Research Review 2010

Transcription

Research Review 2010
Research Review 2010
A
Research review
was published in 2010
Contents
05
05 Global
Global research
research
06
06
Arup’s
Arup research
researchbusiness
business
10
10
Climate
Climatechange
changeextremes:
extremes;the
the
26
38Monitoring geothermal piles
combined
combinedeffects
effectsofofstorm
stormtides
tides
and
andcatchment
catchmentflooding
floodingininCairns
Cairns
at Keble College, Oxford
30
42Sustainable and quake resistant
priorities
priorities
14
14
Geomechanical
Geomechanicalcharacterisation
characterisationofof
07
07 Meet
Meet the
the team
team
08
08
Regional
Regional research
research
champions
champions
Manhattan
Manhattanschist,
schist,aastudy
studyofofthe
the77
line
lineextension
extensioncavern
cavern
façade for existing buildings
34
46Measuring change of coastal
defence structures using advanced
3D laser mapping techniques
18
12
An
Aninvestigation
investigationofoffire
fireload
load
density
densityfor
forresidential
residentialbuildings
buildings
and
in Hong
offices
Kong
in Hong Kong
38
50Digital infrastructure and changing
practices in engineering design
22
12
Retrofitting
Retrofittingprivate
privatehomes
homesat scale:
investigating
at scale; investigating
the business
thecase
business case
26Transient thermal comfort
16
Transient in
thermal
comfortstations
modelling
underground
42
54Beasties in the creative workplace
46
58Neighbourhood Pedestrian
Analysis Tool (NPAT)
modelling in underground stations
30Singapore buried channel research
18Singapore buried channel research
34The Life Cycle Tower: a high-rise
22
Life construction
Cycle Tower; a high-rise
inThe
timber
in timber construction
50
62Human induced vibrations
on footbridges
56
66 Contact
The importance
information
of research
at Arup
Prof. Jeremy Watson, heads the
research strategy and research
business for Arup. His background is
in research and technical management
in both industry and academe. Jeremy
is a chartered engineer, and fellow of
the Institute of Engineering and
Technology. He is also a visiting professor
at the Universities of Southampton and
Sussex, a board member of the UK
Technology Strategy Board, and Chief
Scientific Advisor for the Department of
Communities and Local Government.
4
Global research
I am delighted to introduce Arup’s Research Review
2010, showcasing research undertaken with
collaborators around the world.
Arup seeks to deliver research outputs comparable with
world-class universities. I continue to be impressed and
excited by the quality and innovation of our client and
internal development projects.
Research is a key contributor to Arup’s success; we deliver
new, validated thinking in support of our clients’ projects
and aspirations. We maximise value by matching business
requirements with developments from the academic sector.
Arup experts work in collaboration with the best private
and public sector partners. An internal research investment
fund supports staff time and provides studentships and
other contributions to university collaborators.
Research has always differentiated Arup’s work and
we encourage it globally. Coordination has been
enhanced by the appointment of five Regional Research
Champions, responsible for local research strategy
and external collaboration.
Arup University
Arup’s research culture has been developed with the
recent launch of the Arup University, providing opportunities
for staff to train at Professional, Masters and Doctoral
levels. The Doctoral programme is administered by the
Research team and delivered in partnership with University
College London. Candidates can gain an accredited
Doctoral degree based on work-based research training.
Research Roadmap
Aimed at Arup’s research partners, including funding
councils, and informed by client-facing workshops, a new
Arup publication offers better coverage and detail than
before. The preferred tool for capturing Arup’s research
priorities, Roadmapping links global drivers with business
opportunities to deliver research agendas. We also offer a
customised service to clients developing research strategies,
and can broker and manage external research providers.
Knowledge is shared through a Research Wiki. Connecting
our offices and networks, this promotes collaboration to
support research objectives and our clients’ requirements.
Collaboration
Many Arup staff members are renowned in their fields,
publishing in peer-reviewed journals. However while we
have much expertise in house, our research offerings are
strengthened by strategic alliances and partnerships.
When working with individual universities and companies
we align our thinking with national research priorities
and leverage research investments through national
funding programmes.
Links with UK research councils are a strong asset.
We have a Strategic Partnership agreement with
Engineering and Physical Sciences Research Council
(EPSRC), which can support mutually developed research
programmes. Arup is currently using this agreement to
develop research at the Thames Gateway Institute for
Sustainability into Sustainable Urban Infrastructures.
Arup acts for the UK in an EU programme on Energy
Efficient Buildings, providing expertise as a board member of
a not-for-profit company which will administer up to €2bn
Public Private Partnership research funds over 10 years. We
provide the UK’s national contact point, helping to shape
E2B calls and promoting UK industry collaboration.
Such partnerships and research initiatives are exciting,
innovative and great to be involved with. As well as
allowing us to grow knowledge and demonstrate excellence
in many technical areas, they ensure we develop
relationships with the best in industry and academia to
provide research that adds value for our clients.
The case studies included in this Research Review go
some way to demonstrating our expertise but they are a
small selection chosen from the many projects available.
I hope that you find this Research Review interesting and
inspirational, and that you will want to collaborate with us.
Prof. Jeremy Watson
Director of Global Research
5
Arup research business priorities
Multidisciplinary projects require
specialist project management skills,
providing a high level of coordination
of a range of expert inputs.
We can help clients understand their
research needs, develop a strategy,
and then develop and manage
the research programme and deliver
multidisciplinary research projects
through offering a variety of services.
Strategic Roadmapping is a facilitated
workshop process which helps clients
to address strategic issues facing
their business and produce a multilevel
plan which has buy-in from across
their organisation.
The process involves assessing key
drivers, identifying the business
opportunities that arise from these
drivers, and reviewing the resources
and processes that are needed to meet
these opportunities; including
research, technical investments and
training. The workshop is usually
supported and informed by a desk
study which reviews current and
emerging issues in the field.
The Research Business Team’s role is
facilitation of the workshops, coordination
of expert inputs to desk studies, and
final reporting.
6
Research Programme Management
can provide high-level coordination
of research programmes, including
theme definition via stakeholder
consultation, calls for proposals,
peer review with external expert panels,
liaison with research councils and fund
administration. We are increasingly
interested in doing this with industry
consortia.
Sandpit workshops are intense three
to five day workshops used to shape
research programmes around a particular
topic and develop and fund project
proposals. They involve delegates in an
‘immersion’ process, learning about the
problem space, creating ideas for
research solutions, and identifying and
developing project ideas.
The Research Business Team can run
all elements of this process, from
assisting the client in managing delegate
applications, identifying stakeholders and
mentors for the sandpit, facilitating the
event itself, and helping manage the
resulting research projects.
Meet the team
Dr Marta Fernandez,
Research Relationships Manager
Dr Jennifer Schooling,
Research Business Manager
Geraldine Ralph,
Research Events and Projects Coordinator
Marta focuses on relationships between Arup’s
internal network and research partners externally,
as well as supporting our efforts to realise the
value of the firm’s Intellectual Property. She
represents the company on the operational group
of the E2B. A chemical engineer by training,
her previous roles included commercialising
early-stage technology in renewable energy
start-ups, and forging links between industry
and academe in the engineering, energy and
environmental sectors. Marta is an honorary
lecturer at University College London (UCL).
Jennifer manages multidisciplinary
research services, enabling clients and staff
to access the many skills within Arup, and
establishing successful collaborations with
external agencies. In previous roles, she
managed engineering research and
development projects for both academic
and commercial applications. Jennifer has
also handled new product development in
the semiconductor equipment industries,
managing new product introductions from
concept design to final launch.
Geraldine is responsible for all areas of
research communications including publications,
disseminating research activities and leveraging
Arup’s research network and partnerships
through events. She also provides project
support to the doctoral module programme
and CASE awards. Geraldine is a member of
the Chartered Institute of Public Relations.
Dr Rick Wheal,
Research Associate
Nausicaa Voukalis,
Research Associate
Jackie Young,
Personal Assistant
Rick supports the team with UK representation on
the Energy Efficient Buildings programme in the
EU. He also has a client-facing role dealing with
the commercialisation of research and realising
benefits throughout the supply chain. Rick has
a background in academe and also building
consultancy services, with a strong emphasis
on sustainability and architectural design.
Nausicaa is a specialist in the firm to
realise the value of its Intellectual Property,
and is responsible for any paralegal issues
required for commercial licensing and
collaboration agreements with Universities.
A chartered architect with an MBA, Nausicaa’s
previous roles also include working on both the
design and construction aspects of projects,
as well as the management of variable schemes,
ranging from small residential designs to
large urban planning projects.
Jackie provides full secretarial support to
Jeremy Watson and admin support to the
Research Team. She is often the point of
contact for both internal and external
research equiries.
7
Regional research champions
Tim Keer | Americas Region
Richard Hough | Australasia Region
Dr Ricky Tsui | East Asia Region
Tim Keer is a Principal in Arup’s New York Office
and is the Research Champion for the Americas
Region. His current responsibilities are focused
on operational improvement. He has recently
launched the Arup Americas Project Management
Academy and is leading various initiatives to
strengthen the region’s performance. His
technical background is in the analysis of
automotive structures and Tim has a particular
interest in the development and application of
techniques for non-linear analysis and design.
Richard Hough is a Principal in the Sydney
Buildings Group, and Australasia Region
D&T Leader. He is also chair of the Regional
Investment Coordination Group which oversees
locally-funded investment projects. This role
fits well with his regional research champion
position, and also with his chairing of the
Regional DTX, which promotes R&D and
innovation in the regional offices and practices.
Ricky is responsible for strategic planning on
research, drives Research and Development
activities in the East Asia Region and establishes
links to external partners. In previous roles,
he obtained and conducted over 20 collaborative
R&D and technology dissemination projects
under Government funding. He has also
extensive machinery design experience and
won several awards.
Dr Mikkel Kragh | Europe Region
Dr Gavin Davies | UK-MEA Region
Mikkel leads the Technology subsector in
the Consulting Practice in Arup’s Milan Office.
He is the Research Champion for the Europe
Region and he was recently appointed regional
Building Physics Skills Network leader for
Europe. Mikkel is Senior Visiting Research
Fellow at the University of Bath where he is
involved in façade engineering and building
physics research. He actively promotes
integrated design as the Chairman of the
Society of Façade Engineering.
Gavin is the Design and Technical leader
and also UK-MEA Research Champion. His
background is as an applied mathematician and
engineer. Gavin now leads Arup’s Environmental
Physics team in London with particular interest
in building physics, microclimate design, fluid
dynamics consultancy, climate change
adaptation and commercial research and
development.
8
Americas Region
Australasia Region
East Asia Region
Europe Region
UK-MEA Region
Arup Partnerships
Boston
Chicago
Houston
Los Angeles
New Jersey
New York
San Francisco
Seattle
Toronto
Adelaide
Auckland
Brisbane
Cairns
Melbourne
Perth
Singapore
Sydney
Bangkok
Beijing
Guangzhou
Ho Chi Minh City
Hong Kong
Hyderabad
Macau
Manila
Mumbai
Seoul
Shanghai
Shenzhen
Tianjin
Tokyo
Wuhan
Amsterdam
Ankara
Belgrade
Berlin
Bucharest
Düsseldorf
Frankfurt
Istanbul
Kraków
Madrid
Milan
Moscow
Rome
St Petersburg
Warsaw
Wroclaw
Abu Dhabi
Belfast
Bristol
Cape Town
Cardiff
Doha
Dubai
Dundee
Durban
Edinburgh
Gaborone
Glasgow
Johannesburg
Leeds
Liverpool
London
Manchester
Newcastle
Nottingham
Port Louis
Sheffield
Solihull
Southampton
Tshwane
Winchester
Wrexham
Abuja
Brunei
Bulawayo
Cork
Dublin
Galway
Harare
Kota Kinabalu
Kuala Lumpur
Lagos
Limerick
Penang
*
The Arup Partnerships is a partnership comprising a number of independent
yet inter-related practices, of which Arup Group Ltd is the largest. The Arup
Partnerships controls the use of the Arup name, co-ordinates the activities
of all the Arup practices, and fosters collaborative working.
*
9
Climate change extremes:
the combined effects of storm tides
and catchment flooding in Cairns
Authors: Sam Koci, Tania Cobham, Ragini Prasad
10
Tropical cyclones are
responsible for significant
coastal flooding due to the
high intensity rainfalls and
storm tides. This problem
increases in small urbanised
coastal catchments and is
predicted to rise with the
effect of climate change.
Elevation (m AHD)
> 10
4 – 10
2–4
1–2
<1
Fig 1. Digital elevation model of Cairns CBD and environs
The influence of storm tidal ocean conditions
on flooding in a coastal catchment within
the environs of the Cairns Central Business
District (CBD) has been investigated for this
study. Hydrodynamic flood modelling of both
freshwater discharge and storm tidal processes
in the catchment was conducted using the
one and two dimensional hydrodynamic
modelling engine, TUFLOW.
The study found both the magnitude of storm
tidal ocean conditions and the relative phasing
between freshwater discharge and storm tidal
conditions, to significantly influence flood
behaviour in the catchment. The results have
significant implications for flood prediction and
the implementation of flood warning systems
in low-lying coastal areas. A series of flood
maps and comparative flood extent maps
have been developed to illustrate the flood
behaviour and the relative influence of storm
tidal ocean conditions.
In order to assess the risks associated
with the cooccurrence of these events, an
understanding of the likelihood of such
cooccurrence is required. Due to the partially
dependent relationship between cyclone
induced storm surges and freshwater flooding,
estimating the likelihood of their cooccurrence
requires a complex joint probability assessment.
To overcome this problem, most flood analyses
use a single, generally conservative tidal signal
to reflect tailwater conditions for the modelling
of any given design event. An assumption
commonly made for coastal catchments is that
peak storm tidal levels coincide with the onset
of peak rainfall over the catchment.
The Cairns area consists of a number of
catchments that include natural and modified
creeks, channels and piped stormwater
drainage systems. The region is bound to the
west by steeply rising mountain ranges leaving
a narrow coastal plain of, on average, less
than 10km width.
4 0000
Tropical cyclones can produce coastal flooding
by generating extreme rainfalls and associated
freshwater runoff, as well as the elevated
coastal water levels generated by storm tides.
In cases in which extreme storm tidal conditions
coincide with immense freshwater discharge,
the extent of resulting inundation can vastly
exceed that of either of the independent
processes. This problem is particularly significant
in small, urbanised catchments, where high
intensity rainfalls can produce peak discharges,
surcharge and flooding in creeks and stormwater
drainage systems, in relatively short time periods.
Elevation (m AHD)
1 5000
2 0000
3 5000
Introduction
Coastal flooding is becoming a key issue for
coastal cities, particularly those along the far
north Queensland coastline. The Queensland
Government’s prediction for climate change in
the region states an increased risk from extreme
events due to increased storm tide events.
These include sea level rise, increased cyclone
intensity and frequency, and a shift southwards
in cyclone tracks. The severe coastal flooding
often caused by tropical cyclones, presents
a significant risk hazard for many coastal
communities.
0 5000
Abstract
Due to the high intensity rainfalls and storm
tides they can produce, tropical cyclones are
often responsible for significant coastal flooding.
Particularly in small, urbanised coastal
catchments, surcharge and flooding in creeks
and storm water drainage systems can be
significantly exacerbated by the elevated oceanic
tailwater conditions associated with coinciding
storm tides. With climate change forecasting
more frequent cyclones and rising ocean water
levels, it is widely expected that the problems
associated with cyclone driven coastal
flooding will increase.
Fig 2. Aerial photography and digital elevation model
of study area
Many parts of the coastal plain in the
Cairns region are relatively low-lying and flat,
with the majority of the CBD and surrounding
areas below 2-4m relative to the Australian
Height Datum (AHD) see Fig 1. During tropical
cyclones, the various drainage paths are
subject to surcharge and flooding from
freshwater discharge, as well as significant
influence from elevated storm tidal conditions.
The flooding caused by urban drainage
surcharge is a major problem in Cairns City,
posing significant risk to human life as well as
an estimated 17,000 properties.
Given the extreme future weather and ocean
conditions predicted as a result of climate
change, there is a clear need to further develop
our understanding of the combined effects of
storm tides, fluvial flooding and the implications
for coastal communities. A better understanding
of the interaction between these processes will
ensure that flood warning systems are set up to
better respond to predicted meteorological
conditions, and more accurate flood prediction
models are used in decision making and
infrastructure design.
11
Increased flood extent due to
increased tailwater conditions
10yr ARI rainfall event
20yr ARI rainfall event
50yr ARI rainfall event
100yr ARI rainfall event
Tailwater boundary
conditions
Static mean sea level
(MSL) signal, at 0m AHD
Inflow discharge (m3 S1)
Inflow boundary
conditions
12
Standard astronomical tide
for the region with a range
of ±0.6m AHD
MHWS tide superimposed
with a storm surge,
producing a peak storm
tidal level of 1.78m AHD
(50yr ARI storm tide)
MHWS tide superimposed
with a storm surge,
producing a peak storm
tidal level of 2.4m AHD
(100yr ARI storm tide)
Table 1. Hydraulic boundary conditions
8
6
4
2
0
3
Tailwater ocean level (m AHD)
Mean high water spring
(MHWS) tide with peak
level of 0.93m AHD
10
2.5
2
1.5
1
0.5
0
0.5
-1
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Fig 3. Relative phasing between inflow tailwater boundary conditions
The primary objectives of this study were to:
• characterise the flood behaviour in the
coastal catchment
• analyse the influence of tidal and storm tidal
ocean conditions on the flood behaviour
• analyse the sensitivity of the flood behaviour
to the relative phasing between freshwater
discharge and tidal and storm tidal processes
Methodology
For this study a hydrodynamic flood model
of a coastal catchment within the Cairns
CBD environs was developed, calibrated and
simulated under a range of hydraulic conditions.
This was used to investigate the influence
of tidal and storm tidal processes on the
flood behaviour.
The key objective of this study was to
understand the likely influence of extreme
tidal conditions on the flood behaviour. It was
therefore decided to apply the proposed
methodology to one suitable catchment within
the Cairns CBD rather than the entire CBD.
The Fearnley Drain catchment was chosen as
being suitable for the study, because of its poor
height relief relative to the ocean and the fact
that there is little hydrologic interaction between
it and the surrounding catchments.
The catchment has an area of approximately
410ha, most of which sits below around 3m
AHD. It includes a mixture of residential and
industrial land use. The majority of stormwater
drainage from the catchment is provided by a
single branch man-made channel, which likely
replaced one or more natural tidal channels
when the land was initially reclaimed. Flap gates
have been installed at one point in the channel
to prevent saltwater intrusion.
12
Fig 4a. Flooding in catchment during cyclone in
January 2009
Fig 4b. Storm tidal conditions during cyclone in
January 2009
A dynamically linked 1D/2D TUFLOW model was
developed for the Fearnley Drain catchment in
order to accurately represent both the narrow
Fearnley Drain channel system, and the broader
floodplain. The terrain characteristics of the
floodplain were represented in the 2D domain,
developed from a high resolution digital elevation
model and aerial photography, Fig 2. The 1D
domain was developed to model the Fearnley
Drain channel system which, due to its narrow
width and the existence of various hydraulic
structures, could not be accurately represented
in the 2D domain. In the absence of historic
flood data for the catchment, the model was
calibrated against the results of a previous
flood study of the region.
Hydraulic boundary conditions
A series of simulations were conducted in
which these hydraulic boundary conditions
were altered to vary the magnitude of the
freshwater inflow and tidal tailwater conditions,
and the relative phasing between the two
processes. Each combination of inflow and
tailwater conditions were modelled with various
different relative phasings, and the results of
the simulations were analysed. Fig 3. shows
the set of relative phasings simulated for the
100yr ARI freshwater discharge, and 2.4m
storm tide event. A similar set were simulated
for all other combinations of inflow and
tailwater boundary conditions.
A range of hydraulic boundary conditions
were applied to the model to simulate both
freshwater inflows and tidal tailwater conditions in
the catchment. The boundary conditions used
were derived from the WBM (2001) study and are
described in Table 1. It is noted that the 1.78m
and 2.4m AHD storm tidal signals developed in
WBM (2001) were based on research by James
Cook University and the Beach Protection
Authority. These tidal signals correspond to
predicted storm tide levels of 50yr and 100yr
average recurrence intervals (ARI) respectively, with
consideration of sea level rise and other predicted
climate change conditions including increased
cyclone frequency and intensity.
Results and discussion
Consistent with historic records and recent
observation of flood events see Fig 4a. and
Fig 4b. the simulated model results, see Fig 4c.
found the low-lying Fearnley Drain catchment to
be significantly affected by flooding due to both
freshwater discharge and tidal processes.
Particularly in the downstream, tidal regions of
the catchment, elevated ocean water levels
caused by extreme tidal conditions were found
to have a major influence on the magnitude and
extent of flooding. The flood maps produced
from the study illustrate this flood behaviour.
Tidal and storm tidal ocean conditions were
found to have a relatively significant influence in
Inundation area (ba)
180
20yr ARI
50yr ARI
100yr ARI
160
140
120
100
80
60
40
100yr ARI (2.4m
AHD) storm tide
0%
0%
17%
117%
20yr ARI
0%
0%
6%
65%
50yr ARI
0%
0%
4%
39%
100yr ARI
0%
0%
4%
36%
Influence of tailwater conditions on peak flood levels in
catchment - 100yr ARI freshwater discharge event
3
2.5
2
1.5
1
Mean sea level
Standard tide
MHWS tide
50Yyr ARI storm tide (1.78m)
100yr ARI storm tide (2.4m)
0.5
0
20
0
50yr ARI (1.78m
AHD) storm tide
10yr ARI
Peak flood level in channel (m AHD)
220
Base flow
10yr ARI
MHWS tide
Table 2. Increased flood extent due to increased tailwater conditions
Influence of tailwater conditions on
extent of unundation
200
Standard tide
Sensitivity to relative phasing
- 10yr ARI/ 2.4m Storm tide
2.450
Peak flood level (m AHD)
Fig 4c. Simulated flood results in catchment
% increase in flood extent from MSL tailwater conditions
2.400
2.350
2.300
2.250
2.200
2.150
Point 1
Point 2
Point 3
2.100
Point 4
Point 5
Point 6
2.050
Static
MSL
Standard
MHWS
1.78m
tide
tide
Storm tide
Tailwater condition
2.4m Storm
tide
0
1000
2000
3000
4000
Upstream distance from coast (m)
1
2
3
4
5
6
7
8
9
10
Time at which peak tidal level phased
(hours from start of storm)
11
Fig 5. Influence of tailwater conditions on extent
of inundation
Fig 6. Peak flood levels produced by 10yr ARI fluvial
flood event under various tailwater conditions
Fig 7. Sensitivity to relative phasing between 10yr
ARI freshwater discharge and 2.4m storm tide
the tidal regions of the catchment and limited
influence in the non-tidal regions. The significant
change in flood behaviour beyond the flap gate
(located approximately 1.4km along the Fearnley
Drain channel from the coast) is illustrated in
Fig 6. as the peak flood levels produced by the
different tailwater conditions remain relatively
constant beyond this distance.
representative points in the catchment,
produced by variation in the relative phasing
between the 10yr ARI freshwater discharge
and 2.4m storm tide, are illustrated in Fig 7.
The figure shows a variation in peak flood levels
of up to 0.2m at one point in the catchment.
As demonstrated by the results of this study,
the factors underlying these assumptions can
have a significant influence on the flood
behaviour. When adopting these assumptions
it is therefore essential that the sensitivity of
the flood behaviour to these factors is
appropriately assessed.
Throughout the tidal regions large variations in
peak flood levels and the extent of inundation
were produced by different tidal and storm tidal
conditions, for all freshwater discharge events.
For the 100yr ARI freshwater discharge event,
peak flood levels in the tidal region produced by
the 100yr ARI (2.4m AHD) storm tide were more
than 0.5m greater than those produced under
normal tidal conditions, see Fig 5. Fig 6. and
Table 2. This shows the increase in flood extent
due to changed tidal conditions.
This demonstrates that tidal conditions up to a
certain level (MHWS) have minimal impact on
flood extents. These results also demonstrate
that there is a threshold limit above which flood
extents significantly increase. Table 2. shows
that this occurs between MHWS and the 50yr
ARI (1.78m AHD) storm tide. This is important
when determining flood warning thresholds in
the development of flood warning systems.
The sensitivity of the flood behaviour to the
relative phasing between hydraulic processes,
was found to depend mainly on the magnitude
of the tidal conditions. The relative phasing with
the smaller tidal events (standard and MHWS
tides) had very little influence on the peak flood
levels recorded at any point in the catchment,
whereas the relative phasing with the larger
storm tide events had a relatively significant
influence on peak flood levels, particularly in the
downstream regions of the catchment. As an
example, the variation in peak flood levels at six
In general, the influence of relative phasing
between the freshwater discharge and tidal
processes was found to diminish with further
distance from the coast as well as increased
magnitude of freshwater discharge and in cases
in which tidal processes completely dominate
the flood behaviour.
Conclusion and next steps
Using the Fearnley drain catchment in Cairns
as a case study, this investigated the influence
of storm tidal ocean conditions on the flood
behaviour in coastal catchments. The study
found both the magnitude of storm tides, and
the relative phasing at which they coincide with
freshwater flooding, to have quite a significant
influence on the flood behaviour in the
catchment. Although the extent of this influence
will depend on a large number of site-specific
conditions, it is likely that this result would also
be observed in other similar coastal catchments.
In coastal areas like Cairns, where tropical
cyclones frequently produce immense rainfall
and storm tide events, an understanding of the
interaction between tidal and catchment driven
flooding is critical in predicting the flood
behaviour and implementing flood warning
systems and mitigation strategies. In many
flood studies of coastal catchments,
accurate estimation of the joint probability of
cooccurrence of significant rainfall and storm
tide events is not feasible and a number of
assumptions are made in adopting the
hydraulic conditions for given design events.
12
Acknowledgements
This project was conducted at James Cook
University, Townsville, under the supervision
of Professor John Patterson.
We would also like to acknowledge the
contributions to the project made by the
following: Bill Syme, BMT WBM for providing
the TUFLOW license, as well as substantial
project advice and assistance; Tim Smith, Nola
Strawbridge, Rudd Rankine Cairns Regional
Council for providing a wide range of data; Tony
Martin Queensland Department of Main Roads
for providing the digital elevation model data and
aerial photography used in the project.
References
Baddiley P., The flood risk in Cairns, Natural
Hazards 30, Kluwer Academic Publishers,
2003, 155-164.
McInnes K. L., HubbertG.D., Abbs D. J. & Oliver
S. E. A numerical modelling study of coastal
flooding-Meteorology and Atmospheric Physics,
Meteorol. Atmos. Phys. 80, Springer-Verlag,
2002, 217-233.
WBM 2001 Cairns CBD and Environs Drainage
Management Plan, Cairns Regional Council.
Queensland Government 2009, Climate change
in the Far North Queensland Region.
13
Climate change extremes: the combined effects of storm tides and catchment flooding in Cairns
Increased flood extent due to increased tailwater conditions
Freshwater
discharge event
Geomechanical characterisation of
Manhattan schist, a study of the
7 line extension cavern
Author: Seth Pollak
14
New York
intro
text City struggles to cope
with growing demands placed
on an aging metro system. The
7 line extension, the first major
system upgrade in decades, is
currently under construction and
will be the first to be completed
when it opens to the west side
of Manhattan in 2013.
Fig 1. Station cavern top heading excavated to full span at contact zone revealing granitic rock to the left and
Manhattan schist on the right
Abstract
Arup was engaged by S3 II Tunnel constructors
to provide initial support design for two
Tunnel Boring Machine (TBM) starter chambers,
three shafts, five cross passages, and the
22m wide x 300m long main station cavern.
Design challenges included low cover and
proximity to sensitive structures, among others.
On site, Arup carried out verification of the
design through geologic mapping, rock mass
classification, detailed joint characterisation,
and monitoring of instrumentation. As this is
the first shallow cavern to be constructed in
Manhattan in 40 years, there was little precedent
for how the rock mass would respond to the
construction sequence and the adopted
support scheme of rock bolts and shotcrete.
