synapse hurley demandresponseasapowersystemresource 2013 may 31 (1)

8/12/2019 Synapse Hurley DemandResponseAsAPowerSystemResource 2013 May 31 (1)

1/76

Demand Response as aPower System Resource

Program Designs, Performance, andLessons Learned in the United States

AuthorsDoug Hurley

Paul PetersonMelissa Whited

May 2013


2/76

Electronic copies of this paper and other RAP publicationscan be found on our website at www.raponline.org .

To be added to our distribution list,please send relevant contact information to

[email protected] .

The Regulatory Assistance Project (RAP)

The Regulatory Assistance Project (RAP) is a global, non-prot team of expertsfocusing on the long-term economic and environmental sustainability ofthe power and natural gas sectors. RAP has deep expertise in regulatory andmarket policies that promote economic efciency, protect the environment,ensure system reliability, and allocate system benets and costs fairly among allconsumers.

RAP works extensively in the European Union, the US, China, and India. We have assisted governments in more than 25 nations and 50 states andprovinces. In Europe, RAP maintains ofces in Brussels and Berlin, with ateam of more than 10 professional experts in power systems, regulation, and

environmental policy. For additional information, visit the RAP websitewww.raponline.org .
mailto:info%40raponline.org?subject=http://www.raponline.org/http://www.raponline.org/mailto:info%40raponline.org?subject=


3/76

1

Demand Response as a Power System Resource

RAP Foreword

The nations of Europe, and the European Unionas a whole are in the midst of a transition acrossthe power sector that, at the highest level, aimsto deliver the benets of market competition

and transnational market integration to customers acrossall of the Member States and to do so while meeting 21 st

century power reliability standards, increasing the shareof variable renewable generation, decreasing the climate

and environmental impacts of power generation, andmanaging total power costs for the benet of families andbusinesses that must compete in world markets. This is,all in all, a tall order.

European policymakers, power providers, andutilities are making signicant progress towards thesechallenging goals, but challenges are also evolving andgrowing. Greater reliance on competitive markets exposescustomers and investors to more volatile prices, andgreater reliance on renewable generation will put pressureon grid transfer capabilities and will make it harderfor system operators to align generation and customerdemand levels in real time. For these and other reasons,it is increasingly apparent that investments on the supply-side alone will be insufcient to optimize the costs andenvironmental footprint of European power systems in thecoming decade. It is crucial now to examine options on thedemand side of the power system,and to design market rulesand public policies that will enlist customers and theiragents as power sector resource providers.

Demand-response (DR) resources can provide

numerous benets to power systems, particularlythose seeking to integrate a large fraction of renewablegeneration, but DR is a challenging new area for mostutilities and grid managers. As EU policymakers, system

operators, and power providers examine options to tapDR resources, it is very useful to build on lessons fromthose power markets and system operators that havehad some years of experience in this arena. Due to theexpressed interest of European policymakers, includingthe electric power team at ACER, Europes Agency forthe Cooperation of Energy Regulators, the Regulatory

Assistance Project commissioned this paper to provide a

detailed and highly expert review of DR tools and resultsin several of the leading markets in North America, whereDR has become a signicant and valued resource. RAPitself has had substantial experience with these policies,having launched and led several of the national andregional initiatives that advanced DR as a resource in USpower, reserves, and ancillary services markets.

This paper was completed by a team of highly-regardedexperts from Synapse Energy Economics, who havebeen deeply involved in many of those measures. Thescoping and drafting was prepared in close collaborationwith Meg Gottstein and Mike Hogan, senior RAP policyadvisors on market design, along with the invaluableproject management assistance of RAP Associate SarahKeay-Bright. We are grateful for the opportunity toprovide this information in the European context, andstand ready to assist Governments, system operators, andother stakeholders on DR policies and programme designsas DR resources rise in importance in Europes powersystems.

Richard CowartDirector of European ProgrammesRegulatory Assistance Project

Brussels, May 2013


4/76

2


Synapse Foreword

Demand Response as a Power System Resource was prepared by Synapse Energy Economics,Inc. for the Regulatory Assistance Project.The report focuses on the ways that demand

response resources effectively participate in and improvethe performance of coordinated electric systems in theUnited States. The report reviews the many types ofservices that demand response can provide and the early

history of demand response programs in the UnitedStates. The bulk of our research examined the specicapplications of demand response in several US regions.This report includes numerous examples of demandresponse successfully providing reliable system servicesat competitive prices, and ends with lessons learned andkey challenges for the near future.

Instead of attempting to translate the Americanexperience into numerous European structures, wehave tried to use well-dened, simple terms to describehow a variety of demand response resources provide

reliability, energy, and ancillary services. Demandresponse resources have varied capabilities and servicesthat they can provide, just as supply resources such ascentral station power plants do. We have minimized theuse of acronyms in an industry that ourishes with thembecause the challenges facing the electric industry arecomplex and technical enough; the language itself shouldnot be an additional barrier to communication and

problem-solving. We appreciate the information, suggestions, and

insights provided by numerous members of the electricindustry, colleagues, and friends that assisted us inwriting this report. We are particularly grateful forcomments from staff members at the Lawrence BerkeleyNational Laboratory and the Regulatory AssistanceProject who reviewed early drafts of the report.

Doug Hurley, Melissa Whited, and Paul PetersonSynapse Energy Economics


5/76

3


List of Acronyms and Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 A. Services Provided by Demand Response in the United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8B. Lessons Learned from Existing Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9C. New Opportunities and Challenges for Demand Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10D. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2. Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 A. What is Demand Response? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13B. Demand Response Provision and Classication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Infrastructure Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Customer Baselines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3. History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 A. Demand Response as a Reliability Resource Prior to Restructuring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19B. Early Wholesale Market Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19C. Recent Developments to Enable Demand Response in US Markets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22D. Demand Response Today . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Fully-Integrated Market-Based Demand Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24Market-Reactive Demand Response Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24Non-Market Local Demand Response Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

4. Demand Response for Resource Adequacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 A. Forward Capacity Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Independent System Operator of New England . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26PJM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

B. Other Capacity Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35New York Independent System Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Midwest Independent System Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36C. Non-Market Demand Response Capacity Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

California Independent System Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

5. Demand Response As An Energy Resource . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 A. PJM Economic Demand Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39B. Independent System Operator of New England . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Price Response Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Day-Ahead Load Response Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

C. New York Independent System Operator Day-Ahead Demand Response . . . . . . . . . . . . . . . . . . . . . . . . . . 43D. Order No. 745 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Table Of Contents


6/76

4


6. Demand Response As A Provider Of Ancillary Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 A. Reserve Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

ERCOT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46PJM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51Independent System Operator of New England . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

New York Independent System Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52B. Regulation and Load-Following Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

Balancing Services for Renewable Energy Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52FERC Order 755 on Compensation of Regulation Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53PJM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53Independent System Operator of New England . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55Bonneville Power Administration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

7. Lessons Learned And Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 A. Lessons Learned . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

Demand Response Can Perform Reliably . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Demand Response Can Provide Signicant Contributions to Resource Adequacy . . . . . . . . . . . . . . . . . . . 59Markets Have Achieved Great Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60The Money Matters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

B. Near-Future Opportunities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61New Developments in Demand Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

Additional Design Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61C. Key Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

Baseline Determinations and Infrastructure Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Regional versus Distribution Utility Dispatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Market Barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Institutional Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

