and services to be analysed in ecoinvent mainly relies on the
market and consumption situation in Switzerland in the year 2000. But, since the Swiss
economy is closely linked to the surrounding countries, a lot of processes are also described
for the situation in Europe (RER). For production processes that take place outside Europe,
but with an important role for the European market (mainly extraction of mineral and energy
resources), the reference year 2000 was applied, if ever possible.
The processes included in ecoinvent represent in most cases the average of then used
technology. Emissions from the past (infrastructure construction), the present (e.g. heating)
and the future (e.g. disposal options) are all included in the inventory analysis without
temporal boundaries. Emissions that occur over large time frames of more than 100 years
are assigned to specific subcategories.
Modelling principles in ecoinvent2000
As far as possible, the ecoinvent database contains data on a unit process level that are
neither vertically nor horizontally aggregated. Average data for a country or region are
calculated with the available data from different suppliers if they use comparable processes.
The analyses of technical processes are based on pure environmental process chain
analysis. Economic input-output analyses are only used in exceptional cases.
83
CHAPTER 5
If data availability is poor, stoichiometric balances are used to determine the raw materials
demand. If no information about the amount of a release or the exact substance emitted is
available, an educated guess is made based on plausibility considerations. In cases where
such an assumption dominated the LCA result, further investigations have been carried out
within the ecoinvent project.
Products and services are distinguished on a level of economic regions within which such a
distinction is meaningful. For cement for instance, a national distinction is useful and
meaningful, since cement is hardly traded across national borders. For globally traded
products such as aluminium, a distinction on the level of continental economies is sufficient,
because these commodities can hardly be traced back on a national or regional level.
Transport happens nearly between any two process steps of a product system. They are
investigated for the real market situation as far as possible.
Inputs and outputs required for the means of production and the infrastructure of a
production process are recorded separately, but ecoinvent recommends using always
inventory data including the infrastructure to avoid the use of data sets that are substantially
incomplete.
When a process requires an energy carrier for process or space heating, distinction is made
in ecoinvent between the case that operational emissions and efficiencies are known and the
case that only the amount of energy and the kind of energy carrier is know. In the latter,
generic boiler datasets are used.
Waste treatment is considered in ecoinvent as part of the technical system and therefore
modelled like all other technical processes. If information about the treatment of specific
wastes is not known, generic treatment processes are applied. For uncontaminated building
materials the following default disposal scenarios are considered appropriate:
-
all solid burnable wastes are considered as disposal in municipal solid waste
incineration
-
all bulk metals are considered to be recycled.
Allocation and recycling rules in ecoinvent2000
Multi-output processes are omnipresent in LCA product systems. In ecoinvent, multi-output
unit processes are entered into the database before allocation. Additionally, the allocation
factors applied are defined on the multi-output process level. The database then creates
single output processes by combining both. By-products of a production process that are
used in a subsequent process (re- or down-cycling) are not reported in the list of inputs and
outputs of the first process, but are allocated to the subsequent process.
84
Uncertainty in ecoinvent2000
Within the LCI of a unit process, the amounts of inputs and outputs are described in
ecoinvent with single figures (mean values). This quantitative description includes
uncertainty, because the mean values are uncertain. Only uncertainty due to variability and
stochastic error can be expressed in quantitative terms on the level of individual inputs and
outputs of a unit process. This type of uncertainty has been treated consistently and in a
quantified way within the ecoinvent project, mostly applying a lognormal uncertainty
distribution.
However, quite often the uncertainty cannot be derived from the available information, since
there is only one source of information providing only a mean value, without any information
on the uncertainty of this value. In this case, a simplified standard procedure is applied in
ecoinvent, including a qualitative assessment based on a pedigree matrix in which basic
uncertainty factors are used, based on expert judgments. This is shown in the last column of
table 5.3, where each number in these strings stands for a qualitative uncertainty factor.
Outlook of LCI data in ecoinvent2000
Ecoinvent provides inventory data and impact assessment data per unit process. For the unit
process, e.g., ‘brick, at plant’, the database provides all elementary flows involved in the
production of 1kg brick, taking into account all processes from extraction up to the point that
the brick is ready for distribution from the production plant. The elementary flows are
identified in the database by a flow name, its unit, a category and a subcategory. Categories
describe the different environmental compartments ‘air’, ‘water’, ‘soil’ and ‘resource’.
Subcategories further distinguish sub compartments which may be relevant for the
subsequent impact assessment step. The categories ‘air’, ‘water’ and ‘soil’ describe the
receiving compartment and are used for pollutant emissions, whereas the category ‘resource’
is used for all kinds of resource consumption.
In addition, the database provides per unit process a list of impact indicators, identified by an
indicator name, its unit, a location, a category and a subcategory. The location indicates the
country or region (Europe or global) that will be affected by the impact. Categories describe
the impact assessment method that is applied (IPCC, Eco-indicator 99,…), whereas
subcategories further define the kind of impact (human health, ecosystem quality,
resources,…). Apart from the inventory data, meta-information is provided on the processes
included, the origin of the raw data, comments on assumptions made, etc. Also an
uncertainty assessment is available per unit process, although only for a limited part of the
elementary flows in the life cycle inventory.
Tables 5.2 to 5.4 present the outlook of the inventory data, the impact indicators and the
uncertainty assessment for 1kg brick, at plant. As the data are protected by a license, only
85
CHAPTER 5
fictitious values are given. Due to the extent of the inventory (986 elementary flows, 60
impact indicators and 59 uncertainty values for brick), the lists are only partially presented.
Input
Output
Name
Category
Sub-Category
Unit
Location
InfrastructureProcess
Unit
brick, at
plant
RER
0
kg
resource
in ground
kg
3.00E-11
resource
soil
in air
industrial
kg
kg
8.17E-10
4
Aluminium, 24% in
bauxite, 11% in crude
ore, in ground
Carbon dioxide, in air
Heat waste
2.36E-11
4
Heat waste
soil
unspecified
kg
1.08E-10
4
Carbon dioxide, fossil
air
kg
6.46E-09
kg
1.85E-10
4
4
low population
density
lower
4
air
stratosphere +
upper
troposphere
4
air
unspecified
kg
3.41E-08
4
Ammonium, ion
water
river
kg
6.48E-08
4
Antimony
water
ground-
kg
7.24E-10
Table 5.2: Brick, at plant: partial list of inventory data (fictitious values) (ecoinvent2000)
Name
Location
Unit
Location
Unit
lubricating oil, at
plant
brick, at
plant
Uncertainty
Type
Standard
Deviation
95%
General
Comment
RER
0
kg
(5,5,3,3,1,
5);
(5,5,3,3,1,
5);
(3,5,5,1,3,
5);
(1,2,3,1,1,
RER
0
kg
0.000381
1
2.34
clay, at mine
CH
0
kg
0.00541
1
1.64
mine, clay
CH
1
unit
8.58E-07
1
1.64
limestone,
crushed, for mill
CH
0
kg
0.000542
1
1.89
UCTE
0
kWh
0.0166
1
2.34
(5,5,3,3,1,
5);
RER
0
kg
1.64E-05
1
2.34
(4,2,3,1,1,
5);
electricity,
medium voltage,
production UCTE,
at grid
sheet rolling,
chromium steel
3);
Table 5.3: Brick, at plant: partial list of uncertainty values (fictitious values) (ecoinvent2000)
86
Name
Location
Category
GWP 100a
GLO
IPCC 2001
GWP 20a
GLO
IPCC 2001
GWP 500a
GLO
IPCC 2001
non-renewable energy
resources, fossil
non-renewable energy
resources, nuclear
renewable energy
resources, biomass
renewable energy
resources, wind, solar,
geothermal
renewable energy
resources, water
Total
Total
acidification &
eutrophication
Ecotoxicity
GLO
GLO
GLO
GLO
GLO
RER
RER
RER
RER
cumulative
energy demand
cumulative
energy demand
cumulative
energy demand
cumulative
energy demand
Sub-Category
Unit
Location
Unit
climate
kg
change
CO2-Eq
climate
kg
change
CO2-Eq
climate
kg
change
CO2-Eq
brick, at
plant
RER
0
kg
2.23
2.29
2.21
fossil
MJ-Eq
23.2
nuclear
MJ-Eq
2.15
biomass
MJ-Eq
0.349
wind, solar,
geothermal
MJ-Eq
0.0741
cumulative
energy demand
IMPACT 2002+
water
MJ-Eq
0.333
human health
points
0.000245
IMPACT 2002+
eco-indicator
99, (E,E)
eco-indicator
99, (E,E)
resources
ecosystem
quality
ecosystem
quality
points
0.000171
points
0.00466
points
0.00543
Table 5.4: Brick, at plant: partial list of impact indicators (fictitious values) (ecoinvent2000)
5.3.3. Extraction of data from ecoinvent2000
Collection of LCI data
Not all 2500 process datasets from the ecoinvent database are of interest for this research
that concentrates on buildings. Therefore, only building related process datasets have been
extracted. Annex A presents the database of building materials, building products and
building systems that have been applied within this research and for which LCI data are
gathered. For each of these commodities, the ecoinvent datasets are specified on which the
LCI data of this research are based: 47 datasets have been extracted to calculate the LCI of
54 building related commodities.
The same way, not all elementary flows have been considered, but a selection is made
among the elementary flows to be included in the research database. The elementary flows
have been limited to energy resources, heat waste and emissions of CO2, NOx, SOx, non
methane volatile organic compounds (NMVOC) and particulates. The impact indicators have
been limited to the cumulative energy demand and the global warming potential for a time
frame of 20, 100 and 500 years. The extracted flows are presented in detail in Annex B.
87
CHAPTER 5
Input data for the optimisation process
One of the main goals of the life cycle analysis in this research is firstly, to calculate the
embodied energy of extremely low energy buildings and subsequently, analyse the relation
between the embodied energy and the energy consumption during utilisation phase.
However, when investigating the elementary flows for a certain unit process, ecoinvent does
not present the embodied energy as such. Several energy related input and output flows are
available, but an in-depth analysis was necessary to distinguish them from each other.
Distinction must also be made between renewable and non-renewable energy carriers, as in
fact, only the non renewable embodied energy is of importance in the comparison with the
energy consumption of non renewable energy during utilisation. The following energy related
flows can be identified in ecoinvent2000 (Frischknecht and Jungbluth 2003):
-
Energy resources: non-renewable energy resources like oil and gas are inventoried
with their weight and volume, respectively. Renewable energy resources like wind,
solar and hydro power are recorded with the direct energy input from nature in MJ
that has been used in the process, namely kinetic energy of the air for wind power,
potential energy of the water for hydroelectric power, the radiation energy from the
sun that meets the technical device for solar energy and the upper heating value of
the extracted biomass (wood, crops, etc.).
-
Waste heat: waste heat released from processes is recorded, but no distinction is
made between waste heat emissions from renewable and non renewable sources.
Waste heat is determined as equal to the gross calorific value of fossil energy carriers
and biomass and the energy content of electricity. However, in order not to count the
waste heat of renewable energy sources in the overall waste heat balance, the
energy uptake by these systems is subtracted, taking into account an average
conversion coefficient.
-
Cumulative energy demand: CED is an impact assessment method developed in
the frame of the ecoinvent project. It accounts for the energy resources at the point of
extraction (even if they are not used energetically).
Frischknecht and Jungbluth (2003) classify the cumulative waste heat balance as the best
measure to estimate the actual use of non renewable energy carriers and to obtain a realistic
picture of the energy consumption at the different stages of a life cycle. Therefore, the waste
heat emissions from all categories and subcategories need to be summed up. As said above,
in order to not account for the waste heat from renewable energy resources, such as wind,
hydro or solar plants, the uptake of kinetic, potential or radiation energy by these systems is
subtracted from the total waste heat balance, using average efficiencies. For biomass, the
gross calorific value is subtracted. However, Frischknecht and Jungbluth (2003) admit that ‘it
88
may be said that waste heat emissions are not fully consistently modelled throughout the
ecoinvent data’. A revised approach is still under discussion within the ecoinvent team.
In order to define the total embodied energy of a material or product as the sum of the non
renewable energy, used in all processes upstream to the utilisation phase, the total waste
heat balance has been taken as the starting point to calculate the total embodied non
renewable energy. However, distinction needs to be made between renewable and non
renewable materials.
-
For a non renewable material, such as plastic, the waste heat can be partially
dedicated to energy use of non renewable energy resources and partially to energy
use of renewable energy resources. The non renewable energy resources are mostly
released as waste heat during the extraction and transport of raw materials and the
production of the material. In case of plastics, they also partially establish the energy
content of the material. So, when calculating the waste heat balance by subtracting
the uptake of energy from renewable energy systems from the total released waste
heat, the result directly equals the consumption of non renewable energy resources
for production of the material, including the processes upstream of the production.
Non renewable embodied energy of a non renewable material:
Non renewable energy resources ≡ energy use in processes + energy content of
material
Renewable energy resources ≡ energy use in processes
→ Non renewable embodied energy = total waste heat – renewable energy
-
Also for a renewable material, such as wood, the waste heat is partially dedicated to
energy use of non renewable energy resources and partially to renewable energy
resources. However, part of the renewable energy resource represents the energy
content of the material (biomass). When then calculating the waste heat balance by
subtracting the uptake of energy from renewable energy systems from the total
released waste heat, the waste heat balance will be negative, because the energy
content of the material has not been used until that stage, but is already counted
within the waste heat balance. So, to calculate the embodied non renewable energy
correctly, the energy content of the material has to be added to the total waste heat
balance.
89
CHAPTER 5
Non renewable embodied energy of a renewable material:
Non renewable energy resource ≡ energy use in processes
Renewable energy resources ≡ energy content of material (biomass) + energy use in
processes
→ Non renewable embodied energy = total waste heat – renewable energy +
biomass
The non renewable embodied energy, linked to the processes upstream of the utilisation
phase, is incorporated in the optimisation process in the energy related criterion, by adding it
to the non renewable primary energy consumption of the utilisation phase. For the ecological
criterion, the GWP100 is selected. At the same time, several additional elementary flows are
calculated during the optimisation process. These flows do not interfere with the optimisation,
but serve as extra information for the final decision making process:
-
The total energy content
-
Total waste heat balance for processes upstream of the utilisation phase
-
Total NOx, SOx and NMVOC emissions and particulates < 2.5µm, each of them
summed up for regions with high population density, low population density, lower
stratosphere and upper troposphere (i.e. due to air traffic) and the unspecified part
5.4
Life cycle inventory model for buildings
5.4.1. Material and product models
Several materials in the research database, such as brick, concrete, plywood, rock wool,
etc., correspond directly to materials defined in the ecoinvent database and thus, the
inventory data could directly be imported from ecoinvent into the research database. This
also appears from the list in Annex A that describes the relations between the research
database and ecoinvent. For other commodities, such as hard or soft timberwood, window
frames, sun shading or installation components, a material or product model needed to be
developed and inventory data from different ecoinvent datasets needed to be combined. The
different product models with the assumptions made on the contribution of the different
materials are described in Annex C.
90
5.4.2. Transport model
Assumptions
All materials or products are assumed to be stored at the production plant. For the transport
model of constructional materials, distinction is made between transport from production
plant to distribution or assemblage centre (step 1) and from there to the construction site
(step 2). For the transport model of installation components, step 1 always assumes
transport of the composing materials to the assemblage site, whereas step 2 reflects
transport of the finished goods to the construction site. For each transport step, assumptions
are made per material or product on:
-
transport distance: transport distances are estimated based on assumptions on the
location of production and/or distribution or assemblage site, in Belgium, in a
neighbouring country, elsewhere in Europe or outside Europe. The transport model
only considers one-way distances, assuming the vehicle is always occupied for nonrelated transport on the way back.
-
transport vehicle: ecoinvent contains inventory data for different transport vehicles.
Elementary flows are expressed in MJ/tkm or kg/tkm, being the flow released or
resource consumed when transferring 1 ton of goods over a distance of 1 km. Data
are extracted for transport by:
-
o
a transoceanic freight ship
o
European average goods transport by rail
o
European average lorry of 32 tons
o
European average lorry of 16 tons
o
European average van of < 3.5 tons
transport weight: for each material or product the specific weight is defined in ton
per unit, being m³ (e.g. brick), m² (e.g. sun shading) or m (e.g. window frames).
Based on the assumptions for transport weight, transport distance and type of vehicle, the
overall energy consumption, GWP or emissions due to transport can be calculated as follows
(e.g. for GWP):
GWP _ tot tr ,matX = (GWPtr , ship * Dtr , ship + GWPtr ,train * Dtr ,train + GWPtr ,lorry 32t * Dtr ,lorry 32t
+ GWPtr ,lorry16t * Dtr ,lorry16t + GWPtr ,van * Dtr ,van ) * weight matX / unit
with:
GWP_tottr,matX
GWPtr,…
Dtr,…
Weight matX/unit
[5.1]
total GWP for transport of material X [tons/unit]
GWP for transport per ship, train, lorry or van [tons/tkm]
distance for transport of material X per ship, train, lorry or van
[km]
weight per unit (m³, m² or m) for material X [tons/unit]
91
CHAPTER 5
Assumptions per material or product on transport distances, transport vehicles and transport
weight are presented in detail in Annex D.
Establishing final transport input data
In order to investigate the impact of the assumptions about transport vehicle and distance on
the final LCI results and to incorporate the uncertainties into the final transport model, some
alternative transport scenarios have been developed and corresponding LCI results
calculated:
-
-
Transport from production to distribution:
o
Alternative scenario 1: all transport by rail
o
Alternative scenario 2: all transport by 32t lorry
Transport from distribution to construction site:
o
o
o
Alternative scenario 5: all transport by van
Based on the original assumptions and the alternatives, average transport energy
consumption and average transport emissions are calculated per commodity. These
average values serve as final input data for the transport model in the optimisation process.
Results of the sensitivity analysis on the transport model as well as the impact of the
different phases within the whole life cycle are presented in section 5.5.
5.4.3. Building model
The energy consumption and emissions related to the building are calculated for the different
subsequent phases, as shown in figure 5.1.
92
Figure 5.1: Scheme of the model for life cycle inventory of buildings
A straightforward calculation algorithm has been developed to calculate the life cycle
inventory data for a whole building:
1. Basic matrices are composed, containing data on embodied energy, embodied
emissions, transport energy and transport emissions for all basic constructional
materials and installation related materials:
o
MAT: matrix with the relation between materials from research database and
datasets from ecoinvent
o
ECO: matrix with extracted data sets from ecoinvent with waste heat, energy
content, non renewable energy demand, GWP100a, NOx, SOx, NMVOC and
particulates < 2.5µm per material
o
DISTANCE: matrix with distances and transport vehicle per material from
research database
o
TRANSPORT: matrix with extracted data sets from ecoinvent with waste heat,
energy content, non renewable energy demand, GWP100a, NOx, SOx, NMVOC
and particulates < 2.5µm per transport vehicle
o
WEIGHT: matrix with specific weight per material from research database
o
ECOinst: matrix with extracted data sets from ecoinvent for materials and
processes used in the production of installation components
93
CHAPTER 5
o
MATinst: matrix with the relation between installation components in the
research database and datasets for composing materials and processes from
ecoinvent. For several installation components, such as boilers or heat
pumps, the amount of composing material is a function of the power (e.g.
boiler) or of the power0.666 (e.g. heat pump, based on Berghmans and Duprez
(1997)) as follows:
Component of X kW = amountmatA*X+constmatA+amountmatB*X+constmaB+…
or
Component of Y kW = amountmatC*Y0.666+constmatC +…
Therefore, the MATinst matrix consists of 3 submatrices: 1 submatrix with
data as a function of the power per the installation component [in kW], 1 with
data as a function of the power0.666 [in kW] and 1 with constant data per
installation component.
o
DISTprod: matrix with distances and transport vehicle for each material used
in the production of installation components
o
DISTsite: matrix with distances and transport vehicle per installation
component from the research database
o
TRANSPinst: matrix with extracted data sets from ecoinvent with waste heat,
energy content, non renewable energy demand, GWP100a, NOx, SOx, NMVOC
and particulates < 2.5µm per transport vehicle
o
WEIGHTinst: matrix with the weight used per composing material in the
production of installation components (7 columns) and with the total weight
per installation component (8th column). Similar to the MATinst matrix, also the
WEIGHTinst matrix consists of 3 submatrices with data as a function of the
power, as a function of the power0.666 and with constant data.
2. With the basic matrices, a MATERIAL matrix is calculated containing the global data
for production and transport of each construction material in the research database
MATERIAL = MAT * ECO + DISTANCE * TRANSPORT * WEIGHT
[5.2]
The same way, an INSTALLATION matrix is calculated containing the global data for
production and transport of each installation component in the research database.
Similar to the MATinst matrix and the WEIGHTinst matrix, the INSTALLATION matrix
consists of 3 submatrices. Each INSTALLATION submatrix is calculated as follows:
94
INSTALLATION _ 1 = LCIinst _ 1 + TRANSinst _ 1
with
LCIinst _ 1 = MATinst _ 1 * ECOinst
TRANSinst _ 1 = TRANSPROD _ 1 + TRANSSITE _ 1
with
TRANSPROD _ 1 = WEIGHTinst _ 1(1 : 7) * DISTprod * TRANSinst
[5.3]
TRANSSITE _ 1 = WEIGHTinst _ 1(8) * DISTsite * TRANSinst
3. For each building variant, the applied volume, area or length per constructional
material is calculated and stored in a VOLUME matrix. The installation power is
determined based on the insulation level; lengths of heating pipes and ventilation
pipes and the number of radiators depend on the type of reference building. Different
volume matrices are calculated for different utilisation periods, taking into account
replacements of certain materials and products, based on assumptions of life span for
each material, product or installation component.
4. Based on the MATERIAL and the VOLUME matrix and the information on the
installation, the waste heat, energy content, non renewable energy demand, GWP100a,
NOx, SOx, NMVOC and particulates < 2.5µm, coupled to production and transport,
are calculated for the whole building variant, including both constructional materials
and installation components.
5. For each building variant, the annual net heat demand, annual end energy
consumption, annual primary energy consumption and annual GWP are calculated
with a steady state or dynamic building simulation programme.
6. The primary energy consumption and GWP is calculated for the utilisation phase,
assuming use of the building by one generation during 30 or 40 years. In order to
analyse the impact of the utilisation phase, these values are calculated also for 60 or
90 years. The extra replacements during these time periods are taken into account.
7. In a final step, the non renewable energy consumption as well as the GWP is
summed up for production, transport and utilisation. These values are applied as
energy related criterion and ecological criterion in the optimisation process.
8. Up to now no assumptions have been made on the destination of the building and its
composing materials after passing from one generation to the next.
Despite the straightforwardness of the building model and the inventory algorithm, the
uncertainty is quite high, since all matrices of the building model contain data based on more
or less uncertain assumptions. Therefore, a sensitivity and uncertainty analysis has been
performed on the partial models and on the life cycle building inventory model as a whole.
Methodology and results are discussed in the section below.
95
CHAPTER 5
5.5
Uncertainty and sensitivity analysis
5.5.1. Sensitivity analysis for the transport model
As explained in section 5.4.2, the impact of the assumptions on transport vehicle and
distance is investigated by comparing the results for the original assumptions with those for
some alternative transport scenarios. All three transport aspects (vehicle, distance and
weight) have a well-defined impact. However, when combining them, their mutual interaction
may neutralise or emphasise their combined impact:
-
Transport vehicle: the smaller the vehicle, the larger its environmental impact per
tkm. Long distance transport by 32t lorry consumes 16 times more non renewable
energy and releases 20 times more NMVOCs and particulates than by ship, whereas
short distance transport by van consumes 8 times more energy and releases 8-10
times more NMVOCs and particulates than by 32t lorry. Only for NOx and SOx, the
differences are smaller between the different vehicles.
-
Transport distance: the smaller the distance, the smaller the environmental impact
of the transport (less kilometres)
-
Transport weight: the smaller the material weight, the smaller its environmental
impact during the transport phase (less tons)
Table 5.5 presents the results for the original assumptions and for the alternative scenarios
for some commonly used building materials.
In case of wood, originally most transport was assumed to be intercontinental (large
distances) by ship (step 1). When modifying the transport vehicle to train or 32t lorry, the
environmental impact per tkm increases and this effect is even more emphasised due to the
large distances. Therefore, for wood the average of all scenarios corresponds to an increase
of more than 100% for softwood and more than 200% for hardwood in comparison with the
original assumptions, not only for the non renewable energy consumption (NRE), but also for
the GWP. The increase of NOx emissions is slightly lower (80% for softwood, 125% for
hardwood), whereas the increase of SOx is only 9%. This is caused by differences in energy
carrier between the different transport vehicles.
In case of bricks, transport was originally assumed by 32t lorry to the distribution site (step 1)
and by 16t lorry to the construction site (step 2) and all transport taking place within Belgium.
Modifying the transport mostly decreases the impact per tkm, except for transport by van.
However, due to the small transport distances, the final result is only slightly affected by the
assumptions on transport vehicle. In comparison with the original assumptions, the average
of all scenarios produces an increase of 5-7% for non renewable energy consumption and
96
GWP. The emission of NOx is even lower than in the original scenario. Only the emission of
SOx is for the average of all scenarios 15% higher than originally.
Original
scenario
Alternative
Alternative
scenarios
scenarios
for step 1
for step 2
By
train
By
By
By
lorry
lorry
lorry
32t
32t
16t
Increase
By
Mean
compared
value
to
original
van
Hardwood
NRE
[MJ/m³]
GWP
[kg/m³]
SOx *10-4
[kg/m³]
NOx *10-4
[kg/m³]
1087
3892
13366
1037
1087
1361
3638
+235%
76
227
902
73
76
94
242
+216%
7.8
6.3
13
7.8
7.8
8.2
8.5
+9%
9.5
12
78
9.3
9.5
10
21
+125%
564
313
564
456
564
1150
602
+7%
38
20
38
31
38
75
40
+5%
5.5
3.6
5.5
4.5
5.5
13
6.3
+15%
3.1
1.4
3.1
2.7
3.1
4.8
3.0
-3%
Brick
NRE
[MJ/m³]
GWP
[kg/m³]
SOx *10-5
[kg/m³]
NOx *10-4
[kg/m³]
Table 5.5: Comparison of the original scenario and alternative scenarios for transport vehicle
for hardwood and brick
In general, it can be concluded that for most commodities the average of all scenarios results
in an increase of the environmental impact in comparison with the original assumptions. Only
for materials or products that are assumed to be transported by van to the construction site,
the average value is lower than the original. Because of the very large variations in transport
of building related products, both to the distribution site as to the construction site, the
average values for energy consumption and emissions are assumed to be acceptable for
integration in the optimisation process.
97
CHAPTER 5
5.5.2. Contribution analysis for the building model
According to Heijungs et al. (2005), a contribution analysis decomposes the aggregated
results of an inventory analysis in a way that the contribution of the different unit processes,
materials or phases to the overall inventory results can be traced back. In order to compare
the contribution of the utilisation phase with the contribution of the phases upstream, figure
5.2 presents the embodied energy as a function of the primary energy consumption for
building use over 30 years for the architectural house, one of the buildings optimised in the
EL²EP-project (more details on the dwelling see chapter 8, plans see Annex G). The points
on the right hand side in figure 5.2 represent the non-insulated reference version of this
dwelling, where the clouds of points on the left hand side represent all kinds of more or less
energy saving variants of this dwelling. The reference values clearly show that the building
structure, including the massive part of the building envelope represents the largest part of
the embodied energy (over 400 GJ), whereas the embodied energy of the installations is
negligible. When increasing the energy performance of the building, the embodied energy of
both the building envelope and the installations increases, but in general, more for the
envelope than for the installations. Building variants with a primary energy consumption over
30 years of more than 1000 GJ can be realised with a limited increase of the embodied
energy. Only extremely low energy buildings with a primary energy consumption of ca. 500
GJ have a total embodied energy higher than the energy use of the utilisation phase.
However, the sum of both remains small.
Furthermore, when comparing the embodied energy with the energy savings they realise
(figure 5.3), it is clear that in most cases the embodied energy represents less than 10% of
the primary energy savings over 30 years. The cloud on the right hand side of figure 5.3
represents variants with outer insulation, in the middle with cavity wall and on the left hand
side with wood frame construction. Remarkably, the embodied energy for the three
constructional types is comparable. The primary energy savings strongly differ, but this is
caused by the difference in reference situation (8cm of mineral wool for wood frame
construction versus no insulation for the other construction types).
98
Embodied energy vs energy consumption
Architectural house, utilisation period of 30 years
Embodied energy [GJ]
900
800
total
700
construction+envelope
600
installations
500
reference
400
ref construction+envelope
300
ref installations
200
100
0
0
2000
4000
6000
8000
10000
12000
Primary energy consumption during use phase [GJ]
Figure 5.2: Architectural house of the EL²EP-project, utilisation period of 30 years: embodied
energy vs. primary energy consumption for building use during 30 years for different variants.
The dark grey cloud represents the embodied energy for all kinds of installations, the light
grey cloud the embodied energy for all kinds of energy saving variants of the building
structure and envelope, whereas the black cloud is the sum of both. The dots on the right
hand side are the values for the non insulated reference case with a cavity wall.
Embodied energy vs energy savings
Architectural house, utilisation period of 30 years
1400
Embodied energy [GJ]
1300
cavity wall
1200
outer insulation
1100
wood frame construction
1000
900
800
700
600
500
400
4000
6000
8000
10000
12000
14000
Primary energy savings during utilisation phase [GJ]
Figure 5.3: Architectural house of the EL²EP-project, utilisation period of 30 years: embodied
energy vs. primary energy savings, compared to the non insulated reference during 30 years.
The clouds represent the overall embodied energy per constructional variant, whereas the
dotted line represents the 10% limit. The black cloud are variants with outer insulation, the
dark grey cloud with a cavity wall and the light grey cloud with wood frame construction.
99
CHAPTER 5
These results lead to the important conclusion that considerations on the embodied
energy of extremely low energy houses can be an interesting issue for future
research, but that in the first place, effort should be paid to the reduction of the energy
consumption during the utilisation phase, as this phase still has the largest potential
for improvement.
5.5.3. Perturbation analysis for the building model
According to Heijungs et al. (2005), a perturbation analysis identifies the sensitive
parameters, being the input parameters of which a small change induces a large change in
the selected results. The factor that couples a small change in input to a change in output is
referred to as the multiplier. Multipliers larger than 1 or smaller than -1 indicate sensitive
parameters; whereas multipliers close to 0 indicate insensitive parameters. The advantage of
a perturbation analysis is that it allows studying inherent sensitivities, even for variables for
which no uncertainty indication is known. Perturbations of 1% of each value successively
have been induced in the basic matrices MAT, ECO, DISTANCE, TRANSPORT, WEIGHT,
MATinst, ECOinst, DISTprod, DISTsite, TRANSPinst and WEIGHTinst. Firstly, the impact on
the MATERIAL matrix and on the INSTALLATION matrix has been analysed and
subsequently the impact on the overall life cycle inventory results of a whole building.
Impact on the basic matrices
As could be expected from the linearity of the building model, the perturbation of a single
parameter has generally little impact on the overall result. Only for some materials in the
ECO and WEIGHT matrices, the multiplier is larger than 1, meaning that a perturbation of 1%
of one of the input data creates an impact on the output larger than 1%. Table 5.6 shows the
materials from the ECO matrix for which the multiplier is larger than 1, as well as the
magnitude of the impact and the material and category that are sensitive for perturbation.
However, the impact remains small, as only for two cases the impact is larger than 2% and
the maximum value is 4.4%. Table 5.7 presents the results for perturbation of the WEIGHT
matrix. Also here the impact remains small, with variations less than 1.3% after perturbations
of 1%.
100
ECO matrix: Impact > 1% after perturbation of 1%
ECO
Sawn timber, hardwood, planed,
kiln dried, at plant
Sawn timber, softwood, planed,
Plywood, outdoor use, at plant
Polyurethane, rigid foam, at plant
MATERIAL
Impact
Heat, waste
balance
GWP 100a
Hardwood
1.1%
1.1%
Wooden window frame
Wood-PUR-wood window
frame
1.5%
1.2%
Softwood
Plywood
Wood-PUR-wood window
frame
1.1%
1.2%
4.4%
1.1%
2.2%
Table 5.6: Results from the perturbation analysis of the ECO matrix, each time for an input
perturbation of 1%
WEIGHT matrix: Impact > 1% after perturbation of 1%
Specific
MATERIAL
weight
unit
Hardwood
Softwood
Plywood
Wooden window frame
Wood-PUR-wood window frame
700
700
600
4.97
6.22
kg/m³
kg/m³
kg/m³
kg/m
kg/m
Impact
Heat, waste
balance
GWP 100a
1.1%
1.1%
1.2%
1.1%
1.1%
1.05%
1.05%
1.3%
Table 5.7: Results from the perturbation analysis of the WEIGHT matrix, each time for an
input perturbation of 1%
Apparently, most impact is realised through perturbation of data related to wood or wood
derivatives, such as plywood and wooden window frames. Furthermore, the sensitive
categories are the waste heat balance and the global warming potential. These are exactly
the categories for which wood acts differently from other building materials. As explained in
section 5.3.3, the waste heat balance for renewable materials, such as wood, is negative, as
the energy content of the material (biomass) has not been used at the stage of production,
but is already counted within the waste heat balance. Also the global warming potential of
wood is negative in contrast to other materials due to the uptake of CO2 emissions during the
growth phase of the trees. For the wood-PUR-wood window frame, the combination of wood
and PUR results in a positive waste heat balance and a slightly negative global warming
potential. Through the presence of the negative LCI data from the wood, the perturbation of
the positive PUR waste heat balance data is emphasised, resulting in a larger variation than
1%. For all other categories and materials, all LCI data are positive and thus, the impact of a
perturbation of 1% remains 1% or less.
This phenomenon is also reflected in the results of the perturbation of the data from the
installation related matrices. Since no renewable materials are applied for the production of
installation components, no sensitive parameters have been found. For every perturbation,
101
CHAPTER 5
the multiplier is 1 or less. The largest impact has been found for normal concrete, applied for
floor heating, where a perturbation of 1% for one of the LCI input data for concrete results in
an increase of 1% for the output value for floor heating.
Impact on the life cycle inventory results of the whole building
To analyse the impact of perturbations on the life cycle inventory results of the whole
building, perturbations of 1% have been induced successively for all data in the VOLUME
matrices of 9 building variants. As could be expected from the perturbation analysis of the
basic matrices, the impact on the overall building results is very limited. For building variants
with wood frame construction, the largest impact is found for plywood, where a perturbation
of 1% for input data for plywood results in an increase of 0.9% of the overall results. For
massive building variants, the largest impact is found for concrete, where a perturbation of
1% for input data for concrete results in an increase of 0.9% of the overall results.
So, it can be concluded from the perturbation analysis that, due to the structure of the
building LCI model, none of the parameters is sensitive for perturbation and thus, no inherent
sensitivities exist. However, in practice, uncertainties exist for all data and by propagation the
uncertainty on the final result may become large. This is analysed in the next section through
an uncertainty analysis with Monte Carlo simulation.
5.5.4. Uncertainty analysis with Monte Carlo simulations
According to Heijungs et al. (2005), an uncertainty analysis aims at a systematic study of the
propagation of input uncertainties into output uncertainties. There are two basic ways of
running an uncertainty analysis: by random sampling, such as a Monte Carlo simulation, and
by analytical formulas for error propagation. As for most input data very little and often only
qualitative uncertainty information is available, the Monte Carlo simulation technique has
been selected.
Methodology
For most input data only one value is known with no quantitative information on the
uncertainty in terms of standard deviation. The known value is therefore assumed to be the
mean value. A normal distribution has been assumed for the input data. Several scenarios
have been selected for the coefficients of variance (= standard deviation/mean value), thus
providing assumptions for the standard deviation and for the width of the distribution curve of
the input data.
In a first step, the same variance coefficient has been selected for all input data and the
coefficients of variance for the output data have been calculated through 1000 Monte Carlo
102
simulation runs. An uncertainty analysis has been made for variance coefficients of 5%, 20%,
50% and 100%.
In a second step, the variance coefficient has been set at 5% for all input matrices, except for
one matrix for which the coefficient of variance has been set at 30%. Again, the coefficients
of variance for the output data have been calculated through 1000 Monte Carlo simulation
runs.
Firstly, the different scenarios have been executed and analysed for the constructional
model, thus only incorporating the matrices that contribute to the MATERIAL matrix (see eq.
[5.2]). Subsequently, the same analysis has been executed for the installation model, thus
only incorporating the matrices from the INSTALLATION matrix (see eq. [5.3]). Finally, the
propagation of input uncertainties has been analysed for the whole building model, by
combining the results of the previous two steps.
Results for the constructional model
Figures 5.4 and 5.5 present the coefficients of variance (CV) for the final result as a function
of the variance coefficient for the data of the input matrices. In this analysis, the CV remains
constant for all submatrices. Results are presented for the non renewable embodied energy
(NRE) and for the global warming potential (GWP).
The figures give the average CV and the min-max error bars for nine variants of the terraced
house, one of the buildings optimised within the EL²EP-project, presented in chapter 8. The
nine variants consist of three variants with cavity wall, three with massive walls and three
with wood frame construction. As the variants with wood frame construction (WFC) appeared
to have a large impact on the average value and on the maximum value, especially for the
global warming potential, distinction is made between the average including the wood frame
variants (black diamonds) and the average excluding them (grey triangles). The solid line
indicates where the CV of the output equals the CV of the input data.
Considering the non renewable embodied energy (figure 5.4), all dwelling variants have a
coefficient of variance for the output that is lower than the coefficients of variance for the
input data. Even the maximum error bars remain below the solid line. Even with standard
deviations up to 100% for the input data, the final result has a standard deviation of 70-85%.
This means that the propagation of errors is limited and that the errors of the different input
data neutralise each other somehow.
Considering the global warming potential (figure 5.5), similar results are found for the
massive variants. With coefficients of variance up to 100% for the input data, the final result
has a coefficient of variance of around 100%. Only for the variants with wood frame
construction, the errors of the different input data are propagated more intensively, resulting
in a CV for the final result that is 1.5 to 2.5 times higher than the CV of the input data. Cause
103
CHAPTER 5
of this larger sensitivity to errors is the fact that the global warming potential for wood and
wood derived products is always negative, whereas the GWP for all other materials is always
positive.
Uncertainty analysis via Monte Carlo simulations:
NON RENEWABLE EMBODIED ENERGY (NRE)
coefficient of variance for final
result NRE
1.2
1.0
0.8
0.6
0.4
incl WFC
0.2
excl WFC
0.0
0
0.2
0.4
0.6
0.8
1
coefficient of variance for submatrices
1.2
Figure 5.4: The coefficients of variance for the non renewable embodied energy of envelope
measures as a function of the coefficient of variance of the input data: average value,
including (black diamonds) and excluding (grey triangles) wood frame variants (WFC).
GLOBAL WARMING POTENTIAL (GWP)
result GWP
3.0
2.5
2.0
1.5
1.0
incl WFC
excl WFC
0.5
0.0
0
0.2
0.4
0.6
0.8
1
1.2
Figure 5.5: The coefficients of variance for the global warming potential of envelope
measures as a function of the coefficient of variance of the input data: average value,
104
Figures 5.6 and 5.7 present the coefficients of variance (CV) for the final result when
increasing the CV for one submatrix. In this analysis, the CV is set at 5% for all submatrices,
except for the submatrix mentioned in the x-as that has a CV of 30%. Results are again
presented for the non renewable embodied energy and for the global warming potential.
Considering the non renewable embodied energy (figure 5.6), the impact of a larger CV for
the different submatrices on the final result appears to be limited. The uncertainty on
submatrices related to transport has the least impact, with a CV for the final result of 5%. The
uncertainty for input data of material volume and specific weight has the highest impact.
However, as figure 5.6 shows, a CV of 30% on the input data for material volume or specific
weight results in a CV of only 10% for the final result. Similar to the results from figure 5.4, no
distinction can be found between massive and lightweight construction, resulting in small
min-max error bars for all submatrices and all variants.
result NRE
0.35
0.30
0.25
0.20
0.15
0.10
incl WFC
0.05
excl WFC
0.00
ECO mat
distance
weight ECO trans
submatrices
volume
Figure 5.6: The coefficients of variance for the non renewable embodied energy of energy
measures when the coefficient of variance of one input matrix is set at 30%. The CV of the
other submatrices remains 5%: average value, including (black diamonds) and excluding
(grey triangles) wood frame variants (WFC).
However, when considering the global warming potential (figure 5.7), the higher sensitivity for
error propagation for wood frame constructions reappears. As can be seen in figure 5.7,
increasing the CV of a submatrix related to material properties (ECOINVENT data for
materials, specific weight or material volume used in the building) results in the highest CV
105
CHAPTER 5
for the overall GWP (CV = 10 to 32%). For massive constructions, the sensitivity is much
lower, with a CV for the overall GWP of 10 to 15%.
result GWP
0.35
0.30
0.25
0.20
0.15
0.10
incl WFC
excl WFC
0.05
0.00
ECO mat
distance
weight
ECO trans
volume
submatrices
Figure 5.7: The coefficients of variance for the global warming potential of energy measures
when the coefficient of variance of one input matrix is set at 30%. The CV of the other
submatrices remains 5%: average value, including (black diamonds) and excluding (grey
triangles) wood frame variants (WFC).
Results for the installation model
The same analysis has been executed for the installation model. Results are presented in
figures 5.8 to 5.11. Similar conclusions can be drawn as for the building model. The
propagation of errors throughout the installation model is limited. Even with a coefficient of
variance of 100% for all submatrices, the coefficient of variance for the final result remains
between 82 and 141%. As in the installation model the input data for both the non renewable
embodied energy and the global warming potential are positive, the results obtained for both
objectives are comparable.
When analysing the impact of each submatrix separately, errors on input data related to
material properties (ECOINVENT data for composing materials and MAT matrix defining the
amount of material used for a certain installation component) have the highest impact. Errors
on the VOLUME matrix refers in the installation model to the uncertainty on length of pipes,
ventilation tubes and the needed power for the heating system. Their impact of the overall
