November 2024 Selecting and Implementing Demand Response Programs to Support Grid Flexibility A Guidance Note for Practitioners © 2024 International Bank for Reconstruction and Development/The World Bank 1818 H Street NW, Washington, DC 20433, Telephone: 202-473-1000 www.worldbank.org This report was prepared by a World Bank Energy and Extractives Global Practice team led by Michele Chait under the strategic guidance and general direction of Jas Singh and Tatyana Kramskaya. It benefited from the peer review comments from Tamara Babayan, Debabrata Chattopadhyay and Ashok Sarkar. This work is a product of the staff of the World Bank’s Energy and Extractives Global Practice. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The World Bank does not guarantee the accuracy of the data included in this work. The financial support of ESMAP is gratefully acknowledged. ESMAP is a partnership between the World Bank and over 20 partners to help low- and middle-income countries reduce poverty and boost growth through sustainable energy solutions. ESMAP’s analytical and advisory services are fully integrated within the World Bank’s country financing and policy dialogue in the energy sector. Through the World Bank (WB), ESMAP works to accelerate the energy transition required to achieve Sustainable Development Goal 7 (SDG7) to ensure access to affordable, reliable, sustainable, and modern energy for all. It helps to shape WB strategies and programs to achieve the WB Climate Change Action Plan targets. Rights and Permissions. The material in this work is subject to copyright. Because the World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. The recommended citation is as follows: World Bank (2024). “Selecting and Implementing Demand Response Programs to Support Grid Flexibility: A Guidance Note for Practitioners.” Washington, DC: World Bank. Any queries on rights and licenses, including subsidiary rights, should be addressed to: World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA. Fax: 202-522-2625; email: pubrights@worldbank.org SUMMARY This Guidance Note has been developed to serve as a primer for utilities and regulators in developing countries on how to select and implement retail demand response (DR) programs to address electric grid issues. DR can be employed to alleviate a range of challenges, including peak load growth, distribution system constraints, and integration of variable renewable energy (VRE) and electric vehicle (EV) charging. An example of peak load growth comes from Bangladesh, where peak loads more than quadrupled over 2008–18; expensive peaking generation units had to be acquired. But a World Bank study found that the adoption of light-emitting diode (LED) bulbs in combination with a direct load control (DLC) program could reduce peak demand at about one-fifth of the cost of installing new peaking generation units (Islam and others 2019). In Jordan, the utility’s inability to integrate increasing shares of VRE in the transmission and distribution (T&D) network is cited as a key hurdle to further growth of the country’s renewable energy program, and improved load management is recommended to match demand and supply in a manner that reduces overall system costs and incremental integration infrastructure investments (IRENA 2021). This Note focuses on retail DR programs because they are applicable to all types of electric utilities, regardless of whether a wholesale market is present. The implementation guidance accounts for the capabilities and constraints of utilities in developing countries, and outlines guidelines for selecting, designing, and implementing successful DR programs, informed by case studies from emerging markets. While DR cannot alleviate all grid constraints, it is important that regulators and utilities consider including DR as part of their toolbox in future infrastructure planning and asset management.1 1. This Note has been prepared alongside the Energy Sector Management Assistance Program’s publication “Harnessing the Potential of Flexible Demand Response in Emerging Markets: Lessons Learned and International Best Practices” (World Bank/ESMAP 2024), which provides a more detailed analysis of DR programs. Selecting and Implementing Demand Response Programs 1 ACRONYMS ADR AUTOMATED DEMAND RESPONSE CP COINCIDENT PEAK CPP CRITICAL PEAK PRICING DERMS DISTRIBUTED ENERGY RESOURCE MANAGEMENT SYSTEM DLC DIRECT LOAD CONTROL DR DEMAND RESPONSE DRMS DEMAND RESPONSE MANAGEMENT SYSTEM DSB DEMAND-SIDE BIDDING DSM DEMAND-SIDE MANAGEMENT ESS ENERGY STORAGE SYSTEM EV ELECTRIC VEHICLE GMP GREEN MOUNTAIN POWER GW GIGAWATT INT INTERRUPTIBLE CONTRACT IOU INVESTOR-OWNED UTILITY KW KILOWATT KWH KILOWATT-HOUR MW MEGAWATT MWH MEGAWATT-HOUR ONS OPERADOR NACIONAL DO SISTEMA ELÉTRICO (BRAZIL) PCT PARTICIPANT COST TEST PTR PEAK-TIME REBATE RIM RATEPAYER IMPACT MEASURE RTP REAL-TIME PRICING SCE SOUTHERN CALIFORNIA EDISON T&D TRANSMISSION AND DISTRIBUTION TOU TIME-OF-USE TPDDL TATA POWER DELHI DISTRIBUTION LIMITED TRC TOTAL RESOURCE COST UCT UTILITY COST TEST VRE VARIABLE RENEWABLE ENERGY 2 Selecting and Implementing Demand Response Programs TABLE OF CONTENTS Global Context------------------------------------------------------------------------------------------------------------------------ 4 Defining DR and DSM--------------------------------------------------------------------------------------------------------------- 4 DR Potential and Benefits-------------------------------------------------------------------------------------------------------- 5 Types of DR