Works Cited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

Terminology Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68


7/76

5


List of Figures

Figure 1: Services Provided by Various Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Figure 2: Demand Side Management Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15Figure 3: Illustration of Baseline Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17Figure 4: Regional System Operators in the United States and Canada . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20Figure 5: Change in Program Enrollments from 1998 to 2003 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21Figure 6: Growth of Demand Response Resources in New England . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27Figure 7: ISO-NE Demand Response. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28Figure 8: July 22, 2011 Real Time Demand Response Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30Figure 9: MW of DR by Program 2007-2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30Figure 10: PJM Capacity Resources (Includes Interruptible Load for Reliability and Auction Products) . . . . . . . . .31Figure 11: Demand Response Cleared in PJM Forward Capacity Auction with

Weighted Average Clearing Price (Excludes Interruptible Load for Reliability) . . . . . . . . . . . . . . . . . . .32Figure 12: Emergency DR Events in PJM in 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33Figure 13: Estimated PJM Demand Response on July 17, 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Figure 14: Estimated PJM Demand Response on July 18, 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Figure 15: Historical Growth in Resources and MW in NYISO Reliability Programs 2001-2010 . . . . . . . . . . . . . .36Figure 16: August 2, 2006 Response from Demand Resources in NYISO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36Figure 17: Peak Load Reductions during DR Events (2001, 2006, 2010, 2011) . . . . . . . . . . . . . . . . . . . . . . . . . .37Figure 18: MISO Registered Demand Response and Behind-the-Meter Generation

(December 2012) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37Figure 19: Energy Market Megawatt Hours (2007 - Nov. 2012) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40Figure 20: Percent Load Curtailment and Simulated Corresponding Price Reduction

in Mid-Atlantic States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40Figure 21: Megawatt Curtailment and Corresponding Value of Spot Market Price Reduction . . . . . . . . . . . . . . . .40Figure 22: Participation of Demand Response and Percent Decrease in Monthly

Average Real-Time Locational Marginal Price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42Figure 23: Participation of DR and ISO-NE Real-Time Locational Marginal Price . . . . . . . . . . . . . . . . . . . . . . . . .42Figure 24: PJM Monthly Energy Market Participation Before and After Order No. 745 . . . . . . . . . . . . . . . . . . . . .44Figure 25: European and US Ancillary Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45Figure 26: ERCOT Load Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Figure 27: ERCOT Load Resource Deployments since April 2006 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Figure 28: ERCOT Demand Response (FebruaryMay) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Figure 29: ERCOT Demand Response (June-September) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Figure 30: ERCOT Demand Response (October-January) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Figure 31: ERCOT Emergency Interruptible Load Service Deployment, February 2- 3, 2011 . . . . . . . . . . . . . . . .49Figure 32: ERCOT Emergency Interruptible Load Service Deployment, August 4, 2011 . . . . . . . . . . . . . . . . . . . .50Figure 33: Demand Response Provision of Regulation Services in PJM (Jan 2012-Feb 2013) . . . . . . . . . . . . . . . .54Figure 34: Aggregated Response of 250 Thermal Storage Heaters Loads to Frequency Signal . . . . . . . . . . . . . . . .54Figure 35: Aggregated Response of Nine Loads to Frequency Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55Figure 36: Example of Bonneville Power Administration Balancing Reserve

Deployment in 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Figure 37: Demand Response Resource Available as Percent of 2010 Peak Demand . . . . . . . . . . . . . . . . . . . . . . .60


8/76

6


List of Tables

Table 1: Demand Response Available at US ISOs and RTOs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23Table 2: Ancillary Services Provided by Demand Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24Table 3: New England Capacity Prices Applicable to All Capacity Resources

Dec 2006 May 2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28Table 4: June 24, 2010 Real Time Demand Response Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29Table 5: July 22, 2011 Real Time Demand Response Performance Period of 100% Dispatch . . . . . . . . . . . . .29Table 6: Load Management Commitments, Compliance, and Test Performance . . . . . . . . . . . . . . . . . . . . . . . . .33Table 7: 2010 PJM DR Performance Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34Table 8: Total Number of Hours of Special Case Resource/Emergency Demand Response

Events in NYISO, by Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35Table 9: Estimated Savings due to Demand Response Depressing Market Prices in NYISO . . . . . . . . . . . . . . . .41Table 10: NYISO Day Ahead Demand Response Total Scheduled Hours . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43Table 11: Ancillary Services that may be Provided by Demand Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .46Table 12: Thirty-Minute Emergency Response Service Pilot

(July-September 2012 Procurement) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50Table 13: PJM Synchronized Reserve Clearing Prices and DR Participation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51Table 14: New England Capacity Resource Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59


9/76

7


ACEEE American Council for an Energy-EfcientEconomy

BPA Bonneville Power AdministrationBRA Base Residual AuctionCAISO California Independent System OperatorCSP Curtailment Service ProviderDADRP Day-Ahead Demand Response ProgramDALRP Day-Ahead Load Response ProgramDOE Department of EnergyDR Demand ResponseDRR Demand Response Resources

DSASP Demand Side Ancillary Services ProgramDSM Demand Side ManagementDSR Demand Side ResourcesEDR Emergency Demand ResponseEDRP Emergency Demand Response ProgramERCOT Electric Reliability Council of TexasEILS Emergency Interruptible Load ServiceEPA Environmental Protection AgencyEPACT Energy Policy Act of 2005ERS Emergency Response ServiceETS Electric Thermal StorageFCA Forward Capacity AuctionFCM Forward Capacity MarketFERC Federal Energy Regulatory CommissionHVAC Heating/Ventilation/Air ConditioningICAP Installed CapacityISO Independent System OperatorISO-NE Independent System Operator New EnglandLaaR Load Acting as a ResourceLMP Locational Marginal PriceLMR Load Modifying Resources

LOLE Loss of Load Expectation

LOLH Loss of Load HoursLOLP Loss of Load ProbabilityLR Load ResourceLSE Load Serving EntityMISO Midwest Independent System OperatorMW MegawattMWh Megawatt hourNARUC National Association of Regulatory

Utility CommissionersNAESB North American Energy Standards BoardNERC North American Electric Reliability

CorporationNYISO New York Independent System OperatorPJM PJM Interconnection (formerly

Pennsylvania-New Jersey-MarylandInterconnection)

PRD Price Responsive DemandPRP Price Response ProgramRFP Request for ProposalRPM Reliability Pricing ModelRTDR Real Time Demand ResponseRTEG Real Time Emergency GenerationRTO Regional Transmission OrganizationSCR Special Case ResourcesSPP Southwest Power PoolSRMCP Synchronized Reserve Market Clearing PriceT&D Transmission and DistributionTDU Transmission and Distribution UtilityTPRD Transitional Price Responsive DemandTVA Tennessee Valley AuthorityUCAP Unforced CapacityVOLL Value of Lost Load

List of Acronyms and Abbreviations


10/76

8


Executive Summary

Demand response refers to the intentionalmodication of electricity usage by end-usecustomers during system imbalances or inresponse to market prices. While initially

developed to help support electric system reliabilityduring peak load hours, demand response resourcescurrently provide an array of additional services thathelp support electric system reliability in many regionsof the United States. These same resources also promoteoverall economic efciency, particularly in regions thathave wholesale electricity markets. Recent technicalinnovations have made it possible to expand the servicesoffered by demand response and offer the potential forfurther improvements in the efcient, reliable deliveryof electricity to end-use customers. This report reviewsthe performance of demand response resources in theUnited States; the program and market designs thatsupport these resources; and the challenges that mustbe addressed in order to improve the ability of demandresponse to supply valuable grid services in the future.

