Smart markets for a Smart Electricity Grid

Smart markets for a Smart Electricity Grid Shmuel Oren The Earl J. Isaac Professor Department of Industrial Engineering and Operations Research University of California, Berkeley http://www.ieor.berkeley.edu/~oren/ Presented to IEEE GREECE PES CHAPTER Athens, Greece, June 6, 2012 All Rights Reserved to Shmuel Oren

POWER INDUSTRY RESTRUCTURING Competitive FERC Regulated State Regulated Customers Demand Management Generation Transmission Distribution Micro-Grids Retail Providers Power Generation SC LSE ISO PX TO SC - Scheduling Coordinator PX - Power Exchange ISO - Indep. System Operator TO - Transmission Owner LSE - Load Serving Entity

Electricity Grid Interconnects

Regional Transmission Organizations in the US (Cover half the states and 70% of load )

Multiple Products and Markets Day-ahead energy market Real-time energy market Forward capacity market Fixed transmission rights (FTRs) auction market Regulation market Operating reserves market Spinning Non-spinning Replacement All Rights Reserved to Shmuel Oren

Unit Commitment Optimization - MIP (Solved for 24 hours in Day Ahead market) Decisions (financially binding): on/off, output level and compensated reserves for each unit in each of 24 hours + locational marginal energy prices and reserve prices for each node and hour Minimize (fuel cost + no-load cost + startup cost) s.t. Load balance constraint at each node Unit output constrain for each generator Unit ramping limits for each gen Unit min up time and min down time for each gen Transmission constraints (DC approximation with thermal proxy limits) Reserves margin requirements Contingencies (n-1) (Cost and constraints data provided as offers in day ahead auction)

Power Flow Optimization (every five minutes) and Locational Marginal Pricing (LMP) For Generators that are Running and Synchronized Decisions: Price of energy (LMP) at node i = Marginal cost of energy at the node Calculated as the dual variable to energy balance constraint for the node in a linearized Optimal Power Flow approximation (DCOPF) Minimize (Generator Fuel Cost) s.t. Energy balance (net supply = load at each node) Generator limits (including dynamic limits such as ramp rates) Transmission Constraints (AC model with voltage and thermal limits) Reserve requirements (Cost curves and generator limits data provided as offers in real time auction every 15 minutes)

LMP / Congestion Example West Limit = 26 MW East ~ ~ 80 MW 90 MW P W 45 40 P E 50 45 Key: 106 120 Q 1 50 64 Q 1 Prices/Supplies under 26 MW limit Prices/Supplies with no transmission limit Marginal value of transmission = $10/MWh (=$50 $40) Total congestion revenue = $10*26 = $260/hr Total redispatch cost = $130/hr Congestion cost to consumers: (40*106+50*64) (45*170) = 7440 7650 = $210/hr All Rights Reserved to Shmuel Oren

Measuring Social Welfare Price ($/MWh) Inverse Demand Function (WTP) - P(q) Benefit Marginal cost curve Local Quantity (MWh) Production = q Production cost C(q) Local Consumption = q + import r

DC- Approx. Optimal Power Flow (DCOPF) Maximizes Net Benefit (or minimizes as bid cost) max r i, in in subject to: in r i l l, i i l in 0 losses =0 Pi ( i ) di Ci ( qi ) K D r K, l L r i Import/Export quantity Local production q i Inverse demand function Balancing constraint Flows constraints All Rights Reserved to Shmuel Oren

KKT conditions for the DCOPF problem LMP Nodal Markup P( q r ) p i N i i i i ( ) D i m m m, i ml jn r j losses =0 0 D r K 0 l L l l, j j l jn 0 K D r 0 l L l l l, j j jn All Rights Reserved to Shmuel Oren

Over-generation, congestion and no storage capability can lead to negative prices 19

