Integrated Resource Planning with PLEXOS

Integrated Resource Planning with PLEXOS World Bank Thirsty Energy - A Technical Workshop in Morocco March 24 th & 25 th, 2014 Rabat, Morocco Dr. Randell M. Johnson, P.E. randell.johnson@energyexemplar.com

Integrated Planning with PLEXOS Multi-Sector Least Cost Planning Water Requirements for Power Plant Cooling Desalination Generation Capacity Planning Transmission Expansion Gas Pipeline Expansion Planning Renewables Integration Ancillary Services Commodity Storages Least Cost Optimizations Mixed Integer Programs Stochastic Optimizations Co-Optimizations 2

Water/Energy Tradeoffs or Opportunities Generate a power expansion plan considering aggregated water use limits When a desal can be the least cost options for both power supply and water production When to use one or the other type of cooling in power plant in response to environmental regulation or water constraints When to build a power plant in a different location given water constraints When to build a transmission line instead of a water pipe to locate the power plant where it makes sense to least cost expansion plan 4 April, 2014 Energy Exemplar 3

Country Electricity Demand (MW) Country Water Demand (MGD) Demand Duration Curves for Co-Optimized Expansion and Operation of Water and Electricity Sectors - Multi Sector CapEx and OpEx Least Cost Optimization - Primary and Secondary demand curve optimizations Percent Time Country Electricity Demand Country Water Demand 4 April, 2014 Energy Exemplar 4

Simplified Diagram of a Long Term Expansion Plan Optimization Problem Generator Type Expansion Decision Ctx Line 1 Load Ctx Line 2 Cooling Type Once Through Capital and O&M Costs - Build cost - Cooling Type - Efficiency - Substation - Others Ctx pp1 Ctx pp4 Ctx pp2 Ctx pp3 Recirculating Dry Thermal Plant Cpp1 Thermal Plant 1 Cpp4 Thermal Plant 4 Cpp2 Thermal Plant 2 Cpp3 Thermal Plant 3 Flow 1 Water Use C w1 Water Use Flow 4 C w4 Water Flow 2 Use Flow 3 C w2 Aqua Duct Vol 1 Vol 2 Vol 3 Cad 1 Flow ad 1 Water Use C w3 Coal Supply Capital Costs O&M Costs Fuel Costs Constraints OR Gas Supply Capital Costs O&M Costs Fuel Costs Constraints Fuel Supply Expansion Decision

Coal Plant Cooling Water Requirement water treatment production cost = water treatment penalty price x flow

Optimized Expansion Decisions Which power plant(s) to build and when considering water use costs and constraints? Which transmission interfaces to expand and when and how much? Should Aqua Duct be built, when, and what size? Should the Power Plant be build in a different water area? And build more power transmission? (co-optimize water pipe and power transmission as well) In the case above, could we consider as well the cost of building transport facilities for the fuels (the coal, gas) to the alternative power plant site? Include combined water-and-power production facilities, so that if one need to meet a water production target, the opportunity to co-generate water-and-power is exploited to minimize the power cost (INV and O&M) and meet the water targets (these water targets may for a separate and different type of water not necessarily the same or connected to the water in the 3 areas shown in the picture)

1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 103 109 115 121 127 133 139 145 151 157 163 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 103 109 115 121 127 133 139 145 151 157 163 Expansion with Stochastic Electric and Water Demands Stochastic Electric Demand Paths Stochastic Decomposition Stochastic Water Demand Paths Hours over a Week Hours over a Week 4 April, 2014 8

Multi Sector Least Cost Decision Making Chart shows the minimization of total cost of investments and of production cost Cost $ Objective: Minimize net present value of the sum of investment and production costs over time Investment cost/ Capital cost C(x) Total Cost = C(x) + P(x) As more investments made production cost trends down however investment cost trends up Minimize capital costs and production costs for electric, water, gas and other demands Minimum cost plan x Production Cost P(x) Investment x 9

Illustrative Least Cost Optimization Investment Cost Production Cost Y I T I Minimize BuildCost i Build i,y + ProdCost i Prod i,t + ShortCost Shortage t y=1 i=1 t=1 i=1 Individual Unit Build Cost Amount Built Individual Unit Individual Unit Production Cost Production VOLL Unserved Energy subject to Supply and Demand Balance: I i=1 Production Feasible: Prod i,t ProdMax i Expansion Feasible: Build i,y BuildMax i,y Integrality: Build i,y integer Reliability: LOLE(Build i,y ) LOLETarget Prod i,t + Shortage t = Demand t t i, t i, y 27 February, 2014 MA AGO y = for all y = year t = interval i = unit Y = Horizon This simplified illustration shows the essential elements of the mixed integer programming formulation. Build decisions cover generation, generation cooling types, water use costs, transmission, gas pipeline, coal transport, water pipe, as does supply and demand balance and shortage terms. The entire problem is solved simultaneously, yielding a true cooptimized solution. 10