A. Services Provided by DemandResponse in the United States

In Section 2 we provide an overview of the manydiverse types of demand response resources and denethe categories of demand response that we reviewedfor this paper. We primarily focus on resources thatare capable of being dispatched by system operators

through direct controls and economic incentives (or acombination). In Section 3 we review historical demandresponse participation and how the programs developedby vertically integrated utilities became the precursors totodays demand response programs.

Sections 4 through 6 provide descriptions of the keyservices that demand response provides to the US electricsystem. These services range from ensuring resourceadequacy and providing ancillary services to reducinghigh energy prices through participation in energymarkets. An overview of these capabilities is shown in theillustration below.

DemandResponse

Peak Load Reduction

Generation Storage

Figure ES-1

Services Provided by Various Resources

Wider Operating Range(lower minimumcapacity)

DispatchableWind/Solar

VoltageSupport

RegulationFast Ramping

FrequencyResponse

DispatchableQuick Start

LoadShifting

Absorptionof Excess

Generation

Source: (Liu 2012)

In particular: We look at demand response resources that support

resource adequacy through wholesale marketdesigns in PJM, New York, and New England. Thesemarket designs treat demand response resources asrough equivalents to traditional generation resourcesto ensure sufcient capacity during peak-load hours.

We also briey look at non-centrally dispatcheddemand response resources that provide a similarcapacity/resource adequacy service in California.These demand response resources are often calledemergency resources, but we prefer the termsresource adequacy or capacity resources to reecttheir expanded use in situations other than systememergency events.

We look at demand response resources thatprovide energy reductions in the day-ahead orreal-time wholesale energy markets and reviewthe market designs currently in place. We also

discuss the numerous new market designs goingthrough the development, review, approval, and


11/76

9


enabling demand responseto compete with generationon a relatively level playingeld. The Midwest and New

York also have robust demandresponse participation in theircapacity programs. In thesedesigns, demand responseresources have demonstratedreliable performanceand provided substantialcontributions to systemresource adequacy goals. Thetable below demonstrates thisby comparing the megawatts

of demand responsedelivered relative to thequantity that had anobligation to be available, aswell as the performance oftraditional resources.

2. Demand response can provide energy servicesthat primarily enhance efcient price formation inwholesale energy markets, but also enhance reliableoperation of the system. Market designs are still beingdeveloped pursuant to Commission Order No. 745 to

ensure that pricing and verication mechanisms areoptimal. We focus largely on PJM and New Englandprogram and market designs for demand responseresources that can provide energy services. The gure

Table ES-3

New EnglandCapacity Resource Performance

implementation process to comply with FERCOrder No. 745.

We look at ancillary services such as spinningand non-spinning reserves, and regulation andload-following services. A comparison of ancillary

services in the United States and Europe is shownin the illustration above while a more detaileddescription of each service is provided in Section6 and the appendix. Texas has proven demandresponse programs that act as spinning and non-spinning reserves to provide resources duringsystem emergencies. Other regions are exploringinnovative ways that demand response can providea variety of ancillary services.

B. Lessons Learned from ExistingPrograms

In Section 7 we discuss lessons learned from demandresponse in the United States. These lessons include thefollowing:

1. Demand response can provide capacity/resourceadequacy services for system peak load days thatare equivalent, and perhaps superior, to servicesprovided by traditional resources. PJM and NewEngland have existing market designs that have

successfully incorporated signicant amounts ofdemand response into their capacity markets by

European System

U.S. System

Sources: Adapted from Ruggero Schleicher-Tappeser diagram based on (NERC 2011) and (Amprion 2013).

Source: (Scibelli 2012)

Figure ES-2

Comparison of European and US Ancillary Services


12/76

10


above shows the potential reductions to PJM hourlyclearing prices from demand response, based on astudy by the Brattle Group.

3. Demand response can provide ancillary services thatinclude various reserve services, dynamic systemregulation, and load-following capabilities that candeliver value to the grid during any hour of the year.The Texas ten-minute reserves programs for demandresponse resources have demonstrated reliableperformance over numerous system operator dispatchevents (see Figure ES-5 below). Current program andmarket designs for demand response participationprovide excellent examples of this type of demandresponse service. Additionally, pilot programs in severalregions and new technology adaptations are suggesting

Figure ES-5

Demand Response Deployment of EmergencyReserves in Texas on August 4, 2011

Source: (Brattle Group 2007)

Source: (ERCOT 2012)

a more robust role for demand response resourcesin providing balancing services to correct systemimbalances on a minute-by-minute or even second-by-second basis.

C. New Opportunities and Challengesfor Demand Response

Opportunities and challenges for demand response arealso explored in Section 7. Future expansion of demandresponse participation in wholesale markets includesseveral promising areas:

Coordination of demand response resources bysystem operators to more accurately match resourceneeds with system conditions. Rather than relyingon individual suppliers or distribution utilities toactivate demand response resources, regional systemoperators could provide more reliable and efcientimplementation that utilizes demand responseresources across multiple utility areas.

Deployment of technologies that enhance theability of system operators to integrate new formsof demand response into normal system operationsduring any hour, rather than just peak demandperiods. These new resources include price-responsive demand that is enabled by advanced

meters and demand response resources with astorage component such as water pumping andspace heating that can increase demand duringperiods of excess generation. Such forms of demandresponse can reduce costs and enhance systemefciency during any hour of the year.

Enhanced or new market designs that providecompensation for a variety of demand response (andother) resources to meet future system needs. Thesecould be changes to existing capacity programs orseparate programs for specic demand response

resources that supply regulation or balancingservices.

Key challenges include: Establishing baselines to accurately measure

demand response contributions to ensureappropriate performance and compensation.

Resolving dispatch issues between traditionaldistribution utilities and regional system operatorsor balancing authorities. As the proportion ofvariable generation resources increases, demandresponse can play a critical role by providing moreexible options to system operators to ensurereliability and improve overall system efciency.

Figure ES-4

Percent Load Curtailment andSimulated Corresponding Price Reduction in

Mid-Atlantic States


13/76

11


In order to achieve that greater exibility, systemoperators must be able to directly dispatch demandresponse resources located across multiple utilitydistribution areas.

Integrating sufcient metering and communications

equipment to provide accurate and timelyinformation about overall electric systemperformance as well as specic demand responseresource operations, while ensuring requirementsare not cost prohibitive for smaller resources.

Modifying current market and operational rulesto remove numerous market barriers to demandresponse participation, such as minimum sizerequirements and prohibitions regarding demandresponse aggregator participation.

Addressing the lack of demand responseparticipation in certain regions that is often relatedto traditional utility incentive structures that donot reward utilities for incorporating demandresponse. In some regions, third-party aggregatorsare obstructed from enrolling demand responseresources due to utility opposition and regulatorconcerns about consumer impacts and benets.

D. Summary and Conclusions

Demand response provides a variety of valuableservices to the US electric grid and has the potential toenhance system efciency to an even greater extent in thefuture. To that end, we offer the following conclusionsand recommendations for facilitating optimal demandresponse participation going forward:

Provide greater revenue certainty. The growth ofdemand response has been strongest where a steadymonthly payment exists, and where multiple streamsof revenue are present to support different types ofloads and different types of customers. Trying to rely

on unpredictable and infrequent high-priced eventsis a business model that is too risky to incentivizesignicant demand response participation. Mechanismsto provide greater revenue security could include targetedinstruments such as procurement of specic criticalservices, or less-targeted instruments such as forwardcapacity markets.

