How to Get More Response from Demand Response

Home , Rolling blackout

Scott Neumann is Chief Technology Ofﬁcer of Utility Integration Solutions, Inc. (UISOL), a consulting ﬁrm based in Lafayette, California. His professional background includes working with Empros, CES International, MessageTek, and UISOL. Scott leads the U.S. efforts within IEC TC57 for the development of standards related to power system operations and associated information exchanges.

Fereidoon (Perry) Sioshansi is President of Menlo Energy Economics, a consulting ﬁrm with experience in demand-side management, How to Get More Response real-time pricing, enabling technologies, and demand response. His professional background from Demand Response includes working at Southern California Edison Company, the Electric Power Research Institute, and Global Energy Decisions. He has a Ph.D. in Economics from Purdue University. Despite all the rhetoric, demand response’s contribution

Ali Vojdani is CEO of Utility Integration to meet peak load will remain elusive in the absence Solutions. His professional background of enabling technology and standardized business includes working with Pacific Gas and Electric, EPRI, Perot Systems, Vitria protocols. Technology, and UISOL. He has a Ph.D. in Electrical Engineering. Scott Neumann, Fereidoon Sioshansi, Ali Vojdani and Gaymond Yee is a Research Coordinator at the California Institute for Energy and the Gaymond Yee Environment (CIEE). His professional background includes working at Silicon Energy, Diablo Research, EnergyLine Systems, and Science Applications International Corporation. He has a Master of Engineering Degree in Mechanical Engineering from the I. Summer’s Heat Wave When operators run out University of California at Berkeley. shows DR’s Potential of generation capacity, the alternatives are either to resort to This article is based on a research project managed by the California Institute for Energy This past summer’s heat wave rolling blackouts – which nobody and the Environment on behalf of the California engulfed most of the U.S. and likes – or to plead to customers to Energy Commission (CEC), with funding from the Public Interest Energy Research (PIER) stretched many utilities and sys- drop discretionary loads. This program. This Demand Response Enabling tem operators close to the limit. In practice, where customers are Technology Development (DR ETD) project California, the Independent Sys- provided financial incentives to was conducted by a team of consultants led by Utility Integration Solutions, Inc., with tem Operator (CAISO) had to drop load during emergencies is contributions from Dynamic Networks, Menlo declare several Stage 1 and Stage 2 known as demand response or Energy Economics, Michigan Group, Nexant, alerts in July,1 signaling that its DR.2 Infotility, Savvion, and TIBCO. Participating stakeholders included the California reserve margin was precariously urrently, system operators Independent System Operator (CAISO), low. Similar episodes were C have limited capabilities to Pacific Gas & Electric Company, Southern California Edison Company, San Diego Gas & repeated across the country in engage in DR for a number of Electric Company, DR service provider July and August as new peak load reasons. Most importantly, the Infotility, and a few representative commercial records were repeatedly set and protocols for sending the signal and residential customers. broken (Figure 1). System opera- that capacity is tight and voluntary tors, once again, were reminded load shedding is needed is time- of the real value of demand consuming, error-prone, and response programs. mostly manual. Since system

operators must balance load in recent report by the Federal III. How Much DR is real-time, any delays to get a signal A Energy Regulatory Really There? out, get confirmation, and get Commission (FERC), for example, tangible results must be fast, provides some evidence.4 One Following the passage of the error-free and automatic – study claims annual savings of Energy Policy Act (EPAct) in characteristics that are lacking $15 billion per year in the U.S. for August 2005, there has been in most current systems. shifting 5 to 8 percent of con- renewed interest in smart meters, This means that in many cases, sumption from peak to off-peak time-variable pricing, and the operator may resort to hours and for depressing peak demand response – the legs of a involuntary load shedding simply demand by 4 to 7 percent.5 stool. EPAct’s main contribution because of inherent delays and Another study looking at the New was two-fold: First, it codifies the inefficiencies in implementing England ISO’s service area claims significance of enabling technol- DR programs. annual savings of $580 million per ogies, which are prerequisites to year for reducing peak demand wider implementation of DR. by as little as 5 percent.6 Second, it instructs both the II. Is DR Cost-Effective? During the August heat wave, Department of Energy (DOE) and PJM Interconnection reported FERC to establish baselines and A number of studies have cost savings totaling $650 million goals for increased reliance on confirmed the cost-effectiveness attributed to DR programs.7 On DR. of DR relative to the alternatives, Aug. 2, 2006, alone, when PJM set s a consequence, DOE namely reliance on peaking units a new peak load record of A published a report in or rolling blackout. The former 144,796 MW, it reported DR sav- February estimating DR’s poten- are expensive – as everyone ings of $230 million.8 Similar tes- tial benefits, offering recommen- recognizes – and the latter even timonials are available from other dations on how these benefits more so.3 ISOs and RTOs. may be captured.9 As a starting