PLEXOS Algorithms Chronological or load duration curves Large-scale mixed integer programming solution Deterministic, Monte Carlo; or State-of-the-art Stochastic Optimization (optimal decisions under uncertainty) Stochastic Variables Set of uncertain inputs ω can contain any property that can be made variable in PLEXOS: Load Fuel prices Electric prices Ancillary services prices Hydro inflows Wind energy, etc Discount rates Others Number of samples S limited only by computing memory First-stage variables depend on the simulation phase Remainder of the formulation is repeated S times Capacity Expansion Planning 11

Constraints Driving Decisions Investment Constraints Renewable Energy Laws 10 30 year horizon Minimum zonal reserve margins (% or MW) Reliability criteria (LOLP Target) Inter-zonal transmission expansion (bulk network) Resource addition and retirement candidates (i.e. maximum units built / retired ) Water Pipe Gas Pipeline Coal Transport Build / retirement costs Age and lifetime of units Technology / fuel mix rules Operational Constraints Energy balance Ancillary Service requirements Optimal power flow and limits Resource limits: energy limits, fuel limits, emission limits, water use, etc. Emission constraints User-defined Constraints: Practically any linear constraint can be added to the optimization problem Capacity Expansion Planning 12

AverageHeatRate (btu/kwh) IncrementalHeatRates (btu/kwh) Gas and Coal Gen Efficiency System expansion for obtaining higher capacity factors leads to better over all efficiency and lower carbon intensity MinCap 50% 65% 85% MaxCap Capacity (MWh) AverageHeatRate(btu/kWh) IncrementalHeatRate(btu/kWh) Full Load HeatRate(btu/kWh) 4 April, 2014 Energy Exemplar 13

PLEXOS Example: Desalination Electricity Seawater Heat Desalination Freshwater 14

2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 DESAL Electricity Consumption Example of Desal Expansion Desal Expansion Decision Option for power plant expansion simultaneous with a desal expansion Existing Desal Capacity Expanded Desal 4 April, 2014 Energy Exemplar 15

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101 105 109 113 117 121 125 129 133 137 141 145 149 153 157 161 165 Water Production Water Production Merit Order Unit A Unit B Hours over Week 4 April, 2014 Energy Exemplar 16

Example Desalination Plant Example of Gold Coast Desalination Plant: Maximum Production = 133 ML/day Energy Consumption = 3.2 kwh/1000l Expression of Demand ML/hour (rate) ML (quantity) Use units consistently: 133/24 = 5.5417 ML/hour 3.2kWh 1000 = 3.2 MWh/ML PLEXOS treats water demand as a load on the electrical system Generation and load are balanced with following equation: j Generation j + USE i,t Dump i,t NetInjection i,t = Load i,t i, t Where USE i,t is unserved energy, Dump i,t is over-generation and NetInjection i,t is the net export from the transmission node 4 April 2014 Generic Decision Variables 17

Objects Class Name Cat- Description Decision Variable Heat - The amount of heat provided to the desalination process in period t Decision Variable Heat Input - Heat input to storage in period t Decision Variable Heat Loss - Heat lost from storage in period t Decision Variable Heat Stored - Amount of heat in stored in period t Decision Variable Heat Stored Lag - Amount of heat stored in period t-1 Decision Variable Water - Freshwater Constraint Heat Definition - Definition constraint for "Heat" Constraint Heat Input Definition - Definition of "Heat Input' as sum of waste heat and ancillary boiler heat. Constraint Heat Load - Defines the heat load associated with desalination. Constraint Heat Loss Definition - Defines "Heat Loss" as a proportion of "Heat Stored" Constraint Heat Stored Definition - Defines "Heat Stored" as a function of previous period "Heat Stored" and "Heat Input" and "Heat Loss" Constraint Heat Stored Lag Definition - Defines "Heat Stored Lag" as "Heat Stored" in the previous period 4 April 2014 Generic Decision Variables 18

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101 105 109 113 117 121 125 129 133 137 141 145 149 153 157 161 165 Production Cost with Desal Demand with Desal (MW) Demand before Desal (MW) 4 April 2014 Generic Decision Variables 19

Economic Generation Dispatch to meet Electrical Demand and Water Demand 4 April, 2014 Energy Exemplar 20