CV is limited as can be seen in figure 5.10 and 5.11. Errors on transport related input data
106
(transport distance, transport weight and transport vehicle) only have a minor impact on the
overall uncertainty.
Uncertainty analysis via MC simulations:
result NRE
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0
0.2
0.4
0.6
0.8
1
1.2
Figure 5.8: The coefficients of variance for the non renewable embodied energy of
installation measures as a function of the coefficient of variance of the input data: average
value
result GWP
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0
0.2
0.4
0.6
0.8
1
1.2
Figure 5.9: The coefficients of variance for the global warming potential of installation
measures as a function of the coefficient of variance of the input data: average value
107
CHAPTER 5
result NRE
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
ECOinst
MATinst
distance
WEIGHTinst ECO trans
VOLUME
submatrices
Figure 5.10: The coefficients of variance for the non renewable embodied energy of
installation measures when the coefficient of variance of one input matrix is set at 30%. The
CV of the other submatrices remains 5%: average value
result GWP
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
ECOinst
MATinst
distance
WEIGHTinst ECO trans
VOLUME
submatrices
Figure 5.11: The coefficients of variance for the global warming potential of installation
measures when the coefficient of variance of one input matrix is set at 30%. The CV of the
other submatrices remains 5%: average value
108
Results for the whole building model
The non renewable embodied energy for the whole building is the sum of the constructional
related NRE and the installation related NRE. The same is valid for the global warming
potential. For the uncertainty analysis of the whole building model, the overall mean value
and the overall standard deviation are calculated for the combinations of each constructional
variant with each installation variant. When combining one constructional variant with one
installation variant, the mean NRE and the standard deviation on the mean NRE are
calculated in the following way:
NRE dwelling = NRE constr + NRE instal
stdev dwelling = stdev 2 constr + stdev 2 instal
[5.4]
The same approach has been applied for the uncertainty analysis of the overall global
warming potential. This approach assumes that the standard deviation of the constructional
part is independent from the standard deviation of the installation part. This is not totally
correct, as the input data for the transport vehicles and for some of the materials are the
same for both models. However, in the analysis of both models, the same coefficients of
variance are considered for the transport matrix and for the ECO matrix. When combining the
results for both models into results for the whole building model, only results with the same
assumptions for the CV are combined. This means that for the Monte Carlo simulations of
both models maybe not the exact same input data for transport are used, but surely the
same normal distribution is applied with the same mean value and the same standard
deviation. As Monte Carlo simulations of the whole building model appeared to be very time
consuming, this approach is assumed to be accurate enough.
Comparing figure 5.12 with figure 5.4 reveals the large similarity of both figures. This can be
explained by the fact that the NRE related to the constructional part of the building is more
than 10 times higher than the NRE related to the installations. The constructional model thus
has the highest impact on the overall result. The same phenomenon appears in the overall
result for the global warming potential (figure 5.13 vs. figure 5.5). Depending on the
combination of constructional variant and installation variant, the constructional components
have a contribution to the overall GWP that is 9 to 20 times higher than the contribution of
the installation. Due to this large impact of the constructional part of the model, the difference
between massive and wood frame constructions reappears in the uncertainty analysis of the
overall GWP. However, because of the addition of positive GWP related to the installation,
the difference between the results including and excluding wood frame constructions for the
109
CHAPTER 5
overall building model (figure 5.13) is smaller than in the case that only the constructional
model is analysed (figure 5.5).
result NRE
1.2
1.0
0.8
0.6
0.4
incl WFC
excl WFC
0.2
0.0
0
0.2
0.4
0.6
0.8
1
1.2
Figure 5.12: The coefficients of variance for the non renewable embodied energy of the
whole building as a function of the coefficient of variance of the input data: average value,
result GWP
2.5
2.0
1.5
1.0
0.5
incl WFC
excl WFC
0.0
0
0.2
0.4
0.6
0.8
1
1.2
Figure 5.13: The coefficients of variance for the global warming potential of the whole
building as a function of the coefficient of variance of the input data: average value, including
(black diamonds) and excluding (grey triangles) wood frame variants (WFC).
110
result NRE
0.35
0.30
0.25
incl WFC
excl WFC
0.20
0.15
0.10
0.05
0.00
ECO mat
MAT
distance
WEIGHT ECO trans VOLUME
submatrices
Figure 5.14: The coefficients of variance for the non renewable embodied energy of the
whole building when the coefficient of variance of one input matrix is set at 30%. The CV of
the other submatrices remains 5%: average value, including (black diamonds) and excluding
result GWP
0.35
incl WFC
excl WFC
0.30
0.25
0.20
0.15
0.10
0.05
0.00
ECO mat
MAT
distance
WEIGHT ECO trans VOLUME
submatrices
Figure 5.15: The coefficients of variance for the global warming potential of the whole
building when the coefficient of variance of one input matrix is set at 30%. The CV of the
other submatrices remains 5%: average value, including (black diamonds) and excluding
111
CHAPTER 5
Similarly, in the analysis of the contribution of each submatrix separately to the overall
uncertainty, the same phenomena reappear as explained above. Considering the non
renewable embodied energy, none of the submatrices has a dominant impact on the
uncertainty of the overall result. There is only a slightly higher impact of the submatrices
related to composing materials than of the submatrices related to transport (figure 5.14). The
submatrix MATinst has the lowest impact, because this matrix only intervenes in the
installation model (eq. [5.3]). Considering the global warming potential, the higher sensitivity
for error propagation for wood frame constructions reappears (figure 5.15). Submatrices
related to material properties (ECOINVENT data for materials, specific weight or material
volume used in the building) result in the highest CV for the overall GWP, however, less
pronounced than in figure 5.7 due to the addition of positive GWP values for the installation
components.
5.5.5. Conclusions on the sensitivity and uncertainty analysis of the LCI model
Due to the lack of detailed information on the uncertainty of the inventory data, the sensitivity
of the inventory model for errors and error propagation has been analysed trough a
perturbation analysis as well as an uncertainty analysis by MC simulations. Both studies
show that the sensitivity for errors and the propagation of errors is limited and that the errors
of the different input data neutralise each other somehow. Errors on input data related to
material properties have the highest impact, whereas errors on transport related input data
(transport distance, transport weight and transport vehicle) only have a minor impact on the
overall uncertainty.
Of more importance, however, are the results from the contribution analysis. They showed
the relative small importance of the embodied energy of a building compared to the energy
consumption during the utilisation phase. This is even more valid when comparing the
embodied energy of energy saving measures with the energy savings they realise. In most
cases, the embodied energy represents less than 10% of the primary energy savings over 30
years. Only extremely low energy buildings with a primary energy consumption of ca. 500 GJ
might have a total embodied energy higher than the energy use of the utilisation phase.
However, the sum of both remains small and the energy savings realised with these
dwellings are large, compared to the energy consumption of average dwellings.
Remarkably, the embodied energy for both massive and light weight buildings is comparable.
All these results lead to the important conclusion that considerations on the embodied energy
of extremely low energy houses can be an interesting issue for future research, but that in
the first place, effort should be paid to the reduction of the energy consumption during the
utilisation phase, as this phase still has the largest potential for improvement.
112
MODEL FOR ECONOMIC EVALUATION
CHAPTER 6. Model for economic evaluation
6.1
Introduction
The global methodology for developing extremely low energy and pollution dwellings consists
of three main pillars: energy, cost and ecology. To evaluate the economic impact of the
building concepts from the point of view of a private building owner, a cost database and a
cost evaluation model is required that can be integrated in the optimisation model. This
chapter first outlines the different economic evaluation criteria used in this research, followed
by a discussion on the strengths and weaknesses of these criteria. Subsequently, the cost
database is presented together with the cost models for building components and installation
components, including the secondary cost effects due to the large insulation thicknesses.
Finally, this chapter is concluded with a discussion on the impact of several aspects on the
final results, such as the uncertainty of cost data, energy price evolutions and life span and
the importance of the rebound effect and of the residual value of low energy dwellings.
6.2
Economic evaluation criteria
6.2.1. Description of the criteria
In the cost evaluation module, a large number of economic criteria is calculated and stored.
This creates not only opportunities for an in depth analysis of the results e.g. to compare the
contribution of constructional measures versus installation measures to the overall cost, or to
compare total costs versus extra costs, etc. It also provides a basis to evaluate the different
economic evaluation criteria and to determine the best criterion to be incorporated into the
optimisation process. The following criteria are calculated for each building variant:
-
Total initial investment cost for components of the building envelope: this contains the
cost for the whole building, being the costs for foundation, roofs, floors, inner and
outer walls, insulation, windows and doors, solar shadings and ventilation grids. It
excludes all costs for heating systems, domestic hot water systems and mechanical
ventilation systems (fans and pipes). The initial investment cost includes material
costs, installation costs and VAT (21% for new built dwellings, 6% for renovation).
-
Extra initial investment cost for components of the building envelope, compared to the
reference: as the reference mostly is a non insulated version of the building with
single or standard double glazing (U = 2.9W/m²K), the extra cost comprises mainly
the cost for insulation and for thermally better performing windows. Depending on
whether the reference building is assumed to be newly built or renovated, the extra
cost for windows only contains the surplus for thermally better performing glazing and
113
CHAPTER 6
window frames compared to the reference situation (new buildings) or it represents
the whole cost for better performing glazing and window frames plus the costs for
removing the old windows (renovation).
-
Investments for replacement of components of the building envelope before
discounting: the total replacement cost will highly depend on the considered utilisation
period of the building (30, 40, 60, 90,… years) and the assumed life span per
component (more than 90 years for all constructional elements, 20 years for double
glazing and ventilation grids, 10 years for sun shading).
-
Total initial investment cost for installation components: this contains the cost for
heating systems, domestic hot water systems and mechanical ventilation systems
(fans and pipes), including material costs, installation costs and VAT.
-
Extra initial investment cost for installation components, compared to the reference:
similar to the case of components of the building envelope, the extra cost depends on
whether the reference building is assumed to be newly built or renovated. In the first
case, the extra cost is only the surplus for the better performing installation compared
to the reference, whereas in the case of renovation, the extra cost contains the
overall cost for the systems plus the costs for removal of the old system. In case of a
well insulated new building, the extra cost for the heating system might be negative,
as the dimensions of the different components of the system mostly decrease in
comparison with the reference. This is further explained in section 6.4 on the
calculation models.
-
Investments for replacement of installation components before discounting: the total
replacement cost will highly depend on the considered utilisation period of the
building (30, 40, 60, 90,… years) and the assumed life span per component (20-25
years for heat production systems, 40 years for emission and distribution
components, 5-10 years for pumps, ventilators and control systems).
-
Annual energy cost for fossil fuels: depending on the application, the final annual
energy consumption for space heating and domestic hot water production is
calculated with a steady-state or dynamic building simulation programme. With this
consumption in MJ/year or kWh/year and the assumptions on energy cost per energy
carrier, the annual energy cost for fossil fuels can be calculated. The assumptions on
energy cost are presented in section 6.3.3.
-
Annual energy cost for electricity: as the optimisation process only focuses on the
building and its equipment for heating, domestic hot water and ventilation, the
electricity consumption is only related to the systems for heating, domestic hot water
and ventilation. Electricity might be the main energy carrier for heating or domestic
hot water production, or it might be used as auxiliary energy for the electrical
114
components of the installation. In both cases, the annual electricity consumption can
be calculated with a steady-state or dynamic building simulation programme. In case
a local electricity production unit is present, such as photovoltaic modules or a CHP
system, a grid-coupled system is assumed and the produced electricity is subtracted
from the consumed electricity on an annual basis. With this final electricity
consumption in kWh/year and the assumptions on electricity cost per kWh, the annual
energy cost for electricity can be calculated.
-
Total annual energy cost: being the sum of the annual energy cost for fossil fuels and
the annual energy cost for electricity
-
Annual energy cost savings related to consumption of fossil fuels, compared to the
reference: the cost savings are calculated as the difference between the annual
energy costs of the reference case and the annual energy costs of the improved
case.
-
Annual energy cost savings related to electricity consumption, compared to the
reference: the cost savings are calculated as the difference between the annual
electricity costs of the reference case and the annual electricity costs of the improved
case.
-
Total annual energy cost savings: being the sum of the annual energy cost savings
for fossil fuels and the annual energy cost savings for electricity. For the improved
case, there might be a shift in energy carrier or the systems might have a much
higher auxiliary electricity consumption, compared to the reference. Therefore,
comparisons should be made on the total annual energy cost or cost savings, in order
to include all secondary cost effects of improved systems.
-
Annual maintenance costs: these costs only consider maintenance costs for the
heating systems and the mechanical ventilation system. This way, maintenance
differences between different systems can be incorporated into the cost evaluation.
Maintenance costs for the building as such are not included.
-
Total present value: with the initial investment costs, the costs for replacement and
the annual energy and maintenance costs, the total present value can be calculated
based on assumptions for utilisation period, discount rate and price evolutions,
according to equation [6.1]
TPV = I 0 +
∑
j = x, y , z
I j (1 + rI ) j
(1 + a) j
n
+∑
i =1
n
K E (1 + rE ) i
K M (1 + rM ) i
+
− R0
∑
(1 + a) i
(1 + a) i
i =1
[6.1]
with:
I0
Ij
the initial investment [€]
the investment for replacement j at time x, y or z [€]
115
CHAPTER 6
KE
KM
n
R0
rI
rE
rM
a
-
annual maintenance cost [€]
considered utilisation phase [year]
discounted residual value of the building at time n
Net present value: with the extra initial investment costs, the extra costs for
replacement and maintenance and the annual energy cost savings, the net present
value can be calculated based on assumptions for utilisation phase, discount rate and
price evolutions, according to equation [6.2]
⎡
NPV = − ⎢ EI 0 +
⎢⎣
∑
j = x, y, z
EI j (1 + rI ) j
(1 + a ) j
n
+∑
i =1
∆K M (1 + rM ) i ⎤ n ∆K E (1 + rE ) i
+ R0 [6.2]
⎥+∑
i
(1 + a ) i
⎥⎦ i =1 (1 + a )
with:
EI0
the extra initial investment compared to a reference case [€]
EIj
the extra investment for replacement j at time x, y or z compared to a
reference case [€]
∆KE
annual energy cost saving compared to a reference case [€]
∆KM annual extra maintenance cost compared to a reference case [€]
n
considered time period for evaluation [year]
R0
residual value of the building at time n
rI
rE
rM
a
The role and importance of the residual value in the cost evaluation is subject to
debate. It is therefore discussed as a separate item in section 6.5.5.
-
Simple pay back time: (SPBT in years) is calculated based on the initial investment
cost and the annual energy cost savings, without taking into account inflation or
discounting of costs and savings:
SPBT =
-
I0
∆K E
[6.3]
Dynamic pay back time: (DPBT in years) is calculated, similar to the simple pay back
time, but taking into account inflation and discounting of the annual energy cost
savings. The DPBT can be calculated from the following equation:
116
DPBT
∑
t =0
-
∆K E
= I0
(1 + a ) t
[6.4]
The internal rate of return (IRR:) is calculated iteratively from equation [6.2], with
NPV=0 and a=IRR.
-
Total discounted cost per saved kWh primary energy: is calculated by dividing the
difference in total present value between the reference case and the improved case
by the difference in primary energy between both cases.
-
Total discounted cost per avoided ton CO2: for each energy carrier, the CO2
emissions coupled to the energy consumption are calculated based on the values in
table 6.1. The emissions from electricity production depend on the location.
Therefore, two slightly different values are used in the projects, described in chapter 7
and 8. The dependence of the CO2 emissions from electricity production on time (day
time versus night time and weekends) is not taken into account here.
Fuel
Natural gas
Electricity
[kg/kWh]
[kg/kWh]
[kg/kWh]
0.264
0.192
0.310*
0.296**
Table 6.1: CO2 emissions for the different energy carriers
*
: based on information from Electrabel (Verbeeck and Hens 2002)
**
: based on the Energy Balance 2002 of the Brussels Capital Region (ICEDD 2004)
By comparing the CO2 emissions of the reference case and the improved case, the
total discounted cost per avoided ton CO2 can be calculated by dividing the difference
in total present value by the difference in CO2 emissions between both cases.
6.2.2. Criteria for optimisation
Although the cost evaluation module calculates and stores a large number of economic
criteria, only one cost criterion can be incorporated in the optimisation process. As explained
in chapter 4, the optimisation is always performed in two steps. In the first step, only energy
saving measures related to the building envelope are considered, whereas in the second
step, installation related variables are optimised.
As in the first step only the net heat demand is calculated and no assumptions are made on
the heating system, there is in this phase not enough information available to calculate the
annual energy cost. Therefore, only the initial investment cost can serve as cost criterion in
117
CHAPTER 6
this phase. In the second phase then, the heating system is defined and so, the annual
energy consumption and the annual energy cost can be calculated.
However, there still remains the choice between the total cost of the buildings and the extra
cost compared to the reference as the most appropriate economic optimisation parameter.
When considering the total cost of the building, comparison of the costs for the building
envelope, the systems and the energy costs shows the very large impact of the building
envelope on the overall cost, although only part of these costs has an impact on the energy
consumption. A large part is fixed costs for foundation, roof construction, inner walls, etc.
This clearly appears from figure 6.1 that presents the total present value for two variants for
the building envelope of the terraced house of the EL²EP-project: a very energy conserving
variant with Umean = 0.22 W/m²K (U0.2) and less energy conserving variant with Umean = 0.54
W/m²K (U0.5). Each of these two building versions is combined with some 130 different
combinations of heating system, thus resulting in a cloud of values per variant. For each
combination of building envelope and heating system, the total present value over a period of
30 years is given, together with the division in TPV for the building envelope (BUI), TPV for
the heating systems (SYS) and TPV for the energy cost (ENER). This figure shows that for
all cases, the discounted costs for the building envelope (initial costs and cost for
replacements) represent 80 to 90% of the overall total present value, whereas costs for
heating systems and energy costs only represents each 5 to 10% of the total present value.
TPV vs total primary energy consumption
terraced house, utilisation phase of 30 years, low energy price scenario
total present value [€]
300000
U0.5 TPV30
U0.5 BUI
U0.5 SYS
U0.5 ENER
250000
200000
150000
100000
U0.2 TPV30
U0.2 BUI
U0.2 SYS
50000
U0.2 ENER
0
800
1300
1800
2300
2800
primary energy consumption [GJ]
Figure 6.1: Terraced house, utilisation period of 30 years, low energy price scenario: total
present value versus total primary energy consumption for two variants of the building
envelope: U0.5 = 0.54 W/m²K (in grey); U0.2 = 0.22 W/m²K (in black); BUI = TPV for the
building envelope, SYS = TPV for the installation system, ENER = TPV for the energy costs
118
Extra cost vs primary energy savings
terraced house, utilisation phase of 30 years, low energy price scenario
extra discounted cost [€]
150000
U0.5 SUM
100000
U0.5 BUI
U0.5 SYS
U0.5 ENER
50000
0
800
1300
1800
2300
U0.2 SUM
2800
U0.2 BUI
U0.2 SYS
-50000
U0.2 ENER
-100000
primary energy consumption [GJ]
Figure 6.2: Terraced house, utilisation period of 30 years, low energy price scenario: extra
discounted cost versus total primary energy consumption for two variants of the building
envelope: U0.5 = 0.54 W/m²K (grey); U0.2 = 0.22 W/m²K (black); BUI = extra cost for the
building envelope, SYS = extra cost for the installation system, ENER = energy cost savings,
SUM= sum of all
When presenting the results for the same combinations as extra costs compared to the
reference (figure 6.2), the difference between both variants is much larger, especially
concerning the extra costs for the building envelope. For the most energy conserving variant
(U0.2), the extra cost for the building envelope is three times higher than that for the less
energy conserving variant (U0.5). For the latter, the extra costs for the building envelope and
for the heating system are of the same magnitude. Figure 6.2 also shows that for both
building envelope variants, the energy cost savings (ENER) are of the same magnitude. For
both cases, the type of heating system also has a large impact on the energy cost saving.
Comparing the optimisation process for TPV as a criterion and NPV as a criterion, however,
shows that both produce the same results. Anyhow, as will be seen in chapter 8, the total
present value has been chosen as optimisation criterion, but for the interpretation of the
results, the focus is put on the net present value, as only solutions with a positive net present
value can be considered economically viable.
119
CHAPTER 6
6.3
Cost database
6.3.1. Cost data for the building envelope
The construction of a cost database has been started already in earlier work on energy
saving measures in buildings (Verbeeck and Hens 2002). This database contained cost data
for materials that improve the thermal quality of the building envelope, such as insulation
materials and thermally better performing glazing types and window frames. Depending on
the building material, costs are expressed as a function of volume (insulation), area (glazing)
or length (window frames). These costs are updated and extended with cost data for all kinds
of materials applied in the building envelope, such as bricks, concrete blocks, concrete for
floors and flat roofs, tiles, finishing materials, ventilation grids, solar shading devices, wood
frame constructions for floors and sloped roofs, etc. The extended data are mainly based on
cost data provided by building contractors that were found willing to make a price offer for the
reference buildings of the EL²EP-project. The price offers comprised the working hour cost.
With these data, a cost database has been created that gave the possibility to calculate the
overall construction cost of a building with massive walls, cavity walls or wood frame
construction. The database has been structured in a way that the costs for replacement of
components could easily be extracted as separate data. Annex E presents part of the cost
database as an example of the structure. Obviously, these cost data will need to be kept up
to date, if used in future work.
6.3.2. Cost data for components of the installation
The cost data for components of the installation are also based on up to date (from 2005)
price information for boilers, radiators, convectors, floor heating systems, storage tanks, fans,
pipes, etc. However, as the insulation level of a building directly affects the needed power
and the dimensions of the heating system, the cost needs to be expressed as a function of
the insulation level. The approach that is adopted is further explained in section 6.4.2.
6.3.3. Energy prices and price evolution
The assumptions for the energy prices are based on private consumer prices. For electricity
and natural gas, the adopted prices come from the Federal Ministry of Economic Affairs; for
fuel, the adopted price comes from Informazout, the Belgian umbrella organisation of fuel
distributors, and is valid for a purchase of more than 2000 litres of fuel. As the studies,
presented in part three are executed at different times, the adopted energy prices differ. For
the BIM-project in chapter 7, the average prices for 2004 are applied, whereas for the EL²EP-
120
project in chapter 8, the prices of May 2006 are applied. Table 6.2 presents the energy prices
for both projects. The prices for electricity and gas include taxes for transport, distribution,
energy taxes and federal taxes. In the BIM-project, only the proportional term is taken into
account, whereas in the EL²EP-project, also the fixed term for gas and electricity is taken into
account.
Energy prices
Natural gas
Fuel
Tarif B
for heating
BIM-PROJECT: average prices for 2004
Proportional
3.36
3.77
term [c€/kWh]
EL²EP-PROJECT: prices of May 2006
Proportional
4.64
5.77
term [c€/kWh]
Electricity
Twofold
Twofold
Exclusive
day price
night price
night price
16.56
8.35
-
18.33
9.64
7.91
Fixed term
103.46
40.40
17.73
[€/year]
Table 6.2: Energy prices for natural gas, fuel and electricity for the BIM-project (chapter 7)
and the EL²EP-project (chapter 8)
Scenarios for energy
Low
Medium
High
% per year
% per year
% per year
Natural gas
0%
2.1%
4.3%
Fuel
0%
1.9%
3.2%
Electricity
0%
2.1%
4.3%
price evolution
Table 6.3: Three scenarios for energy price evolution for natural gas, fuel and electricity
Within the BIM-project and the EL²EP-projects, three different scenarios are used for the
price evolution of gas, fuel and electricity: a low, medium and high scenario. The values for
the medium and high scenario are based on the EU POLES scenarios from 2000 until 2030
for gas and fuel (EU 2004). The values are presented in table 6.3. However, only the growth
factors of the EU POLES scenarios are adopted, not the starting values.
6.4
Calculation models
6.4.1. Building components
For each building variant, the applied volume, area or length per constructional material is
calculated and stored in a VOLUME matrix. Different volume matrices are calculated for
121
CHAPTER 6
different utilisation periods, taking into account replacement of materials and products based
on the assumptions of life span for each material or product. These volume matrices are the
same as used within the life cycle inventory model.
6.4.2. Components of the installation
As the dimensions of the heating system strongly depend on the insulation level and on the
type of building, a detailed dimensioning of the heating system for each building variant
according to the ruling standards would be necessary. As this approach is infeasible within
an optimisation method, cost curves have been determined for each reference building that
express the cost of the heating system as a function of the insulation level. These cost
curves are based on a detailed analysis of 7 insulation levels per reference building (Hens
2005). For each of these levels, the needed power is calculated and the heating system is
designed according to NBN EN 12831 (2003). Analysis of the price information from
producers of boilers and radiators already revealed that the price of components of the
installation strongly depend on their maximum power level (Verbeeck and Hens 2002). So
with the results of the detailed dimensioning of the heating system for different insulation
levels and the price information of the installation components, cost curves have been
determined that express the investment cost for the different components as a function of the
insulation level. These cost curves could easily be integrated in the cost module. The
overdimensioning of heating systems, as occurs permanently in practice, is not taken into
account in these curves.
6.4.3. Integration of secondary cost effects
Application of large insulation thicknesses does not only affect the investment cost through
the direct cost of the insulation, but also leads to secondary costs effects. Due to the large
insulation thickness, the overall thickness of the wall increases significantly, thus leading to
necessary adaptations in the construction in comparison with a traditional dwelling. The
thickness of the foundation has to increase to be able to bear the wall, the coupling between
roof and wall has to be adapted as well as the coupling between window and wall. The large
insulation thickness also results in more material use at the corners and asks for adapted
solutions to fix the insulation material and to avoid thermal bridges.
The extra material use for the foundation and the walls is integrated in the cost database.
The extra cost to avoid thermal bridges appeared to be negligible in comparison to the
overall building cost. Details on the approach are beyond the scope of this research, but can
be found in Van Londersele and Janssens (2007).
122
6.5
Discussion of different assumptions
6.5.1. Uncertainty on cost data
The uncertainty on cost data is quite high in the building sector. Architects use basic prices
when estimating the overall cost of a building project, but the final cost will strongly depend
on the building contractor who effectively executes the job. Prices can be twice as high
differing from one contractor to another. In the attempt to integrate realistic price information,
the cost database is based on price offers from building contractors. However, only very few
contractors were found willing to present a price offer. Moreover, none of them was familiar
with the application of large insulation thicknesses, wood frame construction or a high level of
air tightness. Furthermore at this moment, most extremely low energy dwellings are built with
a large contribution of the building owner in the construction process, e.g. for the installation
of the large packs of insulation or for sealing all air leakages to realise the needed high level
of air tightness. This practice cannot be integrated in a general cost database. The prices in
this research are based on real prices for traditional building practice, but with a high
probability for a deviation between calculated costs and real construction costs.
Nevertheless, the uncertainty analysis of the life cycle inventory model can serve as an
indication for the uncertainty on the cost data, as both the inventory and the costs are based
on a similar methodology, being the multiplication of volume data with inventory or cost data.
The uncertainty analysis of the life cycle inventory model showed that due to the linearity of
the model, the propagation of errors is limited and overestimation of certain parameters
partly compensates underestimation of others. This can also be assumed for the cost model.
Furthermore, the calculated costs should not be considered as exact values, but more as an
indication of the cost differences between different levels of energy performance of buildings.
This way, although it can be expected that the real cost of the buildings will be higher than
the estimated cost, the estimations can be considered acceptable to compare different
building concepts and to determine tendencies and priorities for extremely low energy
dwellings.
6.5.2. Uncertainty on energy price evolutions
The uncertainty on the evolution of the energy prices may be considered as even higher than
the uncertainty on the cost data. Energy prices always strongly depended on evolutions on
global constraints, such as national and international energy policies, geopolitics, relations
between nations and the global economic evolution. This will not change in the future, surely
not in view of the pending environmental treats, the evolution in fossil stocks, etc.
(D’haeseleer 2006). The only way to deal with this uncertainty is by taking into account
123
CHAPTER 6
different scenarios for the energy price evolution and to analyse the robustness of the results
for the scenarios. Of importance are the tendencies and priorities that can be derived and
their dependency on the energy price scenarios. This will be discussed in detail on the basis
of the results of the BIM-project (chapter 7) and the EL²EP-project (chapter 8).
6.5.3. Impact of the utilisation period
Within this research, the choice has been made not to consider the whole lifecycle of the
buildings, but to adopt an utilisation period that corresponds more or less to the time one
generation occupies the building. As explained in section 2.5 and 5.2.2, this choice is based
on the assumption that building owners are not willing to invest in measures that are only
economically viable on a longer term than their period of ownership of the house. This period
of ownership is estimated at 30 to 40 years. For the BIM-project, an utilisation period of 40
years, for the EL²EP-project, an utilisation period of 30 years is adopted. However, in the
EL²EP-project, all calculations of primary energy consumption, global warming potential and
total and net present value are executed also for 60 and 90 years, especially to analyse the
impact of the utilisation period on the final results.
Figure 6.3 presents the results for the optimal solutions for the terraced house of the EL²EPproject, considering an utilisation period of 30, 60 and 90 years. The difference in utilisation
period has an impact on the net present value through the energy costs and the costs for
replacements. The results are valid for the low energy price scenario (+0%, see table 6.3)
and a discount rate of 4%. Only the solutions with a positive net present value over the
considered periods are presented.
A first conclusion is that almost the same variants appear as economically viable, regardless
of the considered utilisation period. In fact, the results for one variant have the same annual
energy cost, but different net present value and thus are lying in one vertical line. Obviously,
the net present value increases with the considered period as more years of energy cost
savings can be incorporated and these energy cost savings outweigh the extra replacement
costs during a period of 60 or 90 years. There are just a few extra variants that only become
viable when the utilisation period is extended, but these variants have only very small net
present value, even after 60 and 90 years. Furthermore, as can be seen in the figure, due to
the discounting, the near future has a larger impact on the net present value than the far
future. This explains the limited difference in net present value between the results for 60 and
90 years.
The most important conclusion, however, is that the hierarchy of energy saving measures
does not change with the considered utilisation period. What is the best solution over 30
years remains the best solution, even for 60 to 90 years. A detailed discussion on this
124
hierarchy, including the dependency on the energy price scenario will be hold in chapter 7
and 8, based on the results of the BIM-project and the EL²EP-project.
Impact of the utilisation period
Terraced dwelling, low energy price scenario
14000
net present value [€]
12000
10000
8000
6000
optima 30 years
optima 60 years
4000
2000
optima 90 years
0
200
400
600
800
1000
Figure 6.3: Terraced house, low energy price scenario: impact of the utilisation period on the
optimal solutions: net present value and annual energy cost are given for the optimal
solutions for an utilisation period of 30, 60 and 90 years
6.5.4. Integration of rebound effect
There is no doubt about the existence of a rebound effect when improving the energy
performance of a building. As has been explained in section 3.3.2, empirical evidence clearly
shows that part of the improvement does not contribute to energy savings but is used to
improve the originally low comfort level. This effect is not incorporated as such in the global
methodology and thus not taken into account in the main results of the projects presented in
chapter 7 and 8. Main reason for not taking it into account is that all buildings are evaluated
for the same overall performance, according to IEA Annex 32 (Hendriks and Hens 2000).
This means that one of the key assumptions in the assessment of (extremely) low energy
dwellings is that a minimum comfort level should be satisfied. Therefore, in order to compare
dwellings on a correct basis from the point of view of energy, emissions and cost, a
prerequisite is that all dwellings at least have an acceptable indoor climate. This is
implemented by imposing temperature profiles per room to be satisfied during the heating
season, by controlling and limiting the risk for summer overheating and by always designing
the ventilation system according to the ruling standard NBN D50 001 (1991). This way, all
building variants are compared on an equivalent basis. However, it is obvious that these
125
CHAPTER 6
assumptions do not cope with reality and that they lead to an overestimation of the energy
consumption, especially for energy-devouring dwellings where the occupant normally
reduces his comfort level in order to limit his energy bill. Therefore, in order to frame the
results in a more realistic context, the impact of the comfort level on the energy consumption
is briefly discussed here.
The net heat demand (NHD) and the end energy consumption (EEC) of the terraced house
of the EL²EP-project are calculated for different comfort levels and different energy saving
variants. Table 6.4 presents the comfort levels for which the net heat demand and end
energy consumption are calculated. For the living room and the kitchen, the maximum set
temperature is valid in the morning and the evening during the week and during the whole
day in the weekend, whereas the minimum set temperature is valid during the night. For the
bathroom, the maximum set temperature is only valid during 2 hours in the morning and 3 to
4 hours in the evening. For the sleeping rooms, the maximum set temperature is only valid
from 16h to 21h. In case of a minimum and maximum set temperature for the hall, the same
heating schedule is followed as for the living room.
Living room +
kitchen
Bathroom
Sleeping
rooms
Hall
Comfort level 0
20°C / 12°C
22°C / 12°C
-
-
Comfort level 1
20°C / 15°C
22°C / 12°C
-
-
Comfort level 2
20°C / 12°C
24°C / 12°C
-
-
Comfort level 3
21°C / 15°C
22°C / 12°C
-
-
Comfort level 4
21°C / 15°C
24°C / 12°C
-
-
Comfort level 5
21°C / 15°C
24°C / 18°C
15°C
15°C
Comfort level 6
21°C / 15°C
24°C / 18°C
18°C / 15°C
18°C / 15°C
Comfort level 7
21°C / 15°C
24°C / 20°C
20°C / 15°C
21°C / 15°C
Comfort level EL²EP
21°C / 15°C
24°C / 21°C
20°C / 15°C
21°C / 15°C
Table 6.4: Terraced house of the EL²EP-project: different comfort levels applied to different
energy saving variants in order to calculate the rebound effect
The different comfort levels are applied to different energy saving variants. Firstly, variants
are derived by improving the mean U-value of the building stepwise from Umean = 2.0 W/m²K
to Umean = 0.34 W/m²K without changing the glass area. Secondly, variants are simulated
with an improved U-value and a different glass area, resulting in an overall mean U-value
between 0.19 and 0.54 W/m²K. The net heat demand is calculated for all comfort levels and
all building variants with TRNSYS, a dynamic building simulation programme. The end
energy consumption is calculated from the net heat demand, assuming an overall efficiency
126
of 70% for the heating system of the non insulated reference case, an efficiency of 80% for
improved variants with Umean > 0.5 W/m²K and 85% for variants with Umean < 0.5 W/m²K.
Results are presented in figure 6.4. In the theoretical case with the highest comfort level for
all variants (comfort level EL²EP), the end energy consumption is 110,700 MJ/year for the
non insulated variant (Umean = 2.0 W/m²K, black square) and 25,160 MJ/year for the best
performing case (Umean = 0.19 W/m²K, cross most at the left hand side). Thus, according to
these theoretical assumptions, the improvement should result in a reduction of the energy
consumption with 77%. However, every day practice shows that in non insulated dwellings, a
very comfort level (comfort level 0) is a more realistic assumption, as for higher comfort
levels the energy cost would be perceived too high. This results in an energy consumption of
59,900 MJ/year for the terraced house (black triangle), thus representing only 54% of the
theoretical energy consumption (black square). According to these assumptions, the energy
conservation of the best insulated variant is only 58%.
Assuming that the energy cost for the non insulated building with comfort level 0 (black
triangle) is acceptable for the building owner, it can be expected that initially improvement of
the energy performance of the building, will not result in any energy conservation, but only in
an improvement of the comfort level to the highest level. This appears in figure 6.4 from the
fact that the energy consumption for a building with Umean = 1.06 W/m²K (grey square at
Umean= 1.06W/m²K) remain identical (ca. 60,000 MJ/a) to the energy consumption of the non
insulated building with low comfort level (black triangle). Since for variants with Umean
between 1.0 and 2.0 W/m²K, the energy consumption at high comfort level (grey squares) is
higher than the energy consumption of the non insulated variant at low comfort level (black
triangle), it is highly probable that the improvement of the building envelope does not lead to
energy conservation, but only to improvement of the comfort level, with a comfort level in
between the lowest and the highest. Only in case of larger improvements of the building
envelope, with Umean < 1.0 W/m²K, real energy savings can be expected, but these savings
will be in any case much lower than theoretically assumed.
Furthermore, in case also the glass area is changed (grey crosses) in comparison with the
non insulated case, it is even possible to have well insulated building variants that still have
the same end energy consumption as the non insulated case, obviously with a much better
comfort level than for the reference. This can be seen in figure 6.4, by comparing the black
triangle (non insulated, low comfort) with the grey cross with Umean ≈ 0.55 W/m²K (well
insulated, high comfort, but also higher glass area than the non-insulated case): they both
have an energy consumption of ca. 60,000 MJ/a. Nevertheless, an improvement of the
heating efficiency can dampen the rebound effect somewhat, as the same comfort level can
be realised with a higher efficiency and thus, with a lower energy consumption.
127
CHAPTER 6
It can be concluded that it is highly probable that the energy savings, theoretically predicted
in the projects in part three will overestimate the real savings. This will have an impact on the
economic viability of measures of which the net present value is only slightly positive, but it
cannot be expected to have a large impact on the hierarchy of potentially energy saving
measures.
Impact comfort level on energy consumption
End energy consumption [MJ/a]
Terraced dwelling
120000
100000
80000
60000
40000
20000
0
0.0
0.5
1.0
1.5
2.0
2.5
mean U-value [W/m²K]
non-insulated, low comfort level
non-insulated, high comfort level
variants with low comfort level
variants with high comfort level
high comfort level and variation glass area
Figure 6.4: Terraced house: impact of the comfort level and the glass area on the end
energy consumption, given as a function of the mean U-value
6.5.5. Integration of the residual value
In earlier work (Verbeeck and Hens 2002), the total present value has been calculated,
without taking into account the residual value of the building at the end of the considered
period. Also in the literature, very little information is found on the integration of the residual
value of buildings in the cost assessment of energy saving investments. Tommerup and
Svendsen (2006) are one of the few authors who discuss the residual value of energy saving
investments in private buildings. They use the technique of linear depreciation over the
expected lifetime of the measures. Banfi et al. (2006) developed a model to evaluate the
consumers’ willingness to pay for energy saving measures in buildings, but they did not
incorporate any choice parameter related to the residual value.
However, there are several trends that are likely to result in higher valuation of energy saving
measures in buildings in the future, maybe leading to an economically interesting residual
128
value in case the dwelling is sold. One issue is the introduction of the energy certificate,
planned for 2009. From that time on, every building that is sold or rented has to dispose of an
official certificate that labels the primary energy consumption of that building under standard
conditions. This way, the energy consumption will be clearly presented as one of the
assessment criteria in contrast to the situation of today where most individuals who are
looking for a house, do not consider energy consumption as an assessment criterion.
Another issue is the slowly increasing interest in (extremely) low energy houses, such as
passive houses. These concepts increase the building cost significantly, especially the cost
for the building envelope, but have very low energy consumption during the utilisation phase.
Most of these buildings are still young and occupied by their first building owners, but it can
be expected that in case these buildings are sold, the original building owners will present
this low energy consumption as one of the main qualities of the building and incorporate it in
the demand price.
So, although integration of the residual value of energy saving measures is not yet common
practice, two different approaches have been tested here in order to analyse its magnitude.
In a first approach, the surplus value of a well insulated dwelling is estimated with a simple
calculation module that is freely available on the website of KBC, one of the main Belgian
banks (KBC 2003). Based on some simple questions on condition, location and dimensions
of the building, the module estimates the value of the building. The needed input concerns
information on land price, dimensions of the parcel, location (rural or urban), constructional
characteristics (including insulation, window frames and glazing), finishes (including heating
system), dimensions of the rooms, construction year, renovations and wear and tear. To
estimate the land price, a separate module is provided. For the input of insulation, window
frames and glazing, the following choices can be made:
-
Insulation: none / not everywhere / everywhere
-
Window frames:
simple (softwood, grey aluminium) / medium (pvc, aluminium,
meranti) / luxurious (hardwood, enamelled aluminium, pvc with metal skeleton)
-
Glazing: single glazing / double glazing, medium price / double glazing, high price
Obviously, these categories are unable to characterise the thermal quality of the building.
Figure 6.4 shows the input sheet for the constructional characteristics of the calculation
module.
129
CHAPTER 6
Figure 6.4: Input sheet for the constructional characteristics of the building valuation module
of the bank KBC (KBC 2003)
To analyse the approach of this calculation module, four existing dwellings have been tested