A study was undertaken to verify assumptions
made during design regarding ground
relaxation, impacts of junction construction
and joint properties. A site specific rock mass
classification correlation has been derived and a
statistical database of joint properties compiled.
A methodology for junction design in Manhattan
Schist is proposed, based on measured ground
movements. The observations and measurements
made during construction of this study will
benefit the design of five similar caverns slated
for future construction in Manhattan.
Introduction
The 7 line extension project is a 2.4km long
two-track subway expansion extending service
from the existing Times Square Station at
41st Street and 7th Avenue out to the west side
of Manhattan and terminating at a new station
to be located at 34th Street and 11th Avenue.
The centerpiece of the study and main focus
of this research is the 300m long-mined station
cavern at 34th Street, shown in plan on Fig 1.
The station consists of a two level public area
(200m long) with interlocking caverns on
either end which facilitate track crossovers.
Challenges associated with construction of
the cavern include:
• urban setting
• less than 1 span rock cover (type 14m
for 21m span)
• close proximity to active rail lines and historic
buildings (minimum 8m)
• lack of precedent with regards to cavern
construction experience in NYC workforce
These factors, coupled with adverse geology,
made excavation of this station anything but
routine. Further complicating the construction
was the need to form six perpendicular junctions
which would be driven and left as stub tunnels
for future entrance connections or utility adits.
The cavern top heading cross section is shown in
Fig 1. The full section has an excavated span of
21m and height of 16.5m. The adopted
construction sequence included a staggered,
multiple drift top heading (3 drifts x 50m 2 each)
excavated through the full length of the main
cavern, followed by benching and interlock cavern
excavation, also using the multiple drift approach.
Methodology
Geologic setting
Two different rock types are present along the
cavern alignment. A central intrusion of granitoid
rock, described as euhedral, crystalline, acidic,
and mica deficient is present over roughly
50% of the excavation. There exists a central
depression within this rock mass which is
infilled with a “mica schist” often refered to
as “Manhattan schist” and minor pegmatite.
The contact between the granitic rock and
the schist is generally intact to moderately
weathered. The southern limb of the intrusion is
located at the cavern/south interlock transition,
where rock grades back into mica schist.
The northern limb is characterised by a
faulted contact between the granitic rock
and mica schist. This contact was found to
be approximately 1m thick and contain
decomposed rock and breccia in a matrix of
green, low plasticity clay. Following the contact,
the schist is characterised as faulted and
Fig 2. Faulted contact zone between granitic rock
(bottom) and schist
sheared with sub-vertical to vertical foliation
fractures and seams. The fault system is a
series of sub-parallel en echelon features
striking obliquely across the excavation trend.
These structures are discrete and bounded by
higher quality schist, similar to that found in the
southern end of the cavern. The total length of
cavern that was excavated through this poor
quality zone was 100m.
In terms of rock mass behaviour, two distinct forms
of Manhattan schist have been observed during
excavation. The first type is what could be termed a
“classic” schist: lack of blockiness, a dominant
fabric orientation where foliation is very close and
discrete foliation joints are difficult to discern. This
fabric is present in rock of the TBM starter tunnels,
located at the south end of the study.
Conversely, the behaviour of the schist in
the cavern is dominated by intersecting
discontinuities, containing both discernable
foliation joints and wider spaced sub-vertical
cross foliation joints. This leads to what is termed
“blocky” ground, one in which the Geological
Strength Index (GSI) has been applied to derive
geotechnical design parameters. The intact rock
strength is also higher, resulting in schist whose
foliation fabric does not necessarily control
rock mass behaviour.
Jointing in the granitic rock is generally orthogonal
with the occasional sub-vertical joint cross
cutting through the mass. Horizontal joints are
typically open and clay filled (up to 15mm), and
moderately to highly continuous with measured
15
RMR (89)
90
Rock mass classification correlation
Granite
70
Log. (Schist)
y = 6.5808Ln(x) + 42.698
R 2 = 0.3426
60
Schist
40
30
20
1
Q (Barton, 1974)
10
Results and discussion
Rock mass characterisation
During the design stage, the engineer is forced
to make many assumptions regarding the
conditions of a rock mass, sometimes only
with 50mm diameter rock core and lab test data.
This is especially true in urban construction
where rock outcrops are rare and those that
do exist have been physically weathered
and/or chemically altered for decades.
If rock mass classification is being done from
rock core, the parameters contained within the
Rock Tunnelling Quality Index (Q system) are
typically easier to determine than those
associated with the Rock Mass Rating system
(RMR). Both of which are types of empirical
classification systems. Several correlations exist
between the two systems, but the most accurate
correlation will be one derived from site specific
data. During geological mapping of the 7 Line
study, over 250 such classifications have been
carried out by a limited number of engineers and
geologists which minimises variability. Fig 3.
shows the following site specific correlations
have been derived for granitic rock and
Manhattan schist.
A case can be made that these correlations are
valid based on the fact that the trend line for
all the data is approximately equal to the well
known correlation given by Bieniawski (1989)
which is commonly applied in rock mass
RMR = 91n Q + 44
(1)
classification:
It is also a difficult, if not impossible task to
accurately characterise large scale joint
properties from borings. If outcrops are
unavailable for scanline observation, an
educated guess must be made for these
parameters, which in most cases leads to a
conservative design. However, properties that
can only be quantified by mapping, such as
joint persistence, large scale waviness, and
roughness, are all key input parameters into
discontinuum modelling and key block analyses.
For example, there will be a drastic difference
in modelling results (support requirements)
between a model with planar, continuous joints
versus one with low persistence joints, which
can only be mobilised by shearing through the
intact rock. Likewise, if the amplitude of large
scale waviness of the joint is great, dilation is
inhibited and the joint can only mobilise if
the intact rock is sheared through.
*
100
21m
16
All the properties measured above can be used
to improve the joint shear strength models used
in discontinuum numerical models.
Ground relaxation and interaction of a
multiple drift excavation sequence
The concept of stress redistribution, along with
subsequent deformations caused by formation
of a plastic zone, ahead of an advancing tunnel
face is well documented since being introduced
by Panet and Guenot (1982). Since then,
numerous papers have been authored on the
empirical and analytical shape of the longitudinal
displacement profile for an advancing circular
tunnel at a constant rate, ie TBM tunnelling.
These profiles are then applied in 2D numerical
models to account for the 3D face effects
through methods such as convergenceconfinement. In this method, stress tractions of
equal and opposite magnitude to the radial initial
stress (o) are applied along the tunnel periphery
and then systematically reduced by a factor, λ
whose value is between 0 and 1 prior to the
support installation stage:
(2)
where σ r is the ‘relaxed’ radial stress.
The “relaxation factor”, λ, is defined by the ratio
of the amount of radial deformation that takes
place in the ground prior to arrival of the face
(μr(o)) to the total amount of radial deformation
that takes place at infinity behind the face (μ r (∞)).
λ = μ r (o) / μ r (∞)
Borehole
extensionmeter
Fig 4. Plan of cavern: top heading construction sequence
and instrumentation arrangement
Therefore, improved quantification of joint
parameters during the design stage can lead to
an optimised, cost effective support scheme
through validated numerical models.
In order to verify the design, joints were
characterised whenever possible by observation
and profile gauge. Large scale waviness was
calculated by taking the ratio of maximum
amplitude to wavelength over the visible trace
length. The Joint Roughness Coefficient (JRC),
as determined from profile gauge measurement,
was used to estimate Joint Roughness (JR) from
tables prepared by Barton (1987). Over 120
joints comprise the database. Table 2. gives
statistics for the primary schist joint set.
σr = (1- λ) σo
*
9m
Fig 3. Site specific rock mass classification correlation for 7 line study
trace lengths of 10-15m. These joints also
produce the majority of water inflow into the
cavern with flows of up to 10 l/min observed.
*
D = 7m
9m
Log. (Granite)
y = 9.4359Ln(x) + 48.136
R 2 = 0.4958
Log. (All)
y = 8.6272Ln(x) + 44.258
R 2 = 0.4013
50
10
0.1
*
Plan of a cavern
80
(3)
For TBM tunnelling, which takes place at a more
or less constant rate, a smooth longitudinal
deformation curve is assumed as the stress
redistribution reaches a quasi steady state
during excavation. This is a valid assumption for
continuous excavation, but not for cyclic drill
and blast tunnelling. This aspect was explored
by observing ground relaxation in the three top
heading drifts.
Drift
No. of
Ave. first
measurements response
[D = 7 m]
Ave.
relaxation
Ratio [λ ± σ]
1
5
1.2D
0.70 ± 0.28
2
2
1.0D
0.71 ± 0.17
3
4
0.7D
0.66 ± 0.07
Table 1. Summary of drift relaxation in cavern top heading
The method of excavation is directly responsible
for the stress path that the ground is subjected
to. The stress path in turn will determine the
extent of the plastic zone around the face and
hence the deformation. For the 7 Line study, the
situation is further complicated by having a
staggered, multiple drift construction sequence
in which the plastic yield zone ahead of and
around the drifts will interact with each other
and cause further deformations.
Multipoint extensometer arrays installed from the
surface in advance of construction were monitored
by a real time data acquisition system so the
movements could be correlated with the blasting
cycle. The general layout is shown in Fig 4.
By monitoring the response of multipoint
borehole extensometers in front of the three
advancing drift faces, a comparison can be
made to the typical assumption of λ = 0.3, which
is commonly used for TBM tunnel modelling.
For the central heading, the relaxation factor
was calculated by taking the ratio of movement
occurring in the instrument up to the point of
face arrival to the subsequent movement
recorded up to the point when Drift 2 arrives at
the instrument. This value is not quite “pure”
relaxation in the sense that bolting is typically
completed 1D behind the face, resulting in
slightly higher λ values (ie less movement
measured due to support installation).
Nevertheless, the values obtained give a good
indication of the range and magnitude of
relaxation that could be applied in a staged
numerical model. Similar observations were
made for Drifts 2 and 3. A summary of the
findings is presented in Table 1. The average
distance of the instrument from tunnel face at
first response was also observed for each drift
to give an indication of the extent of the plastic
zone ahead of the face.
The general trend of the data suggests that the
proportion of movement that occurs ahead of
the face is over half of the final value in each
of the drifts, ie greater than 50% relaxation.
This is because the sudden strain release
<1
1-3
3-6
6-9
>9
0
27
59
14
0
% per total surveyed
2-4
4-6
6-8
8-10
10-12
%
6
28
25
22
14
6
Jr
0.5
1
1.5
2
3
4
JRC
%
Nature of infill
%
Large scale waviness
%
>12
0
3
75
11
11
0
0-2
2-4
4-6
6-8
8-10
>10
3
3
%
i°
17
31
28
19
None
surface stained
non-cohesive
clay
96
0
4
0
None
0.01
0.01-0.02
>0.02
42
0
0
58
Run 2D numerical model of
main cavern with standard
support and record roof
deformation
Reduce stiffness of intact
rock or rock mass
(depending on model type)
until 15% increase in roof
deformation is produced
Check capacity of
standard support
Table 2. Summary of Manhattan schist foliation joint characteristics
Adit
Span ratio
[DAdit/
DCavern]
Rock mass classification
[Q/RMR89]
Cavern
Adit
E1
0.6
1.3 / 47
1.3 / 44
E2
0.6
1.0 / 39
T3
0.6
1.0 / 43
Δδm/δmo
[%]
Δδa/δao
[%]
Fail
N/A
172
2.7 / 49
8
280
0.7 / 46
13
93*
Table 3. Rock mass behaviour in junction areas of 7 line main cavern. | m (o): main cavern crown extensometer
movement prior to adit | a (o): main cavern adit side extensometer movement prior to adit | *Spiles used over adit
caused by blasting produces more plastic
damage in the rock mass than gradual strain
relief, as is the case when using mechanical
excavation methods. Applied in a model, this
would lead to larger deformations but smaller
load on the support. The difference in increasing
the bolt pattern spacing by even 0.3m could
have significant cost savings over the length of
a 300m cavern.
Rock mass behaviour at junction locations
Junction design is commonly carried out in
several ways. The first is an empirical approach
using the Q system (namely Jn x 3 where
Jn = joint set number). The increase in joint
number is to account for the addition of a third
dimension, formed by the intersection, along
which the potential for kinematic wedge failure
is increased. The second way is to utilise a
structural beam-spring model to design the
thickness of the shotcrete.
This requires an estimate of rock load on
the lining and does not account for any
rock-structure interaction (ie no arching effects).
Shotcrete capacity is usually designed to keep
combinations of moment and thrust within the
elastic envelope, neglecting the post cracking
benefits of steel fibers. Both methods are
typically conservative. Complex 3D models
can be useful, but are time consuming and
sometimes difficult to interpret. Another question
is how far to extend the additional reinforcement
around either side of the penetration. One adit
diameter is a typical rule of thumb value used
during design.
Construction of junctions gave the opportunity
to study how the rock mass behaved by
observing ground movements recorded by the
extensometers. In particular, three penetrations
formed perpendicular to the main cavern were
studied. Direct comparison between the
junctions was possible as they were all of the
same size and rock mass classification of the
cavern roof revealed nearly identical rock mass
quality. Note that the typical magnitude of
cavern crown deformation recorded was
10-15mm. Table 3. summarises the results.
The effects of spiling above the T3 junction
breakout are evident in reducing the deformation
by approximately 50% compared to the other
two junctions.
The adits were blasted only after the main
cavern top heading had been fully excavated
and supported. Roof movement in the cavern of
less than 15% additional strain (compared to
cavern movement prior to junction excavation)
was observed due to E2 and T3 junction
construction, both of which were in Manhattan
schist. The E1 extensometer was damaged
during blasting, so no reading was possible.
The extent of the plastic zone around the
junctions was smaller than assumed during
design. Extensometers located 4.5m offset from
the edge of the adit showed no response during
excavation. This corresponds to a zone of
influence of less than 0.5D adit either side of the
penetration. In addition, extensometers located
on the far side of the cavern (opposite the adit)
showed no response to adit construction.
Install heavier support
and re-run model
OK
Increase thickness
of shotcrete 0.5D
around adit for potential
wedge failure and adopt
Fig 5. Methodology proposed
Ground behaviour in the vicinity of junction
penetrations was also studied which showed
that an additional strain of less than 15% in the
cavern crown was recorded following junction
excavations. The extent of influence into the rock
mass due to the junction was limited to 0.5Dadit
around the penetration and to cavern centerline.
Based on the data, a new methodology for
designing perpendicular junctions in Manhattan
schist has been proposed.
With several caverns in the planning and design
stages in similar geological conditions, the
information gained during construction should
go a long way in helping Arup develop support
designs which are cost effective and tailored to
the unique combination of rock mass and in
situ stress conditions found beneath the streets
of New York.
Acknowledgements
Based on the results presented above, a new
design methodology is proposed for
perpendicular junctions formed in Manhattan
caverns is shown in Fig 5.
This method is relatively quick and produces a
support design that is based on site specific
data. The increase in shotcrete thickness can be
restricted to local containment of wedges
around the penetration without needing to be
extended across the entire cavern span, keeping
the design cost effective.
Conclusion and next steps
The successful construction of the 7 Line study
has provided a unique opportunity to study
critical assumptions and carry out rock mass
behaviour observations for the purpose of
refining shallow cavern design methodology in
Manhattan schist. Joint characterisation has
quantified difficult to define properties such as
large scale waviness, persistence, and
roughness for each rock and joint type.
The effects of multiple drift excavation on
ground relaxation were studied via multipoint
borehole extensometers which revealed that in
all cases, more than half of the total recorded
strain occurred ahead of the drift faces.
This is an amended version of the paper
‘Geomechanical Characterisation of Manhattan
Schist – A Study of the 7 Line Extension Cavern’
presented at the 2010 International Tunnelling
Association World Congress, Vancouver. Thanks
to Dr. Chris Snee of SneeGeoconsult, Brian
Balukonis of GZA Geoenvironmental, Inc., S3 II
Tunnel Constructors and New York City Transit
Metropolitan Transportation Authority
Capital Construction.
References
Barton, N., Lien, R., Lunde, J., Engineering
classification of rock masses for the design
of tunnel support, Springer-Verlag, Rock
Mechanics 6, 1974, 189-236.
Barton, N., Predicting the behaviour of
underground openings in rock. Maunel Rocha
Memorial Lecture, Lisbon. Oslo: Norwegian
Geotechnical Institute. 1987.
Bieniawski, Z.T., Engineering Rock Mass
Classification, John Wiley and Sons,1989.
Panet, M., Guenot, A., Analysis of convergence
behind the face of a tunnel. In: Tunnelling ’82.
IMM, London, 1982.
17
Geomechanical chatacterisation of Manhattan schist; a study of the 7 line extension cavern
Persistence of joints (m)
An investigation of fire load
density for residential buildings
and offices in Hong Kong
Authors: Mingchun Luo, M Liu, SM Lo and KK Yuen
18
The average fire load in
Hong Kong is higher than in
Europe. The implication is that
local information is important
for determining the fire load
density and a reliable fire
safety engineering study
should be supported by
local data.
Abstract
Performance-based fire safety engineering
design is now widely adopted. The potential
and strength of a performance-based fire
safety engineering approach becomes an
important tool for specialist consultants.
However, one major deficiency is that it requires
appropriate design fires to determine the
fire scenarios as well as the probability of fire
occurrence. A design fire can be characterised
by the heat release rate which may be inferred
by the fire load density of the space concerned.
Accordingly, a comprehensive study on the
fire load density is required to be carried out
in order to support the development of
performance-based fire safety engineering
design. Planned field surveys will be
conducted in selected typical buildings.
As the physical dimensions and available
combustible materials can be roughly estimated,
the approximate weight of those combustible
materials can also be estimated and, by
assuming the type of materials, a rough range
of the fire load density can then be established.
This study presents the results of the surveys
for residential buildings and also presents
the summary of the survey results for offices
in Hong Kong.
Data collected
For information only
Factors affecting fire loads in residential buildings
1 Series number
2 Address
Dependent variable in
fire load calculation
Other information
Variables in electrical
equipment
1 Residents’ gender
1 Types of equipment
3 Number of elderly
2 Number of equipment
4 Number of children
3 Volume occupied
5 Number of students
4 Weight
6 Number of partition walls
Shown in part I
GENERAL INFORMATION
2 Number of residents
Variables in Furniture
1 Types of furniture
2 Number of furniture
3 Volume occupied
4 Weight
5 Material used
Variables in
living rooms/bedrooms
Variables in
bathrooms
Variables in
kitchens
Shown in part II
Shown in part II
Shown in part II
SURVEY OF
COMBUSTIBLE
MATERIALS – LIVING
ROOMS/BEDROOMS
SURVEY OF
COMBUSTIBLE
MATERIALS –
BATHROOMS
SURVEY OF
COMBUSTIBLE
MATERIALS – KITCHENS
Fig 1. Structure for the survey form
Introduction
In Hong Kong and many large cities in China,
rapid urbanisation has caused the construction of
high-rise buildings. In view of the rapid growth of
population and commercial activities in the urban
areas, the construction of high-rise buildings will
continue. Regarding fire safety protection design
of such complex high-rise building, the codecompliant or prescriptive approach may not be
the only way to assure the fire safety level.
Performance based fire safety engineering design
is now widely adopted. The potential and strength
of performance-based fire safety engineering
approach manifests itself to become an important
tool in specialist consultantancy.
However, one major deficiency is that it requires
the appropriate design fires to determine the
fire scenarios as well as the probability of fire
occurrence. A design fire can be characterised
by the heat release rate, which may be inferred
by the fire load density of the space concerned.
Accordingly, a comprehensive study on the fire
load density is required to be carried out to
support the development of performance-based
fire safety engineering design. Planned field
surveys will be conducted in selected typical
buildings. As the physical dimensions and
available combustible materials can be roughly
estimated, the approximate weight of those
combustible materials can also be estimated
and, by assuming the type of materials, a rough
range of fire load density can then be established.
This study presents some results of the surveys.
Methodology
In this project, the procedure for conducting
the study was as follows:
• specify the type of residential buildings to be
investigated
• specify the number of samples to be
carried out
• define the scope of combustible materials to
be counted eg papers, books, computer
equipment, furniture, partitions and
miscellaneous
• observe the physical dimensions and
characteristics of the various content items to
determine the weight of the counted items.
This can be done by estimating the
dimensions and other information from the
photographs taken on site. Weight of
materials other than wood were converted to
an equivalent weight of wood using a
multiplying factor based on the ratio of the
heats of combustion
• data collected by the survey were analysed
using statistical measures (eg t-distribution to
summarise the results for the fuel load
density in various buildings
19
A
Ra es's equation
te =
te
FLD
Af
AT
Av
h
C
0.067 Af FLD
(AT Av h) 0.5
(1)
= the equivalent duration of fire
D
= fire load density
= floor area
= total area of wall
= surface area of ventilation
= height of the opening
QL&B = ∑ 0.3VLC ρLC H LC + ∑ VLF ρLF HLF + ∑m LE HLE
(2)
FLDL&B =
(3)
∑0.3VLC ρLC H LC +VLF ρLF HLF + ∑m LEHLE
A L&B
FLDW =
(5)
∑0.3VWC ρWC HWC + ∑VWF ρWF H WF + ∑mWEHWE
AW
Kitch en
QK = ∑0.3VKC ρKC HKC + ∑VKF ρKF H KF + ∑m KE H KE
(6)
FLDW =
(7)
FLDW / FLDK
AW / AK
VWC /VKC
ρWC / ρKC
HWC / HKC
VWF /VKF
ρWF / ρKF
HWF / HKF
mWE / mKE
HWE / HKE
Where
HLF
m LE
HLE
(4)
QW / QK
Living room and bedroom
Q L&B
FLDL&B
AL&B
VLC
ρLC
HLC
VLF
ρLF
QW = ∑0.3VWC ρWC HWC + ∑VWF ρWF H WF + ∑m WE H WE
∑0.3VKC ρKC HKC + ∑VKF ρKF H KF + ∑mKE HKE
AK
Where
The average design fires can be computed by fire
load density and equivalent duration of fire
B
Wash room
= fire load of living room/bedroom, in MJ
= fire load density of living room/bedroom, in MJ/m2
= total area of living room/bedroom, in m2
= fire load of washroom/kitchen, in MJ
= fire load density of washroom/kitchen, in MJ/m2
= total area of washroom/kitchen, in m2
= volume of furniture which is a container, in m3
= corresponding density of furniture which is a container, in kg/m3
= corresponding calorific values of furniture which is a container, in MJ/kg
= volume of furniture in living room/bedroom, in MJ/m2
= corresponding density of furniture, in kg/m3
= corresponding calorific values of furniture, in MJ/kg
= mass of electrical equipment in living room/bedroom, in kg
= corresponding calorific values of electrical equipment, in MJ/kg
For the fire load density corresponding to a domestic flat, it is calculated as the ratio of the total
quantity of the fuel commodities to the floor area of the space. The equations are derived as follows:
= volume of furniture which is a container, in m3
= corresponding density of furniture which is a container, in kg/m3
= corresponding calorific values of furniture which is a container, in MJ/kg
= volume of furniture in living room/bedroom, in MJ/m
2
= corresponding density of furniture, in kg/m3
= corresponding calorific values of furniture, in MJ/kg
= mass of electrical equipment in living room/bedroom, in kg
= corresponding calorific values of electrical equipment, in MJ/kg
E
Total fire load d ensity
QT = QL&B + QW + QK
(8)
Where
QT
FLDT
FLDT =
QL&B + QW + QK
AL&B + AW + AK
(9)
= total fire load, in MJ;
= total fire load density, in MJ/m2;
Fig 2. List of equations for calculating fire load density
• with the estimated fire load density from survey,
the equivalent duration of a fire can then be
determined by Raes’s equation, see Fig 2. (1)
• the average design fires can then be
computed by fire load density and equivalent
duration of fire
Data collection
To determine the fire load density of a domestic
building, a simple approach proposed by Kumar
and Rao, in which the weight of an object is
related to its visual physical characteristics was
employed. The approach is better than the direct
weighing method as it is convenient, acceptable
to residents, and also time-saving.
The fire load survey was conducted between
October 2005 to March 2006 covering a total
floor area of about 2,787m², including 50 flats
in 13 domestic buildings with the height up to
thirty-six storeys. The flats and buildings were
selected randomly due to the availability of
residents.
In this study, fire load survey for six types of
offices were also conducted. The structure of
the survey form is constructed and shown in
Fig 1. The overall procedures for the collection
of data are summarised in Fig 3.
To improve the practicability of the survey, the
following assumptions have been made and
summarised as follows:
• the fire load in basements (if any) and
staircases are not included in the present
20
investigation as a basement is normally not
provided for residential buildings and
staircases are assumed to be a protected
zone in which combustibles are limited
• for paper and books, the mass is not
dependent on their various compactions
• the fire loads do not vary for items produced
on the basis of new technologies and new
materials, hence the data used for calculation
is consistent for similar types of materials
• for containers, it is assumed that only 30%
of total volume is made of the corresponding
material and 70% of area is space
• the fire load contributed by doors, windows,
and ventilators in between two rooms is
considered as a fire load for one room only
• the dimensions of commonly-used furniture
and electrical equipment are the same
• there is a uniform distribution of combustibles
in the building
• all combustibles will be involved in a fire
• the combustion will be complete and the
rate of heat production will be the same for
all combustibles
• it is assumed that the volume of clothing
occupied 50% of the containers
Determination of fire load density
For containers, such as boxes, drawers etc, it is
assumed that only 30% of total volume is made
of the corresponding material, while 70% of the
area is empty space. Based on this assumption,
two equations are derived for determining
the fire load of a living room and bedroom,
see Fig 2. (2) and (3). Similarly, the equations
for the washroom and kitchen can be seen in
Fig 2. (4)(5) and (6)(7).
Results and discussion
The survey results for domestic buildings are
presented in Fig 4. and Fig 5., and show that all
the residential flats in Hong Kong have a fire load
density higher than 1000 MJ/m2; for over 20%
of flats, the fire load density is over 1,500 MJ/m2.
The total fire load density can be calculated from
the equations seen in Fig 2. (8) and (9).
Table 1. summarises the results of a fire load
survey for six type of offices. The fire load
densities vary from 546 MJ/m 2 to 1,408 MJ/m 2.
The weighted average fire load density for these
six types of offices is over 800 MJ/m 2.
To compare the results with other countries,
the fire load density of a Europe Survey
has been extracted from British Standard,
PD7974-1:2003 and presented in Table 2.
The average fire load densities of dwellings
and offices are 780 MJ/m 2 and 420 MJ/m 2
respectively. These values are significantly
lower than the survey results in this study.
This is due to the fact that the average floor
area per person occupied in Hong Kong is
much less than that in the Western Countries
because of the limited land available. From
Raes’s Equation, Fig 3. (1), the equivalent
duration of fire (t e) is in direct proportion to
the floor area (A f) and the fire load density
(FLD). With a higher fire load density, the
equivalent duration of a fire will be longer.
Survey on current public housing
structure 50 nos. of domestic flats in
public person and properties
Office 1
Office 2
Office 3
Office 4
Office 5
Office 6
Area (m2)
330
230
280
820
750
580
Fire Load Density (kJ/m2)
784
1408
546
658
857
756
Table 1. Summary of fire load density of office buildings
Preliminary analysis from the collected datum
in previous stages
• Overall floor area, floor area of living room,
washroom and kitchen
• Number of bedrooms, number of residents,
number of elderly, number of students and
number of children
Fireload density *
Occupancy
Fractile (MJ/m 2) **
Average
(MJ/m 2)
80%
90%
95%
Dwelling
780
870
920
970
• Density of common combustible materials
Hospital
230
250
440
520
• Calorific values of common combustible materials
Hospital storage
2000
3000
3700
4400
Hotel bedroom
310
400
460
510
Data processing in deriving the fire load
density of each flat
Offices
420
570
670
760
• The volume of a furniture is calculated by
multiplying the measured dimensions
Shops
600
900
1100
1300
• Measured dimensions and material type of furniture
• Mass of furniture = volume x density
Manufacturing
300
470
590
720
• Heat release from a combustible furniture =
calorific value x mass of a furniture
Manufacturing & storage ***
1180
1180
2240
2690
Libraries
1500
2250
2550
–
Schools
285
360
410
450
Note 2: T
he values given in this table included only the variable fire loads (ie, building contents).