Measures-------------------------------------------------------------------------------------------------------------- 6 Matching DR Programs to a Utility’s Grid Needs------------------------------------------------------------------------ 9 Assessment of DR Program Technology Requirements-------------------------------------------------------------- 11 Additional Steps in Developing and Implementing DR Programs------------------------------------------------ 13 Solutions for Mitigating DR Adoption Barriers in Developing Countries-------------------------------------- 13 DR Program Checklist-------------------------------------------------------------------------------------------------------------- 13 Conclusions----------------------------------------------------------------------------------------------------------------------------- 19 Additional Resources--------------------------------------------------------------------------------------------------------------- 20 References------------------------------------------------------------------------------------------------------------------------------ 20 Glossary--------------------------------------------------------------------------------------------------------------------------------- 22 Appendix A. DR Program Benefits and Cost-Effectiveness Metrics------------------------------------------------------- 23 Appendix B. DR Program Case Study Examples for Developing and Developed Countries ---------------------- 26 Appendix C. Findings and Lessons Learned------------------------------------------------------------------------------------ 29 Selecting and Implementing Demand Response Programs 3 GLOBAL CONTEXT Worldwide, the electricity sector is undergoing significant change. Utilities are struggling to meet the growing electricity demand that often accompanies economic growth. These difficulties are often more pronounced in the developing world. While decarbonization of electricity supply will translate into greater shares of renewable energy in generation portfolios, electrification of energy demand and the ever-growing and complex power needs of industry, households, and transport will also have to be managed appropriately. With electricity supply and demand becoming more dynamic, volatility is likely to increase. Since actions on the supply side alone are typically expensive and inadequate for the optimal functioning of the electricity system, DR is emerging as a critical load management tool for both utilities and consumers. For consumers, DR measures can leverage the many developments on the customer side of the meter (e.g., rooftop solar photovoltaics and battery storage, as well as smart appliances, internet-connected control systems and EVs). Yet, worldwide, DR is underutilized, particularly in developing countries, where electricity growth is expected to more than double by 2050 (IEA 2022) 2 . Developing countries also face grid reliability issues and often experience load shedding events, which may require costly capacity investments3 . Utilities in these countries may also incur a high cost of capital, and experience capital constraints and revenue shortfalls due to below-cost electricity rates and/or nonpayment of electricity bills. And, amid the continued growth of VRE, electrification, and distributed resources, utilities will require more load management services and the ability to compensate customers for providing these services 4 . DR programs can address these issues, enabling utilities in developing countries to better optimize their systems. This will in turn defer or avoid costly grid upgrades and expensive peaking plants, while increasing reliability and helping to alleviate utility budget shortfalls. DEFINING DR AND DSM DR refers to a temporary change (either an increase or decrease) in a customer’s electricity consumption to help optimize the electricity system (generation, transmission, and distribution). DR provides an opportunity for consumers, both large and small, to play a role in an electricity grid’s operation by volunteering to modify their consumption temporarily in exchange for compensation. DR has been successfully used by the United States and many other countries to help utilities manage demand by reducing peak loads, shifting loads to off-peak periods, managing critical grid events, and so forth. DR is one component of a broader suite of measures that alter the quantity and/or timing of a customer’s electricity consumption. These measures include energy efficiency, behind-the-meter technologies (e.g., solar generation, battery storage), and customer response to time-differentiated price signals. Collectively, these activities and technologies are referred to as demand-side management (DSM). In the past, the primary objective of most DSM programs was to provide cost-effective energy and capacity resources to help defer the need to invest in new generation, transmission, and distribution capacity. 2. International Energy Agency’s (IEA’s) Stated Energy Policy Scenario (STEPS). 3. Distribution constraints arise for many reasons, including load or distributed generation that exceeds distribution line capacity and transformer overload. Distributed VRE can lead to power factor issues on distribution circuits, congestion, and voltage issues. EV charging increases loads on the system in the geographic areas where charging occurs. These loads can lead to congestion in the distribution system, transformer overload, and power quality issues on the distribution grid, such as voltage drop or phase imbalance. 