for all combinations of insulation, window frames and glazing. This revealed the methodology
of the module. The total estimated value of a dwelling strongly depends on the construction
year and the location. Based on the estimated value of the reference case of a dwelling (no
insulation, single glazing and simple window frames) and the estimated surplus value of the
improved variants of the same building, it appears that the calculation module employs a
basic amount that is 0.7 - 0.8% of the total value of the dwelling. The condition of insulation,
window frames and glazing can be valued with a score (bad = 0; better = 1; good = 2). The
estimated surplus value of the energy saving measures appears to be always an exact
multiple of this basic amount, with the multiple being more or less equal to (valuation
score+1). Table 6.5 gives some details of the tested dwellings and the results of the analysis.
For these buildings, the surplus value varies from 2,500€ to 16,800€, mainly depending on
the age of the building and the number of measures. Applying this approach to the buildings
of the EL²EP project and assuming a building age of 30 years, the surplus value will be
13,000 to 14,000€ depending on the location. This value can be interpreted as the
130
discounted value. Compared to the overall total present value, having an order of magnitude
of 200,000€ and more (figure 6.1 section 6.2.2) this residual value is quite limited. So, it can
be expected that according to this approach, the incorporation of the residual value will not
have a significant impact on the economic objective in the optimisation process.
Dwelling
1
Dwelling
2
Dwelling
3
Dwelling
4
Terraced
Detached
Terraced
urban
Antwerp
Rural
Limburg
urban
Fl-Brabant
1905
1967
1995
1974
Basic amount [€]
832.30
1,950.12
2,394.88
1,836.10
% of total value of the dwelling
(non insulated)
0.73%
0.82%
0.85%
0.82%
GENERAL INFORMATION
Type of building
Location
Year of construction
Semidetached
rural
Fl-Brabant
SURPLUS VALUE compared to the NON INSULATED SITUATION
insulation
window
frames
glazing
score
Surplus value [€]
(= number x basic amount)
None
Medium
Double
medium
2
2,496.90
(3xba)
3,900.23
(2xba)
7,184.63
(3xba)
5,508.30
(3xba)
None
Luxury
Double
high
4
4,993.80
(6xba)
9,750.57
(5xba)
11,974.40
(5xba)
9,180.50
(5xba)
Overall
Medium
Double
medium
4
4,161.50
(5xba)
7,800.45
(4xba)
11,974.40
(5xba)
9,180.50
(5xba)
Overall
Luxury
Double
high
6
6,658.40
(8xba)
13,650.79
(7xba)
16,764.26
(7xba)
13,062.54
(7xba)
Table 6.5: Details of the dwellings, tested with the KBC building valuation module and results
of the valuation of energy saving measures
In a second approach, the residual value for energy saving measures is calculated based on
linear depreciation of the investment cost over the expected lifetime of the measure. Two
situations are possible:
-
the expected lifetime of the measure is longer than the assumed utilisation period.
This is the case for insulation.
-
the expected lifetime of the measure is shorter than the assumed utilisation period.
This is the case for glazing, pumps, fans, heat production systems, storage tanks,
control systems, solar collectors and photovoltaic modules. For each of these
elements, one or more replacements are considered within the assumed utilisation
period, depending on the assumed lifetime.
131
CHAPTER 6
With UP as the assumed utilisation period, LT the assumed lifetime for a certain measure
and CM the investment cost of the measure, the undiscounted residual value can be
calculated as follows:
If LT > UP → RV = (1 −
UP
)CM
LT
If (n.LT ) < UP and (n + 1).LT > UP → RV = (1 −
[6.5]
UP
)CM
(n + 1) LT
[6.6]
The discounted residual value R0 in equation [6.1] and [6.2] can then be calculated as
follows:
R0 =
RV
(1 + a )UP
[6.7]
Figure 6.5 gives the discounted residual value (discount rate = 4%) as a function of the
annual net heat demand for different versions of two buildings of the EL²EP-project. Only
investments for insulation, glazing and a natural ventilation system are taken into account in
these results. As the figure shows, the residual value strongly depends on the net heat
demand, especially for low energy houses with a net heat demand < 60MJ/m³a.
Incorporating the production system for heating increases the residual value with 270 to
360€. Even the incorporation of alternative energy systems, such as a solar collector and PV
modules, has a limited impact on the residual value: ca. 1,200€ for a solar collector and ca.
1,400€ for PV modules when assuming a lifetime of 25 years for both systems. With this
approach, discounted residual values are obtained within the range from 2,000 to 16,000€,
the latter for an extremely low energy dwelling with 4m² solar collectors and 1kWpeak PV
modules.
132
Residual value
discounted over 30 years, a=4%
discounted residual value [€]
16000
14000
terraced
detached
12000
10000
8000
6000
4000
2000
0
0
50
100
Net heat demand (MJ/m³a)
150
Figure 6.6: Discounted residual value for the terraced (black) and detached (grey) reference
house of the EL²EP-project as a function of the net heat demand
This approach produces results with the same order of magnitude as the estimation module
of KBC bank, even for extremely low energy dwellings. In contrast to the KBC calculation
module, this approach offers a more differentiated residual value depending on the effective
investment cost of the energy saving measures. Weakness of this model is that location is
not taken into account. However, in comparison with the total present value, both
approaches produce residual values that are a factor 10 smaller than the total present value
of the building envelope. Anyhow, the impact on the economic objective in the optimisation
process will be limited. The residual value is therefore not taken into account in the
optimisation process. Which approach will approximate most to the reality of the market of
energy conserving buildings will strongly depend on the future evolution of the energy prices
and of the energy policy, such as the way the energy certificate will be introduced.
133
BIM-PROJECT
PART THREE: APPLICATIONS
CHAPTER 7.
7.1
BIM-project
Introduction
In the Brussels Capital Region, the most urban Region in Belgium, the energy consumption
in buildings represents 71% of the total regional energy consumption (ICED 2004).
Additionally, the European Energy Performance of Buildings Directive obliges the member
states to implement a calculation methodology for the integrated energy performance of
buildings by 2006 at the latest (EPBD 2002). In this context, the government of the Brussels
Capital Region by means of the Brussels Institute for the Management of the Environment
(BIM), ordered a technical-economical study on the cost-effectiveness of energy saving
investments in buildings. Aim of the study was to determine the most feasible energy saving
measures, considering the current technologies and their costs. These results should then
serve as scientific basis for the requirements to be set in the Brussels energy performance
regulation for new and retrofitted buildings. The project has been executed in 2005 in
collaboration with the engineering office 3E (www.3e.be). The Division of Building Physics
was responsible for the residential sector, whereas 3E was responsible for the nonresidential sector. Within the project, the methodology, developed in this doctoral research,
has been applied to trade off costs and energy savings of current energy saving
technologies, also taking into account the embodied energy. As no optimisation was needed
in this project, the methodology has been applied without the optimisation module. Details on
the implemented energy saving measures are described in the section below. All building
simulations have been executed with the calculation procedure of the Energy Performance
Regulation of the Flemish Region, further called the EPB (EPB Besluit Bijlage I en II 2005).
The outline of this calculation procedure is briefly presented below. The assumptions on life
cycle inventory and costs for this project are presented in section 7.3 and 7.4. In the last
sections, the results are summarised and discussed. Although the developed methodology is
used for both the residential and non-residential buildings, only the application to the
residential sector is presented in this chapter, as only this part has been executed as part of
this doctoral research. In order to keep the extent of this PhD dissertation under control, only
the most important assumptions and results are presented. Further details and results can be
found in Verbeeck and De Coninck (2005).
135
CHAPTER 7
7.2
Implementation of energy saving measures
7.2.1. Reference buildings
In order to obtain representative results for the Brussels Capital Region, four reference
buildings are selected based on a statistical analysis of the Brussels residential sector.
Because of the almost strictly urban character of the Brussels Capital Region, the reference
buildings are restricted to terraced houses and apartment buildings, thus representing 90%
of the Brussels building stock. For the terraced houses, two dwellings are selected: an old
mansion for the retrofit case and a modern terraced dwelling as a new construction. For the
apartment buildings, distinction is made between a small apartment building of 4 flats and a
large apartment building of 20 flats. The same lay out is used for both new construction and
retrofit. This selection results in 6 different reference cases: three for new construction and
three for retrofit. Table 6.1 presents the reference situation of the dwellings and the
apartment buildings. Annex F gives the plans and façades of the reference buildings.
Reference buildings
Apartment buildings
BIM-project
small
large
modern
(new)
mansion
(retrofit)
1945-1970
1945-1970
-
before 1945
Heated floor area
building
flat
436m²
109m²
1804m²
89m²
173m²
220m²
Heated volume
building
flat
1220m³
305m³
5054m³
249m³
514m³
747m³
Heat loss area
building
flat
549m²
56m²
2164m²
39m²
228m²
324m²
Compactness
building
flat
2.2
5. 5
2.3
6.4
2.3
2.3
Insulation level (Umean
W/m²K)
retrofit
new construction
K151 (U=2.1)
K71 (U=1.0)
K150 (U=2.2)
K72 (U=1.0)
Cavity wall
Non insulated
Cavity wall
2.5cm
insulation
Cavity wall
Non insulated
Cavity wall
2.5cm
insulation
Construction period
Façade
retrofit
new construction
136
Terraced dwellings
K118 (U=1.7)
K69 (U=0.98)
Massive wall
Non insulated
Cavity wall
2cm insulation
BIM-PROJECT
Glazing + frames
retrofit
new construction
Single
Wood
Double
Wood
flat
non insulated
flat
new construction
5cm insulation
Roof
retrofit
Single
Aluminium
w.o.
thermal break.
Double
Double
Aluminium
Aluminium with
with thermal
thermal break
break
Single
Wood
flat
non insulated
sloped
flat
5cm insulation 5cm insulation
sloped + flat
non insulated
Heating
Radiators
Radiators
retrofit
Natural gas
new construction
Natural gas
Natural gas
and fuel
Natural gas
emission
Radiators
Radiators
Energy carrier
Natural gas
and fuel
Natural gas
Heating boiler
Standard
Standard
collective
per flat
new construction High efficiency High efficiency High efficiency
collective
per flat
retrofit
Control system
Hot water production
Standard
Central, room
thermostat
Central, room
thermostat
Central, room
thermostat
Central, room
thermostat
On boiler
On boiler
On boiler
On boiler
Table 7.1: Reference situation for the reference buildings of the BIM-project
7.2.2. Measures on the building envelope
Insulation
In a first step, the impact of insulation, glazing, sun shading and air tightness is analysed.
Insulation measures are taken for the roofs, façade and floor. For each building type, the
insulation thickness in each envelope component varies from zero to a maximum.
Thicknesses rise from zero to the maximum in 2.5 cm steps for the floor and façade and 5
cm steps for the roofs. Table 7.2 gives the maximum applied insulation thickness and the
applied insulation material per envelope component.
137
CHAPTER 7
Insulation
min thickness
[cm]
max thickness
[cm]
step
[cm]
material
λ
[W/mK]
Flat roof
0
20
5
XPS
0.031
Sloped roof
and/or attic floor
0
20
5
MW
0.044
Façade
0
10
2.5
MW
0.044
Floor
0
10
2.5
PUR
0.021
Table 7.2: details on the insulation measures of the BIM-project
Glazing and window frames
Apart from the insulation, the impact of thermally better performing glazing and window
frames is analysed. Tables 7.3 and 7.4 present the different types of glazing and window
frames with their thermal characteristics: for the glazing types, the U-value and the g-value,
being the direct and indirect solar heat gain coefficient and for the window frames, the Uvalue. Aluminium window frames without thermal break are only considered as part of a
reference situation for retrofit. They are not considered for new buildings.
Glazing
U-value
[W/m²K]
g-value (solar factor)
[-]
5.7
0.76
3
0.72
Low e double glazing, air filled
1.8
0.61
Low e double glazing, argon filled
1.3
0.61
1
0.61
0.6
0.50
Single
Standard double
Low e double glazing, krypton filled
Low e triple glazing, argon or krypton filled
Table 7.3: Properties of the glazing types, applied in the BIM-project
Window frames
U-value
[W/m²K]
Wood
1.8
Aluminium without thermal break
5.9
Aluminium with thermal break
2.4
PVC with 2 chambers
2.2
Highly insulating window frame
0.65
Table 7.4: Properties of the window frames
In order to control the summer comfort, the impact of glass area and sun shading is analysed
for some of the combinations of insulation and glazing. For this analysis, the glass area
138
BIM-PROJECT
varies between a minimum and a maximum value, and is combined with manually operating
outer sun shading. The method for assessing the summer comfort is explained in the section
on the building simulation programme.
Air tightness
Also the air tightness of the building is incorporated as a parameter in the first step of the
analysis. The level of air tightness is expressed through the n50-value, being the number of
times per hour the air volume is refreshed by uncontrolled infiltration of outdoor air when
creating a pressure difference of 50 Pa between indoor and outdoor. For the first step, only
two levels of air tightness are considered:
-
the case where no attention is paid to the air tightness: the default value of the
building simulation programme has been taken, which depends on the compactness
C of the building:
-
n50 = 10.5/C +1.5
[1/h]
the level of air tightness when a designed natural ventilation system is planned:
n50 = 3
[1/h]
In the second step, also lower levels of air tightness are considered, in combination with
mechanical ventilation systems.
7.2.3. System-related measures
In the second step, the measures on the building envelope are combined with system-related
measures. A large number of system variables are taken into account, mainly based on the
options available in the building simulation procedure used. Distinction is made between
systems for the dwellings and systems for the apartment buildings.
Dwellings
Space heating
The space heating system in a dwelling can be a local or a central heating system with
different energy carriers. For the local system, gas stoves, direct electric heaters and electric
accumulation heaters are available. For the central heating system there is a choice between
high efficiency or condensing boilers which run on natural gas or fuel, or an electrically driven
heat pump. The heat is assumed to be emitted by high temperature radiators with or without
a rear radiation shield, convectors, low temperature radiators or floor heating. The room
temperature can be controlled by a thermostat or thermostatic valves. The exit temperature
of the water in the boiler can be fixed or variable.
139
CHAPTER 7
Seven possibilities are considered for the hot water production:
(1) a direct water heater on electricity
(2) a direct water heater on gas with a pilot-light
(3) a direct water heater on gas without a pilot-light
(4) a storage tank heated by gas
(5) a storage tank heated by electricity
(6) a storage tank connected to the heating boiler
(7) a storage tank connected to the heat pump
Each space heating system is combined with different systems for hot water production, as
long as the combination is realistic. This way, no space heating on fuel or electricity is
combined with a hot water system on natural gas, as this combination is highly unrealistic.
However, space heating with natural gas or fuel can be combined with hot water production
on electricity.
In case of direct water heaters, two heaters are assumed: one in the bathroom and one in
the kitchen. The same assumption is made in case of an electrically heated storage tank. In
case of a storage tank on gas, on the boiler or on a heat pump, only one tank per dwelling is
assumed. These assumptions do not affect the production efficiency, but have an impact on
the distribution losses as shorter pipe lengths can be assumed in case of two separate hot
water systems.
Concerning the distribution losses, the building simulation programme distinguishes between
pipes lying inside and outside the heated volume. However, pipe insulation can only be
considered within the programme for circulation pipes in which the hot water is continuously
circulating. Such system is normally only applied in apartment buildings with a central hot
water production system. So no pipe insulation is considered for the terraced dwellings, but
most pipes lie inside the heated volume and in that case, 100% recovery of the distribution
losses is assumed.
Solar collectors
Each hot water system can be combined with a solar collector. Per dwelling only one system
is considered, being a system with a collector area of 4m², placed on the south-oriented
sloped roof, as both dwellings are north-south oriented. For the mansion, the roof slope is
34°, for the terraced house, the slope is 37°.
140
BIM-PROJECT
Ventilation
In the first step, a natural ventilation system is assumed, whereas in the second step four
extra ventilation systems are simulated with the appropriate level of air tightness. Each
ventilation system is combined with each system for space heating and hot water production.
This way, synergy of different measures can be taken into account. Basic assumptions for
the ventilation systems are a good design and installation of the system. Current, energy
saving technologies are selected, however, without selecting always the most energy saving
option. This assumption is adopted to create realistic energy savings and to avoid
overestimation of the savings on ventilation losses. For all ventilation systems, the ventilation
flows are calculated per dwelling according to the ruling ventilation standards (NBN D50-001
1991). The following ventilation systems are considered:
-
natural ventilation with self regulating grids
-
mechanical extraction ventilation with natural supply through self regulating grids and
extraction with a DC ventilator; level of air tightness n50 = 3/h
-
mechanical supply and extraction, with a DC ventilator and without heat recovery;
level of air tightness n50 = 1/h
-
mechanical supply and extraction, with a DC ventilator, a level of air tightness n50 =
1/h and with a heat recovery unit with 70% recovery efficiency
Photovoltaic systems
Each combination of space heating, hot water production and ventilation system can be
combined with a photovoltaic module. In the building simulation programme EPB, different
PV-systems can be selected:
-
integrated in the roof or free standing
-
with a central transformer or with an AC module per panel
-
4 types of PV-cells: mono crystalline, poly crystalline, amorphous with 1 or 2 junctions
Per dwelling, only one system is considered, being 10m² poly crystalline PV-cells with a
central transformer, integrated in the south-oriented sloped roof, as both terraced dwellings
are north-south oriented. This is the most commonly applied PV-system in Belgium. The
produced electricity is taken into account in the total primary energy consumption.
141
CHAPTER 7
Apartment buildings
The system-related variants for the apartment buildings are mainly the same as for the
dwellings, except for some differences that are presented below.
Space heating
For the small apartment building, there is the option between individual heating (a boiler per
flat) and collective heating (a boiler per apartment building). For the large apartment building,
only a collective heating system is considered, as the flats are too small to install a boiler per
flat.
The option for floor heating is not considered for the apartment buildings, as it is at the same
time unrealistic and difficult to implement in the EPB software.
For the apartment buildings, an extra option for space heating is considered, being
cogeneration of heat and power (CHP, gas engine type), in combination with radiators or
convectors and controlled by thermostatic valves. In case of CHP, the option of PV-cells is
no longer considered, as the gas engine itself produces electricity. The produced electricity is
taken into account in the total primary energy consumption.
In case of CHP, the option for hot water production with CHP is included.
The combination with solar collectors is only considered for a centralised hot water
production system. The collector system is installed on a frame on the flat roof, according to
the ideal orientation (south) and slope (34°). The collector area is 12m² for the small
apartment building and 40m² for the large building.
Ventilation
For the ventilation systems, the same systems as well as the same assumptions have been
made as for the dwellings. However, in case of mechanical supply and extraction, a
collective system is assumed.
As for the dwellings, each combination of space heating, hot water production and ventilation
system can be combined with a photovoltaic module, except in case of CHP. Per apartment
building, only one collective system is considered, being poly crystalline PV-cells with a
central transformer, mounted on a freestanding frame on the flat roof, south oriented with a
slope of 34°. For the small building 30m² is assumed, for the large building 60m².
142
BIM-PROJECT
7.2.4. Building simulation programme
The energy consumption has been simulated with the calculation procedure for the Flemish
Energy Performance Regulation EPB (EPB Besluit Bijlage I en II 2005), as at the time of the
project, no calculation procedure for the Brussels Capital Region was available yet. The
procedure is mainly based on the EN ISO 13790 (2004). Firstly, both the net heat demand
and the gross heat demand, which includes the system efficiency for distribution, emission
and control of the heat, are calculated as well as the end energy consumption for space
heating and hot water production. In the latter, not only the production efficiency and the
energy consumption of pumps and fans, but also the contribution of solar collectors is taken
into account. The above procedure also calculates the primary energy consumption by firstly
adding the fuel or gas consumption to the primary energy used for electricity production for
the Belgian plant mix, and then subtracting the primary energy equivalent to the electricity
production of PV-panels or CHP, if present. The climate data are monthly average values for
the Test Reference Year of Brussels, Belgium. The EPB does not take into account the
electricity consumption for household and lighting, as it is considered to be too dependent of
the occupants’ behaviour.
In the cool Belgian climate, summer overheating can be avoided in residential buildings by a
good building design. Therefore, the EPB evaluates the risk for overheating and thus, the risk
that an active cooling system will be installed. In case the risk for overheating exceeds a
certain threshold, a penalisation is incorporated in the energy consumption by assuming a
fictitious cooling load proportional to the violation of the threshold. In case the risk for
overheating exceeds a maximum value, the design is no longer accepted and needs to be
adapted.
Concerning the performance of renewable energy systems, the EPB contains modules to
calculate the energy contribution of solar collectors and PV-panels.
Solar collectors can be applied for only sanitary hot water or for both space heating and
sanitary hot water. Default performance characteristics for the collector are incorporated in
the EPB. They are mainly those of a standard flat plate collector in the Belgian climate. The
parameters which remain to be specified are the orientation, the slope and the collector area.
At the time of the project, the calculation procedure of the Flemish EPB was already
available, however, without the official EPB-software. Therefore, the EPB calculation
procedure has been incorporated in a Visual Basic software package that has been
developed at the Division of Building Physics.
143
CHAPTER 7
7.3
The life cycle inventory was of minor importance for the customer within the BIM-project.
However, it is added as extra information to the results to counter remarks on the fact that
the embodied energy of energy saving measures might be too high compared to the energy
savings they realise. For this project, only the extra embodied energy is calculated, being the
energy needed for the production of the energy saving measures. The energy payback time
(EPBT) is calculated by dividing the extra embodied energy of the energy saving measures
by the annual primary energy savings realised with the measures.
7.4
Cost assumptions
The economic evaluation criteria are described in section 6.2. The cost-benefit analysis of
the BIM-project does not consider the total building cost, but only the costs for the energy
saving measures. For the first step, only the total investment cost for the energy saving
measures on the building envelope are considered, as well as the extra investment cost
compared to the reference case. For the second step, also the investment cost for heating
and ventilation systems are taken into account as well as the energy prices and price
evolution. This way, not only the total and net present value are calculated for the different
combinations of energy saving measures, but also the static and dynamic payback time and
the internal rate of return. The assumptions on the energy prices are presented in section
6.3.3. The utilisation period is assumed to be 40 years. As said before, this is not considered
to be the life span of the building, but the period the building is used by one generation. For
measures with a life span shorter than 40 years, re-investment is considered at the same
cost as the initial investment. No assumptions are made on what happens with the building
after 40 years. No secondary costs are assumed in the BIM-project, as the maximum
assumed insulation thicknesses remain feasible within the traditional building practice.
Neither is the rebound effect taken into account.
7.5
Results
7.5.1. Energy saving measures on the building envelope
Figure 7.1 presents the extra investment cost and the annual primary energy consumption
for all combinations of energy saving measures on the envelope of the new terraced dwelling
(in grey). The large black dot at the right represents the reference case. The black squares
are the Pareto optima, representing the trade-off between investment cost and primary
energy consumption. These combinations realise certain annual primary energy consumption
144
BIM-PROJECT
at a minimal extra investment. Combinations that are not on the trade-off curve, might have
the same primary energy consumption, but at higher extra investment cost. Three separate
clouds of solutions can be distinguished. The lower cloud represents all combinations of
insulation, glazing and air tightness, combined with the cheapest window frames. The middle
cloud gives the variants with the expensive highly insulating window frames, whereas the
upper cloud represents the variants with both the very expensive low e triple glazing and the
highly insulating window frames.
New terraced dwelling: measures on the building envelope
18000
16000
extra investment [€]
14000
all results
Pareto optima
reference
12000
10000
8000
6000
4000
2000
0
10000
-2000
12000
14000
16000
18000
20000
22000
24000
26000
annual primary energy consumption [kWh/a]
Figure 7.1: New terraced dwelling: extra investment cost vs. annual primary energy
consumption for all combinations of energy saving measures on the building envelope
Figure 7.1 shows that some solutions perform energetically better than the reference case,
but at a lower investment cost (a negative extra investment cost). These are variants with
PVC window frames. These frames are cheaper, but perform thermally better than the
wooden window frames of the reference case. However, in practice the choice for the
material for window frames is mainly determined by cost and esthetical considerations, more
than by considerations on energy performance.
Figure 7.2 shows the extra investment per m² heated floor area as a function of the annual
energy savings. The solutions of the trade-off curve are presented as dark grey squares. In
addition, some solutions with extra insulation thickness are shown (light grey squares).
Finally, in black at the left hand side, variants are presented in which one single measure is
applied to the maximum value, while all other parameters have their reference value. So,
‘SR’ represents the variant with maximum insulation (20cm) in the sloped roof pitches only,
145
CHAPTER 7
whereas for the façade, floor, windows and air tightness, the reference values are
maintained. ‘FA’ represents the variant with maximum insulation in the façade, ‘FL’ maximum
insulation in the floor, ‘AF’, in the attic floor, ‘n50’ the best air tightness and ‘window’ the
variant with the best glazing and window frames. These results show that applying one single
measure has a limited impact on the energy savings, whereas the synergy of measures can
realise large energy savings, even at a limited extra investment cost.
New terraced dweling: measures on the building envelope
extra investment/m² heated floor area
[€/m²]
120
100
80
60
40
Pareto optima
reference
20
FL n50 AFFA
window
SR
0
0
2000
4000
extra insulation
6000
8000
10000
12000
14000
-20
annual primary energy savings [kWh/a]
Figure 7.2: New terraced dwelling: extra investment cost per m² heated floor area vs. annual
primary energy savings for all optimal combinations of energy saving measures on the
building envelope
For the new terraced dwelling, up to 5000 kWh or 20% of the annual primary energy
consumption can be saved at a negative or very low extra investment cost by applying a
somewhat higher insulation thickness in roof, façade or attic floor than in the reference case
and combining it with well performing PVC window frames. By these limited measures the
legal insulation standard K45 (Umean = 0.63 W/m²K) can be reached at a limited extra cost.
More than 9000 kWh or 38% of the annual primary energy consumption can be saved at an
extra investment cost of less than 20€ /m² heated floor area. This can be realised with a
combination of 15-20cm insulation in the sloped roof pitches and on the attic floor, 6-8cm
insulation in the façade, 2.5cm in the floor and low e double glazing with a U-value of 1-1.3
W/m²K. This combination represents an overall Umean of 0.4 W/m²K.
Comparison of the upper dark grey dots and the light grey dots show that extra insulation
realises higher energy savings at a lower cost than the very expensive low e triple glazing
and highly insulating window frames, as applied in passive houses. However, this should be
146
BIM-PROJECT
nuanced. In extremely well insulated dwellings, the summer comfort becomes a crucial focus
and reduction of the glass area and application of outer sun shading is most often
indispensable. This secondary effect of high insulation thicknesses complicates the analysis
of the investment cost. The impact of glass area in extremely low energy dwellings will be
discussed in chapter 8 on the EL²EP-project.
Figures 7.3 and 7.4 present the extra investment per m² heated floor area as a function of
the annual primary energy savings per m² heated floor area for new construction (fig.7.3) and
retrofit (fig.7.4). These figures show that if the results for the different building types are
presented per m² heated floor area, similar results are achieved for all buildings. This means
that similar energy savings can be realised at similar extra investment cost for all buildings.
Only when sun shading is needed to control the summer comfort, the investment cost clearly
increases. Obviously, as comparison of figure 7.3 and 7.4 shows, higher energy savings can
be realised in old energy devouring buildings, but this asks for higher investment costs.
The variants of figure 7.3 and 7.4 are also the variants for which all installation variants have
been simulated. More details on the energy saving measures, the overall Umean-value and the
results for extra investment cost and annual primary energy savings for all variants can be
found in Verbeeck and De Coninck (2005).
New construction
extra investment/m² heated floor
area
120
100
80
60
40
20
0
-20
0
20
40
60
80
100
-20
annual primary energy saving/m² heated floor area [kWh/m²]
large apartment
small apartment
terraced house
Figure 7.3: Extra investment cost per m² heated floor area as a function of the annual
primary energy savings per m² heated floor area for all new reference buildings of the BIMproject
147
CHAPTER 7
Retrofit
investment/m² heated floor area
200
180
160
140
120
100
80
60
40
20
0
0
50
100
150
200
250
annual primary energy savings/m² heated floor area [kWh/m²]
flat large gas
flat large fuel
flat small
mansion gas
mansion fuel
Figure 7.4: Extra investment cost per m² heated floor area as a function of the annual
primary energy savings per m² heated floor area for all retrofitted reference buildings of the
BIM-project
7.5.2. Energy saving measures on the overall building
For the assessment of the impact of energy saving measures on the whole building, the
concept of Pareto-front or trade-off curve is adopted, as described in section 1.3.2. By
comparing the results for two criteria (mostly an energy- and a cost-criterion), the optimal
trade-off curve can be determined. However, this curve might depend on the adopted
assessment criteria. What the optimal trade-off curve consists of and the dependence on the
assessment criteria is analysed in this section. Therefore, the results are firstly analysed from
the point of view of extra investment cost (figure 7.5) and then, from the point of view of total
present value (figure 7.6). For reason of readability, the analysis is limited to results of the
new terraced house and the results are presented as graphs. Unless specified otherwise, all
graphs are valid for the middle energy price scenario (+2.1%). Discussion of the results with
relation to the impact of price evolutions, new construction versus retrofit, etc. is presented in
section 7.6. More results and figures for all reference buildings can be found in Verbeeck and
De Coninck (2005).
Extra investment versus total primary energy consumption
Firstly, the results are analysed from the point of view of extra investment cost. Figure 7.5
presents therefore all results for the new terraced house for the criteria ‘extra investment
148
BIM-PROJECT
cost’ and ‘total primary energy consumption over 40 years’ in light grey. The trade-off curve
for these criteria, representing the optimal variants from the point of view of extra investment
cost, is given in dark grey (PE-INV). The black triangle represents the reference case.
Obviously, the results on the right hand side of the reference case are of no interest, as they
result in higher energy consumption. These results represent variants with low thermal
quality combined with electrical heaters. From the point of view of primary energy
consumption, these combinations perform worse than the reference case with a gas boiler.
The optimal solutions with lower energy consumption represent in the order of decreasing
energy consumption and increasing extra investment cost:
1. increase of the thermal quality of the building up to Umean = 0.5 W/m²K in combination
with a high efficiency boiler (PE ≈ 2000 GJ);
2. additionally better air tightness and application of a condensing instead of a high
efficiency boiler (PE ≈ 1600 GJ);
3. application of extra energy saving, but expensive measures, such as a solar collector,
heat pump, mechanical ventilation with heat recovery or higher thermal quality (Umean
= 0.4 W/m²K) (PE ≈ 1000 -1500 GJ);
4. the ultimate step to achieve primary energy consumption < 1000 GJ is by combining
two or more extra, expensive measures, such as a heat pump with a solar collector or
a heat pump with highly insulating glazing and window frames.
New terraced dwelling:
extra investment vs. total primary energy consumption
60000
Extra investment [€]
50000
All results
Reference
PE-INV
40000
30000
20000
10000
0
-10000
0
1000
2000
3000
4000
5000
6000
7000
Total primary energy consumption [GJ]
Figure 7.5: New terraced dwelling: trade-off curve for extra investment cost (INV) and total
primary energy consumption (PE) over 40 years (in dark grey); the black triangle is the
reference situation
149
CHAPTER 7
Total present value versus total primary energy consumption
When the same results are analysed from the point of view of total present value, a different
trade-off curve appears. This can be seen in figure 7.6, that contains the same variants, but
now presented for the criteria ‘total present value’ and ‘total primary energy consumption
over 40 years’. The trade-off curve is given in black (PE-TPV). By way of comparison, also
the variants from the trade-off curve of figure 7.5 are given in dark grey. This figure shows
that in contrast to the extra investment cost, that reaches a minimum at 5500 GJ, the total
present value reaches a minimum around 2000 GJ for the new terraced dwelling. This
optimum represents a variant with an overall Umean = 0.5 W/m²K (15cm insulation in the roof
and attic floor, 6cm in the façade, Uglass = 1 W/m²K, Uframe = 2.2 W/m²K), a high efficiency
boiler with variable water temperature and radiators designed for high temperature and
domestic hot water production with a direct water heater on gas without pilot-light. With this
solution, the primary energy consumption will decrease with nearly 40% compared to the
reference with at the same time, a decrease of the total present value of 19%. Further
decrease of the primary energy consumption is possible, but at higher total present value
than the minimum. A decrease up to 1500 GJ is possible at lower TPV than the reference.
For solutions with primary energy consumption < 1500 GJ, the TPV will be even higher than
the reference case.
New terraced dwelling:
total present value vs. total primary energy consumption
140000
Total present value [€]
120000
100000
80000
60000
All results
Reference
PE-INV
PE-TPV
40000
20000
0
0
1000
2000
3000
4000
5000
6000
7000
Figure 7.6: New terraced dwelling: these are the same variants as in figure 7.5, but now
presented for total present value (TPV) and total primary energy consumption (PE) over 40
years. The trade-off curve from the point of view of TPV is given in black, the trade-off curve
from the point of view of extra investment cost in dark grey. The black triangle represents the
reference case.
150
BIM-PROJECT
Comparison of both cost criteria, extra investment cost and TPV, clearly shows that, both at
short and at long term, the total present value is the best basis for selection of energy saving
measures as the solution with minimum TPV represents the economically most viable
solution. Furthermore, in case there is interest to realise higher energy savings than the
economic optimum, the trade-off curve of TPV and total primary energy consumption gives
the hierarchy of most cost effective energy saving measures.
7.5.3. Extra embodied energy due to energy saving measures
The Energy Payback Time (EPBT) is defined as the proportion of the embodied energy for
energy saving measures to the annual energy savings they realise. Figure 7.7 presents the
EPBT for the optimal measures for the building envelope for the new and retrofitted terraced
dwelling, whereas figure 7.8 gives the EPBT for the optimal combinations of envelope- and
installation-related energy saving measures.
Terraced dwellings: measures on the envelope
120000
extra embodied energy [MJ]
new dwelling
old mansion
EPBT=1 year
100000
EPBT=1.5 year
80000
60000
40000
20000
0
0
20000
40000
60000
80000
100000
120000
annual primary energy saving [MJ/a]
Figure 7.7: The extra embodied energy of energy saving measures on the building envelope
as a function of the annual primary energy savings for the new terraced house (black) and
the old mansion (grey). The straight lines represent the energy payback time (EPBT) of 1
year (dotted line) and 1.5 year (full line)
151
CHAPTER 7
Terraced dwellings: measures on envelope and installation
180000
new dwelling
old mansion
EPBT=1 year
EPBT=1.5 year
Extra embodied energy [MJ]
160000
140000
120000
100000
80000
60000
40000
20000
0
0
20000
40000
60000
80000 100000 120000 140000 160000 180000
Annual primary energy saving [MJ/a]
Figure 7.8: The extra embodied energy of energy saving measures on the whole building as
a function of the annual primary energy savings for the new terraced house (black) and the
old mansion (grey). The straight lines represent the energy payback time (EPBT) of 1 year
(dotted line) and 1.5 year (full line)
Both figures show that the EPBT for all studied cases is less than 1.5 year. This means that
the energy, needed for production of the energy saving measures, is regained in less than
1.5 year by the energy savings realised by these measures. In case of retrofit, the EPBT is in
most cases even less than 1 year, due to the much higher energy savings. This is obvious as
the first centimetres of insulation, already present in new dwellings, are the most effective.
Only for highly extended measures, such as the combination of large insulation thicknesses
with a heat pump and a solar collector, the EPBT is 1.5 year or more. Similar results have
been found for the apartment buildings.
7.6
Discussion
7.6.1. Impact of the scenarios with varying economic parameters
Scenarios for the discount rate and the energy price evolution mainly have an impact on the
total present value. The extra investment cost only depends on the discount rate for the reinvestment costs, not for the initial investment cost, whereas the primary energy consumption
is completely independent of the varying economic parameters within the assumptions of the
methodology. Therefore, the impact of the scenarios is discussed in this section on the basis
of trade-off curves for primary energy consumption and total present value.
152
BIM-PROJECT
Discount rate and VAT
On demand of the customer of the BIM project, 3 scenarios have been considered for the
discount rate for energy saving measures in the residential sector:
-
4% for the public sector, 21% VAT for investments in new construction
-
4.5% for private builders, 21% VAT for investments in new construction
-
6.5% for building contractors, 0% VAT for investments in new construction
New terraced dwelling: impact of discount rate
120000
100000
rate=4%
rate=4.5%
rate=6.5%
80000
60000
40000
20000
0
0
500
1000
1500
2000
2500
Figure 7.9: New terraced house: impact of the discount rate on the trade-off curve between
total present value and total primary energy consumption: three scenarios for the discount
rate are considered
As figure 7.9 shows, the different scenarios for the discount rate and VAT result in a vertical
shift of the total present value, but the optimal solutions that correspond to these values
remain identical. The higher the discount rate and the lower the VAT, the lower is the
investment cost and the total present value. The energy cost is not affected by these
scenarios, as regardless of the builder (public, private or professional) all residential buildings
are assumed to be occupied by households.
Energy price evolutions
Comparison of the three scenarios for energy price evolution shows that the impact of the
energy price decreases with decreasing primary energy consumption (figure 7.10, interest
rate = 4.5%). This is obvious as the contribution of the energy costs to the total present value
decreases with decreasing energy consumption.
153
CHAPTER 7
More important, however, is the conclusion that the economic optimum (minimal TPV) as
well as the hierarchy of most cost effective energy saving measures is almost independent of
the evolution of the energy price and remains the same as described in section 7.5.2. Only
the system for domestic hot water production slightly differs at the lowest energy price
scenario: the extra point at the right represents a direct heater on gas with pilot-light.
In case of higher energy prices, more extremely energy saving combinations will become
viable. However, for extremely low energy buildings the initial investment cost is so high that
the energy cost becomes of minor importance for the total present value as for these
buildings the initial investment cost strongly overweighs the discounted energy cost over the
utilisation period. This will further be discussed in chapter 8.
New terraced dwelling: impact of energy price scenarios
120000
100000
PE-TPV energy + 0%
PE-TPV energy + 2.1%
PE-TPV energy + 4.3%
80000
60000
40000
20000
0
800
1000
1200
1400
1600
1800
2000
2200
Figure 7.10: New terraced house: impact of the energy price scenarios on the trade-off
curve between total present value and total primary energy consumption: three scenarios for
the evolution of the energy price are considered. The discount rate = 4.5%
7.6.2. New construction versus retrofit
To compare the results for new construction and renovation, the results for the apartment
buildings are analysed, because for these buildings, both new construction and renovation
are considered for the same geometry. Figure 7.11 presents the results for the trade-off of
total present value and total primary energy consumption for new construction (dark grey)
and renovation (light grey) of the small apartment building.
154
BIM-PROJECT
Obviously, the reference situation strongly differs between new construction and renovation.
In case of the small apartment building, the primary energy consumption for the old building
is twice as high as for the new building. The optimal solutions, however, are almost identical
for both cases. To reduce the primary energy consumption to the range of 4000 - 6000 GJ,
the combination of extensive insulation (Umean = 0.3 W/m²K) with a collective high efficiency
or condensing boiler (PE > 5000 GJ) or with individual high efficiency boilers (PE < 4800 GJ)
is most recommended and this is valid for both new construction and renovation.
Small apartment building: new construction vs. renovation
300000
250000
200000
150000
100000
Ref new construction
Ref renovation
new construction
renovation
50000
0
0
2000
4000
6000
8000
10000 12000 14000 16000 18000 20000
Figure 7.11: Small apartment building: comparison of the trade-off curve between total
present value and total primary energy consumption for the case of new construction (dark
grey) and the retrofit case (light grey). The reference situations for both cases are given in
black.
For the retrofit case, also solutions with PE ≈ 2000 GJ are achieved. These results represent
variants with a heat pump, but without insulation of the circulation pipes. This choice can be
argued for retrofit because of the difficulty of installing the pipe insulation, but this is not
restrained as an option for new construction. However, of more importance is the feasibility
and cost of installing a heat pump, which, in case of retrofitting an apartment building, is
highly uncertain. So these results should be treated with caution.
A general drawback of retrofit is that unexpected situations might occur, resulting in a more
uncertain investment cost than in new construction. But, on the other hand, most energy
saving measures in case of retrofit can be applied independently from each other and even
spread in time, if wanted. This approach is an advantage of retrofit that is more difficult to
apply in case of new construction.
155
CHAPTER 7
7.6.3. Position of alternative technologies against optimal solutions
Production systems for domestic hot water
In nearly all optimal solutions, the domestic hot water system consists of a direct heater on
gas without pilot-light. In order to analyse the impact of the domestic hot water system on the
overall energy consumption and cost, figure 7.12 presents for some variants of the new
terraced dwelling the results for all domestic hot water systems and compares them with the
optimal solutions on the trade-off curve (dark grey triangles). All other parameters remain
constant. As a reference the whole trade off curve is given (light grey crosses).
Firstly, the difference between direct heaters with and without pilot-light is very small. The
other systems combine higher energy consumption with higher TPV. The worst results are
achieved with systems with storage tank due to the higher heat losses. However, in practice,
the choice for one system or the other will mainly be determined by considerations on
investment cost, available space and comfort of use.
Production systems for domestic hot water
PE-TPV
direct gas - pilot light
direct gas + pilot light
direct electricity
storage tank gas
storage tank electricity
120000
100000
80000
60000
40000
20000
0
0
500
1000
1500
2000
2500
Figure 7.12: New terraced house: position of the different production systems for domestic
hot water to the trade-off curve for total present value and total primary energy consumption.
The direct water heater on gas without pilot light is part of all the optimal solutions.
Furthermore results for two other types of direct water heater and two systems with storage
tank are shown.
In figure 7.12 no combinations with solar collectors are considered. The impact of solar
collectors is analysed separately in figure 7.13 that presents the trade-off curve for the new