If significant quantities of combustible materials are used in the building construction,
this should be added to the variable fire load to give the total fire load
•
Conclusion and next steps
Results of a survey for estimating the fire load
density of residential flats and offices have been
presented in this study. With reference to the
figures given in a British Standard, the average
fire load density of a dwelling is 780 MJ/m 2 and
for offices 420 MJ/m 2. These are far below the
survey results obtained in Hong Kong.
This may indicate that the average floor space
occupied by the people in Hong Kong is less
than that in the UK and the corresponding fire
load density may then be higher. The implication
is that in each country, local information is
important for determining the fire load density
and that a reliable fire safety engineering
study should be supported by local data.
Table 2. has listed the fire load densities of a range
of classifications used in the UK. It is expected
that the fire load density will be different in Hong
Kong. This study will be extended to other types
of occupancies, particular retail shops.
Table 2. Fire load densities (extracted from PD 7974-1:2003)
Cumulative Frequency Distribution - Whole Flat Units
In addition the survey of fire load density of
various buildings in Hong Kong, the heat release
rate (the design fire size) of these buildings can
be estimated more accurately. The survey data
provides strong support to the performancebased design in Asia and a good reference to
other regions.
The study will be further extended to conduct
fire risk analysis for various buildings by
combination of the likelihoods and the
consequences. The consequences of the fire,
and the resulting loss of property and life, is
related to the severity of the fire. Fire risk
assessments are one of the major aspects in
Performance-Based Fire Safety Engineering
approach. Based on the determination of the
appropriate design fires, it allows the evaluation
of fire risk level as well as assessment of
fire protection requirement.
1901-2000
1801-1900
Fire Load Density (MJ/m2)
Fire Load Density (MJ/m2)
Fig 4. Fire load density histogram, whole flat units
1701-1800
1901-2000
1801-1900
1701-1800
1601-1700
1501-1600
0
1601-1700
2
1501-1600
4
1401-1500
6
1301-1400
8
1201-1300
10
100
90
80
70
60
50
40
30
20
10
0
1001-1100
Cumulative Fequency (%)
12
1101-1200
Fire Load Density Histogram – Whole Flat Unit
1401-1500
It is well known that either the fire load (the
amount of combustible materials) or ventilation
(the amount of oxygen supply) can be a control
factor to the heat release rate (design fire size).
For a high fire load space (a residential apartment
or an office), the maximum design fire size will be
determined by ventilation. Under fire conditions,
ventilation will mainly come from the windows
after the glass is dislodged and the doors are
open. In general, the area of glazing windows
in Hong Kong is greater than that in the UK.
Compared with the UK situation, a fire in Hong
Kong could last longer and be more severe as
indicated from the results of this survey.
•
Derived from surveys: see CIB W14 Workshop Report,1985 *
The 80% fractile is the value that is not exceeded in 80% of the rooms or occupancies **
Storage of combustible materials at less than 150 kg/m 2 ***
1001-1100
Fig 3. Flow chart for data collection
•
N0. of Flats
Determine the fire load density at
various type buildings
1301-1400
Analysis of data and discussion
• Determine the correlation of between the
parameters (ie. type of occupants, number of
occupants) fire load density
Note 1: L
imits. The fire load densities given in this table assume perfect combustion, but in real fires,
the heat of combustion is usually considerably less.
1201-1300
• Fire load density = total heat release / overall
floor area
1101-1200
• Total heat release in a flat = adding up the hear
release from all combustible furniture
Fig 5. Cumulative frequency, whole flat units
Acknowledgements
The work described in this study was the
result of a collaboration with the Department of
Building and Construction and City University
of Hong Kong.
References
Raes H., The Influence of a Building’s
Construction and Fire Load on the Intensity
and Duration of a Fire, Fire Prevention
Science and Technology, Vol.16: 1977, 4-16.
Kumar S., Rao C.V.S.K., Fire loads in
office buildings, Journal of Structural
Engineering-ASCE, Vol.123, Issue 3:
1997, 365-368.
British Standard Institute, PD7974-1: 2003:
The Application of Fire Safety Engineering
Principles to Fire Safety Design of Buildings,
British Standard Institute, BSI, 2003.
21
An investigation of fire load density for residential buildings and offices in Hong Kong
Data collection
Retrofitting private homes at scale:
investigating the business case
Authors: Ann Cousins, Matt Gitsham
22
The household sector represents
27% of total UK emissions and
achieving substantial cuts here
is imperative. Whilst there
has been some retrofitting
activity in social housing, action
has been very slow moving in
the privately owned sector,
which accounts for 93% of
Bristol City housing.
Abstract
This research looked at the opportunities
and barriers, in terms of finance and energy
efficiency, for the refurbishment of private
housing in large-scale contracts. The study
looked at energy savings and costs for a range
of retrofit measures, such as insulation or new
windows, as applied to 12 different housing
types (defined by age and size), typical to the
Bristol area.
The main findings were that there are significant
financial savings to be made in contracts of
more than 50 houses, as large contractors
become interested. Further savings can be made
where houses are all within walking distance of
each other, rather than spread across a city, as
the costs of management and logistics increase
as homes are further apart.
Achieving an A-Rated home is difficult with the
existing stock; indeed none of the packages of
measures studied achieved this. Larger, older
houses present the greatest potential for carbon
savings, as they have higher CO 2 emissions prior
to any retrofit. Retrofitting existing homes will
contribute significantly to achieving the UK’s
carbon emission targets.
Introduction
There are an increasing number of Government
targets and wider initiatives for new homes
producing low carbon homes that enable
sustainable lifestyles.
This study aimed to identify the opportunities
and barriers, in terms of finance and energy
efficiency, for the refurbishment of private
housing in large-scale contracts. The study used
the west of England as a case study and took
a systems-based approach to analyse the scale
of savings available. Rather than creating a
consumer-facing tool, the primary audience for
this work would be investors, contractors and
local authorities, who want to consider the
economics for large refurbishment contracts,
similar to those carried out in social housing.
Methodology
Categorising the housing stock
Data provided by the Centre for Sustainable
Energy (CSE) presents the housing stock in the
West of England broken down by local authority
area, age of property (presented in approximately
30 year ranges), built form (detached, semi,
terraced etc), number of bedrooms, and access
to gas. This data showed 237 housing types in
the area of the Bristol City Region.
A final list of 12 predominant housing types
were chosen to be examined representing
approximately 55% of the west of England
housing stock.
Determining the base case
A review of evidence was done to determine
whether it would be possible to identify the
current state of the housing stock in the west
of England. It was decided that for completeness,
we should define the base case houses as being
completely unimproved (excluding the addition
of central heating), with no insulation or any
other additional measures.
To identify further details relating to the standard
construction of an “average” house, reduced
data Standard Assessment Procedure (SAP)
was used to produce a base case SAP rating
for each housing type.
Package One
Energy Saving Lighting
Draught proofing
Roof insulation
Cavity wall insulation
Package Two
New windows
New boiler and controls
Package Three
Internal wall insulation
Floor Insulation
Package Four
External wall insulation
Floor Insulation
Solar thermal
Solar PV
Table 1. Measures contained within each package
Retrofit measures
An initial list of potential housing retrofit
measures to enable householders to live
more sustainable lifestyles was drawn up.
This included measures to contribute to
CO 2 reduction, climate change adaptation,
wellbeing and other sustainability measures.
In order to prioritise this long list, it was decided
to concentrate on carbon reduction measures.
To enable the production of a meaningful model,
we chose areas that could be measured using
SAP. We excluded wind energy, as the effectiveness
of this technology varies significantly depending
on location. Further work was done to define
each measure (eg material used and thickness of
insulation) and relevance to each housing type.
SAP calculations
SAP is a logarithmic scale; adding different
retrofit measures to a base case house has
differing results depending on the combination
of measures applied at any one time. It was
therefore decided the most effective way of
providing the results required to fulfil the aims
and objectives of the study, was to define
four packages. Each was designed to efficiently
combine measures, as appropriate, and to
become progressively more expensive
(with package four being the most expensive).
Table 1. shows the measures contained
within each package.
23
Base house
Package one
SAP EPC
CO2
(kg/yr)
band
3-bed semi,
1930 to 1980
Package two
CO2
CO2
SAP EPC
(kg/yr) saved
band
(kg/yr)
Package three
CO2
CO2
SAP EPC
(kg/yr) saved
band
(kg/yr)
Package four
CO2
CO2
SAP EPC
(kg/yr) saved
band
(kg/yr)
CO2
CO2
SAP EPC
(kg/yr) saved
band
(kg/yr)
8916
35
F
5244
3672
59
D
2627
6289
79
C
2433
6483
80
C
2219
6697
82
B
3-bed semi,
1900-1930
11895
27
F
9361
2534
39
E
5467
6428
63
D
2818
9077
80
C
2253
9642
84
B
3-bed Semi,
Victorian
15549
17
G
11680
3869
31
F
7864
7685
49
E
3472 12077
76
C
2790 12759
81
B
3-bed Semi,
Georgian
13051
18
G
10601
2450
28
F
6925
6126
48
E
3084
9967
76
C
2473 10578
81
B
2-bed terrace,
1900-1930
6993
28
F
4978
2015
44
E
2762
4231
68
D
1566
5427
81
B
1133
5860
87
B
2-bed terrace,
Victorian
7280
35
F
5239
2041
51
E
3098
4182
70
C
1763
5517
82
B
1323
5957
87
B
2-bed terrace,
Georgian
7817
31
F
5626
2191
47
E
3425
4392
67
D
1828
5989
81
B
1325
6492
87
B
4-bed detached,
post 1980
6923
46
E
5036
1887
61
D
3567
3356
71
C
2922
4001
76
C
2386
4537
81
B
2-bed flats,
1945-1980 (top)
5785
1
G
3190
2595
27
F
1459
4326
78
C
1363
4422
79
C
919
4866
87
B
2-bed flats
1945-1980 (middle)
4640
23
F
3756
884
34
F
1730
2910
81
B
1377
3263
84
B
-
-
-
-
3-bed detached,
1945-1980
9434
27
F
5638
3796
52
E
3375
6059
70
C
2844
6590
75
C
2386
7048
79
C
3-bed terrace,
1945-1980
7895
39
E
5133
2762
59
D
2133
5762
82
B
2005
5890
83
B
1545
6350
87
B
Table 2. CO 2 savings and SAP ratings
Each package is applied in order ie you cannot
install package two until you have installed
package one, but packages three and four are
mutually exclusive. A choice must be made
between internal and external wall installation.
Average breakdown of costs
1% 9%
Results and discussion
Costs and savings
There are savings to be made through a whole
house approach rather than by individual
measures. ROK’s cost analysis suggests that
an average of over 4% can be saved by
installing measures in packages. These savings
are made predominantly due to the fact that
more than one measure can be carried out by
the same trade at the same time, cutting down
on overhead costs. There may also be savings
in preparation time and plant (eg scaffolding
could be used for more than one measure).
Our benchmarking exercise demonstrated that
there are significant variations in costs of single
measures to single dwellings. A large scale
approach would remove this variability and
give the consumer some consistency in
what they could expect to be charged.
There are significant savings to be made by
retrofitting enough houses for a large contractor
to be interested in the study. The exact number
24
Carbon emmisions in the home
28%
Labour
Heating homes
Plant
57%
Materials
Overhead & Profit
Costs
Cost information was gathered predominantly
from ROK, which provides building repairs
and refurbishment services and a small local
specialist retrofit firm Footprint Building.
This was benchmarked against prices available
from the Department for Energy and Climate
Change (DECC), the Energy Savings Trust (EST),
the Centre for Sustainable Energy (CSE), and
World Wildlife Fund (WWF). The local builders
costs will be used to compare the costs of
retrofitting individual homes (1-19), with the
costs from ROK of retrofitting a large number
of homes (20+) as one contract. Maintenance
costs were excluded from our analysis.
3%
15%
25%
Heating water
Appliances & lights
Cooking
62%
Fig 1. Average breakdown of costs
Fig 2. Carbon emissions in the home
of homes at which a large contractor might be
interested is not an exact science, but following
discussions with ROK and Footprint, we have
assumed that this would be at around 20
homes. Further savings will occur when a large
contractor is able to get further bulk savings.
This is assumed to be for 50 houses.
they are not familiar with and working with an
unknown team, and the cost will reflect this risk
for them. As the market changes, and installing
these more “advanced” measures becomes more
mainstream, it is likely that the cost of large
contractors will come down to reflect this change.
In order to achieve the best financial savings,
retrofit at scale will have to happen in clustered
groups. Cost savings from carrying out
measures at scale are likely to be cancelled
out by the extra transport, management and
logistics costs of spreading houses too thinly.
For the majority of retrofit measures, the
most significant costs are associated with the
materials. As can be seen in Fig 1. across all
of the measures studied, except the renewable
options, over 60% of the cost is associated with
materials. Whilst ROK’s buying power is already
likely to mean that many of these materials are
already at a significantly reduced cost, it is
possible that with more widespread take up,
the market will become more competitive
and prices will come down.
At present, there are a small minority of
measures that are cheaper from a small
specialist builder. The most obvious of these
are solar PV and solar thermal, which are
currently cheaper through a small specialist
contractor. However, mainstream contractors
currently have limited experience in renewable
technology. They will be installing systems that
SAP and CO2 savings
Table 2. outlines the SAP and CO 2 results for
each of the four packages. As can be seen,
the SAP improvements and CO 2 savings are
less significant in packages three and four. This
is not necessarily a reflection that the additional
measures in packages three and four are less
effective, but rather that the energy savings
available become smaller as you progress
through the refurbishment options.
The Government sets out that SAP ratings are
based on the energy costs associated with
space heating, water heating, ventilation and
lighting. Carbon savings can be achieved
without any saving in the cost of energy, by the
installation of renewable energy technologies.
This can been seen in the example of the
three-bed semi-detached property constructed
between1930-1980, which only moves from a
SAP rating of 79 to 82 between packages two
and four, but saves 408kg CO 2/year.
The base case flats are assumed to be fitted
with electric heating, so the significant impact
for them, particularly in terms of SAP rating is
when the new A-rated gas combi-condensing
boiler is added in package two.
Improvement results per house
# Houses
Intended refits packages
Houses
dispersed/
clustered
3-bed semi, 1930 to 1980
50
Clustered
Existing
Existing
Existing
Install
80
82
B
B
14,630
214
68.36
3-bed semi, 1900 – 1930
50 Dispersed
Existing
Existing
Existing
Install
80
84
B
B
24,289
565
42.99
3-bed Semi, Victorian
50
17
31
G
F
1,461
3869
0.38
3-bed Semi, Georgian
50 Dispersed
18
28
G
F
1,332
2450
0.54
2-bed terrace, 1900 – 1930
15
Clustered
Existing
Existing
Existing
Install
81
87
B
B
19,264
433
44.49
2-bed terrace, Victorian
15 Dispersed
Existing
Existing
Existing
Install
82
87
B
B
21,431
440
48.71
2-bed terrace, Georgian
15
Clustered
Install
31
47
F
E
1,334
2191
0.61
4-bed detached, post 1980
15 Dispersed
Install
46
61
E
D
2,610
1887
1.38
New windows/Boiler and controls
Internal wall insulation/Floor Insulation
External wall insulation/Solar thermal/Solar PV
Clustered
One
Two
Three
Four
Install
Install
SAP
before
(/100)
SAP
after
(/100)
SAP
rating
before
SAP
rating
after
Cost/
house
(£)
CO2
saving
(Kg/yr)
£/Kg
CO2
2-bed flats, 1945–1980 (converted)
1
Clustered
Existing
Install
27
78
F
C
5,720
1731
3.30
2-bed flats 1945–1980 (purpose built)
1 Dispersed
Existing
Install
34
81
F
B
5,272
2026
2.60
3-bed detached, 1945 – 1980
1
27
52
F
E
2,446
3796
0.64
CO2 saved
(kg/yr with
package 4)
% CO2
saved
Clustered
Install
Fig 3. Refit Packages and improvement results
It is interesting to note that the properties that
attain the best SAP rating, after implementing
all four packages, are the smaller, terraced
houses and flats. They will have the smallest
external wall area, and so lose least heat in this
way. It also reflects the fact that smaller homes
cost less to run, as there is less space to
provide heat and light to.
As can be seen from the results in Table 2.
it is not easy to achieve an A-rated home,
which requires a SAP rating of 93. This would
require improvements in u-values (a measure of
thermal resistance), or an increase in renewable
technologies, either at the individual home or
community scale. Either of which will add
significant extra costs to the retrofit.
What contribution can be made to the UK
carbon reduction targets?
The UK Climate Change Bill requires that the
UK reduces its greenhouse gas (GHG) emissions
by 80% by 2050. The Low Carbon Transition
Plan sets out the emissions reduction required
from each sector and states that our homes will
need to be near zero carbon by 2050. One of
the aims of retrofitting existing housing has to be
to help reach this target. The solutions proposed
as part of this study only consider single house
solutions; any locally distributed energy was
outside of our scope, but may help in reaching
a zero carbon housing target by 2050.
Table 3. sets out the percentage CO 2
reduction that can be obtained from applying
all four packages of retrofit measures to each
of the housing types.
The average percentage reduction obtained is
78.69%. The 4-bed detached house, post 1980
will have had higher energy efficiency rating in its
base case, which explains the lower percentage
saving of just 65.5%. The middle-floor, purpose
built flats could not install package four as it is
assumed that they did not own the roof space,
so could not install solar panels. This explains
the lower percentage of carbon savings at just
70.3%. This highlights the need to look beyond
single house measures in the long term.
It is worth noting that whilst the CO 2 savings
represent accurate savings from our base
case houses, the base case houses represent
minimum likely standards, rather than the
average of the current housing stock in Bristol.
It is also important to understand the elements
that make up a home’s CO 2 emissions, these
are set out in Fig 2. Emissions associated with
appliances and cooking are not included in our
calculations. There are clearly opportunities to
reduce emissions associated with these areas
by fitting homes with A+ rated appliances.
Installing packages one and two on approximately
55% of all homes in the west of England
(429,105 homes) results in an average CO 2
saving of 62.9% per house in the first year, at an
average cost of £0.84 per kg of CO 2 saved.
These figures are estimated average costs of
carbon savings and the base case homes do
not represent the accurate current condition
of all homes in the Bristol City Region, but the
minimum possible standards. However, this
approximate approach is useful to indicate
potential contributions available for reducing the
28.6% of emissions that currently come from the
housing sector. It is estimated that the west of
England’s total carbon emissions could be
reduced by 9% by implementing such a
programme and that emissions from housing
could be reduced by up to 30%.
Conclusion and Next steps
The study resulted in the production of a model
to assess SAP improvement, carbon reduction
and costs of retrofitting each of the house
types considered, which is shown in Fig 3.
In conclusion, the principal benefits of
retrofitting private homes at scale include:
• consistency in price for individual homeowners
• reduced costs (of up to £12,000) if a large
contractor is involved (20 houses or more)
• further cost reductions (up to an additional
£1,500) for over 50 homes
• providing a significant contribution to the
UK’s target to reduce carbon emissions by
80% by 2050
The results of this research will feed into Forum
for the Future’s Refit West Study, which is aiming
to retrofit 1000 private homes in the west of
England by 2011.
The model produced features the predominant
housing types in the west of England, but could
easily be adapted to other areas of the country.
(Beds/type)
3/semi 1930-1980
6,697
75.1
3/semi 1900-1930
9,642
81.1
3/semi Victorian
12,759
82.1
3/semi Georgian
10,578
81.1
3/terrace 1900-1930
5,860
83.8
3/terrace Victorian
5,957
81.8
3/terrace Georgian
6,492
83.1
4/detached post 1980
4,537
65.5
2/flats 1945-1980 (top)
4,866
84.1
2/flats 1945-1980 (middle)
3,263
70.3
3/bed detached 1945-1980
7,048
74.7
3/bed terrace 1945-1980
6,350
80.4
Average:
78.6
Table 3. Carbon savings for each housing type
Acknowledgements
This research formed part of a Forum for the
Future Engineers for the 21st Century study; a
collaboration between Arup, Forum for the Future,
ROK and inputs from the Centre for Sustainable
Energy (CSE) and Footprint Building.
References
Defra, The Government’s Standard Assessment
Procedure for Energy Rating of Dwellings,
BRE on behalf of Defra, 2005, 3.
Boardman B., University of Oxford Environmental
Change Institute, Home Truths: A Low Carbon
Strategy to Reduce UK Housing Emissions By
80%, ECI Research Report 34, University of
Oxford, November 2007.
CAG Consultants and Energy Action Scotland
for WWF, Carbon Countdown for Homes:
How to make Scotland’s existing homes low
carbon, WWF, September 2008.
Centre for Sustainable Energy, Association for the
Conservation of Energy and Moore R. for WWF,
How Low: Achieving optimal carbon savings from
the UK’s existing housing stock, WWF, March 2008
Energy Saving Trust, Roadmap to 60%:
eco-refurbishment of 1960s flats, Energy Saving
Trust August 2008.
Department of Energy and Climate Change,
Heat and Energy Saving Strategy Consultation,
HM Government, February 2009.
Department of Energy and Climate Change, The
UK Low Carbon Transition Plan: National strategy
for climate and energy, HM Government, July 2009.
Michael Dyson Associates Ltd for Bristol City
Council, Report of Bristol Private Sector House
Condition Survey, Bristol City Council, July 2008.
25
Retrofitting private homes at scale: investigating the business case
Draught proofing/Roof insulation/
Cavity wall insulation
Transient thermal comfort modelling
in underground stations
Authors: Davar Abi-Zadeh, Mohammad Tabarra
26
Concept of local and overall sensation and comfort in the UCB thermal comfort model
Physiology
Tskin
Local sensation vs. local skin temperature
Whole body colder
Local thermal comfort model
Very comfortable
5
Comfortable
-10
-5
0
5
10
Local thermal
sensation
Comfortt
Local sensation
The highly transient nature
of underground environments
leads to difficulty in using
traditional comfort indices
to properly address comfort
conditions. This has significant
impact on the comfort and
well-being of passengers
Local thermal
comfort
Slightly comfortable
Slightly uncomfortable
Uncomfortable
Whole body warmer
-5
T Local
Very uncomfortable
-T Local,set
Local thermal sensation
Overall thermal sensation
1
dominant segment
0.8
Abstract
Underground stations or subway environments
are highly transient and non-uniform.
This is largely due to the massive exchanges
in air volume which occur due to the train
piston effect. In addition, when considered
from the point of view of the passenger and
not just the environment, the conditions that
will be experienced throughout a whole journey
on an underground system vary greatly due
to passenger movement between different
areas, eg external, station mezzanine,
platform, train.
This article presents a new thermal comfort
modelling approach for subways and stations,
which can evaluate local and whole body
thermal sensation and comfort under transient
conditions. The model provides a useful tool for
the evaluation of alternative low-energy design
and operating strategy and mitigation
measurement for operating scenarios.
Introduction
Underground systems represent a significant
challenge when it comes to passenger comfort.
In design, analysis and modelling terms,
underground systems represent highly transient
environments with a great number of variables
all having a significant impact on comfort, eg:
• variation in temperature, air flows and relative
humidity between the outdoor environment,
the station concourse, the platform and
the train
• variations in passengers’ metabolic levels as
they change between running, walking,
standing and sitting
• clothing variations between passengers
There are a number of traditional thermal
comfort indices in current usage, such as
Predicted Mean Vote (PMV), Predicted Percent
of Dissatisfied (PPD) and Effective Temperature
(ET). In addition, the relative warmth index
(RWI) is recommended by the United States
Department of Transportation as a thermal index
for underground station temperature control.
Most of these indices were developed in the
1970s, at which time heat transfer models of
the human body were not so well developed.
asymetrically
weighed segment
0.6
0.4
non-dominant
0.2
0
-6
back
face
e Overall thermal
d
hand
sensation
-4
-2
0
S Local
2
4
-S Mean
6
8
Fig 1. Concept of local and overall sensation and comfort in the UCB thermal comfort model
They are generally derived from quasi-steadystate conditions, such as a typical office
environment and even when applied with
significant experience, cannot fully incorporate
all the factors affecting thermal comfort in a
non-uniform, transient environment or situations
as experienced by underground passengers.
This can lead to unsuitable designs, as
engineers may mistakenly attempt to recreate,
for an underground environment, the level of
comfort required in a typical office, when much
less stringent conditions may actually be
required. This presents a wasted opportunity
to utilise appropriate design in order to minimise
energy usage in underground stations and
other transient situations.
Methodology
UCB thermal comfort model
It is clear that there is a need for better tools to
model such environments. The University of
California, Berkeley (UCB) has over the last 12
years, developed one such tool. Their UCB
Thermal Comfort Model software is designed to
predict thermal comfort under both steady-state
and transient conditions. The model predicts
thermal comfort for 16 body parts. Each part is
divided into core, muscle, fat, and skin layers.
An underlying model of the blood circulation
system simulates the heat exchanges between
the tissue layers, such as muscle and skin. Local
sensation is calculated based on the local skin
temperature of the body part and the mean skin
temperature, which represents the whole-body
thermal state. In transient conditions, the
derivatives of local skin and core temperatures
are included. Local comfort is calculated based
on local and overall sensations. The local
sensation and comfort levels are integrated to
get an overall sensation and an overall comfort.
The underground station model
In this collaborative research between the
University of California, Berkeley, and Arup,
we used the software to develop a model of
geometry and environmental conditions in an
underground system. The principles can
also be applied to transit facilities, such
as railroad stations, multi-modal transit
centres, and airports. Fig 1. shows the
customised graphic user interface of the
Arup underground simulation.
Software outputs
While the software can provide many different
data, such as skin temperature, core temperature,
sweating or shivering rate, PMV, PPD etc, the two
key outputs are the thermal comfort and thermal
sensation. These figures provide the indices
which can be used for comparison of different
environmental conditions and the impact they
have on comfort. The scales are shown in Fig 2.
How these scales are used must be considered
carefully depending on the application.
There is a difference between a comfortable
condition and one which is still ‘acceptable’ to
the user. This is particularly affected by user
expectation. Studies in the offices show that
people consider a slightly uncomfortable
condition acceptable. The comfort value of
minus 1 was recommended as the threshold
for acceptability for an underground station
environment in this study, because people’s
expectations are slightly lower in underground
stations than in buildings.
Modelling scenarios
Typical modelling scenarios were developed for
passengers passing through an underground
station. Each scenario is composed of a
series of transitory “phases”. Each phase defines
its specific environmental and personal clothing
and metabolic conditions.
27
4.0
Very comfortable
4.0
Very hot
4
Mezzanine
2.0
Hot
3
2.0
Warm
2
1.0
Slightly warm
3.0
Comfortable
1.0
Slightly comfortable
0
0
Neutral
Slightly uncomfortable
-1.0
Outdoor
Platform
In train
Outdoor 32º C/50%,
Mezannine 34º C/50%,
Platform 36º C/50%,
150/100 W/m2 dir/dif solar
1
Comfort
3.0
Outdoor 32º C/50%,
Mezannine 34º C/50%,
Platform 36º C/50%,
700/100 W/m2 dir/dif solar
Stairs
Outdoor 32º C/50%,
Mezannine 34º C/50%,
Platform 36º C/50%,
500/100 W/m2 dir/dif solar
0
-1
-1.0
Slightly cool
Outdoor 29º C/50%,
Mezannine 31º C/50%,
Platform 33º C/50%,
500/100 W/m2 dir/dif solar
-2
-2.0
Uncomfortable
-3.0
-4.0
Very uncomfortable
-2.0
Cool
-3.0
Cold
-4.0
Very cold
Fig 2. Thermal comfort and thermal sensation scales
-3
-4
0
5
10
15
Time (minutes)
20
25
30
Outdoor 29º C/50%,
Mezannine 31º C/50%,
Platform 33º C/50%,
150/100 W/m2 dir/dif solar
Fig 3: Effect of external conditions on underground station thermal comfort
The following inputs are defined for each
transitory phase:
• surrounding space geometry – choose from a
predefined geometry, modify existing
geometry, or build a new customised
geometry from scratch
• environmental conditions: defines air
temperatures, air velocities, relative humidity,
solar radiation, azimuth and altitude, surface
temperatures etc
• passenger metabolic level and clothing:
define metabolic rates relating to different
levels of activity: walking, standing, sitting
etc,and clothing, whether summer or winter.