October 2006, Vol. 19, Issue 8 1040-6190/$–see front matter # 2006 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2006.09.001 25 Table 1: What is DR? Price-Based Options Incentive-Based Programs Time-of-use (TOU): a rate with different unit prices for usage Direct load control: a program by which the program during different blocks of time, usually defined for a 24-hour day. operator remotely shuts down or cycles a customer’s electrical TOU rates reflect the average cost of generating and delivering equipment (e.g., air conditioner, water heater) on short power during those time periods. notice. Direct load control programs are primarily offered to residential or small commercial customers. Real-time pricing (RTP): a rate in which the price for electricity Interruptible/curtailable (I/C) service: curtailment options typically fluctuates hourly reflecting changes in the wholesale price integrated into retail tariffs that provide a rate discount or of electricity. Customers are typically notified of RTP prices on a bill credit for agreeing to reduce load during system day-ahead or hour-ahead basis. contingencies. Penalties may be assessed for failure to curtail. Interruptible programs have traditionally been offered only to the largest industrial (or commercial) customers. Critical peak pricing (CPP): CPP rates are a hybrid of the TOU and Demand bidding/buyback programs: customer offers bids RTP design. The basic rate structure is TOU. However, provision is to curtail based on wholesale electricity market prices or made for replacing the normal peak price with a much higher an equivalent. Mainly offered to large customers CPP event price under specified trigger conditions (e.g., 1 MW or over). (e.g., when system reliability is compromised or supply Emergency demand response programs: programs that prices are very high). provide incentive payments to customers for load reductions during periods when reserve shortfalls arise. Capacity market programs: customers offer load curtailments as system capacity to replace conventional generation or delivery resources. Customers typically receive day-of notice of events. Incentives usually consist of up-front reservation payments, with penalties levied for failure to curtail when called upon to do so. Ancillary services market programs: customer bid load curtailments in ISO/RTO markets as operating reserves. If their bids are accepted, they are paid the market price for committing to be on standby. If their load curtailments are needed, they are called by the ISO/RTO, and may be paid the spot market energy price. Source: Benefits of DR in Electricity Markets and Recommendations for Achieving Them, US DOE, Feb. 06.

point, DOE divided DR programs signal that there is an impending potential to be around into two basic categories, price- emergency and ask for customers 20,500 MW, or 3 percent of the and incentive-based (Table 1), to shed load in exchange for peak load. Surprisingly, DOE and identiﬁed various sub- predetermined (the former case) found that DR capacity has actu- categories under each. Following or negotiated (the latter case) ally declined from its peak in these deﬁnitions, the focus of the prices/incentives. 1996, when more utilities engaged present article is on demand he DOE study estimated in such programs and had more bidding/buyback and emergency T current (2004) U.S. DR load under control. demand response programs. capacity around 9,000 MW, DOE offered a long list of These programs, broadly speak- roughly 1.3 percent of the U.S. recommendations under six ing, require the system operator to peak load. It estimated DR’s major categories:

26 1040-6190/$–see front matter # 2006 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2006.09.001 The Electricity Journal reports by DOE and FERC, blame the problem on lack of enabling technology – which certainly is a major obstacle. Without afford- able smart meters, reliable and inexpensive two-way communi- cation and widespread use of time-variable tariffs, the true potential of DR will never be realized. But there are two other highly critical aspects of enabling tech- Figure 2: How Much Demand Response Is There? Source: Assessment of Demand nology, which remain as serious Response and Advanced Metering, FERC, Aug. 8, 2006 obstacles to successful and cost- effective implementation of DR, Fostering price-based DR, tion from existing programs is namely: Improving incentive-based estimated to be about Fast, reliable, and automated DR, 37,500 MW,’’ the lion’s share communications among multiple Strengthening DR analysis among the investor-owned utili- players in the DR domain in real- and valuation, ties, followed by the ISOs time, and Integrating DR into resource (Figure 2). FERC’s main conclu- Standardized protocols for planning, sions are that, ‘‘The potential customer enrollment and notifi- Adopting enabling technolo- immediate reduction in peak cation, business processes and gies, and electric demand that could be settlement. Further enhancing federal achieved from existing DR Unless these two issues are action. resources is between 3 and 7 successfully addressed, wide- n August, FERC also complied percent of peak electric demand scale implementation of DR shall I with the mandate of EPAct by in most regions,’’ but points out remain limited and problematic, releasing an assessment of the that the low penetration of especially if there is interest to state of technology and demand enabling technologies limits what reach a significant number of response.10 The FERC report can be achieved in the immediate small consumers. provides a comprehensive survey future. It is a classic chicken-and- Take the former. Currently, of the penetration of enabling egg problem. Without wide- system operators have limited technologies and the prevalence – spread penetration of smart capabilities to engage in large- or rather lack thereof – of time- meters and time-variable pricing scale DR involving multiple variable tariffs. Noting that less there is little future for DR. players and large numbers of than 6 percent of electric meters participating customers for a are currently able to record or number of reasons. Most impor- report interval usage, FERC, IV. What is the Holdup? tantly, the protocols for sending using rather diplomatic language, the signal that capacity is tight said, ‘‘the use of DR (in the U.S. Given the significant size of the and voluntary load shedding is power industry) is not wide- DR resource and its cost-effec- needed is time-consuming, error- spread.’’ tiveness, why aren’t we seeing prone, and mostly manual. Since In terms of DR potential, FERC more DR deployment when system operators must balance concluded, ‘‘Nationally, the total emergencies do occur? Most load in real time, any delays to get potential DR resource contribu- studies, including the two major a signal out, get confirmation, and

October 2006, Vol. 19, Issue 8 1040-6190/$–see front matter # 2006 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2006.09.001 27 tangible results must be fast, assess how many customers may and how much they are owed as a error-free, and automatic – char- participate within a short time result of their contribution is acteristics that are presently frame and given a particular currently a back-office nightmare lacking. This means that in many incentive offered. The ISO, faced (Figure 4). In many states, cases, the operator may resort to with uncertainties of how many including California, there are involuntary load shedding simply megawatts of load can actually be multiple existing programs because of inherent delays and shed in real-time, may resort to offered by different utilities to time-consuming complications in involuntary but certain load different customers with widely implementing DR programs. shedding. varied incentives, terms, and con- urrently when an emer- The problem becomes even ditions.11 Record keeping, invoi- C gency occurs, the system more intractable if the system cing, collecting and settlement operator sends an alert to multi- operator is engaged in real-time processes become intractable with ple utilities as well as others bidding, adjusting its incentives thousands or millions of custo- informing them of an impending in response to how many customers and multiple intermediaries. crisis and requesting a response. mers volunteer to shed load. For oth problems are going to This signal typically goes from the such interactive schemes to work, B grow in complexity and ISO to multiple utilities, who, in a higher level of sophistication, scale as more interval meters are turn, pass it on internally and to automation, aggregation, and installed and more customers participating customers using confirmation is needed. participate in time-variable pri- multiple means and channels The second problem may not be cing schemes. California, for (Figure 3). The process of sending as obvious but is equally daunt- example, has decided to convert and aggregating the responses ing. Since many customers and virtually all electrical meters in from multiple parties is notor- intermediaries are likely to parti- the state to the smart variety, able iously cumbersome and time- cipate in DR programs, keeping to handle time-variable pricing.12 consuming, making it difficult to track of who did what and when Other jurisdictions, including the Province of Ontario in Canada,13 are moving in the same direction. In the absence of standardized protocols, the problem of managing multiple signals and com- mands, receiving confirmations, recording the response, and settling accounts to millions of customers will simply become unmanageable. Everyone agrees that the participation of vast numbers of small commercial and residential users is critical if DR programs are to reach their full potential. Yet simple tasks such as attracting and handling the registration of residential users currently are labor-intensive and cumbersome manual processes. Each program Figure 3: Schematic of the ‘‘As-Is’’ State in California offered by each utility to a seg-

