H2FCSUPERGEN

WP2: Hydrogen systems
Sheila Samsatli, Nouri Samsatli, Nilay Shah
Hub Research and Advisory Board Meeting
University of Newcastle
30-31st July 2014
Our models that are relevant to hydrogen
1. Modelling of energy storage in a whole energy
system
•
Spatio Temporal Model for Energy Systems
(STeMES)
2. Hydrogen infrastructure design
•
Hydrogen Supply Chain Model (HSC)
3. Potential of producing renewable hydrogen
from wastes and biomass in the UK
• ETI Biomass Value Chain Model (ETI BVCM)
Addressing the challenge of modelling
energy storage in a whole energy
system
Future energy targets are driving the deployment
of renewable technologies
London Array – the world’s largest
offshore wind farm
A solar farm
Aquamarine Power’s Oyster Wave Energy
Photo credits: http://www.londonarray.com/media-centre/image-library/offshore/
http://www.solarselections.co.uk/blog/wp-content/uploads/2012/09/ROC-solar-farm.jpg
The future role for energy storage in the UK, Energy Research Partnership, 2011
The intermittency challenge
Source: The future role for energy storage in the UK, Energy Research Partnership, 2011
•
•
•
Renewable energy is generated when it’s not needed
Large dip in generation during high demand
Peaking generators, e.g. gas turbines, used to balance supply and
demand are expensive and produce GHG emissions
Energy storage solution
Photo credit: The future role for energy storage in the UK, Energy Research Partnership, 2011
Royal Institution Battery 1807
•
•
•
Llyn Stwlan reservoir
Hot water storage tank in the
basement of a smart house
Enables “wrong-time” energy generation from intermittent renewables
Reduces need for peaking generators
Improves energy use efficiency
Specific energy (MJ/kg)
140
H2, gas
at 70 MPa
120
H2, liquid
Conventional fuels
Synthetic fuels
100
Electrochemical
80
60
Mechanical
NG
at 250 MPa
NG, liquid
Gasoline
Crude oil
40
Coal
Ethanol
Dry wood Methanol
20
Batteries: Pb-acid, Ni-Cd, Li-ion
0
Hydro, CAES, Flywheel
0
10,000
20,000
30,000
40,000
Energy density (MJ/m 3)
Reference: Wikipedia
50,000
Source: Williams, R. Is liquid air the missing link in energy storage. Focus April 2013.
Modelling challenge: the dynamics of storage technologies occur over short
time scales (<hourly), very different from the time interval in energy system
planning models (>yearly)
Tractability is an issue!
Challenges
• We need a dynamic energy system model with a very
wide range of time scales
• Planning: years or decades
• Seasonal: variations in demands and availability
• Hourly (or shorter):
• Dynamics of storage technologies
• Variations in demands, intermittency of renewable resources
• Still need to model spatial aspects
• Demands and availability depend on location
• Determine location/size of technologies and storage facilities
• Requires integer variables
• Transport of resources (centralised vs. distributed)
• Very large scale model
Hydrogen Supply Chain (HSC) model
Natural gas
Steam
methane
reforming
Natural gas
Steam Methane
Reforming
Coal
Coal
gasification
Coal
Coal gasification
Biomass
Biomass
gasification
Biomass
Biomass gasification
Electricity
Electrolysis
Electricity
Electrolysis
Liquid
hydrogen
Gaseous
Hydrogen
Tanker truck
Tanker truck
Tanker truck
Tanker truck
Railway
tanker car
Liquid
hydrogen
storage
Railway
tanker car
Railway tanker car
Liquid H2
storage
Railway tanker car
Tube trailer
Compressed
hydrogen gas
storage
Tube trailer
Railway
tube car
Railway tube car
Gaseous H2
storage
Fuelling
stations
(liquid)
Fuelling station
Tube trailer
Railway tube car
Railway
tube car
Fuelling
stations
(gas)
Fuelling station
Tube trailer
Spatial element: Great Britain represented by 34 108×108 km2 square cells
Temporal element: 2015-2044 divided into 5 6-year periods
Last upgrade: made it a dynamic model with time intervals of 4 seasons in a year and
4 6-hr periods in a day
Example case study
Summer 2015-2020
Summer 2027-2032
Summer 2039-2044
Winter 2039-2044
No. of variables > 1M, No. of constraints > 0.5M, Integers = 15k
Took 3 days to solve full MIP!