Work is in progress to update the clothing editor
with comprehensive customisable options.
A number of phases were combined to create a
complete transient scenario. Two scenarios were
simulated; normal mode and congested mode.
Normal mode models the movements of a
passenger and conditions for an underground
train in normal operations, ie moving from one
station to another without stopping in tunnels
for extended periods. The conditions are
illustrated in Fig 4.
Congested mode examines the thermal stress of
passengers when conditions start to worsen due
to the train being stalled in a tunnel between
stations. The simulation looks at the time taken
before a person’s physiological state reaches a
danger threshold.
Results and discussion
Response to environmental transition
The transient nature of the model allows certain
conditions to be simulated, the effects of which
cannot be assessed using steady-state models.
Of particular note is the effect of transition
between one environment and another. In this
case, several conditions were simulated to see
whether there is a ‘carry-over’ effect as the
passenger moves from varying external ambient
conditions into the station.
A number of different conditions were
considered, varying ambient temperature and
solar radiation. Two conditions for ambient
temperature were used: 29°C and 32°C.
28
Outdoor Walking
(32°C, 50% RH, 10 min)
Mezzanine
(34°C, 50% RH, 1 min)
Platform
(conditions vary, 5 min)
Stairs (34°C,
50% RH, 1 min)
Train air conditioned
(25.5°C, 55%, 10 min)
Train arrival
24 sec, step into it
Fig 4. Simulation scenarios for a typical normal mode operation
Corresponding platform temperatures of 33°C
and 36°C were defined based on experience.
Fig 3. shows the results of varying these
conditions. It shows that in general, solar
load experienced while outside does not
affect the comfort on the station platform.
In fact the residual discomfort caused by the
external solar radiation gradually disappears as
the passenger wallks through the mezzanine/
stairs down to the platform area, with the
comfort difference finally reducing to 0.5 scale
units. This shows the residual memory of
discomfort is short-lived and justifies the
transient modelling approach. Such a method
allows a comparison of comfort conditions that
can be achieved using different cooling
mechanisms by considering the integral effects
of the whole scenario with multiple phases.
Testing platform cooling solutions:
Platform air conditioning
Simulations were carried out to compare
traditional cooling of the platform air with
local cooling by radiant panels.
In the first set of simulations, external ambient
conditions and conditions in the stair and
mezzanine were kept constant. Two platform
temperatures were simulated 27.8°C and 29.4°C.
Results are shown in Fig 5. They indicate that,
while a lower air temperature results in a more
comfortable condition after five minutes waiting
on the platform, the difference is not great
(0.5 scale units) and the conditions only just meet
the comfort criterion of minus 1 that is considered
acceptable. The increase is small considering
the volume of treated air required to cool the
whole platform by 1.6°C. This indicates that
investigation of other cooling methods may
result in more or equally favourable solutions.
The sharp drop in comfort at 17 minutes is due
to the arrival of the train pushing warm tunnel air
into the platform.
Testing platform cooling solutions:
Radiant panels
Radiant cooling has been successfully integrated
in large public buildings and presents a viable
cooling solution for underground stations with low
latent load. In this study, thermal comfort was
simulated for cooled floor, ceiling and vertical
panels. The cooled panels are all at 20ºC, a
value that would not cause condensation in
most situations. Results are shown in Fig 5.,
Fig 6., Fig 7., Fig 8. and Fig 9.
At the end of the period on the platform, the
best comfort results are achieved with cooled
vertical panels. In all cases, the cooled panels
achieved a comfort condition above the required
value of minus 1. Vertical cooled panels
achieved results almost identical to the 27.8°C
platform air condition, and a cooled floor panel
achieved results only slightly below this. A
cooled ceiling provided the least comfortable
condition, but is still above the required criterion.
Mezzanine
1.5
Outdoor
1.5
Stairs
Platform
In train
1.0
Platform
0
-0.5
-1.0
Threshold for thermal
environmental acceptability
0
-0.5
-1.0
Piston effect
5
29.4°C
10
15
Time (minutes)
20
25
Ttunnel, ºC
0
In train
5
10
15
Time (minutes)
20
25
0
-1.0
Piston effect
-2.0
30
0
29.4°C platform
RHtrain, max, %
10
15
Time (minutes)
20
1.5
Time to RHtrain,
max,min
Mezzanine
1.0
36
10
100
15
36, with ventilation
36
7
71
12
31, no ventilation
34.6
21
100
15
31, with ventilation
34.2
17
62
16
27.8°C platform
Outdoor
Stairs
Platform
In train
0.5
0
-0.5
-1.0
Piston effect
-1.5
Table 1. Time taken to reach stable temperature and humidity conditions when a train is stalled in a tunnel
At the four different conditions, the train was
modelled to see how long it took for the
temperature to stabilise. The changes in
temperature and humidity were then used to
model the response of passenger core body
temperature. The conditions are shown in Table 1.
The difference in relative humidity between
ventilated and unventilated is quite significant.
Without ventilation, the car reaches saturation
(100% RH) in 15 minutes. With ventilation, the
air does not reach saturation, with a maximum
of 71% at 36°C and 62% at 31°C. Maintaining
conditions at a low RH will be key in reducing
the thermal stress of passengers, therefore
it is crucial that trains are ventilated in the
congested mode.
The response in passenger core temperature in
the worst case (36°C/no ventilation) is presented
in Fig 9. for two different metabolic rates,
representing seated passengers and standing
passengers. For seated passengers, the time to
reach the threshold core body temperature is
approximately 40 minutes. With the higher
metabolic rate of a standing passenger, the
threshold is reached after only 33 minutes.
Conclusion and next steps
Several scenarios have been simulated using
a transient approach to thermal comfort, to
determine how thermal comfort in an underground
system is affected by various external and station
conditions. Although the effects of ambient
conditions may carry over for a short period
when the passenger is in the mezzanine/stair
area of the station, there is little to no carry over
effect on comfort once the platform is reached.
Lowering the platform air temperature from
29.4ºC to 27.8ºC improves comfort 0.5 scale
units. This is not a great effect when considering
the energy required to treat the air in the
platform space. It indicates potential for
low-energy strategies like spot-cooling to be
used to cool the platform air directly in the
occupied zone.
Radiant cooling strategies, such as cooled floor,
ceiling and vertical panels can potentially provide
comfort equal to lowering the platform air
temperature as with conventional HVAC. In an
underground platform, 20ºC radiant panels can
have the same comfort effect as reducing the air
temperature by 1.6ºC. The comfort with vertical
panels is slightly better than the cooled floor, which
is in turn slightly better than the cooled ceiling.
During the congested mode, train ventilation is
very important to reduce interior humidity, which
has a significant effect on passenger heat stress.
In the case where 36°C and no ventilation were
modelled, passengers reached the safety
threshold for thermal stress after approximately
33 minutes while standing, or 40 minutes while
seated. In extreme congested cases passengers
should try to reduce their metabolic level, by
staying calm and seated.
Overall, this research demonstrates that, with
the right tools, a new approach can be taken to
thermal comfort by modelling transient effects
and considering multiphased scenarios, which
can be used to investigate and demonstrate
alternative cooling solutions.
-2.0
0
5
10
29.4 ºC platform
15
Time (minutes)
20
25
30
50
60
27.8 ºC platform
29.4 ºC platform, 20 ºC vertical panels
Fig 8. Effect of cooled wall panel on comfort
39.0
Stand
Core temperature (ºC)
Four conditions were simulated for tunnel air
temperatures of 31°C and 36°C, and either
mechanically ventilating the train car at 0.26m³/s
or no mechanical ventilation. The geometry of
the train follows the London Underground’s
Central Line stock, with 34 seated and 97
standing passengers in a car.
30
Fig 7. Effect of cooled ceiling panel on comfort
36, no ventilation
Congested mode
Passenger heat stress was analysed in the
congested mode to determine how long
passengers can tolerate the elevated
temperatures before their core temperature
begins to reach unsafe levels. A core
temperature of 37.5°C is used as a threshold,
set slightly lower than that used for US Army
soldiers (38°C) to represent the average citizen.
25
29.4°C platform, 20°C ceiling panel
Fig 6. Effect of cooled floor panel on comfort
Time (min) to reach
Ttrain, max, min
5
29.4°C platform
27.8°C platform
29.4°C platform, 20°C floor panel
27.8°C
Ttrain, max, ºC
Platform
-0.5
30
Fig 5. Effect of platform air temperature on comfort
Stairs
-1.5
Thermal comfort
0
-2.0
Outdoor
0.5
-1.5
-1.5
Mezzanine
1.0
In train
Thermal comfort
Thermal comfort
Thermal comfort
Outdoor
1.5
Stairs
0.5
0.5
-2.0
Mezzanine
1.0
Seat quietly
38.5
38.0
37.5
37.0
36.5
0
10
20
30
Time (minutes)
40
Fig 9. Time for the core temperature to reach 37.5°C for
seated and sitting metabolic rates
Acknowledgements
The authors wish to acknowledge technical
input from Dr H Zhang of University of
California, Berkeley.
References
United State Department of Transportation,
1976, Subway Environmental Design Handbook,
Vol I, Principles and Applications. 1976, 2-4.
Zhang,H., DongEun K., Arens E., Buchberger
E., Bauman F., and Huizenga C, Comfort,
Perceived Air Quality, and Work Performance
in a Low-Power Task-Ambient Conditioning
System. Center for Built Environment, Internal
Report, 2008.
Wang D., Zhang H., Arens E, Huizenga C.,
Time-series observations of upper-extremity
skin temperature and corresponding thermal
sensations. Building and Environment,
Vol. 42, No. 12, 2007, 3933-3943.
Davar Abi-Zadeh, Nigel Casey, Stefan
Sadokierski, Barry Hodges, 2003, King’s Cross
Underground Station Redevelopment: Analysis
of Air Temperature, Velocity, Pressure and
Quality Conditions, 11th International Symposium
on Aerodynamics and Ventilation of Vehicle
Tunnels, UK, 2003.
Tabarra M., and Guan Y.,Temperature Monitoring
of New York Subway System: Stations and
Trains, 12th International Symposium on
Aerodynamics and Ventilation of Vehicle Tunnels,
Portoroz, BHRG, Slovenia, 2006.
29
Transient thermal comfort modelling in underground stations
2.0
Singapore buried
channel research
Authors: Timothy Mote, Anthony Bowden, and Michael McGowan
30
104˚0’0”E
1˚20’0”N
Ancient river channels lie buried
beneath the superficial geology
of Singapore. These channels
have a significant negative
impact on construction work,
particularly where they go
undetected in the early stages
of construction projects.
103˚50’0”E
1˚20’0”N
103˚40’0”E
Meters 5,000
0
5,000
10,000
Study Area
By using mapped occurrences of palaeochannel
segments, the systems are projected back up
gradient and down gradient to reveal their full
extent. A key step in the interpretation is
understanding a transition from bedrock
controlled valleys into meandering river
systems based on subsurface geology resolved
from major infrastructure projects with large
spatial footprints.
A Geographic Information System (GIS) was
used to integrate over 1200 borings that Arup
has compiled from numerous jobs in Singapore.
Interpretation on the location and extent of
the palaeochannels was made and modelled
in 3D to visualise their projection in relation to
planned infrastructure.
Knowledge of the location of these channels
and where they can be expected to be
found is of great value to further work Arup
carries out in Singapore. The knowledge
can provide warning to both the firm and
our clients of potentially difficult and hazardous
ground conditions.
Study area
Kallang Formation Bukit timah granite
Palaeozoic volcanics
Kallang formation
Huat Choe Formation
Gombak norite
Old alluvium
Jurong Formation Sajahat formation
Tekong formation
Jurong formation
Dyke Rock
1˚10’0”N
Huat choe
formation
Old
Alluvium
1˚10’0”N
Abstract
In Singapore, buried meandering river
channels (palaeochannels) are obscured by
present day surficial cover making prediction
of their extent difficult. The location of the
palaeochannels beneath the superficial geology
of Singapore was mapped within the context
of the geologic history and subsurface
investigations of the area. Prediction of yet
unidentified channels was made based on a
development of a geological model.
Dyke rock
103˚40’0”E
103˚50’0”E
104˚0’0”E
Fig 1. Study area and geologic map
Introduction
The Central Business District (CBD) of Singapore
is founded on flat lying terrestrial and marine
sediments at the southern end of Singapore
Island. This CBD is located across the opening
of a number of filled northwest to southeast
trending valleys. In the past, tectonic activity
and variable sea-level conditions both enhanced
the development of the river channels and
subsequently filled the channels with soft
unconsolidated marine and alluvial sediments.
The filled channels or palaeochannels have been
shown to reach a thickness of 50m below the
present ground surface. These palaeochannels
follow the direction of existing rivers, with the
most extensive being the Kallang Basin,
extending for a considerable area around
the Kallang River.
Detailed knowledge of the location and extent
of these palaeochannels after they emerge
from the bedrock controlled valleys and flow
towards the coast is limited, as the palaeo-fluvial
systems are now filled with younger sediments.
These palaeochannels have a significant
negative impact on construction work,
particularly where they go undetected in the
early stages of projects. The channels are often
filled with marine clay and viewed as a challenge
for foundation and underground construction.
Over the last couple of decades, significant
underground infrastructure projects have been
developed in Singapore, including the Mass
Rapid Transit (MRT) network. Subsurface
investigations from these and other projects
have identified palaeochannels generally
coincident with the current rivers, as expected.
A number of investigations have also
encountered palaeochannels in unanticipated
areas away from the current rivers.
These investigations provide key insights to the
location and extent of the palaeochannels, but
they are limited within the project’s footprint
and do not characterise their greater location
and extents across the CBD.
To characterise the potential existence and
occurrence of palaeochannels under the
CBD, published and unpublished subsurface
data are modelled in the context of current
understanding of the regional geologic and
geomorphologic history.
Geologic setting
The study area is centred on the CBD of
Singapore Fig 1. Singapore is an island
city-state located at the southern tip of the
Malay Peninsula. The centre of Singapore Island
is built on the Bukit Timah granite extending
8km north to south and 7km east to west. The
Bukit Timah granite forms Bukit Timah Hill,
the topographic high (166m) region north and
northwest of the CBD. This topographic high
creates the headwaters of the river valleys
that flow to the CBD.
The CBD is built over thick layers of marine clay
and other young sediments of both terrestrial
and marine origin comprising the Kallang
Formation. These sediments are present in
low-lying coastal areas, particularly where
reclamation has been carried out and in-filled.
The Kallang covers most of the CBD Fig1., the
immediate offshore zone and the deeply incised
river valleys which penetrate to the centre of
Singapore Island. The formation has the widest
distribution of all Quaternary (last 2 million years)
sediments on Singapore, blanketing most of the
areas below 5m above sea level and also forming
valley fill at higher elevations. It consists of late
Pleistocene (approximately 2 million years ago)
31
Controlled bedrock to
meandering river transition
MRT
Reclamation
Kallang
Confirmed occurence
500
0
500
Not encountered
1,000 Meters
Fig 2a. Plan view of confirmed occurrences of palaeochannel systems
to Holocene (younger than 10,000 years) and
recent deposits which are of marine, alluvial,
littoral and estuarine origin. Within the Kallang
the most important unit is the marine member
locally called the marine clay. It occurs over a
quarter of Singapore island but does not
outcrop. Its thickness is extremely variable
with a thickness of at least 55m. This unit
is geotechnically problematic with in situ
moisture content close to the liquid limit.
Development of palaeochannels
During the last 65 million years (Tertiary sub-era)
block faulting and movement along pre-existing
faults created large river valleys draining
southeast. The upper reaches of these older
fluvial systems were structurally controlled and
followed a regional structural trend, northwest
to southeast. As the rivers approached the
continental margin they emerged from the
bedrock controlled valleys, transitioning into
a braided fluvial system depositing their
sediment to form the Old Alluvium formation
with a thickness up to 149m.
Tectonic activity in the late Tertiary consisting
of block faulting and warping of the Older
Alluvium, created a northwest trending tectonic
depression bounded on both sides by rising
granite hills. These tectonic controls combined
with a low sea-level and enabled a new fluvial
system to scour the Old Alluvium.
After the cessation of the tectonic activity in
the Pleistocene the river valleys were filled in.
This in-filling resulted from major changes in
sea level which occurred during late glacial and
post glacial times. As sea levels rose after the
final glaciations the valleys were flooded and
filled with marine sediments. This superimposed
a Kallang-aged fluvial system onto an older Old
Alluvium-aged system. Sediments of the Kallang
continue to be deposited in the present day.
32
Fig 2b. Oblique view of confirmed occurrences of palaeochannels
Methodology
Data compilation
This study compiled subsurface data from
sources including geotechnical investigations
for some major infrastructure projects in
Singapore in published reports as well as
unpublished geotechnical engineering reports.
To support 3D modelling and interpretation the
data was stored in a GIS. A total of 1245 boring
logs were examined: of those, 656 recorded the
thickness of Kallang.
Many of the historic boring logs were lacking
this key information. When the logged borings
were lacking coordinate data, location maps of
the logs were georeferenced and locations were
digitised. Critical to the development of the 3D
model is the elevation of the base of the Kallang.
This is traditionally calculated from the elevation
of the boring top and the depth to the base of
the Kallang. Where borings reported soil
descriptions and did not identify geologic units,
an interpretation was made to pick the base
of the Kallang. This was generally based on the
bottom of the marine clay or a transition from
markedly softer to harder material.
A digital version of the geology of Singapore as
mapped at 1:25,000 was integrated into the
GIS to provide geologic context. A regional
digital elevation model (Aster source with 30m
resolution) was added to the system and
processed to categorise slope highlighting
topographic control and a potential indication
of shallow bedrock.
All pertinent data sets were integrated into a
3D geologic model to allow for interpretation
of the extent of the palaeochannels in a
geomorphologic context.
Palaeochannel modelling interpretation
By using both factual data and a conceptual
understanding of the geomorphological
environment in which these palaeochannel
systems were developed, their present location
beneath the Kallang can be interpreted.
The interpretation is based on:
• location of known palaeochannels
• location of areas where palaeochannels
are confirmed not to be present
• orientations of current river systems
• general hydrologic gradient towards
the coast
• geomorphological understanding of
fluvial system formed prior to deposition
of the Kallang
• rivers are bedrock controlled in the west of
the study area through Bukit Timah and
Jurong
• rivers are meandering east of the study area
in Old Alluvium
• regional structural trend of northwest to
southeast
• current topographic relief as indicators of
bedrock highs
The existence of palaeochannels within the
bedrock controlled valleys trending northwest to
southeast towards the CBD is well established.
The current rivers, after emerging from these
valleys, generally follow this orientation as well.
There is a secondary conjugate northeast to
southwest structure shown by smaller valleys
and segments of current drainages, specifically
reflected in the Kallang River system. A key
control on the interpretation of the general trend
of the palaeochannels follows these primary and
secondary orientations.
Singapore buried channel research
Bedrock controlled channels
Meandering channels (inferred)
cut into Old alluvium
500
0
500
1.000 Metres
Study area
Kallang formation
Study area
High
Reclamation
Old alluvium
Reclamation
Moderate
Controlled bedrock
to meandering river
transition
Jurong formation
MRT
Low
Confirmed occurence
Not present
Bukit timah granite
Fig 3. Plan view showing transition from bedrock controlled channels
into meandering channels
The confirmed locations of palaeochannels
from subsurface investigations are used to
project the palaeochannels across the study
area following hydrologic gradients down
towards the coast and up toward the source
valleys. Where subsurface investigations have
shown that palaeochannels are not present,
zones lacking palaeochannels in hydrologic
shadows are developed.
When the palaeochannels emerge from the
bedrock controlled valleys they transition into
meandering systems. To the west of this line,
palaeochannels are inferred to be structurally
controlled and relatively straight, while to the
east of the line meanders are be expected.
The curvature in the meanders, hence
uncertainty in their location is anticipated to
increase towards the east.
The interpretation was further enhanced by using
the current topography to represent indicators
of a more resistant structure at depth and a less
likely location of a palaeochannel.
Results and discussion
Using subsurface investigations to initiate the
interpretation, the occurrence of palaeochannel
systems are mapped emerging from the upland
valleys into the coastal plain Fig 2a. and Fig 2b.
Palaeochannel likelihood
The likely occurrence or absence of
palaeochannel systems across the study area
is interpreted from the known occurrences of
palaeochannels and an understanding of the
geomorphological context. Where a known
palaeochannel is shown to occur it can be
connected back up to source zones following
hydrologic gradients and continued down
gradient to the coast. Where previous subsurface
investigation has shown palaeochannels are not
present, it defines zones where palaeochannels
can be expected to not be encountered.
500
0
500
1.000 Metres
Fig 4. Plan view of palaeochannel likelihood mapping
Zones are developed and classified by a high,
moderate or low likelihood of palaeochannel
occurrence see Fig 4.
High likelihood that a palaeochannel system will
be encountered in the subsurface: these areas
are generally down hydrologic gradients from a
mapped palaeochannel and have an orientation
that fits in the context of the regional
geomorphological trends.
Moderate areas where evidence that a
palaeochannel will either occur or not occur is
lacking. Often these are areas with limited
subsurface investigation and as such regional
trends can not be conclusive.
combination of the subsurface investigation
data in a geomorphological context allows for
development of zones of potential occurrence
of palaeochannels.
Further refinements could be made to this
interpretation by the addition of further
subsurface investigation data, particularly in
areas not covered. Of particular value, would
be data from other major infrastructure projects
and offshore seismic reflection data which may
be available for the various reclamation areas
near the CBD.
Acknowledgements
Low likelihood that a palaeochannel would be
encountered in the subsurface: these areas are
in the hydrologic shadows or down gradient from
areas mapped where palaeochannels are not
encountered.
Conclusion and next steps
The likely occurrence of palaeochannels across
the CBD is interpreted using existing subsurface
investigations and an understanding of the
palaeo-geomorphological setting. By using
mapped occurrences of palaeochannels defined
from subsurface investigations, the systems can
be tracked back up gradient or down gradient to
the coast following regional structural trends.
A key step in the interpretation is an
understanding that the palaeochannels follow
bedrock controlled valleys from their source in
the uplands until they emerge and transition into
meandering river systems. Where palaeochannels
are shown not to be present, zones of low
palaeochannel occurrence are defined based on
the hydrologic shadows that these zones create.
A version of this research was presented at the
Underground Singapore 2009 Conference.
The authors would like to acknowledge the
many people who helped complete this
research: Kathryn Nation, Ian Darlington, and
Matthew Uidam in Sydney; Heng Kok Hui and
Chris Deakin in Singapore; and Raymond Koo,
Vicki Lau and Jack Pappin in Hong Kong.
References
Bird M I., Chang C. H., Shirlaw J.N., Tan T.S.,
The T.S., Proc. of the Underground Singapore
Conference 2003.
Chiam S. L., Wong K. S., Tan T. S., Ni Q., Khoo
K. S., and Chu J., Proc. of the Underground
Singapore Conference 2003.
Defence Science and Technology Agency,
Geologic of Singapore, 2nd Edition. Defence
Science and Technology Agency, SIC/2-5/GG,
ISO, 2009.
Shirlaw J. N., Broome P. B., Chandrasegaran
S., Daley J., Orihara K., Raju G.V.R., Tang
S.K., Wong I.H., Wong K.S., Yu K., Proc. of the
Underground Singapore Conference 2003.
Sharam J.S., Chu J., and Zhao J., Tunnelling
and Underground Space Technology, Vol. 14,
no. 4, 1999.
In certain areas, the lack of dense spatial data
precludes detailed delineation of individual
palaeochannels across the CBD, but a
33
The Life Cycle Tower:
a high-rise in timber construction
Authors: Jan Wurm, Tim Göckel, Martin Unger
34
There has been increasing
interest in high-rise timber
construction due to the growing
awareness of economical and
sustainable architecture.
This has lead to an increase
in multistorey timber
constructions across Europe.
Abstract
Timber is carbon neutral and 100% renewable
but its use was originally associated with either
low-budget housing or low-rise eco-homes.
The objective of this study is to demonstrate
the feasibility of a commercial timber high-rise
construction of 20 storeys in a densely
populated urban context at detail design level.
This study will focus on issues related to the
structural design and fire engineering. The layout
of the building with a footprint of approximately
45m x 30m and a 1.35m grid reflects the
requirement of a mixed-use development for
office and residential units. The structural layout
in the currently preferred scheme features a
central core of Glue Laminated Timber
(Glulam) and Glulam Perimeter Columns.
Timber-Concrete Composite Decks provide
transfer of horizontal loads. For fire protection,
a number of structural, technical and design
measures have been developed to compensate
for the combustibility of the material.
Rhomberg is going to build a three storey
prototype of the system on their premises
mid-2010, while investigating sites to transform
the design into the first built timber high-rise.
This showcase study will provide a great
opportunity to lift Arup’s design and engineering
skills in timber construction to a new level.
Fig 1. Principal layout of the life cycle tower with
20 storeys (building notched open at corner for
better visualisation)
Fig 2. Model of the timber core where the edges of
interfacing slabs are tied into the model. The structure
of the first three retail floors (shown in blue) is modelled
in reinforced concrete.
Methodology
In 2000 the prototype of a six-storey timber
frame was built in the UK as the first of its
kind in the world: the so called TF 2000 project.
Extensive research investigating stability,
robustness and fire safety was conducted.
Since 2000, about 100 platform frame
buildings between five and seven storeys
have been built in the UK. The current climax
of this development is the recently completed
nine-storey panelised timber frame building in
Murray Grove, a few miles away from the City
of London. A similar development can be
seen in Europe. In Switzerland revised codes
allow the construction of up to six storeys. In
Germany, in 2008, where the use of timber is
usually restricted to buildings below 13m, a
seven-storey residential development in Berlin
has been approved and built as a Glulam
Skeleton Structure.
Wiehag GmbH, the specialist Austrian timber
contractor, and architect Prof. Hermann
Kaufmann complete the team. The funding
granted by Arup’s Design and Technical
Executive (DTX) was essential to enable the
team to carry out the required Research &
Development Work in connection with this study.
An Austrian consortium of specialists, under
the lead of architectural firm Schluder, initiated
a first feasibility study investigating concepts
for timber structures above eight-storeys
(eight+). This initiative was supported by the
programme ‘The House of the Future’ by the
Austrian Research Fund, Österreichische.
Forschungsförderungsgesellschaft mbH (FFG).
Rhomberg Bau GmbH, a developer based in
Bregenz, who investigated the commercial
aspects of high-rise timber construction during
this feasibility phase, set up the new consortium
‘Life Cycle Tower’ (LCT) involving Arup as
multidisciplinary consultant to focus on the
realisation of a pilot project, securing additional
funds from the FFG. Arup was commissioned as
consulting engineer for structures, building
services, façade engineering, fire engineering,
building physics and materials consulting.
The objective of this study is to take the detail
design for a timber frame high rise to 80%
completion, setting the base for a commercially
viable study to follow.
The main qualitative design drivers are:
• high degree of offsite prefabrication and
reduced work on site, aiming to minimise
construction programme
• highly flexible structure and layout to
accommodate residential, office and hotel use
• energy efficiency and design for the
minimisation of carbon footprint
• no compromise on any performance criteria
in comparison to conventional structure
Key challenges are:
• compliance with fire performance
• compliance with acoustic performance
• cost
An agreed outline specification defining the
performance criteria forms the basis of the highly
integrated design process. The criteria are
based on the more conservative among
European Standards or the National Standards
of the key market places of Austria, Switzerland
and Germany.