4. VRE generation is not dispatchable. Grids with high solar penetration experience strain in the morning as generation increases and again in the evening as generation ceases. During these periods, generation or load may need to be curtailed. For example, a study on Karnataka, India, found that the system value DR increases with increasing VRE penetration, and that agricultural DR is more flexible and can operate for more hours per day than residential, commercial, and industrial DR (NREL 2015). While utility investments to increase generation and distribution system capacity, replace transformers, install voltage regulators or capacitors, or add battery storage can address the grid issues described above, these investments can be costly. Conversely, DR is a “non-wires” solution that can be harnessed to relieve these grid issues at a much lower cost. 4 Selecting and Implementing Demand Response Programs Today, DSM programs have expanded to help address VRE, EV charging, and other developments. All DSM measures, including DR, provide valuable services to help support grid operations and reduce system costs, which, in turn, benefits all consumers. While both DR and DSM refer to altering the quantity and/or timing of a customer’s electricity consumption, the time frames of service provision can differ. While DR services are provided during a limited number of hours per year, certain DSM measures, such as energy efficiency and behind-the-meter solar generation, may operate daily over many years. DR POTENTIAL AND BENEFITS In the United States, where DR programs are mature, retail and wholesale market DR programs achieved over 62 gigawatts (GW) of peak load reduction in 2022 (FERC 2023). In each market region of the North American Electric Reliability Corporation, these savings equated to 0.7 percent to 10.2 percent of peak demand reduction and averaged 6.5 percent (FERC 2023) 5 . Globally, DR programs are estimated to generate potential capacity savings several times greater than the levels achieved in the United States. For example, the National Renewable Energy Laboratory (NREL) estimated global DR peak load reduction potential in 2023 at about 200 GW (NREL 2018). The International Energy Agency’s Net Zero Emissions by 2050 Scenario assumes 500 GW of DR will be deployed globally by 2030 (IEA 2024). Unfortunately, most of this potential remains as yet unrealized. Figure 1. Potential Benefits of DR DR = demand response; EV = electric vehicle; VRE = variable renewable energy. 5. Market regions are Midwest Independent System Operator (MISO), California Independent System Operator (CAISO), PJM Interconnection (PJM), the Electric Reliability Council of Texas (ERCOT), New York Independent System Operator (NYISO), Southwest Power Pool (SPP), and ISO New England (ISO-NE). Selecting and Implementing Demand Response Programs 5 The marginal costs that can be reduced or avoided through DR program load reductions include generation, transmission, and distribution capacity investments, financing costs, variable operation and maintenance costs, fuel and other variable generation costs, T&D losses, and the costs of ancillary services—which can collectively be quite high. DR provides these benefits cost-effectively because the total cost of implementing a DR program should be less than the cost of serving loads during the targeted hours—otherwise, the program would not be implemented. Figure 1 summarizes the potential benefits of DR. (Appendix A includes a detailed summary of the benefits and economics of DR programs, including typical cost-effectiveness tests.) TYPES OF DR MEASURES DR measures can typically be categorized as price based or quantity based6 . Price-based measures incentivize customers to modify their loads in response to time-differentiated pricing (e.g., reducing load during higher-price on-peak hours or increasing load during lower-price off-peak hours). Participants can achieve bill reductions by modifying their electricity consumption consistent with price signals. Quantity-based measures compensate participants for achieved load reductions, typically on a dollar per kilowatt ($/kW) and/or $/kilowatt-hour (kWh) basis. In both cases, it is the customer who needs to determine how best to achieve load shift or reduction, whether through self- generation, energy storage, temporary usage reduction, or other means. Price- and quantity-based DR programs can be implemented in wholesale electricity markets7 and via retail DR programs. While not all utilities operate within a wholesale market 8 , retail DR measures are applicable to all types of electric utilities. Table 1 summarizes eight common retail compensation measures, including their type, the corresponding compensation, load reduction, eligibility for participation and enrollment, and the certainty of load reduction. The measures are listed in a roughly descending order of the magnitude of peak demand reduction per participant. A utility’s retail and/or wholesale activities are overseen by an entity such as a regulator, ministry, or municipal government. This entity reviews and approves the utility’s tariff, including the costs of DR programs, the participant compensation, and authorized cost collection. The utility, or a third party engaged by the utility or regulator, would be responsible for monitoring programs’ impact, program evaluations, and other reporting. All the DR measures listed would require some form of regulatory approval to implement. DR programs can target nearly all types of utility customers. The customer classes targeted depend on the regulator’s or utility’s objective. While the implementation of certain DR measures such as time- differentiated rate structures may be mandatory, customer response to DR events remains optional. 