1. November 2001 CHRISTENSENASSOCIATES RTP as a Demand Response Program – How Much Load Response Can You Expect? Peak Load Management Alliance Fall Conference November 2001 Steven Braithwait Christensen Associates

2. November 20012 CHRISTENSENASSOCIATES Christensen Associates  Economic and engineering consulting for energy industry  25 years of experience in designing and evaluating retail pricing strategies – –TOU (traditional and competitive) –Real-time pricing (NiMo, Georgia Power, KCP&L) –Market-based interruptible load programs

3. November 20013 CHRISTENSENASSOCIATES Sound Bites from Yesterday  “Square peg in round hole” Customers aren’t generators  Trends in DR programs? Look at competitive markets  “Dollars on the table” Focus on NEW $$$ from efficient pricing  “Keep it simple” Here are tomorrow’s prices, you decide…

4. November 20014 CHRISTENSENASSOCIATES Topics  Real-time pricing as a DR program  Capturing the benefits of DR and RTP  Evidence of RTP load response  Technology facilitates residential “RTP”

5. November 20015 CHRISTENSENASSOCIATES What is Real-time Pricing (RTP)?  A market-based pricing strategy –Customers face hourly spot market prices –Advance notice – day-ahead; hour-ahead –Price protection through hedging/price caps –Works in regulated and competitive markets  A demand response program –RTP customers provide load reductions at times of high wholesale prices

6. November 20016 CHRISTENSENASSOCIATES RTP at Georgia Power Company  1,700 C & I customers  5,000 MW (80% of C & I sales)  Day-ahead (75%) & hour-ahead (25%)  Load response: 500 – 1,000 MW (at prices of $500 – $2,000/MWh)

7. November 20017 CHRISTENSENASSOCIATES Benefits of Demand Response  Connects wholesale and retail markets  Demand response at high prices can reduce wholesale price spikes

8. November 20018 CHRISTENSENASSOCIATES Connected Markets: Demand Response Yields Lower Wholesale Prices $/MWh GWh Q normal Retail Price P normal Q spike L normal D normal E P spike L hot WP B Q hot P hot E´ D hot A

9. November 20019 CHRISTENSENASSOCIATES Capturing the Benefits of DR  To capture benefits, the amount of demand response must be measured and anticipated (e.g., in unit dispatch and power purchases)

10. November 200110 CHRISTENSENASSOCIATES Measuring Non-RTP DR  DR programs pay for “performance”  However, cannot “measure” performance (i.e., load reductions) by metering  Load response (LR) must be estimated: LR = Baseline load - Actual load Problems in estimating CBL; $$ at risk

11. November 200111 CHRISTENSENASSOCIATES Measuring and Forecasting RTP Load Response  For billing: No “verification” problem – RTP customers pay for what they consume  For forecasting: –Develop load response model based on analysis of historical experience –Advantage of aggregating over customers

12. November 200112 CHRISTENSENASSOCIATES RTP Load Response Curve for Georgia Power (Load Response as a Percent of Total RTP Load)

13. November 200113 CHRISTENSENASSOCIATES Evidence of RTP Load Response  Georgia Power Real-Time Pricing (RTP) –1,700 large C & I customers; 5,000 MW of load  Duke Power Hourly Pricing –100 large industrial customers; 1,000 MW  GPU Energy “Critical price” TOU –1997 residential pilot program

14. November 200114 CHRISTENSENASSOCIATES Summary of RTP Price Responsiveness  Significant, consistent load response  Small to modest price elasticities  Wide range across customer types  Most price responsive customers: –Electricity intensive (e.g., most intensive industrials; residential customers with most major appliances) –Enabling technology (e.g., own generation; storable production process; automatic controls)

15. November 200115 CHRISTENSENASSOCIATES Highest DA prices Load on highest-price day GP RTP Load Response (DA): Very High-Price (Load response = 500 MW; 20% of reference load)

16. November 200116 CHRISTENSENASSOCIATES 0.000 0.050 0.100 0.150 0.200 0.250 0.300 0.350 0.400 HAOSG - INon-Int - IOSG - CNon-Int - C Moderately high price Very high price Diversity of Customer Price Responsiveness Price Elasticity by Customer Type and Price Level

17. November 200117 CHRISTENSENASSOCIATES Duke Power RTP Load Response (per Tom Taylor, Rates and Regulation)  100 industrial customers; 1,000 MW  Total load response when Price > $.25/kWh –200 MW, or 20% of expected load  20 customers reduced load by > 5%  Significant price elasticities for 25% of customers

18. November 200118 CHRISTENSENASSOCIATES GPU “Critical-price” TOU Rate Effect of Technology on Load Response  TOU rate, plus critical price ($.50/kWh)  Interactive communication system –customers pre-select thermostat settings and circuit priority at different price triggers –utility can send critical price signal  Similar programs at AEP, Gulf Power

19. November 200119 CHRISTENSENASSOCIATES “Critical-price” TOU Rate Design 0 0.1 0.2 0.3 0.4 0.5 1814182024 Critical price Rate 6173 Rate 9122 Standard Rate

20. November 200120 CHRISTENSENASSOCIATES Residential TOU Load Response – Critical Price Day 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 1 23456789101112131415161718192021222324 Average Hourly Usage: kWh/hr. Control Treatment OP S Peak S OP

21. November 200121 CHRISTENSENASSOCIATES Conclusions  RTP offers demand response in a natural retail market setting  Methods are available for anticipating RTP load response at different price levels  Evidence is available on amount of RTP load response to expect

22. November 200122 CHRISTENSENASSOCIATES For Additional Information:  Customer Response to Market Prices – How Much Can You Expect When You Need it Most?, Steven Braithwait and Michael O’Sheasy, EPRI Pricing Conference, July 2000.  Residential TOU Response in the Presence of Interactive Communication Equipment, Steven Braithwait, in Pricing in Competitive Electricity Markets, Ahmad Faruqui, Ed.  The Choice Not to Buy: Energy Savings and Policy Alternatives for Demand Response, Steven Braithwait and Ahmad Faruqui, in Public Utilities Fortnightly, March 15, 2001. Contact: Steve Braithwait (sdbraithwait@lrca.com) Christensen Associates 608-231-2266

23. November 200123 CHRISTENSENASSOCIATES Types of Demand Response Programs  Demand-side bidding – customers bid load reductions into the wholesale market  “Buy-back,” or pay-for-performance interruptible –Suppliers buy load reductions, relative to baseline, at price tied to market price  Real-time (hourly) pricing –Full-time –Part-time; whenever cost exceeds specified level

24. November 200124 CHRISTENSENASSOCIATES How to Estimate Reference Load?  Historical load on same day-type (e.g., summer Tuesday, with “hot” weather)  Rolling average of loads on “non-event” days (e.g., previous 10 weekdays)  Average load in previous hours (e.g., previous 3 hours) Key objective – avoid “gaming” possibilities

25. November 200125 CHRISTENSENASSOCIATES “Disconnected” Electricity Markets: Fixed retail price  no demand response $/MWh GWh Retail Price P normal Q normal L normal P spike Q spike L hot WP