35
Fig 3. One floor showing the central core with services
shafts along the short sides. Currently the wall thickness
of the Glulam core is 360mm
Client Workshops for coordination and the
setting of objectives of work stages are held
every six to eight weeks. In between, research
and development is carried out as desk studies
inside the study teams of involved disciplines,
and alternatives and options are developed
and compared within the set criteria.
Work is periodically reviewed by internal
workshops involving specialists from other
Arup offices, such as timber specialist Andrew
Lawrence from the Technical Development
and Support Group in London. In addition,
workshops with external reviewers are also
held to discuss the most challenging issues.
The building
As the prototype design for the Life Cycle
Tower needs to address all relevant issues and
restrictions of a life study, the client selected an
urban development site in Austria. The compact
footprint measuring 27m x 43m is orientated
along westeast axis. The layout and plan are
based on a 1.35m grid in order to accommodate
both office and residential use. The central
core measures 8.10m x 18.90m.
Fig 4. Current FEM model with all structural elements on the left and core only on the right
Fig 5. The floors are built with composite decks with installations running between timber beams. The partition walls
connect to the concrete deck providing obstruction to horizontal fire spread
The first three storeys contain retail areas and
are built as a conventional reinforced concrete
construction on a structural grid of 8.10m.
The floor-to-floor height of the remaining
17 storeys in timber construction is 3.50m.
The columns
The columns along the perimeter are located at
2.70m centres and are 480mm wide and 250mm
deep at its maximum. The specified strength
category is GL24h following EN 14080.
Structure
Early studies in conjunction with the results of
the feasibility study ‘eight +’ showed that a
cross laminated timber frame, as realised in the
Murray Grove Project, would not be feasible for
a highrise structure. This is mainly due to the
structural restrictions at the interfaces. The most
feasible solution is a shell of skeleton timber
frame using Glulam components stabilised
by a central core.
The core
Two concepts for the core are currently being
investigated. A structural timber core of vertical
Glulam beams connected by steel plates and
rods would require a thickness of 300mm. In
order to minimise joints, the maximum available
length of Glulam beams of 35m would need to
be considered. This solution is technically
feasible and cost effective. For comparison we
are pursuing in parallel a conventional concrete
core as a back-up option and for even taller
structures a braced façade.
As such, the primary components for vertical
load transfer are the perimeter columns together
with the central core. Slabs and the vertical walls
of the core provide stability and the transfer of
lateral loads. The applied loads based on
Eurocode EC1 include life load of 3.20kN/m² for
rental floor areas and 5kN/m² for the corridors.
The superimposed load of the structure was
calculated to be 1.50kN/m². The static wind load
increases from 0.66kN/m² on the bottom of the
building up to 1.44kN/m² on the top. In the
following section the intermediate results of
our study are summarised.
36
The slab
A solid timber construction for the slabs
would meet neither the acoustic or the thermal
performance requirements due to its lack of
mass. As an in situ concrete slab construction
would counteract the primary advantage of
the fast and “dry” timber construction,
prefabricated composite decks have been
selected. The decks span generally over
8.10m although an increased span of 9.45m
is currently being analysed. The prefabricated
decks are 2.70m wide and correspond to
the grid of the primary columns.
The build-up of 180mm Glulam beams with
180mm concrete decking has been established
following a parametric study also taking the
frequency and sensitiveness to oscillation into
account. To provide shear stiffness across
the floor slab, the edges of the decks are
monolithically joined with in situ concrete.
Fire
In order to gain building permission it has to be
demonstrated to the authorities that the timber
design of the Life Cycle Tower does not impose
any higher risks to the inhabitants compared to
a conventional building.
The building addresses all fire regulations
except the target that no combustible material
should be used for the primary structure.
Therefore a number of additional measures
compensating for the increased permanent
fire load are considered:
• floors have full sprinklers installed
• prevention of smoke and fire spread
using cavity free building elements for
floors and walls
• design of structural components addresses
reduction of section by 0.7mm/min
• fire spread between units across the floor is
obstructed by partitions walls supported
directly on concrete
The life cycle tower: a high-rise in timber construction
Void filled with insulation material and encapsulated encased with plaster fireboards
Composite reinforced concrete-timber slab
Installation area above suspended ceiling
Fig 6. Longitudinal section through composite deck in the zone of escape corridor
Timber core wall encapsulated
with gypsum plaster board
Composite reinforced concrete-timber slab
Installation area above suspended ceiling
Corridor wall
Fig 8. Example for a timber concrete composite
decking system
Corridor soffit
Fig 7. Cross section through composite deck in the zone of escape corridor
Timber GL24h
Concrete C30/37
Steel
Section
480mm x 250mm
d=300mm
d=220mm, t=16mm
Density
380kg/m³
2400kg/m³
7850kg/m³
Embodied energy
Co-efficient
4,6MJ/kg
1,3MJ/kg
32MJ/kg
Weight, l=3,5m
160kg
600kg
280kg
Embodied energy
per element
740MJ
780MJ
9000MJ
Table 1. Comparison of embodied energy of structural column in timber, concrete and steel
• vertical fire spread between floors is
obstructed by continuous concrete decking
(timber columns are discontinuous)
• systematic prefabricated construction is
ensuring control over all details and
preventing uncontrollable solutions onsite
Results and discussion
Comparison embodied energy
A conventional construction with the same
layout and loads would have led to a column
of 300mm diameter (C30/37) in concrete
respectively to a CHS of 220mm diameter
and a 16mm plate thickness in steel.
Taking into account the specific density of
the materials and their embodied energy
coefficients, the timber column and concrete
column feature a similar quantity of embodied
energy while the steel column requires more
than 12 times the energy see Table 1.
With approximately 40 columns per floor and
17 floors, a total volume of around 300m³
of timber will be used. Over 260 tons of CO 2
will be stored; it has been calculated that
with the timber structure an equivalent of
approximately 1000 ton of CO 2 will be emitted
during manufacturing, construction and service
of 50 years – this is 10 times less than with a
conventional structure.
Conclusion and next steps
The structural design of the Life Cycle Tower
with 20 storeys as a timber skeleton frame of
Glulam elements in connection with a composite
slab system has been developed to detail design
level. The design of the central core in timber
construction is technically feasible, and cost
effective in comparison to a conventional
concrete core.
At present, the detail design for building
services and façade engineering is being
reviewed and a detailed cost analysis is carried
out by our client. We anticipate finalising the
design by the end of March 2010.
In order to build a prototype building of three
storeys on the premises of Rhomberg, a third
application for funding was submitted to FFG at
the end of 2009. The prototype study will allow
the detail design and the final specification to be
completed. In parallel we will continue the
dialogue with external experts and the local
building authorities aiming to get agreement to
our approach to the fire protection strategy while
Rhomberg is investigating suitable sites for
realising the Life Cycle Tower. The foundation
for the first timber highrise has been built.
Fig 9. Visualisation of the currently envisaged façade
concept incorporating PV-modules and “green walls”
Acknowledgements
We would like to thank Rhomberg Bau GmbH
for founding and managing the “Life Cycle
Tower” Consortium and the “Österreichische
Forschungsförderungsgesellschaft mbH (FFG).
In addition we would like to acknowledge
research student Aliénor Dahmen for her
very valuable contribution.
References
Schluder Architektur ZT GmbH, Final Report of
the R&D project 8+, Vienna 2008.
Hein C., Göckel T., VDI Conf. on Building
with Innovative Materials (VDI No. 2084),
Germany 2009.
Braune A., Benter M., CO 2 – Check Lifecycle
Tower, Conf. PE International, Germany 2010.
37
Monitoring geothermal piles
at Keble College, Oxford
Authors: Duncan Nicholson, David Whitaker, Natasha Kefford
38
Geothermal pile systems
are quite a new technology
in the UK. There is a lack
of longterm data on the
sustainability and performance
of these systems.
Abstract
This study describes the monitoring of a
geothermal pile system installed below the
Sloane Robinson Building, Keble College,
Oxford. The ground loops were installed in
both the secant wall and the internal load
bearing piles, and connected to a heat pump.
The design provides 45kW of heating and
cooling capacity with annual loads of 74MWh
and 55MWh respectively.
Fig 1. Completed Sloane Robinson building
This collaborative research project has been
undertaken with funding from the South East
England Development Agency (SEEDA). The
objective was to gather and interpret data from
the Building Management System (BMS) to
assess the long term performance of the
building and compare it with the design
predictions and specifications.
Introduction
Between 2002 and 2006 Arup lead a UK
Department of Trade and Industry (DTI) project
into the ground storage of building energy.
This work highlighted the lack of case histories
detailing the performance of ground energy
systems in England. To address this, an
agreement was reached between Arup,
(SEEDA), and Cementation Foundations who
were members of the DTI research project.
This was to undertake geothermal pile
monitoring at the Sloane Robinson building,
Keble College, Oxford.
The results from the first annual report indicate
that the building heating cycle has performed as
designed and the geothermal pile system has
not exceeded the stipulated temperature limits.
Geothermal pile system
The geothermal pile system at Keble College
was completed and commissioned in 2001,
and comprised:
However, the building cooling cycle differs from
that predicted in the design. The geothermal
pile system has been asked to deliver less of
the daily peak summer load than designed due
to the installation of additional cooling systems.
The geothermal pile system has, however,
delivered the higher than expected cooling
loads during spring and autumn.
For the period analysed (March 2007 to March
2008), the geothermal pile system appears to
have caused the temperature of the ground
to increase by 2°C. If this trend continues,
there may be a drop in efficiency during
cooling as the system will no longer be able
to operate in direct cooling mode, ie the ground
temperature will no longer be less than 19°C.
• 15 load bearing piles, dia. 750mm, length 12.5m
• 14 load bearing piles, dia. 600mm, length 7.5m
• 6
1 secant wall piles, dia. 450mm, embedment
length 5m, (see image left).
Within these piles, a total of 41 ground loops
were installed. Each ground loop comprised
150m of 20mm diameter plastic tubing
connected to a series of geothermal piles.
The system uses a heat pump to heat and cool
the Sloane Robinson building. The design is
based on a peak heating or cooling capacity of
45kW, and annual heating and cooling loads of
74MWh and 55MWh respectively. About 80% of
the cooling load is supplied by direct cooling.
Although the building loads are unbalanced,
ie the heating load exceeds the cooling load,
the actual loads placed on the ground are
approximately in balance depending on the
actual annual climatic conditions.
The heat pump system was designed to
function with ground loop temperatures of
27°C (max summer) and 1°C (max winter).
Fig 2. Geothermal pile reinforcement cage
In Fig 2. this shows an geothermal pile
reinforcement cage. The geothermal pile
system was designed by Enercret and installed
by Cementation. The completed building is
shown in Fig 1.
The research project
The objective of the research project is to assess
the performance of the geothermal pile system
at Keble, using data gathered from the Building
Management System (BMS).
The proposal of work developed by SEEDA and
Arup in 2007 was to:
• collect and review data collected by the BMS
• analyse the data to assess the performance
of the geothermal pile system
• specify additional BMS instrumentation or
BMS upgrades, if required
• carry out further monitoring, up to a period
of three years
• annual reporting
39
Methodology
Temperature data for 2007/2008
There are 52 monitoring points which feed data
to the BMS, with data saved from February 2007
onwards. The data used in this project are:
26
21
• fluid temperature in the pipes returning from
the ground
16
• heat pump flow rate (ground loop side)
• heat pump flow rate (building side)
• building external air temperature
Temperature (˚C)
• fluid temperature in pipes leaving the heat pump
11
6
An initial assessment of the operation of the
geothermal pile system, between March 2007
and March 2008, was made by comparing the
temperatures of the circulating fluid as it flows to
and returns from the ground, and the average
daily outside air temperature.
Return from the ground
Average daily outside air
(II) T
his is calculated by the formula Q
(energy) = flow x delta T x specific heat
capacity of water, using the specific heat
capacity of water (4.182kJ/kg°C)
(III) U
sing the heat pump coefficient of
performance (COP) of 4.4
(IV) A
s the monitoring period is a quarter of
an hour the total needs to be divided by
four to calculate equivalent kWh
For periods of direct cooling the heat pump
is switched off. Therefore its influence is
excluded from the calculation.
To check if the system is performing as planned,
annual energy abstracted from and rejected to
the ground, were summed to estimate the
annual balance. The return temperatures from
the ground over one year were graphed.
Results and discussion
Temperature trends
The temperature data in Fig 3. presents
February 2007 to February 2008 and shows:
• periods during which heating predominates,
when the temperature of the fluid returned from
the ground exceeds that sent to the ground
• periods during which cooling predominates,
when the temperature of the fluid returned
from the ground is lower than that sent to
the ground
40
Feb 08
Jan 08
16
Geo heat exchange flow
Geo heat exchange return
Outside air
20
14
19 Mar
17 Mar
15 Mar
13 Mar
11 Mar
09 Mar
27 Feb
29 Dec
24 Dec
0
19 Dec
-10
14 Dec
4
09 Dec
-5
07 Mar
6
05 Mar
0
8
03 Mar
5
10
01 Mar
Temperature (˚C)
Temperature (˚C)
Geo heat exchange flow
Geo heat exchange return
Outside air
12
10
04 Dec
(I) D
ifference between temperature to and
from the heat pump. A negative temperature
signifies that the system is in heating mode
Dec 07
Intermittent heating mode
Continuous heating mode
25
15
3.58 l/s
-2.3°C (I)
34.4kW (II)
44.5kW (III)
11.1kWh (IV)
Nov 07
Fig 3. Temperature data for 2007/2008
As an example, heating during March 1st,
2007, 5am:
Flow rate: Delta T: Energy from the ground:
Energy to building:
Equivalent kWh:
Oct 07
Sep 07
Aug 07
Fig 4: Temperature data for 2007/2008.
Jul 07
Jun 07
May 07
Apr 07
Mar 07
-4
Feb 07
As a check on both the original design and the
way in which the building is operated, the design
parameters for the system have been compared
with BMS data.
Flow to the ground
1
Fig 4. Continuous heating mode
Fig 5. Intermittent heating mode
• brief spikes in the temperature profile
coinciding with periods during which the
geothermal pile system is not being operated
By contrast, Fig 5. shows the temperatures from
1st to 18th March during which the system was
providing heat intermittently. Again, the outside
air temperature was below 10°C for most of the
period. However, there must have been solar
gain and/or internal gains to the building
because the system was not providing
continuous heating, and the ground temperature
was stable (approximately 8°C).
Building Loads: heating mode
A comparison of the design and operational
heating energy was performed, Table 1.
The data shows an extremely good match
between the design and actual heating loads.
This is evidence that both the building and the
heat pump are functioning as designed.
The system was designed to operate most
efficiently with minimum average winter inflow/
outflow temperatures greater than 1°C. The
actual minimum ground temperature (7.36°C)
was much higher than expected, which implies
that the heat pump is operating very efficiently
in the heating cycle.The heating profile of the
system can be further split into:
• continuous building heating mode, Fig 4.
• intermittent building heating mode, Fig 5.
When the outside air temperature is
consistently below 10°C, and there is no solar
gain to the building, the geothermal pile system
is constantly providing heating to the building.
During this period, the temperature rejection to
the ground is >2°C lower than the return flow
from the ground, Fig 4. These conditions
occurred between 6th and 26th December
when ground return temperature dropped
from 13.8°C to 8.8°C.
The system is designed to maintain a
temperature difference between flow to and
from the ground of 2°C. There are also intervals
during which the temperature of the water
returned to the ground loop was higher than
that of the ground. This could have been due to
either the system providing cooling or the
system being off and the temperature of the
water next to the sensor rising towards room
temperature. The latter is more plausible.
Building loads: cooling mode
A comparison of the design and actual cooling
loads was performed, see Table 2. The monthly
cooling design loads exceed and show a
different monthly distribution to the recorded
figures. The figures differ as follows:
• the peak cooling design loads (July, August)
are much higher than the recorded loads
• the building requires cooling in more
months than predicted
When the system was originally installed,
overheating in the summer was reported, and
additional cooling systems were subsequently
installed. In peak summer periods, the additional
Ground return temperature
30
28
30
26
Design cooling
energy (kWh/
month)
Recorded cooling
energy (kWh/
month 2007)
10,285
10,422
98
1,707
6,545
3,788
April
851
3,234
May
1,870
2,335
May
3,925
5,043
October
2,550
4,378
June
10,566
9,530
November
8,585
9,006
July
20,902
8,842
December
13,600
13,194
August
15,505
7,856
Total
43,435
43,123
September
2,747
5,968
589
3,177
0
2,368
March
Table 1. Design parameters verses calculated delivery
from the heat pump (heating mode).
October
cooling systems are operating and there is
therefore a smaller cooling demand placed on the
geothermal pile system. The additional months of
cooling (November and December) reflect either a
different usage of the building than originally
planned, or a warmer winter than expected.
December
At the start of the cooling period the temperature
differential between system inflow and outflow
(ΔT) is approximately 2.5°C, and the ground
loop temperature is 15°C. Therefore the system
is operating efficiently. The temperature of the
ground loop increases during the cooling period
and ΔT becomes <2.5°C. To maintain the same
level of cooling the flow rates of the system have
risen and the efficiency will have dropped.
This trend continues throughout the summer,
Fig 7. and (ΔT) starts at 2.5°C and ends at
<1°C. Significant temperature spikes represent
periods where the system is not operational and
November
12/04/08
22/02/08
03/01/08
14/11/07
25/09/07
06/08/07
Fig 8. Ground return temperature
April
The cooling provided by the geothermal pile
system operates most efficiently at the beginning
of the summer. Fig 8. shows a period of
continuous cooling in early summer.
A
10
17/06/07
Sep 07
Aug 07
Jul 07
Jun 07
May 07
Apr 07
27 Jun
25 Jun
23 Jun
21 Jun
17 Jun
15 Jun
13 Jun
11 Jun
09 Jun
07 Jun
19 Jun
Recorded heating
energy (kWh/
month), 2007
The geothermal pile system was designed to
provide direct cooling when ground supply
temperature is <19°C. This was designed to
meet up to 80% of the building cooling demand.
The operational data shows that 77% of cooling
is provided directly. However, in the summer, the
geothermal pile system is no longer meeting the
total cooling demands for the building.
Therefore, this is not the equivalent of 80% of
the building cooling demands.
B
Temperature
Fig 7. Summer cooling, rise in ground return
temperature, reduction in delta T
The maximum average design summer system
temperature was 27°C, whereas that recorded
was 25.88°C. Thus the geothermal pile cooling
system is performing within system limits.
15
0
6
Total
20
5
Flow to the ground
Return from the ground
8
Fig 6: Continuous direct cooling mode
March
14
10
0
Design heating
energy (kWh/
month)
16
12
Geo heat exchange flow
Geo heat exchange return
Outside air
5
18
128/04/07
10
20
09/03/07
15
22
18/01/07
20
25
24
Temperature (Degrees ˚C)
Fluid Temperature (˚C)
Temperature (˚C)
25
0
1,219
55,183
48,944
Table 2: Design parameters verses calculated delivery
from the heat pump (cooling mode)
the temperature of the feed pipes reaches room
temperature. At the end of the summer the
temperature from the ground approaches 19°C
(the design limit for direct cooling).
Ground temperature
The annual energy balance between the system
and the ground can be calculated by summing
the energy abstracted and rejected to the ground.
Between March 2007 to March 2008, the system
rejected approximately 17,000kWh to the ground.
It could thus be inferred that the ground
temperature will have risen over this period.
The return temperature from the ground rises in
the summer and falls in the winter, Fig 8. as
would be expected. However, this data suggests
that over a period of one year, the temperature
of the ground has risen by approximately 2°C
(from 11°C (A) at the start of the monitoring period
to 13°C (B) at the end). Without additional data it
is not possible to say whether this is a long term
trend or the result of an unseasonably warm year.
However, the temperature data coupled with the
fact that the system rejected 17,000kWh to the
ground during this period implies that the system
may be causing the increase in ground
temperature.
Conclusions and next steps
The BMS at Sloane Robinson building, Keble
College, Oxford has provided reliable monitoring
data from March 2007 to March 2008. Longer
term records could not be retrieved. Reviewing
BMS records provides a simple method of
assessing the geothermal pile system.
The geothermal pile system started operation in
2001. The mean ground temperature between
March 2007 and March 2008 is similar to the
expected ambient temperature of 12°C.
This implies a reasonable balance between
annual heating and cooling loads. The return
temperatures remained within the temperature
limits of 27°C and 1°C. This indicates that the
heat exchange is large enough to handle the
loads sustainably.
The building’s heating energy demand from the
BMS data is similar to the design predictions.
Between March 2007 and March 2008, the
geothermal pile system appears to have caused
the ground temperature to increase by 2°C.
This figure compares the difference between
the average ground return temperatures in
February 2007 and February 2008. However
this appears to be a function of the climate for
that particular year.
The building’s cooling cycle has not performed
as designed. The system has met less of the
peak summer loads than expected due to the
installation of additional cooling systems.
Conversely, the system has met the higher than
expected cooling loads during spring and autumn.
Recommendations
The current monitoring research project should
be continued. The ground return temperatures
should be monitored to understand whether
there is a steady annual increase in temperature.
The building should be analysed to understand
why the cooling loads are different to those that
were expected. The supplementary cooling
system should be assessed. An electricity meter
should be installed on the heat pump to better
understand the performance of the heat pump
and the running costs of the system.
Acknowledgements
We would like to thank SEEDA, Keble College,
University of Oxford, Cementation Foundations
Contractor (contruction contractor), Pillinger
(monitoring/maintenance contractor). Ryan Law
Geothermal Engineering Ltd
References
Cementation Foundations Skanska, Geothermal
Piles used at Keble College, Oxford. The
Structural Engineer, 2004, 19.
41
Monitoring geothermal piles at Keble College, Oxford
Summer cooling, rise in ground return
temperature, reduction in delta T
Continuous direct cooling mode
Sustainable and quake resistant
façade for existing buildings
Authors: Susumu Matsunobu, Yutaka Misawa, Piet Lycke
42
(A) Integrated façade Configuration
Abstract
This research challenges the customary
reinforcement methods which only focus on the
structural performance. Instead it proposes an
integrated-façade-system which not only consists
of the necessary seismic structural retrofit, but
also significantly improves the environmental
and architectural quality of the existing building.
A Buckling Restrained Brace (BRB) to increase
the seismic resistance was developed with
minimum dimensions and in such a way that it
can be perfectly integrated and harmonised with
a façade louver system.
The louver system makes it possible to control
and optimise the heat load and the daylight
entering the building at the perimeter. As big
earthquakes are mostly accompanied by fire,
additional research has been done on the
behaviour of the fire and its flames in this
particular case. Furthermore, deploying the
system on the outside of the building also
ensures continuity in its usage.
Initially the system has been singled out to be
used for the main gathering points in the cities’
disaster evacuation plans such as in schools.
However due to its flexibility it could easily be
extended to other facilities in the future such as
hospitals.
Introduction
Major cities in Japan have had a long standing
reputation for wanting to have the newest and
most modern buildings. This has been intrinsically
linked to repeatedly demolishing and replacing
the existing infrastructures with elaborate new
ones. This short life span of buildings in Japan
is remarkable and in the current climate of
sustainable design, promoting the continued
usage of buildings through improvement of their
functionality is key. This will be an important
factor in the process of decreasing the usage of
resource materials and reducing the emission
of CO 2, both associated with the construction
process and the operation of the building.
Meanwhile, Japan’s history has been
characterised by several severe earthquakes.
The most recent one in Kobe in 1995 (magnitude
7.3 on the moment magnitude scale) caused
6,434 deaths and 43,792 injuries. It was
demonstrated that buildings from before the
Louvers,
Double skin etc
+
=
Structure
bracing
member
Structure
Integrated façade
Integrated façade on building
+
Plasticity effect
=
Seismic load
Collapse
Seismic load
(B) Integrated façade: Structural function
Seismic load
The severe earthquake in Kobe,
Japan, 1995, demonstrated that
the city’s infrastructure, designed
before the change of Japan’s
building regulations in 1981,
failed to secure the safety of its
inhabitants. This study focuses
on how to reinforce existing
structures sustainably.
Deformation
Deformation
Energy
Absorption
No damage
to main
structure
Reduce
maximum
deformation
Deformation
(C) Integrated façade: Enviornmental Function
Solar radiation
(Summer)
Exhaust heat
Solar
radiation
(Winter)
Louver
Structural bracing
member
Solar radiation
(Summer)
Solar
radiation
(Winter)
External louver
Glass
External wall
Structural bracing
member
Reflection
Fresh air
Fig 1. Functions of the integrated-façade-system
change of Japan’s building regulations in 1981
were unable to resist a major quake. Considering
the fact that about 50% of the current building
infrastructure in Japan dates from before 1981,
a significant amount of buildings are in need of a
seismic resistance structural retrofit to meet the
current regulations and to reduce the amount of
casualties in the event of a major disaster. In this
way, it is clear that the renovation of buildings
will become influential over the current
scrap-and-build approach.
This research is challenging the widely
spread retrofit methods which focus purely
on the structural aspect and proposes an
integrated-façade-system. This integrates
façade engineering techniques in the fields of
structural engineering, environmental and
architectural design and adds extra value to the
reinforcement of the existing buildings.
The functions aimed for in this integratedfaçade-system are shown in Fig 1. The structural
function seeks to reduce the maximum response
displacement and to improve the load-bearing
capacity through energy absorption by a bracing
system. The environmental function seeks to
reduce the seasonal energy input through the
addition of control mechanisms that block direct
sunlight during summer and effectively use the
light in winter. As the deployment of this
integrated-façade-system on the outside of the
building ensures continuity of its use, the system
can be designed in such a way that it is also
visually aesthetic. The implementation of the
system therefore demonstrates potential.
Methodology
Evaluation axes for integrated-façade-systems
In order to judge the effectiveness of the
proposed integrated-façade-system, evaluation
axes were established in accordance to the
fundamentally required functions mentioned
above: ‘Design’-axis, ‘Shelter function’-axis and
‘Environment’-axis. More detailed requirements
concerning the individual axes have been
defined as shown in Fig 2. As the louvers
become a not merely functional element in the
integrated-façade-system, but also can satisfy
aesthetic demands, a ‘cost’-axis has been
introduced, intersecting with each axis.
Studied integrated-façade-systems
The variety of integrated-façade-systems is
dependant on the diversity in louver design, the
structural bracing member and their possible
combinations.
An overview of existing louver types has been
created based on a bibliographic survey. Apart
from the existing types, a new bound louver type
has been proposed, Fig 3. This type has
openings of varying size and shows, with its
unique design, almost unlimited possibilities for
the future.
As structural bracing member, a Buckling
Restrained Brace (BRB) is used to increase the
seismic resistance. The BRB is a patented
product by Arup Japan and Kanagawa University
and was developed with minimum dimensions
and in such a way that it can be perfectly
integrated and harmonised with the louver
system. Considering one floor, one span and
only one possible intermediate node, some of
the possible BRB arrangements are shown in
Fig 4. Expanding the method to multiple floors
43
Construction level
Integrated façade system
Evaluation axis level
Design axis
Shelter func. axis
Environment axis
Design element level
Design
Response control
Daylight factor
Compatibility
Load bearing cap.
Daylight use
Creativity
Rigidity
Maintenance
Fig 3. Louver types and bound louver type
Cost
Fig 2. Evaluation axes of the integrated-façade-system
and eliminating the options which have no clear
structural benefit leaves us with five considerable
arrangements.
When combining louvers and BRB, it is seen that
only in the case of diagonal, oblique lattice or
bound louvers the BRB can be integrated into
the louvers. In the other cases the systems are
separated. Focusing on the combined systems
(diagonal, oblique lattice and bound louvers),
one separated system (horizontal louvers) and
excluding the combination of diamond BRB
arrangement with bound louvers, with results
as shown in Fig 5.