28 1040-6190/$–see front matter # 2006 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2006.09.001 The Electricity Journal operator to manage their internal business processes including customer enrollment in DR programs, meter management, load shedding, and post-DR settlement processing. he DRBizNet project has an T ambitious goal to reduce the costs and increase the capabilities of DR business transactions a hundred times – reducing costs by an order of magnitude and increasing speed and functional- ity by similar magnitude. According to Gaymond Yee, the project manger at California Institute for Energy and the Environment (CIEE), ‘‘The recent Figure 4: Schematic of the Alternative State field demonstration of DRBizNet proved the project’s ambitious efficiency and cost-effectiveness ment of the market uses a unique spread use of DR, namely allow- goals, paving the way for great set of forms and protocols for ing efficient real-time collabora- benefits to the people of California customer enrollment. Likewise, tion among multiple – and elsewhere – if the techno- the process of encouraging vast stakeholders, typically the grid logy is widely adopted.’’17 numbers of consumers to engage operator, utilities, and their par- California’s two regulatory and in DR in real-time during an ticipating customers as well as policy bodies, the California emergency is time-consuming other intelligent agents.16 Public Utilities Commission and cumbersome, reducing their he ultimate goal is to allow (CPUC) and the California Energy contribution while increasing the T requests from the grid Commission (CEC), respectively, implementation costs. operator to curtail load to be are strong proponents of energy transmitted flawlessly and conservation, demand-side man- instantaneously to hundreds, agement (DSM), and DR. Over the V. Potential Solutions thousands, or millions of partici- years, CEC has aggressively pating customers and their will- maintained California’s position Many efforts are underway on ingness to shed load immediately as the state with the lowest per how to tackle these key interface registered and aggregated. With capita electricity consumption in and logistical issues.14 Among the such a facility at its disposal, the the U.S. For its part, the CPUC has promising solutions is the grid operator could receive been a strong advocate of invest- Demand Response Business Net- acknowledgment of the amount ment in advanced metering work (DRBizNet) offering a cost- of load reduction available in real infrastructure (AMI) and has set effective approach to implemen- time, enabling it to engage in DR ambitious goals for DR for the tation of DR in real time.15 This rather than rolling blackouts. IOUs in California.18 A number of project, like several others, is Additionally, standardized other states, notably in the focused on addressing one of the business protocols allow a better Northeast, are also considering toughest challenges to wide- way for utilities and the grid similar initiatives to expand the

October 2006, Vol. 19, Issue 8 1040-6190/$–see front matter # 2006 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2006.09.001 29 penetration of DR. Likewise, FERC sion congestion, or unusual or real-time pricing – or DR can has been actively encouraging the spikes in the load similar to those make a big difference. These various independent system experienced during this past demand-side schemes essentially operators (ISOs) and regional summer’s heat wave. Hence, it is avoid the need to fire up that last, transmission operators (RTOs) to no substitute for resource plan- most expensive peaking unit or to expand their DR programs. ning, maintaining adequate engage in involuntary load shed- reserve margins, effective price ding. At the peak of the 2000–2001 hedging on the part of loads, or California electricity crisis VI. Not a Panacea having functional markets for (Figure 5), a fairly small reduction ancillary services and alike. in load could have avoided the Realistically, however, DR is But the experience of the past costly rolling blackouts19 – with not a panacea for addressing all few years in competitive whole- their significant economic and the problems associated with sale markets suggests that intro- political ramifications. managing peak load during a ducing relatively little elasticity in li Vojdani, the project’s lead crisis, whether they are caused by demand through time-variable A investigator, is among shortage of generation, transmis- prices – be it critical peak pricing those who are convinced that the

Introducing relatively little elasticity in demand through time-variable prices can make a big difference.