terraced dwelling with and without solar collectors. The grey triangles represent the trade-off
curve for all solutions without solar collector, whereas the black dots represent the trade-off
for all solutions with solar collector. From this figure, it appears that solar collectors form part
156
BIM-PROJECT
of the optimal solutions, but only after application of measures such as insulation and better
performing heating system. They do contribute to lower energy consumption, but with a
significant impact on the total present value, resulting in a higher TPV than the reference
case. They should be considered as the ‘cherry on the cake’ for those who are willing to
invest more in energy saving measures than what is economically viable.
Position of solar collectors
120000
100000
optima - solar collectors
optima + solar collectors
reference
80000
60000
40000
20000
0
0
500
1000
1500
2000
2500
3000
3500
4000
Figure 7.13: New terraced house: position of solar collectors within the optimal solutions of
the trade-off curve for total present value and total primary energy consumption. Solutions
without solar collector are given in grey, solutions with solar collector in black. The reference
situation for the new terraced house is given as a black triangle.
Since PV-systems are building-independent, without impact on the energy consumption for
space heating or domestic hot water, the contribution of PV-systems can be calculated
separately from the building simulation and subtracted from the final energy consumption.
Figure 7.14 presents the trade-off curve for all variants with and without taking into account
PV-systems. The black blocks represent the trade-off curve for all solutions without PVsystem, whereas the grey triangles represent the trade-off curve for all solutions with PVsystems. Between 1600 and 2000 GJ only solutions without PV are present. These solutions
are all well insulated variants (Umean = 0.5 W/m²K) with a high efficiency or condensing boiler.
From 1600 GJ downwards, both curves diverge, with the lowest energy consumption for
variants with a PV-system. In fact, the grey curve contains exactly the same variants as the
black curve, but then combined with a PV-system. By analogy with the solar collectors, it
appears that also PV-systems form part of the optimal solutions, but only after application of
157
CHAPTER 7
measures such as insulation and better performing heating system. They do contribute to
lower energy consumption, but with a TPV that is higher than for the reference case.
New terraced dwelling: impact of PV-systems
120000
100000
Reference
optima - PV
optima + PV
80000
60000
40000
20000
0
0
500
1000
1500
2000
2500
3000
3500
4000
Figure 7.14: New terraced house: impact of PV-systems on the trade-off curve for total
present value and total primary energy consumption. Optimal solutions without PV-systems
are given in black, optimal solutions when PV-systems are also taken into account are given
in grey. The reference situation for the new terraced house is given as black triangle.
7.7
Conclusions
The insulation standard introduced in Flanders in 1992, set an upper limit for the insulation
level of K55, corresponding to an overall Umean = 0.75-0.8 W/m²K for the reference buildings
of the BIM-project. Since January 1st 2006, this insulation standard is strengthened up to
K45, corresponding to an overall Umean = 0.6-0.7 W/m²K for the Brussels reference buildings.
However, the SENVIVV study (1998) showed that in practice the average insulation level of
new buildings in Flanders is K70 or higher, thus not complying at all with the legal standard.
The same results were assumed to be found in the Brussels Region. Therefore, the K70
insulation level was the starting point for the new reference buildings of the BIM-project,
corresponding to an overall Umean = 1.0 W/m²K for these buildings.
The analysis of all combinations of envelope- and installation-related energy saving
measures shows that the economic optimum corresponds to an insulation level K25-K30
(Umean = 0.35-0.5 W/m²K for the Brussels reference buildings), thus much lower than the legal
standard K45. The largest insulation thickness is applied in the roof (15-20cm), but also in
158
BIM-PROJECT
the façade (with 6-10cm) and in the floor (up to 10cm) higher thicknesses than usual can be
applied within the economic optimum, in combination with low emission glazing with a Uvalue ≈ 1.0 W/m²K and a window frame with a U-value ≈ 2.0 W/m²K.
For the heating systems, a high efficiency boiler with variable water temperature combined
with radiators is sufficient. To complete this economic optimal solution, the insulation level
and heating system should be accompanied with a good air tightness of the building
envelope and a well designed natural ventilation system. All these measures result in an
energy performance level of ± E60, according to the Flemish EPB (EPB Besluit Bijlage I en II
2005). This is much lower than the current legal requirement of E100.
In case lower primary energy consumption needs to be reached, the application of a
condensing boiler or a heat pump is the next step to take. A heat pump performs better than
a condensing boiler, but is much more expensive. Further steps to increase the energy
performance of the building are possible by means of a balanced ventilation system with heat
recovery, a solar collector for domestic hot water or a grid-coupled photovoltaic system.
These measures improve the energy performance of the building, but at a high price, far
beyond the economic optimum. Two remarks need to be added on these final steps:
-
To effectively limit the ventilation losses by means of a mechanical ventilation
system with heat recovery, not only a good air tightness of the building envelope
and the ventilation pipes is indispensable, but also an adaptation of the
occupants’ behaviour. Sleeping in winter with open windows is no longer an
option. For some occupants, this system is a bridge too far.
-
If there is the motivation, but not yet the budget to install a solar system, the
installation can be postponed, as both a PV-system and a solar collector can
easily be installed afterwards.
In a nutshell, the most logical hierarchy for energy saving measures in dwellings is:
1. Invest in a good insulation level with good air tightness and a well designed
natural ventilation system. The economical optimal insulation level K25-K30
(Umean ≈ 0.3-0.4 W/m²K) lies, at least in Belgium, far beneath the legal
insulation requirement K45 (Umean ≈ 0.6-0.7 W/m²K).
2. Select a well performing heating system, at least a high efficiency or
condensing boiler. If the budget is available, a heat pump is a good
alternative: better performing, but at a higher price.
3. Finally, if for any reason, further steps to decrease the energy consumption
are wanted and the budget is available, there are the options for a balanced
ventilation system with heat recovery and/or a solar driven system (solar
collector or PV-system). However, one must realise that, although these
159
CHAPTER 7
measures improve the energy performance, they are not cost effective and
far beyond the economic optimum.
This guideline is valid for both new construction and renovation. Some insulation measures,
such as insulation of the façade or the floor, are less easy to implement in case of retrofit or
at much higher price than for new construction. So the economic viable insulation level will
normally be somewhat higher for retrofit than for new buildings, but the hierarchy of
measures remains identical and is independent of energy price evolutions and discount rate.
160
EL²EP-PROJECT
CHAPTER 8.
8.1
EL²EP-project
Introduction
The EL²EP-project is a research project funded by the Flemish government through IWT, the
Institute for the Promotion of Innovation by Science and Technology in Flanders
(www.iwt.be). The basic objective of the project was the development of extremely low
energy and low pollution dwellings, in short “el²ep-dwellings”, by tracing out the best passive
(building envelope) and active (installations) energy saving measures based upon life cycle
optimisation of costs, energy and emissions.
Similar to the BIM-project, energy saving measures are applied to the building envelope and
the heating system in two steps. In a first step, the impact of insulation, glazing, sun shading
and air tightness is analysed, whereas in the second step, the measures on the building
envelope are combined with system-related measures. However, different from the BIMproject and crucial for the EL²EP-project is the optimisation process that is included. Where
in the BIM-project all possible combinations have been calculated in order to analyse the
cost-effectiveness of different combinations of measures, the aim in the EL²EP-project is to
develop extremely low energy dwellings that are optimised for costs, energy and emissions.
The optimisation is performed in two steps. Firstly, the net energy demand is reduced to a
minimum by optimising the energy saving measures on the building envelope. In the second
step, the focus is shifted towards the most appropriate technologies to meet this very low
energy demand in an optimal way.
The EL²EP-project has been executed from January 2002 till December 2006 in a
collaboration of several research divisions from the Katholieke Universiteit Leuven and the
Universiteit Gent. Due to the extent of the project, the research was subdivided in subtasks,
executed by different researchers. The global methodology, explained in this PhD
dissertation, was the core task of the project in which the outcome of the other subtasks had
to be integrated. However, it is not feasible within the extent of this dissertation to present
and discuss all assumptions and results of all subtasks of the EL²EP-project. Therefore, the
main focus of this chapter will be put on the presentation and discussion of the final outcome
of the EL²EP-project, being the final concepts for optimised extremely low energy dwellings
and their position to the economic optimal solutions.
The evolutionary multi-objective optimisation process is an important component within the
EL²EP-project that has not been applied within the BIM-project. The model itself has already
been explained in chapter 4, but in this chapter attention will be paid to the strengths and
weaknesses of the optimisation methodology for this project.
161
CHAPTER 8
Finally the concepts developed within the project will be compared with existing concepts for
extremely low energy dwellings, such as passive houses and zero energy houses.
8.2
Implementation of energy saving measures
The approach adopted within the EL²EP-project is strongly similar to what is already
explained for the BIM-project, especially for implementing the energy saving measures.
Therefore, the assumptions will be presented in brief, with an emphasis on what differs from
the BIM-project.
8.2.1. Reference buildings
In order to obtain representative results for the Belgian building practice, a number of
dwellings are designed following the statistical average of the Belgian residential sector
(Verbeeck and Hens 2002). The sociological reference is a family of four persons living in a
middle class house with three bedrooms and a total floor area of 140-175m². To include the
compactness, being the proportion of the heated volume to the heat loss area, five typologies
are defined, resulting in five different dwellings:
-
a terraced house
-
a semi-detached house
-
two detached houses: one with a simple square plan and one ‘architectural’ dwelling
with a fragmented plan
-
an apartment flat
With the compactness varying from 0.85m for the architectural dwelling to 4.17m for the flat
and a constant heated air volume of ca. 470 m³ for all dwellings, the heat loss area varies
over the dwellings from 105m² to 560m². The non insulated version of the selected dwellings
is defined as the reference situation. Although the glass area is one of the variables in this
project, the reference version has a fixed average ratio of glass to floor area per type of
room, being 24.5% for the living room, 18% for the kitchen, 17% for the office, 12% for the
bedrooms and 7% for the bathroom. Annex G gives the plans, sections and façades of the
selected buildings.
8.2.2. Measures on the building envelope
Similar to the BIM-project, firstly the impact of insulation, glazing, sun shading and air
tightness is analysed. Insulation measures are applied to the roof, façade and floor.
162
EL²EP-PROJECT
However, differently from the BIM-project, the insulation material is not fixed per envelope
component and insulation thicknesses can vary in steps of 1cm and up to larger thicknesses.
Table 8.1 gives the maximum insulation thickness and the applicable insulation materials per
envelope component. In contrast to the BIM-project, where per dwelling only one
constructional type has been considered, the EL²EP-project includes thermal capacity as a
parameter by incorporating three construction types per selected dwelling:
-
a building type with cavity wall and massive inner walls and floors
-
a building type with a brick façade with external insulation and with massive inner
walls and floors
-
a complete lightweight wood frame construction
Envelope component
Maximum
thickness [cm]
Applicable materials*
Flat roof
30
MW, PUR, CG
Sloped roof pitches
40
MW, PUR, EPS,XPS, CF
Attic floor
40
MW, PUR, EPS, XPS
Cavity wall or external insulation
30
MW, PUR, EPS, XPS
30
MW, CF
15
MW, PUR, EPS, XPS
Façade
Floor
*
: MW=mineral wool, PUR=polyurethane, CG=cellular glass, EPS=expanded polystyrene,
XPS=extruded polystyrene, CF=cellulose fibre
Table 8.1: Details on the insulation measures of the EL²EP-project: maximum thickness and
applicable materials per envelope component
The glazing types are selected from the window library of PREBID, the building description
interface of the building simulation programme TRNSYS. It contains 66 different glazing
types with a U-value varying from 2.8 W/m²K to 0.4 W/m²K and a g-value, being the direct
and indirect solar heat gain coefficient, varying from 0.76 to 0.21. For calculating the edge
correction of the U-value of the glazing, 5 spacer types are available in PREBID. A choice list
of 14 window frames is composed, based on WTCB (1999) and Passivhaus (2000) that
contains window frames of wood, PUR, PVC and/or aluminium with a U-value varying from 6
W/m²K to 0.65 W/m²K, including frames that meet the passive house standard. The glass
area can vary per window between a minimum and a maximum value. Minimum and
maximum are defined per room, depending on the minimum requirement for light intensity
and the constructional limitations of the façade. To affect summer comfort, there is a
possibility to add a movable internal or external shading device with an opaque fraction (1-τ)
of 0 to 100%. If the option of shading is selected, it is applied to all windows with a south to
163
CHAPTER 8
west orientation and controlled by the indoor temperature in each room. Four levels of air
tightness are considered, being the average n50 of new built dwellings (SENVIVV 1998), n50 =
3/h (applying for designed natural ventilation systems), n50 = 1/h (standard for mechanical
ventilation systems) or n50 = 0.6/h (passive house standard). Finally, four ventilation
scenarios are considered to analyse the impact of extra summer or night ventilation as an
additional measure to control summer comfort.
8.2.3. System-related measures
After the optimisation process of the building envelope, the measures on the building
envelope are combined with system-related measures. This includes systems for distribution,
emission, production and storage of heat, systems for local electricity production and control
systems. The measures are nearly the same as in the BIM-project. However, in contrast to
the BIM-project, no local space heating systems are considered, but more attention is paid to
new technologies, such as different types of heat pumps (air-to-water and ground-to-water),
different systems for cogeneration of heat and power (CHP, both Stirling motor and gas
engine), different levels of efficiency of heat recovery in mechanical ventilation systems and
better performing control systems. In order to simulate passive house concepts, the
combination of a mechanical ventilation system with heat recovery and integrated electrical
heater is also incorporated into the list.
8.2.4. Building simulation programme
All energy simulations are executed with TRNSYS, a transient systems simulation
programme with a modular structure (TRNSYS 2005). In this programme, the user has to
specify the components that constitute the system and the way in which they are connected.
The TRNSYS library includes many of the components commonly found in thermal and
electrical energy systems, as well as routines to handle weather input data or other timedependent boundary conditions. The modular nature of TRNSYS gives the programme large
flexibility, and facilitates the addition of mathematical models that are not included in the
standard TRNSYS library. TRNSYS has become reference software for researchers and
engineers in the field of low energy buildings, HVAC systems, renewable energy systems,
etc.. The input data for ventilation and infiltration are calculated with COMIS (2003). This
multi-zone infiltration and ventilation simulation model predicts the airflows in and through the
building, taking into account both internal and external boundary conditions. COMIS can
easily be coupled to TRNSYS.
164
EL²EP-PROJECT
8.2.5. Boundary conditions
In contrast to the energy performance calculation procedure EPB that uses default values for
occupancy, internal gains, one single indoor set temperature for the whole dwelling and
monthly average weather data for Brussels, these boundary conditions need to be specified
as input data in TRNSYS. For the internal gains, six household scenarios have been
specified: from a household with a single, working person over a household of two persons to
a family of two parents and two children, of which one adult stays at home. These scenarios
describe the occupancy and use of lighting and electrical appliances in each room for each
hour of a week- or weekend-day and thus determine the internal gains and the set
temperature for each room. The weather data are hourly average data for the Test
Reference Year of Brussels, Belgium. The risk for summer overheating is evaluated with the
method of weighted temperature exceeding hours (WTE-hours) according to ISSO/SBR
(1994). This Dutch method, developed for the assessment of summer comfort in office
buildings, is adapted for dwellings, based on Maeyens (2001). The limit for summer comfort
is set at 130 WTE-hours per room. This limit is incorporated as a constraint in the
optimisation process, together with the maximum insulation level of K45, as explained in
section 4.2.3.
8.3
The life cycle inventory of the EL²EP-project is more extended than in the BIM-project.
Description of the assumptions can be found in chapter 5. In contrast to the BIM-project,
where an utilisation period of 40 years is adopted, within the EL²EP-project, an utilisation
period of 30 years is considered. In addition, the primary energy consumption, global
warming potential and net present value are also calculated for an utilisation period of 60 and
90 years, especially to analyse the impact of the utilisation period on the final results. The
impact on the net present value has been discussed already in section 6.5.3. The impact on
the life cycle inventory will be discussed in this chapter. However, in none of the cases, the
life cycle inventory takes into account thorough refurbishment or demolition of the building.
8.4
Cost assessment
Most assumptions for the cost assessment in the EL²EP-project have already been explained
in chapter 6. As mentioned in section 6.2.2, the total present value has been chosen as
economic optimisation criterion, but for the interpretation of the results, the focus is put on
the net present value, as only solutions with a positive net present value, compared to the
reference situation, can be considered economically viable. The impact of energy price
165
CHAPTER 8
scenarios and discount rate is already briefly discussed for the BIM-project, but will be
discussed further for the extremely low energy dwellings. The rebound effect, as discussed in
section 6.5.4, is not incorporated as such in the methodology and thus not taken into account
in the main results of the EL²EP-project. The main reason for not considering the rebound
effect, is that a constant comfort level is assumed for all simulated building variants in order
to compare building concepts on an equivalent basis. In reality, however, this comfort level
will not be achieved in badly insulated dwellings and therefore, the energy savings might be
overestimated in this theoretical approach. A more thorough discussion can be found in
section 6.5.4.
8.5
Optimisation
An important part of the EL²EP-project that was not applied within the BIM-project, is the
optimisation process. Where in the BIM-project, all possible combinations have been
calculated with the Flemish EPB in order to analyse the cost-effectiveness of different
combinations of measures, the EL²EP-project aims at the development of optimised low
energy building concepts. The evolutionary multi-objective optimisation model that is applied
in the EL²EP-project has already been explained in chapter 4 and summarised in the scheme
in figure 4.1. In section 8.6, the strengths and weaknesses of the optimisation methodology
are discussed.
8.6
Results
8.6.1. Optimisation of the building envelope measures
In the EL²EP-project, the assessment and optimisation criteria are energy consumption,
ecological impact and costs. As in the first phase of the project, only measures on the
building envelope are considered and systems are not included yet, primary energy
consumption cannot be calculated. In order to assess the energy-related impact of the
measures, the annual net energy demand for heating and the embodied non renewable
energy of the whole building are selected as energy criteria. Initially, this resulted in four
optimisation objectives, namely net heat demand, embodied energy, embodied global
warming potential and investment cost. However, the first simulations showed such a strong
linear correlation between the embodied energy and the embodied global warming potential
that the latter would have very little impact as extra objective on the optimisation process.
Therefore, only three objectives were finally maintained for the optimisation process, with the
extra advantage of an easier visualisation and interpretation of the results.
166
EL²EP-PROJECT
Results
Table 8.2 shows the values for net energy demand, non renewable embodied energy and
investment cost for the reference situation of all five dwellings. Distinction is made between a
reference version with a non insulated cavity wall, a massive brick wall as used for external
insulation systems and a wood frame structure. For the latter, 8 cm of mineral wool is
included, as this construction type is never used without insulation. Obviously, this results in
a lower net energy demand for the reference situation in case of a wood frame structure.
For the apartment building, not the whole building is considered, but only a centred flat. This
allows a comparison of all selected buildings based on compactness. For the flat, no wood
frame construction is considered, as this is highly unusual in the Belgian building practice.
Remarkable is the very high net energy demand for the architectural house. This is caused
by the very fragmented plan and thus the much higher heat loss area in comparison with the
other dwellings.
Net energy
Non renewable
Investment
demand
embodied
cost
[MJ/(m³a)]
energy
[€/m²]
[MJ/m³]
Terraced house
Cavity wall
Massive wall
184
215
560
565
740
710
136
600
740
Cavity wall
Massive wall
269
305
710
690
690
665
220
730
720
Cavity wall
Massive wall
252
309
810
780
780
730
175
915
810
Cavity wall
Massive wall
585
662
930
890
730
670
435
1160
700
128
144
480
485
620
610
Semi-detached house
Detached house
Architectural house
Apartment flat (centre)
Cavity wall
Massive wall
Table 8.2: Reference situation for the five selected building types of the EL²EP-project
167
CHAPTER 8
Terraced
Semi-
Detached
Archi-
Flat
house
detached
house
tectural
(centre)
house
Number of optimal
house
44
147
214
133
68
K14-40
K12-40
K11-43
K12-40
K6-35
0.19-0.54
0.14-0.46
0.12-0.46
0.12-0.40
0.15-0.83
25-56
16-52
24-58
24-39
17-33
20%
16%
17%
17%
12%
Without heat recovery
50-100
50-170
40-120
30-300
30-60
70% recovery of ventilation
20-70
25-75
20-80
20-140
< 20
630-900
770-1100
900-1400
1000-1500
430-550
±10%
±10%
±10%
±10%
±10%
25-90
30-100
30-120
50-150
50-60
±10-30%
±10-30%
±10-30%
±10-30%
±10-30%
800-1100
720-1120
780-1240
800-1200
600-770
solutions
Insulation level
Umean [W/m²K]
Glass area [m²]
Average % of floor area
Net energy demand
[MJ/m³a]
Non renewable
embodied energy
[MJ/m³]
Embodied GWP [kg/m³]
Investment cost
[€/m²floor area]
Table 8.3: Results for the optimisation of the building envelope of all selected building types
of the EL²EP-project
Table 8.3 presents the results for the optimisation of the building envelope for the five
selected buildings. Firstly, the number of optimal solutions found with the optimisation
process is shown. Then, per result, the range is given in which the optimal solutions lie. For
the net heat demand, two results are presented for the following reason. For the
development of extremely low energy concepts, heat recovery of ventilation losses is
indispensable to minimise the net heat demand. However, a heat recovery unit is needed
then and this is only implemented in the optimisation process of the second phase. In order
to incorporate already the impact of heat recovery in the net heat demand, two results are
given: the net heat demand without recovery of ventilation losses which will be used for
concepts without heat recovery, and the net heat demand, assuming a heat recovery
efficiency of 70%. Units with higher recovery efficiencies (up to 90-95%) are available on the
market, but by considering an efficiency of 70%, imperfect execution or functioning of the
system is taken into account. As can be noticed from table 8.3, the heat recovery of
ventilation losses has a significant impact on the net heat demand.
168
EL²EP-PROJECT
Table 8.3 shows that the optimisation process results in equivalent solutions for all five
dwellings for the aspects of insulation level and net energy demand. It also appears that the
embodied energy and the embodied global warming potential are positively affected by the
compactness. The more compact the dwelling, the lower the embodied energy and the
embodied global warming potential.
Net energy demand versus summer comfort
In order to investigate if lower net energy demand than found by the optimisation process,
can be realised by increasing the solar gains, an in-depth analysis of the distribution of gains
and losses is performed for the terraced dwelling. Solutions are created with maximum
insulation thicknesses for all envelope components, glazing with a U-value of 0.4 W/m²K,
window frames with a U-value of 0.65 W/m²K and an air tightness n50 of 0.6/h in order to
minimise the heat losses. No heat recovery of ventilation losses is taken into account here.
To investigate the effect of the solar gains, the glass area is varied between the minimum
and maximum values, from 21m² to 82m². Movable external sun shading with an opaque
fraction of 90% and extra summer and night ventilation is applied for all variants. The internal
gains remain constant.
Figure 8.1 shows the distribution of the gains and losses and the total net energy demand as
a function of the glass area for the case of the terraced house with a cavity wall. By
increasing the glass area, both the heat gains and the heat losses increase. Initially, the heat
gains dominate and the net energy demand as a (positive) sum of gains and losses
decreases. From a glass area of ca. 45m² on however, the decrease fades out and the net
energy demand remains constant at ca. 50 MJ/m³a. By further increasing the glass area, the
effect on the net energy demand becomes negligible, whereas the summer comfort quickly
deteriorates due to the solar gains. As external sun shading and extra ventilation is already
applied for all variants, there remain no passive measures to improve the summer comfort.
This way, the solid black dots, from a glass area of ca. 60m² on in figure 8.1 represent the
solutions that no longer fulfil the requirement for summer comfort. Only active cooling could
improve the summer comfort for the variants with large glass area, but this option is a priori
excluded for (extremely) low energy buildings. As the results show, the magnitude of the
glass area should be applied thoughtfully in concepts for extremely low energy dwellings and
the effect on the summer comfort should always be controlled by building simulations in
order to avoid installation of an energy devouring cooling system afterwards. The only way
for further decreasing the net energy demand is by heat recovery of the ventilation losses, as
already shown in table 8.3.
169
CHAPTER 8
Impact of glass area
Terraced house, cavity wall
losses / gains [MJ/(m³a)]
200
150
100
50
0
-50
20
30
40
50
60
70
80
90
-100
-150
-200
glass area [m²]
losses
gains
net heat demand
no summer comfort
Figure 8.1: Terraced house with cavity wall: effect of the glass area on net heat demand and
summer comfort. The heat losses (dark grey), heat gains (light grey) and total net heat
demand (crosses) are given as a function of the glass area. From a glass area of 60m², good
summer comfort can no longer be achieved with passive measures (outer shading and/or
extra ventilation) (solid black dots).
Impact of the constructional type
Initially, the thermal capacity was set as one of the variables for optimisation. This way,
massive and light weight construction variants were mutually compared. However, based on
the original assumptions, wood frame constructions were favoured for two of three
objectives, namely embodied energy and investment costs and therefore, obviously, most of
the optimal solutions comprised lightweight constructions. However, in Belgium, where the
majority of the dwellings is built with cavity walls and brick production forms an essential part
of the industrial activity, it is highly unrealistic and from the point of view of the building sector
not opportune, to promote a total shift of the building sector towards only wood frame
constructions. Therefore, after some adaptations, finally, the thermal capacity was no longer
one of the variables for optimisation, but the optimisation process is executed separately for
each of the construction types. This created the possibility to deduce optimised low energy
concepts for each of the construction types.
Figures 8.2 to 8.4 present the optimisation results from phase 1 for the three constructional
types for the terraced dwelling.
170
EL²EP-PROJECT
Net energy savings - extra non renew. embod. energy
Terraced house
extra non renewable embodied
energy [GJ]
160
cavity
140
outer insulation
wood frame
120
100
80
60
40
20
0
0
20
40
60
80
100
Net energy savings [MJ/m³a]
120
140
160
Figure 8.2: Terraced house: optimal solutions for three constructional types, cavity wall
(black dots), massive wall with outer insulation (grey triangles) and wood frame construction
(grey squares): extra non renewable embodied energy versus net energy savings
Net energy savings - extra investment cost
Terraced house
60000
extra investment cost [€]
cavity
outer insulation
wood frame
50000
40000
30000
20000
10000
0
0
20
40
60
80
100
120
140
160
Net energy savings [MJ/m³a]
Figure 8.3: Terraced house: optimal solutions for three constructional types, cavity wall
(black dots), massive wall with outer insulation (grey triangles) and wood frame construction
(grey squares): extra investment cost versus net energy savings
171
CHAPTER 8
Extra investment cost - extra non renewable embodied energy
Terraced house
extra non renewable embodied
energy [GJ]
160
140
cavity
outer insulation
120
wood frame
100
80
60
40
20
0
0
10000
20000
30000
40000
50000
60000
extra investment cost [€]
Figure 8.4 Terraced house: optimal solutions for three constructional types, cavity wall (black
dots), massive wall with outer insulation (grey triangles) and wood frame construction (grey
squares): extra non renewable embodied energy versus extra investment cost
Each figure provides the results for two of the three objectives relative to the reference
situation. All grey open circles represent the results for all solutions calculated during the
optimisation process, with each circle corresponding to one combination of insulation,
glazing, glass area, sun shading, ventilation and air tightness. The solid black and solid grey
dots show the overall optimal solutions for the three constructional types. Obviously, the
highest net energy savings are realised with a wall with external insulation, as for this type
the initial net energy demand was highest (figure 8.2). For the same reason, lower savings
are realised with the wood frame structure, as the first centimetres of insulation, already
present in the reference situation, are the most effective. This also explains the lower extra
embodied non renewable energy for the solutions with a lightweight structure. The ranges of
extra investment cost, however, are similar for the three construction types (figures 8.3 and
8.4).
All the optimal solutions from phase 1 form the starting point for phase 2, as only to these
optimised constructional concepts, different installations are applied in search for the most
optimal combinations. This results in a number of globally optimised concepts for low to
extremely low energy dwellings.
172
EL²EP-PROJECT
8.6.2. Globally optimised concepts for extremely low energy dwellings
Figures 8.5 and 8.6 present the optimal solutions for the terraced dwelling. In figure 8.5 the
net present value is given as a function of the total primary energy consumption over the
utilisation period, whereas in figure 8.6, the x-axis represents the annual energy cost. The
results are valid for an utilisation period of 30 years, a low energy price scenario and a
discount rate of 4%. In the analysis of all results (grey squares), distinction has been made
between all optima resulting from the optimisation process (all black squares) and the optima
with a positive net present value compared to the reference case (solid black squares). The
assumptions on the energy price scenarios are presented in section 6.3.3.
Some optimal solutions in figures 8.5 and 8.6 seem to be suboptimal as there exist solutions
with higher NPV for the same total primary energy consumption. They are nevertheless part
of the Pareto front. This misleading visual effect is caused by the fact that the optimisation
process considers three objectives (total present value, total primary energy consumption
and global warming potential), whereas the figure only shows two objectives. The actual
Pareto front or trade-off curve is a 3D surface and the Pareto optima in figures 8.5 and 8.6
(open and solid black squares) are in fact a 2D projection of that 3D Pareto front. This
projection does not coincide completely with the 2D trade-off curve of net present value and
total primary energy consumption. This explains why some optimal results in figures 8.5 and
8.6 lie below the 2D trade-off curve.
Net present value vs. total primary energy consumption
20000
0
-20000
-40000
all results
optima
optima NPV30>0
-60000
-80000
-100000
800
1300
1800
2300
2800
total primary energy consumption [GJ]
Figure 8.5: Terraced house, utilisation period of 30 years, low energy price scenario,
discount rate = 4%: the net present value over 30 years (NPV30) is given as a function of the
total primary energy consumption over 30 years. All results are presented in grey, whereas
the trade-off curve is given in black. The solid black dots have a NPV30 >0.
173
CHAPTER 8
Obviously the optima with a positive net present value are of most interest, as they are
economically viable, even in the improbable case that the energy prices remain constant.
The results represent the same hierarchy of energy saving measures as achieved from the
BIM-project and summarised in section 7.7. As can be seen from figure 8.6, energy saving
viable concepts can be realised for the terraced house with an energy cost ranging from 400
to 700€ per year.
Similar results have been found for the other building types, but to limit the extent of this
dissertation, not all results for all five buildings are presented here. What the economic viable
concepts are, in which manner extremely low energy houses can be realised and for which
energy price scenario these concepts become viable, will be discussed in section 8.7,
illustrated with graphs for different building types. All details and results for all five buildings
can be found in Verbeeck et al. (2007).
Net present value vs. annual energy cost
20000
0
-20000
-40000
-60000
all results
optima
-80000
optima NPV30>0
-100000
200
400
600
800
1000
1200
discount rate = 4%: the net present value over 30 years (NPV30) is given as a function of the
annual energy cost. All results are presented in grey, whereas the trade-off curve is given in
black. The solid black dots have a NPV30 >0.
8.6.3. Embodied energy versus energy savings
Figure 8.7 presents the energy payback time for energy saving measures on the building
envelope and the installations of the semi-detached house. The optima are valid for an
utilisation period of 30 years and contain solutions for all three constructional types (massive
and light weight). Similar to the results of the BIM-project, these results show that even for
174
EL²EP-PROJECT
extremely low energy dwellings, the energy payback time remains less than 2 years. It can
be concluded from these results that the embodied energy of the energy saving measures
should not be a reason of great concern, regardless of the applied insulation materials or
installation systems.
Energy Payback Time
Semi-detached house, measures on building envelope and installations
extra embodied energy [GJ]
350
300
EPBT = 1 year
EPBT = 2 years
250
all results
optima
200
150
100
50
0
0
50
100
150
annual primary energy savings [GJ]
Figure 8.7: Semi-detached house, utilisation period of 30 years: the extra embodied energy
for energy saving measures on both the building envelope and installations is given as a
function of the annual primary energy savings. All results are presented in grey, whereas the
results of the trade-off for NPV and total primary energy consumption are given in black.
8.7
Discussion
Firstly, the strengths and weaknesses of the optimisation model are discussed. Secondly, the
impact of several aspects, such as constructional type, compactness, energy carrier and
heat production system are discussed. Then, several cost-related aspects are discussed,
such as impact of the constructional cost versus installation cost and the impact of price
evolutions. Finally, the extremely low energy dwellings that result from the EL²EP-project, are
compared to the economic optimum and to existing concepts for extremely low energy
dwellings, such as passive houses and zero energy houses.
175
CHAPTER 8
8.7.1. Strengths and weaknesses of the optimisation model
The multi-objective optimisation programme proved to be very useful in the search for
optimised (extremely) low energy dwellings. For a good balance between computational time
and approximation of the ideal Pareto front, the population size was set at 100 chromosomes
and the number of generations at 60. This resulted for all five building types in a wide spread
of Pareto optima. However, starting from these results, an in depth analysis and fine tuning
still is necessary in order to end up with all realistic building concepts. Therefore, some
adaptations and extra calculations need to be performed.
Firstly, the U-values of all envelope components need to be analysed. Despite the boundary
condition for the overall insulation level, this constraint does not guarantee that the U-value
per envelope component does not exceed the maximum value set in the EPB standard as
presented in table 8.4. If the U-value of an envelope component is too high, the insulation
thickness must be increased in order to fulfil the requirement.
Building component
Maximum U-value
[W/m²K]
Façade
0.6
Roof
0.4
Window
Overall
2.5
Glass
1.6
Floor above cellars
0.4
Floor slab on the ground
0.4
Wall/floor between flats
1.0
Common wall
1.0
Table 8.4: Maximum U-value per envelope component
according to the Flemish EPB (EPB 2005)
Secondly, the homogeneity of the insulation level of the building envelope needs to be
analysed. As the optimisation process does not compare the insulation thicknesses of the
different envelope components, some results might have very inhomogeneous thicknesses,
e.g. 2 cm roof insulation and 20 cm façade insulation or windows that combine an Uglass of
0.7 W/m²K with an Uframe of 3.6 W/m²K. These variants need to be adapted to more
homogeneous combinations and recalculated.
Theoretically, both the maximum U-value and the homogeneity could have been
incorporated as extra constraints in the optimisation process. However, this would have
176
EL²EP-PROJECT
significantly increased the complexity of the penalty functions and therefore, the above
control procedure has been selected.
In addition, some extra variants can be constructed by implementing the maximum insulation
thickness for each envelope component and the best performing window characteristics on
the optimal variants that result from the optimisation process.
Finally, for those variants that violate the boundary condition for summer comfort, extra
measures such as outer sun shading and/or night ventilation can be included, if not yet
present, and a new evaluation of the summer comfort can be performed.
After this procedure of control and extension of the draft Pareto front, new simulations need
to be performed for all adapted variants and these results form the basis to determine the
final Pareto front. This way, the evolutionary multi-objective optimisation process is a useful
tool to development (extremely) low energy dwellings, but final control of the results is
indispensable to avoid the presence of unrealistic building concepts within the final optimal
solutions.
8.7.2. Impact of the constructional type
Figure 8.8 presents the results for the semi-detached dwelling. The optimal solutions are
determined after ordering the results by constructional type. The highest net present value
can be realised with a construction with outer insulation, whereas the lowest energy
consumption is achieved with a wood frame construction. The variants with a cavity wall lie in
between. This also appears from figures 8.9 and 8.10 for which the sorting by constructional
type is executed after determining the optimal solutions. To illustrate the effect of projecting
the 3D optimisation results in 2D graphs, figure 8.9 presents the optimal variants for the
objectives net present value – total primary energy consumption, whereas figure 8.10
presents the same optimal variants for the objectives net present value – global warming
potential. From these figures it is clear that, although the variants with wood frame
construction achieve lower net present values than the variants with outer insulation, they
perform better for the global warming potential. The variants with cavity wall lie both for net
present value and global warming potential in between and therefore they disappear from the
optimal solutions when the optimisation process is executed for all constructional types
together (figure 8.9 versus figure 8.8).
177
CHAPTER 8
Pareto optima per constructional type
Semi-detached house, utilisation period = 30 years, high energy scenario
80000
60000
40000
20000
0
-20000
-40000
optima cavity wall
-60000
optima outer insulation
-80000
optima wood frame construction
-100000
800
1300
1800
2300
2800
3300
3800
Figure 8.8: Semi-detached house, utilisation period of 30 years, high energy price scenario:
Pareto optima are determined after sorting the results by constructional type: optimal variants
with cavity wall (black dots), optimal variants with massive wall with outer insulation (grey
triangles) and optimal variants with wood frame construction (grey squares)
Pareto optima ordered by constructional type
80000
60000
40000
20000
0
-20000
optima cavity wall
-40000
-60000
-80000
800
1300
1800
2300
2800
3300
3800
Pareto optima are determined for all results together and then sorted by constructional type:
optimal variants with cavity wall (black dots), optimal variants with massive wall with outer
insulation (grey triangles) and optimal variants with wood frame construction (light grey
squares)
178
EL²EP-PROJECT
Pareto optima ordered by constructional type
80000
60000
40000
20000
0
-20000
optima cavity wall
-40000
-60000
-80000
40
60
80
100
120
total global warming potential [ton]
140
Pareto optima are determined for all results together and then sorted by constructional type:
optimal variants with cavity wall (black dots), optimal variants with massive wall with outer
insulation (grey triangles) and optimal variants with wood frame construction (light grey
squarse)
8.7.3. Impact of the compactness
The compactness C is the proportion of the heated air volume in m³ to the total heat loss
area in m² and is expressed in m. Figure 8.11 presents the trade-off curve for the total
present value, expressed in €/m² heat loss area, and the total primary energy consumption,
expressed in MJ/m² heat loss area, for all five building types. The compactness per building
type is mentioned in the legend of the figure. With the compactness varying from 0.85 to
4.17m and a constant heated air volume of ca. 470m³ for all dwellings, the heat loss area
varies over the dwellings between 105m² and 560m². Table 8.2 in section 8.6.1 showed that
in the non-insulated reference situation, the net heat demand per m³ heated volume is more
than four times higher for the less compact (architectural) house than for the most compact
one (flat). Because of the large heat loss area of the less compact house, it is obvious that
this house has the highest energy saving potential. This explains why in figure 8.11 the
architectural house performs best. It also leads to the conclusion that, if energy saving
measures beyond the legal standard are indispensable for new common dwellings, they are
even more indispensable for modern architecture with a fragmented plan. In practice this
means that, where K25-30 (Umean = 0.3-0.4W/m²K) is the economic optimal insulation
179
CHAPTER 8
standard for houses with a compactness of 1.5-2, for houses with a low compactness the
economic optimum lies at K20 (Umean = 0.4W/m²K). This way, the lack of compactness should
be compensated by more severe insulation measures and thus, similar energy consumption
at similar total present value could be achieved for all building types. This is shown in figure
8.12, where the energy consumption and TPV are given in absolute values and not per m²
heat loss area, as in figure 8.11.
Impact of compactness
utilisation period = 30 years, low energy price scenario
Total present value/m² heat
loss area [€/m²]
1800
1600
1400
1200
1000
800
flat C=4.17m
600
terraced C=2.04m
semi-detached C=1.49m
detached C=1.22m
architectural C=0.85m
400
200
0
0
2000
4000
6000
8000
10000
Total primary energy consumption/m² heat loss area [MJ/m²]
Figure 8.11: All dwellings, utilisation period of 30 years, low energy price scenario: trade-off
for TPV/m² heat loss area and total primary energy consumption per m² heat loss area. C is
the compactness, being the proportion of heated air volume and heat loss area. As the
heated air volume of all these dwellings is almost the same, this figure shows that dwellings
with a low compactness have the highest energy saving potential, due to their large heat loss
area.
As figure 8.12 shows, similar consumption levels are realised at similar TPV for all building
types. Only in case of the flat, much lower energy consumption can be realised at much
lower TPV due to the very high compactness. Both figures lead to the conclusion that less
compact houses have the highest energy saving potential, due to their very bad performance
if not insulated, whereas in absolute terms, when applying a certain combination of energy
saving measures, highly compact dwellings will perform better energetically at lower TPV,
because of their very small heat loss area.
180
EL²EP-PROJECT
Impact of compactness
utilisation period = 30 years, low energy price scenario
350000
300000
250000
200000
150000
flat C=4.17m
100000
terraced C=2.04m
semi-detached C=1.49m
detached C=1.22m
50000
architectural C=0.85m
0
0
500
1000
1500
2000
2500
3000
Figure 8.12: All dwellings, utilisation period of 30 years, low energy price scenario: trade-off
for total present value and total primary energy consumption. Similar energy consumption
levels can be realised for all building types at similar TPV. Only for the flat, much lower
energy consumption can be realised at much lower TPV due to the very small heat loss area.
8.7.4. Impact of energy carrier and heat production system
Energy carrier
To analyse the position of the different energy carriers in the optimal solutions, a similar
analysis has been executed as for the constructional types. Firstly, all results are ordered by
energy carrier and then, the optimal solutions are determined. This is presented for the
detached house in figure 8.13. This figure shows that, if for any reason, one energy carrier is
more preferred than the other, it is always possible to realise a concept that performs equally
to concepts with other energy carriers, as for all energy carriers, comparable levels of energy
consumption can be realised at comparable levels of NPV.
However, if there is no preference for any energy carrier and the intention is to realise