Steep Supply Function and Inelastic Demand leads to High Price Volatility $100 $90 $80 $70 $60 Marginal Cost ($/MWh) Resource stack includes: Hydro units; Nuclear plants; Coal units; Natural gas units; Misc. Marginal Cost $50 $40 $30 $20 $10 Supply Curve Demand shocks $0 0 20000 40000 60000 80000 100000 Demand (MW) All Rights Reserved to Shmuel Oren

Energy Price ($/MWh) $400 On-peak Balancing Market Prices at ERCOT January 2002 thru July 2007 $350 $300 UBES 2/25/03 $523.45 Balancing Energy (UBES) Price Spot Price Energy- ERCOT SC $250 $200 $150 Jan-02 Mar-02 May-02 Jul-02 Sep-02 Nov-02 Jan-03 Mar-03 May-03 Jul-03 Sep-03 Nov-03 Jan-04 Mar-04 May-04 Jul-04 Sep-04 Nov-04 Jan-05 Mar-05 May-05 Jul-05 Sep-05 Nov-05 Jan-06 Mar-06 May-06 Jul-06 Sep-06 Nov-06 Jan-07 Mar-07 May-07 Jul-07 24 $100 $50 $0

Electricity Supply Chain Generators Customers (end users served at fixed regulated rate) Wholesale electricity market (spot market) LSE (load serving entity Similar exposure is faced by a trader with a fixed price load following obligation (such contracts were auctioned off in New Jersey and Montana to cover default service needs) 26

Demand (MW) Electricity Price ($/MWh) Price and Demand Correlation Electricity Demand and Price in California 18000 16000 14000 12000 10000 8000 6000 4000 2000 Load Price 180 160 140 120 100 80 60 40 20 0 9, July ~ 16, July 1998 0 Covering expected load with forward contracts will result in a contract deficit when prices are high and contract excess when prices are low resulting in a net revenue exposure due to load fluctuations 27

Volumetric Static Hedging Model Setup One-period model At time 0: construct a portfolio with payoff x(p) At time 1: hedged profit Y(p,q,x(p)) = (r-p)q+x(p) Spot market p LSE r Load (q) x(p) Objective Portfolio for a delivery at time 1 Find a zero cost portfolio with exotic payoff which maximizes expected utility of hedged profit under no credit restrictions. 28

Mathematical Formulation Objective function x( p) max E U[( r max x( p) p) q U[( r Utility function over profit x( p)] p) q Joint distribution of p and q x( p)] f ( p, q) dqdp Constraint: zero-cost constraint 1 Q E [ x( p)] 0 B Q: risk-neutral probability measure B : price of a bond paying $1 at time 1! A contract is priced as an expected discounted payoff under risk-neutral measure 29

Mathematical Formulation Objective function x( p) max E U[( r max x( p) p) q U[( r Utility function over profit x( p)] p) q Joint distribution of p and q x( p)] f ( p, q) dqdp Constraint: zero-cost constraint 1 Q E [ x( p)] 0 B Q: risk-neutral probability measure B : price of a bond paying $1 at time 1! A contract is priced as an expected discounted payoff under risk-neutral measure 30

Impact of Hedging on VaR 31

Financial Transmission Rights (FTR) Market Under LMP Participants in wholesale spot market or bilateral contracts paying congestion charges are exposed to the LMP difference between the injection and withdrawal nodes FTRs were designed to hedge transmission service customers against congestion charge ( nodal price differences) risks FTRs are defined as LMP swaps between each pair of nodes and are settled based on the realized nodal price difference (to offset congestion charges) FTRs are available as options or as two sided contracts which may become a liability when the path is opposite to the direction of the actually congested flow Auction conducted monthly for new FTRs and for trading of outstanding FTRs All Rights Reserved to Shmuel Oren

FTR Auction (ERCOT) Initial design had 72 time slices (24 monthly blocks divided into 3 time blocks Bids (and offers) can cover any subset of the 72 products OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE Year One OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE 72 Periods Optimized Simultaneously = 1300 Hours Year Two Clearing mechanism maximizes auction revenue subject to simultaneous feasibility test (SFT) in every time slice SFT ensures that physical grid could support physical exercise of all outstanding FTRs OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE OP WD WE All Rights Reserved to Shmuel Oren