6. Price-based measures may also be referred to as “implicit” and quantity-based measures may also be referred to as “explicit.” 7. Participant compensation for wholesale market DR services is provided through relevant wholesale market measures such as emergency response, automated demand response (ADR), and ancillary services such as frequency regulation or ramping services to integrate VRE. 8. In countries with more sophisticated wholesale energy and/or capacity markets, for-profit third parties known as aggregators can bundle the load reductions of multiple customers as a single resource. In this case, the aggregator enables these retail customers to reduce or increase their electricity consumption during times of grid need in exchange for retail and/or market-integrated, supply-side DR compensation. Wholesale markets enable the aggregator to (1) provide wholesale market services, (2) customize offerings under load management programs, (3) compensate participants adequately through utility and market revenues, and (4) obtain adequate revenues for its services. Any retail compensation paid to customers and/or to third parties, as well as the utility-borne program administration and equipment costs (net of any funding via grants or government funding), should be collected in retail electric rates. 6 Selecting and Implementing Demand Response Programs Table 1. Summary of Retail DR Measures Retail DR Compensationa Determination of Load Participation Load Reduction Probability Measure Change Quantity Interruptible Customers opt in to receive a The customer-specific firm The program is applicable only to The probability of load contract (INT) monthly dollars per kilowatt ($/kW) service level is determined by very large customers. reduction is highest because discount on their bills throughout the customer and is specified in Participation is optional. The of the strong nonperformance the year—the discount calculated the contract between the level of utility enrollment effort penalty, and few annual events based on the difference between customer and the utility. is moderate: large individual load are typically called. the actual customer demand and Demand must be reduced to the reductions are achieved but the customer-specific firm service– firm service level during events. individual contracts are required. level kW. Customers pay significant penalty on usage above the customer-specific firm service level during events. Typically, an accompanying utility tariff defines events, compensation, and penalties, but the programs may be implemented with contracts only. Quantity Automated Customers opt in to integrate ADR The program has a minimum The program is applicable to The load reduction probability demand capabilities into their building load reduction requirement. nonresidential customers. For is strong because the customer response energy management or heating, Load reduction per ADR signal is utilities, customer enrollment load is directly controlled (ADR) ventilation, and air-conditioning considered to determine load can be difficult because customer during an ADR event. However, systems. Participants receive $/kW change. outreach and changes in participants may override the bill credits for the achieved load customer systems are necessary. ADR signal. reductions. Quantity Direct load There are multiple compensation Load reduction per DLC signal is Participation is optional for The load reduction probability control (DLC) structures, including fixed monthly considered to determine load residential and nonresidential is lower if customers are able bill credits, bill credits based on change, unless a customer has customers. For utilities, to override the DLC signal. DLC customer-selected level of air the ability to override the signal. enrollment can be difficult events typically occur during conditioner cycling, the ability of because outreach to many less than 1 percent of annual customers to override the DLC customers is necessary. hours. signal, and avoided time- differentiated energy and demand charges. Quantity Peak-time Customers opt in to receive both Customers establish their load Nonresidential customers opt in The load reduction probability rebate (PTR) dollars per kilowatt-hour ($/kWh) reduction targets. The achieved to participate. The program has a is moderate due to the level of and $/kW credits for the achieved load reduction is relative to the minimum load reduction compensation for load reductions during an event. highest loads in a recent requirement and a penalty performance, the penalty for historical baseline. Penalty may structure. For utilities, nonperformance, and be assessed if the minimum enrollment is difficult because customer-specific targets. share of target reduction is not outreach to many customers is Events occur during less than 1 achieved. necessary. percent of annual hours. 