Remove restrictions on demand responseparticipation. Regions that do not limit demandresponse resources and allow demand response toprovide multiple types of services (energy, reserves,

regulation) have demonstrated greater participation bydemand response resources.

Ensure demand response providers face adequateincentives. Independent demand response aggregatorshave a greater nancial incentive to sign up as many

customers with load reduction capabilities as possible.Utility providers can provide reliable demand response,but often have conicting nancial incentives. Shiftingtraditional utility compensation to a more nuancedcompensation system will help align incentives towards amore efcient overall use of resources.

The residential market remains largely untappedfor now. Few demand response providers have evenapproached the residential market to date due tothe amount and variety of load available from largecustomers. However, cost-effective technology to providesmall amounts of demand response from a very largenumber of residential customers is not far away, and maylead to widespread implementation by the end of thedecade.

Demand response, with sufcient compensation,can help integrate variable resources. The ability ofstorage-type demand response that can ramp in bothdirections to both reduce load and also absorb excessgeneration is a new and developing area of demandresponse that has proven reliable for balancing services

(regulation and load-following) at a small scale. However,it remains to be seen whether the revenue generatedthrough such services is sufcient to sustain demandresponse providers. A re-determination of the value ofbalancing services under resource mixes with greaterquantities of variable resources (such as wind and solar)may need to occur. In addition, compensation thatrewards the speed and accuracy of the response willhelp incentivize demand resources to participate in thismarket.

Regulatory support is necessary to level the

playing eld. Finally, regulatory support at both thestate and federal level has been critical. In order fordemand response to ourish, the necessary policy andregulatory framework rst had to be established togovern the treatment of demand response and enable itto be compensated in a manner comparable to generationresources. The consistent attention to these issues bythe Federal Energy Regulatory Commission has provenessential to the success of demand response resources todate.


14/76

12


1. Introduction

Demand response (DR) encompasses numeroustypes of load-modifying resources thatprovide a variety of electric system functions.Over the last few decades, the original utility

programs that were developed primarily to provide loadreductions during system emergencies have evolved intomore sophisticated programs capable of providing a rangeof targeted services. Demand response has transitionedfrom simply a means for shaving peak demand intoa valuable tool enabling grid operators to managethe challenges of the modern grid. Demand responseprogram structures have likewise expanded from simpleincentives that enable a utility to temporarily interruptconsumption to more sophisticated market arrangements,including three-year forward commitments to provide aguaranteed level of energy reduction based on a centraloperator dispatch signal, and balancing services that mayemploy storage to better integrate renewable resources. 1

Signicant effort has been invested in current demandresponse programs and market constructs to ensure thedevelopment of appropriate incentives, regulations, andtechnologies. There have been bumps in the road; someprograms have worked better than others. Despite beingoriginally viewed solely as a seasonal peak load reducer,demand response has demonstrated that it can providecost-effective, year-round reliability services, daily energyservices, and ancillary services that include reserves,load-following, and regulation. Programs and markets are

likewise evolving to incorporate these diverse functionsand encourage greater demand response participation.

This paper provides a cursory review of earlier demandresponse programs, a detailed look at recent programs,and summarizes the lessons learned, the most promisingfuture applications, and key challenges facing demandresponse in the United States. The report is organized in

sections:Section 2 covers background issues related to demand

response resources. This includes dening the differenttypes of demand response resources and the servicesthey can provide; the benets that demand response canprovide to bulk power system operations; and the speciccharacteristics of demand response resources reviewed inthis report.

Section 3 covers the history of demand responseinitiatives in the context of the transition from verticallyintegrated utilities to more competitive regional marketsoperated by system operators who provide coordinateddispatch and exercise day-to-day operational control.

Sections 4 through 6 review current programsdeveloped and implemented by numerous entities acrossthe United States. We review program designs, the levelof participation by demand response resources, and

the performance of demand response resources duringspecic events. The reviews are organized into threebroad categories: capacity/resource adequacy, energymarkets, and ancillary services.

Section 7 describes the lessons learned, near-termopportunities for demand response, and key challenges toexpanded participation by demand response resources.

The Terminology Appendix provides denitions ofterms used in this report, and an acronym list is providedprior to the Executive Summary.

1 In this report we use balancing services to mean regula-tion and load-following services that are employed undernormal system conditions to correct supply and demandimbalances within seconds or minutes.


15/76

13


A. What is Demand Response?

Electricity demand varies signicantly by timeof day as well as by season. Historically, thebalancing of electricity supply and demand wasperformed only by increasing or decreasing

the electrical output of power plants, but this oftenrequires large investments in capital-intensive facilitiesthat are infrequently used, or the dispatch of increasinglyinefcient (and therefore expensive) generators.

Demand response was originally developed by electricutilities in order to increase exibility on the demand sideby temporarily shifting or reducing peak energy demand,thereby avoiding costly energy procurements and capacityinvestments for a small number of hours of need. Withthe shift toward competitive electricity markets, demandresponse has become an important tool used by manyutilities and system operators in the United States toenhance grid reliability and market outcomes.

The Federal Energy Regulatory Commission (FERC)currently denes demand response as:

Changes in electric usage by demand-side resources from their normal consumption patterns in response tochanges in the price of electricity over time, or to incentive payments designed to induce lower electricity use at timesof high wholesale market prices or when system reliabilityis jeopardized.(FERC 2012)

As variable resources such as wind and solar provide

an increasing proportion of electricity to the grid, newforms of demand response are being developed with

capabilities that surpass traditional peak load-reducingdemand response. 2 This next generation of demandresponse is automated and often linked to some form ofenergy storage in order to quickly respond to changes insystem frequency or to increase demand during periodsof oversupply, which improves the utilization of manyrenewable resources, as well as traditional thermal units.

In recognition of this expanded role, FERCs 2010National Action Plan for Demand Response emphasizesthat demand response includes actions that can changeany part of a customers load prole, not just the periodof peak demand. This denition specically includesthe smart integration of changeable consumption withvariable generation, such as through energy storage(using devices such as electric vehicle batteries andthermal storage), and the associated provision of ancillaryservices such as regulation and reserves (FERC 2010).

The benets offered by demand response arenumerous, but they fall into three general categories:economic efciency, system reliability, and environmentalbenets. The economic benets consist primarily of lowerwholesale market prices due to demand responses abilityto displace the most expensive peak generation resources,as well as the deferment or avoidance of more costlynew capacity construction by attening the demandcurve.3 The exibility of demand is key to ensuringwholesale market efciency; enhancing the elasticity ofdemand more accurately reects consumers willingness

to pay and mitigates the ability for suppliers to exercisemarket power. Additional economic benets may include

2. Background

2 Variable as used in this paper refers to any source of elec-tricity production where the availability to produce elec-tricity is largely beyond the direct control of the operator.It can be simply variablechanging production indepen-dently of changes in demandor variable and uncertain.

Another term for this latter category is intermittent.

3 A 2007 report found that in the PJM regional electricitymarket, a three percent load reduction in the 100 highestpeak hours corresponds to a price decrease of six to twelvepercent. (Brattle Group 2007)


16/76

14


4 Participant nancial benets consist of the bill savings andincentive payments that customers receive in return forcurtailing, shifting, or otherwise modifying their load. Riskmanagement benets are related to the ability of demandresponse investments to diversify generation portfolios andavoid large capital investments in new power plants thatmay experience shocks in terms of fuel costs, constructioncosts, or future environmental regulations (Binz, et al.2012).Through its ability to increase the elasticity of

demand, demand response also deters generators fromexercising market power (US Department of Energy 2006).