Actual ERCOT Implementation (live 12/1/2010) Annual auction comprised of six separate optimizations Calendar Period TOU Blocks Period Approach OP WD WE Decouple TOU and Decouple Years / Monthly 12 Year 1 12 Year 2 12 Year 1 12 Year 2 12 Year 1 1 WD 1 WD 1 WE 1 WE 1 OP 12 12 12 12 12 C Monthly granularity 12 Year 2 1 OP 12 WD Year One Year Two WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD WD 12 Periods Optimized Simultaneously Year One 12 Periods Optimized Simultaneously Year Two WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE WE 12 Periods Optimized Simultaneously 12 Periods Optimized Simultaneously OP Year One Year Two OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP OP 12 Periods Optimized Simultaneously 12 Periods Optimized Simultaneously

Simultaneous Feasibility Guarantees Revenue Adequacy (congestion revenues cover FTR settlements ) 100 G1 G2 2 1 3 3 1 2 3 3 1 G3 2/3 3 1/3 1/3 L3 2 G1 G2 300 G1 G2 220 100 ( G1 G2) 100 1 3 320 MW 300 MW FTR 2 3 2 1 A B 100 MW O 20 MW X Changing capacity of line 2 to 3 Z 2 3 E Y 1 2 300 MW 1 3 C D 400 MW FTR1 3 Two sided FTRs must stay within the outer nomogram One sided FTRs (options) must stay within the inner nomogram because we cannot rely on counterflows to alleviate congestion. 35

Renewables Making Headlines

Conventional Solution Source: CAISO

The DR Alternative to Expanding Flexible Thermal Generation

Emerging Issues Renewables and distributed resources integration Mitigation of supply and demand uncertainty and resource variability Connectivity (planning and financing of transmission) Stochastic optimization for planning and dispatch Smart Grid Integration: Enable customer choice, demand response, load management and energy efficiency with smart Metering Improve grid visibility and control with SCADA and Phasor Measurement Units (PMUs) Dynamic Co-optimization of grid topology Market implications of smart grid integration Business models and dispatch algorithms for demand response aggregation Impact of mass PHVs EV deployment on the grid Utilization and impact of mass storage Impact of carbon and RPS regulation on grid planning, operation and energy markets

Stochastic Unit Commitment N

Scenario Selection

Wind Modeling and Data Sources

Day Types

WECC Model

Model Summary

Topology Control Example Original optimal cost: $20,000 (A=180MW,B=30MW,C=40MW) at {2} Original feasible set: {0,1,2,3} Open Line A-B, optimal cost: $15,000 (A=200MW, B=50MW) at {8} Feasible set with Line A-B open {0, 4, 5, 6} Feasible set with optimal transmission switching: {0, 1, 7, 5, 6} (non-convex) 50

Optimal Transmission Switching DCOPF

Results DCOPF IEEE 118 IEEE 118 opened lines for J=10 Note: this diagram has additional gens than our model 52

Results DCOPF IEEE 118 Transmission switching solution saves 25% of total generation cost J 53

More Test Results IEEE 118 Bus Model: 54 DCOPF transmission switching solution with no contingencies saves 25% of total generation cost (10 lines switched off) Up to 16% savings with N-1 DCOPF transmission switching (for feasible solutions) IEEE 73 (RTS 96) Bus Model Up to 8% savings with N-1 DCOPF transmission switching (for feasible solutions) Savings of 3.7% in Unit Commitment for 24 hours with N-1 security constraints ISO-NE 5000 Bus Model: DCOPF transmission switching with approximate solution produced 5-13% savings Current ARPA-E ($5M) Project on Topology Control Test of concept on TVA 7000 bus network

Questions?