7 Selecting and Implementing Demand Response Programs Table 1. Summary of Retail DR Measures (continued) Retail DR Compensationa Determination of Load Participation Load Reduction Probability Measure Change Quantity Demand-side Customers typically bid for kW load The achieved energy reduction is The program is applicable only to The probability of load bidding (DSB) reductions at most monthly. If a bid relative to the pre-event very large customers. reduction is highest because is selected, participants receive $/ baseline. Participation is optional. The level of the strong nonperformance kWh compensation for the achieved of utility enrollment effort is penalty, and few annual energy reductions during an event. moderate: large individual load events are typically called. Tariffs may provide monthly bill reductions are achieved but reductions based on the accepted individual contracts are required. bids. Bid prices are not known until after the bids are received, and the utility may reject uncompetitive bids. Price Critical peak Revenue neutral for utilities. Rates Per customer response. For the Can be applied to all customer The level of load change is pricing (CPP) are materially higher during CPP class average customer, there is classes. Tariffs may provide for driven by CPP rates during rates event hours (typically no more than no penalty for failing to reduce optional or mandatory event periods. Low reliability 20 events per year, lasting no longer usage during CPP events other enrollment; CPP event of load reduction due to no than four hours on any given day),b than exposure to higher CPP participation is optional for penalty for nonperformance. and lower off-peak rates to rates. customers. Utility enrollment maintain revenue neutrality. effort depends on whether tariffs are optional or mandatory. Price Time-of-use Revenue neutral for utilities. Rates Per customer response. For the Participation is optional for The load reduction probability (TOU) rates differ by time period (i.e., on-peak class average customer, there is residential and nonresidential is lower if customers are able and off-peak) and perhaps by no penalty for failing to reduce customers. For utilities, to override the DLC signal. season. The number of on-peak on-peak usage. enrollment can be difficult DLC events typically occur hours greatly exceeds CPP event because outreach to many during less than 1 percent of hours. Therefore, the on-peak TOU customers is necessary. annual hours. rate is much lower than the CPP rate.c Price Real-time Revenue neutral for utilities. Prices Per customer response. Can Not desirable for less Level of load change driven by pricing (RTP) vary hourly, signaling bulk grid induce load flexibility during sophisticated customers (i.e., RTP price level during hours rates prices. hours when prices are very high residential, small nonresidential when prices are very high or or very low. customers). Optional or very low. mandatory participation tariffs. Customer load shift in response to the RTP price signal is optional. a. For price-based measures, the assumed rates are revenue neutral, meaning for each customer class, the rates are designed to collect the revenue required for that class. b. A CPP rate may have 15 events lasting 4 hours each, for a total of 60 hours annually (less than 1 percent of annual hours). CPP rates are therefore designed to signal traditional demand response services: very high prices during a limited number of hours when grid strain is the highest. c. TOU rates fix on-peak and off-peak periods over a season or year and provide price signals to reduce as well as increase usage. For example, a TOU rate may have a 4-hour on-peak period daily from June through September, totaling 488 on-peak hours. This price structure supports the integration of VRE and EVs, but TOU is less suitable for demand response services due to the potential for customer fatigue, and the much lower on-peak price signal versus a CPP structure. 8 Selecting and Implementing Demand Response Programs Customer classes should be targeted based on the relative load of the individual class during the hours when load reduction is needed, the expected per-customer peak load reductions, and the ease of program implementation for the utility. In terms of compensation, for price-based measures, participants receive compensation through reduced on-peak billing for load reductions during peak periods and lower costs for usage during off-peak periods. Quantity-based measures typically compensate participants for the load reduction achieved during events. Quantity-based measures, which are applicable to larger, nonresidential customers, require a consumption baseline to calculate the usage and the potential for load reductions during a DR event. This baseline may be determined by the customer based on a set level of service or based on historical average electricity use. Quantity-based DR measures can be achieved through utility load control or, for customer-managed load reductions, feature large penalties for failing to achieve the targeted load reductions. Consequently, these measures are the most likely to reduce load. The certainty of load reduction is high for interruptible contracts (INT), due to the penalty provisions for nonperformance. It is for DLC and automated demand response (ADR), because utilities carry out load curtailments. Demand-side bidding (DSB) and peak-time rebate (PTR) generally offer less certainty since their nonperformance penalties are lower. By contrast, price-based measures offer lower load reduction certainty because they only reward customers for responding during events. Time-of-use (TOU) provides the weakest on-peak price signal because the on-peak to off-peak price ratio is often relatively low due to the relatively high number of on-peak hours. Critical peak pricing (CPP) offers a stronger price signal than TOU because the number of CPP event hours is smaller. However, the penalty for not responding is weak. Real-time pricing (RTP) rates may signal very high prices during a