st century national energy & transportation ...jun 07, 2011 · 21st century national energy...
TRANSCRIPT
21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
httpwwweceiastateeduresearchresearch-projectsnetscore-21htmlhttpwwwyoutubecomNETSCORE21
A project funded by the US NSF via the 2008 Solicitation for Emerging Frontiers in
Research amp Innovation - Resilient and Sustainable Infrastructures (EFRI-RESIN)
1
James McCalleyHarpole Professor of
Electrical amp Computer EngineeringIowa State UniversityPSERC Webinar June 7 2011
Robert Brown ME
JimMcCalley EE
Dionysios Aliprantis EE
Nadia Gkritza CE
Lizhi Wang IE
Arun Somani CpE
DiWu EE
JinxuDing CpE
YingZhou IE
DiegoMejia EE
JosephSlegers EE
CatherineRentziou CE
ZhaoyangDuan IE
JoseVillarrel EE
JoshGifford ME
KeithJohnson EE
McNair Scholar
LizbethGonzales EE
Yang GuEE
EduardoIbanez EE
Acknowledgement to NETSCORE21 Faculty amp Students
Steve Lavrenz
CE
Jeff BrownBusiness
Qi QihuiEE
2
Eirini KastrouniCE
1 Objective and orientation2 Modeling approach 3 Data4 Results5 Current efforts6 Conclusions
Presentation Outline
3
OBJECTIVE OF WORK DESCRIBED IN TODAYrsquoS SEMINAR
Provide 40-year national modeling processfor energy and transportation systems
Orientation Long-term multi-sector (fuel electric transportation) national planning
4
bull A way to probe future infrastructure trajectories
bull Separates ldquogoodrdquo from ldquobadrdquo choices
bull Informs societal dialogue and political debate
sustainable resources depletable resources
ENERGY SYSTEM
TRANSPORTATION SYSTEM
2050
ENERGY SYSTEM
TRANSPORTATION SYSTEM
TODAY
5
Orientation Long-term multi-sector (fuel electric transportation) national planning
NETPLAN V1
Evolutionary algorithmSelects new solution population based dominance and crowding in terms of
cost sustainability resiliency
Investment biases minimum invest-ments subsidies emission limits
Multiobjective optimization
NSGA-II Search amp selection
Evaluation(fitnessfunctions)
6
SustainabilityMetrics
ResiliencyMetrics
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Conceptual Cost-Minimization Model
7
LP Cost Minimization Model Features
8
Commodity amp
passenger networks load energy system
Energy loads commodity transportsystem
Electric can be modeled with DC power flow
Stor
age
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Robert Brown ME
JimMcCalley EE
Dionysios Aliprantis EE
Nadia Gkritza CE
Lizhi Wang IE
Arun Somani CpE
DiWu EE
JinxuDing CpE
YingZhou IE
DiegoMejia EE
JosephSlegers EE
CatherineRentziou CE
ZhaoyangDuan IE
JoseVillarrel EE
JoshGifford ME
KeithJohnson EE
McNair Scholar
LizbethGonzales EE
Yang GuEE
EduardoIbanez EE
Acknowledgement to NETSCORE21 Faculty amp Students
Steve Lavrenz
CE
Jeff BrownBusiness
Qi QihuiEE
2
Eirini KastrouniCE
1 Objective and orientation2 Modeling approach 3 Data4 Results5 Current efforts6 Conclusions
Presentation Outline
3
OBJECTIVE OF WORK DESCRIBED IN TODAYrsquoS SEMINAR
Provide 40-year national modeling processfor energy and transportation systems
Orientation Long-term multi-sector (fuel electric transportation) national planning
4
bull A way to probe future infrastructure trajectories
bull Separates ldquogoodrdquo from ldquobadrdquo choices
bull Informs societal dialogue and political debate
sustainable resources depletable resources
ENERGY SYSTEM
TRANSPORTATION SYSTEM
2050
ENERGY SYSTEM
TRANSPORTATION SYSTEM
TODAY
5
Orientation Long-term multi-sector (fuel electric transportation) national planning
NETPLAN V1
Evolutionary algorithmSelects new solution population based dominance and crowding in terms of
cost sustainability resiliency
Investment biases minimum invest-ments subsidies emission limits
Multiobjective optimization
NSGA-II Search amp selection
Evaluation(fitnessfunctions)
6
SustainabilityMetrics
ResiliencyMetrics
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Conceptual Cost-Minimization Model
7
LP Cost Minimization Model Features
8
Commodity amp
passenger networks load energy system
Energy loads commodity transportsystem
Electric can be modeled with DC power flow
Stor
age
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
1 Objective and orientation2 Modeling approach 3 Data4 Results5 Current efforts6 Conclusions
Presentation Outline
3
OBJECTIVE OF WORK DESCRIBED IN TODAYrsquoS SEMINAR
Provide 40-year national modeling processfor energy and transportation systems
Orientation Long-term multi-sector (fuel electric transportation) national planning
4
bull A way to probe future infrastructure trajectories
bull Separates ldquogoodrdquo from ldquobadrdquo choices
bull Informs societal dialogue and political debate
sustainable resources depletable resources
ENERGY SYSTEM
TRANSPORTATION SYSTEM