concepts e.g. with highest NPV or lowest energy consumption, this figure also shows that the
highest NPV is achieved with systems on natural gas or fuel, whereas heat pumps on
electricity realise the lowest energy consumption.
181
CHAPTER 8
Pareto optima per energy carrier
Detached house, utilisation period = 30 years, low energy price scenario
60000
40000
20000
0
-20000
-40000
optima gas
-60000
optima fuel
optima electricity
-80000
-100000
1000
1500
2000
2500
Figure 8.13: Detached house, utilisation period = 30 years, low energy price scenario: firstly,
all results are sorted by energy carrier and then, Pareto optima are determined: optimal
variants with natural gas (dark grey squares), optimal variants with fuel (light grey squares)
and optimal variants with electricity (black triangles).
Pareto optima for all results, ordered by energy carrier
60000
40000
20000
0
-20000
-40000
-60000
optima gas
optima fuel
optima electricity
-80000
-100000
1000
1500
2000
2500
Pareto optima are determined for all results and then, the optima are sorted by energy
carrier: optimal variants with natural gas (dark grey squares), optimal variants with fuel (light
grey squares) and optimal variants with electricity (black triangles).
182
EL²EP-PROJECT
This also appears from figure 8.14 for which firstly, the optimal solutions are determined and
then afterwards, the optima are sorted by energy carrier. In the range of the economic
optimum, fossil fuels give the best results, whereas for extremely low energy houses,
electricity appears to be the most appropriate energy carrier. The heat production systems
that are represented by these optima, are discussed in the next section.
Heat production system
To analyse the position of the different heat production systems, the optimal solutions of
figure 8.14 are sorted by heat production system. This is presented in figure 8.15. This figure
shows a clear hierarchy of heat production systems when considering the net present value
versus the total primary energy consumption.
Pareto optima for all results, ordered by heat production
60000
40000
20000
0
-20000
high efficiency boiler
-40000
condensing boiler
-60000
-80000
-100000
1000
air-water heat pump
soil-water heat pump
cogeneration heat and power
1500
2000
2500
Pareto optima are determined for all results and then, sorted by heat production system: high
efficiency boilers (open triangles), condensing boilers (dark grey squares), air-to-water heat
pumps (light grey solid triangles), soil-to-water heat pumps (open dots) and cogeneration of
heat and power (black squares).
The economic optimum is achieved with high efficiency boilers (open triangles). With a
condensing boiler (dark grey squares) lower energy consumption can be realised, but at
lower NPV than with a high efficiency boiler. The cheapest type of heat pumps (air-to-water,
light grey triangles) results in equal to lower energy consumption than with condensing
boilers, but in most cases also the NPV is lower and even can become negative. For the
lowest energy consumption ranges, the better performing, but more expensive soil-to-water
183
CHAPTER 8
heat pump (open circles) is more appropriate, but mostly not economically viable. Very few
variants with a CHP-system (black squares) are part of the optimal solutions. For these
variants, the cheapest, but still expensive, version of a CHP-system is applied, being a
condensing boiler with an integrated Stirling motor. Although the results appear to be
comparable to the results with a soil-to-water heat pump, these results should be interpreted
with care, as up to now, the uncertainty on the performance and on the cost is higher for
CHP than for heat pumps. It can be expected that more reliable values will become available
in the future.
8.7.5. Constructional cost versus installation cost
Analysis of the evolution of the total present value shows that, compared to the non-insulated
reference situation, firstly the TPV decreases with decreasing total primary energy
consumption up to a minimum and then increases with further decreasing total primary
energy consumption. As explained in equation [6.1], the total present value mainly consists
of the initial investment cost, the discounted replacement cost and the discounted energy
cost. The investment cost contains the cost for the construction and the building envelope as
well as for the installation. These costs are assumed to be independent of the energy price
scenario. In reality, of course, investment costs will evolve with the energy cost evolution, but
this is not taken into account here.
To analyse the contribution of the different cost components to the total present value as a
function of the energy consumption, figure 8.16 presents the total present value (black solid
diamonds), the discounted constructional cost (open grey diamonds), the discounted
installation cost (light grey solid squares) and the discounted energy cost (grey crosses). The
results are valid for the terraced house for an utilisation period of 30 years, a discount rate of
4% and a low energy price scenario. The figure shows the large contribution of the cost for
the building structure and envelope to the TPV. The discounted constructional cost starts at
133,000 € for the non insulated version and increases up to nearly 220,000€ for the most
energy saving variants. In figure 8.16, levels of constant constructional cost can be observed.
This is caused by the fact that the first optimisation step results in a certain number of
optimised concepts for the building envelope. Only to these concepts, all kinds of installation
types are applied in the second optimisation step. Obviously, the total primary energy
consumption of a building does not only depend on the building envelope, but also on the
installation type that is added. This explains why the same cost for structure and envelope
can be observed at different levels of primary energy consumption. The discounted
installation cost starts at 26,000€ for the non-insulated reference situation and almost
doubles when evolving to the most energy saving variants. However, the contribution to the
184
EL²EP-PROJECT
TPV is much smaller than for the constructional cost and furthermore, for energy
consumption levels in between, even lower costs than 26,000€ are possible. Obviously, the
discounted energy cost decreases with decreasing energy consumption, but this cost is very
small compared to the costs for the whole building.
Constructional cost, installation cost and energy cost
Present value over 30 years [€]
Terraced house, utilisation period = 30 years, low energy price scenario
300000
250000
200000
150000
100000
50000
0
800
1300
1800
2300
total present value
discounted installation cost
2800
discounted constructional + envelope cost
discounted energy cost
Figure 8.16: Terraced house, utilisation period of 30 years, low energy price scenario: the
total present value is given as a function of the total primary energy consumption over 30
years for all combinations of energy saving measures on the building envelope and the
installations (solid black diamonds)) as well as the contribution to the present value of the
construction and building envelope cost (open grey diamonds), the installation cost (light
grey solid squares) and the energy cost. All costs are discounted for a discount rate of 4%.
Figure 8.16 presents the overall cost for the building, whereas a large part of the building
cost consists of a fixed cost for the building structure, mostly independent of the insulation
level. To analyse the impact of the energy saving measures only, figure 8.17 presents the
extra initial investment cost for the building envelope and the installation, compared to the
non-insulated version of the terraced house as a function of the annual energy cost. In the
non-insulated situation, the terraced house has an annual primary energy consumption of ca.
114,000 MJ, corresponding to an annual energy cost of ca. 1,600€. A decrease of the annual
energy cost with 50% can be realised with a total extra initial investment cost of ca. 8,000 to
9,000 €, by combining energy saving measures on the building envelope for an extra cost of
ca. 12,000€, with a well-performing, but smaller and thus, cheaper heating system (3,000 to
4,000€ less than the reference case). As figure 8.17 shows, further decrease of the annual
energy cost is possible, but finally leads to an exponential increase of the extra initial
investment cost. An annual energy cost of less than 300€ requires for the terraced house at
185
CHAPTER 8
least an extra initial investment cost of 60,000€ for the energy saving measures in the
building envelope and 20,000€ for the installation, compared to the non-insulated version.
Extra initial investment cost for energy saving measures
Terraced house, utilisation period = 30 years
Extra initial investment cost [€]
90000
80000
70000
total extra cost
extra cost envelope
60000
extra cost installations
50000
40000
30000
20000
10000
0
-10000200
400
600
800
1000
1200
Annual energy cost [€/year]
Figure 8.17: Terraced house, utilisation period of 30 years: the total extra initial investment
cost is given as a function of the annual energy cost for all combinations of energy saving
measures (black solid diamonds) as well as the contribution of the extra initial investment
cost for energy saving measures on the building envelope (open grey diamonds) and on the
installation (light grey crosses). The annual energy cost of the reference situation is ca.
1,600€.
8.7.6. Impact of price evolutions and discount rate
Energy price scenarios
Figure 8.18 presents the optimal solutions for the architectural house for an utilisation period
of 30 years. Similar to the outcome of the BIM-project, it appears from these results that
almost the same optimal combinations are found for all three energy price scenarios. The
only difference is that some solutions that are not economically viable for low energy prices,
become viable when the energy price increases significantly. The hierarchy of measures,
however, does not depend on the price scenario.
The fact that for some levels of energy consumption more than one optimum is presented
per energy price scenario can be explained by the optimisation for three objectives. E.g. for
the lowest energy consumption in figure 8.18, the points for the high energy price scenario
(light grey) represent a solution with outer insulation (lowest NPV) and a solution with wood
frame construction (highest NPV). However, as the solution with wood frame construction
has a much lower GWP, both solutions can be part of the same 3D Pareto surface, that is
projected in 2D in figure 8.18. Furthermore, for the highest energy price scenario, the
186
EL²EP-PROJECT
difference in energy price evolution can be observed in the optimal solutions, especially for
higher energy consumption levels. Some combinations with fuel that are not restrained for
lower energy prices, form part of the optimal solutions, as the assumed price increase for fuel
(3.2%) is lower than the assumed price increase for gas and electricity (4.3%). This explains
the scatter of results for the highest energy price scenario at higher energy consumption
levels (> 1600 GJ for the architectural house).
As the architectural house is extremely energy-devouring in the reference situation due to its
large heat loss area, compared to its heated volume, most of the optimal solutions appear to
be economic viable, even for the improbable scenario that the energy prices remain
constant. The impact of the compactness is further analysed in the next section.
Impact of energy price scenarios
Architectural house, utilisation period = 30 years, discount rate = 4%
200000
Net present value [€]
optima low price scenario
150000
optima middle price scenario
optima high price scenario
100000
50000
0
-50000
1000
1200
1400
1600
1800
Figure 8.18: Architectural house, utilisation period of 30 years: Net present value vs. total
primary energy consumption is presented for the optimal solutions per energy price scenario
(low, middle, high). The optimal combinations appear to be almost independently from the
energy price scenario. Only the net present value increases with increasing energy prices.
Discount rate
All results presented elsewhere in chapter 8 have been calculated for a discount rate of 4%,
as this is assumed to be a realistic estimation of the real interest rate. However, in order to
control the robustness of the results with relation to the assumptions, different scenarios for
the discount rate are analysed: 2%, 4% and 8%. The larger the discount rate, the less
importance is given to expenses in the far future. This way, a large discount rate can be
applied to simulate the fact that most consumers are mostly concerned about the initial
investment cost and estimate the energy cost of less importance. On the other hand, a low
discount rate of 2% could simulate the economic effect of subsidies and fiscal corrections,
187
CHAPTER 8
experienced by the private consumer, meanwhile from the society point of view, a discount
rate of 4% might be a more realistic assumption. For clarity’s sake, subsidies and fiscal
depreciation have not been taken into account in this project. However, there could be the
case in reality that through subsidies and fiscal corrections, certain energy saving
investments might appear economically viable for the private consumer, meanwhile from the
point of view of society, they are not the best economic choice. Therefore, it is important to
analyse the impact of the discount rate on the economic viability of solutions and to control
the robustness of the results regarding the assumptions for discount rate.
Figure 8.19 presents the optimal solutions for the terraced house for a discount rate of 2%,
4% and 8%. For variants with average insulation level, the energy cost still has a significant
impact on the net present value, whereas for extremely low energy houses, the initial
investment cost dominates the net present value. This is reflected by the fact that for variants
with total primary energy consumption more than 1000GJ, the net present value diverges for
the different discount rates. Figure 8.19 also shows that with higher discount rate, fewer
variants are economically viable and the curve of optima becomes flatter in the range of
higher energy consumption. These phenomena are intensified in case of combination of high
energy prices and high discount rates (figure 8.20). For extremely low energy houses
however, the discount rate has a much lower impact.
Impact of discount rate
Net present value over 30
years [€]
50000
40000
30000
20000
10000
0
-10000500
-20000
-30000
-40000
1000
1500
2000
2500
a = 2%
a = 4%
a = 8%
-50000
Figure 8.19: Terraced house, utilisation period of 30 years, low energy price scenario: The
optimal solutions are presented per discount rate scenario (a=2%, a=4%, a=8%). The
optimal combinations appear to be almost independently from the discount rate scenario.
Only the net present value decreases with increasing discount rate with a flattening curve in
the range of higher energy consumption.
188
EL²EP-PROJECT
Impact of discount rate
Terraced house, utilisation period = 30 years, high energy price scenario
years [€]
50000
40000
30000
20000
10000
0
-10000500
1000
1500
2000
2500
-20000
a = 2%
-30000
a = 4%
-40000
a = 8%
-50000
Figure 8.20: Terraced house, utilisation period of 30 years, high energy price scenario: The
optimal solutions are presented per discount rate scenario (a=2%, a=4%, a=8%). The
optimal combinations appear to be almost independently from the discount rate scenario.
The shift in net present value is even more intensified than for the low energy price scenario.
Impact of material cost evolution
Terraced house, utilisation period = 30 years, low energy price scenario,
years [€]
20000
10000
0
-10000 0
500
1000
1500
2000
2500
-20000
-30000
-40000
-50000
inv = 0%
inv = 2%
inv = 4%
-60000
-70000
discount rate = 4%: The optimal solutions are presented per material cost scenario (annual
increase of 0%, 2% and 4% above inflation). The highest impact is found for the most energy
saving variants. However, the optimal combinations appear to be almost independently from
the material cost scenario.
189
CHAPTER 8
Material cost evolution
All results presented elsewhere in chapter 8 are calculated for an increase of the
(re)investment cost of 0% (above inflation). This means that rI in equation [6.1] is set to 0.
However, due to changing economical situations or increasing energy prices, higher
investment costs might be possible. Therefore, different scenarios have been analysed.
Figure 8.21 presents the optimal solutions for the terraced house for an annual price
increase for (re)investment costs with 0%, 2% and 4%. As could be expected, higher
material cost will have the largest impact on extremely low energy houses, as here the
investment costs are highest. However, the impact is much smaller than the impact of the
energy price evolution or the discount rate.
Conclusions for the cost scenarios
For the calculation of total present value and net present value, assumptions need to be
made on discount rate, energy price evolution and material cost evolution. These
assumptions have an impact on the economic viability of energy saving variants. However, of
greater importance are the following conclusions:
-
The economic optimum is not only independent of these cost scenarios, it also
remains economically viable for all analysed cases.
-
Also the hierarchy of energy saving measures, as concluded from the BIM-project
and further extended in section 8.7.8, is independent of these scenarios. Obviously,
the economic viability of some measures, such as the application of heat pumps or
mechanical ventilation with heat recovery, will depend on the cost scenarios, but in
contrast to what sometimes is commonly assumed, there is no shift at all in this
hierarchy.
So, generally, it can be concluded that both the economic optimum and the hierarchy of
energy saving measures are very robust for the assumptions made.
8.7.7. Position of extremely low energy dwellings to the economic optimum
From the analysis of the results for the five building types of the EL²EP-project, a similar
economic optimal combination of energy saving measures can be deduced as for the BIMproject. The economic optimum is very robust and consists of:
-
an insulation level K25-27 for most common houses (Umean ≈ 0.3-0.35W/m²K for
compactness of 1.5-2). Only for houses with a very fragmented plan (compactness <
1), the economically optimal insulation level decreases, down to K20 (Umean ≈
0.2W/m²K)
190
EL²EP-PROJECT
-
combined with a high efficiency or condensing boiler on gas or fuel, variable water
temperature, high temperature radiators and a natural ventilation system.
The energy performance can further be improved by means of a mechanical ventilation
system with a good performing heat recovery unit (here an overall performance of 70% is
assumed) or an air-to-water heat pump. The seasonal performance factor (SPF) of an air-towater heat pump is less than the SPF of a soil-to-water heat pump, but the investment cost is
much lower. Therefore, the air-to-water heat pump turns out to be economically viable more
quickly than the soil-to-water heat pump. For most dwellings, these measures are just below
the limit of economic viability.
Further decreasing the energy consumption is possible by implementing a better insulation
level (K15-20), by combining this very good insulation level with a heat pump and with
mechanical ventilation with heat recovery or even by combining it with a CHP-system. This
leads to very low total primary energy consumption and annual energy costs, but these
solutions are far beyond the economic optimum for most dwellings.
Only for houses with a very low compactness, which are very energy-devouring in a badly
insulated version, these extended combinations of energy saving measures turn out to be
economic viable at higher energy price evolutions.
Table 8.5 presents a summary of the results of the global optimisation of all five building
types. For each result and each building type, the range in which the optimal solutions lie, is
given. Cost results are valid for the low energy price scenario and a discount rate of 4%.
Most values are expressed per m³ heated air volume (energy related values) or per m²
heated floor area (cost related values).
Terraced
Semi-
Detached
Archi-
Flat,
house
detached
house
tectural
centre
house
house
Number of optima
79
138
125
36
35
econ. viable
16
48
62
15
17
K15 → K27
K13 → K34
K13 → K40
K13 → K20
K9 → K38
0.2 → 0.37
0.15 → 0.4
0.14 → 0.43
0.13 → 0.20
0.15 → 0.69
15 - 90
15 - 90
25 - 100
30 - 55
< 40
2000 - 4500
2240 - 4650
2230 - 5330
2500 - 3700
1300 - 2500
Insulation level
Umean [W/m²K]
Net energy
demand [MJ/m³a]
Total primary
energy
consumption over
30 years [MJ/m³]
191
CHAPTER 8
Total GWP over
30 years [kg/m³]
90 - 330
100 - 290
95 -330
120 - 230
100 - 190
40 - 500
80 - 800
3 - 840
100 - 720
50 - 340
300 - 880
300 - 900
300 - 1000
300 - 650
130 - 510
1100 - 1300
1000 -1600
1000 - 1700
1000 - 1500
800 - 1050
-260 → 60
-360 → 300
-500 → 200
-200 → 400
-160 → 70
9 → 50
3 → 57
-4 → 62
2 → 23
-3 → 35
Extra investment
cost [€/m² floor
area]
Annual energy
cost [€]
Total present
value over 30
years, low energy
scenario [€/m²]
Net present value
over 30 years, low
energy scenario
[€/m²]
Static payback
time [years]
Table 8.5: Results for the global optimisation of all five dwellings of the EL²EP-project: only
results from the trade-off curve of total present value and total primary energy consumption
over an utilisation period of 30 years and low energy price scenario are presented here.
Some values are expressed per m³ heated air volume, some per m² floor area. For each
result, the range of values is given.
8.7.8. Position of existing concepts for extremely low energy dwellings
Passive houses
The concept of a passive house was developed in the late eighties by Adamson at Lund
University, Sweden and Feist, founder of the Passivhaus Institut, Germany (Feist 2006). It
stands for a building concept in which a comfortable indoor climate can be maintained
without active heating and cooling systems. For the European context, prerequisite is an
annual heating requirement less than 15kWh/m² floor area and a total annual primary energy
consumption of less than 120 kWh/m² floor area for heating, domestic hot water and
household electricity. The latter requirement is set to avoid that the extremely low heating
demand would be achieved by an increase of energy use for other purposes, such as
domestic appliances (www.passivhouse.com). Basic features to realise a passive house are
a good compactness and extremely high insulation level (average U-value for the opaque
elements < 0.15W/m²K and an overall U-value for windows < 0.8W/m²K), a very air tight
building envelope (n50 < 0.6/h), a conscious use of the southern orientation, glass area and
sun shading in order to have a passive use of solar energy without creating an overheating
192
EL²EP-PROJECT
problem in summer, a mechanical ventilation system with a highly performing heat recovery
unit (efficiency > 80%), preheating of the fresh ventilation air by underground ducts, domestic
hot water production by alternative systems such as solar collectors or heat pumps and use
of energy saving domestic electrical appliances.
Within the EL²EP-project a measuring campaign was set-up in two Belgian passive houses
to evaluate both the energy consumption and the indoor climate in winter and summer
conditions. Presentation and discussion of these results can be found in De Meulenaer et al.
(2005(1)) and (2005(2)).
In order to evaluate the position of the passive house concept to the economic optimum and
the building concepts developed within the EL²EP-project, the passive house standard is
applied to one reference dwelling, being the terraced house as it is the building with the
highest compactness. The evaluation is executed only for the heating requirement.
Therefore, only variants with a maximum insulation thickness and passive house windows
are selected that have a comfortable indoor climate in winter and summer and an annual end
energy consumption for heating less than 15kWh/m² floor area. Furthermore, the traditional
heating system with boiler or heat pump and radiators or floor heating is replaced by an
electrical heating system in the mechanical ventilation system. This is simulated by assuming
a production efficiency of 100% and an overall efficiency of 90% for emission, distribution
and control. For the heat recovery efficiency, two scenarios are considered: an efficiency of
90% and one of 70%, the latter because due to imperfect installation and sealing of the
recovery unit, the real performance can be much lower than what is mentioned in the product
information.
Figure 8.22 and 8.23 compare the optimal solutions from the EL²EP-project (dark grey
squares) with the passive house variants for the low and high energy price scenarios for net
present value and total primary energy consumption. Both variants with 70% heat recovery
(light grey squares) and 90% heat recovery (black triangles) are presented.
The importance of the heat recovery efficiency clearly appears from both figures. As the heat
losses by conduction and infiltration are very low due to the very high insulation level and
very high air tightness, the overall heat losses are mainly determined by the ventilation
losses. Heat recovery units with 90% efficiency are on the market, but excellent
workmanship is crucial to really achieve this high efficiency in practice. Control of the air
tightness of the ventilation system after execution by means of a duct blaster or pressure test
is therefore indispensable in order to detect the unwanted leakages. It also appears from the
figures that with the PH variants with 90% heat recovery the lowest total primary energy
consumption (for heating) can be achieved, between 800 and 900 GJ over 30 years for the
terraced house. This is even lower than what is achieved with the EL²EP-optima.
193
CHAPTER 8
Position of passive house concepts
40000
20000
0
-20000
-40000
all optima of EL²EP
PH 70% heat recovery
-60000
-80000
800
1300
1800
2300
2800
Figure 8.22: Terraced dwelling, utilisation period of 30 years, low energy price scenario: All
optimal solutions from EL²EP-project are given for net present value vs. total primary energy
consumption (dark grey) as well as the passive house variants with a heat recovery
efficiency of 70% (light grey) and 90% (black).
Position of passive house concepts
40000
Terraced house, utilisation period = 30 years, high energy price scenario
20000
0
-20000
all optima of EL²EP
-40000
-60000
-80000
800
1300
1800
2300
2800
Figure 8.23: Terraced dwelling, utilisation period of 30 years, high energy price scenario: All
optimal solutions from EL²EP-project are given for net present value vs. total primary energy
consumption (dark grey) as well as the passive house variants with a heat recovery
efficiency of 70% (light grey) and 90% (black).
194
EL²EP-PROJECT
However, most variants are not economically viable for none of the energy price scenarios.
From the 33 variants, only 2 have a positive net present value over 30 years in case of a
discount rate of 4% and the high energy price scenario. Only at higher energy prices and/or
lower discount rates, more PH-variants will become economically viable. They demand an
total investment cost of 160,000 to 165,000€ for the wood frame structure and the building
envelope and 12,000€ for the installation. This represents an extra investment cost of 35,000
to 40,000€ for the building envelope compared to the reference. The installation cost
however, lies 4,600€ lower than the installation of the reference case. Compared to the
economically optimal solution (20,000€ extra for the building envelope and minus 2,0004,000€ for the installation), the investment cost for the passive houses is much higher, but it
results in much lower energy consumption: 870-900 GJ for the passive houses versus 1500
GJ for the economic optimum. For the EL²EP-optima with lowest energy consumption, the
extra cost for the building envelope is with 25,000 to 30,000€ somewhat lower than the PHvariants, but through the application of heat pumps or CHP in combination with a mechanical
ventilation system, the installation cost lies 12,000 to 15,000€ higher than the reference and
16,600 to 19,600€ higher than the PH-variants.
As in PH-concepts, a heating element is integrated in the ventilation system, a large cost
reduction can be realised with at the same time, large energy savings. However, by
combining ventilation and heating in one system, the control of the indoor air temperature
becomes more difficult as the air flows needed for ventilation differ from the air flows needed
for heating. This could also be concluded from the measurements in 2004 in the passive
house near Ghent, Belgium. With an end energy consumption for heating less than 9
kWh/m²a and a total primary energy consumption less than 30 kWh/m²a, this is an extremely
energy efficient dwelling. However, the indoor climate remains a weak point. In winter, the
temperature in the living room is rather low (average of 19°C), whereas in summer, the
temperature is rather high (average of 24°C). In the sleeping rooms on the contrary, rather
high temperatures were measured. Even in winter the average temperature is above 20°C.
Furthermore due to very low air flows, the relative humidity is high (average of 51% and
average vapour difference between indoor and outdoor of 600Pa) (De Meulenaer et al.
2005(1)). The future challenge for passive houses therefore should be to combine the
extremely low energy consumption with a good indoor comfort in winter and summer.
Zero Energy houses
A more extreme concept of energy saving house is the zero energy house. Generally, a zero
energy house represents a building concept in which the energy consumption of the house is
completely covered by own energy production, at least on annual basis. The latter, ‘on
195
CHAPTER 8
annual basis’, is an important prerequisite and distinguishes this concept from an energy
autarkic house for which the energy production needs to cover the energy consumption at
every moment. In contrast to an energy autarkic house, a zero energy house is coupled to
the electricity grid and delivers the produced electricity to the grid at moments of
overproduction to take it back at moments of underproduction. The outcome of a zero energy
house can slightly differ if the zero balance is defined on the level of end energy
consumption, primary energy consumption or energy cost. However, in case of only electrical
energy consumption with forward and backward counting electricity counter, the three
definitions lead to the same result.
As a final step within the EL²EP-project, zero energy houses are created. Starting point are
the passive house variants of the terraced dwelling. The simulations for section 8.7.1 deliver
the annual end energy consumption for heating, ranging from 1425 kWh/a to 1611 kWh/a. As
the energy production also has to cover the energy consumption for hot water, lighting and
electrical appliances, some scenarios are applied for domestic hot water and electricity
consumption for households. Hens and Verdonck (1997) developed calculation modules for
domestic hot water and electricity consumption based on measurements in situ. The
calculation modules consider the magnitude of the building and of the households and the
presence of different appliances. They also contain business as usual scenarios as well as
energy saving scenarios that take into account the use of energy efficient appliances, low
energy light bulbs, low flow showerheads, hotfill washing machines and dishwashers, etc.
With these modules, the annual energy consumption for domestic hot water, lighting and
electrical appliances is calculated for the different scenarios. In addition, also the extra
investment cost for the energy efficient scenarios is calculated based on consumer prices of
January 2007 (Vandenborre 2007). Depending on the scenarios, the energy consumption for
domestic hot water, lighting and electrical appliances ranges from 3900 to 9500 kWh/a.
For the energy production, both thermal solar collectors and photovoltaic modules are
considered. For the contribution of thermal solar collectors, calculations are done for collector
areas of 4m² up to 20m², based on the calculation module of the Flemish EPB. For the
contribution of photovoltaic modules, the PV-calculation module of the EL²EP-project is
applied (Vermeyen 2007). This calculation module contains data on 169 different PV-cells
and 176 different invertors. Starting point for this module is the roof surface, including the
exact location of roof windows. Based on these data, the module calculates the available roof
area and determines a list of 25 best PV-types to cover the roof in the most optimal way,
taking into account both the dimensions and the performance of the cells. In a next step, the
most appropriate invertor is determined. Finally, the annual electricity production is
calculated for the Test Reference Year of Uccle, Belgium, taking into account all boundary
conditions. In addition, the investment cost is calculated with and without subsidies, as well
196
EL²EP-PROJECT
as the green certificates, the electricity cost saving (based on 0.15€/kWh) and the net
present value over the life span of the PV modules (25 years). Also the embodied energy for
PV-modules and invertor is calculated. The calculation module gives the opportunity to
determine the energy production and consumption per hour, including or excluding the
presence of a battery. This way, also energy autarkic houses can be simulated.
Table 8.6 presents the building variants with the best results, being a combination of 30-31m²
PV-modules and 4-8m² thermal solar collectors.
Electricity consumption
Electricity
Solar
PV-
Extra
NPV30
NPV25
[kWh/a]
Produc-
collec-
system
invest-
building
solar
tion
tor
ment
systems
cost
Heating
Hot
House
water
-hold
[kWh/a]
[m²]
[m²]
[€]
[€]
[€]
1
1427
984
1943
4868
4
30
44558.12
-20412.74
-1014.97
2
1427
494
2876
4868
8
30
48124.62
-20412.74
-6943.97
3
1449
620
2876
5065
6
31
42074.00
-18867.06
-4325.00
4
1463
620
2876
5065
6
31
42074.00
-42613.60
-4325.00
5
1467
620
2876
5065
6
31
42074.00
-53898.58
-4325.00
6
1478
620
2876
5065
6
31
42074.00
-11390.64
-4325.00
7
1480
620
2876
5065
6
31
42074.00
-5303.42
-4325.00
8
1525
620
2876
5065
6
31
42074.00
-30764.86
-4325.00
9
1527
620
2876
5065
6
31
42074.00
-25067.20
-4325.00
10
1528
620
2876
5065
6
31
42074.00
-40629.16
-4325.00
11
1529
620
2876
5065
6
31
42074.00
-35122.30
-4325.00
12
1548
620
2876
5065
6
31
42074.00
-11759.32
-4325.00
13
1551
620
2876
5065
6
31
42074.00
-5672.42
-4325.00
Table 8.6: Results for the zero energy variants of the terraced dwelling of the EL²EP-project:
Electricity consumption and production on annual basis are given, together with the needed
surface for solar collectors and PV-modules. The extra investment cost for the energy
producing systems is given (excl.building!), as well as the net present value over 30 years for
the building and the net present value over 25 years of the solar systems. The overall
investment cost for the building, installations and energy systems is given in Annex H.
The last three columns give information on the investment cost and net present value. The
extra investment cost represents the cost for the solar collectors, PV-modules and the extra
cost for more energy efficient appliances. The cost for the building is not included in the extra
investment cost. The last but one column gives the net present value over 30 years for the
197
CHAPTER 8
building as PH-concept, whereas the last column gives the net present value of the solar
systems over their lifespan (25 years). NPV30 is calculated taking into account that all
energy consumption is covered by the solar systems. All values are valid for the high energy
price scenario and a discount rate of 4%. More details are presented in Annex H.
For the variants from row 2 to 13, the south oriented roof surface of the terraced house
(sloped roof + small flat roof) is just not large enough to integrate 2 roof windows, 30-31m² of
PV-modules and 6-8m² of solar collectors. A small part of the energy producing systems
needs to be installed on an extra surface, e.g. on a terrace or in the garden. Only in case of
very energy efficient electrical appliances, (hardly on the market yet), resulting in an
electricity consumption < 2000kWh/a for lighting, cooking and electrical appliances, the south
oriented roof surface of the terraced house is large enough to integrate the roof windows and
PV-modules on the sloped roof and 4m² of thermal solar collectors on the small flat roof, thus
covering the total electricity consumption of 4350 to 4460 kWh/a by an annual production of
4463 kWh. This scenario is represented in the first row of table 8.6. The building variant is
the same as in row 2. Some passive house building variants could come to energy breakeven by combining 30-32m² PV-modules with 12-14m² thermal solar collectors, but for these
solutions even more extra surface is needed. With the embodied energy of 30m² PVmodules estimated at 220GJ (Vermeyen 2007) and the PH-building concept representing ca.
440GJ embodied energy, the PV-modules are responsible for one third of the overall
embodied energy.
It is clear that none of the zero energy houses that are analysed here, are economically
viable within the assumptions for energy prices and discount rate. Only in case of even
higher energy prices and/or lower discount rates, some of the variants might become costeffective. However, the largest barrier is the extremely high investment cost of these houses,
being with the current prices, 70,000 to 90,000€ extra compared to the reference case of the
terraced house and 35,000 to 50,000€ extra compared to the passive house concept.
8.8
Conclusions
Firstly, the EL²EP-project results in the same economic optimum and the same logical
hierarchy of energy saving measures as found in the BIM-project. Best results can be
achieved with compact houses, but in case of modern architecture with fragmented plan, the
lack of compactness should be compensated by more extensive insulation measures. For
this kind of houses it is important to realise that, because of their large heat loss area, even
very elaborate energy saving measures are economically viable.
In contrast to what is sometimes assumed, there is no reason for great concern about the
embodied energy of the energy saving measures. The embodied energy strongly increases
198
EL²EP-PROJECT
with increasing insulation level, but at the same time large energy savings are realised during
the utilisation phase of the building. Regardless of the insulation materials and installation
systems, applied in extremely low energy dwellings, the total embodied energy of the energy
saving measures represents less than 10% of the total primary energy saved by these
measures and the energy payback time is for all case less than 2 years. This means that
reducing the energy consumption during the utilisation phase still should remain the first
concern.
Also concepts for extremely low energy dwellings have been determined and compared with
existing concepts, but none of them appear to be economically viable for the current energy
prices or discount rates. Only in case of much higher energy prices, some of these concepts
will become cost-effective over 30 years. However, the largest barrier for all these concepts
is the extremely high investment cost. Without financial support or incentives, these concepts
will be limited to a minor part of consumers with a high environmental consciousness that is
willing to invest such a large budget in an extremely energy saving house.
199
CONCLUSIONS AND FURTHER RESEARCH
CHAPTER 9.
Since the oil crisis of the early seventies, energy consumption in buildings is a hot item.
During the years, the interest slowly shifted from reduction of energy consumption during the
utilisation phase towards the overall sustainability of buildings over their lifetime, not at the
least challenged by the problem of climate change. Many research initiatives have been
launched to improve energy efficiency and sustainability in buildings. However, all this
research is only useful, if at the end, it results in effective changes in the environmental
performance of the building stock. This demands large changes on all levels of society.
However, up to now, no one has the right answer how to effectively induce them. Also this
work shows only a glimpse of the complexity of sustainable development, but hopefully, it is
one small step further towards a more sustainable world. Therefore, in this last chapter, the
main achievements are presented together with perspectives for future research.
9.1
Main results and conclusions
Subject of the research project here presented was the development and implementation of
a global methodology to develop and evaluate on a scientific basis, residential buildings that
are globally optimised from the point of view of energy, ecology and costs. The application of
the methodology resulted in concepts and guidelines for globally optimised buildings.
Developed global optimisation methodology
Basic principle for the developed methodology is a well-founded evaluation of the
environmental impact and financial cost during the whole life cycle of the building and its
installations. The methodology consists of three pillars, closely linked to each other: (1) the
multi-objective optimisation strategy that combines the technique of genetic algorithms with
the concept of Pareto optimality, (2) a life cycle inventory model for buildings as a whole and
(3) a cost assessment model that evaluates the economic impact of the building concepts
from the point of view of the private building owner. The global optimisation methodology is
developed in the frame of the GBOU-EL²EP-project, a research project funded by the
Flemish government that aimed at the development of extremely low energy and low
pollution dwellings through life cycle optimisation. It also proved its usefulness for the BIMproject, a technical-economical study on the cost-effectiveness of energy saving investments
in buildings in the Brussels Capital Region. Normally, the methodology will be applied again
in the near future, in a project commissioned by the Flemish Energy Agency VEA on the
201
CHAPTER 9
economic viability of strengthening the legal requirements set by the Energy Performance
Regulation for buildings EPB.
Several conclusions could be drawn from the application of the developed methodology in
the EL²EP- and BIM-project.
Life cycle inventory of buildings and embodied energy
A straightforward calculation algorithm has been developed to calculate the life cycle
inventory of a whole building. However, despite the straightforwardness, the uncertainty is
quite high, since all data are based on more or less uncertain assumptions. Therefore, a
perturbation analysis and an uncertainty analysis by Monte Carlo simulations have been
performed on the inventory model. Both studies showed that the sensitivity for errors and the
propagation of errors is limited and that the errors of the different input data neutralise each
other somehow.
Of more importance, however, are the results from the contribution analysis of the life cycle
inventory. They showed the relative small importance of the embodied energy of a building,
compared to the energy consumption during the utilisation phase. This is even more valid
when comparing the embodied energy of energy saving measures with the energy savings
they realise. In most cases, the embodied energy represents less than 10% of the primary
energy savings over 30 years. Only extremely low energy buildings might have a total
embodied energy higher than the energy use of the utilisation phase. However, the sum of
both remains small and the energy savings realised with these dwellings are large, compared
to the energy consumption of average dwellings.
A remarkable conclusion from the life cycle inventory is also that the embodied energy for
both massive and light weight buildings is comparable. All these results lead to the
conclusion that considerations on the embodied energy of extremely low energy houses can
be an interesting issue for future research, but that in the first place, effort should be paid to
the reduction of the energy consumption during the utilisation phase, as this phase still has
the largest potential for improvement.
Economic optimum and hierarchy of energy saving measures
In both the EL²EP-project and the BIM-project, the methodology has been applied to a
number of buildings that are representative for the Belgian building stock. The parameters for
optimisation were related to energy saving measures, both for the building envelope and the
heating system. The optimisation itself has been performed in two steps. In the first step,
only envelope-related energy saving measures have been considered, such as insulation,
better glazing, glass area, sun shading, air tightness and natural ventilation scenarios. In the
second step, the measures on the building envelope have been combined with system202
related measures. This included systems for distribution, emission, production and storage of
heat, systems for local electricity production and control systems.
From the optimisation and analysis of the combinations of envelope-related and installationrelated energy saving measures, an economic optimum could be deduced that appeared to
be very robust for the assumptions on energy prices, discount rate and utilisation period. It
consists of:
-
an insulation level K25-30 for most common houses (Umean ≈ 0.3-0.4 W/m²K for
compactness of 1.5-2). Only for houses with a very fragmented plan (compactness <
1), the economically optimal insulation level decreases, down to K20 (Umean ≈ 0.2
W/m²K);
-
combined with a high efficiency or condensing boiler on gas or fuel, variable water
temperature, high temperature radiators and a natural ventilation system.
The energy performance can further be improved by means of a mechanical ventilation
system with a good performing heat recovery unit or an air-to-water heat pump. The
seasonal performance factor (SPF) of an air-to-water heat pump is less than that of a soil-towater heat pump, but the investment cost is much lower. Therefore, the air-to-water heat
pump turns out to be economically viable more quickly than the soil-to-water heat pump. For
most dwellings, these measures are just at the limit of economic viability.
Further decreasing the energy consumption is possible by implementing a better insulation
level (K15-20), by combining this very good insulation level with a heat pump and with
mechanical ventilation with heat recovery or even by combining it with a CHP-system or a
solar driven system. This leads to very low total primary energy consumption and annual
energy costs, but these solutions are far beyond the economic optimum for most dwellings.
Only for houses with a very low compactness, which are very energy-devouring in a badly
insulated version, these extended combinations of energy saving measures showed to be
economically viable at higher energy price evolutions.
In a nutshell, the most logical hierarchy for energy saving measures in dwellings is then:
1. Invest in a good insulation level with good air tightness and a well designed
ventilation system. The economic optimal insulation level K25-K30 (Umean ≈ 0.30.4 W/m²K for compactness of 1.5-2) lies, at least in Belgium, far beneath the
legal insulation requirement K45 (Umean ≈ 0.5-0.6 W/m²K).
2. Select a well performing heating system, at least a high efficiency boiler or
condensing boiler. If the budget is available, a heat pump is a good alternative:
better performing, but at a higher price.
203
CHAPTER 9
3. Finally, if for any reason, further steps to decrease the energy consumption are
wanted and the budget is available, there are the options for limiting the
ventilation losses through a balanced ventilation system with heat recovery and/or
installing a solar driven system (solar collector or PV-system). However, one must
realise that, although these measures improve the energy performance, they are
far beyond the economic optimum and mostly beyond economic viability.
This guideline is valid for both new construction and renovation. Some insulation measures,
such as insulation of the façade or the floor, are less easy to implement in case of retrofit or
at much higher price than for new construction. So, the economic viable insulation level will
normally be somewhat higher for retrofit than for new buildings, but the hierarchy of
measures remains identical and is independent of energy price evolutions and discount rate.
For the calculations, assumptions had to be made on discount rate, energy price evolution
and material cost evolution and these assumptions have an impact on the economic viability.
However, the results appeared to be very robust for the assumptions. The economic optimal
combination of energy saving measures is not only independent of the cost scenarios, it also
remains economically viable for all analysed cases. Furthermore, the hierarchy of energy
saving measures is independent of these cost scenarios. Obviously, the economic viability of
some measures, such as the application of heat pumps or mechanical ventilation with heat
recovery, will depend on the cost scenarios, but in contrast to what sometimes is assumed,
there is no shift at all in this hierarchy.
Extremely low energy dwellings
Finally, extremely low energy and low pollution building concepts have been developed. By
combining a very high insulation level (Umean ≈ 0.13-0.2) with a heat pump and mechanical
ventilation with heat recovery or even with solar collectors, a PV-system or a CHP-system,
extremely low energy consumptions can be realised. This was not only proven by the
simulations, but also by in situ measurements on passive houses. However, up to now, these