5 We note, however, that the use of distributed back-upgeneration from fossil fuels (e.g., diesel fuel or naturalgas) as demand response can reduce demand responsesenvironmental benets. For this reason, some regions haverestricted the use of fossil-fuel powered back-up generatorsthat may qualify to provide demand response.

participant nancial benets and risk management. 4 Demand resources may also be called upon by system

operators to maintain the reliability of the electric systemin the event of an emergency and avoid brownouts orblackouts. In addition to reducing capacity constraints,

some demand resources can be used to provide ancillaryservices such as reserves or balancing by quickly increasingor decreasing demand. System stability is thus improvedthrough better aligning the movement of generation supplyand electricity demand. The services provided by variousresources from generation to storage to demand response are depicted in Figure 1. The gure demonstrates thedegree to which demand response can provide bothunique services (i.e., peak load reduction), as well as someof the services offered by traditional generation and storageresources.

The environmental benets of demand responseresources will vary from region to region. Environmentalbenets are mostly associated with the displacement ofmarginal fossil fuel resources. Each region of the countryhas a different generation resource mix with variouspercentages of fossil resources (coal, oil, and natural gas),nuclear, hydroelectricity, renewables, etc. During differentseasons of the year and depending on the time of day, thedisplacement of marginal emissions due to the dispatch ofdemand response resources can vary substantially. 5

Additionally, environmental benets may result fromdemand responses ability to facilitate the integration ofrenewable resources. The exibility of demand responseallows the electric system to accommodate higher penetra-tions of variable resources such as wind and solar, whoseenergy output can uctuate quickly and lead to excessgeneration supply on the system. Demand resources thatinclude a form of energy storage are particularly well-equipped to facilitate the management of periods of over-

supply through the provision of load-following or regula-tion services that can both increase or decrease demand.

Demand responses load modifying capability thusenables more efcient use of current electricity generationresources, while yielding economic, reliability, and

environmental benets. Yet demand response is not ahomogenous resource; it is provided by a highly diverseset of actors in numerous different ways, and withvarying capabilities. This diversity precludes any simplecharacterization of demand response types and alsocontributes to the exibility of demand response to meetmultiple system needs. An overview of the various formsof demand response is given in the following section.

DemandResponse

Peak Load Reduction

Generation Storage

Figure 1

Services Provided by Various Resources

Wider Operating Range(lower minimumcapacity)

DispatchableWind/Solar

VoltageSupport

RegulationFast Ramping

FrequencyResponse

DispatchableQuick Start

LoadShifting

Absorptionof Excess

Generation

Source: (Liu 2012)


17/76

15


B. Demand Response Provision andClassication

Demand response can be provided by all categoriesof customers (industrial, commercial, and residential)employing many different technologies or strategies toachieve shifts in demand. Common examples include:

Reducing or interrupting consumption temporarilywith no change in consumption in other periods

Shifting consumption to other time periods Temporarily utilizing onsite generation in place of

energy from the grid 6

In addition, demand response can provide frequencyregulation and load-following services. During periodsof excess energy production, demand response resourcesthat have an element of storage may increase the energyused for heating or pumping water, charging batteries,compressing air, or freezing ice for cold storage. Therate at which these activities occur can be automaticallyadjusted to align consumption with generation output.

Demand responseresources interfacewith wholesale marketsin two distinct ways:either as resourcesthat are dispatched by

a system operator, oras non-dispatchableresources that may elect(voluntarily) to adjusttheir consumptionbased on price signals.Dispatchable resourcestypically bid directlyinto wholesale marketsor enter into contractsto receive paymentsfor demand reduction,whether in response to areliability event or highmarket prices.

In contrast, non-dispatchable resources

6 As noted previously, the environmental benets of demandresponse may be reduced by fossil-fueled back-up genera-tion. In this report we refer to demand response generally,

generally participate in price-based demand responseprograms such as real-time pricing, critical peak pricing,and time-of-use tariffs. These price-based programs provideusers with ongoing price signals to encourage lower energyconsumption during periods of high electricity prices,

but are generally not considered rm resources, as theyare not dispatchable and grid operators do not knowthe degree to which customers will respond. These non-dispatchable price-based programs may become moreprevalent for residential and small commercial customersas smart meters are deployed. Advanced meters alsocreate opportunities for aggregating residential and smallcommercial customers in ways to provide dispatchableservices to system operators, which may be compensatedthrough either contract prices or market prices. The varioustypes of demand side resources available, spanning bothdemand response and energy efciency, are depicted inFigure 2, below.

For this report, we focus primarily on dispatchabledemand response resources that may be dispatched

Figure 2

Demand Side Management Categories

Source:(NERC 2011)

irrespective of the means employed to shift or reduce load,in part due to the lack of data regarding the amount ofdemand response provided by back-up generation.


18/76

16


7 A notable early failure of telemetry was with electricwater heater control programs that relied on a timingclock attached to the individual water heater. The timingdevice that switched the electric water heater on and offwas affected by local power outages and the timers couldonly be reset by the local utility. The result was that manyof the electric water heater timers were not properlysynchronized and some of performed opposite to theprogram design.

8 For example, the Midwest Independent System Operator(MISO) initially required real-time telemetry for demandresponse resource participation in all types of ancillaryservices, but later found this to be unnecessary forprovision of reliable spinning and non-spinning reserves(Pfeifenberger and Hajos 2011).

for the variety of purposes described above, including:ensuring resource adequacy (capacity), providing otherreliability functions such as reserves and balancingservices, and responding in energy markets to highmarket prices. In general, this report examines the ways

that demand response can provide specic services at thecontrol and discretion of system operators to improvethe overall performance and stability of electric powersystems.

Demand response participates in all regions of theUnited States. The participation occurs across a fullcontinuum of structures from integrated, centrallymanaged, mandatory wholesale markets at one extremeto vertically integrated utility areas that have voluntarybalancing services at the other extreme.

One key distinction along this continuum of marketparticipation is the ability of fully-integrated demandresponse to set the market clearing price. Any reductionin load will reduce the overall cost of serving electricityduring that timeframe, but fully integrated demandresponse in areas with wholesale markets can have amuch larger price impact due to its ability to reduce themarket clearing price for all market participants. There isfurther discussion of this distinction in section 6.D.

In regions where demand response does notparticipate in wholesale markets (whether or not these

markets exist), demand response may be carried out bydistribution utilities to avoid high peak energy costs,ensure system reliability, or provide balancing services.

INFRASTRUCTURE R EQUIREMENTSIrrespective of the market structure in which

dispatchable demand response operates, to receivecompensation, these resources must comply withdispatch signals from the system operator and changesin demand must be measured and veried. Measurementand verication typically requires a certain level

of metering accuracy and telemetry infrastructureinvestment. 7 In particular, programs that offer incentivesfor participation (and/or penalties for non-compliance)must calculate the baseline load and measure the changefrom this baseline that occurs during a demand responseevent in order to calculate the total change in demand.