Inventory profile for a whole year
London and the South East (cell 29) in 2039-2044
Winter
Spring
Summer
Autumn
Week
Limitations of the HSC model
• Multi-echelon model
• Pathways inflexible
• Distribution within cells too complex
• Too many binary variables
• Big M formulation
• Too large to be extended
• Adding a pipeline transport mode resulted in intractable
problems
• Difficult to add new technologies and resources
• Still not enough time intervals
Back to basics:
A very simple MILP model with storage
Resource balance: I  U   P    Q
Q  D
rch
rch
rp
pch
p
rc ' ch
Production capacity constraint: Ppch  NPpc p
c'
max
p
rcc ' h
p, c, h
rch
 Srch  Src, h1 r , c, h
A simple model is
intractable for the time
horizon needed for a
planning model!
max
Resource availability constraint: U rch  urch
r , c, h
Storage capacity constraint: Srch  srcmax r , c, h
Objective function definition
r = 3
p = 4
 h
c = 14 No. of integer variables = 56
h - contiguous hourly interval
No. of variables
No. of constraints
Solution time (s)
24
6,139
3,703
7
168
42,427
25,879
744
720
181,531
110,887
22,248
2,160
544,411
332,647
>155,520
8,760
2,207,611
1,349,047
??!
STeMES
• Spatio Temporal Model for Energy Systems
• RTN representation of energy pathways
• same framework as the BVCM and TURN model in SynCity
toolkit
•
•
•
•
MILP formulation
Efficient representation of time
Detailed storage formulation
Transport losses modelled in detail
Hierarchical non-uniform time discretisation
Years
y
y=1
ss
Seasons
yy
s=1
dd
Days
Hours
h=1
d=1
hh
s
d
h
Total number of time intervals T = |y| × |s| × |d| × |h|
e.g. for one year, T = 1 × 4 × 2 × 24 = 192 << 8760
Without storage – very easy!
With storage – extra variables for initial inventories; extra constraints to link
inventories within and between time levels
Elec. Grid
(not currently
modelled)
Renewable
potential
Rail
Truck
Electrolysis
To another
location
Electricity
Pipeline
Fuel cell
CGH2
Demand
CGH2
From
another
location
Liquefaction
LH2
Regasifier
Resource interconversion is modelled
using tasks to represent conversion
technologies and states to represent
resources (STN).
Underground
storage
CGH2 Storage
Metal
hydride
LH2 storage
Transport
Transport task – used to model connections between cells
i
i'
r1
r2
r3
r1
1
Transport
r2

r2
r3
Resource r2 is transported from cell i to cell i’, which requires r1 from cell
i and results in waste r3 being generated in both cells
Storage
An example set of storage tasks to store resource r1.
r2
Put
1
r1
r3
Hold

Inventory
r1
Get
r5
The “put” task transfers r1 from the cell to the store, requiring some r2 and producing some
wastes r3 (e.g. CO2). The “hold” task maintains r1 in storage, which also requires some r2 but
at less than 100% efficiency, the losses being converted to r3. Finally, the “get” task retrieves
r1 from storage and delivers it to the cell, requiring some r5.
STeMES prototype
•
•
•
Developed and tested for a hypothetical island of 14 50×50km cells
Wind generation installed at two locations
Choice of storage technologies
• Salt cavern available for use as hydrogen storage facility
• Other H2 storage technologies: gaseous (tank), liquid, metal hydride
•
Target: transport demand to be met by hydrogen (CGH2)
•
Objective: Minimum cost
•
Decisions
• Location and size/number of hydrogen
production and storage facilities
• Operation of production facilities
• Operation of storage facilities
• when to charge and discharge
• Transportation of hydrogen
12
13
10
11
6
7
8
3
4
5
Iltasmas
9
Cavern
1
2
14
Spatio-temporal input data
Hourly demand for CGH2 (MW) in cell 07
200
Hydrogen Demand (MW)
Ave. daily demand for CGH2 (MW)
180
160
140
120
100
80
60
40
20
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time of day
Hourly wind generation potential (MW)
Spring WD
Spring WE
Summer WD
Summer WE
Autumn WD
Autumn WE
Winter WD
Winter WE
Wind energy potential (MW)
rgy potential (MW)
600
500
400
300
200
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time of day
Results |y|= 1, |s|= 4, |d|= 2, |h|= 24
17
21
h = 113
824
16
515
76
22
23
19
18
14
12
11
234910
20
Demand (MW)
1
Demand (MW)
Snapshot of the network during
weekday (d=1) in spring (s=1)
2
3
4
5
6
7
8
6
7
8
9
10
11
9
10
11
12
13
14
200
150
100
50
0
1
2
3
4
5
12 13 14 15
Time of day
16
17
18
19 20
21
22
23
24
Storage charging rate (MW)
Storage charging
rate (MW)
250
200
150
100
50
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
12 13 14
Time of day
15
16
17
18
19
20
21
22
23
24
Time of day
Storage discharging rate (MW)
Hourly transport of CGH2 by pipeline
Installed electrolyser capacity (3 small units)
Installed underground storage capacity
Storage
discharging rate
(MW)
250
200
150
100
50
0
1
2
3
4
5
6
7
8
9
10
11
Resource Utilisation
Without storage, the scenario is infeasible
BUT with storage, only a fraction of the available wind energy is needed!