PLEXOS Example: Co-Optimization of Generation and Transmission Expansion 21

Security Constrained Unit Commit /Economic Dispatch SCUC / ED consists of two applications: UC/ED and Network Applications (NA) SCUC / ED is used in many power markets in the world include CAISO, MISO, PJM, etc. Unit Commitment / Economic Dispatch (UC/ED) Energy-AS Co-optimization using Mixed Integer Programming (MIP) enforces resource chronological constraints, transmission constraints passed from NA, and others. Solutions include resource on-line status, loading levels, AS provisions, etc. Resource Schedules in 24 hours for DA simulations, or in sub-hourly for RT simulations Violated Transmission Constraints Network Applications (NA) DC-Optimal Power Flow (DC-OPF) solves network power flow for given resource schedules passed from UC/ED enforces transmission line limits enforces interface limits enforces nomograms

Simple Example: G&T Co-Optimization Intermediate/advanced exercises: 1. Create a locational model by defining new GT (operating on Oil) candidate close to the load. 2. Solve the trade-off expansion problem of building Oil-fired GT or reinforcing the transmission system (building a second circuit L1-3 at 10 Million $$). WACC = 12% and Economic Life Year = 30, not earlier than 1/1/2015 Enable following Line expansion properties: Coal_Gen Gas_Gen (2x) New CCGT New_GT x2 Line 1-Coal_Mine Max Flow (MW) L1-2 500 L2-3 500 L1-3 500 2-River & Market L1-2 L2-3 L1-3 L1-3_new 3-Load_Center GT-oil Load 23

Generation and Transmission Expansion Results Generation Expansion Transmission Expansion 24

Australian ISO Use of PLEXOS of Co-Optimization of Generation and Transmission in planning Least-cost expansion modelling delivers a co-optimized set of new generation developments, inter-regional transmission network augmentations, and generation retirements across the NEM over a given period. This provides an indication of the optimal combination of technology, location, timing, and capacity of future generation and inter-regional transmission developments. The least-cost expansion algorithm invests in and retires generation to minimize combined capital and operating-cost expenses across the NEM system. This optimization is subject to satisfying: The energy balance constraint, ensuring supply matches demand for electricity at any time, The capacity constraint, ensuring sufficient generation is built to meet peak demand with the largest generating unit out of service, and The Large-Scale Renewable Energy Target (LRET) constraint, which mandates an annual level of generation to be sourced from renewable resources. 25

PLEXOS Example: Co-Optimization of Ancillary Services for Energy Storage to Balance Renewables 26

Co-Optimization of Ancillary Services Requirements for Renewables Integration of the intermittency of renewables requires study of Co- Optimization of Ancillary Services and true co-optimization of Ancillary services is done on a sub-hourly basis More and more the last decade, it has been recognised that AS and Energy are closely coupled as the same resource and same capacity have to be used to provide multiple products when justified by economics. The capacity coupling for the provision of Energy and AS, calls for joint optimisation of Energy and AS. 27

Ancillary Services Reliable and Secure System Operation requires the following product and Services (not exhaustive): 1. Energy 2. Regulation & Load Following Services AGC/Real time maintenance of system s phase angle and balancing of supply/demand variations. 3. Synchronised Reserve 10 min Spinning up and down 4. Non-Synchronised Reserve 10 min up and down 5. Operating Reserve 30 min response time 6. Voltage Support Location Specific 7. Black Start (Service Contracts) 28

Energy Storage Providing Ancillary Services and Time Shift of Energy 4 April, 2014 Energy Exemplar 29

PLEXOS Example: Sub-Hourly Energy and Ancillary Services Co-Optimization 30

PLEXOS Base Model Generation Result Peaking plant in orange operating at morning peak Some displacement of hydro to allow for ramping Variable wind in green 31

Spinning Reserve Requirement CCGT now runs all day to cover reserves and energy Coal plant 2 also online longer Oil unit not required Less displacement of hydro generation for ramping 32

PLEXOS higher resolution dispatch 5 Minute Sub-Hourly Simulation Oil unit required at peak for increased variability Increased displacement of base load to cover for ramping constraints 33

Energy/AS Stochastic Co-optimisation!!! So far the model example has had perfect information on future wind and load requirements. Uncertainty in our model inputs should affect our decisions Stochastic optimisation (SO) The goal of SO then is to find some policy that is feasible for all (or almost all) the possible data instances and maximise the expectation of some function of the decisions and the random variables What decision should I make now given the uncertainty in the inputs? 34

Energy/AS Stochastic Co-optimisation Even though load lower (wind unchanged) more units must be committed to cover the possibility of high load and low wind These units must then operate at or above Minimum Stable Level 35