Meter requirements for dispatchable demand responsemay include 15-minute or ve-minute interval meters,while telemetry requirements vary. Telemetry is generallyadded to ensure stable operation of the network, and,depending on the size of the demand resource and typeof service offered, telemetry requirements may range from

after-the-fact metering to four-second real time telemetryequipment to enable system operators to monitor loadsand ensure that the contracted change in demand ismet (Isser 2008). In order to set the real-time marketclearing price, demand resources are typically required

to have sufcient telemetry and capability to receive asystem operator dispatch. However, for small resources,advanced metering and telemetry requirements canbe prohibitively expensive and may not be necessary(Pfeifenberger and Hajos 2011). 8

Telemetry is required for regulation by all systemoperators, but not all system operators require it forspinning reserves. In some regions, data granularity istwo seconds for telemetry, but it can be batch sent onceevery minute for demand response. This requirement iseasier to accommodate, particularly for demand responseaggregators. As metering accuracy increases, so do costs.Some system operators also require that data are reportedfrom each individual resource, while others only requirethat data be available and veriable in aggregate form(MacDonald, et al. 2012).

A key element in enabling demand response toprovide these various services has been the developmentof appropriate market regulations. Through a series ofrecent orders, FERC has established wholesale marketrules that facilitate demand response resources ability to

participate actively in markets to provide energy, capacity,or ancillary services in a manner similar to generationresources. These regulations and case studies of demandresponse utilization will be explored in greater detail insections 4 through 6.


19/76

17


C USTOMER B ASELINESThe amount of electric service provided by genera-

tion whether central station power plants or distributed can be measured through metering the actual energyoutput from the plant over a certain time interval. In

contrast, demand reduction at a customer facility is achange in usage pattern and cannot be as easily calcu-lated. Underlying the performance of demand responseis the concept of a baseline the amount of energy thecustomer would have consumed absent a dispatch signalfrom the system operator.

Figure 3

Illustration of Baseline Concept

Baseline

Expected orContracted

LoadReduction

MeasuredLoad

DeploymentInstruction

ReductionDeadline

Release NormalOperations

E l e c t r i c i

t y U s a g e

Time

Source:(NERC 2011)

Figure 3 illustrates the basic idea behind baselines. Ona day without a dispatch signal, the customer would haveused electricity over time along the green line labeledbaseline. Given the dispatch signal, the customer willreduce their load during a few hours, by the amount ofthe gap between the baseline and the dotted line labeledMeasured Load. As indicated above, knowledge of load

over time requires time-interval meters and a method forrecording their output for reporting to the grid operatoror other authority.

The trouble, of course, is that few customers have aload prole that is so regular that their baseline is simpleto calculate. While retail outlets may have a predictabledaily and weekly usage pattern, their load often shiftssignicantly with holidays. Primary and secondaryschools have a drastically different usage during summerand school holidays. Factories often undergo shutdownsfor routine maintenance, and order requests from theircustomers may uctuate over time. As such, different

regions have employed differing methods for calculatingbaselines. Numerous studies have analyzed diverseapproaches to ensuring that customer baselines areaccurate and remain accurate over time, under variousdispatch scenarios.

An example will be helpful. At one time, theIndependent System Operator of New England had abaseline methodology that assumed that all customerswould respond on only a very small number of days.

As such, the method for setting baselines required thata new customer report interval load data to the systemoperator for ten days before being declared fully readyto respond. Once this initial period ended, the baselinewas always set by the most recent ten days of load, whichwas accurate enough to capture weekly load patterns,and seasonal differences. A small number of customers,however, found a loophole. These customers turnedoff on-site distributed generation units for maintenanceduring the initial reference period. During this time, thefacilities were pulling the entire electric load needed tooperate from the grid, and their baselines reected thislevel of usage. Once the initial ten-day period ended, theyrestarted the onsite generation, and the load pulled fromthe grid dropped signicantly, every day. They offered andcleared their resources as demand response every day, andthe baseline was not reset for many months. They were

paid for apparent reduction of load that was not a changefrom their normal usage. Although they appear to havebeen acting within the rules of the program, all outsideentities aware of the situation - including the FERC Ofceof Enforcement - view this activity as a violation of thespirit of demand response. The baseline methodologyfor New England has since been changed such that nobaseline can remain at a set level for more than sevenconsecutive business days.

Any method for calculating baselines must meet twobasic criteria. First, it must be one that is trusted by the

grid operator to be reliable. The entity that dispatchesthe demand response resources must be condent thatthey are receiving the change in load that they expectand that is being reported to them. This is true both forthe reliable operation of the electric system and for theaccurate settlement of payments for the electric serviceprovided, whether it be energy, capacity, reserves, orother service. Second, it must be a method with whichthe customer and/or the DR provider can comply, bothnancially and feasibly. Reporting interval meter datato the regional operator days after a demand responseevent is reasonably inexpensive and easy, and is often


20/76

18


sufciently accurate and timely. On the other hand,sub-second, real-time metering and communicationequipment may be feasible but not nancially viable,

especially for large aggregations of smaller customers. Without feasible, trustworthy baselines, demand

response will not succeed.


21/76

19


A. Demand Response as a ReliabilityResource Prior to Restructuring

Demand response in the United Statesoriginated in the 1970s, in part due to thespread of central air conditioning, whichresulted in declining load factors and

needle peaks during hot summer days. 9 The advent ofintegrated resource planning in the late 1970s and1980s drew attention to the high system costs of meetingthese peak loads and encouraged utilities to look for loadmanagement alternatives (Cappers, Goldman, and Kathan2009). 10 Rate design (particularly time-of-use pricing)and incentive programs became standard demandresponse programs at many regulated utilities.

Incentive programs such as direct load control andinterruptible/curtailable programs allow a utility to curtaila portion of a customers load in exchange for a monetary

incentive, such as a credit on the customers monthlybill or a lower overall electricity rate. Direct load controlprograms involve the installation of control technologieson a customers appliance typically an air conditioner,water heater, or pool pump and are primarily offered toresidential consumers. Interruptible/curtailable programstarget large industrial and commercial consumers whohave the capability of completely shutting down theiroperations or reducing their demand by a predeterminedamount upon notice by the utility or system operator.In exchange, these large commercial and industrial

customers often receive lower electricity rates. Although interruptible/curtailable rate programs

were popular during the 1980s and 1990s, in manycases customers were rarely called upon to reduce theirload. For example, Southern California Edison had thelargest interruptible load program in California, but didnot invoke a single interruption for fourteen years until

June 2000, despite customers receiving bill reductionsof approximately 15 percent for participating (Marnay, etal. 2001). In Vermont, many ski areas received reducedelectric rates in exchange for the ability to interruptsnow-making equipment during times of peak winterloads. A major issue arose with one ski area when an

3. History

unusual weekend peak load coincided with a holidayweekend and the ski area was unwilling to shut downits snow-making equipment during one of its busiest skiweekends. The low utilization of these programs resultedin unofcial economic incentives for large customersand a limited ability or willingness to curtail load byparticipants when called more frequently, as describedmore below (Fryer, et al. 2002).

Until the late 1990s, the US electric industry consistedprimarily of vertically integrated utilities that managedtheir own generation and distribution assets. Thedemand response programs of the 1970s through muchof the 1990s were largely conducted by such utilitiesin a structured, regulated environment, and thereforeconsumers were not exposed to real-time wholesale pricesignals, nor were consumers compensated for the fullsystem value of their demand reduction. This began tochange in the 1990s as the US electric industry initiatedthe restructuring process.