Design axis
Using the earlier defined evaluation criteria for
the design axis – compatibility with existing
structure, aesthetic appeal and creativity – a
computer graphics questionnaire, was held
amongst 50 randomly selected architecture
students to evaluate the considered louver-BRB
designs. Construction 1 scored highest in
compatibility, while construction 8 had the best
aesthetic appeal and constructions 11, 12 and
13 were seen as most creative. Construction 1,
2 and 3 scored notably low in creativity as they
are the least innovative.
Shelter function axis
To evaluate the structural axis, the BRB
arrangements are applied to a hypothetical
four-storey building from 1971. This is before the
revision of Japan’s reinforced concrete calculation
standard. A push-over analysis is carried out,
taking the presence or absence of partition walls
and plasticity or non-plasticity of the beams as its
parameters and shows that the structure has the
lowest strength index when no partition walls and
the plasticity of the beams were considered.
Based on this result, the required quantity of
reinforcement is defined to comply with the
current regulations, this is for a seismic index
Is=0.7 when the toughness index is at a
maximum of F=1.05 (inter-storey drift
approximately 1/250). When reinforcing the
structure by means of H-sections
(250x250x9x14mm) used in the strong axis
direction, satisfying results are reached for X,
V, N or diamond BRB arrangements. However,
in the case of diagonal or bound arrangement,
the requirements are not met. When replacing
the H-sections by rectangular steel sections
of 250x250x12mm, 350x250x11mm,
450x250x12mm and 500x250x14mm (which
respectively approximately have the same, two,
four, and six times the geometrical moment of
inertia of the original H-section), an increase in
44
Fig 4. BRB arrangement methods
1
2
3
4
5,6,7
8
9
10
11,12 ,13
Fig 5. Computer graphics of the considered louver-BRB arrangements
and achievement of the required load-bearing
capacity and rigidity can be seen as the steel
section enlarges.
mock-up test with two types of diagonal louvers
with a different pitch was done to confirm its
efficiency: a rectangular section (short side
facing the outside) and a square section
(corner facing the outside) as shown in Table 2.
Finally, so as to judge its influence, a dynamic
response analysis of the structure with the applied
reinforcement is carried out. The cases with
H-sections or rectangular sections 250x250x12mm
or 500x250x14mm reinforcements were considered
and El Centro-NS, Taft-EW and Hachinohe-NS
seismic waves, were introduced.
Over a two-day period and at an interval of
one hour from 09:00 to 15:00, two types of
measurements were taken: the illuminance was
registered by an illuminometer and a photograph
was taken using a fish-eye lens.
The result in Fig 6. shows that the inter-storey
drift is smaller than 1/250 for the cases where
the BRB is arranged in X,V,N and diamond
shapes. In the case of diagonal arrangement
with H-sections the inter-storey drift exceeds
1/250, however, when rectangular sections are
applied, values smaller than 1/250 become
possible as the section size increases.
We can conclude that the installation of the louvers
has no negative influence on the amount of
daylight in the interior. On the contrary, the
louvers are diffusing the daylight which improves
the daylight levels further away from the opening.
As the louvers are controlling the amount of direct
sunlight entering at the perimeter, visual discomfort
of the building users will also be reduced.
Environment axis
The intergrated-façade-system has different
levels of enviornmental functionality. A first
objective is to reduce the heat load introduced
in the interior by reflecting the solar radiation.
According to the orientation and the solar
altitude in a particular situation, optimal heat
shielding can be obtained by adjusting the
shape and pitch of the louvers.
Cost axis
The cost is evaluated on the quantity of
components used in the BRB-louver
arrangements and is for one floor span.
At the same time, louvers can also serve to
control the amount of daylight entering the
perimeter. As in many cases the installation of
louvers is rather seen to have a negative effect
on the brightness of the space and the view to
the exterior, a full-scale (one floor, one span)
First, the amount of steel in the columns and
beams according to the BRB arrangements is
considered. For X, V, N or diamond shape, the
steel quantity is equal, but when rectangular
steel sections are used (inter-storey drift smaller
than 1/250) it increases considerably. As the
manufacturing and transportation costs rise
together with the rise in material quantity, it could
be assumed that the cost when using rectangular
sections is higher. However, this difference needs
to be handled with care as the study does not
Taft EW wave
5
1/150
1/250
1/150
No reinforcement
2
4
RF floor
3
3
2
0
10
20
KEY:
X shape
30
V shape
40
N shape
1/150
No reinforcement
4
RF floor
RF floor
1/250
No reinforcement
4
1
Hachinohe NS wave
5
1/250
1
50mm
Diamond
3
2
0
10
Diagonal (H-250)
20
30
Diagonal ( -250)
40
1
50mm
0
10
20
30
40
50mm
Diagonal ( -500)
Fig 6. Dynamic response analysis results
15
10
Environment
axis
Cost axis
Does not
apply
4500
BRB yield not at the same time
Not
sufficient
Rigidity
All BRB yield at the same time
Daylight
Factor
Average value of daylight factor Minimum value of daylight factor
is under 10% at 450mm
is under 10% at 450mm
Does not
apply
Daylight use
Average value of daylight factor
at 2250mm increases when
installation of the louvers
Does not
apply
Minimum daylight factor at
2250mm increases when
installation of the louvers
Maintenance
Linked with the louver area of the cost axis
3/ (steel quantity of each studied construction/smallest steel quantity
of all studied constructions)
Louver
surface area
3/ (Louver surface area of each studied construction/smallest louver
surface area of all studied constructions)
Table 1. Rating standards for the considered BRB-louver arrangements
Results and discussion
Based on the results of the investigations of
design, shelter function, environment and cost
axes, a quantitative evaluation of the considered
BRB louver arrangements is attempted.
Rating standards, as shown in Table 1. are
defined for each criterion and comparatively stable
performances are discovered in the evaluation of
the studied arrangements. Whilst the option with
horizontal louvers gains the highest evaluation,
all considered arrangements are verified as being
realistic and potentially applicable options.
Additional research: fire safety
Severe earthquakes are usually accompanied
with the outbreak of fire. As the integratedfaçade-system is seen to be applied to major
gathering points in the cities’ disaster evacuation
plans in the first instance, it is important to know
the behaviour of the fire to ensure the safety of
the building’s users and to prevent the fire from
spreading to upper floors and adjacent buildings.
No research had been done on the influence of
external louvers on the way the fire spreads. and
a full scale mock-up test was completed to
ensure safety of the proposed system.
b) Cvalue
type
Max./min.
Average value
Direct daylight factor
15
10
5
0
900
0
2700
Measurement point (mm)
4500
Does not
yield
Quantity
of steel
include the amount of steel required for (transverse)
web stiffeners in the case of H-sections; this
becomes more critical as the load-bearing capacity
of the structure increases. Next, the total surface
area of the louvers in one floor, one span is
considered as an increase in surface area is
connected to an increase in maintenance cost in
addition to the material cost. It could be assessed
that this value is the smallest for diagonal louvers.
20
The experiment was done for several different
distances between the louvers and the opening.
The behaviour of the ejected plume from the
opening was verified through temperature and
radiated heat measurement around the opening
and the louvers and through study of the shape
of the plume.
As neither the installation of the louvers nor their
distance to the opening had an effect on the
temperature on the central axis of the opening,
it is asserted that the ejected plume separates
two main directions due to their presence.
Furthermore, it can be considered that, after
ejecting from the opening, the plume will rise in
accordance with the inclination of the diagonal
louvers. At last, the thermo-camera confirmed
that the shape of the plume is depending on the
distance of the louvers to the opening and that
the point of that shape-transformation is at
about 600mm-800mm from the opening.
Conclusion and next steps
The undertaken research on the proposed
integrated-façade-system shows that it is more
than a static unique idea which consists of
combining a necessary seismic retrofit of
existing buildings with measures to significantly
improve its environmental and architectural
quality. On the contrary, it is a dynamic, flexible
system which is full of possibilities and has the
potential to be applied to a wide range of
buildings and to anticipate on what the future,
with ever growing requirements towards energy
conservation and sustainability, will bring.
100 mm
250 mm
100 mm
BRB:
Buckling restrained brace
100 mm
5
Shelter
Response
Max. response; Interstorey drift Max. response; Interstorey
0
0 drift 1/150 or less
Control
1/250 or less
function.
900
2700
4500
900
2700
0
0
axis
Load
bearingpointSufficient
load bearing
Measurement
(mm)
Measurement point (mm)
Capacity
reinforcement
a) R type
350 mm
Creativity
2
20
C type
25
Max./min. value
Average value
1
Direct daylight factor
100 mm
Questionnaire
Compatibility
R type
Score
350 mm
5
Design
25
Daylight factor (%)
Exterior
10
Max./min. value
Average value
3
Direct daylight factor
Louver section
Daylight factor (%)
Design
15
axis
Without louver
Evaluation
criteria
Daylight factor (%)
25
Evaluation
20 axis
BRB:
Bucklingrestrained brace
Table 2. Mock-up used for illuminance measurements
Acknowledgements
This research was carried out using a Ministry
of Land, Infrastructure, Transport and Tourism
Grant; We would like to thank Prof. Iwata and
Prof. Iwamoto of Kanagawa University.
References
Hikone Shigeru, Misawa Yutaka, Makoto
Nakamura, Iwamoto Shizuo, Iwata Mamoru:
Lighting environment of diagonally arranged
louver on integrated façade system, Journal
of Architectural Institute of Japan, No.644,
2009, 1187-1193.
Makoto Nakamura, Hikone Shigeru, Misawa
Yutaka, Iwamoto Shizuo, Iwata Mamoru: The
integrated façade consists of louvers and
buckling-restrained-braces as a building system,
Journal of Architectural Institute of Japan,
No.647, 2010, 121-129.
Misawa Yutaka, Hikone Shigeru, Aburano
Kenji, Omiya Yoshifumi, Iwamoto Shizuo, Iwata
Mamoru: Fire experiment on diagonally arranged
external louver for integrated façade system,
Journal of Architectural Institute of Japan, 2010.
Fire and Disaster Management Agency:
Great Hanshin Awaji Earthquake 2006. Ministry
of Land, Infrastructure, Transport and Tourism.
Kaneki Yohei, Hikone Shigeru, Yamashita
Tetsuo, Iwata Mamoru: Seismic Strengthening
by the buckling restrained braces arranged
diagonally, Journal of structural and construction
engineering, 73 (634), 2008, 2215-2222.
Japan Construction Disaster Prevention Society:
2001 Revised Edition Seismic Retrofit Design
Guidelines and Explanation for Existing Reinforced
Concrete Construction Buildings, 2001.
The Building Centre of Japan: Structure design
guideline for high-rise building, 2002.
45
Sustainable and quake resistant façade for existing buildings
EL centro NS wave
5
Measuring change of coastal defence
structures using advanced 3D laser
mapping techniques
Authors: Ilse Steyl, Dean Crowley, Patrick Kuhn and Simon Bray
46
Main Rivers
The challenge to effectively
manage coastal assets to
encompass changing use, sea
level rise, coastal squeeze and
development pressures, can be
great. Monitoring these assets
is central to this management
process. Using advanced
technology could assist in
time and cost savings.
Study Area
Urban Areas
Groynes
Lymington
New Milton
Highcliffe
Groynes
Bournemouth
nt
ole
S
he
Milford on Sea
T
Keyhaven
Christchurch Bay
Abstract
Technological evolution and the increased
development of remote sensing technology
provide an opportunity to rapidly acquire data
and accurately measure changes over wide
spatial areas. The monitoring of ecologically and
economically valuable coastlines and inter-tidal
zones is difficult, mainly due to poor access,
large extent and tidal restrictions. This can result
in reactive rather than proactive management.
In this study the use of terrestrial Light Detection
and Ranging (LIDAR) equipment, as a cost
effective and accurate tool for monitoring
coastal defence structures and measuring
change, is discussed. The survey location was
at Highcliffe in Christchurch Bay along the south
coast of Britain, where rock armour groynes
have been constructed to protect the coast
against erosion. Recorded information
highlighting the impacts of erosion in the area
has been recorded since the 1700s.
LIDAR surveys of three of the ten rock armour
groynes were undertaken on three occasions
during low tide: 22 October 2008, 19 February
2009 and 9 September 2009. Data were
processed using remote sensing and Geographic
Information System (GIS) software. Areas of
erosion and deposition between the three
survey dates were calculated. The research
demonstrated the value of using terrestrial laser
scanning to collect detailed data for small area
monitoring of coastal defences. The relative
ease and speed of collecting data can allow
for frequent return visits to monitor structures
and their integrity.
Totland
0 0.5 1
2
3
Isle of Wight
4 kilometres
Fig 1. Study Area, Christchurch Bay, Hampshire
Introduction
Economically and ecologically valuable
coastlines and intertidal zones are difficult to
monitor. This is mainly due to poor access,
large extent and tidal restrictions. As a result,
management can be piecemeal and reactive
rather than proactive. It is a dynamic challenge
to effectively manage coastal assets to
encompass changing use, sea level rise, coastal
squeeze and development pressures. Access to
cutting edge techniques enabling improved
understanding of coastal processes and
mechanisms can be very valuable.
Typically topographical and ecological survey
methods have depended on field observations,
or more recently airborne remote-sensing
techniques. Technological evolution has led to
remote survey systems becoming faster and
more affordable. This, along with the compact
nature of modern systems, has paved the way
for the further development of Terrestrial LIDAR
Scanning (TLS) possibilities. TLS allows rapid
data acquisition and precise measurements over
a wide spatial scale at relatively low costs.
TLS has been used in the urban environment for
examining building dimensions or deformation
monitoring of constructions for example. Its use
has been extended to the natural environment and
is now also commonly used in surveys as diverse
as archaeology, landslide dynamics, monitoring
of volcanic activity and fluvial morphodynamics.
However, TLS has not yet been extensively used
in coastal surveying, although Nagihara et al (2004)
employed it for mapping sand dune morphology,
whilst Rosser et al. (2005) used the technology
to monitor cliff erosion.
Shoreline Management Plans and coastal
strategy studies have consistently identified the
need for development of coastal management.
The use of LIDAR in monitoring programmes is
listed as one of the objectives of the Strategic
Coastal Monitoring Programme for the South
East Region.
This research was divided into two separate
phases, each focusing on a distinct environment,
measuring specific facets. The first phase of the
research examines:
• The use of terrestrial LIDAR as a cost
effective and accurate tool for monitoring the
change of coastal defence structures:
analysing design and monitoring to enhance
proactive management
• Using terrestrial LIDAR to map intertidal algal
community distribution
Monitoring coastal defence structures
The focus of this study is on the mapping of
coastal defence structures and change
measurement. The survey location is at Highcliffe
on Christchurch Bay along the south coast of
Britain, Fig 1. A number of rock armour groynes
have been constructed to mitigate erosion along
the coastline.
The geology of the area is dominated by the
Barton beds and the overlying Headon Hill
Formation (of the Solent Group), all of which
form part of the Hampshire Basin and deposited
between approximately 42.1-35.4 million years
ago. The whole of the cliff face fronting
Christchurch Bay is a Site of Special Scientific
Interest (SSSI). The cliff face exposes the only
complete late-middle to early-late Eocene
sequence in the world.
Records of erosion at Highcliffe date from the
early 1700s. Coastal defences in the form of
better drainage were constructed during the
mid-1800s, but these were eventually destroyed.
Further work was undertaken in the beginning of
the 1900s and by the 1960s a cliff stabilisation
scheme was completed as well as timber groynes
constructed to enhance the protection of the
coast by absorbing wave energy. By 1991, the
timber groynes were replaced with alternative
long and short rock armour groynes, see Fig 2.
47
Fig 2a. Rock armour groynes at Highcliffe
Fig 2b. Rock armour groynes at Highcliffe
Fig 4a. Example of groyne after cleaning
Fig 4b. Example of groyne after cleaning
Stabilisation work along the cliff included
improved drainage and planting of salt
tolerant grass mixes.
of the registrations during post-processing, while
providing the added benefit of shifting the scans to
their correct world coordinates. The survey control
was undertaken in conjunction with a qualified
surveyor, who surveyed strategically placed
reflectors, at the same time as they were captured
by the laser scanner. Due to time constraints, it
was not feasible to utilise survey control for all
scans, so it was limited to use on the scan
positions on the cliff above the groynes, as these
areas were not impacted by the incoming tide.
Methodology
Data collection and processing
Three of the long rock armour groynes labelled
H2, H4 and H6, were surveyed on three
separate occasions during low tide: 22 October
2008, 19 February 2009 and 9 September 2009.
Data collection was undertaken using a Riegl
LMS Z420i terrestrial laser scanner, Fig 3.
This has a range of up to a 1000m with a level of
accuracy of 100mm. Properties of scattered light
are measured to find the range to a distant
target/surface. A set of three dimensional
coordinates is generated, normally referred to
as a ‘point cloud’. Mounted on top of the laser
scanner is a Nikon D100 digital SLR camera,
allowing digital photos to be captured whilst
scanning, provide colour data for each point.
At each groyne, scanning was undertaken from
at least three different vantage points. One at
beach level east of the groyne, another at beach
level west of the groyne and one or more on the
cliff looking south onto the groyne. During some
return surveys it was necessary to scan the side
of the groyne from more than one location due
to the shifting of shingle on the beach, creating
shadow zones. This caused gaps in the data
collection that needed to be filled. At each of
these scan positions, two single scans were
performed; one a 360° overview scan at low
resolution and the other a fine scan (high
resolution) focusing on the groyne. The 360°
scans were required mainly for the purposes of
registration for the post-processing stage, while
the fine scans provided the detailed point data
on the relevant areas of interest.
Survey control was undertaken simultaneously
with the laser scanning on the first and third
visits, in an endeavour to maximise the accuracy
48
The survey scans were translated, rotated and
registered to the controls scans using I-Site
v3.1.1 software. Once all the scans were
registered together, a quality check was
undertaken to ensure no errors were made in the
registration and each scan fitted seamlessly into
place. Data were then cleaned manually in I-Site
to remove any stray points such as people,
vehicles, or rain drops, reflections from water
and glass. This included any other reflective
surface and distant points not relevant to the
study such as the upper cliffs and groynes
further along the beach. At this point the data
was filtered removing any points nearer than
25mm to any other point. This effectively thinned
the point surface near the high density ‘point
circle’ immediately around the scan position,
resulting in a more manageable file size, Fig 4.
Data was then exported as a text file (with the
associated x, y, z coordinates for each point)
and imported into ArcGIS 9.3.1. A surface model
was created, which represents the survey area
as a continuous raster surface (in this case each
cell was 25cm x 25cm). The raster surface is
generated through interpolation of the point
data. Interpolation predicts the values for cells
from a limited number of sample points.
The method used to generate the surface for the
groynes is called natural neighbour interpolation.
This method finds the closest subset of input
samples to a query point and applies weights to
them based on proportionate areas (rather than
distances) in order to interpolate a value.
Fig 3. Riegl LMS Z420i
Three surfaces were created representing the
data for each visit. To determine the areas of
change where erosion and deposition occurred
between the three time periods, geoprocessing
was undertaken using tools within the 3D
Analyst™ extension within ArcGIS.
The surface was generated through The 3D
Analyst™ and allows effective visualisation and
analysis of surface data through the application
of a cut-and-fill procedure, in which the elevation
of a surface is modified by the removal or
addition of surface material. The surface of a
specific location at two different time periods are
used to identify where surface material has been
removed, added or where no change has
occurred. Volumes and areas demonstrating the
change between each surface was generated.
Fig 6. is a representation of the changes that
occurred between the three survey dates.
A comparison was made between the first and
second surveys, the second and third surveys
and lastly between the first and third surveys.
The changes in elevation of surfaces can be
visualised by generating a profile of a specified
area. Fig 5. shows the changes in profile over
time for groyne H2.
Results and discussion
The preliminary analysis showed clearly which
areas are prone to erosion and deposition. In all
three surveys, the area with the highest build-up
of shingle was to the west of groyne number H2,
whilst the areas around the base of the groynes
were more prone to erosion. It is also interesting
to note that the area to the east of groyne
number H4 is more prone to erosion.
The highest build-up of material took place to the
west of groyne number H2 during the winter
storms between the first and second visit.
This can clearly be seen in Fig 6. which shows
the profile of groyne H2. The area between
H4 and H6 are more prone to erosion.
Profile of groyne H2 – third visit
3.0
3.0
3.0
2.5
2.5
2.5
2.0
2.0
2.0
1.5
1.0
1.5
1.0
0.5
0.5
0
0
0
10
20
30
Distance (m)
40
Elevation
3.5
50
1.5
1.0
0.5
0
0
10
20
30
Distance (m)
40
50
0
10
20
30
Distance (m)
40
50
Fig 5. Changes in profile between visits for groyne H2
0
25
50
100
150
200
N
Deposition
Erosion
Fig 6a. Erosion and accretion between first and second survey
Acknowledgements
This research was undertaken in colaboration
with University of Southampton (Dr Simon Bray).
We would like to thank Steve Woollard and
Mike Hinton from Christchurch Borough Council
and Dr Dafydd Lloyd Jones (EMU Ltd).
References
Dale W., Archaeology of West Hants:
A Natural History of Bournemouth and District,
including Archaeology, Topography, Municipal
Government, Climate, Education, Fauna, Flora
and Geology. Bournemouth Natural Science
Society, Bournemouth. Morris, D (ed), 1914.
Fig 6b. Erosion and accretion between second and third survey
Harland WB., Cox AV., Llewellyn PG., Pickton
CAG., & Walters R., A Geologis Time Scale.
Cambridge University Press, New York, 1982.
Hinton MT., The causes, effects and
mitigation strategies relating to coastal
landslides at Highcliffe and Naish Farm on
the Dorset – Hampshire border, Christchurch,
Dorset County Council, 2007.
Lichti DD., Gordon SJ., Stewart MP.,
Ground-Based Laser Scanners: Operation,
Systems and Applications, Geomatica, 2002.
Fig 6c. Erosion and accretion between first and third survey
Fig 5: representation of the changes that occurred between the three survey dates.
Fig 6. Erosion and accretion between first and third survey
During the survey period, repairs were needed to
groyne H6, which confirms the results of the
analysis. The deposition of material to the west of
H2 was also very significant and more detailed
analysis of the volumes will be undertaken.
The survey data collected for the three groynes
took approximately five to six hours on each of
the three survey days. The survey control data
was collected over a similar time frame by a
separate survey team. The cleaning and
processing of the data for each survey was
undertaken over a period of two person days,
with an additional day for survey control to be
referenced and checked.
The rapid acquisition of the data and relative
ease of providing the raw data in a format to
allow for the generation of terrain models,
geostatistical analysis and undertake
measurements, allows for cost savings and the
storage of data in an accessible format.
Monitoring of the coastal defences can therefore
be undertaken in a structured way, allowing for
rapid analysis.
Conclusion and next steps
The research demonstrated the value of using
terrestrial laser scanning to collect high detailed
data for small area monitoring of coastal
defences. The relative ease of collecting the data
and speed at which it takes place can allow for
frequent return visits to monitor structures and
their integrity. The information can be used for
routine monitoring of coastal defences and could
highlight where repair work is needed before
structures deteriorate or collapse.
The data collected for this study will be further
analysed in conjunction with data from other
sources to give a better understanding of the
vertical and horizontal changes along the
groynes. More detailed analysis of the volumes
and areas of the deposited and eroded areas
will be undertaken. This will be done through
comparison between different interpolation
techniques, for example kriging.
Maas HG., & Vosselman G.,Two algorithms
for extracting building models from raw
laser altimetry data, ISPRS Journal of
Photogrammetry & Remote Sensing 54,
1999, 153-163.
Milan DJ., Heritage GL., & Hetherington D.,
Application of a 3D laser scanner in the
assessment of erosion and deposition volumes
and channel change in a proglacial river.
Earth surface Processes & Landforms. Vol. 32,
2007, 1657-1674.
Mouginis-Mark PJ., & Garbeil H., Quality of
TOPSAR topographic data for volcanology
studies at Kilauea Volcano, Hawaii: An
assessment using airborne lidar data.
Remote Sensing of Environment. Vol. 96 (2),
2005, 149-164.
Nagihara S., Mulligan KR., & Xiong W., Use of
a three-dimensional laser scanner to digitally
capture the topography of sand dunes in high
spatial resolution. Earth Surface Processes and
Landforms, Vol. 29, 2004, 391-398.
Rosser NJ., Petley DN., Lim M.,Dunning SA.,
& Allison RJ., Terrestrial laser scanning for
monitoring the process of hard rock costal cliff
erosion. Quaterly Journal of Engineering Geology
& Hydrogeology. Vol. 38, 2005, 363-375.
Rowlands A., & Sarris A., Detection of
exposed and subsurface archaeological
remains using multi-sensor remote sensing.
Journal of Archaeological Science. Vol. 34 (5),
2007, 795-803.
Schulz W.H., Landslide susceptibility revealed by
LIDAR imagery and historical records, Seattle,
Washington. Engineering Geology. Vol. 89 (1-2),
2006, 67-87.
Sibson R., A brief description of natural
neighbour interpolation In: Interpolating
multivariate data, John Wiley & Sons,
New York, 1981.
Worthing Borough Council: Strategic Coastal
Monitoring in the South East.
49
Measuring change of coastal defence structures using advanced 3D laser mapping techniques
Profile of groyne H2 – second visit
3.5
Elevation
Elevation
Profile of groyne H2 – first visit
3.5
Digital infrastructure and changing
practices in engineering design
Authors: Jennifer Whyte, Sunila Lobo, Mark Neller, Sarah Bowden
50
New text
intro
skills are needed to
compete, as integrated software
solutions provide a digital
infrastructure for projects.
This changes the practice of
information management and
engineering design on next
generation projects.
Abstract
Integrated software is becoming used as a
digital infrastructure for the delivery of large
building and infrastructure projects. Its
introduction fundamentally changes engineering
design as project stakeholders work together
through shared sets of technologies.
In this research, the organisation and
management of design through integrated
software was investigated across three Arup
projects: Motorway 6 (M6) Toll, Channel Tunnel
Rail Link (CTRL) and SAFElink Motorway.
On each of these projects design teams have
been at the forefront of developing new ways of
using the digital infrastructure for delivery and of
adding value to the client in data handover.
The research highlights how strong in-house
technology capabilities are important to Arup in
responding to diverse client requirements and
system configurations on projects and in
contributing to innovation on these projects.
It identifies new skills that are needed within the
industry as integrated software solutions provide
a digital infrastructure for project delivery and
draws out lessons for information management
on the next generation of projects.
Introduction
Capabilities in using and developing integrated
software are increasingly central to the delivery
of major building and infrastructure projects.
Such tools open up possibilities to improve
design processes by collating and making
available data for better decision-making.
Their introduction is part of a broad move
towards advanced manufacturing in which the
opportunities for technological innovation shift.
In construction, integrated software involves
a centralised repository of data; standard
methods for using that data and the data itself,
which often takes the form of a 3D Computer
Aided Design (CAD) model, with associated
attributes. Studies into the use of such digital
representations have described them as
propagating waves of innovations across
project supply chains.
Fig 1. The three projects: M6 Toll; SAFElink; and the Channel Tunnel Rail Link (CTRL)
The changes brought about by integrated
software are having profound impacts on the
way that design and construction work is
organised. On one of the projects studied in this
research, rainfall data provided an example of
how the digital infrastructure changes the
visibility of information across firm boundaries.
There was an issue with rain on the construction
site. It had been raining heavily and given the
geology of the area this was turning ponds
orange in a neighbouring golf club that was
about to host a major competition.
To design temporary holding ponds, an engineer
seconded from one part of the project to another
needed to know how much it rained. However,
as the license for this data had been agreed at
firm level then the information could not be
released. To fix the problem the rainfall had to
be recalculated. While firm-level ownership of
information is normal practice, the integration of
the design data through an integrated software
solution strengthened the sense of the project
being at the relevant organisational level for such
a purchase. Access to a project-wide dataset
made this project-level more visible in engineers’
day-to-day work.
Hence, in the research community, attention is
shifting from a focus on the technical challenges
of integrated software to organisational
questions. Return on investment from the use of
building information models is beginning to be
analysed, with 70% of owners reporting positive
returns in a recent survey.