2050
ENERGY SYSTEM
TRANSPORTATION SYSTEM
TODAY
5
Orientation Long-term multi-sector (fuel electric transportation) national planning
NETPLAN V1
Evolutionary algorithmSelects new solution population based dominance and crowding in terms of
cost sustainability resiliency
Investment biases minimum invest-ments subsidies emission limits
Multiobjective optimization
NSGA-II Search amp selection
Evaluation(fitnessfunctions)
6
SustainabilityMetrics
ResiliencyMetrics
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Conceptual Cost-Minimization Model
7
LP Cost Minimization Model Features
8
Commodity amp
passenger networks load energy system
Energy loads commodity transportsystem
Electric can be modeled with DC power flow
Stor
age
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Orientation Long-term multi-sector (fuel electric transportation) national planning
4
bull A way to probe future infrastructure trajectories
bull Separates ldquogoodrdquo from ldquobadrdquo choices
bull Informs societal dialogue and political debate
sustainable resources depletable resources
ENERGY SYSTEM
TRANSPORTATION SYSTEM
2050
ENERGY SYSTEM
TRANSPORTATION SYSTEM
TODAY
5
Orientation Long-term multi-sector (fuel electric transportation) national planning
NETPLAN V1
Evolutionary algorithmSelects new solution population based dominance and crowding in terms of
cost sustainability resiliency
Investment biases minimum invest-ments subsidies emission limits
Multiobjective optimization
NSGA-II Search amp selection
Evaluation(fitnessfunctions)
6
SustainabilityMetrics
ResiliencyMetrics
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Conceptual Cost-Minimization Model
7
LP Cost Minimization Model Features
8
Commodity amp
passenger networks load energy system
Energy loads commodity transportsystem
Electric can be modeled with DC power flow
Stor
age
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
bull A way to probe future infrastructure trajectories
bull Separates ldquogoodrdquo from ldquobadrdquo choices
bull Informs societal dialogue and political debate
sustainable resources depletable resources
ENERGY SYSTEM
TRANSPORTATION SYSTEM
2050
ENERGY SYSTEM
TRANSPORTATION SYSTEM
TODAY
5
Orientation Long-term multi-sector (fuel electric transportation) national planning
NETPLAN V1
Evolutionary algorithmSelects new solution population based dominance and crowding in terms of
cost sustainability resiliency
Investment biases minimum invest-ments subsidies emission limits
Multiobjective optimization
NSGA-II Search amp selection
Evaluation(fitnessfunctions)
6
SustainabilityMetrics
ResiliencyMetrics
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Conceptual Cost-Minimization Model
7
LP Cost Minimization Model Features
8
Commodity amp
passenger networks load energy system
Energy loads commodity transportsystem
Electric can be modeled with DC power flow
Stor
age
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
NETPLAN V1
Evolutionary algorithmSelects new solution population based dominance and crowding in terms of
cost sustainability resiliency
Investment biases minimum invest-ments subsidies emission limits
Multiobjective optimization
NSGA-II Search amp selection
Evaluation(fitnessfunctions)
6
SustainabilityMetrics
ResiliencyMetrics
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Conceptual Cost-Minimization Model
7
LP Cost Minimization Model Features
8
Commodity amp
passenger networks load energy system
Energy loads commodity transportsystem
Electric can be modeled with DC power flow
Stor
age
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Conceptual Cost-Minimization Model
7
LP Cost Minimization Model Features
8
Commodity amp
passenger networks load energy system
Energy loads commodity transportsystem
Electric can be modeled with DC power flow
Stor
age
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
LP Cost Minimization Model Features
8
Commodity amp
passenger networks load energy system
Energy loads commodity transportsystem
Electric can be modeled with DC power flow
Stor
age
LP-Cost MinimizationSelects investments time location over 40 years
for nationrsquos energy amp transportation systems
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Energy system modeling for cost minimization model
bull Generalized flow transportation model
bull Commodity energy
bull Pathsndash Electric transmission
ndash Gas pipelines
ndash Liquid fuel pipelines
ndash Conversion
bull Decision variablesndash Flow across the system
ndash Capacity investment in arcs9
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Transportation modelingbull Multicommodity flow
ndash Coal cereal grains foodstuffs chemicals gravel woodndash Routes fixedndash Arc demand forecasted