30 1040-6190/$–see front matter # 2006 Elsevier Inc. All rights reserved., doi:/10.1016/j.tej.2006.09.001 The Electricity Journal 11. Examples include various types of interruptible loads, air-conditioner cycling programs, Flex-Your-Power, critical peak pricing (CPP), time-of-use (TOU) rates, and real-time pricing (RTP) options. Each utility offers a variety of programs targeted at different classes of customers.

12. On July 20, 2006, Paciﬁc Gas & Electric Company (PG&E) received approval from the California Public Utilities Commission to convert its entire system – 5.1 million electric meters – by 2011 at a cost of $1.74 billion. CPUC wants the other two IOUs to follow a similar path within a similar time frame.

Figure 5: Rolling Blackouts during California’s Electricity Crisis in 2001. Source: James L. 13. The Province of Ontario has Sweeney, The California Electricity Crisis, 2004, Hoover Press proposed to convert all electric meters to digital variety by 2010. response (DR) is distinguished from industry needs packaged solutions 14. For example, the State of California these programs by virtue of its such as DRBizNet for managing has established the Demand Response sophistication, where participating Research Center (DRRC) at the the demand-side of electricity far consumers are enticed to shed Lawrence Berkeley National discretionary load in response to better than it has been possible up Laboratory (LBNL) singly devoted to 20 incentives offered – typically by the to now. And that is important DR issues. system operator. given the increasing demand for 15. There are, of course, a myriad of electricity, increasing fuel prices, 3. There is a large body of evidence confirming the cost of maintaining an other promising solutions, including and environmental concerns such inventory of peaking units, which are one developed at DRRC called as global warming.& only used infrequently. Similarly, a AutoDR. number of studies have documented 16. The project was successfully Endnotes: the costs of service interruptions, which demonstrated during a field tend to be highly disruptive to demonstration at the California 1. Stage 1 Alert is called when customers. Energy Commission (CEC) in operating reserves fall below 7 4. Assessment of Demand Response and mid August with participation percent, prompting CAISO to issue a Advanced Metering, FERC, Aug. 8, of CAISO and California’s three public alert and ask for voluntary 2006. IOUs. conservation. Stage 2 Alert is called when operating reserves fall below 5 5. MCKINSEY Q., 2002 17. DRBizNet Press Release, Aug. 11, 2006, available at percent, at which point interruptible 6. FERC, supra note 4. loads are curtailed and demand www.DRBizNet.org. response programs are invoked. Stage 7. PJM Press release dated Aug. 17, 18. California IOUs are currently 3 Alert is called when reserves fall 2006. required to develop and maintain below 1.5 percent, instituting 8. ‘‘These (DR) voluntary DR programs to manage up to 5 involuntary load shedding. curtailments reduced wholesale percent of their peak demand, a 2. Programs to shed load in a targeted energy prices by more than $300 percentage that is likely to be raised in fashion and with prior consent of per megawatt-hour during the the future. customers come under a variety of highest usage hours,’’ according to 19. James Sweeney in ELECTRICITY plans including interruptible loads and Andrew L. Ott, PJM vice president – MARKET REFORM: AN INTERNATIONAL air-conditioner cycling programs. markets. PJM press release dated Aug. PERSPECTIVE, F. Sioshansi and W. Other programs provide incentives for 17, 2006. Pfaffenberger, Eds. (Oxford, England: customers to reduce usage during peak 9. Benefits of DR in Electricity Markets Elsevier, 2006). demand hours by shifting load to off and Recommendations for Achieving peak hours, such a time-of-use (TOU) Them, DOE, Feb. 2006. 20. DRBizNet Press Release, Aug. 11, rates, critical peak pricing (CPP), and 2006, available at www.DRBizNet. real-time pricing (RTP). Demand 10. FERC, supra note 4. org.