Wind availability (MW)
600
Spring WD
Summer WE
Winter WD
Spring WE
Autumn WD
Winter WE
Wind utilisation (MW)
Summer WD
Autumn WE
Wind energy utilisation (MW)
Wind energy potential (MW)
500
400
300
200
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time of day
Spring WD
Summer WE
Winter WD
600
Spring WE
Autumn WD
Winter WE
Summer WD
Autumn WE
500
400
300
200
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time of day
Cell 1
Cell 1
CGH2 production rate (MW)
•
The rate of operation of
electrolyser is effectively constant
CGH2 Production rate (MW)
•
•
Spring WD
Summer WE
Winter WD
Spring WE
Autumn WD
Winter WE
Summer WD
Autumn WE
50
25
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time of day
Cell 1
1
160
319
478
637
796
955
1114
1273
1432
1591
1750
1909
2068
2227
2386
2545
2704
2863
3022
3181
3340
3499
3658
3817
3976
4135
4294
4453
4612
4771
4930
5089
5248
5407
5566
5725
5884
6043
6202
6361
6520
6679
6838
6997
7156
7315
7474
7633
7792
7951
8110
8269
8428
8587
Amount of CGH2 in the storage (GWh)
Results |y|= 1, |s|= 4, |d|= 2, |h|= 24
Hourly inventory of CGH2 in the storage for a whole year
15
Spring
10
Summer
Hour (contiguous)
Winter
5
Autumn
0
Benchmarking
No. of integers = 336, relative tolerance = 0.1%
|s|
|d|
|h|
a
1
1
1
24
32,823
98,631
7
b
1
1
2
24
64,959
196,911
272
c
1
2
2
24
129,399
393,639
2,543
d
1
4
2
24
258,279
787,095
69,480
No. of
variables
No. of
constraints
Solution
time (s)
All runs determined 45.4 MW of electrolysis capacity installed in cells 1 and 14,
H2 transport by pipeline and underground storage.
However, the runs with fewer time intervals underestimated the storage capacity.
15
Run c – (2 season types)
10
5
15
Run d – (4 season types)
10
5
0
Hour (contiguous)
1
160
319
478
637
796
955
1114
1273
1432
1591
1750
1909
2068
2227
2386
2545
2704
2863
3022
3181
3340
3499
3658
3817
3976
4135
4294
4453
4612
4771
4930
5089
5248
5407
5566
5725
5884
6043
6202
6361
6520
6679
6838
6997
7156
7315
7474
7633
7792
7951
8110
8269
8428
8587
0
1
160
319
478
637
796
955
1114
1273
1432
1591
1750
1909
2068
2227
2386
2545
2704
2863
3022
3181
3340
3499
3658
3817
3976
4135
4294
4453
4612
4771
4930
5089
5248
5407
5566
5725
5884
6043
6202
6361
6520
6679
6838
6997
7156
7315
7474
7633
7792
7951
8110
8269
8428
8587
Amount of CGH2 in the storage (GWh)
•
|y|
Amount of CGH2 in the storage (GWh)
•
Run
ID
Hour (contiguous)
If underground storage is not an option
|y|= 1, |s|= 1, |d|= 2, |h|= 24
Demand (MW)
Demand (MW)
d=1
h=5
200
cell 1
2
6
7
3
4
5
6
7
10
11
8
9
10
11
12
13
14
100
0
1
2
3
4
5
8
9
12 13 14
Time of day
15
16
17
18
19
20
21
22
23
24
15
16
17
18
19
20
21
22
23
24
19
20
21
22
23
24
Storage charging
rate (MW)
Storage charging rate (MW)
d=1
h = 19
cell 1
200
7
14
150
100
50
0
1
2
3
4
5
6
7
8
9
10
11
12 13 14
Time of day
Storage
discharging rate
(MW)
Storage discharging rate (MW)
•
Hourly transport of CGH2 by pipeline
Installed electrolyser capacity (3 small units)
Installed CGH2S capacity (1 small unit each)
cell 1
200
7
14
150
100
50
0
1
2
3
4
5
6
7
8
9
10
11 12 13
Time of day
14
15
16
17
18
Network is more distributed with small compressed
gaseous H2 storage technologies installed in cells
where generation are located and cell the with highest
demand.