bull Infrastructurendash Highway railway ports
bull Fleetndash Trains trucks barges
bull Decision Variablesndash Amount of each arcrsquos freightallocated to each possible modendash Investment on infrastructure and fleet
bull Passenger transportation not fully developed yet
10
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Mathematical formulation for cost minimization problem
11
Minimize operational and investment cost
Meet energy demand
DC power flow
Meet electric peak demand
Meet transportation demand
Max fleet capacity
Max transportation infrastructure capacity
Energy flows and investments
Transportation flows and investments
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Compact notation and decomposition
12
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Every mode of transportation produces a demand in the energy networks
Transportation system loading on energy
MWHR = MWHRTON times TON
13
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
ldquoEnergy commoditiesrdquo (eg coal) are represented in the transportation network (as transported tons) and the energy network (as MWh) Both flows are coordinated
Energy system loading on transportation
TONS = TONSMWHR times MWHR
14
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Summary of networks represented in cost-minimization problem
15
Energy and energy commodity networks have demand specified at nodes whereas freight and passenger networks have demand specified at arcs Allocation of transportation load across modes (infrastfleet) is decision
People or vehicles
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Additional ModelingComputational Attributes
16
bull DC flow representation available for transmission
bull Different time steps modeled for different networks
bull NETPLAN is C++ pre post-processor coordinator for NSGA-II and CPLEX optimizerbull Load for each elect syst
time interval (month) may be segmented to reflect peak amp off-peak conditions
bull Source code httpgithubcomeibanezNETPLAN(but no support)
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
17
US data set
Petroleum
Natural Gas
Coal
Emissions
Electric Power Generation
ImportExport
Transmission
End Use
EIA Forms 7A 176 191 857 895
MSHA Form 7000-2
FERC Forms 423 549B 580
DOE NMA DOTFHWABTS FRAAAR OFE API
DOEEIA
EPA (eGRID)
DOE
EIA Form 767 860 906
FERC Form 423
ISOs
FERC Form 715EIA Form 412
NERC ISOs
DOE
EIA Form 826 861 FERC Form 714
NERC ISOs
NEBCDOEOFPISOs
Commodity Flow Survey
Transportation energy data book
EIA ldquoCoal Transpor-tation Rates amp Trendsrdquo
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
NETSCORE21 Technology DatabaseTechnologies1 Nuclear2 Pulverized Coal3 NGCC4 CT5 Hydro6 Inland Wind7 Oil8 IGCC9 Solar PV10 Fuel Cell
Attributes (Low Med Hi)bull Invest Cost (million$MW)bull Fixed OampM Cost ($kW-yr)bull Variable OampM Cost ($MWh)bull Heat Rate (MMBTUMWh)bull Calculated Efficiency ()bull Fuel Use NOx (kgMWh)bull Fuel Use SOx (kgMWh)bull Fuel Use PM (kgMWh)bull Fuel Use NMVOC (kgMWh)bull Fuel Use GHG (kgMWh)bull Construction GHG (kgMWh)bull Direct Land Usage (m2MWh)bull Lifetime (years)bull LeadLag Time (years)bull FOR ()bull Capacity Factor ()bull Sources
11 Geo Thermal12 Solar Thermal13 MSW14 LF Gas Recovery15 IBGCC16 OTEC17 Offshore Wind18 Tidal Power19 IPCC20 Wave Power
18
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
19
Model implementation Energy
24 states comprise coal resourcesDemand is all power by stateCoal resources connected to all statesCoal network uses yearly step sizes
COALNat GASGulfTxCanadian resources amp storage modeledDemand nonpower (1 grwth) power by stateGas pipelines modeled between adjacent statesGas network uses monthly step sizes
ELECTRIC
Each NEMS region models 15 gen typesState demand trnsfrmd to regions (15 grwth) Trans modeled between adjacent regionsElectric network uses monthly step sizes
PETROLEUMHave not yet developed detailed model So now using single petroleum source node with unlimited supply
Petroleumsource
Diesel$380gal
Gasoline$400gal
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Model implementation Transportation
20
2008 Freight transport wo coal 2048 Freight transport wo coalFREIGHTModes are rail-diesel rail-electric and highway-dieselProjected freight demand is obtained from DOT ldquoCommodity Flow SurveyrdquoState-to-state freight transport is pre-fixed (no route optimization) added to coal transport as demanded by energy networkDistances capacities (based on existing demand) estimated for each arcLocational fuel demand based on terrain estimated for each mode (gal1000ton-miles)Transport network uses yearly time stepsPASSENGERModes are highway-gasoline and highway-PHEV20New vehicle sales based on (a) existing vehicle population distributed among 13