building concepts can not be realised in an economically viable way for the present energy
prices and discount rates. Only in case of much higher energy prices, some will become
cost-effective over 30 years. Furthermore, in situ measurements show that, despite the very
low energy consumption, the comfort level of these dwellings still needs to be improved.
However, the largest barrier for all these concepts is the extremely high investment cost.
Without financial support or incentives, they will be limited to a minor part of consumers with
a high environmental consciousness that is willing to invest such a large budget in an
extremely low energy house.
204
9.2
Perspectives for future research
This work has only been a small step in the large search for sustainable buildings and it
certainly is not the last one. Several paths for further research can be proposed:
-
The EL²EP-project focused on rather common heating systems that are mainly
developed for averaged levels of energy consumption. None of these systems is well
adapted to extremely low energy houses. In passive houses, heating is provided by
incorporating the heating system into the ventilation system, but in situ
measurements show that here, thermal comfort remains a weak point. Therefore, one
path for further research could be an in depth search for more appropriate, well
performing systems that not only have a very low energy consumption, but also are
capable of guaranteeing good thermal comfort and indoor air quality all over the year.
-
A second direction of research is inspired by the report of the Federal Planning
Bureau on the climate policy after 2012 (Devogelaer et al. 2006). Several scenarios
for emission reduction for 2020 and 2050 are analysed. However, the proposed
scenarios for the residential sector appear to be at best theoretical reduction
potentials, but certainly not technical and even less economical reduction potentials.
The work in this PhD research on the energy saving potential for dwellings could
serve as a more reliable input for a macro-economic study of the technical and
economical emission reduction potential in the residential sector.
-
Further research could also focus on the minimisation of the embodied energy of
extremely low energy dwellings. Up to now, the energy saving potential in buildings is
extremely high and effort should first go to increase the effective energy performance
of buildings during their utilisation phase. However, at a very long term, it might be
thinkable that only extremely low energy dwellings will be constructed. To be
prepared for that moment, the focus could shift towards the development of buildings
that combine low embodied energy with extremely well energy performance during
use of the building.
205
SAMENVATTING
SAMENVATTING
Inleiding
Sinds de oliecrisis van begin jaren zeventig is energiegebruik in gebouwen steeds meer een
aandachtspunt geworden. Door de problematiek van de klimaatsverandering is deze
aandacht vanaf de jaren negentig bovendien stilaan verschoven naar globale duurzaamheid
van gebouwen over hun volledige levensduur.
Een van de huidige trends betreft de ontwikkeling van extreem lage energiewoningen,
gaande van passiefhuizen met een dusdanig lage netto energiebehoefte dat een traditioneel
verwarmingssysteem niet meer nodig is, over nulenergiewoningen waarin het energiegebruik
op jaarbasis volledig wordt gedekt door de energieproductie, tot energie-autarke woningen
die losgekoppeld zijn van elk extern toevoernet en volledig instaan voor hun eigen
energiebehoefte. Deze concepten worden soms als dé ultieme duurzame gebouwen
beschouwd, maar, in tegenstelling tot traditionele gebouwen, vragen ze zeer veel
technologie, met een veel hogere ingebouwde energie en pollutie, een hoger gebruik van
grondstoffen en een veel hogere investeringskost tot gevolg. En, hoewel deze gebouwen
zeer weinig energie gebruiken tijdens de gebruiksfase, wordt zelden of nooit aangetoond dat
de globale balans van energie, emissies en kosten over hun levenscyclus positief is.
Een
belangrijke
methode
voor
de
beoordeling
van
globale
milieu-impact
is
levenscyclusanalyse (LCA). Echter, meestal wordt LCA toegepast op materialen en
componenten, terwijl grotere gehelen, zoals gebouwen, zelden voorwerp zijn van een
gedetailleerde LCA. Ook wordt LCA vooral toegepast voor kwantificatie van in- en
uitstromen, maar zelden voor het optimaliseren van beslissingsprocessen.
Bovendien is optimalisatie van een gebouw als geheel een zeer complex probleem door het
grote aantal parameters en variabelen, de niet-lineaire relaties en tweedeorde effecten.
Evolutionaire optimalisatietechnieken, zoals Genetische Algoritmes (GA), worden steeds
vaker toegepast omwille van hun potentieel voor dergelijke complexe problemen. Zij halen
hun inspiratie bij de genetische processen van biologische organismen, die over de
generaties heen evolueren volgens de principes van natuurlijke selectie en ‘survival of the
fittest’. Door dit proces te imiteren zijn zij in staat oplossingen te ontwikkelen voor reële
problemen. Hun toepassing voor gebouw-gerelateerde problemen is echter nog zeldzaam.
In dit onderzoek wordt getracht een methodologie op te stellen die toelaat gebouwconcepten
te ontwikkelen en te evalueren vanuit het oogpunt energie, ecologie en kosten. Basisprincipe
is een wetenschappelijk gefundeerde evaluatie van de milieu-impact en kosten over de hele
levensduur van een gebouw en zijn installaties, door koppeling van LCA en kostenbatenanalyse met geavanceerde optimalisatietechnieken.
207
Literatuurstudie
De methodologie is opgebouwd uit drie sterk gekoppelde pijlers: (1) de multi-objectieve
optimalisatietechniek die genetische algoritmes combineert met het Pareto-concept, (2) het
model voor levenscyclusinventarisatie van een gebouw als geheel en (3) de kostenbatenanalyse.
Multi-objective optimalisatie
Genetische algoritmes
Door de complexiteit van vele hedendaagse problemen, wordt bijna elk ingenieursdomein
geconfronteerd met optimalisatieproblemen waarin een groot aantal parameters, variabelen,
objectieven en randvoorwaarden betrokken zijn. Zo ook het probleem van optimalisatie van
extreem lage energiewoningen. Het is nooit de bedoeling geweest om een vergelijkende
studie te maken van alle mogelijke optimalisatietechnieken, wel om de mogelijkheden van
evolutionaire optimalisatietechnieken, en in het bijzonder van genetische algoritmes, te
testen voor gebouw-gerelateerde optimalisatieproblemen, aangezien deze technieken nog
zelden zijn toegepast binnen dit domein. Deze techniek laat toe om een meer gerichte
zoektocht te maken in het geheel van mogelijke oplossingen. In tegenstelling tot andere
optimalisatietechnieken rekent een genetisch algoritme niet met 1 oplossing, maar met een
set van mogelijke oplossingen. Gevolg hiervan is dat een optimalisatierun niet resulteert in 1
optimale oplossing, maar in een groep van optimale of suboptimale oplossingen. De techniek
is geïnspireerd op de evolutieleer van Darwin en werkt grosso modo volgens het principe van
‘survival of the fittest’. De ‘fittest’ is in de context van dit onderzoek al naargelang het
uitgangspunt, de woning met het laagste energieverbruik, de laagste pollutie en/of de laagste
kostprijs. Het principe bestaat erin dat van een referentiewoning door variatie van de
parameters (isolatie, beglazing, luchtdichtheid,…) verschillende varianten gemaakt worden.
Een groep van varianten samen vertegenwoordigt 1 populatie. Elke variant uit de populatie
wordt beoordeeld op zijn prestaties qua energieverbruik, pollutie en kostprijs. Uit deze
populatie worden telkens, twee per twee, varianten gekozen die door onderlinge combinatie
en mutatie nieuwe varianten van de referentiewoning voortbrengen. De kans dat een variant
gekozen wordt voor ‘voortplanting’ hangt rechtstreeks af van hoe goed hij presteert (cfr.
survival of the fittest). Op deze manier evolueert men in principe na verschillende ‘generaties’
naar de meest optimale oplossingen.
Concept van Pareto-optimaliteit
Naast optimalisatie van meerdere variabelen wordt ook geoptimaliseerd voor meerdere
objectieven (energie, pollutie en kosten) en wordt rekening gehouden met bijkomende
208
SAMENVATTING
randvoorwaarden, zoals het zomercomfort, de binnenluchtkwaliteit, etc. Van cruciaal belang
is hiervoor het beslissingsproces. Er zijn verschillende mogelijkheden:
-
Beslissing vóór optimalisatie: hierbij weegt men de verschillende objectieven ten
opzichte van elkaar en herleidt men de meerdere objectieven tot een gewogen som
alvorens te optimaliseren. Voordeel is dat nog slechts voor 1 objectief (de gewogen
som) moet geoptimaliseerd worden; nadeel is dat voldoende achtergrondkennis
nodig is voor de weging. Bovendien zijn de gevonden oplossingen enkel optimaal
voor de gekozen wegingsfactoren.
-
Beslissing na optimalisatie: hierbij wordt elk objectief als gelijkwaardig behandeld
tijdens de optimalisatie en wordt de wisselwerking tussen de objectieven bepaald
door de niet-gedomineerde oplossingen te zoeken (= het Pareto-front). Een oplossing
is niet-gedomineerd als er geen andere oplossing is binnen de set van oplossingen
die gelijk of beter presteert in één objectief zonder slechter te presteren in het andere
objectief. Deze aanpak is gekozen binnen dit onderzoek.
De literatuur over evolutionaire multi-objectieve optimalisatiemethodes omvat twee
domeinen: artikels waarin de algoritmes zelf het belangrijkste onderzoeksobject zijn (Coello
1996, 2000 and 2002, Deb en Goldberg 1989, Fonseca en Fleming 1993 en 1995, Hancock
1994, Horn et al 1994, Michalewicz 1996 en 1999, Smith en Coit 1995, Van Veldhuizen 1998
en 1999,…) en artikels over een toepassing van de techniek op een specifiek
optimalisatieprobleem (Asiedu et al 2000, Erickson et al 2002, Fleming en Purshouse 2002,
Lotov 2001, Ozturk et al 2004(1) and 2004(2), Wang et al 2005(1) and 2005(2),…). Het
onderzoek dat hier wordt voorgesteld, behoort tot deze laatste groep. Tot hier toe is deze
techniek nog maar weinig toegepast in gebouw-gerelateerde domeinen. Meestal betreft het
optimalisatie van componenten van installaties (Asiedu et al. 2000, Wang en Jin 2000,
Fleming en Purshouse 2002,…). Sporadisch is het al toegepast voor de optimalisatie van het
energiegebruik in gebouwen als geheel (Coley en Schukat 2002, Ozturk et al. 2004(1),
2004(2), Wang et al. 2005(1), 2005(2)).
Levenscyclusanalyse
Algemeen
Sinds begin jaren 70 zijn door een groeiend milieubewustzijn verschillende methodes
ontwikkeld voor de beoordeling van de milieu-impact van materialen en producten. In deze
evolutie heeft de industrie een grote rol gespeeld, maar dit heeft ook een impact gehad op de
methodes en onderzoeksonderwerpen zelf: de focus ligt meestal op productontwikkeling en
veel minder op grotere gehelen, zoals gebouwen.
209
De meeste methodes zijn gebaseerd op een vorm van levenscyclusanalyse (LCA), zoals die
sinds eind jaren 90 is vastgelegd in de ISO 14000 reeks. Deze normen bepalen slechts het
kader voor LCA. Een belangrijke eerste stap in LCA is het definiëren van doel en omvang
van de studie, omdat dit de systeemgrenzen van het bestudeerde product in tijd en ruimte
vastlegt. De volgende stap is een levenscyclusinventarisatie (LCI). Hierbij wordt een model
voor het product opgesteld dat het product opsplitst in deelprocessen, onderling gekoppeld
door productstromen. Eindresultaat van een LCI is een inventaris van alle inkomende en
uitgaande stromen horende bij het product. In een volgende stap worden de verschillende
stromen gewogen en gecombineerd tot milieu-indicatoren, die toelaten de levenscyclus van
het product te beoordelen voor een aantal impactcategorieën, zoals klimaatsverandering,
ozonaantasting, verzuring, menselijke gezondheid, etc. Voordeel van deze indicatoren is hun
gebruiksvriendelijkheid en de schijnbare eenvoud voor interpretatie. Maar, zoals bij elke
toepassing van wegingsfactoren, kan het resultaat sterk afhankelijk zijn van de gemaakte
aannames. Een laatste essentiële stap in LCA die interfereert met elke voorgaande, is de
levenscyclusinterpretatie.
Verschillende bestaande methodes voor beoordeling van milieu-impact zijn gebaseerd op dit
LCA-kaderwerk, maar variëren in graad van complexiteit en detail voor de gebruikte data,
processen en milieu-aspecten. Voorbeelden zijn procesketenanalyse en input-outputanalyse.
Een belangrijk aspect bij levenscyclusinventarisatie is de beschikbaarheid en kwaliteit van de
data voor input- en outputstromen. Verschillende inspanningen zijn al gedaan op het vlak
van verzamelen en documenteren van data. Echter, studies over onzekerheidsanalyses of
over de invloed van de datakwaliteit op de eindresultaten zijn zeer zeldzaam in de literatuur.
Heijungs et al. (2005) bespreekt vijf numerieke methodes om de interpretatie van de LCIresultaten te ondersteunen. De eerste drie, een contributie-, een perturbatie- en een
onderzekerheidsanalyse, zijn toegepast op de resultaten van dit onderzoek.
Omwille van de coherentie in methodologie en datastructuur, is voor dit onderzoek enkel
gebruik gemaakt van de commerciële LCA databank ecoinvent2000, ontwikkeld door een
groep van Zwitserse onderzoeksinstellingen (www.ecoinvent.ch). Deze databank bevat
gegevens voor meer dan 2500 processen die representatief zijn voor de Europese context.
LCA van gebouwen
In de literatuur en in de bouwpraktijk zijn verschillende methodes voorhanden om de milieuimpact van gebouwen te beoordelen, gaande van checklists over labels tot gedetailleerde
LCA modellen. IEA-Annex 31 (www.annex31.org) heeft eind jaren 90 een analyse gemaakt
van de bestaande methodes voor gebouwen en van de verschillende aspecten van milieuimpact van gebouwen. Hieruit bleek dat de toepassing van LCA op gebouwen niet zo evident
is. Ten eerste zijn gebouwen, in tegenstelling tot veel industriële producten, geen
210
SAMENVATTING
serieproducten. Dit maakt de extrapolatie van een LCA-studie op 1 gebouw niet
vanzelfsprekend en bemoeilijkt ook de vergelijking van verschillende gebouwen. Bovendien
stroken sommige aannames binnen LCA niet met de karakteristieken van een gebouw, zoals
de aanname van tijdsstabiliteit die impliceert dat als een product zijn levenseinde bereikt, het
afval behandeld wordt op dezelfde wijze als het bij het begin van zijn leven gebruikelijk was.
Voor gebouwen met een levensduur van 80 jaar en meer is dit een zeer onrealistische
aanname. Bovendien ondergaan de meeste gebouwen tijdens hun levensduur meerdere
renovaties die kunnen leiden tot grondige aanpassingen aan het gebouw. Bijgevolg leidt het
inrekenen van de volledige levensduur van een gebouw, inclusief de renovaties en de
eindfase, tot zeer grote onzekerheden. Bijkomend heeft de gebruiksfase van een gebouw
een dermate hoge impact, zeker voor energiegebruik en emissies, dat algemeen aanvaard
wordt om voor gebouwen de verschillende fases (productie-transport-constructie-gebruikafbraak) afzonderlijk te bekijken of ze te groeperen in ingebouwde impact (als gevolg van
productie-transport-constructie) en impact tijdens de gebruiksfase.
Een laatste belangrijk aspect bij gebouwen is de functionele eenheid waarvan de milieuimpact wordt bepaald. In de literatuur is een groot aantal gevallenstudies te vinden over de
milieu-impact van bouwmaterialen of –componenten (Börjesson en Gustavsson 2000,
Erlandsson et al. 1997, Jönsson 2000, Weir en Muneer 1998). Echter, door de functionele
eenheid te beperken tot een materiaal of een component, kunnen een aantal
randvoorwaarden waaraan een gebouw zou moeten voldoen (thermisch comfort,
binnenluchtkwaliteit, etc.) of interacties tussen componenten en hun impact op de prestaties
van het gebouw niet in rekening gebracht worden. Daarom zijn in dit onderzoek gebouwen
als geheel geanalyseerd zodat de globale prestatie van het gebouw kan beoordeeld worden
en enkel gebouwen met eenzelfde prestatieniveau vergeleken worden.
Kosten-batenanalyse
Algemeen
Kosten-batenanalyse is een techniek die algemeen wordt toegepast om de wenselijkheid van
investeringen te beoordelen. Globaal gezien kan een kosten-batenanalyse verder gaan dan
enkel het inrekenen van de reële financiële kosten van een project en kan het ook andere
effecten, zoals milieueffecten, monetariseren en inrekenen. In dit onderzoek echter, wordt de
meer strikte benadering van kosten-batenanalyse toegepast, enkel voor de financiële
evaluatie van energiebesparende investeringen door een individuele gebruiker. De milieuimpact is immers al meegerekend via de levenscyclusinventarisatie. Deze kostenbatenanalyse wordt niet als enige beslissingscriterium gebruikt, maar wordt als economisch
objectief opgenomen in de multi-objectieve levenscyclusoptimalisatie.
211
Financiële beoordelingscriteria
Belangrijk voor een kosten-batenanalyse is het verdisconteren van alle kosten en baten.
Toegepast op gebouwen, veronderstelt dit initiële investeringen in het gebouw en
vervangingsinvesteringen na een bepaalde tijd die samen leiden tot een bepaalde jaarlijkse
energiekost die constant wordt verondersteld in de tijd zolang geen grondige renovaties
worden toegepast. De totale actuele kost en de netto contante waarde kunnen als volgt
berekend worden, waarbij de totale actuele kosten alle huidige en toekomstige kosten als
absolute waarden neemt, en de netto contante waarde de kosten en baten ten opzichte van
een referentie bekijkt:
TAK = I 0 +
∑
I j (1 + rI ) j
j = x, y, z
⎡
NCW = − ⎢ I 0 +
⎢⎣
(1 + a) j
∑
j = x, y , z
n
K E (1 + rE ) i
K M (1 + rM ) i
− R0
+∑
+∑
(1 + a) i
(1 + a) i
i =1
i =1
n
I j (1 + rI ) j
(1 + a ) j
[S.1]
K M (1 + rM ) i ⎤ n ∆K E (1 + rE ) i
+ R0
+∑
⎥+∑
(1 + a ) i ⎥⎦ i =1 (1 + a ) i
i =1
n
[S.2]
met:
I0
Ij
KE
∆KE
KM
n
R0
rI
rE
rM
a
Andere
de initiële investering [€]
de vervangingsinvestering j op tijdstip time x, y of z [€]
jaarlijkse energiekost [€]
jaarlijkse energiekostbesparing vergeleken met de referentie [€]
jaarlijkse onderhoudskost [€]
beschouwde tijdsperiode voor evaluatie [jaar]
restwaarde van het gebouw op tijdstip n
verandering van de investeringskost bovenop inflatie
verandering van de energiekost bovenop inflatie
verandering van de onderhoudskost bovenop inflatie
discontovoet of reële intrestvoet, gecorrigeerd voor inflatie
gerelateerde financiële methodes
om
energiebesparende
investeringen
te
beoordelen, zijn de ‘internal rate of return’ die wordt gedefinieerd als de discontovoet waarbij
de netto contante waarde 0 is, de statische terugverdientijd (STVT) die bepaald wordt uit de
verhouding van initiële investeringskost en de jaarlijkse energiebesparing zonder rekening te
houden met actualisatie, en de dynamische terugverdientijd die de besparing op de
energiekost wel verdisconteert. De STVT wordt vooral gebruikt voor vergelijking van
maatregelen met een korte levensduur (< 5jaar), maar is minder geschikt als
beslissingsbasis voor maatregelen met een lange levensduur.
Een belangrijk aspect is de keuze van de discontovoet. Het drukt het standpunt van de
maatschappij uit en bepaalt, in het geval van energiebesparende investeringen in gebouwen,
212
SAMENVATTING
of rendabel is geld in deze maatregelen te investeren dan wel beter in andere economische
sectoren. Algemeen wordt de intrestvoet voor risicovrije investeringen als een goede maat
gezien
voor
de
discontovoet,
maar
omwille
van
de
grote
onzekerheden
over
langetermijnprojecties voor de intrestvoet, de evolutie van de inflatie en van de
energieprijzen is een gevoeligheidsanalyse noodzakelijk voor het beoordelen van de impact
van aannames voor discontovoet en energieprijzen op de uiteindelijke resultaten.
Rebound effect en beslissingsmodellen voor energiebesparende maatregelen
Gedurende de laatste jaren is veel inspanning geleverd door wetenschappers om technisch
haalbare en economisch rendabele energiebesparende maatregelen te bepalen. Toch blijkt
uit de praktijk dat het niet zo eenvoudig is om mensen aan te zetten tot energiebesparende
investeringen. Bovendien blijken de energiebesparende effecten van deze investeringen in
praktijk vaak veel lager dan theoretisch voorspeld. Oorzaak van dit laatste is het
reboundeffect dat ertoe leidt dat een deel van de energiebesparing wordt omgezet in een
hoger comfortniveau en dat consumenten het bespaarde geld besteden aan andere
goederen, die op zich ook energie verbruiken. Het bestaan van dit effect wordt algemeen
aanvaard. Echter is er in de literatuur geen eensgezindheid over de grootte van het effect
(Energy Policy 2000). Voor woningverwarming wordt het meestal begroot op een reductie
van 10 tot 30% van de theoretisch voorspelde energiebesparing. Het reboundeffect wordt
kort besproken aan de hand van enkele resultaten, maar is niet ingerekend in de
methodologie.
Bijkomend worden kort een aantal beslissingsmodellen uit de literatuur gepresenteerd die
trachten een verklaring te geven voor het falen van het energiebeleid (Dixit en Pindyck 1994,
Kahneman en Tversky 2000, STEM 2004). Deze worden niet verder uitgewerkt, maar
trachten vooral de resultaten uit dit onderzoek in een bredere context te plaatsen.
Methodologie
Multi-objectieve optimalisatie
Referentiewoningen
Doel is woningen te optimaliseren. Om representatieve resultaten voor België te krijgen,
worden een aantal referentiewoningen ontworpen volgens het statistische gemiddelde van
het Belgische woningenpark waarbij de geometrie wordt vastgelegd voor de niet geïsoleerde
toestand (zie Annex F en G). Enkel de glasoppervlakte kan variëren.
Energiebesparende maatregelen
De parameters voor optimalisatie betreffen energiebesparende maatregelen voor zowel de
gebouwschil als de installaties, maar het optimalisatieproces is opgesplitst in twee stappen.
213
In een eerst stap worden enkel maatregelen op de gebouwschil beschouwd, zoals isolatie,
betere beglazing, glasoppervlakte, zonwering, luchtdichtheid en scenario’s voor natuurlijke
ventilatie. Isolatie wordt voorzien in de daken, op de zoldervloer, in de gevels en onder de
vloeren en de isolatiedikte per schildeel kan variëren tussen 0cm en een maximale dikte (3040cm, afhankelijk van het schildeel). Ook kunnen verschillende isolatiematerialen worden
toegepast. Voor de beglazing is er keuze tussen verschillende glastypes met een U-waarde
tussen 2.8 W/m²K en 0.4 W/m²K en een g-waarde (directe en indirecte zonnewinsten) tussen
0.76 en 0.21. Voor de raamprofielen is er keuze tussen profielen van hout, PUR, PVC en/of
aluminium met een U-waarde tussen 6 W/m²K en 0.65 W/m²K. De glasoppervlakte kan
variëren tussen een minimum en een maximum waarde, afhankelijk van de minimale
vereisten voor daglichttoetreding en de constructieve beperkingen van de gevel. Om het
zomercomfort te kunnen beïnvloeden is er de mogelijkheid om interne of externe
beweegbare zonwering te voorzien. Vier luchtdichtheidsniveau’s worden beschouwd, van de
gemiddelde n50 voor nieuwbouw (SENVIVV 1998) over n
50
= 3/h en n50 = 1/h tot de
passiefhuisstandaard n50 = 0.6/h. Bijkomend zijn vier natuurlijke ventilatiescenario’s
toegevoegd om het effect van extra zomer- of nachtventilatie mee te nemen voor de
beheersing van het zomercomfort.
In de tweede optimalisatiestap worden de maatregelen voor de gebouwschil gecombineerd
met installatietechnische maatregelen voor distributie, afgifte, controle, productie en opslag
van warmte en systemen voor lokale elektriciteitsproductie. Hierbij worden zowel
hoogrendements-
en
condensatieketels
beschouwd,
als
warmtepompen
en
warmtekrachtkoppeling. De watertemperatuur kan constant zijn of afhankelijk van de
buitentemperatuur en zowel regeling met kamerthermostaat als met thermostatische kranen
worden beschouwd. Naast het natuurlijke ventilatiesysteem uit de 1ste stap worden in de 2de
stap
ook
ventilatiesystemen
met
mechanische
afvoer
en
volledig
mechanische
balansventilatie met of zonder warmteterugwinning beschouwd. Ook toepassing van
zonnecollectoren en fotovoltaische systemen wordt in sommige gevallen meegenomen in de
optimalisatie.
De methodologie zou eenvoudig kunnen worden toegepast voor gelijktijdige optimalisatie
van alle energiebesparende maatregelen, maar deze optie is niet weerhouden. Belangrijkste
reden is het verschil in levensduur tussen de gebouwschil en de installaties. Door de
langdurige impact van ingrepen op de gebouwschil op het energiegebruik, wordt verkozen
om eerst de gebouwschil te optimaliseren (minimale warmtevraag) en dan pas de meest
geschikte installaties te zoeken die aan deze vraag kunnen voldoen. Deze aanpak is
geïnspireerd op vroeger werk (Verbeeck en Hens 2005) waaruit al een logische hiërarchie
van energiebesparende maatregelen is afgeleid.
214
SAMENVATTING
Genetisch algoritme
Voor de optimalisatietechniek is geopteerd voor een combinatie van genetische algoritmes
met het Pareto-concept. Voor het genetisch algoritme is een chromosoom opgesteld dat, in
combinatie met de vaste geometrie die is aangenomen per gebouw, in staat is elke mogelijke
gebouwvariant volledig en ondubbelzinnig te definiëren. Het chromosoom voor een rijwoning
met deels plat, deels hellend dak, een niet verwarmde zolder en 12 ramen ziet er als volgt
uit:
Elementnummer
Plat Hellend Zolderdak
dak
vloer
1 2 3
4 5
6
7
8
9 10 11
12
13
14
15
16
Elementwaarde
17 0
25
1
1
65
11
3
2
89
19
3
18
Gevels
0
Vloer
8
2
Raamkarakteristieken
40
Luchtdichtheid
41
Zomerventilatie
42
0
2
1
Raam 1
Raam 2
…
Raam 11
Raam 12
17
18
19
20
…
37
38
39
89365
0
35287
1
…
20276
0
60379
Elk element in het chromosoom staat voor een bepaalde eigenschap: isolatiedikte,
isolatiemateriaal, bouwtype (massief of houtskeletbouw), glastype, raamprofieltype, type
zonwering, etc. Per raam wordt de variatie van de oppervlakte gegeven in cm² en of het een
toe- of een afname is. Alle waarden in het chromosoom kunnen random variëren tussen
voorgedefinieerde minimum en maximum waarden. Idem voor het chromosoom voor de
installatie.
Verwarming
Zonnecollectoren
PV systemen
Energie-
Pro-
Con-
Af-
Venti-
Warm
Aan-
Opp
Aan-
Opp
Element
drager
ductie
trole
gifte
latie
water
wezig
[m²]
wezig
[m²]
nummer
1
2
3
4
5
6
7
8
9
10
waarde
2
1
1
3
2
3
0
6
1
10
Genetische operatoren
Met deze chromosomen wordt een populatie samengesteld van bv. 100 random
gegenereerde varianten van eenzelfde gebouw. Op deze populatie wordt het genetische
proces van selectie, recombinatie en mutatie van chromosomen toegepast om zo een
nieuwe populatie te creëren. De kans dat een variant wordt geselecteerd voor voortplanting
wordt statistisch bepaald en is afhankelijk van zijn fitheid. Voor het bepalen van de juiste
waarden voor de genetische operatoren (populatiegrootte, recombinatiegraad, mutatiegraad,
generatiekloof, etc.) is een parameterstudie gedaan.
215
Kostfuncties, fitheidsfuncties en straffuncties
Energieprestatie, ecologische impact en financiële impact zijn de kostfuncties voor dit
onderzoek en bepalen mee de fitheid van een gebouwvariant. De energieprestatie wordt
berekend via een gebouwsimulatieprogramma (EPB of TRNSYS), de ecologische impact via
het
levensinventarismodel
en
de
financiële
impact
via
een
kosten-batenanalyse.
Randvoorwaarden voor binnenluchtkwaliteit, wintercomfort en daglichttoetreding zijn zo
geïntegreerd dat elke gebouwvariant er aan voldoet, onafhankelijk van de optimalisatie.
Randvoorwaarden die wel worden beoordeeld tijdens het optimalisatieproces zijn het
isolatiepeil (maximum K45) en het zomercomfort (maximum 130 GTO-uren volgens
ISSO/SBR (1994)). De fitheid van een variant in een populatie wordt berekend via een
fitheidsfunctie die de prestaties van de variant voor energie, emissies en kosten vertaalt naar
een Pareto-score. Deze score is gelijk aan het aantal varianten dat deze variant domineert.
Een variant is niet-gedomineerd als er geen andere oplossing is binnen de populatie die
gelijk of beter presteert in één objectief zonder slechter te presteren in het andere objectief.
Niet-gedomineerde varianten hebben score 0. De randvoorwaarden voor isolatiepeil en
zomercomfort worden geëvalueerd via een straffunctie. Als 1 of meerdere randvoorwaarden
niet voldaan zijn, wordt een strafwaarde toegevoegd aan de Pareto score, evenredig met de
overschrijding van de randvoorwaarden. Verschillende tests zijn uitgevoerd om de meest
geschikte straffuncties voor dit onderzoek af te leiden.
Levenscyclusinventaris
Doel en reikwijdte
Hoofddoel van de levenscyclusinventaris is het opstellen van een LCI databank voor
bouwmaterialen en –componenten en het ontwikkelen van een LCI-model voor gebouwen
als geheel dat kan geïntegreerd worden in het optimalisatieproces. Maar deze LCI geeft ook
de mogelijkheid om de energiebesparingen bij extreem lage energiewoningen te vergelijken
met de ingebouwde energie, nodig om deze concepten te creëren. Deze balans moet altijd
positief blijven over de levenscyclus van het gebouw. De LCA is beperkt tot een inventaris
van energiestromen en emissies. De enige impactindicator die wordt berekend, is de global
warming potential. Energiestromen en emissies van de verschillende fasen kunnen worden
opgeteld en vormen zo het energie- en pollutiecriterium voor de optimalisatie. Als functionele
eenheid worden in dit onderzoek gebouwen als geheel geanalyseerd zodat de globale
prestatie van de gebouwen kan beoordeeld worden en enkel gebouwen met eenzelfde
prestatieniveau vergeleken worden. Ook wordt niet de volledige levensduur van het gebouw
in rekening gebracht, maar enkel de impact van 1 generatie (30 – 40 jaar). Dit betekent dat
de fasen van extractie, productie, transport, gebruik en vervangingen worden ingerekend,
maar niet de eindfase van het gebouw.
216
SAMENVATTING
Input data en inventarismodellen
Alle LCI data komen uit de ecoinvent2000 databank (Frischknecht 2003). Hierbij zijn 47
datasets voor bouwgerelateerde processen uit de databank gebruikt om de LCI van 54
gebouwgerelateerde goederen te berekenen. De in- en outputstromen zijn beperkt tot
energie, afvalwarmte, CO2, NOx, SOx emissies, niet metaan VOC’s en partikels, de
impactindicatoren tot de global warming potential over een periode van 20, 100 en 500 jaar.
Voor elke fase zijn rekenmodellen ontwikkeld. Voor de exploitatie en productiefase voorziet
ecoinvent de meeste input. Sommige materialen (baksteen, rotswool,…) zijn als dusdanig
aanwezig in ecoinvent, voor andere producten (raamprofielen, installatiecomponenten,…)
moet een productmodel opgesteld worden. Voor de transportfase is onderscheid gemaakt
tussen transport van bouwmaterialen en installatiecomponenten. Voor bouwmaterialen wordt
onderscheid gemaakt tussen transport van de productieplaats naar een verdeelcentrum
(stap 1) en van daar naar de werf (stap 2). Voor installatiecomponenten houdt de 1e stap
altijd in dat samenstellende materialen worden getransporteerd naar de assemblageplaats,
terwijl in de 2e stap afgewerkte producten naar de werf worden vervoerd. Voor elke
transportstap zijn aannames gemaakt per materiaal of product over afstand, voertuig en
gewicht. Op basis van deze aannames is het energiegebruik en zijn de emissies als gevolg
van transport berekend.
De data uit deze twee fasen worden in een algoritme gebruikt om de LCI data voor een
gebouw als geheel te berekenen. Hierbij wordt per gebouwvariant voor elk gebruikt materiaal
of product het totale volume, oppervlakte of lengte berekend, inclusief de vervangingen die
gebeuren over de beschouwde periode. Op die manier wordt de totale niet hernieuwbare
energie, GWP100a, NOx, SOx, NMVOC en partikels < 2.5µm gekoppeld aan de productie- en
transportfase van het gebouw als geheel berekend. Voor de gebruiksfase wordt per
gebouwvariant het jaarlijks primair energieverbruik en de jaarlijkse GWP berekend met een
stationair of dynamisch gebouwsimulatieprogramma (EPB of TRNSYS) en dit jaarlijks
verbruik wordt omgerekend naar een verbruik en GWP over 30 of 40 jaar. In een laatste stap
wordt de niet hernieuwbare energie en GWP voor alle fasen opgeteld. Er worden geen
aannames gemaakt over de bestemming van het gebouw na deze eerste generatie.
Onzekerheids- en bijdrage-analyse
Ondanks de eenvoud van het LCI-model is de onzekerheid groot, omdat alle matrices meer
of minder onzekere data bevatten. Daarom is een onzekerheids- en gevoeligheidsanalyse
uitgevoerd op het LCI-model. Beide studies tonen aan dat de gevoeligheid voor en de
voortplanting van fouten beperkt is en dat de fouten op de verschillende inputdata elkaar
kunnen neutralizeren. Vergelijking van ingebouwde energie en energie tijdens de
gebruiksfase leidt tot de belangrijke conclusie dat de ingebouwde energie een relatief kleine
217
rol speelt in vergelijking met het energiegebruik tijdens de gebruiksfase, zelfs bij extreem
lage energiewoningen. De gebruiksfase moet daarom ook in de toekomst de absolute
voorrang krijgen.
Kosten-batenanalyse
Kostcriteria
Om het financiële aspect van de gebouwconcepten te evalueren vanuit het standpunt van de
bouwheer, is een kostenberekeningmodule geïntegreerd in het optimalisatiemodel. In deze
module wordt een groot aantal financiële criteria berekend en opgeslagen die kunnen
worden gebruikt voor een grondige analyse van de resultaten: initiële investeringskosten,
investeringskosten voor vervangingen, jaarlijkse onderhoudskosten, jaarlijkse energiekosten
voor fossiele brandstoffen en electriciteit (geen huishoudelijk verbruik) en TAK en NCW voor
verschillende aannames van gebruiksperiode, discontovoet en prijsevoluties. Voor de netto
contante waarde dient de niet geïsoleerde variant van de gebouwen als referentie.
Toch kan slechts 1 kostcriterium meegenomen worden in het optimalisatieproces. In de
eerste optimalisatiestap, waar enkel de gebouwschil wordt bekeken en de installatie nog niet
gekend
is,
wordt
de
investeringskost
als
kostcriterium
gekozen;
in
de
tweede
optimalisatiestap wordt de TAK als kostcriterium gekozen. De evaluatie van de resultaten
gebeurt aan de hand van de NCW, omdat enkel varianten met een positieve NCW
economisch rendabel zijn over de beschouwde periode.
Kostdatabank
De kostdatabank is samengesteld op basis van reële prijzen, inclusief werkuren en BTW.
Omdat de grootte van een installatie sterk afhankelijk is van de thermische kwaliteit van het
gebouw, zijn voor installatiecomponenten ook kostencurves opgesteld die de kost van een
verwarmings-component of –systeem uitdrukken als een functie van het isolatiepeil (Hens
2005).
De aannames voor energieprijzen zijn gebaseerd op consumentenprijzen. De prijzen voor
elektriciteit en gas komen van het Federaal Ministerie voor Economische Zaken en zijn
inclusief taksen voor transport en distributie, energietaksen en federale taksen. De prijzen
voor stookolie komen van Informazout en zijn geldig voor een aankoop van minstens 2000l.
Voor de evolutie van de energieprijzen, zijn drie scenario’s aangenomen, gebaseerd op de
EU POLES scenario’s voor 2000-2030 voor gas en stookolie (EU 2004): laag scenario
(+0%), midden scenario (+ 1.9% voor stookolie, +2.1% voor gas en elektriciteit), hoog
scenario (+3.2% voor stookolie, +4.3% voor gas en elektriciteit)
218
SAMENVATTING
Opbouw van het optimalisatieprogramma
Het programma voor de globale methodologie is geschreven in MATLAB, omdat dit een
makkelijke koppeling met andere programma’s toelaat. Het programma is gevalideerd aan
de hand van resultaten uit vroeger werk (Verbeeck and Hens 2002). Figuur S.1 geeft de
schematische opbouw van het programma.
Figuur S.1: Schematische opbouw van de globale methodologie
219
Resultaten
De ontwikkelde methodologie is toegepast in twee projecten: het BIM-project, een technischeconomische studie van de rendabiliteit van energiebesparende investeringen voor de
context van het Brussels Hoofdstedelijk Gewest (De Coninck en Verbeeck 2005) en het
EL²EP-project, een studie gefinancierd door het IWT, waarin via levenscyclusoptimalisatie
van energie, emissies en kosten concepten voor extreem lage energie- en pollutiewoningen
zijn ontwikkeld (Verbeeck et al. 2007). De resultaten in deze samenvatting zijn grotendeels
beperkt tot het EL²EP-project.
Globaal geoptimaliseerde concepten
Figuur S.2 geeft de optimale oplossingen die volgen uit de tweede optimalisatiestap, voor de
rijwoning van het EL²EP-project. Natuurlijk zijn vooral oplossingen met positieve NCW van
belang, omdat zij economisch rendabel zijn, zelfs in het onwaarschijnlijke geval dat de
energieprijzen constant blijven. Tabel S.1 geeft een overzicht van de resultaten voor alle vijf
woningen die in het EL²EP-project geoptimaliseerd zijn. Voor elk aspect wordt de range
gegeven waarbinnen de geoptimaliseerde concepten vallen. Energie-gerelateerde waarden
zijn meestal gegeven per m³ verwarmd volume, kost-gerelateerde waarden per m² verwarmd
vloeroppervlakte.
Netto contante waarde vs. totaal primair energiegebruik
Rijwoning, laag energieprijsscenario
netto contante waarde [€]
20000
0
-20000
-40000
-60000
alle resultaten
optima
-80000
optima NCW30>0
-100000
800
1300
1800
2300
2800
totaal primair energiegebruik [GJ]
Figure S.2: Rijwoning, gebruiksperiode van 30 jaar, laag energieprijsscenario,discontovoet =
4%: de NCW over 30 jaar (NCW30) is gegeven in functie van het totaal primair
energiegebruik over 30 jaar. Alle resultaten worden in het grijs weergegeven en de trade-off
curve in het zwart. De volle zwarte punten hebben een NCW30 >0.
220
SAMENVATTING
Rijwoning
Hoekwoning
Villa
Bungalow
Flat,
midden
Aantal optima
79
138
125
36
35
16
48
62
15
17
Isolatiepeil
K15 → K27
K13 → K34
K13 → K40
K13 → K20
K9 → K38
Ugem [W/m²K]
0.2 → 0.37
0.15 → 0.4
0.14 → 0.43
0.13 → 0.20
0.15 → 0.69
15 - 90
15 - 90
25 - 100
30 - 55
< 40
2000 - 4500
2240 - 4650
2230 - 5330
2500 - 3700
1300 - 2500
90 - 330
100 - 290
95 - 330
120 - 230
100 - 190
40 - 500
80 - 800
3 - 840
100 - 720
50 - 340
300 - 880
300 - 900
300 - 1000
300 - 650
130 - 510
1100 - 1300
1000 -1600
1000 - 1700
1000 - 1500
800 - 1050
-260 → 60
-360 → 300
-500 → 200
-200 → 400
-160 → 70
9 → 50
3 → 57
-4 → 62
2 → 23
-3 → 35
Econ. rendabel
Netto energiebehoeften [MJ/m³a]
Totaal primair
energiegebruik over
30 jaar [MJ/m³]
Totale GWP over
30 jaar [kg/m³]
Extra investeringskost [€/m²]
Jaarlijkse
energiekost [€]
TAK over 30 jaar,
laag energieprijsscenario [€/m²]
NCW over 30 jaar,
laag energieprijsscenario [€/m²]
Statische terugverdientijd [jaren]
Tabel S.1: Resultaten voor de globale optimalisatie van de vijf woningen uit het EL²EPproject: enkel resultaten van de trade-off voor TAK en totaal primair energiegebruik over 30
jaar, geldig voor het lage energiescenario en een discontovoet van 4% zijn weergegeven
221
Discussie
Economisch optimum en hiërarchie van energiebesparende maatregelen
Uit de resultaten kan een economisch optimale combinatie van energiebesparende
maatregelen worden afgeleid. Dit economische optimum is gelijk voor alle woningen en
bestaat uit:
-
Een economisch optimaal isolatiepeil K25-27 voor de meeste woningen (Ugem ≈ 0.30.4W/m²K voor een compactheid van 1.5-2). Enkel voor zeer weinig compacte
woningen (compactheid < 1), ligt het economisch optimaal isolatiepeil lager, bij K20
(Ugem ≈ 0.2W/m²K),
-
+ een hoogrendements- of condensatieketel op gas of stookolie, radiatoren
ontworpen op hoge temperatuur, maar werkend op een variabele watertemperatuur,
en een goed ontworpen natuurlijk ventilatiesysteem met aandacht voor luchtdichtheid
-
Dit komt overeen met een energieprestatieniveau E57 ± 6 voor de meeste huizen,
berekend volgens de Vlaamse EPB.
-
In vergelijking met de niet geïsoleerde referentie betekent dit een afname in jaarlijks
primair energiegebruik met 66% ± 6%; in vergelijking met de E100 eis, een afname
met 43% ± 6%.
Verder kan uit de resultaten een logische hiërarchie van energiebesparende investeringen
worden afgeleid: Vertrekkend van het economische optimum kan de energieprestatie verder
verbeterd
worden
door
toepassing
van
een
mechanische
balansventilatie
met
warmteterugwinning of een lucht/water warmtepomp. Deze warmtepomp heeft een iets
lagere SPF dan een grond/water warmtepomp, maar is ook veel goedkoper. Voor de meeste
woningen zitten deze maatregelen nog net op de grens van wat economisch rendabel is. Het
energiegebruik kan nog verder gereduceerd worden door een nog beter isolatiepeil (K15-20),
gecombineerd met een mechanische balansventilatie met warmteterugwinning of zelfs met
een micro-WKK of systemen op zonenergie, zoals zonnecollectoren of PV-systemen. Dit
leidt tot zeer lage energiekosten, maar deze oplossingen liggen ver onder wat economisch
rendabel is.
Belangrijk is dat zowel het economische optimum als de hiërarchie van energiebesparende
maatregelen onafhankelijk is van de geanalyseerde kostprijsscenario’s voor energie,
materiaalkost en discontovoet (2%, 4%, 8%).
222
SAMENVATTING
Concepten voor extreem lage energiewoningen
Op basis van de hiërarchie van energiebesparende maatregelen kunnen woningen met
extreem laag energiegebruik worden gerealiseerd. Gelijkaardige verbruiken worden ook in
de praktijk al gerealiseerd met concepten zoals passiefhuizen en nulenergiewoningen.
Echter, analyse van deze woningen toont aan dat quasi geen enkele variant economisch
rendabel is voor de aangenomen energieprijsscenario’s. Enkel bij veel hogere energieprijzen
en/of lagere discontovoeten, zullen extreem lage energiewoningen rendabel worden. Echter,
de grootste hindernis voor dit type woningen is de zeer hoge investeringskost. Passiefhuizen
vragen een extra investering van 30.000 – 35.000€ t.o.v. niet geïsoleerde woningen en ca.
16.000-18.000€ t.o.v. het economisch optimum. Dit resulteert wel in een veel lager verbruik:
ca. 900GJ over 30 jaar t.o.v. ca. 1500GJ voor het economische optimum. Uit in situ metingen
in passiefhuizen blijkt echter dat in praktijk het thermische comfort en de binnenluchtkwaliteit
in deze woningen nog niet optimaal is.
Besluiten
De ontwikkelde optimalisatiemethodologie bleek een interessant middel om de economisch
meest rendabele combinatie van energiebesparende maatregelen af te leiden, evenals een
hiërarchie van energiebesparende investeringen. Analyse van de kostprijsscenario’s toonde
bovendien aan dat zowel het economische optimum als de hiërarchie van maatregelen
onafhankelijk is van deze scenario’s. Ook zijn concepten voor extreem lage energiewoningen
afgeleid, maar geen van hen bleek economisch rendabel voor de geanalyseerde scenario’s.
Bovendien is de zeer hoge investeringskost een groot nadeel van deze woningen. Zonder
financiële steun zullen deze concepten beperkt blijven tot een kleine minderheid van zeer
milieubewuste bouwheren die bereid zijn deze investering te doen.
Als laatste kan besloten worden dat de ingebouwde energie van energiebesparende
maatregelen geen reden tot bekommernis is. De ingebouwde energie neemt sterk toe met
het isolatieniveau, maar tegelijk leidt het tot zeer hoge energiebesparingen tijdens de
gebruiksfase. Ongeacht de materialen en systemen die worden toegepast, blijkt de
ingebouwde energie te worden terugverdiend in minder dan 2 jaar.
223
ANNEX A
ANNEX A: Relation between research database and
ecoinvent database
Table A.1 presents the constructional materials and products of the research database and
the related data and process sets from the ecoinvent database. Table A.2 contains the same
information for the installation components.
Material
in
PhD
Datasets from ecoinvent2000
database
valid in
zone
Reinforcing steel
Reinforcing steel, at plant
RER
Roof tiles
Roof tile, at plant
RER
Bricks
Brick, at plant
RER
Ceramic tiles
Ceramic tiles, at regional storage
CH
Mortar bed
Cement mortar
CH
Plaster mixing
CH
Concrete blocks
Concrete block, at plant
DE
Floor slab
Concrete normal, at plant
CH
Inner plasterwork
Stucco
CH
Plaster mixing
CH
Stucco
CH
Plaster mixing
CH
Gypsum board
Gypsum plaster board, at plant
CH
Bitumen
Bitumen sealing, at plant
RER
Air barrier (PE-foil)
Fleece, polyethylene, at plant
RER
Underlay
Fleece, polyethylene, at plant
RER
Hard timberwood
Sawn timber, hardwood, planed, kiln dried, at plant
RER
Preservative treatment, sawn timber, pressure vessel
RER
Wood preservative, organic salt, Cr-free, at plant
RER
Sawn timber, softwood, planed, kiln dried, at plant
RER
RER
RER
Plywood
Plywood, outdoor use, at plant
RER
OSB
Oriented strand board, at plant
RER
Mineral wool
Rock wool, packed, at plant
CH
Polyurethane
RER
Expanded polystyrene
Polystyrene foam slab, at plant
RER
Extruded polystyrene
Polystyrene, general purpose, GPPS, at plant
RER
Extrusion, film
RER
Cellulose fibre
Cellulose fibre, inclusive blowing in, at plant
CH
Foam glass
Foam glass, at plant
CH
Outer plaster
Soft timberwood
225
Uncoated glass
Flat glass, uncoated, at plant
RER
Coated glass
Flat glass, coated, at plant
RER
Aluminium
Aluminium, production mix, cast alloy at plant
RER
Sheet rolling aluminium
RER
Chromium steel 18/18, at plant
RER
Sheet rolling chromium steel
RER
Synthetic rubber, at plant, including extrusion profiles
RER
Stainless steel
Butyl (EPDM)
for window sealings
RER
Section bar, extrusion, aluminium
RER
Electricity, medium voltage, at grid
BE
Heat, heavy fuel oil at industrial furnace, 1MW
RER
PVC, at regional storage
RER
Injection moulding
RER
Steel, low alloyed, at plant
RER
Section bar, rolling, steel
RER
BE
RER
RER
RER
BE
RER
Injection moulding
RER
RER
RER
BE
Alu-PUR-wood window
RER
frame
RER
RER
Injection moulding
RER
RER
RER
RER
BE
RER
Wood-PUR-wood
RER
window frame
RER
RER
RER
Injection moulding
RER
Aluminium
window
frame
PVC window frame
Wooden window frame
PUR window frame
226
ANNEX A
BE
Inner sunshading
Glass fibre, at plant
RER
(textile + fixation)
RER
RER
RER
Outer sunshading
RER
RER
RER
RER
Table A.1
Installation component
Datasets from ecoinvent2000
in PhD database
valid in
zone
RER
RER
Steel, low-alloyed, at plant
RER
Sheet rolling steel
RER
RER
RER
Copper, at regional storage
RER
RER
RER
RER
Sheet rolling steel
RER
RER
RER
RER
Electrical
RER
accumulation heating
RER
RER
Sheet rolling steel
RER
RER
RER
RER
High efficiency boiler
RER
fuel
Sheet rolling steel
RER
High efficiency boiler
RER
natural gas
Sheet rolling steel
RER
Condensing boiler
RER
Sheet rolling steel
RER
Gas stove
Direct electrical heater
227
Heat pump
Direct water heater
Storage tank
Water pipes
Radiator
Floor heating
Solar
collector,
storage tank
Ventilation pipes
Ventilator
Heat recovery unit
Table A.2
228
incl.
RER
RER
RER
Sheet rolling steel
RER
RER
Polypropylene, granulate, at plant
RER
Extrusion, plastic pipes
RER
RER
RER
RER
RER
RER
RER
RER
RER
RER
Polyethylene, HDPE, granulate, at plant
RER
RER
Sheet rolling steel
RER
RER
RER
RER
Polyethylene, HDPE, granulate, at plant
RER
RER
RER
Section bar, rolling steel
RER
RER
Rock wool, at plant
RER
RER
RER
RER
RER
RER
RER
Sheet rolling steel
RER
RER
RER
RER
Polystyrene, high impact, HIPS, at plant
RER
ANNEX A
For combined heat and power generation and photovoltaic modules, all LCI data are
provided by the co-research groups of the GBOU-EL²EP-project (CHP by KUL-TME, PV by
KUL-ELECTA). References for these data can be found in Verbeeck et al. (2007).
229
ANNEX B: Extracted flows from the ecoinvent database
Table B.1 shows the elementary flows that are extracted for each dataset from ecoinvent into
the research database. The bold figures in italic are calculated within the research database
and exported into the optimisation algorithm, not to interfere in the optimisation, but to serve
as extra information input.
Extracted flows from ecoinvent database
Elementary flow or impact category
Subcategory
units
Cumulative energy demand
non-renewable total
MJ/kg
fossil
nuclear
MJ/kg
renewable total
biomass
wind, solar, geothermal
water
MJ/kg
total
Heat, waste balance
total
Heat, waste
total
In air
In soil
In water
Energy resource
high population density
MJ/kg
low population density
MJ/kg
low population density, long term
MJ/kg
lower stratosphere + upper troposhere
MJ/kg
unspecified
MJ/kg
industrial
MJ/kg
unspecified
MJ/kg
ground-, long-term
MJ/kg
river
MJ/kg
unspecified
MJ/kg
gross caloric value,in biomass
MJ/kg
kinetic, flow, in wind
MJ/kg
potential, stock, in barrage water
MJ/kg
solar
MJ/kg
total
GWP 20a
kg/kg
GWP 100a
kg/kg
GWP 500a
kg/kg
CO2
total
kg/kg
NOX
total
kg/kg
SOX
total
kg/kg
230
ANNEX B
NMVOC
total
kg/kg
Particulates, < 2.5 µm
total
kg/kg
kg/kg
CO2, in air, resource
CO2, biogenic
CO2, fossil
NOX
SOX
NMVOC, unspecified origin
Particulates, > 2.5 µm, and < 10µm
Particulates, > 10 µm
Particulates, < 2.5 µm
kg/kg
kg/kg
lower stratosphere + upper troposphere
kg/kg
unspecified
kg/kg
kg/kg
kg/kg
kg/kg
unspecified
kg/kg
kg/kg
kg/kg
kg/kg
unspecified
kg/kg
kg/kg
kg/kg
kg/kg
unspecified
kg/kg
kg/kg
kg/kg
kg/kg
unspecified
kg/kg
kg/kg
kg/kg
kg/kg
unspecified
kg/kg
kg/kg
kg/kg
kg/kg
unspecified
kg/kg
kg/kg
kg/kg
kg/kg
unspecified
kg/kg
PROCESSES INCLUDED
Table B.1
231
ANNEX C: Models for life cycle inventory of building
products
Unit
Composing materials
model
Material
Mortar bed
1 kg
Inner plasterwork
1 kg
Outer plaster
Hard timberwood
Soft timberwood
Extruded
1 kg
1 m³
1 m³
1 kg
polystyrene
Aluminium
1 kg
Stainless steel
Aluminium window
1 kg
1m
frame
PVC window frame
Wooden
window
1m
1m
frame
PUR window frame
Alu-PUR-wood
window frame
232
Product
1m
1m
Cement mortar
0.67 kg
Plaster mixing
1 unit
Stucco
0.67 kg
Plaster mixing
1 unit
Stucco
0.67 kg
Plaster mixing
1 unit
1 m³
1 unit
4 kg
Sawn timber, softwood, planed, kiln dried, at plant
1 m³
1 unit
4 kg
Polystyrene, general purpose, GPPS, at plant
1.02 kg
Extrusion, film
1.02 units
1 kg
1 unit
1 kg
1 unit
3.1 kg
3.1 units
2.53 kWh
13.46 MJ
4.18 kg
Injection moulding
4.18 units
Steel, low alloyed, at plant
1.79 kg
Section bar, rolling, steel
1.79 units
0.48 kWh
-3
7.1 10 m³
7.1 10-3 unit
-3
28.4 10 kg
2.4 kWh
1.48 kg
Injection moulding
1.48 units
1.03 kg
1.03 units
0.48 kWh
1.55 kg
1.55 units
ANNEX C
Wood-PUR-wood
1m
window frame
Inner sunshading
1 m²
Outer sunshading
1 m²
0.74 kg
Injection moulding
0.74 units
-3
3.55 10 m³
-3
3.55 10 unit
-3
14.2 10 kg
2.7 kWh
6.73 MJ
-3
7.1 10 m³
-3
7.1 10 unit
-3
28.4 10 kg
0.74 kg
Injection moulding
0.74 units
2.4 kWh
0.22 kg
0.30 kg
0.3 kg
0.3 units
0.26 kg
0.36 kg
0.5 kg
0.5 units
Table C.1
unit
Product model
Composing materials
Installation
Per kW
component
Gas stove
Per
kW
1 unit
Aluminium, production mix, cast
of x kW
alloy at plant
Per unit
0.666
0.11kg
0.05 kg
0.11units
0.05units
4.97 kg
2.48 kg
Sheet rolling steel
4.97units
2.48units
0.25 kg
0.13 kg
0.25units
0.13units
0.022kg
0.01 kg
0.11kg
0.05 kg
0.11units
0.05units
4.97 kg
2.48 kg
Sheet rolling steel
4.97units
2.48units
0.25 kg
0.13 kg
0.25units
0.13units
0.022kg
0.01 kg
0.11kg
0.05 kg
Direct electrical
1 unit