At a project level, integrated software solutions
have become a strategic rather than purely
operational matter and by focusing on strategic
decisions about IT use on projects the research
community is responding to and supporting this
shift. At an industry level, the growing maturity of
the technologies for supporting integrated
working has lead to initiatives that bring
researchers and practitioners together. This
supports standardisation of information
management processes, both in the UK and
internationally. As the example of rainfall data
suggests, by increasing the available
information, new software has both intended and
emergent properties. They get used alongside
older work practices and recombined in new
forms of practice.
While digital technologies are transforming the
delivery of both buildings and infrastructure,
most research has paid attention to their use in
building design. Here, researchers detail how
‘an accurate virtual model of a building is
constructed digitally’, providing a building
information model. Less attention has been paid
to the benefits and challenges of implementing
such an approach in the infrastructure sector.
Yet there are some particular challenges in
infrastructure, which involve global supply
chains, multiple stakeholders and significant
organisational complexity in project delivery.
51
The aim of this study is to develop a practical
understanding of how digital modelling,
collaboration and project management tools
support the delivery of global infrastructure
projects. Hence the research seeks to
understand how engineers and other
stakeholders use information on projects by
looking at their every day practices, rather than
the formally documented processes. The
focus is on uncovering the challenges and
opportunities that engineers face and
understanding how their practices are
institutionalised in business and regulatory
environments. The purpose is to inform the next
generation of engineering and design projects.
Methodology
The research is part of a trajectory of work on
technologies and practices at the University of
Reading that has drawn lessons from large UK
projects such as Heathrow Terminal 5 (T5),
Crossrail, and the London 2012 Olympics, and
which involved a range of leading industrial
partners. The academic research team worked
with Arup to analyse technologies and practices
on three Arup projects: Motorway 6 (M6) Toll,
Channel Tunnel Rail Link (CTRL), and SAFElink,
see Fig 1. They looked for lessons that could be
transferred to ongoing work and disseminated
more widely to improve UK construction.
Data on these projects, and the information
management processes and tools used, was
collected using three set-up meetings, thirty six
semi-structured interviews, non-participant
observation, and access to internal documents,
databases and information. The research was
designed as an embedded case-study, studying
three projects within the same firm, with data
collection in 2008-9. Analysis is ongoing through
a process that iterates between the data and the
literature to develop a theoretical contribution.
Preliminary results were discussed in feedback
meetings with each of the project teams and
with Arup Directors.
Results and discussion
By studying three projects that represent
milestones in Arup’s innovative use of
information management on projects, this
research charts the growing use of these
technologies across the infrastructure division.
Though remembered as all being equally large
projects within Arup, the projects are of different
orders of magnitude, scale and complexity, as
shown in Table 2.
The M6 Toll near Birmingham is the first toll road
built in the UK under the private finance initiative;
CTRL, now known as High Speed 1 in its
operational phase, is a high-speed railway link
between London and the Channel Tunnel in the
UK; and SAFElink is a motorway link between
Brisbane and Ipswich in Australia.
CTRL is an order of magnitude larger than the
other two projects. Its timescale for its delivery
overlapped with that of M6 Toll and staff that
worked on the early stages of M6 Toll then spent
time at CTRL before returning to the former
project. On the other side of the globe, SAFElink
is a more recent project that is pioneering new
technologies and processes.
52
Knowledge and
standards
Best practice,
Regulations,
Specifications
CAD
Drawings,
Calcs,
Engineering
Briefing
Functional req,
Estimates, Conditions,
Requirements
Project
Data
Demolition
Refurbishment
Rebuild, Demolition,
Restoration
Facilities Managment
Letting, Sale,
Operations Maintenance,
Guaranties
Construction
Management
Scheduling,
Logistics, 4D
Procurement
Product Database,
Price Database
VRML & Simultions
Visualisations,
3D Models,
Light & Sound,
Life Cycle
Specifications
Spec Sheets,
Classifications,
Standards,
Estimates
Fig 2. Diagram showing significant areas of construction projects,which benefited from using central source project data
All the projects were important in the
development of in-house capabilities for the
use of integrated software solutions. Design
teams gained experience developing and using
integrated processes and tools. The projects
involved coalitions of design and engineering
firms working together to deliver large scale
infrastructure, with the work divided up across
the many offices, disciplines, teams and firms
involved. On one project, for example, twenty
offices were involved in the delivery of project
deadlines at one stage.
Here the study observed how the integrated
software solutions become vital in the
coordination of design work and acted as a
digital infrastructure for project delivery. The
main innovations were the use of a collaboration
extranet software, Integration, as a collaboration
tool for project wide document dissemination on
M6 Toll, the implementation of IT tools and
processes across CTRL and their stability for the
duration of the project and work using 3D
modelling on SAFElink.
The three projects had features that led to
different strategic decisions about IT use
outlined below.
M6 Toll
The two design firms involved were based in
home offices rather than being colocated. Arup
introduced Integration to exchange documents
and data across the project. This software
facilitated this method of working in an efficient
and effective manner, enabling the sharing of
published documents. The main role of this
digital repository was to make visible knowledge
about the audit trail of design deliverables and
where the project was in the review process.
It managed the workflow approval process, it
overcame the version control problem, enabled
digital global information sharing and preserved
final design knowledge.
CTRL
The colocation of a project team away from the
firms’ offices meant it was possible to have all
designers using the same software tools and to
hold software stable over the limited duration of
the project, rather than upgrading the software
alongside corporate IT systems. The project was
highly successful in implementing a robust IT
strategy and processes for shared use of data.
Holding software stable simplified matters for the
project itself since it did not have to deal with
the costs and disruption of upgrades, but it did
mean that Arup staff returning to their offices
were unfamiliar with new technology
implemented on the corporate network while
they had been on the project.
SAFElink
A small and more recent project, but the largest
project run out of the Brisbane office. The Arup
design contract was ‘novated’ to the contractor,
who was unable to provide the IT infrastructure to
support the wide range of software tools required
by the design team. Arup therefore used the
corporate systems and these were augmented
by a project extranet collaboration tool (Incite) to
support data exchange between several Arup
offices and the contractor. Processes and data
management procedures were developed to
support the 3D CAD modelling tool deployed
to efficiently facilitate multilocation and
multidisciplinary design teams and the complex
approvals process. As a result of the success of
this approach Arup is now looking to disseminate
it more widely across the firm.
The management of information through the
whole life-cycle of the built environment is one
motivator for using integrated software solutions.
The ability to add value to the client in
data-handover has steadily improved across
these projects. On M6 Toll, at the end of the
project documents were taken from the extranet
CTRL
SAFElink
Teams left to use own design tools. Integration
used as a collaboration tool to exchange
documents and data. Software was updated
through the project. Because of an overlap in
timescales and staff there was a transfer of
learning to M6 Toll from CTRL
Client organisation selected the best tool for
the job. Limited duration of the project meant
little incentive to keep up with latest software
releases, so software versions were fixed
from 1996 for the project duration
Arup used own IT system, a 3D CAD
modelling environment and a project extranet
(Incite) to facilitate efficient iterative design,
review and approvals. Software was updated
through the project
Technology through
life
Final design documents were handed over in
paper and on CD to the client.
Design and other documents were handed
over to the client in an electronic format.
Arup will hand over 3D models and a
structural model for bridges to the client.
Project features
influencing IM
strategy
Two design teams, geographically separate
and from two different organisations,
operating on independent IT networks. Arup
developed the Integration solution to provide
an IT infrastructure.
Teams were colocated and used client’s
IT network and software.
Brisbane was the lead Arup office, with
design undertaken globally across a range
of Arup offices. Significant multilocation
interdisciplinary review and complex
contractor and client approvals. Not all
parties on the Arup IT network.
Table 1. Preliminary analysis on IT use on the projects
and handed over to the client in electronic and
paper forms. On CTRL the documents were
handed over to the client in an electronic format,
while on SAFElink Arup will hand over 3D models
and a structural model for the bridges. This is
due to Arup’s growing capabilities and also the
maturity of clients being able to manage
advanced data-sets.
Though there was a significant investment in
training on each of these projects, there were
difficulties in interfacing with the wider business
and regulatory environment. In the approval
process the terminology used in the system was
found to be extremely important in enabling a
digital review of the design rather than a paper
based review. For example, on one project it
became important to enable the client design
advisor to “receive” rather than “approve”
documents so that they could engage with the
system. This small change in terminology
enabled the team to use the system. It then
became a record of decisions that were made as
to reject a drawing, the reviewer had to input the
rationale for his or her decision, making these
judgments more widely visible to the appropriate
parties within the project.
These findings are summarised in Table 2.
The strong similarities between the three cases
are that each project made use of advances in
IT tools and that this use has been influential
across Arup. On each of these projects, design
teams have been at the forefront of developing
new ways of using the digital infrastructure for
delivery and of adding value to the client in data
handover. The many areas of project delivery in
which these technologies become significant is
summarised in the Arup diagram show in Fig 2.
Conclusion and next steps
Integrated software is becoming used as a digital
infrastructure for large building and infrastructure
projects. The research indicates the increasing
sophistication of this digital infrastructure being
used to integrate work across the many offices,
disciplines, teams and firms involved in project
delivery. While most studies have focused on its
role in large building projects, here the focus has
been on the role of integrated software in three
large infrastructure projects.
Information management becomes a core skill
area for infrastructure projects. The introduction
of integrated software as a digital infrastructure
for delivery is changing practices in engineering
design, making information more broadly
available across projects and increasing the
distribution of design work. For clients and for
design, engineering and construction firms, new
skills are needed to compete on the next
Type of
project
M6 Toll
CTRL
SAFElink
First toll road built in the UK
under a private finance
initiative
Railway line built between
London and the Channel
Tunnel, Folkstone, UK
Motorway link between
Brisbane and Ipswich,
Australia
Overall cost
approx. £500m
approx. £6bn
approx. £400m
Project dates
1986-2003
1990-2007
2006-2010
Main
innovation
Use of Integration as a
collaboration tool for
project wide document
dissemination.
IT tools and processes
implemented and fixed for
the duration of the project
Work undertaken using 3D
modelling and this was
provided as a deliverable
to the client
Table 2. Summary of project characteristics
generation of projects. These include 3D
modelling skills; wider software skills;
standardised approaches to information
management and a host of other skills in
developing and implementing new processes for
the coordination of ongoing design data.
Arup is developing the strong in-house
technology capabilities needed to add value
through, and interact with, different information
management approaches. Although there are
industry-level initiatives to create standards
across projects, each major project is currently
organised differently. Hence, for Arup,
understanding the capabilities of digital
technologies and how to exploit them effectively
are essential to delivering design excellence
efficiently to the benefit of the client. It is these
technology skills that have enabled Arup to be
central to the innovations on the projects
studied, responding to diverse requirements and
different system configurations on projects.
There are many remaining research questions
about how such systems are changing practices
and the emerging modes of design in the digital
economy. Two areas in which further research is
needed are the handover of information and
links between the design and operation of
infrastructure, and the different types of technology
strategies and their implications for different types
of projects. New practices are being developed on
projects across the globe. At the University of
Reading, the future plan involves developing a
repository of international best practice through
a new exploration group, the Design Innovation
Research Centre. This project is a first step in
this wider programme of research.
Within Arup, the knowledge gained from this
study has been used to assist in the production
of an Information Management Toolkit and
associated training course. These are intended
to help equip project managers with the
knowledge and information they need to make
the right Information Management decisions to
support the requirements of their commissions.
Acknowledgements
This research was written in collaboration with
the University of Reading (Jennifer Whyte,
Sunila Lobo).The authors acknowledge their
funder, the Engineering and Physical Sciences
Research Council (EPSRC) through the project
‘Infrastructure through Life: Technology use
in Global Projects’, part of the Innovative
Construction Research Centre (ICRC) at the
University of Reading.
References
Avanti (2006) Project Information Management:
a Standard Method & Procedure. London,
Avanti Toolkit 2 Version 2.0.
Boland, R.J., Lyytinen, K. & Yoo, Y. 2007,
Wakes of Innovation in Project Networks:
The Case of Digital 3-D Representations in
Architecture, Engineering, and Construction.
Organisation Science 18 (4): 631-647.
Eastman, C., Teicholz, P., Sacks, R. & Liston,
K. 2008, BIM Handbook: A guide to building
information modeling for owners, managers,
designers, engineers and contractors, Wiley.
ICE IS Panel 2008, Briefing: Knowledge and
information management for major projects
Proceedings of the ICE - Management,
Procurement and Law 161 (1): 9-16.
Harty, C. & Whyte, J. 2010, Emerging hybrid
practices in construction design work: the
role of mixed media, Journal of Construction
Engineering and Management, 136: forthcoming.
Kallinikos, J. 2005, The consequences of
information. London, Blackwell.
McGraw Hill, 2009, The Business Value of BIM:
Getting Building Information Modeling to the
Bottom Line, SmartMarket Report, New York,
McGraw Hill Construction.
Pavitt, K. 2002, System integrators as postindustrial firms. First draft of What are advances
in knowledge doing to the large industrial firm
in the new economy (Welcome Lecture) DRUID
Summer Conference on Industrial Dynamics of the
New and Old Economy – who embraces whom?
Shen, G., Brandon, P. & Baldwin, A., Eds.
2009, Collaborative Construction Information
Management. London, Spon Research.
Whyte, J. & Levitt, R. 2010, Information
Management and the Management of Projects,
In Oxford Handbook on the Management of
Projects (P. Morris, J. Pinto and J. Söderlund
eds) Oxford, Oxford University Press.
53
Digital infrastructure and changing practices in engineering design
M6 Toll
Strategic decisions
about IT use
Beasties in the
creative workplace
Authors: Duncan Wilson, Julie McCann, Asher Hoskins,
Chris Roadknight, Jane Tateson
54
Wireless Sensor Networks
are a key technology in the
emerging ‘internet of things’.
By monitoring office activity
using such a device as the
‘Beastie’, we can gain a
better understanding of
workplace performance.
Abstract
This paper presents an evaluation of an
installation of the Beastie wireless sensor
network to monitor a creative workplace, in
this case the University of the Arts Innovation
Centre in London. The sensor network was
tasked with passively monitoring of the
environment and space usage taking into
account environmental conditions and activity.
The paper focuses on one of the many
platforms used in the trial, the Beastie, using
an architectural description language
implementation called Tesserae, to measure
meeting room and social space usage. This data
was correlated using feature mapping at node
level on tiny 8-bit devices, and state changes
were propagated up the network to the
database. Some monitoring results are provided
and the performance of the algorithms on the
small devices discussed.
Fig 1. Beastie board
Introduction
Many approaches have been used to
measure the usefulness of a work space, such
as integrated workplace performance, Post
Occupancy Evaluations (POE) and research into
comfort. POE, for example, is predicated upon
there being a relationship between people’s
performance and the environment within which
they act. To obtain insight into how well a space
works, correlations with the ambient environment
usually involve taking temperature readings, etc.
from the Building Management Systems (BMS) or
sets of wired sensors placed in the building. This
solution is costly in terms of human resources
and technologies and moreover, due to the offline
nature of the system, the turnaround time of
results is quite slow.
Our method extended beyond the current
practice of getting a general feel of the
inhabitants’ experience in a one-off
questionnaire; we attempted to examine the
relationship between their subjective experiences
and environmental conditions over time. To this
end, Wireless Sensor Networks (WSNs) captured
tangible environmental factors such as light
levels, heat levels, and noise levels and
correlated these to the workforce’s reports on
intangible factors such as perceptions of
personal energy levels, well-being, stress, and
feelings of connectivity. The purpose of the
project was to investigate how pervasive
computing can be used to understand the
creative workplace.
Methodology
The results presented are based on a trial
carried out at the University of the Arts, Central
Saint Martins’ (CSM’s) Innovations Centre,
London, in Spring 2007 which ran for just under
a month. The experiment was part of a two-year
Technology Strategy Board (TSB) funded project
called ‘Bop’ and used a series of three distinct
metrics covering:
• behaviours, ie the extent to which certain
activities were in evidence
• feelings, ie how individual occupants felt
about how well their needs or wants
were met
The main requirements for the devices used for
this project were as follows:
• small, low-cost sensor nodes that could be
easily configured and deployed en masse
• programming environment and language(s)
that would allow non-sensor systems
programmers to deploy their code easily
• nodes that would accurately and efficiently
provide data 24 hours a day, 7 days a week,
to a database repository
• nodes that could withstand the wear-and-tear
of a creative office given that they would be
retrofitted and, in some cases, free-standing
devices
The highly heterogeneous architecture used
to achieve this consisted of four different
hardware platforms using many different systems
and languages. This study focuses on one
platform, the Beastie, which used an
architectural description language
implementation called Tesserae to measure
meeting room and social space activity.
WSN Architecture
Four different WSN platforms were deployed in
the whole project to facilitate the variety of
sensors used – Crossbow motes, Arduinos,
Bengt Polls and Beasties. Data fusion for all the
devices took place at a centralised MySQL
server where each data logger would push data
in order to provide a pseudo-synchronous view
of the WSNs. To achieve the link between the
base stations and the centralised database,
bespoke software was developed to forward
data to the database.
• activities, ie the extent to which certain
kinds of activities were being done
55
Light and proximity levels and their effect on state (sensor value vs time)
300
Ranger3
250
Light
State
8 Bit Reading
200
150
100
Fig 4. Beastie node counting sensor
50
0
0
500
1000
1500
2000
2500
3000
3500
4000
Data Sequence
Fig 2. Light and proximity levels and their effect on state (sensor value vs time)
State changes in a space due to part-time work (sensor vs time)
30
Fig 5. Beastie node
Light and proximity levels and their effect on state (sensor value vs time
25
300
8 Bit Reading
State
20
Ranger3
250
15
Light
State
200
10
150
5
0
100
State
0
5000
10000
15000
20000
25000
30000
Sequence
50
Fig 3. State changes in a space due to part-time work (sensor valve vs time)
State changes in a meeting space (sensor vs time)
0
30
0
500
1000
1500
2000
2500
3000
Fig 6. Beastie node in corridor
3500
4000
State
meeting space activity detection system, were
A web-based Representational State Transfer
Data Sequence
programmed using Tesserae. This allowed the
(REST) 25
query interface was used to interact
development of component-based systems to
with the database and perform queries on
provide better software engineering for WSNs.
filtered data based on location variables.
20
Tesserae is highly lightweight and encourages
Via the Bricks architecture, a user (or an
the programmer to produce better-engineered
application) can seamlessly query multiple
15
solutions by separating out the implementation
sensors that are part of the same location or
of a component from its composition within the
the same location subgroup. Bricks will
10
architecture; the behaviour of components is
automatically
process the values from all the
defined through the use of interfaces.
sensors that logged data, hiding the complexity
5
of the underlying
WSNs and providing a
State
Measuring room states on the Beastie platform
persistent interface for the end-user.
Each
Beastie
was
placed
in
a
social
meeting
0
0
5000
10000
15000
25000
30000
space and 20000
contained three
sensors–temperature,
Beastie hardware
and software
light level and multiecho ultrasound. The latter
The Beasties are simple, wireless, microcontrollerSequence
produced a ping signal which recorded the first
devices designed for prototyping wireless sensor
16 echoes. Through initial calibration of each
node setups. Each Beastie had an 8-bit micro
space, we found that the first, third and eighth
controller, low-power digital radio (433MHz), a
echo readings were adequate to represent a
power supply and an expansion bus connector.
distance covering activity within near, middle
The Beasties communicated using an autonomic
and far-off ranges, respectively. These data
network, which was self-configuring, selfwere combined with the temperature and
optimising (device usage was automatically kept
light readings and placed into a feature map
at an optimum level) and self-repairing (the
which was processed and stored on each
network adapted to take account of new devices
individual Beastie.
as they were added to or removed from the
system). Fig 1. shows the Beastie platform.
These data derived a number of states for each
Applications, both ambient sensing and the
space. For example, dark and empty, bright and
empty, hot bright and someone there, bright and
empty, etc. Using the meeting schedule for the
56
more formal spaces allowed us to interpret the
results from the feature map. This in turn allowed
us to follow state changes in space usage such
as night-time, cleaner entering a space and
vacuuming, space in use, not in use, etc. and
night-time again.
Fitting the SOM to the Beastie
The ranger algorithm used on the Beasties in the
CSM trial was a modified self-organising map
similar to a Kohonen network. An initially random
set of points was placed in an n-dimensional
space, which was moved towards each sensor
reading as it was plotted into the space. The
map used in this trial had five 8-bit dimensions,
one for each of the sensor readings taken at one
time, and 32 points or ‘states’. The map was
stored in electrically erasable programmable
read only memory so that it would survive
battery changes. Every minute, the Beastie
would ‘wake up’ and take light, temperature and
three ultrasound measurements (the first, third
and eighth echoes from a single ping) to
produce five variables to be fed into the map.
Some modifications were made to the algorithm
both to enable its deployment on low-powered
devices but also to avoid excessive tuning and
overtraining. As we were expecting to see
multiple states, our self-organising map
0
0
5000
10000
15000
20000
25000
30000
Beasties in the creative workplace
Sequence
State changes in a meeting space (sensor vs time)
30
25
State
20
15
10
5
State
0
0
5000
10000
15000
20000
25000
30000
Sequence
Fig 7. State changes in a meeting space (sensor vs time)
algorithm was designed to not converge to a
single state or even to an optimal number of
states, but to make use of the 32 initial states
for the period of the trial. Thus only the single
point in the map nearest the sensor reading,
ie the ‘winner’, was moved. For simplicity,
the winner was always moved an absolute
distance closer to the reading rather than a
percentage closer.
Results and discussion
Over the four-week trial, more than 1m sensor
readings were gathered, processed in situ and
wirelessly forwarded to the base station. An
example of a daily trace of data is shown in
Fig 2. For clarity, only two of the five sensor
readings are shown along with state changes.
The response times of the ultrasound and light
levels appear to have an important effect, as
would be expected. Although day and night are
expressed as states there also appear to be
substrates within these, which could be
attributed to out-of-hours working and
gatherings of people.
Looking over a period of four weeks, the
changes in states offer a good indication of
changes in usage of various work spaces. Fig 3.
shows the part-time weekly work practice of a
member of staff and Fig 7. shows how a meeting
space was used differently over time. Further
analysis of data-to-state relationships is beyond
the scope of this paper but it is apparent that
the devices deployed were effective in gathering
and analysing a large quantity of diverse data.
Evaluation and requirements
The main adaptations that had to be made to
the algorithms were to restrict them to integer
maths and simple functions (no square root
for calculating distances between points, for
example) and to limit the amount of memory
they used (the feature map had to fit in the
1K RAM of the Beastie).
Although the devices worked well to identify a
range of different activity levels within a space,
there is evidence that within the crowded office
environment, the sensitivity was limited to
picking up coarse changes. Improvement may
have been possible by further hand-tuning the
placement of the device and the echoes used as
Fig 8. Ultrasound used to analyse occupancy
sensor inputs. However, in the spirit of a
deployment that is self-optimising, it would be
better to implement an algorithm to enable the
devices to choose the echoes themselves,
based on the maximum variability of readings,
during a training period.
Greater sophistication could also be achieved by
subtracting persistent environmental reflections
from the data points recorded, so that only
changes in the environment are registered. In the
open spaces of the foyer and corridor, the
background-reflected sound signals had a
simple profile, so changes to that profile, due to
human activity, stood out clearly. Within the
crowded office environment, however, complexreflected sound signals were generated because
of office clutter, which masked changes caused
by physical activity. This did not give erroneous
results, it merely restricted the sensitivity of the
device, resulting in a more limited number of
states being observed.
Conclusion and next steps
Two common themes ran through the
experiences obtained by the developers
on the project:
• the hardware was fiddly and required more
than a basic knowledge of electronics to
configure
• the software was immature and not
suitable for a non-computing programmer
(even programmers had problems
deploying the systems)
These observations reflect the fact that the
main component developers in the Bop project
are not computer scientists/programmers
but technologists, designers and system
integrators. This is compounded by the
fact that unlike general server or PC-based
technologies, WSN technologies are relatively
immature, stemming from university and
corporate research laboratories and have
been designed for experimental rather than
mass-market usage.
database technologies, to allow the less
experienced developer to easily be creative.
However, this does highlight a situation that the
WSN community has to face and is attempting
to address.
Whilst the initial trend in selling the advantages
of WSN focused on ease of deployment
(ARC report a 10% cost of a wired design)
the current focus of research has been driven by
the desire to use WSN to support energy saving
activity. Consequently, there is a body of
research focused on systems that trade-off
energy consumption against user preferences
and an increasing trend for Green IT surveys.
Acknowledgements
This paper has been adapted from a longer
article published with Julie McCann and Asher
Hoskins (Imperial College) and Chris Roadknight
and Jane Tateson (BT) in Intelligent Buildings
International 1, 2009, 222-229.
This work was supported in by the UK
Technology Strategy Board.
References
Bordass, W., Leaman, A. and Eley, J., A Guide
to Feedback and Post-Occupancy Evaluation,
York, Usable Buildings Trust, 2006.
Kohonen, T., Self-Organizing Maps, 3rd ed,
Berlin, Springer, 2000
McCann, J.A., Huebscher, M. and Hoskins,
A., Context as autonomic intelligence in a
ubiquitous computing environment, International
Journal of Internet Protocol Technology (IJIPT)
special edition on Autonomic Computing,
Inderscience 2007 2 (1), 2001, 30-39
Qiao, B., Liu, K. and Guy, C., A multi-agent
system for building control, in Proceedings of
the IEEE/WIC/ACM International Conference
on Intelligent Agent Technology (IAT’06).
Hong Kong, December 2006, 653-659.
Roberts, S., An integrated framework to improve
the workplace, Facilities Management January,
2006, 10-12.
Yet many have been marketed en masse, even
though they do not have the development
tool-sets, robustness, ease-of-use, interface and
integration, and general maturity, as do general
57
Neighbourhood Pedestrian
Analysis Tool (NPAT)
Authors: Varanesh Singh, Eric Rivers, Carla Jaynes
58
There is a need to perform
pedestrian analysis at a
broad level with the goal of
identifying and prioritising
pedestrian improvements.
At present there are limited
solutions by which to collect
and analyse this data.
Abstract
This research looks at the development of a
Neighbourhood-level Pedestrian Analysis Tool
(NPAT) that addresses practitioners’ concerns
and provides agencies a means by which to
analyse a large scale pedestrian environment in
a user-friendly, graphical environment while
leveraging their existing data collection
methods and information sets.
Using the Geographic Information Systems (GIS)
platform as the basis for the NPAT, the design
team identified key inputs and parameters that
must be incorporated into the tool. With this
information in hand, the design team evaluated
two methods to perform the analysis: a polygon
method where pedestrian space is represented
as shapes, and a link/node method where
pedestrian space is represented as a series of
links and nodes, similar to typical vehicular
capacity analyses.
The design team selected the polygon method
because it allows for easier analysis and better
represents pedestrian movement and results.
This method was then applied to a small test
case with favourable results. The design team
also coordinated with the New York City
Department of Transportation and received a
favourable response including the potential
for future collaboration.
The research shows that there is a feasible
method to perform the broad-brush analysis
of pedestrian spaces. There are some areas
requiring further development which will be
addressed in future phases of research
and design.
Fig 1. The pedestrian environment
Introduction
Many cities face the challenge of providing
adequate pedestrian amenities for its residents
and visitors among the other competing
requirements within the public right of way.
With congestion becoming more common for all
modes of travel, cities need a way to understand
pedestrian-specific congestion on a large scale in
order to balance and prioritise pedestrian space
improvements and make the best use of limited
funds and resources.
Currently, the accepted criterion used by
agencies to evaluate performance of pedestrian
space is Level of Service (LOS). Existing
methodologies and standards include those
established by the Highway Capacity Manual
(HCM) which is based upon Fruin’s thresholds
for density, flow and delay. The processes within
the HCM require analysis at the intersection
level, which can be onerous in terms of inputs
and time if analysis is sought at the corridor or
neighbourhood level.