small number of hours; however, customers may find it difficult to respond to granular (e.g., hourly) price signals. The number of participants and the actual load reductions are not known with certainty prior to a program’s implementation. This is because customers’ responses are voluntary, although, depending on the measure, they may face financial penalties for failing to reduce loads. (Appendix B presents several DR program case studies from both developed and developing countries.) MATCHING DR PROGRAMS TO A UTILITY’S GRID NEEDS Implementing a DR program requires multiple steps, including selecting the program, assessing the technology requirements, identifying target customer class(es), acquiring program funding, determining compensation for participating customers, obtaining regulatory approvals, and implementing and evaluating the program. Each step is discussed in the following sections and summarized in the DR Program Checklist (see p. 13). Figure 2. depicts the first step: selecting the DR program best suited for addressing the identified grid needs. Selecting and Implementing Demand Response Programs 9 Figure 2. Grid Services Provided by Each Type of DR Measure, including Certainty of Load Reduction EV = electric vehicle; RE = renewable energy. 》 The area in green shows all the DR programs that can achieve peak load reductions. This means that a utility seeking to achieve peak load reductions could consider implementing any of these DR measures. However, the strength of the load reduction varies by measure as noted previously. Load reduction certainty is expected to be the highest for INT and ADR programs and moderate for DSB and PTR programs. 》 The area in orange indicates that utilities desiring to alleviate distribution constraints should consider implementing ADR or DLC measures. This is because, with appropriate technologies in place, these services can be targeted to the specific geographic regions experiencing distribution constraints, over the appropriate time periods. While it is theoretically possible to implement a DSB program for a geographic area, the administrative costs of doing so would likely be prohibitively high. Similarly, INT would be theoretically possible to implement geographically; however, it could be difficult for industrial loads to address distribution issues occurring outside the time frames of defined events. 》 The area in blue indicates that the integration of VRE and EV charging can be achieved by both price- and quantity-based measures. Regarding price-based measures, RTP and TOU rate structures are recommended because these measures incentivize both the increase and decrease of customer loads. For quantity-based measures, load control through DLC and ADR can both increase and decrease a customer’s loads. Utilities desiring to integrate VRE or EVs should consider implementing RTP, TOU, ADR, and/or DLC programs. 10 Selecting and Implementing Demand Response Programs ASSESSMENT OF DR PROGRAM TECHNOLOGY REQUIREMENTS The next step is to identify any utility - and/or customer-sited technologies that may be required to implement the DR measure(s) identified. While it is possible to implement certain small-scale DR programs with no incremental technology investments, event notification systems and billing software modifications are needed to implement DR programs at scale. For example: 》 If large commercial and/or industrial customers already have hourly or subhourly billing data and receive service on a time-differentiated tariff with demand charges, then a small number of these customers could be billed under an INT program and notified of events manually. 》 Calculation of the PTR and DSB measure baselines requires historical, hourly usage data; such data will be available if a utility already has the requisite metering and data management systems in place. Billing DSB and PTR programs at scale would require modifications to billing systems to track and compensate customer load reductions. 》 ADR and DLC programs may require utilities or customers to make incremental investments in technologies such as communication systems, meters, smart thermostats, and/or Wi-Fi devices, as well as in the integration of utility and customer systems. 》 A basic residential DLC program may not require complex billing system modifications because participants can be compensated via a fixed monthly bill credit. However, residential customers may not have meters capable of billing time-differentiated rates. Technology issues are summarized in Table 2. DR measures are grouped according to the three typical objectives of DR programs: peak demand reduction, alleviation of distribution constraints, and integration of EVs and/or VRE. Example DR measures are provided for each category. Once the goal of the program has been identified, the presence of metering, billing, event communication, and customer–utility technologies will drive an individual utility’s ability to implement the desired DR measure(s) in relevant customer classes. For each DR measure, implementation efforts will vary depending on the capabilities of the individual utility and the customer class(es) it seeks to target. Regarding event notification, a DR management system (DRMS) or a distributed energy resources management system (DERMS) will be needed if large numbers of customers must be notified of upcoming events. These are types of utility management information systems and are essential for issuing mass event notifications or instructing load changes. These systems not only include geographic information systems