B. Early Wholesale Market Programs

The US electric industry began to shift towardgreater competition in the 1990s following the EnergyPolicy Act of 1992 that allowed independent powergenerators to participate in wholesale markets and FERCOrder 888 that mandated open access to transmissionsystems. Wholesale markets themselves had undergonea transition from cost-of-service principles to greater

competition in the late 1980s when FERC began to grant

9 Some northern states also experienced sharp peaks duringwinter cold snaps, particularly in regions with highpenetrations of electric space heat.

10 Integrated resource planning refers to the evaluation ofdemand and supply resources by public utilities andstate regulatory commissions to cost-effectively provideelectricity service. Integrated resource planning differsfrom earlier planning techniques in that it also considersenvironmental factors, demand-side alternatives, andrisks posed by different investment portfolios (Hirst andGoldman 1991).


22/76

20


markets that span much of the Midwest and northernGreat Plains; the Southwest Power Pool (SPP) operatesan energy imbalance service market in the CentralPlains states; California ISO (CAISO) covers themajority of California; and the Texas market is managed

by the Electric Reliability Council of Texas (ERCOT).ERCOT is the only ISO region that is not subject toFERC regulation. ISOs and RTOs also oversee andoperate the high-voltage bulk power system, coordinateelectricity generation, and conduct long-term regionalplanning (US Department of Energy 2006).

Other regions including the Southeast, Southwest,Inter-mountain West, and Pacic Northwest chose toretain the traditional vertically-integrated utility model.Balancing authorities operate in these areas to maintainthe minute-to-minute balance between electricitysupply and demand within their borders. Many utilitiesand/or balancing authorities in these regions operatedemand response programs, such as the BonnevillePower Administration in the Pacic Northwest and theTennessee Valley Authority in the Southeast.

While the process of restructuring increasedcompetition and established regional wholesalemarkets, it also shifted responsibility for maintaininggrid reliability away from utilities to system operators,thus reducing incentives for traditional utility-run

demand side managementprograms. As utilitiesdivested their generationassets, many no longersaw value in maintainingsuch programs to ensurereliable and efcient gridoperations (Electric EnergyMarket Competition TaskForce 2007). Demand sidemanagement spending

peaked in 1993 atapproximately $2.7 billionnationwide. By 2003, this

11 FERC granted powerproducers the right tosell at market rates onlyif they could show thatthey lacked marketpower and that the prices

reected actual marketdynamics of supply anddemand.

wholesale power producers the ability to sell at market-based rates based on the dynamics of supply anddemand (Joskow 2001). 11

As the process of restructuring gathered steam inthe late 1990s, many states elected to experiment with

competitive markets, transforming their vertically-integrated utilities into stand-alone generation companies,regulated distribution companies, and regional gridoperators. The regional grid operators are referred to aseither independent system operators (ISOs) or regionaltransmission organizations (RTOs), and exist in manyregions of the United States and Canada, as shown inFigure 4.

Regional system operators are responsible formanaging the wholesale power markets (including real-time and day-ahead energy markets, ancillary servicesmarkets, and, in some cases, capacity markets). Thesewholesale power markets represent approximately two-thirds of US electricity demand (Market Committee of theISO/RTO Council 2007).

ISO-New England (ISO-NE) operates the powermarket in the Northeast; New York ISO (NYISO) managesthe markets in New York State; PJM Interconnection(PJM) is responsible for the markets that cover the Mid-

Atlantic states and parts of Ohio, Indiana, and northernIllinois; the Midwest ISO (MISO) operates the power

Figure 4

Regional System Operators in the United States and Canada

Source:(ISO/RTO Council 2011)


23/76

21


value had fallen by more than half to $1.3 billion. Thechange in program enrollment from 1998 to 2003 foreach reliability region is shown in Figure 5, illustratingthat program enrollments dropped precipitouslyalong with funding, although it is worth noting that

enrollments declined in virtually every region, even thosewhere no divestment or restructuring occurred. 12

12 Reliability regions were created under the North AmericanElectric Reliability Corporation (NERC) after the NortheastBlackout of 1965. Note that the reliability regions do notcorrespond directly to the ISO/RTO areas. In addition,some of these reliability regions have merged since 2003.

13 In Maryland, Baltimore Gas & Electrics retail ratesincreased 72 percent in 2006, leading to the dismissal ofthe states ve public utility commissioners (Pfeifenberger,

Interruptible and Direct Load Control by NERC Region

Figure 5

Change in Program Enrollmentsfrom 1998 to 2003

Note: Data from NERC 1998 and 2003 summer assessments.NPCC data is for for 1998 and 2002.

Source: (Goldman 2005)

Soon it became evident that competitive wholesalemarkets, as initially implemented, would notautomatically produce all the benets initially expected,

and that additional structures would be needed tomanage energy price volatility, ensure system reliability,and guard against market power. An example of thedifculty involved in the restructuring process wasthat encountered during the Western States PowerCrisis of 2000-2001. During this period, Californiastotal wholesale power costs more than tripled andwholesale market prices soared to almost $400 permegawatt hour (Weare 2003, Joskow 2001). Traditionaldemand response programs were subjected to new andunexpected stresses.

During this time, California customers oninterruptible/curtailable rate programs, many of whom

had never previously been called upon to curtail, weresuddenly subject to frequent interruptions 23 timesduring the last eight months of 2000 alone. Due to thehigh number of interruptions, many customers beganto leave the program or chose to ignore curtailment

instructions and suffer penalties (Goldman, Eto, andBarbose 2002).

Following the Western States Power Crisis andsubsequent sharp electric rate increases in someregions,13 policymakers concluded that investments indemand response were necessary to ensure the efcientfunctioning of the wholesale markets and reliability of thegrid (Cappers, Goldman, and Kathan 2009).

At the same time, the potential for demand responseto bring benets to electricity customers was expandedby restructurings greater emphasis on wholesale markets.Previously, vertically integrated utilities had carried outdemand response programs to prevent blackouts orcontrol costs during peak periods, but with the expansionof wholesale markets, the scale of impact was greatlybroadened. Instead of waiting for a utility to curtail loadduring a few hours a year, demand response providerswere empowered to participate on an ongoing basis inthe market to reduce volatility, improve the elasticityof demand, and potentially reduce the market clearingprice for energy purchases for a much larger number of

customers across entire regions.Moreover, the pool of potential participants widenedconsiderably due to the restructured wholesale marketscreation of opportunities for entrepreneurs to ndinnovative means to supply demand response (York andKushler 2005). First, however, the necessary policy andregulatory framework had to be established to governthe treatment of demand response and enable demandresponse to be compensated in a manner comparableto generation resources. In the following section, wedescribe these developments, which were initiated

in large part by FERC proactively responding to theidentication of market barriers to demand response.

Basheda, and Schumacher 2007). Other examples of priceincreases following restructuring include a 50 percentincrease in Connecticut and a 59 percent increase inDelaware (Pfeifenberger, Basheda, and Schumacher 2007).In several situations, the rate increases reected up to tenyears of suppressed rates due to rate caps negotiated at the

start of restructuring. In California, total wholesale powercosts rose from $7.4 billion in 1999 to $27 billion per yearfrom 2000 through 2001 (Weare 2003).


24/76

22


14 The FERC Order refers to the full market price of energyas the locational marginal price (LMP) to reect the fact

that in FERC regulated markets the energy price includes alocational signal (to varying degrees).

C. Recent Developments to EnableDemand Response in US Markets

The transition to competitive wholesale marketsentailed signicant market design efforts, which initiallyfocused on supporting the participation of traditionalgeneration resources, rather than demand-side resources.This supply-centric focus created numerous barriers todemand response participation. These included restrictiverules that increased the cost of participation; limitationson the entities allowed to bid in demand response; afailure to provide compensation for certain services;and, in some cases, the outright prohibition of demandresponses participation in the market (FERC Staff 2009).