heater
of x kW
alloy at plant
Electrical
1 unit
accumulation
of x kW
alloy at plant
233
heating
0.11units
0.05units
4.97 kg
2.48 kg
Sheet rolling steel
4.97units
2.48units
0.25 kg
0.13 kg
0.25units
0.13units
0.022kg
0.01 kg
High efficiency
1 unit
4.95 kg
67.27 kg
boiler fuel, incl
of x kW
Sheet rolling steel
4.95units
67.27units
High efficiency
1 unit
1.05 kg
28.5 kg
boiler natural
of x kW
Sheet rolling steel
1.05units
28.5units
Condensing
1 unit
1.05 kg
30 kg
boiler
of x kW
Sheet rolling steel
1.05units
30 units
Heat pump, incl
1 unit
pipes
of x kW
alloy at plant
tank
gas
Direct water
1 unit
heater
Storage tank
Water pipes
1 unit
1m
6.5 kg
6.5units
52 kg
Sheet rolling steel
52units
6.5 kg
24 kg
24units
23.3 kg
23.3units
35 kg
35units
0.68units
0.68 kg
0.37 kg
0.02 kg
0.04units
Polyethylene, HDPE, granulate, at
0.02 kg
plant
Radiator HT
Radiator LT
Floor heating
1 unit
18.1 kg
-2.6 kg
of x kW
Sheet rolling steel
18.1units
-2.6units
1 unit
36.3 kg
-5.13 kg
of x kW
Sheet rolling steel
36.3units
-5.13units
1 unit
1600 kg
of x kW
5.5 kg
11units
Polyethylene, HDPE, granulate, at
5.5 kg
plant
Solar collector,
incl. storage
234
1 m²
alloy at plant
3.43 kg
ANNEX C
tank
Ventilation
1m
pipes
Ventilation fan
1 unit
for dwelling
Heat recovery
unit
3.43units
Section bar, extrusion aluminium
3.43units
6.85 kg
13.99units
Section bar, rolling steel
0.44units
7.59 kg
Rock wool, at plant
4.03 kg
20.16 kg
3.67 kg
3.67units
1 kg
alloy at plant
1 unit
1unit
15 kg
Sheet rolling steel
15units
2 kg
1.5 kg
alloy at plant
1.5units
16.75 kg
Polystyrene, high impact, HIPS, at
16.75 kg
plant
Table C.2
235
ANNEX D: Assumptions for the transport model
Table D.1 presents per commodity the assumed transport distances and transport vehicles
from production site to distribution site. Only one-way distances are considered, as the
transport vehicle is assumed to be used for non-related transport on the way back. Table D.2
presents the distances and vehicles for the transport from distribution site to construction
site. Table D.3 and D.4 provides the same data for the installation components. However,
step 1 (table D.3) concerns transport of the composing materials from their production site to
the assemblage site, whereas step 2 concerns transport of the installation components from
the assemblage site to the construction site (table D.4).
Constructional building products
Step 1: from production site to distribution site
Location
production
Transport vehicle
distribution
Commodity
ship
train
distance
distance
32t distance
[km]
[km]
[km]
200
100
EU
B
Roof tiles
B
B
100
Brick
B
B
100
Ceramic tiles
I
B
700
100
EU
B
100
100
Concrete blocks
B
B
200
Concrete
B
B
100
EU
B
B
B
Bitumen
EU
B
600
100
Air barrier
EU
B
200
100
Underlay
EU
B
200
100
Hard timberwood
GLO
B
8000
100
Soft timberwood
Scandin
B
2000
100
Plywood
EU
B
200
50
OSB
EU
B
200
50
Mineral wool
NL
B
200
Polyurethane
B
B
100
Expanded PS
B
B
100
Extruded PS
D
B
400
Cellulose fibre
D
B
400
Reinforcing steel
Mortar
Stucco
Gypsum board
236
100
100
100
ANNEX D
Foam glass
B
B
100
Uncoated glass
B
B
100
100
Coated glass
B
B
100
100
Aluminium
EU
B
200
100
Stainless steel
EU
B
200
100
Butyl (EPDM)
EU
B
200
100
Alu window frame
B
B
200
PVC window frame
B
B
200
Wooden window frame
B
B
200
PUR window frame
B
B
375
D
D
320
window frame
D
D
250
Inner sunshading
B
B
100
Outer sunshading
B
B
100
Alu-PUR-wood window
frame
Wood-PUR-wood
Table D.1
Constructional building products
Step 2: from distribution site to construction site
Location
distribution
Transport vehicle
construction
Commodity
32t
16t
distance
distance
van distance
[km]
[km]
[km]
Reinforcing steel
B
B
30
Roof tiles
B
B
30
Brick
B
B
30
Ceramic tiles
B
B
30
Mortar
B
B
30
Concrete blocks
B
B
30
Concrete
B
B
30
Stucco
B
B
30
Gypsum board
B
B
30
Bitumen
B
B
30
Air barrier
B
B
30
Underlay
B
B
30
Hard timberwood
B
B
30
Soft timberwood
B
B
30
Plywood
B
B
30
237
OSB
B
B
30
Mineral wool
B
B
30
Polyurethane
B
B
30
Expanded PS
B
B
30
Extruded PS
B
B
30
Cellulose fibre
D
B
30
Foam glass
B
B
30
Uncoated glass
B
B
30
Coated glass
B
B
30
Aluminium
B
B
30
Stainless steel
B
B
30
Butyl (EPDM)
B
B
30
Alu window frame
B
B
30
PVC window frame
B
B
30
Wooden window frame
B
B
30
PUR window frame
B
B
30
D
B
30
window frame
D
B
30
Inner sunshading
B
B
30
Outer sunshading
B
B
30
Alu-PUR-wood window
frame
Wood-PUR-wood
Table D.2
Installation components
Step 1: from production site to assemblage site
Transport vehicle
ship distance
train distance
32t distance
Basic material
[km]
[km]
[km]
Aluminium, cast alloy
150
150
100
Steel, low-alloyed
150
150
100
Chromium steel 18/18
150
150
100
Copper
150
150
100
Polypropylene, granulate
100
100
100
Polyethylene, HDPE, granulate
100
100
100
Polystyrene, high impact, HIPS
100
100
100
200
Rock wool
Coated glass
Concrete, normal
Table D.3
238
100
100
100
ANNEX D
Installation components
Step 2: from assemblage site to construction site
Transport vehicle
32t distance
16t distance
van distance
[km]
[km]
[km]
Gas stove
0
30
15
Direct electrical heater
0
30
15
Electrical accumulation heating
0
30
15
High efficiency boiler (fuel)
0
30
15
High efficiency boiler (natural gas)
0
30
15
Condensing boiler
0
30
15
Heat pump
0
30
15
Combined heat and power
0
50
Direct water heater
0
30
15
Storage tank for domestic hot water
0
30
15
Water pipes
0
30
15
HT radiator
0
30
15
LT radiator
0
30
15
Floor heating
0
30
15
Solar collector
0
50
PV-cells
0
50
Ventilation grids
0
30
15
Ventilation pipes
0
30
15
Ventilation fan
0
30
15
Heat recovery unit
0
30
15
Commodity
Table D.4
239
ANNEX E: Structure of the cost database
Table F.1 presents part of the cost database. Prices are given per constructional variant
(cavity wall, massive wall or wood frame) and per unit (m³, m², m or part).
Opdrachtgever :
Onderwerp:
Laboratorium Bouwfysica
Faculteit Toegepaste Wetenschappen
Kasteelpark 51, 3001 Heverlee (Leuven)
tel 016/321347 f
Prijsofferte:
Marktonderzoek van
volgens Samenvattende opmeting
kostprijzen ahv meetstaat van SO 97 (a) van de Vlaamse
referentiewoningen
Huisvestingsmaatschappij (VHM)
RIJWONING
KLEURCODES
Variant 1: Spouwmuur
Artikel
1
Beknopte aanduiding van de werken en prestaties
Isolatiepakket (verschillende materialen, verschillende
Variant 1 en 2 : Zware binnenconstructie
Hoeveelheid berekend EENH. Eenheidsprijs
door de
in cijfers
ontwerper inschrijver
Gedeeltelijke som
(afgerond op de
cent)
ONDERBOUW
10
GRONDWERKEN ONDERBOUW
10.12
10.21
10.31
12
voorafgaande afgraving terrein - verwijderen teelaarde
bouwputten - gewone bouwputten
sleuven - funderingszolen
FUNDERINGEN OP STAAL
12.11
14
funderingszolen - stortklaar beton / ongewapend
METSELWERK ONDERBOUW
14.13
14.23
38.50
98.33
17.06
m³
m³
m³
17.93
11.95
17.93
690.31 €
1,175.08 €
305.96 €
5.69
m³
119.50
679.72 €
5.50
7.99
3.53
53.06
1.00
m³
m³
14.44
14.63
15
funderingsmuren - betonblokken / hol
dragende keldermuren - betonblokken / hol 29 cm
dragende keldermuren - betonblokken / hol 19 cm
waterdichting - noppenbanen
verluchtingselementen - keldergatroosters
VLOERLAGEN ONDERBOUW
m²
st
250.95
427.87
427.87
19.42
89.63
1,379.82
3,417.72
1,509.49
1,030.50
89.63
15.11
15.21
15.41
16
zuiverheidslagen - stortklaar beton / ongewapend
draagvloeren op volle grond - stortklaar beton / gewapend
vochtwerende lagen - folies / PE
THERMISCHE ISOLATIE ONDERBOUW
3.07
11.26
61.41
m³
m³
m²
113.53
157.74
1.79
348.58 €
1,775.70 €
109.92 €
16.21
16.22
isolatieplaten ondergrondse wanden - polyurethaan (PUR)
isolatieplaten ondergrondse wanden - polystyreen / geëxtru
11.06
0.00
m²
m²
17.69
14.94
195.56 €
0.00 €
30.92
7.02
278.28
m³
m³
m²
311.30
311.30
4.75
9,625.22 €
2,184.87 €
1,321.81 €
3.22
35.77
m³
m²
370.45
4.75
1,192.67 €
169.92 €
2
BOVENBOUW
20
OPGAAND METSELWERK
20.12
20.22
20.34
21
binnenspouwblad - baksteen / geperforeerd
dragende binnenmuren - baksteen / geperforeerd
supplementen - voegen / zichtbaar metselwerk
NIET DRAGEND METSELWERK
21.12
21.33
22
scheidingswanden - baksteen / geperforeerd
supplementen - voegen zichtbaar metselwerk
MUURISOLATIE BOVENBOUW
22.11
spouwisolatie / thermisch - minerale wol (MW)
minerale wol (MW) dikte XX cm
0.00
m²
11.95
0.00
m³
0.00 €
0.00
m²
239.00
12.38
0.00
m³
m²
247.60
15.54
0.00 €
0.00
310.80
15.56
0.00 €
311.20
39.32
31.31
7,179.38 €
0.00 €
1,093.35 €
22.12
spouwisolatie / thermisch - polystyreen / geëxtrudeerd (XP
polystyreen / geëxtrudeerd (XPS) dikte XX cm
22.13
spouwisolatie / thermisch - polyurethaan (PUR)
polyurethaan (PUR) dikte XX cm
0.00
m³
22.13
spouwisolatie / thermisch - geëxpandeerd polystyreen (EPS
geëxpandeerd polystyreen (EPS) dikte XX cm
76.90
m²
23.07
m³
22.31
22.32
drukvaste isolatie - cellulair glas (CG)
drukvaste isolatie - cellenbetonblokken
0.00
34.92
m
m
240
€
€
€
€
€
ANNEX F
ANNEX F: Reference buildings of the BIM- project
Terraced house, new construction
3.10m
4.15m
Keuken
Badk
2.70m
Slaapkamer 1
Eetkamer
2.40m
Berging
9.00m
1.50m
9.00m
Stookplaats
4.10m
1.50m
Wc
3.80m
2.40m
4.10m
4.15m
Slaapkamer 2
Living
7.00m
7.00m
Gelijkvloers
1e verdiep
Berging
2.10m
Slaapkamer 3
3.30m
9.00m
2.40m
4.10m
3.30m
Bureau
7.00m
2e verdiep
Dwarsdoorsnede
241
Achtergevel
Voorgevel
Mansion, retrofit
3.40m
Keuken
4.70m
Badk
1.80m
Stookplaats
Eetkamer
4.40m
13.00m
4.10m
Living
Kelderverdiep
7.00m
242
Gelijkvloers
ANNEX F
4.60m
1.80m
4.30m
Slaapkamer 3
Slaapkamer 1
9.20m
4.00m
Slaapkamer 2
Bureau
Berging
Berging
1e verdiep
2e verdiep
Dwarsdoorsnede
243
Voorgevel
Achtergevel
Small apartment building, new construction and retrofit
3.30m
2.00m
4.00m
Slaapkamer 1
Badk
+
wc
Slaapkamer 2
3.40m
4.50m
12.00m
Berging/
stookpl
Hall
1.50m
4.60m
4.30m
Living
Eetkamer
Keuken
6.40m
3.00m
10.00m
244
ANNEX F
Voorgevel
Achtergevel
245
Large apartment building, new construction and retrofit
4.90m
1.80m
Slaapkamer 1
Slaapkamer 1
Badk
Badk
2.90m
Badk
2.90m
3.30m
Hall
Slaapkamer 1
Slaapkamer 1
4.00m
Badk
0.50m
Hall
Hall
Hall
2.00m
Wc
Wc
Wc
Wc
3.55m
14.00m
1.80m
Eetkamer
Eetkamer
Eetkamer
Eetkamer
4.80m
6.80m
1.90m
4.70m
Living
Keuken
Keuken
Living
Keuken
Living
7.00m
Voorgevel
246
Keuken
Living
ANNEX F
Achtergevel
Zijgevel
247
ANNEX G: Reference buildings of the EL²EP- project
‘Architectural’ dwelling (detached dwelling with fragmented plan)
A
gelijkvloers
N
1
2
10
4
9
5
3
A'
gevelsteen (9cm)
luchtspouw (3 cm)
isolatie (6 cm)
metselwerk (14 cm)
pleister (1 cm)
roofing (0.5 cm)
drukvaste isolatie (6cm)
membraam (-)
hellingsbeton (6cm)
draagvloer (15 cm)
pleister (1cm)
kiezels (3 cm)
roofing (0.5 cm)
membraam (-)
hellingsbeton (6cm)
draagvloer (15 cm)
lattenwerk (2cm)
houtafwerking (2cm)
vloerbedekking (1cm)
plaatsingsproduct (1 cm)
dekvloer op basis van cement
(7 cm)
membraam (-)
drukvaste isolatie (4 cm)
membraam (-)
draagvloer (15 cm)
tegels (1 cm)
plaastinfsproduct (1 cm)
hellingsbeton (6cm)
membraam (-)
draagvloer (15 cm)
Snede AA'
248
6
7
ANNEX G
Aanzicht Noorden
Aanzicht Zuiden
Aanzicht Westen
Aanzicht Oosten
249
Detached dwelling with a simple square plan
B
A
1
2
8
10
3
9
B'
A'
gelijkvloers
B
A
N
6
5
7
4
10
B'
eerste verdieping
250
ANNEX G
B
A
13
B'
A'
kelder
11
12
dekvloer op basis van cement (7 cm)
membraam (-)
draagvloer (15 cm)
pleister (1 cm)
(7 cm)
membraam (-)
membraam (-)
draagvloer (15 cm)
gevelsteen (9cm)
luchtspouw (3 cm)
isolatie (6 cm)
metselwerk (14 cm)
pleister (1 cm)
Cellenglas
bitumen (-)
betonblokken (29 cm)
betonnen draagvloer (20cm)
(=AFP)
Doorsnede BB'
Doorsnede AA'
251
Noordzijde
Zuidzijde
252
Westzijde
Oostzijde
ANNEX G
Semi-detached house
A
A'
13
kelder
2
1
10
A
3
A'
9
gelijkvloers
N
6
A
4
7
10
A'
5
eerste verdieping
253
en
nn
pa
en )
n
tte (1cm
nla
ne erk (-) )
n
w
pa ten ak cm
lat derd (23
on lken
ba
)
cm
(5
11
gevelsteen (9cm)
luchtspouw (3 cm)
isolatie (6 cm)
metselwerk (14 cm)
pleister (1 cm)
dekvloer op basis van cement (7
cm)
membraam (-)
draagvloer (15 cm)
pleister (1 cm)
(7 cm)
membraam (-)
membraam (-)
draagvloer (15 cm)
Snede AA'
Zuidzijde
254
Cellenglas
ANNEX G
Oostzijde
Westzijde
255
Terraced house
Noordzijde
Zuidzijde
B
A
B
A
2
1
3
13
B'
kelder
256
10
A'
B'
gelijkvloers
A'
ANNEX G
B
A
B
9
6
4
A
8
10
10
5
B'
7
A'
B'
eerste verdieping
A'
tweede verdieping
)
cm
(4 )
en cm
n n (1
pa rk
en we
en tten
latt la
en k + cm) )
n
a
n
pa derd (2 3 (1cm
on lken on
rt
ba ska
gip
11
11
(6 cm)
membraam (-)
membraam (-)
draagvloer (15 cm)
(7 cm)
membraam (-)
draagvloer (15 cm)
pleister (1 cm)
gevelsteen (9cm)
luchtspouw (3 cm)
isolatie (6 cm)
metselwerk (14 cm)
pleister (1 cm)
roofing (0.5 cm)
membraam (-)
hellingsbeton (6cm)
draagvloer (15 cm)
pleister (1cm)
Cellenglas
Doorsnede AA'
(7 cm)
membraam (-)
membraam (-)
draagvloer (15 cm)
Cellenglas
Doorsnede BB'
257
Apartment flat
5
7
6
10
4
3
9
2
258
1
WFC
WFC
WFC
WFC
CAV
OUT
CAV
OUT
CAV
OUT
CAV
OUT
2
3
4
5
6
7
8
9
10
11
12
13
0.19
0.19
0.21
0.21
0.20
0.20
0.17
0.17
0.22
0.21
0.20
0.19
0.19
[W/m²K]
Umean
14
14
16
16
15
15
13
13
17
16
15
14
14
level K
Insulation
26
26
47
47
40
40
26
26
47
40
26
26
26
[m²]
Aglass
1551
1548
1529
1528
1527
1525
1480
1478
1467
1463
1449
1427
1427
Heating
620
620
620
620
620
620
620
620
620
620
620
494
984
2876
2876
2876
2876
2876
2876
2876
2876
2876
2876
2876
2876
1943
Household
5047
5044
5024
5023
5022
5021
4975
4974
4963
4958
4944
4797
4354
consumption
Total
160,220
164,330
179,810
182,280
172,620
176,050
159,850
163,960
185,060
177,410
160,960
162,500
162,500
Building envelope
12,000
12,000
12,000
12,000
12,000
12,000
12,000
12,000
12,000
12,000
12,000
12,000
12,000
Installations
[€]
[kWh/m²a]
Hot water
Investment cost
Electricity consumption
+
-5,672.42
-11,759.32
-35,122.30
-40,629.16
-25,067.20
-30,764.86
-5,303.42
-11,390.64
-53,898.58
-42,613.60
-18,867.06
-20,412.74
-20,412.74
installations
Building
[€]
NPV30
Legend:
WFC=wood frame construction; CAV=cavity wall; OUT=massive wall with outer insulation
* Umean is including windows; U-value of the glazing is 0.6W/m²K, U-value of the window frames is 0.65W/m²K; all south oriented windows have outer sun
shading
WFC
1
Type
*
Building
Table H.1 presents more extensive details on the zero energy houses from table 8.6.
ANNEX H: Details on zero energy houses
ANNEX H
4868
5065
5065
5065
5065
5065
5065
5065
5065
5065
5065
5065
2
3
4
5
6
7
8
9
10
11
12
13
6
6
6
6
6
6
6
6
6
6
6
8
4
10,330
10,330
10,330
10,330
10,330
10,330
10,330
10,330
10,330
10,330
10,330
13,300
7,370
[€]
[m²]
31
31
31
31
31
31
31
31
31
31
31
30
30
[m²]
Area
1 - 16.8 - Suntechnics - STM 210 F
PV-type
Invertor-type
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
4.60 - 94.4 - Kaco - PVI 4000i
PV-modules
Legend:
*
GCC = net present value of the future green current certificates over 20 years
**
electricity = net present value of the produced electricity over 25 years
4868
1
Invest
Area
collectors
production
[kWh/m²a]
Solar
Electricity
33,945
33,945
33,945
33,945
33,945
33,945
33,945
33,945
33,945
33,945
33,945
37,031
37,031
[€]
Invest
**
42,897.00
42,897.00
42,897.00
42,897.00
42,897.00
42,897.00
42,897.00
42,897.00
42,897.00
42,897.00
42,897.00
41,181.00
41,181.00
Electricity
GCC* +
-4,325.00
-4,325.00
-4,325.00
-4,325.00
-4,325.00
-4,325.00
-4,325.00
-4,325.00
-4,325.00
-4,325.00
-4,325.00
-6,943.97
-1,014.97
solar systems
Overall NPV25 for
Net present value over 25 years
REFERENCES
REFERENCES
Adalberth K., 1997(1), Energy use during the life cycle of buildings: a method, Building and
Environment 32 (1997) n° 4, 317-320
Adalberth K., 1997(2), Energy use during the life cycle of single-unit dwellings: examples, Building and
Environment 32 (1997) n°4, 321-329
Alanne K., Saari A., Ismet Ugursal V., Good J., 2006, The financial viability of an SOFC cogeneration
system in single-family dwellings, Journal of Power Sources 158 (2006) 403-416
Annex31 2004, Sensitivity and Uncertainty, Background report of IEA Annex 31 on Energy-Related
Environmental Impact of Buildings, CMHC on behalf of IEA ECBCS, Canada
Anon. 2001(1), Greening the building life cycle. Life cycle assessment tools in building and
construction. Building LCA, tools description, Environment Australia, Department of the Environment
and Heritage
Anon. 2001(2). Background Report LCA Tools, Data and Application in the Building and Construction
Industry, Centre for Design, RMIT University for Environment Australia, Department of the
Environment and Heritage
Antonides G., Kroft M., 2005, Fairness judgments in household decision making, Journal of Economic
Psychology 26 (2005) 902-913
Asiedu Y, Besant R.W., Gu P, 2000, HVAC Duct system design using genetic algorithms, HVAC&R
Research, April 2000, 149-173
Astrup Jensen A., Hoffman L., Moller B., Schmidt A., 1997, Life cycle assessment (LCA), a guide to
approaches, experiences and information sources, Environmental Issues Series n° 6, European
Environment Agency
Baker J.E., 1987, Reducing bias and inefficiency in the selection algorithm, Proceedings of the
Second International Conference on Genetic Algorithms, Lawrence Erlbaum Associates, Hillsdale, NJ,
pp.14-21
Banfi S., Farsi M., Filippini M., Jakob M., 2006, Willingness to pay for energy-saving measures in
residential buildings, Energy Economics 2006, article in press
Beasley D., Bull D.R., Martin R.R., 1993, An overview of Genetic algorithms: Part 1, Fundamentals,
University Computing, 1993, 15(2) 58-69
Beasley D., Bull D.R., Martin R.R., 1993, An overview of Genetic algorithms: Part 2, Research Topics,
University Computing, 1993, 15(4) 170-181
Berghmans J., Duprez H., 1997, Toepasbaarheid van de warmtepomp in woningen in België,
Eindrapport CO2-project ‘Kennis van CO2-emissies’ (SP-3), fase 1, TME report
Berkhout P.H.G., Muskens J.C., Velthuijsen J.W., 2000, Defining the rebound effect, Energy Policy 28
(2000) 425-432
Binz A., Erb M., Lehmann G., 2000, Ökologische Nachhaltigkeit im Wohnungsbau, Eine Bewertung
von Erneuerungsstrategien, Forschungsprogrammem “Rationelle Energienutzung in Gebäuden”,
Fachhochschule beider Basel, Institut für Energie Muttenz, April 2000
Birol F., Keppler J.H., 2000, Prices, technology development and the rebound effect, Energy Policy 28
(2000) 457-469
Blanchard S., Reppe P., 1998, Life cycle analysis of a residential home in Michigan, Report N° 19985, September 1998, Center for Sustainable Systems, University of Michigan
Booker L.B., 1987, Improving search in genetic algorithms, in: Davis L. (editor), ‘Genetic Algorithms
and Simulated Annealing’, Morgan Kaufmann Publishers, San Mateo, CA, pp. 61-73
Börjesson P., Gustavsson L., 2000, Greenhous gas balances in building construction: wood versus
concrete from life-cycle and forest land-use perspectives. Energy Policy 28 pp 575-588
261
Bossier F., Bracke I., Van Horebeek F., 2002, The impacts of energy and carbon taxation in Belgium,
Analysis of the impacts on the economy and on CO2 emissions, Federal Planning Bureau, February
2002
Brand G., Braunschweig A., Scheidegger A., Schwank O., 1998, Bewertung in Oekobilanzen mit der
Methode der ökologischen Knappheit Oekofaktoren 1997, BUWAL Schriftenreihe Umwelt Nr. 297.
BUWAL, Bern
Camerer C.F., Loewenstein G., 2002, Behavioral economics: Past, present and future, draft version
October 2002
Carlson R., Tillman A-M., 1998, Data model for product related environmental assessment: SPINE,
International workshop on ‘Systems engineering models for waste management’, 25-26 February,
Göteborg, Sweden,
Chevalier J.L., Le Teno J.F., 1996, Requirements for an LCA-based model for the evaluation of the
environmental quality of building products, Building and Environment, vol. 31, n° 5, pp487-491
Chevalier J.L., Krogh H., Tarantini M., 2002, IEA-SHC Task 27 : Environmental performance
assessment of glazing and windows : context, overview, main concerns. Task 27 Workshop, Ottawa,
Canada, October 2002
Chow T.T., Zhang G.Q., Lin Z., Song C.L., 2002, Global optimization of absorption chiller system by
genetic algorithm and neural network, Energy and Buildings 34 (2002) 103-109
Christian J.E., Beal D., Kerrigan Ph., 2004, Toward simple, affordable zero energy houses,
Proceedings of the Performance of Exterior Envelopes of Whole Buildings IX International
Conference, Clearwater Beach, Florida, USA, December 5-10, 2004
Citherlet S., Di Guglielmo F., Gay J-B., 2000, Window and advanced glazing systems life cycle
assessment, Energy and Buildings 32 (2000) 225-234
Clinch J.P., Healy J.D., 2000, Domestic energy efficiency in Ireland: correcting market failure, Energy
policy 28 (2000) 1-8
CMLCA Chain Management by Life Cycle
http://www.leidenuniv.nl/cml/ssp/software/cmlca
Assessment,
educational
software
tool
Coello Coello C.A., 1996, An empirical study of evolutionary techniques for multiobjective optimisation
in engineering design, PhD thesis, Department of Computer Science, Tulane University, New Orleans,
LA, April 1996
Coello Coello C.A., 2000, Constraint-handling using an evolutionary multiobjective optimisation
technique, Civil Engineering and Environmental Systems, 17 (2000) 319-346
Coello Coello C.A., Montes E.M., 2002, Constraint-handling in genetic algorithms through the use of
dominance-based tournament selection, Advanced Engineering Informatics 16 (2002) 193-203
Coley D.A., Schukat S., 2002, Low-energy design: combining computer-based optimisation and
human judgement, Building and Environment 37 (2002) 1241-1247
Comer & Associates, LLC 2000, Next steps on the road to zero energy buildings, Report of meeting,
October 23-24, 2000, National Renewable Energy Laboratory, Golden, Colorado
COMIS 2003, COMIS Multizone Air flow model http://epb.lbl.gov/comis/users.html
Consoli F., Allen D., Boustead I., Fava J., Franklin W., Jensen A.A. et al. 1993, A code of practice.
Guidelines for life-cycle assessment, Pensacola, USA: SETAC
Cuijpers C., 1996, How rational behaviour undermines improved energy-efficiency: a micro-economic
analysis (in Dutch), Energie & Milieu nr.5, September/oktober 1996
Darby S., 2006, Social learning and public policy: Lessons from an energy-conscious village, Energy
Policy 34 (2006) 2929-2940
Dasgupta D., Michalewicz Z., 1997, Evolutionary algorithms in engineering applications, Springer
Verlag, Berlin
262
REFERENCES
Deb K., Goldberg D.E., 1989, An investigation of niche and species formation in genetic function
optimisation, Proceedings of the Third International Conference on Genetic Algorithms, Morgan
Kaufmann, San Mateo, CA, pp. 42-50
De Coninck R., Verbeeck G., 2005, Technical-economic analysis of the cost-effectiveness of energy
saving investments (in Dutch), Report by order of the Brussels Institute for the Management of the
Environment, 128pp
De Meulenaer V., Verbeeck G., Van der Veken J., Hens H., 2005(1), Performance assessment of
passive houses based on extensive measuring, Proceedings of 7th Nordic Building Physics
Symposium, Reykjavik, Iceland, June 13-15, 2005
De Meulenaer V., Van der Veken J., Verbeeck G., Hens H., 2005(2), Comparison of measurements
and simulations of a passive house, Proceedings of the 9th International Building Performance
Simulation Association Conference, Montreal, Canada, August 15-18, 2005
Devogelaer D., Gusbin D., Bassilière D., Bossier F., Bracke I., Thiery F., Vandevyvere W., Henry A.,
Gouzée N., 2006, Het klimaatbeleid na 2012: Analyse van scenario’s voor emissiereductie tegen 2020
en 2050. Federaal Planbureau, Juli 2006
Dewick P., Miozzo M., 2002, Sustainable technologies and the innovation-regulation paradox, Futures
34 (2002) 823-840
D’haeseleer W. (ed.), 2006, Energie, vandaag en morgen. Beschouwingen over energievoorziening
en –gebruik,Technologisch Instituut, Koninklijke Vlaamse Ingenieursvereniging, uitgeverij Acco
Dixit A.K., Pindyck R.S., 1994, Investment under uncertainty, Princeton University Press, Princeton,
New Jersey
Doka 2004, LCA hotlist, Links about Life Cycle Assessment, http://www.doka.ch/lca.htm
Duflou J., De Wulf W., 2006, Interdisciplinair College Duurzame Ontwikkeling, les 4 Duurzame
Productontwikkeling
Ecobalance, 2006, http://www.ecobalance.com/ LCA History
EMOO Web page with complete list of references related to evolutionary multi-objective optimisation,
http://www.lania.mx/~ccoello/EMOO/
EN ISO 13790:2004, Thermal performance of buildings - Calculation of energy use for heating
Energy Policy, 2000, Special Issue on the rebound effect, Volume 28, Issue 6-7, 351-500
Enquête 1998, Mens en Ruimte
Enquête 2001, Enquëte Energiegebruik Huishoudens in Vlaanderen in 2001, Iris Consulting,
December 2001
Enquête 2003, Enquëte Energiezuinig gedrag Vlaamse huishoudens in 2003, Iris Consulting,
September 2003
Enquête 2005, Enquëte Energiegebruik in huishoudens in Vlaanderen 2005, GfK Group, December
2005, http://www.energiesparen.be/energiegegevens/statistieken.php
EPB Besluit 2005, Besluit van de Vlaamse Regering tot vaststelling van de eisen op het vlak van de
energieprestaties en het binnenklimaat van gebouwen, March 11th 2005
EPB Besluit Bijlage I 2005, Bepalingsmethode van het karakteristiek jaarlijks primair energieverbruik
van woongebouwen, March 11th 2005
EPB Besluit Bijlage II 2005, Bepalingsmethode van het karakteristiek jaarlijks primair energieverbruik
van kantoor- en schoolgebouwen, March 11th 2005
EPBD 2002, Directive 2002/91/EC of the European Parliament and of the Council of 16 December
2002 on the energy performance of buildings, Official Journal of the European Communities, January
4th 2003
http://europa.eu.int/eur-lex/pri/en/oj/dat/2003/l_001/l_00120030104en00650071.pdf
Erickson M., Mayer A., Horn J., 2002, Multi-objective optimal design of groundwater remediation
systems: application of the niched Pareto genetic algorithm (NPGA), Advances in Water Resources 25
(2002) 51-65
263
Erlandsson M., Levin P., Myrhe L., 1997, Energy and environmental consequences of an additional
wall insulation of a dwelling, Building and Environment 1997 Vol.32 No 2 pp.129-136
Erlandsson M., Borg M., 2003, Generic LCA-methodology applicable for buildings, constructions and
operation services – today practice and development needs, Building and Environment 38 (2003) 919938
EU 2004, European Commission, Directorate General for Energy and Transport: European energy
and transport scenarios on key drivers, September 2004
Fanger P. O., 1972, Thermal comfort. Analysis and application in environmental engineering, New
York
Feist W., 2006, 15 Jähriges Jubiläum für das Passivhaus Darmstadt-Kranichstein, September 2006
Fleming P.J., Purshouse R.C., 2002, Evolutionary algorithms in control systems engineering: a survey,
Control Engineering Practice 10 (2002) 1223-1241
Fonseca C.M., Fleming P.J., 1993, Genetic algorithms for multiobjective optimisation: formulation,
discussion and generalization, Proceedings of the Fifth International Conference on Genetic
Algorithms, San Mateo, CA
Fonseca C.M., 1995, Multiobjective genetic algorithms with application to control engineering
problems, PhD thesis, Department of Automatic Control and Systems Engineering, University of
Sheffield, September 1995
Fonseca C.M., Fleming P.J., 1995, An overview of evolutionary algorithms in multiobjective
optimisation, Evolutionary Computation 1995 3(1):1-16
Frischknecht R., Jungbluth N. (ed.), 2003, Overview and methodology, Data v1.01 (2003), ecoinvent
report n° 1, Dübendorf, December 2003
Frischknecht R., Rebitzer G., 2005, The ecoinvent database system: a comprehensive web-based
LCA database, Journal of Cleaner Production 13 (2005) 1337-1343
Fossdal S., 1996, Windows in existing Norwegian buildings in a sustainable perspective, Sustainalbe
use of materials, seminar at BRE, Watford, UK, September 1996
Garcia-Quijano J.F., Deckmyn G., Moons E., Proost S., Ceulemans R., Muys B., 2005, An integrated
decision support framework for the prediction and evaluation of efficiency, environmental impact and
total social cost of domestic and international forestry projects for greenhouse gas mitigation:
description and case studies, Forest Ecology and Management 207 (2005) 245-262
GA Toolbox, 1994, http://www.shef.ac.uk/acse/research/ecrg/gat.html
GEMIS 4.14, 2002, Global Emission Model for Integrated Systems, (now GEMIS 4.3, 2006 available
on http://www.oeko.de/service/gemis/ )
Gens R., 2001, Mehrkriterielle Entscheidungsfindung – optimiertes Pareto-Ranking für Matlab,
Technischer Bericht, Technische Universität Ilmenau, Germany
Gluch P., Baumann H., 2004, The life cycle costing (LCC) approach: a conceptual discussion of its
usefulness for environmental decision-making, Building and Environment 39 (2004) 571-580
Goedkoop M., Spriensma R., 1999, The Eco-indicator 99: A damage oriented method for life cycle
impact assessment, PRé Consultants, Amersfoort, The Netherlands
Goedkoop M., Spriensma R., 1999, Methodology Annex: The Eco-indicator 99: A damage oriented
method for life cycle impact assessment, PRé Consultants, Amersfoort, The Netherlands
Goldberg D.E., Richardson J., 1987, Genetic algorithms with sharing for multimodal function
optimisation, Proceedings of the Second International Conference on Genetic Algorithms, Lawrence
Erlbaum, pp. 41-49
Goldberg D.E., 1989, Genetic Algorithms in Search, Optimisation and Machine Learning, AddisonWesley, Reading, Massachussets
Greening L.A., Greene D.L., Difiglio C., 2000, Energy efficiency and consumption – the rebound effect
– a survey, Energy Policy 28 (2000) 389-401
264
REFERENCES
Grefenstette J.J., 1992, Genetic algorithms for changing environments, in: Männer R., Manderick B.
(editors), Parallel Problem Solving from Nature, 2, North-Holland, Amsterdam, pp. 137-144
Guinée J.B., Gorree M., Heijungs R., Huppes G., Kleijn R., van Oers L. et al. 2002, Handbook on life
cycle assessment – operational guide to the ISO standard. Dordrecht, The Netherlands: Kluwer
Academic Publishing
Haas R., Biermayr P., 2000, The rebound effect for space heating. Empirical evidence from Austria,
Energy Policy 28 (2000) 403-410
Hancock P.J.B., 1994, An empirical comparison of selection methods in evolutionary algorithms,
Evolutionary Computing, AISB Workshop, volume 865 of Lecture Notes in Computer Science,
Springer-Verlag Berlin, pp.80-94
Hauschild M., Wenzel H. (eds), 1998, Environmental assessment of products vol. 2, London, UK:
Chapman & Hall
Heijungs R., 1996, Identification of key issues for further investigation in improving the reliability of lifecycle assessments, Journal of Cleaner Production, 1996, vol. 4, n° 3-4, pp. 159-166
Heijungs R., Kleijn R., 2000, Numerical approaches towards life cycle interpretation: five examples,
CML-SSP Working Paper 2000.001, Leiden 5 June 2000
Heijungs R., Suh S., Kleijn R., 2005, Numerical approaches to Life Cycle Interpretation. The case of
the ecoinvent’96 database, International Journal of LCA, 2005, 10(2) pp. 103-112
Hendriks L., Hens H., 2000, Building Envelopes in a Holistic Perspective, IEA Annex 32, Report on
Methodology, July 2000, Amsterdam, The Netherlands
Henly J., Ruderman H., Levine M.D., 1988, “Energy Saving Resulting from the Adoption of More
Efficient Appliances: A Follow-up,” Energy Journal, Vol. 9, No. 2 (1988), pp 163-170
Hens H., Verdonck B., 1997, Wonen, verwarmen: energie en emissies, CO2 project, fase 1, deel 3
“Analyse van toepassingen”, Electrabel + SPE
Hens H., 2005, Cost efficiency of PUR/PIR insulation, Report 04/14, by order of the Federation of
European Rigid Polyurethane Foam Association
Hinchy M.D., Naughten B.R., Donaldson P.K., Belcher S., Ferguson E., 1991, The issue of domestic
energy market failure, Technical Paper 91.5, Australian Bureau of Agricultural and Resource
Economics
Hobday R. (ed.), 2005, Technical Synthesis Report Annex 31 Energy-Related Environmental Impact
of Buildings, FaberMaunsell Ltd, UK
Holland J.H., 1975, Adaptation in Natural and Artificial Systems, Ann Harbor : University of Michigan
Press
Horn J., Nafpliotis N., Goldberg D.E., 1994, A niched Pareto genetic algorithm for multiobjective
optimisation, Proceedings of the First IEEE Conference on evolutionary computation, IEEE World
Congress on Computational Intelligence, Vol. 1, 1994 (ICEC ’94) pp 82-87
Hubbard R.G., 1994, Investment under uncertainty: Keeping one’s options open, Journal of Economic
Literature, vol 32, n° 4 (Dec 1994)
Huijbregts M.A.J., 1998, Application of uncertainty and variability in LCA (Part 1) – a general
framework for the analysis of uncertainty and variability in life cycle assessment, International Journal
of LCA 1998 3(5), pp. 273-280
ICEDD 2004, Energy balance of the Brussels Capital Region Gewest 2002, May 2004
IMPACT 2000+, http://www.epfl.ch/impact
ISO 14040, 1997, Environmental management – Life cycle assessment – Principles and framework
ISO 14041, 1998, Environmental management – Life cycle assessment – Goal and scope definition
and inventory analysis
ISO 14042, 2000, Environmental management – Life cycle assessment – Life cycle impact
assessment
265
ISO 14043, 2000, Environmental management – Life cycle assessment – Life cycle interpretation
ISO 14048, 2001, Environmental management – Life cycle assessment – Data documentation format
ISO/TR 14049, 2000, Technical Report Environmental management – Life cycle assessment –
Examples of application of ISO 14041 to goal and scope definition and inventory analysis
ISO 15686-1, 2000, Building and Constructed Assets – Service Life Planning – General Principles
ISSO/SBR 1994, Energie-efficiënte kantoorgebouwen: binnenklimaat en energiegebruik, Rotterdam
Jaccard M., Bataille C., 2000, Estimating future elasticities of substation for the rebound debate,
Energy Policy 28 (2000) 451-455.
Jacob M., 2006, Marginal costs and co-benefits of energy efficiency investments The case of the
Swiss residential sector, Energy Policy 34 (2006) 172-187
Jönsson A., 2000, Tools and methods for environmental assessment of building products –
methodological analysis of six selected approaches, Building and Environment 35 (2000) 223-238
Jorgensen B.S., Syme G.J., Nancarrow B.E., 2006, The role of uncertainty in the relationship between
fairness evaluations and willingness to pay, Ecological Economics 56 (2006) 104-124
Kalorigou S.A., 2004, Optimization of solar systems using artificial neural-networks and genetic
algorithms, Applied Energy 77 (2004) 383-405
Kahneman D., Tversky A. (ed.), 2000, Choices, values and frames, Russel Sage Foundation,
Cambridge University Press
KBC 2003, Woningschatter, www.kbc.be/wonen, version of 2003
Knockaert J., Proost S., 2005, Transport sector, in Willems B., Eyckmans J., Proost S., Economic
aspects of climate change policy, a European and Belgian perspective, Acco, Leuven, p.99-110
Lee W.L., Yik F.W.H., 2004, Regulatory and voluntary approaches for enhancing building energy
efficiency, Progress in Energy and Combustion Science 30 (2004) 477-499
Lenzen M., Treloar G., 2002, Embodied energy in buildings: wood versus concrete – reply to
Börjesson and Gustavsson, Energy Policy 30 (2002) 249-255
LISA 3.0, 2000, LCA in Sustainable Architecture Tool, http://www.lisa.au.com
Lotov A.V., 2001, Methodology and application of Pareto frontier visualization in DSS for the design of
water quality improvement strategies in multi-regional river basins
Maeyens J., 2001, Design rules for summer comfort in dwellings (in Dutch), Master thesis, University
of Ghent, Belgium
Maurice B., Frischknecht R., Ceolho-Schwirts V., Hungerbühler K., 2000, Uncertainty analysis in life
cycle inventory. Application to the production of electricity with French coal power plants, Journal of
Cleaner Production 8 (2000) pp.95-108
Meadows D.H., Meadows D.L., Randers J., Behrens W.W., 1972, Limits to Growth, Potomac
Associates, New York
Michalewicz Z., Xiao J., 1995, Evaluation of Paths in Evolutionary Planner/Navigator, Proceedings of
the 1995 International Workshop on Biologically Inspired Evolutionary Systems, Tokyo, Japan, May
30-31, pp. 45-52
Michalewicz Z, Dasgupta D, Le Riche R.G., Schoenauer M, 1996, Evolutionary algorithms for
constrained engineering problems, Computers & Industrial Engineering Journal, Vol.30, No.2,
September 1996, pp.851-870.
Michalewicz Z., 1999, Genetic Algorithms + Data Structures = Evolution Programs, Third, Revised and
Extended version, Springer-Verlag Berlin
Milne G., Boardman B., 2000, Making cold homes warmer: the effect of energy efficiency
improvements in low-income homes, Energy Policy 28 (2000) 411-424
Ministry of Economic Affairs http://www.mineco.fgov.be
266
REFERENCES
Mithraratne N., Vale B., 2004, Life cycle analysis model for New Zealand houses, Building and
Environment 39 (2004) 483-492
Moons E., 2003, The development and application of economic valuation techniques and their use in
environmental policy – A survey, Working Paper Series n° 2003-7, September 2003, Center for
Economic Studies - Energy, Transport & Environment, Catholic University of Leuven, Belgium
NBN EN12831, 2003, Heating systems in buildings – Method for calculation of the design heat load
NBN D50-001, 1991, Ventilation systems for housings, Belgian Institute for Normalisation, Octobre
1991
Norris G., The many dimensions of uncertainty analysis in LCA, http://www.athenasmi.ca
NREL-LCI, Working Paper No.
http://www.nrel.gov/lci.planning.html
6
Data
quality,
variability
and
uncertainty
in
LCI,
NREL, 2002, US LCI Database Project, Final Phase I report, Athena Sustainable Materials Institute,
Franklin Associates, Sylvatica, http://www.nrel.gov/lci/phase1.html
Obitko M., 1998, Introduction to genetic algorithms http://cs.felk.cvut.cz/~xobitko/ga/
Ozturk H.K., Canyurt O.E., Hepbasli A., Utlu Z., 2004, Residential-commercial energy input estimation
based on genetic algorithm (GA) approaches: an application of Turkey, Energy and Buildings 36
(2004) 175-183
Ozturk H.K., Canyurt O.E., Hepbasli A., Utlu Z., 2004, Three different genetic algorithm approaches to
the estimation of residential exergy input/output values, Energy and Buildings, 39 (2004) 807-816
Pareto V. 1896, Cours d’economie politique, Volume 1, Lausanne F. Rouge
Passivhaus Institut, Darmstadt, 2000, Zertifikat Fensterrahmen, Eurotec GmbH, Holzwarmfenster
serie 0.5; HEUSER Türen + Fenster-Metallbau GmbH, Super-Warmfenster U 07 Serie H 3200-120
PU; Woschko Winlite GmbH, Woschko Winplus
Peuportier B., Kohler N., Boonstra Ch., 1997, European project REGENER Life cycle analysis of
buildings. Proceedings of the 2nd International Conference Buildings and the Environment, Paris,
France, June 9-12 1997, pp. 33-40
Peuportier B.L.P., 2001, Life cycle assessment applied to the comparative evaluation of single family
houses in the French context, Energy and Buildings 33 (2001) 443-450
Pohlheim H., 1999, Visualization of evolutionary algorithms – set of standard techniques and
multidimensional visualization, GECCO’99 Proceedings of the Genetic and Evolutionary Computation
Conference, San Francisco, CA: Morgan Kaufmann, pp.533-540
Powell D., Skolnick M. M., 1993, Using genetic algorithms in engineering design optimisation with nonlinear constraints, Proceedings of the Fifth International Conference on Genetic Algorithms, Morgan
Kaufmann, San Mateo, CA, pp.424-430
Pré Consultants, 2007, Eco-Indicator 99 method, http://www.pre.nl/eco-indicator99/default.htm
Proost S., Van Dender K., Courcelle C., De Borger B., Peirson J., Sharp D., Vickerman R., Gibbons
E., O’Mahony M., Heaney Q., Van den Bergh J., Verhoef E., 2002, How large is the gap between
present and efficient transport prices in Europe?, Transport Policy 9 (2002) 41-57
Proost S., Van Regemorter D., 2000, How to achieve the Kyoto Target in Belgium? – modeling
methodology and some results, Working Paper Series n° 2000-09, December 2000, Center for
Ramesohl S., 1999, Opening the black box – What can be learned from socio-economic research for
energy policy analyses? IEA International Workshop on technologies to reduce greenhouse gas
emissions: Engineering-economic analyses of conserved energy and carbon, Washington, D.C., USA,
May 5th-7th 1999
Rebitzer G., et al., 2004, Review: Life cycle assessment Part 1: Framework, goal and scope definition,
inventory analysis and applications, Environment International 30 (2004) 701-720
267
REGENER 1997, European methodology for the evaluation of environmental impact of buildings.
Application of the life cycle analysis to buildings, Detailed description and review, Final report of EU
REGENER project, January 1997
Richardson J.T., Palmer M.R., Liepins G., Hilliard M., 1989, Some guidelines for genetic algorithms
with penalty functions, Proceedings of the Third International Conference on Genetic Algorithms,
Morgan Kaufmann, San Mateo, CA, pp 191-197
Rousseau S., Proost S., 2004,The relative efficiency of market-based environmental policy
instruments with imperfect compliance, Working Paper Series n° 2004-15, November 2004, Center for
Sanne C., 2000, Dealing with environmental savings in a dynamical economy – how to stop chasing
your tail in the pursuit of sustainability, Energy Policy 28 (2000) 487-495
Scheuer C., Keoleian G.A., Reppe P., 2003, Life cycle energy and environmental performance of a
new university building: modeling challenges and design implications, Energy and Buildings 35 (2003)
1049-1064
SENVIVV, 1998, Study of the energy aspects of new built dwellings in Flanders, Final report (in
Dutch), WTCB, Brussels, Belgium
SETAC, 2006, http://www.setac.org
Schipper L., 2000, Editorial On the rebound: the interaction of energy efficiency, energy use and
economic activity. An introduction, Energy Policy 28 (2000) 351-353
Sidoroff S., 2004, Data needs and sources. Background Report of IEA Annex 31
Siedlecki W., Sklanski J., 1989, Constrained genetic optimisation via dynamic reward-penalty
balancing and its use in pattern recognition, Proceedings of the Third International Conference on
Genetic Algorithms, Morgan Kaufmann, San Mateo, CA, pp. 141-150
SimaPro, 2006, Inventory data in the Life Cycle Tool SimaPro 7, Pré Consultants, The Netherlands,
http://www.pre.nl/simapro/inventory_databases.htm
Smith A.E., Coit D.W., 1995, Penalty Functions, Section C 5.2 of Handbook of Evolutionary
Computation, Baeck T., Fogel D., Michalewicz Z. (editors) Oxford University Press and Institute of
Physics Publishing
Sonneman G.W., Schuhmacher M., Castells F., 2003, Uncertainty assessment by a Monte Carlo
simulation in a life cycle inventory of electricity produced by a waste incinerator, Journal of Cleaner
Production 11 (2003) 279-292
SPINE, 2000, Swedish national LCA database, http://www.globalspine.com
STEM, 2004, Determinants of the energy consumption of households, Research group STEM,
University of Antwerp, Final report (in Dutch), October 2004
TG4, 2003, Final Report Task Group 4: Life cycle costs in construction, version 29 October 2003
Thormark C., 2002, A low energy building in a life cycle – its embodied energy, energy need for
operation and recycling potential, Building and Environment 37 (2002) 429-435
Tommerup H., Svendsen S., 2006, Energy savings in Danish residential building stock, Energy and
Buildings 38 (2006) 618-626
TRNSYS, 2005, Transient Systems Simulation Program, Solar Energy Laboratory, University of
Wisconsin, Madison http://sel.me.wisc.edu/trnsys/
Udo de Haes H.A., Lindeijer E., 2002, The conceptual structure of life-cycle impact assessment. In:
Udo de Haes et al., Life-cycle impact assessment: striving towards best practice. Pensacola, Florida,
USA: SETAC Press 2002
UWME, 2006, Design for Environment Lab, Inventory of Inventory
http://faculty.washington.edu/cooperjs/Definitions/inventory%20squared.htm
Data
Sources
van den Berg N. W., Huppes G., Lindeijer E.W., van der Ven B.L., Wrisberg M.N., 1999, Quality
assessment for LCA, CML Report 152, Leiden, The Netherlands
Vandenborre, 2007, www.vandenborre.be website of retail store of domestic electrical appliances
268
REFERENCES
Van Londersele E. Janssens A., 2007, Subreport 2b on ‘Constructional optimisation: constructional
solutions and details’ (in Dutch) of Final reports of IWT GBOU 020212 project Development of
extremely low energy and low pollution dwellings through life cycle optimisation, February 2007.
van Rooijen S.N.M., van Wees M.T., 2006, Green electricity policies in the Netherlands: an analysis of
policy decisions, Energy Policy 34 (2006) 60-71
Van Steertegem M., 2001, MIRA-T Milieu- en natuurrapport Vlaanderen: thema’s, Vlaamse
Milieumaatschappij, Garant Uitgevers
Van Veldhuizen D.A., Lamont G.B., 1998, Evolutionary computation and convergence to a Pareto
front, July 1998, Late Breaking Papers at the Genetic Programming 1998 Conference, pp 221-228,
Stanford University, California
Van Veldhuizen D.A., 1999, Multiobjective evolutionary algorithms: classifications, analyses and new
innovations, PhD thesis, Department of Electrical and Computer Engineering. Graduate School of
Engineering. Air Force Institute of Technology, Wright-Patterson AFB, Ohio, May 1999.
Venkatarama Reddy B.V., Jagadish K.S., 2003, Embodied energy of common and alternative building
materials and technologies, Energy and Buildings 35 (2003) 129-137
Verbeeck G., Hens H. 2002 Energiezuinige renovaties: economisch optimum, rendabiliteit, rapport
Project Kennis van de CO2 emissies, fase 4, Electrabel en SPE
Verbeeck G., 2003, Optimisation of extremely low energy residential buildings, Aim and state of the
PhD research after nearly 1 year, Doctoral seminar November 21th 2003
Verbeeck G., Hens H., 2005, Energy savings in retrofitted dwellings: economically viable? Energy and
Buildings (37) 2005, 747-754
Verbeeck G., Vanderveken J., De Meulenaer V., Hens H., Willems L., Vanlondersele E., Janssens A.,
Peeters L., D’haeseleer W., Vermeyen P., Driesen J., 2007, Final reports of IWT GBOU 020212
project Development of extremely low energy and low pollution dwellings through life cycle
optimisation, February 2007
Verdonck B., Hens H., 1998, Economic rebound in the quest for energy conservation by envelope
insulation, IEA Annex 32, Copenhagen Meeting April 15th-17th 1998
Verdonck B., 1999, Economic rebound effect (in Dutch), Internal document Laboratory of Building
Physics, Catholic University of Leuven, Belgium
Vermeulen W.J.V., Hovens J., 2006, Competing explanations for adopting energy innovations for new
office buildings, Energy Policy 34 (2006) 2719-2735
Vermeyen
P.,
2007,
PV
calculation
module
and
manual
(in
Dutch),
http://homes.esat.kuleuven.be/~pvermeye/ Subreport 3b ‘Optimisation of installations: PV-systems
and DC-net’ of Final reports of IWT GBOU 020212 project Development of extremely low energy and
low pollution dwellings through life cycle optimisation, February 2007
von Bahr B., Steen B., 2004, Reducing epistemological uncertainty in life cycle inventory, Journal of
Cleaner Production 12 (2004) 369-388
Voorspools K.R., Brouwers E.A., D’haeseleer W.D., 2000, Energy content and indirect greenhouse
gas emissions embedded in ‘emission-free’ power plants: results from the Low Countries, Applied
Energy 67 (2000) 307-330
Vrugt J.A, Gupta H.V., Bastidas L.A., Bouten W., Sorooshian S., 2003, Effective and efficient
algorithm for multiobjective optimisation of hydrologic models, Water Resources Research (2003)
Vol.39, No 8, 1214
Wang S., Jin X., 2000, Model-based optimal control of VAV air-conditioning system using genetic
algorithm, Building and Environment 35 (2000) 471-487
Wang W., Zmeureanu R., Rivard H., 2005, Applying multi-objective genetic algorithms in green
building design optimisation, Building and Environment 40 (2005) 1512-1525
Wang W., Rivard H., Zmeurenau R., 2005, An object-oriented framework for simulation-based green
building design optimisation with genetic algorithms, Advanced Engineering Informatics 19 (2005) 523
269
Weidema B.P., 1999, SPOLD ’99 format – an electronic data format for exchange of LCI data
(1999.06.24). SPOLD www.spold.org
Weir G., Muneer T., 1998, Energy and environmental impact analysis of double-glazing windows,
Energy Conversion Management 1998 Vol. 39 No.3/4 pp 243-256
Wenzel H., Hauschild M., Alting L., 1997, Environmental assessment of products, vol. 1, methodology,
tools and case studies in product development, London, UK: Chapman & Hall
World Commission on Environment and Development, 1987, Our Common Future, Oxford University
Press, Melbourne, 1987
Wright J.A., Loosemore H.A., Farmani R., 2002, Optimisation of building thermal design and control by
multi-criterion genetic algorithm, Energy and Buildings 34 (2002) 959-972
WTCB, 1999, Technische voorlichting 214, Glas en glasproducten, functies van beglazing, december
1999
Zitzler E., 1999, Evolutionary algorithms for multiobjective optimisation: methods and applications,
PhD thesis, TIK-Schriftenreihe nr.30, Institut für Technische Informatik und Kommunikationsnetze,
ETH Zürich, Switzerland
270
CURRICULUM VITAE
CURRICULUM VITAE
Personal information
Date of birth
March 4th 1967
Place of birth
Antwerp
Gender
female
Current adress
Leuvensebaan 84, B-3220 Holsbeek
016/48.70.66
Martial status
Married to Stijn Ombelets, mother of Casper (♂
°2000), Mona(♀ °2002), Ybe (♀ °2005)
Education
1985
High School, ASO Latin-Greek, Sancta Maria Instituut,
Deurne
1986
Preparatory year mathematics K.U.Leuven
1991
Masters degree in civil engineering, architecture
K.U.Leuven, distinction
Previous and present positions
Sept 1991 – Sept 2000
Research Associate
Division
of Building Physics,
Civil
Engineering,
K.U.Leuven
Oct 2000 – Dec 2001
Engineer
Energy
administration,
Brussels
Capital
Region
(Brussels Institute for the Management of the
Environment)
Jan. 2002 – Jan 2006
Research Associate and PhD student
Division
of Building Physics,
Civil
Engineering,
K.U.Leuven
Feb. 2006 -
Lecturer and research associate at the Department of
Architecture of the University College PH Limburg
Research activities
Sept 1991 – Sept 2000
IWT-VLIET 930251 project: Transparent Insulation:
numerical models and evaluation in situ
Comfort studies in office buildings and large stores
CO2 projects ELECTRABEL on energy consumption
and
CO2
emissions
in
the
residential
sector:
constructional and installation related aspects
Jan. 2002 – Jan. 2006
CO2 project ELECTRABEL: Viability of energy savings
in retrofitted dwellings
BIM project: Technical-economical analysis on the
viability of energy saving investments
271
GBOU-project: Development of extremely low energy
and
low
pollution
dwellings
through
life
cycle
optimisation
Feb. 2006 -
VEA-study
on
the
economical
viability
of
strengthening the EPB requirements (to be started)
Development of a sustainability design tool for
architects (to be started)
Since 2003
Preparing a PhD, related to GBOU project, entitled:
Optimisation of extremely low energy residential
buildings
Scientific publications
In international journals:

Hens H, Verbeeck G., Verdonck B. 2001. Impact of energy efficiency measures on the CO2emissions in the residential sector, a large scale analysis, Energy and Buildings, 33, 275-281

Verbeeck G., Hens H. 2005 Energy savings in retrofitted dwellings: economically viable?
Energy and Buildings (37) 2005, 747-754

Verbeeck G., Hens H., 2007 Life cycle optimization of extremely low energy dwellings,
Journal of Building Physics (paper accepted)
In proceedings of international congresses:

Verbeeck G., 1996, Radiative and Conductive Heat Transfer through Transparent Insulation
Material and the Effect on the λ-value, Proceedings of the 4th Symposium on Building Physics
in the Nordic Countries and the IEA Annex 24 Closing Seminar, 9-11 september ‘96, Innopoli,
Espoo, Finland

Hens H., Verbeeck G., 1997, Performance assessment of TIM-envelopes, CLIMA 2000, 30
August - 2 September 1997, Brussels, Belgium

Verbeeck G., Hens H., 1997, Case study on the thermal comfort in a series of stores. CLIMA
2000, 30 August - 2 September 1997, Brussels, Belgium

Hens H., Verbeeck G., Verdonck B., 1998, Impact of Energy Efficiency Measures on the CO2
Emissions in the Residential Sector, a Large Scale Analysis, Proceedings of the Epic'98
Conference on Energy Performance and Indoor Climate in Buildings

Rabenseifer R., Verbeeck G., Hens H., , 1999, The implementation of the EN 832 into
national standards, 10th International Symposium for Buildings Physics and Buildings
Climatology, 27-29 September 1999, Dresden, Germany

Verbeeck G., Hens H., Hendrickx C., Ramaekers B., 1999, A collective heating system with
solar collectors for hot water production: a case study of 29 social houses in Belgium, Second
EDF Research Workshops, 28 October 1999, Clamart, France

Hens H., Verbeeck G., 2001, Heating efficiency, the great unknown, 7th World Congress
Clima, 15-18 September 2001

Hens H., Verbeeck G., Verdonck B., 2001, Energy consumption in the residential sector: a
database analysis, 7th World Congress Clima, 15-18 September 2001

Hens H., Verbeeck G., Verdonck B., 2001, Energy use for household: can it be modelled?,
Proceedings of the Clima 2000 Conference, Napoli, 15-18 September, 2001

Hens H., Verbeeck G., Stijnen L., Tomasetig B., 2002. Energy consumption in a low energy
estate, confronting measurements with overall data and predictions, Proceedings of the 11th
Bauklimatisches Symposium, Dresden, 26-30 September
272
CURRICULUM VITAE

Hens H., Ali Mohamed F., Verbeeck G., 2002, Using Energy Life Cycle Costs as an
Instrument for Optimisation, Proceedings of the Sustainable Building 2002 Conference, Oslo,
Norway, September 23-25

Hens H., Ali Mohamed F., Verbeeck G., 2002, Hydronic Radiator Heating with thermostatic
valves: Improves Thermal Comfort or Upgrades Efficiency?, Proceedings of the EPIC 2002
AIVC Conference, Lyon, France, October, 23-26

Hens H., Verbeeck G., 2002 Lage energie gebouwen: ontwerp en uitvoering, European
Green Cities Network, EGCN Conference and Training, 1-2 October 2002

Van der Veken J., Saelens D., Verbeeck G., Hens H., 2004, Comparison of steady-state and
dynamic building energy simulation programs, Proceedings of the Performance of Exterior
Envelopes of Whole Buildings IX International Conference, Clearwater Beach, Florida, USA,
December 5-10, 2004

De Meulenaer V., Verbeeck G., Van der Veken J., Hens H., 2005, Performance assessment
of passive houses based on extensive measuring, Proceedings of 7th Nordic Building Physics
Symposium, Reykjavik, Iceland, June 13-15, 2005

De Meulenaer V., Van der Veken J., Verbeeck G., Hens H., 2005, Comparison of
measurements and simulations of a passive house, Proceedings of the 9th International
Building Performance Simulation Association Conference, Montreal, Canada, August 15-18,
2005

Verbeeck G., Hens H., 2006, Development of extremely low energy dwellings through life
cycle optimisation, In 3th International Conference on Research in Building Physics, Montreal
Canada, August 27-31, 2006.

Hens H., De Meulenaer V., Van der Veken J., Verbeeck G., 2007, Balanced ventilation with
heat recovery: does it really enhance energy efficiency? In Proceedings of the 12th
Symposium for Building Physics, Dresden March 29th to 31st 2007

Verbeeck G., Hens H., 2007, Life cycle optimization of extremely low energy buildings,
CLIMA2007, Helsinki, Finland, June 10-14, 2007 (paper accepted)

Verbeeck G., Hens H., 2007, Life cycle inventory of extremely low energy dwellings,
PLEA2007, Singapore, Novembre 2007 (abstract in review)
273