Next steps
• Real case studies (e.g. UK scenarios)
• Add more resources and technologies
• Exploit the full potential of the nonuniform hierarchical discretisation
method
• E.g. Use fewer non-uniform hourly
intervals
• Additional decomposition methods
• Benders decomposition did not work
• Test in-house approaches
Source: The Electricity Storage Network. Development of
electricity in the national interest. May 2014
Conclusions
• Storage is a key-enabling technology for meeting the energy
demands using renewable resources
• Without storage the example problem is infeasible
• With storage, only a small fraction of available primary resource is used
and the generation technology operates effectively at a constant rate
• To model storage accurately, hourly or shorter intervals are
needed
• In the example, four seasons are also needed
• Model tractability is a big challenge
• Even the simplest model cannot handle a whole year at an hourly level
• Hierarchical time decomposition allows a whole year (and longer
planning horizon) to be considered by exploiting periodicity in the data
Potential of producing renewable
hydrogen from wastes and biomass in
the UK
Pathways from biomass to hydrogen
Bioresource
Biological
Anaerobic
digestion
Fermentation
Thermochemical
Metabolic
processing
CH3CH2OH/CO2
Gasification
H2/CO
CH4/CO2
CH4/CO2
CH4/CO2
Synthesis
Pyrolysis
Severe
CH4/CO2
Bio-shift
Reforming
shift
Reforming
shift
CH3OH/CO2
Photobiology
Reforming
shift
Pyrolysis
H2/CO2
High pressure
aqueous
H2/C
H2/CO2
H2/O2
H2/CO2
H2/CO2
Reforming
shift
H2/CO2
•
Reforming of methane (produced from biomass) to hydrogen
•
Electrolysis using electricity produced from biomass
Reforming
shift
H2/C
H2/CO2
Biomass Value Chain Model (BVCM)
• Can be used to understand how bioenergy systems can be
implemented without negative sustainability-related impacts
• Models pathway-based bioenergy systems over five decades
(from 2010 to 2050)
• Supports decision-making around land use, interactions with
food production and acceleration of bioenergy technologies
Energy demand
resource,
technology and
logistic data
Optimal bioenergy value chain
structure
• allocation of crops to available land
• choice of technologies
• energy provision ( electricity, heat,
hydrogen, transport fuels,
biomethane)
• transport networks required
Model Elements: Time, Space and Climate
Time
• 5 decades
• Up to 4 seasons per year
Spatial representation
• UK divided into 157 square cells of
length 50km
Climate scenarios
• Low and medium scenarios
Model Elements: Land Use
Four levels of land “aggression”
Level 1 as “easy, established technology”
• Arable land
• Heterogeneous agricultural land (e.g. Non-permanent crops associated with
permanent crops )
Level 2 as “pioneering plant establishment”
• Shrub and/or herbaceous vegetation association, e.g. natural grassland
• Open spaces with little or no vegetation
Level 3 as “challenging techno-ecological and economic land use change”
• Permanent crops, e.g. fruit trees and berry plantations
• Pastures
Level 4 as “last resort”
• Forests
• Artificial non-agricultural vegetated areas (e.g. green urban areas and parks)
Model Elements: Land Use
Level 1
Level 2
Level 3
Level 4
3
4
Model Elements: Bioresources
• Biomass data (yield, cost, and emissions) have been typically
estimated at high resolution (1x1km)
• High resolution data have been aggregated at a scale
adequate for optimisation (50x50km)
Waste potential
Keys:
2010s
2020s
2030s
2040s
2050s
Model Elements: Technologies
• The BVCM currently includes 72 distinct technologies, some
with multiple scales:
 pre-treatment and densification technologies
 technologies for hydrogen production
 technologies for liquid and gaseous fuel production
 technologies for heat, power, and combined heat and power generation
 waste to energy technologies
 carbon capture and storage technologies
Model Elements: Bioresources
•
•
•
•
•
•
•
Miscanthus
Short Rotation Coppice (SRC) – Willow
Winter wheat
Oilseed rape
Sugar beet
Short Rotation Forestry (SRF)
Long Rotation Forestry (LRF)
The BVCM currently includes 94 resources, comprising
bioresources, intermediates, final products, by-products and
wastes.