Acknowledging the numerous additional factors
that contribute to a pedestrian environment,
cities and agencies are starting to look towards
a different type of measurement, Quality of
Service (QOS). QOS measures the quality of the
pedestrian experience as people move through a
space, including characteristics of the built and
natural environments and factors such as safety,
comfort of design, and land use.
Despite abundant existing data, there are
currently limited means to perform pedestrian
analysis at a broad-brush level. Large scale
micro-simulation models, require extensive
data and can be expensive to build and
calibrate. While less expensive spreadsheet
analysis can be performed in a shorter time
frame, this type of analysis is often piecemeal
and does not provide graphical representation of
the results. Given these limitations, there is a
need for a tool that can economically and
holistically perform pedestrian analysis.
This research looks at the development of a
Neighbourhood-level Pedestrian Analysis Tool
(NPAT) that addresses the concerns noted above.
it also provides agencies and practitioners a
means by which to analyse a large scale
pedestrian environment in a user-friendly,
graphical environment while leveraging their
existing data collection methods and
information sets.
This tool will ultimately allow practitioners
to pinpoint and prioritise those sidewalks or
crosswalks within a neighbourhood in need of
attention and ensure that walking remains the
most attractive mode of transportation.
Methodology
The research presented below represents the
first of a three phase process. The methodology
of phase 1 was to evaluate various technical
processes and select a preferred means for the
NPAT. The resulting technical process was then
applied to a case study to determine if the
process and outputs were adequate and
warranted further action. Future phases will
include the automation and refinement of many
of the processes identified in this phase.
Prior to conducting the core research, the
design team identified parameters which the
NPAT must incorporate, the first parameter
being the platform for analysis. The design team
selected Geographic Information Systems (GIS)
for this phase of research because of its
accessibility for major cities, as well as the fact
that most geographic data is stored in GIS.
The final parameters identified are key inputs
and outputs which are described in detail below.
Inputs
The required inputs relate to pedestrian space
and its characteristics as well as the
characteristics of the pedestrians and how
they move through the space.
59
A
D
B
E
C
F
Fig 2. Visual representation of pedestrian walkway LOS (adapted from HCM)
Defining the walkable limits of a space is
important in establishing the area of analysis.
For this tool, four distinct walkable regions
for pedestrian analysis have been identified
as important for determining volume, speed,
and density.
These regions contain unique characteristics
that require they are treated differently.
They are summarised below:
• sidewalks are defined as any space that is
primarily pedestrian and lacks modal conflicts
• crosswalks consist of road space allocated
for pedestrians crossing from one corner to
another
• corners consist of the portion of the sidewalk
allocated to crossing behaviours such as
queuing, changing directions or waiting
• areas of mixing occur where perpendicular
flows of traffic meet. These include areas
such as building entrances or subway exits
The pedestrian space for analysis must also take
into account obstacles such as lamp poles, fire
hydrants, newspaper boxes, trees etc which affect
the way in which pedestrians utilise a space.
For the purposes of analysis, obstacles are
typically accounted for by subtracting out the
obstacle’s area plus a standard buffer distance
around the obstacle.
Beyond the geometry of the space, it is also
important to consider the following inputs:
• character of the pedestrian space: this is
most critical when conducting a QOS
analysis; characteristics of the built
environment such as the type of adjacent
roadway, adjacent land use, and presence of
street furnishings all affect the QOS
• aspects of the natural environment such as
wind, shade and sun effects are also
important in evaluating the space
• characteristics of the pedestrians: this is an
important component as it impacts both LOS
and QOS calculations; these characteristics
include age, walking speed, trip purpose and
relevant cultural or demographic information
60
• pedestrian demand data: this is the most
critical piece of information and dictates how
many pedestrians move through the space
within a given period; this is typically
collected in 15 minute intervals, but the
intervals can range between 5 minutes
and 1 hour
Outputs
The outputs are the quantitative results that will be
used by practitioners to evaluate the performance
of the pedestrian space. LOS is the primary
output of evaluation for this phase of the research
and consists of a range of pedestrian densities
and volumes categorised into a scale of A
through F as demonstrated in Fig 2.
Other outputs could include conflict analysis and
QOS results. While QOS outputs are still being
developed and refined by the pedestrian
planning community, they can easily be
incorporated into the NPAT at a later date.
Results and discussion
With the key inputs and outputs defined, the
design team evaluated methods for achieving the
analysis within the GIS environment. Two separate
processes were evaluated: one process using
polygons and another using a link/node system.
Polygons would represent the actual boundaries
of the space and would visually represent the
walkable area including any space removed to
account for obstacles. A link and node system
would represent the spaces schematically. A
node would represent a corner or area of mixing
and each link would represent the sidewalk or
crosswalk.
While the link/node system typically used in
vehicular analysis, the design team deemed the
polygon method to be the more suitable process
because pedestrian LOS, delay and quality is so
dependent on the physical definition of the
space rather than the number of lanes.
The resulting polygon process was broken down
into four modules to collect the inputs and
produce outputs. The details of each module are
described below.
Process of defining pedestrian space
Curbline +
building footprint
Divide polygon
Building area
Building area
Sidewalk
Divided sidewalk
Fig 3. Process of defining pedestrian space
Walkable space module
The module for identifying the walkable space
for pedestrians creates a new GIS shapefile of
the area between building lines and curblines
using the “Intersect” function.
This newly created shapefile then takes into
account obstacles by creating a buffer around
each obstacle and subtracting it out of the
walkable area using the same “Intersect”
function. The design team found that there are
limitations to this type of buffer analysis in
heavily used pedestrian areas, including the
potential to arrive at a negative amount of
pedestrian space. Furthermore, in congested
situations, pedestrians can accept lower buffer
distances. These challenges will be reviewed
further in the next phase of work.
Polygons representing crosswalks are also
created as part of the pedestrian space
boundary. These must be manually drawn unless
cities have this information stored in graphical
format. For the purposes of this research,
crosswalks will be addressed in detail in the next
phase of work.
The resulting walkable space is then further
subdivided into smaller analytical areas that
allow for more detailed pedestrian analysis. The
analysis regions are determined based on a
critical width of each of the polygons in the
pedestrian space boundary. The resulting
shapefile consists of manageable areas that can
be categorised as sidewalks, crosswalks,
corners or areas of mixing.
The process of producing this shapefile involves
creating a new line file that specifies where the
pedestrian space boundary polygon will be cut.
Based on these lines, the polygons are
subdivided using the “split polygons by
polylines” feature in the topology tool. Fig 3.
shows the process of defining pedestrian space.
et
101
Ch
u rc
hS
t re
402
lto
nS
t re
dw
405
404
501
Thru
(Dest1)
To
(Dest2)
Demand
104
103
102
53
102
103
104
78
103
501
404
105
404
501
103
211
103
502
201
175
201
502
103
93
403
404
-
257
404
405
-
75
103
104
Br
oa
102
403
et
ay
Fu
From
(Orig)
Neighbourhood pedestrian analysis Tool (NPAT)
401
502
504
LOS
A
301
507
201
B
D
E
F
Fig 4. Graphical display of results
Characteristics module
This module allows the user to incorporate
inputs that pertain to characteristics of space
and pedestrians. The majority of these inputs will
be used to produce a QOS evaluation. But some
of the information, such as those related to
pedestrian speed and trip purpose can be used
to adjust the LOS outputs.
Information on the characteristics of the space
and pedestrians are based on geographic
conditions and can be spatially joined to each
pedestrian analysis region. For example,
pedestrian speed is related to the land use and
will tend to be higher in business districts.
Demand module
Pedestrian demand data input into the tool will
be stored in tabular format and then spatially
joined to the shapefiles representing the
available pedestrian space. Using a path system,
the pedestrian demand data is linked to
polygons based on the start polygon, through
polygon, and end polygon of each movement.
One of the benefits of using the path system is
that it stores data in a format that can later be
used to estimate an Origin-Destination (O-D)
matrix for a given area. This O-D matrix can be
used to get a better understanding of travel
patterns, help identify deficiencies in a network
and provide a key input for micro-simulation
models. The ability to estimate an O-D matrix
requires a sufficient amount of data and an
algorithm that can process the data intelligently.
As a result, this process has been identified as a
longer term goal.
Results module
Outputs will be generated through scripts which
perform the LOS calculations and then associate
the final result with the corresponding shape.
The shape can then be symbolised or colourcoded to graphically present findings based on
established performance measures. For the first
phase of the research, the results are calculated
separately and then joined to the shapefiles.
Future phases of the research will include
processes to automate this step of the analysis.
Fig 5. Demand module interface
Case test and follow-up
A case test of the described process was
performed on a small area of Lower Manhattan
in New York City. This location was selected
because of Arup’s recent work with New York
City Department of Transportation (NYCDOT) in
this area and the availability of existing data.
The case test resulted in a colour-coordinated
map of polygons shaded to correspond to LOS
values. The results generally corresponded to
the LOS observed on site. Fig 4. shows a
graphical representation of the results.
The results of the case study showed that
significant manual work is required to generate
the pedestrian space. Shapefiles have to be
adjusted and field verified before adequate
pedestrian space shapefiles can be created.
Future phases of the model must address ways
to improve this process.
Arup met with staff of the NYCDOT to present
the idea of the research. Based on the
information discussed, NYCDOT is interested in
the potential of the NPAT and supports Arup’s
pursuit in further developing the tool. There is
interest in using the city as wider-area case test,
to be included in phase 2 of the research study.
NYCDOT and Arup have agreed to meet and
further discuss the potential of this tool.
NYCDOT have also recommended a larger and
diverse neighbourhood for use to be included in
phase 2 of the research. They have also agreed
to provide data and other resources to assist in
this wider-area case study.
Conclusions and next steps
The research demonstrated the feasibility of
creating an affordable tool to effectively analyse
pedestrian activity at a mesoscopic scale.
However, further work is required to automate
several key tasks and to incorporate additional
evaluation measures.
In addition to developing a proof of concept,
the team has identified a potential partner in
the NYCDOT who has the need and interest for
such a tool.
Further refinement of the NPAT and collaboration
with NYCDOT could yield a valuable analytical
tool for agencies in North America and around
the world.
A goal of future phases is to determine
methods of automating a majority of the
processes required for the analysis. The
automation will reduce the amount of
manual input, thus making the process more
user-friendly and providing an overall more
efficient and effective tool.
Future phases will involve developing more
rigorous methods to validate the results
to existing conditions, a process which is
important in determining the legitimacy and
strength of the tool.
Further coordination with NYCDOT is also
identified including a wider area test study in a
larger more diverse area of New York City.
Acknowledgements
We would like to thank New York City
Department of Transportation for providing input
and feedback before and during the research.
References
City of New York, City Environmental Quality
Review Technical Manual, 1 st ed., City of
New York, 2001
Fruin, J.J., Pedestrian Planning and Design,
2 nd ed., Elevator World, Inc., 1987
San Francisco Department of Public Health,
Pedestrian Environmental Quality Index, version
1.1, Available online [http://www.sfphes.org/
HIA_Tools/PEQI_Methods_2008.pdf], 2008
Transportation Research Board National
Research Council, Highway Capacity Manual,
4 th ed., National Academy of Sciences, 2000
61
Human induced vibrations
on footbridges
Authors: Iemke Roos, Peter Burnton
62
The past few decades has
seen demand for better quality
pedestrian and cycleway
facilities. Together with
developments in materials, this
has led to the design of longer,
more complex and slender
footbridges that can be more
sensitive to dynamic forces.
Abstract
Considerable public money is spent on
footbridges and the bridges are expected to
offer comfortable passage to the public.
These bridges can be more sensitive to dynamic
forces brought on by pedestrians, resulting in
vibrations of the bridge deck and possibly
affecting the overall ‘comfort factor’ of the
bridge. Older, simplified design rules, often
based around the movement of a single
pedestrian, are considered to be no longer
adequate.
This study explores the natural frequency,
damping, pedestrian load models and the public
human response to footbridges. The objective is
to compare several load models described in
current codes of practice (British, European and
Australian) intended for practical engineering
application.
To validate the load models, the computer
generated responses are compared to the real
behaviour of two bridges, the Goodwill Bridge
and the Milton Road Bridge, both located in
Brisbane and designed by Arup. This article
focuses on vibrations in the vertical plane,
although horizontal or Syncronous Lateral
Excitation (SLE) effects were also considered.
Proposal annex C
UK national annex
Australian standard
Single pedestrians
(walking)
3
3
3
roup of Pedestrians
G
(walking)
3
3
7
Takes into account:
Joggers
7
3
7
Crowd
3
3
7
Non-moving harmonic
Moving and non-moving
harmonic
Moving harmonic
Loading Time
Until Steady State
Depending on velocity
Depending on velocity
Load Frequency
Natural Frequency
Natural Frequency
Between 1.75Hz
and 2.5Hz
Load model characteristics:
Load
Dynamic load factor dependent on:
Group size
3
3
Not mentioned
Natural Frequency
3
3
Not mentioned
Degree of
synchronisation between
pedestrians
3
3
Not mentioned
Application conditions:
Vertical
fv < 5 Hz
fv < 8Hz
1.5Hz < fv < 3.5Hz
Table 1. Comparison of codes
Introduction
Anyone who has walked over a bridge has
probably felt or seen small movements of the
deck, moving up and down, caused by traffic,
pedestrians or even wind. These vibrations are
usually small and only perceptible with a static
reference point or when standing still on the
bridge. The magnitude of these movements
depends on many factors: length of the bridge,
stiffness of the bridge, load type, load
magnitude, load position, sensitivity of the
observer and many more.
To fully understand the response of the bridge it is
essential to model the loads correctly. Pedestrian
loads are difficult to model because of the
unrelated variables such as: weight of the
pedestrian, walk velocity, number of pedestrians,
distribution of the pedestrians over the bridge, etc.
The centre of gravity of the human body is
located at about 55% of its height and makes a
sinusoidal motion during walking, both in vertical
and horizontal directions. The force thus has
three components: a vertical, a longitudinal and
a lateral component. The vertical component is
up to 40% of the body weight. The other
components are considerably smaller.
Walking, running or jumping each produce a
different loading curve over time and different
frequencies for the load pulse.
The vertical force component during walking
shows a characteristic double hump, which
is the result of the impact of the heel on
the ground followed by the push-off force.
The force magnitude tends to increase
with increasing step frequency.
The footfall force envelopes overlap as in walking
both feet are briefly on the ground at the same
time, Fig 2. Both feet can be off the ground at the
same moment when running.
Synchronisation of pedestrians is more likely to
occur at higher pedestrian densities, when
people are not able to walk freely. At a density of
1.0 person per m² the freedom of movement is
greatly inhibited. When the density reaches
about 1.5 persons per m², walking becomes
difficult and pedestrians are dependant on the
pace and direction of other bridge users. The
pedestrian velocity decreases as the density
increases and consequently the dynamic forces
on the bridge decreases.
Synchronisation between runners is less likely to
occur, as the velocity is quite high and thus the
density is lower. Typically people do not adjust
their stride to the vertical movement of a bridge
and hence synchronisation between pedestrians
and the natural movement of the bridge
structure is not significant for the assessment
of vertical vibration.
Perception of vibrations by pedestrian bridge
users is subjective and psychology is an
important variable. Each person considers an
uncomfortable vibration differently, depending on
the environment, activities around them, type of
bridge, what they are doing, cultural influences,
age, etc.
This study has used Goodwill Bridge and Molton
Road Bridge, Brisbane to validate the load
behaviour model.
63
Methodology
Goodwill Bridge
The Goodwill Bridge, Fig 1. is a bridge for
pedestrians and cyclists that spans over the
Brisbane River. It links the southern part of the
Central Business District (CBD) of Brisbane with
South Bank which offers many public attractions
as well as a railway station and bus service. It
was opened in 2001 and is used by
approximately 40,000 people per week. The
450m long footbridge has three distinct parts:
the Rampart on the South Bank riverside, the
Arch as the main span over the river and the
steel girder spans on the CBD approaches. The
Arch spans 102m and provides 13m navigations
clearance above the tidal water level. Users are
a mix of commuters and recreational users.
Under normal circumstances, small vibrations
can be perceived when standing still on the main
span of the bridge. These vibrations are not
usually felt when walking on the bridge. Larger
vibrations in the main span have been noticed
when groups of joggers use the bridge. These
vibrations can be felt by people standing,
walking or jogging, but have not been reported
as uncomfortable.
The vibrations at Goodwill Bridge had been
measured at three strategic locations on the
deck during a community marathon event.
Measurements were also taken during normal
use of the bridge. This data was made available
for this study.
Milton Road Bridge
The Milton Road Bridge, Fig 2. was built to link
the 52,500 seat Suncorp Stadium with the
Milton Train Station. The steel truss bridge is
largely used by people going to, or coming from,
the stadium in large groups. The 86m long and
8.45m wide twin span bridge spans over Milton
Road with a 6.5m vertical clearance.
Vibrations have been noticed near the midspan
of the longer span of the bridge by people
standing still but have not been reported as
uncomfortable. Vibrations are best perceived
when one or two pedestrians are crossing the
bridge and at a location where an adjacent tree
provides a static reference relative to the bridge
movement. This bridge is essentially used by
large crowds moving between the stadium and
the train station. Therefore, people do not
usually stop on the bridge.
The codes
The load models considered for this study have
been selected from the European, British and
Australian codes of practice.
Proposal Annex C for Eurocode 1
In 2001 a Proposal Annex was issued for
Eurocode. This Annex has not been officially
approved but issued as guidance for designers.
Annex C gives guidance on determination of the
natural frequencies, structural damping and
dynamic load models.
British National Annex for Eurocode 1
of EN 1991-2
The aim of the UK National Annex is to provide
sufficient guidance to account for the effects of
vibration of complicated structures and those in
64
CBD
Brisbane river
South Bank
The Arch
The Pier
Goodwill Bridge
Direction
water flow
The Rampart
Fig 1. View of Goodwill Bridge, Brisbane
Single pedestrian
Group of
pedestrians
Joggers
Crowd
Proposal Annex C
UK National Annex
Australian Standard
a max = 0.184m/s²
a max = 0.015m/s²
a max = 0.015m/s²
u max = 1.25mm
u max = 0.09mm
u max = 0.09mm
Milton Road Bridge
a max = 0.360m/s²
a max = 0.110m/s²
a max = 0.005m/s²
u max = 0.52mm
u max = 0.16mm
u max = 0.02mm
Goodwill Bridge
a max = 0.552m/s²
a max = 0.027m/s²
u max = 3.76mm
u max = 0.16mm
Milton Road Bridge
a max = 0.257m/s²
a max = 0.216m/s²
Goodwill Bridge
u max = 0.37mm
u max = 0.32mm
Goodwill Bridge
-
a max = 0.182m/s²
-
u max = 0.70mm
Milton Road Bridge
-
a max = 2.202m/s²
Goodwill Bridge
Milton Road Bridge
-
u max = 0.29mm
a max = 4.116m/s²
a max = 2.087m/s²
u max = 28.16mm
u max = 14.16mm
a max = 1.520m/s²
a max = 0.456m/s²
u max = 3.68mm
u max = 0.66mm
Table 2. Values for maximum acceleration and displacement for each bridge and pedestrian load type
(a max = maximum acceleration, u max = maximum displacement)
sensitive locations, without imposing undue
conservatism that might constrain designers in
achieving an economic solution.
Australian Standard AS5100.2
AS 5100-2004 is the Australian Standard for
Bridge Design. Clause 12.4 of Part 2 deals with
vibration of pedestrian bridges. This clause is
similar to that in earlier Australian Bridge Design
Codes and is representative of earlier Codes in
the UK where the structural response to a single
pedestrian is used for the compliance criteria.
Table 1. provides a comparison of key aspects
of the approach in each Code. In addition there
is variation in the specific loading and
acceptable comfort criteria in each Code.
Results and discussion
The Goodwill Bridge and the Milton Road Bridge
were analysed according to the three Codes
discussed above. The results of these analyses
are then compared to the measured and
reported performance of the two bridges.
The maximum values from the analyses are
summarised in Table 2. The table contains
both the maximum acceleration a max and the
maximum displacement u max for each bridge
and pedestrian load type.
The results satisfy the compliance requirements
stated in the Codes, except those generated
with crowd load cases. Both Annex C and UK
Annex appear to overstate the influence of
crowds compared to the observed experience at
the two bridges.
This over estimation could be explained by the
fact that the pedestrian density is not taken into
account in the amplitude of the dynamic forces
of pedestrians. Pedestrians tend to walk slower
in high density situations and as a consequence
produce smaller dynamic forces. Crowd loads
are also applied as a point rather than a patch
load which will increase the analysis result. The
degree of pedestrian synchronisation may also
be overstated for the case of the two bridges.
Human induced vibrations on footbridges
To Milton train station
Vibrations
Load
Load
View along the bridge
(looking to the Stadium)
Walking
Running
Ti
m
e
Le
To Suncorp Stadium
Ri
ft
gh
fo
tf
Ti
ot
oo
m
Le
e
Ri
ft
gh
fo
tf
t
ot
oo
t
Fig 3. Patterns of running and walking forces
Acknowledgements
This study is entirely based on the M.Sc.
Thesis undertaken by Iemke Roos, Delft
University of Technology while he was embedded
in the Arup Bridge Design team in Brisbane.
We thank Iemke and the Academic Staff in the
Faculty of Civil Engineering and Geosciences at
Delft University of Technology.
Connection trusses to the main chords
Mid-support columns
View under the bridge
Fig 2. The Milton Road Bridge
Proposal Annex C calculations produce
accelerations that are too large compared
to those observed on the two bridges.
However UK National Annex and Australian
Standard (in the case of a single pedestrian),
generate accelerations and displacements that
are slightly lower than observed with the
exception of the load case representing a crowd.
Results generated with the UK National Annex
are smaller than those calculated with Proposal
Annex C. This is mainly because moving loads
are used in the UK Annex meaning that the load
is not located at the point of maximum influence
for more than a single stride. Moving harmonic
loads appear to best represent pedestrian loads
however they are more complex to both model
and interpret.
The static load alone does not explain the
magnitude of the differences between the two
codes. The dynamic load factor is thought to be a
further contributor to this difference. When
considering the load cases representing a single
pedestrian and a group of pedestrians on the
Milton Road Bridge, the responses calculated with
Proposal Annex C for a single pedestrian are three
times higher than the ones calculated with the UK
National Annex. However, there is nearly no
difference in the case of a group of pedestrians.
The Milton Road Bridge has a relative high
natural frequency which influences the dynamic
load factor of Proposal Annex C considerably.
The dynamic load factor in the UK National
Annex is more complicated. Different parameters
are used to assess the dynamic load factor such
as the natural frequency and degree of
synchronisation. The number of pedestrians has
a considerable influence and is the key reason
why the dynamic load factor is higher for a
group of pedestrians compared to a single
pedestrian. The acceptance criteria for bridge
vibration given in the Codes do not vary with
differing bridge form, use, or location.
The acceptance criteria are different in each of
the Codes but do not vary for the different
bridges when any one Code is considered. One
can foresee a situation where the lack of
flexibility in the acceptance criteria could have a
significant impact on the design outcome.
Conclusions and next steps
Of the three codes considered, the UK Annex
offers the best representation of pedestrian
induced vertical vibration. Loading for most load
types are possibly a little understated.
The representation of crowd load effects is
significantly overstated in both codes
considered. Moving loads offer a better load
representation than static loads but are more
complex to analyse and interpret. The Annex C
approach overstates vertical vibration effects for
all pedestrian load types given in the code. The
Australian code only considers a single
pedestrian and may prove to be inadequate for
some bridges.
References
H. Bachmann, W.J. Ammann, F. Deischl, J.
Eisenmann, I. Floegl, G.H. Hirsch, G.K. Klein,
G.J. Lande, O. Mahrenholtz, H.G. Natke, H.
Nussbaumer, A.J. Pretlove, J.H. Rainer, E.
Saemann, L. Steinbeisser, Vibration Problems
in Structures: Practical Guidelines, 1995
D.R. Leonard, Human Tolerance Levels for
bridge Vibrations, Ministry of Transport RRl
Report No. 34, Road Research Laboratory,
Harmondsworth, 1966
J.W. Smith, The vibration of Highway Bridges
and the effects on human comfort, Ph.D.
Thesis, University of Bristol, September 1969
C. Barker, S. DeNeumann, D. MacKenzie, R.
Ko, Footbridge Vibration Limits – Part 1:
Pedestrian Input, Footbridge 2005 International
Conference
D. MacKenzie, C. Barker, N. McFadyen, B.
Allison, Footbridge Vibration Limits – Part 2:
Pedestrian Input, Footbridge 2005 International
Conference
C. Barker, D. MacKenzie, Design Methodology
for Pedestrian induced Footbridge Vibrations,
Footbridge 2008 International Conference
fib Bulletin No. 32, Guidelines for the design
of footbridges, 2005
J. Blanchard, B.L. Davies, J.W. Smith, Design
Criteria and Analysis for Dynamic Loading of
Footbridges, Symposium on Dynamic
Behaviour of Bridges, 1977
J. Blaauwendraad, CT2022 Dynamica van
Systemen, TU Delft, 2006
A. Romeijn, CT5125 Steel Bridges, part 1,
TU Delft, 2006
The next steps in this study could include the
consideration of flexibility in the acceptance
criteria given in the Codes; improved
understanding of crowd loading within the
context of bridge vibration analysis; and
improvement to the way moving dynamic loads
are modelled in structural analysis software.
65
The importance of research at Arup
The Arup Research Review is one
of our more recent publications.
It presents some of the work which
has been conducted by our Arup
colleagues at the cutting edge of
our industry over the past year.
It captures one of the most important
principles of Arup’s ethos, that being
we bring something special to what
we do, wherever possible.
Client value can be added in many
ways, one of the most fundamental
contributions from our firm has been
in the area of ‘breaking the mould’.
We like to make things possible that
have previously been considered
impossible.
The Research Review demonstrates
that this tradition is still alive and well
in Arup today. Not all research ends
up with immediate application in areas
that bring direct benefit to our clients.
That is the nature of cutting edge
investigation.
However, high quality research
combined with creative thinking and
innovative design, means that
unconventional solutions to the
problems brought to us by our clients
are always a possibility. That, in turn,
means the possibility of step-change
and consequent advantage, is always
present. This in a nutshell, is the
bedrock of all that we do.
There are many examples we could cite
since the firm was founded, ranging
from the creative use of concrete,
through geotechnics, fire engineering
and advanced computer simulation.
All these contributions were made
possible because the firm has
employed members of staff capable of
working in the commercial environment
at the leading edge of research and
development.
66
John Miles
Head of Energy, Resources and
Industry Market
Research Review | May 2010
Editors:
Geraldine Ralph, Anna Goswell
Designer:
Terry Nicholls
Published by:Arup, 13 Fitzroy Street, London, WIT 4BQ
Printed by:
Fulmar Colour
ulmar print using alcohol-free printing
F
methods and vegetable-based inks
were used throughout. Solvents and
waste used in the printing process are
recycled. Fulmar Colour Printing is a
CarbonNeutral Company, and is certified
to ISO 14001, and is a FSC and PEFC
accredited company.
ISBN number:
978-0-9516602-8-7
Research Team
Regional Research Champions
Arup
Prof. Jeremy Watson
Americas
13 Fitzroy Street
Director, Global Research
Tim Keer
London W1T 4BQ
T +44 (0)20 7755 2235
E [email protected]
E [email protected]
Dr Marta Fernandez
Richard Hough
T +44 (0)20 7636 1531
F +44 (0)20 7580 3924
W www.arup.com
Research Relationships Manager
E [email protected]
T +44 (0)20 7755 5105
E [email protected]
East Asia
Dr Jennifer Schooling
Australasia
Dr Ricky Tsui
E [email protected]
Research Business Manager
Europe
T +44 (0)20 7755 2912
E [email protected]
Dr Mikkel Kragh
Geraldine Ralph
UK-MEA
Gavin Davies
Project Coordinator
T +44 (0)20 7755 6433
E [email protected]
E [email protected]
E [email protected]