regions in proportion to electric demand (b) 12 year life (c) 1 annual growthAssumptions made on each vehiclersquos driving distance and electric gasoline demand
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Summary of cost-minimization model
21
bull 748394 variables 472920 constraints
bull ~17 minutessolution on 16 GHz processor 24 GB RAM
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
How to validate this modelbull Perform sensitivity analysis on solutionsbull Simulate past period of timebull Repeat analysis with other models
ndash NEMS ReEDS MARKALTIMES PRISM IPMbull Compare model results to those of other studies
ndash EIArsquos ldquoAnnual Energy Outlookrdquondash DOErsquos ldquo20 Wind Energy by 2030rdquondash NERCrsquos 10 year forecastndash Union of Concerned Scientists 2030 report (NEMS)ndash NAE 2035 reportndash NREL Renewable Energy Futures Reportndash EEI Potential Impacts of Env Regulation on US Gen Fleet
22
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Min cost solution
23
bull Strong investment in nuclear IGCC geothermal and on-shore windbull Dip in total capacity in years 25-28 due to retirement of NGCC and CTs (30
year lives assumed) compensated by heavy investment in windbull Investment in NGCC and CTs are high but little energy covers peak
Added gen capacity Total gen capacity
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Min cost solution
24
bull Nuclear NGCC and CTs show consistent investment levels across areasbull Distribution of remaining gen technologies mainly driven by wind CF
Gen capacity investment by regionCapacity factors
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Freight transport (millions tons)
25
Min cost solution Passenger transport (vehicles)
With no change in existing prices transportation growth occurs only in petroleum-based vehicles
With a doubling of gasoline prices PHEV purchases dominate
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Min cost solution Coal production (millions of tons)
26
GHG emissions (millions of tons)
bull Coal demand GHG emissions decrease as nuclear wind geothermal replace pulverized coal
bull As electricity demand increases following year 20 use of coal increases in both pulverized coal (low inv cost) and IGCC (low op cost)
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Min cost solution LMPs over 40 years by region
27
Avg LMPs for each decade by region
Prices become more spatially uniformas most economic resources are utilized in each region
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Min cost solution Energy production for different emissions reductions
28
0 reductionEnergy generated shifts from PC amp IGCC to geothermal wind (on+off-shore)
20 reduction 40 reduction
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Multiobjective Solver NSGA-II
29
NSGA-II evolutionary algorithm proposes candidate solutions in terms of minimum investment levels for certain technologies
Cost minimization with minimum investments produces technology portfolios and energy flows
Sustainability metrics are computed based on energy flows
Resiliency metrics are computed based on computed system failures tested with calculated portfolio
Metrics returned to NSGA-II next generation generated via tournament selection recombination mutation followed by sorting based on dominance and crowding
A solution dominates another one if all its objective values are equal or better and at least one of them is strictly better
Gives the Pareto-optimal front the set of solutions for which no objective value may be improved without degrading at least one other objective value
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Resilience Assessment
30
Concept Resilience must consider events and consequences which exhibit measureable changes with design variation
RESILIENCE Ability to minimize and recover from the consequences of an event
Extreme Events Simulate total failure of each of 14 major technologies at year 25Societal consequences Average the one year national operational cost increase across all 14 events with respect to the no-event case
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 10 11 12
Nod
al p
rice
Month
Nodal price at node k
With disruption
Without disruption
bull 40 yearsbull National
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Model Size and Computation Time
31
bull Min cost model has
748394 variables 472920 constraints
bull 20 solutionsgeneration
bull 82 generations
bull 472 hours computing on single CPU
bull Average min per LP solution 17
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