Example of a Resource-Technology Chain: SRC Willow
Gaseous fuels
Anaerobic digestion
Heat & power
Syngas
Biogas upgrading
Biomethane
Gasification
DME
Gasification + bioSNG
Gasification + DME
Hydrogen
Biomethane (for grid)
Gasification + H2
Boiler combustion (heat)
Syngas boiler
Biodedicated
steam cycle (CHP)
Electricity
Biodedicated
steam cycle (electricity)
Cofired steam cycle
(electricity)
Cofired steam cycle
(CHP)
Gas compression
Hot water
Electricity &
Steam
Stirling engine
Organic Rankine cycle
IC engine
Electricity &
Hot water
Gas turbine
Dedicated biomass IGCC
SRC (willow) resources
Co-fired IGCC
SRC (Willow) Chips
SRC (Willow)
Pellets
SRC (Willow)
Torrefied chips
SRC (Willow)
Torrefied pellets
Liquid fuels
First gen ethanol
First gen biodiesel
First gen butanol
Pretreatment
Lignocellulosic ethanol
Ethanol
Lignocellulosic butanol
Butanol
Gasification + FT diesel
FT diesel
Gasification +
methanol catalysis
Methanol
Gasification +
mixed alcohol process
Higher alcohols
Chipping
Pellettising
Legend
Torrefaction
Gasification +
syngas fermentation
Torrefaction +
pelletising
Pyrolysis oil upgrading
Upgraded pyrolysis oil
Hydrotreatment
Naphta
Gasification +
FT jet/diesel
FT jet
Sugar-to-diesel
Fuel gas
Harvested resource
Oil extraction
Intermediate resource
Pyrolysis
Agricultural co-product
Final product
Technology
Sugar extraction
Pyrolysis oil
Wood-to-diesel
Model Elements: Logistics
4 transport modes: ship, road, rail, inland waterways
UK Ports
Road Tortuosity
Railway Length
Inland Waterways
Key questions
• How much hydrogen can be produced from wastes and
locally-available biomass resources without undermining
food production?
• How much would it cost to meet the transport fuel
demand using carbon neutral hydrogen from biomass?
• What is the optimal pathway to produce hydrogen from
biomass?
Transport fuel demand
Time
Period
2010s
2020s
2030s
Transport Fuel
Demand (TWh/yr)
7.13
28.52
71.30
2040s
121.20
2050s
142.59
Case study 1: Hydrogen from locally-available biomass
• Maximise H2 production
• Land constraint: allocate 2% of land available in Level 1
and 15% in Levels 2-4 to bioenergy
 (2.35 out of 22.28 Mha )
• No import of resources abroad
Case study 1: Hydrogen from locally-available biomass
No constraint on GHG emissions
Carbon neutral
55% of 2050
transport demand
44% of 2050
transport demand
Total = Bn£ 32.7
Total = Bn£ 73.7
Case study 1: Hydrogen from locally-available biomass
No constraint on GHG emissions
Carbon neutral
Case study 1: Hydrogen from locally-available biomass
No constraint on GHG emissions
Carbon neutral
Case study 1: Hydrogen from locally-available biomass
Land area allocated to SRC Willow and H2 Gasification plants built
Carbon neutral case
Total area
allocated
for crops
Area for
SRC
Willow
H2 Gasification
H2 Gasification
with CCS
2010s
2020s
2030s
2040s
2050s
Case study 1: Hydrogen from locally-available biomass
CO2 capture, transport and sequestration, and waste utilisation
Carbon neutral case
CO2 captured
CO2 transport
CO2 sequestered
Waste wood
utilisation
2010s
2020s
2030s
2040s
2050s
Case study 2: 100% transport fuel demand from bio-hydrogen
No constraint on GHG emissions
Total = Bn£ 51.5
Carbon neutral
Total = Bn£ 62.9
Case study 2: 100% transport fuel demand from bio-hydrogen
Carbon neutral case
What about biofuels?
Total = Bn£ 71.6
Least cost, carbon neutral
scenario if biofuels were to satisfy
100% of transport fuel demand
Conclusions
• Carbon neutral hydrogen produced from wastes and locallyavailable biomass can meet up to 44% of the transport fuel
demand in 2050 without undermining food production.
• CCS technologies and waste play an important role in meeting
the demands for carbon neutral hydrogen from biomass.
• SRC Willow and gasification plants will most likely be used in
producing hydrogen from biomass.
• Import of biomass resources is necessary if we are to satisfy all
transport fuel demand using hydrogen from biomass.