References for Ancillary Co-Optimization Applications of PLEXOS WECC Balancing Authority Cooperation Concepts to Reduce Variable Generation Integration Costs in the Western Interconnection: Intra-Hour Scheduling (using PLEXOS), DOE Award DE-EE0001376, 02:10 12:12 Integration of Wind and Solar Energy in the California Power System: Results from Simulations of a 20% Renewable Portfolio Standard (using PLEXOS) Examination of Potential Benefits of an Energy Imbalance Market in the Western Interconnection (using PLEXOS), Prepared under Task Nos. DP08.7010, SM12.2011 sponsored DOE 36

PLEXOS Example: Consideration of Adequacy of Supply for System Expansion Planning 37

Summer 2017 Peak Load Forecast Distribution (MW) Calculated 1-in-10 LOLE for ISO-NE Control Area Forecast Probability 2017 Peak Load Forecast 10/90 28,325 20/90 28,590 30/70 28,940 40/60 29,340 50/50 29,790 60/40 30,265 70/30 30,750 80/20 31,445 90/10 32,210 95/5 32,900 32900 32210 31445 30750 30265 29790 29340 28940 28590 28325 160 PLEXOS Simulations of High Level ISO-NE Control Area Model Results: NICR = 33,855 MW LOLE ~ 0.1 12-13 March, 2014 32000 33000 34000 35000 36000 37000 38000 Installed Capacity (MW) - Simulated load risk in calculating Loss of Load Expectation (LOLE) - Simulated multiple capacity levels MA AGO 38

95% 96% 97% 98% 99% 100% 101% 102% 103% 104% 105% 106% 107% 108% 109% 110% Cost of Lost Load ($) LOLE (days) $4,000,000,000 Cost of Lost Load 0.6 $3,500,000,000 $3,000,000,000 $2,500,000,000 $2,000,000,000 PLEXOS calculation of average load weighted cost of lost load 0.5 0.4 0.3 $1,500,000,000 0.2 $1,000,000,000 System LOLE degrades rapidly below Net Installed Capacity Requirement (NICR) $500,000,000 $0 PLEXOS calculation of average load weighted LOLE Value of Lost Load % NICR LOLE 0.1 0 Assumption: VOLL = $20,000/MWh 12-13 March, 2014 MA AGO 39

Applications of Integrated Planning Tools Planning Objectives PLEXOS Capability Environmental Policies Optimization of Annual, Mid-Term, and Short Term constraints Water usage, Environmental Constraints, others Generation Capacity Planning Least cost capacity expansion planning, cooling types (once through, recirculating, dry), Capacity Expansion Type, Minimization of Production Costs including water treatment costs Renewables Integration and System Flexibility Requirement Assessments Least Cost Resource Change within and Across Regions Minimizing production costs and consumer costs Sizing Natural Gas Network Components and Natural Gas Storage Sub-Hourly Co-Optimization of Ancillary Services with Energy and Transmission Power Flows Stochastic Optimization and Stochastic Renewables Models PHEV, EE, DR, SG, Energy Storage Models Co-Optimization of Generation and Transmission Expansion Generation Retirements and Environmental Retrofit Models Reliability Evaluation, Interregional Planning Co-Optimization of Production cost of Water, Electrical, and Natural Gas Sectors Electrical Network Contingencies and Natural Gas Network Contingencies Co-Optimization of Natural Gas Network Expansion along with Electricity Sector Expansion Electrical Network Contingencies and Natural Gas Network Contingencies Integrated Reliability Evaluation Integrated Reliability Evaluation to Ensure LOLE and other Metrics Maintained with Co- Optimization of Electric and or Gas Sector Expansion or True Monte Carlo 40