Pareto-Optimal Solutions after 200 Generations
32
S No Cost (M$) EmCO2 (Short ton) Resiliency (M$)1 437E+06 532E+10 337E+052 438E+06 527E+10 320E+053 443E+06 525E+10 362E+054 511E+06 507E+10 134E+045 513E+06 512E+10 133E+046 518E+06 502E+10 137E+047 524E+06 505E+10 127E+048 537E+06 491E+10 116E+049 556E+06 484E+10 108E+04
10 563E+06 479E+10 986E+0311 593E+06 461E+10 874E+0312 599E+06 444E+10 794E+0313 605E+06 451E+10 733E+0314 611E+06 442E+10 795E+0315 617E+06 424E+10 720E+0316 631E+06 431E+10 642E+0317 637E+06 416E+10 668E+0318 639E+06 397E+10 629E+0319 648E+06 384E+10 595E+0320 652E+06 403E+10 532E+03
Pareto Optimal Solutions from NSGA-II
6000
8000
10000
12000
14000 384
4244
4648
552
x 1010
5
52
54
56
58
6
62
64
66
x 106
Emission (Short tons)Resiliency (M$)
Cos
t (M
$)
Fig 10 Pareto front in 3-D solution space
Events For each 40 year investment strategy simulate total failure of each of 14 technologies at year 25Resiliency metric Averaged the 1 year operational cost increase across all 14 events with respect to the no-event case
Cos
ts (M
$) x
106
Least cost least resilient
Highest cost most resilient
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
33
Yearly Generation Investment and CO2 Emission for Most Resilient Solution
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
34
Yearly Generation Investment and CO2 Emission for Least Resilient Solution
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
35
Current Model Improvement Effortsbull Impact of variable generation
bull Cycling costs (increased maint amp FOR decreased life)bull Investment costs of more high-ramp capabilityCTs demand control storage large control areas
bull Transmission optimizationbull Emissions control equipment
bull Fluidized gas desulfurizationbull Carbon capture amp sequestration
bull Hydrogen production amp transportbull Data enhancement
bull Enhancement of generation amp transmission databull Liquid petroleum refining amp transport
bull Identification of key uncertainties amp modelingbull Deployment on parallelized HPCbull Improved passenger transport
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
36
Current Study Effortsbull What technologies and topologies should be used in designing a national electric transmission superhighway systembull What is the best mix of electricity petroleum and biofuels to supply our automotive needsbull To what extent can electric high-speed rail reduce energy use and transportation-related emissions while competing with air and highway travel
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-
37
Conclusionsbull We developed NETPLAN a computational model that is
bull multisector fuels electric and freightpassenger transportbull multiobjective cost resilience and sustainability metrics bull an optimization model (not equilibrium) amp so policy-drivingbull long-termnational and represents transmissiontransport
bull We conceive of large catastrophic Katrina-like events to define resilience in terms of their cost-consequencebull The model allows exploration of how different technolo-gies costs resilience and emissionsother environmental objectives affect long-term investment portfoliosbull NETPLAN is useful for identifying policy directions which balance cost resilience and environmental needsbull We need to make use of software tools which perform systematic engineering evaluation to peer into the future and appropriately guide legislative decision-making
- 21st Century National Energy amp Transportation Infrastructures Long-Term Planning for Cost Sustainability and Resilience
- Acknowledgement to NETSCORE21 Faculty amp Students
- Slide Number 3
- Slide Number 4
- Slide Number 5
- Slide Number 6
- Conceptual Cost-Minimization Model
- LP Cost Minimization Model Features
- Energy system modeling for cost minimization model
- Transportation modeling
- Mathematical formulation for cost minimization problem
- Compact notation and decomposition
- Transportation system loading on energy
- Energy system loading on transportation
- Summary of networks represented in cost-minimization problem
- Additional ModelingComputational Attributes
- Slide Number 17
- NETSCORE21 Technology Database
- Model implementation Energy
- Model implementation Transportation
- Summary of cost-minimization model
- How to validate this model
- Min cost solution
- Min cost solution
- Slide Number 25
- Min cost solution
- Min cost solution
- Min cost solution
- Multiobjective Solver NSGA-II
- Resilience Assessment
- Model Size and Computation Time
- Pareto-Optimal Solutions after 200 Generations
- Slide Number 33
- Slide Number 34
- Slide Number 35
- Slide Number 36
- Slide Number 37
-