As a result, modications by FERC and state regulatorshave been necessary in order to provide legacy utilitydemand response programs as well as new providersof demand response with the opportunity to fullyparticipate in the organized wholesale markets (Cappers,Goldman, and Kathan 2009).

Federal support for demand response was underscoredwith the Energy Policy Act of 2005, which stated thatthe ofcial policy of the United States was to encouragedemand response, facilitate the deployment of enablingtechnology and devices, and to eliminate unnecessarybarriers to demand responses participation in energy,

capacity, and ancillary service markets. The Act alsodeclared, the benets of such demand response thataccrue to those not deploying such technology anddevices, but who are part of the same regional electricityentity, shall be recognized, implying that an accurateassessment of the benets of demand response must takeinto account impacts on all regional customers.

In order to implement the policy goals articulatedin the Energy Policy Act of 2005, modications to theoperation of wholesale markets have been required to setdemand response on an equal footing with generationresources. Over the past few years, FERC has takenmultiple steps to remove remaining barriers to demandresponse. These include:

Order No. 890, issued in February 2007. This ordermodied the Open Access Transmission Tariff toallow non-generation resources such as demandresponse to provide certain ancillary services(e.g., regulation and frequency response, spinningreserves, and supplemental reserves services) on acomparable basis to services provided by generation

resources. The order also directed transmissionproviders to treat demand response comparably to

traditional resources in the transmission planningprocess.

Order No. 719, adopted in October 2008. Knownas the Wholesale Competition Final Rule, thisorder established regulations to improve the

competitiveness of wholesale electric marketsthrough demand response. The rule required systemoperators to study, and if necessary, reform marketrules to ensure that the market price for energyreects the value of energy during an operatingreserve shortage, in order to encourage the entry ofnew resources, including demand response.Order No. 719 also directed system operators toaccept bids from demand response resources inthe provision of certain ancillary services on abasis comparable to other resources, and permitsaggregators of retail customers to bid demandresponse on behalf of retail customers directly intothe wholesale market, unless prohibited by law orregulation (FERC Staff 2009).

Order No. 745, issued in March 2011. This ruleaddresses the compensation for demand responsein wholesale energy markets, requiring that demandresponse be compensated the full market price ofenergy, 14 when it is determined that the resource iscapable of balancing supply and demand and is cost-

effective. Further, the order species that costs are tobe allocated among customers who benet from thelower market price of energy resulting from demandresponse.The basis for this compensation requirement isto provide comparable compensation to bothgeneration and demand response providers, basedon the premise that they provide comparableservices to the grid operator. In addition,compensation based on the full market price ofenergy is designed to facilitate the recovery of

demand response technology investment costs,thereby encouraging greater participation ofdemand response in wholesale markets.

Order No. 755, issued in October 2011. This orderpertains to compensation of resources providingregulation service. FERC found that resourcesproviding such services differ in their ramping


25/76

23


ability and the accuracy of theirresponse, yet compensationby system operators did notaccount for such differences.Thus FERC found rates unjust

and required system operatorsto base payment in part on theperformance of each resource.

The long-term impacts of FERCsmore recent orders are yet to bedetermined. In several cases, systemoperators have not yet nalized theircompliance lings detailing theirmarket rule changes. In other cases, thecompliance lings have been approved,but the implementation of marketdesigns to accommodate changes arenot scheduled to occur for severalyears. For example, ISO-NE recentlymade its third compliance lingrelating to Order 755 requirementsto make regulation market changes toprovide for two-part bidding, uniformpricing, and two-part payments toproviders of regulation service. Thesechanges will allow new technologies (other than traditional

generation resources) that can provide regulation servicesto compete in ISO-NEs regulation market. However, theearliest implementation of the changes would be for the2015-2016 power year.

In regard to Order No. 745, each ISO or RTO hassubmitted compliance lings to FERC including tariffrevisions to implement the orders requirements. PJMimplemented the changes in April 2012 and observedincreased demand response participation in the summer of2012 (discussed in Section 5). ISO-NE has commenced amulti-year transition period to fully implement the rules.

The other major system operators, MISO, NYISO, SPP,and CAISO, are in the process of nalizing their Order No.745 lings. ERCOT is not subject to FERC regulation ofits markets and has elected to determine compensation forERCOT demand response programs separately from theenergy market.

D. Demand Response Today

The amount of demand response available to systemoperators has begun to rebound in many regions sincethe low levels reached at the beginning of restructuring.

In 2010, the potential resource contribution of demandresponse to system operators in the United States totaled31,702 MW. As a percentage of peak demand, theseresources provided between 2 and 10 percent of eachregions peak demand, as shown in Table 1, above (FERC2011).

Currently there are numerous ways in whichdispatchable demand response can operate. In regionswith organized wholesale markets, demand responseresources can typically bid directly into the market orbe dispatched in response to market signals. However,the degree to which demand response is integrated into

the wholesale market varies, with some regions allowingdemand response to set the market clearing price,while other regions restrict demand responses ability toinuence market prices. Finally, across the United States,

Table 1

Demand Response Available at US ISOs and RTOs 15

15 The decline in demand response enrollment for CAISO isdue to the way that DR capacity is assessed and reported,from planning estimates to ex post estimates. The declinefor NYISO is not consistent with NYISO annual reportsand is likely due to differences in denitions and the way

data were reported between the FERC survey and NYISOsannual report.

Source: (FERC 2011)


26/76

24


Table 2

Ancillary Services Provided byDemand Response

and particularly in areas without wholesale markets,utilities may maintain their own demand responseprograms such as direct load control for water heatersand air conditioning units.

The remainder of this report focuses primarily on

demand response programs in regions with wholesalemarkets, but also includes results from a pilot programin the Pacic Northwest that operates outside of anywholesale market. In general, regional dispatch ofdemand response through system operators provides amore exible and sophisticated means of addressing avariety of system needs. Therefore, the majority of thecase studies in this report relate the US experience to datewith centrally-dispatched demand response.

F ULLY -I NTEGRATED M ARKET -B ASED D EMAND R ESPONSE

In some organized wholesale markets under FERCregulation, demand response has been fully integratedinto the various electricity markets. Fully-integratedmarket-based demand response implies that demandresponse can set the market clearing price, rather thanmerely reacting to the clearing price. These resources arealso dispatched by the system operator. Today, market-based demand response performs the following roles:

Energy Resource: Demand response that

participates in the energy market is dispatched foreconomic reasons. Demand response providers maybid their demand reduction directly into either theday-ahead market or the real-time market. If the bidis less than the market clearing price, the resource isdispatched by the system operator and receives theenergy market price as payment.

Capacity Resource: Demand response can play akey role in ensuring resource adequacy. In regionsthat have capacity markets, demand responseproviders may bid in a set amount of load that

can be curtailed during a capacity shortfall. Theseproviders must have the capability of curtailingload on short notice (usually within 30 minutes totwo hours). The number of times that the providermay be called upon to provide a demand reductionvaries with the specic market product, e.g., rangingfrom a maximum of ten times a year to an unlimitednumber of interruptions. Providers of this type ofdemand response receive capacity payments.

Ancillary Services Resource: Demand respo

synapse hurley demandresponseasapowersystemresource 2013 may 31 (1)

Documents