Appendix

PLEXOS Optimization Methods Linear Relaxation - The integer restriction on unit commitment is relaxed so unit commitment can occur in non-integer increments. Unit start up variables are still included in the formulation but can take non-integer values in the optimal solution. This option is the fastest to solve but can distort the pricing outcome as well as the dispatch because semi-fixed costs (start cost and unit no-load cost) can be marginal and involved in price setting Rounded Relaxation - The RR algorithm integerizes the unit commitment decisions in a multi-pass optimization. The result is an integer solution. The RR can be faster than a full integer optimal solution because it uses a finite number of passes of linear programming rather than integer programming. Integer Optimal - The unit commitment problem is solved as a mixed-integer program (MIP). MIP solvers are based on the Branch and Bound algorithm, complemented by heuristics designed to reduce the search space without comprising solution quality. Branch and Bound does not have predicable run time like linear programming. It is difficult in all but trivial cases to prove optimality and guarantee that the integer-optimal solution is found. Instead the algorithm relies on a number of stopping criteria that can be user defined in determining unit on and off states. Dynamic Programming - Dynamic programming (DP) is a technique that is well suited to the unit commitment problem because it directly resolves the min up time and min down time constraints, and over long horizons. Its weakness is in the way unit commit is decomposed so that units are dispatched individually. Thus if all units in the system were dispatched using dynamic programming it would be difficult to converge on a solution where system-wide demand was met exactly, and where any other system-wide or group constraint such as fuel or emission limits were obeyed. For certain classes of generator though DP can be fast and highly effective. The DP is most suited to units with a high capacity factor, and if applied to all units in the system is likely to produce significant under/over generation. Thus you must carefully select units for which the DP is applied e.g. high capacity factor plant with long min up time or min down time values are most suitable. Stochastic Programing - The goal of SO is to find some policy that is feasible for all (or almost all) of the possible data instances and maximize the expectation of some function of the decisions and the random variables Multiple Optimization Methods Offer Advantages to Solving Planning Problems 42

Solving Unit Commitment and Economic Dispatch using MIP Unit Commitment and Economical Dispatch can be formulated as a linear problem (after linearization) with integer variables of generator on-line status Minimize Cost = generator fuel and VOM cost + generator start cost + contract purchase cost contract sale saving + transmission wheeling + energy / AS / fuel / capacity market purchase cost energy / AS / fuel / capacity market sale revenue Subject to Energy balance constraints Operation reserve constraints Generator and contract chronological constraints: ramp, min up/down, min capacity, etc. Generator and contract energy limits: hourly / daily / weekly / Transmission limits Fuel limits: pipeline, daily / weekly/ Emission limits: daily / weekly / Others 43

Illustrative DC-OPF LP Formation minimize subject to f f k j bg g min j k min i k f PTDF j k bgf k, j max j i, j ( g and other constraints k l j bg i k c f k g k j f r j j k f 2 j l k bg ) systemenergy balance; g max i power flow in line flow limits in line j j; r, l flow limit for branch group i k j k generation and load at bus k line resistance of line j; Where bgk i,j = branch-group factor for line j in group i PLEXOS Detailed Nodal Transmission Load Flows integrated with Optimization Algorithms 44

Illustrative Formulation of Energy/AS Co-optimization Min subject to k k as g g t k as t,min m,k t,min k t m,k o o t k t k c g as as g Generator Ramp Rate Constraints Generator Min Up/Down Time Constraints Generator Energy Constraints Transmission Constraints Fuel Constraints k l t k t k t sc (o t,min m t m,k loss as m as o g Emission and Other Constraints j t j t,m (AS constraint for AS m) t m,k o t,max t m,k k ) t t t t 1 k k k k k t k m t (System Energy Balance) t,k,m (Generation AS Capacity Limits) t,max k o t k ac t m,k as t m,k t,k (Generation and AS Capacity Limits) 45

Objective: Illustrative Formulation of Co-Optimization of Natural gas and Electricity Markets Co-Optimization of Natural Gas Electricity Markets Minimize: Electric Production Cost + Gas Production Cost + Electric Demand Shortage Cost + Natural Gas Demand Shortage Cost Subject to: [Electric Production] + [Electric Shortage] = [Electric Demand] + [Electric Losses] [Transmission Constraints] [Electric Production] and [Ancillary Services Provision] feasible [Gas Production] + [Gas Demand Shortage] = [Gas Demand] + [Gas Generator Demand] [Gas Production] feasible [Pipeline Constraints] others 46

Stochastic Optimization (SO) Fix perfect foresight issue Monte Carlo simulation can tell us what the optimal decision is for each of a number of possible outcomes assuming perfect foresight for each scenario independently; It cannot answer the question: what decision should I make now given the uncertainty in the inputs? Stochastic Programming The goal of SO is to find some policy that is feasible for all (or almost all) of the possible data instances and maximize the expectation of some function of the decisions and the random variables Scenario-wise decomposition The set of all outcomes is represented as scenarios, the set of scenarios can be reduced by grouping like scenarios together. The reduced sample size can be run more efficiently 47

Day-ahead Unit Commitment, Continued Stochastic Optimisation: Two stage scenario-wise decomposition Stage 1: Commit 1 or 2 or none of the generators Reveal the many possible outcomes Take the optimal decision 2 Stage 2: There are hundreds of possible wind speeds. For each wind profile, decide the optimal commitment of the other units and dispatch of all units Take Decision 1 Is there a better Decision 1? Expected cost of decisions 1+2 RESULT: Optimal unit commitment for generators 48