to enable load instruction in the targeted geographic areas, they also allow visibility into the locations and load reduction capabilities of large numbers of participants, allow the estimation of load reductions, enable load aggregation, support program management, and help evaluate programs’ performance. Enabling peak load reduction from a large number of customers requires the ability to: (1) enroll customers, perform customer service activities, and conduct marketing; (2) notify customers that events will occur; (3) remotely control/dispatch load reductions at a given time (and possibly subregion) for certain residential and nonresidential customer programs; (4) evaluate individual responses; (5) compensate participants; and (6) potentially develop new DR products. A utility may choose to manage all these activities internally or outsource certain program management activities to a third party. The third-party entity can be contracted by the utility to enroll customers, notify customers of events, and determine compensation to participants. The utility sends event notifications to the third party and compensates the third party for the achieved load reductions. Selecting and Implementing Demand Response Programs 11 In countries where utilities do not have the ability to issue mass event notifications or update their billing software, logistically only a relatively small number of customers would be able to participate in a DR program. In such cases, the utilities should consider implementing INT programs to achieve the strongest peak load reductions. INT programs, which have no incremental customer- or utility-side technologies, can be implemented for (usually large) customers that already have metering capable of measuring hourly or subhourly loads. An ADR program would entail additional metering, notification, billing, and customer systems integration requirements. For utilities seeking to ADR and DLC programs will require appropriate metering, data and billing, and notification capabilities. While a utility customer class that does not receive service in time-differentiated rate structures (e.g., residential customers) could participate in a central air conditioner DLC program, for example. The utility would require an adjustment to its billing system to provide participants with bill credits, besides requiring a load control device (i.e., a programmable thermostat, an EV charger with appropriate communications capabilities) and a means of communicating with the device (i.e., Wi-Fi). VRE and EV integration services can be achieved via RTP and TOU rate structures, as well as via ADR and DLC, since all these measures can incentivize both increases and decreases of customer loads. Appropriate metering and billing software can make it possible to implement time-differentiated rate structures such as RTP and TOU in any customer class. An individual utility may need time to modify its billing systems and meters to carry out time-differentiated rate structure programs if these capabilities are not already present in the targeted customer classes. As noted above, depending on a utility’s existing capabilities, additional technologies and system updates may be required to implement ADR and DLC programs. Table 2. Impact of Technology Issues on DR Program Selection Source: Original compilation. ADR = automated demand response; C&I = commercial and industrial; DERMS = distributed energy resources management system; DLC = direct load control; DRMS = demand response management system; EV = electric vehicle; INT = interruptible contract; RTP = real-time pricing; TOU = time of use. 12 Selecting and Implementing Demand Response Programs ADDITIONAL STEPS IN DEVELOPING AND IMPLEMENTING DR PROGRAMS Determining customer classes. Issues specific to individual customer classes will determine whether a class is a good candidate for providing the desired DR services. For example, if peak load reductions are desired, then the utility should analyze the contribution of each customer class to the utility’s coincident peak demand. If the utility desires to initially implement a small-scale program, then it should seek participants with the ability to supply large potential usage reductions during events. Other issues include the relative sophistication of customers regarding electricity usage and pricing, knowledge of the benefits of DR, and potential customer fears surrounding participation. Compensation. Participant compensation is a function of a utility’s avoided costs during the periods that DR services will be provided. Avoided costs will therefore need to be calculated. Compensation should be high enough to encourage participation, but should not exceed utility avoided costs net of program implementation costs. Regulatory approvals. Regular utility engagement with regulators can support a smooth and robust program regulatory process. Regulatory dialogue will help ensure, among others, timely approvals, clarity on data requirements, and agreements on program funding. Regulators may need to approve changes to rate structures to ensure upfront program costs can be financed and repaid and that program cost savings are appropriately reflected. SOLUTIONS FOR MITIGATING DR ADOPTION BARRIERS IN DEVELOPING COUNTRIES As noted earlier, despite its benefits, DR remains a relatively untapped resource worldwide, and very few countries (outside the United States, Australia, Canada, and Japan, for example) have widely implemented DR measures. Beyond selecting and designing suitable programs, it is necessary to identify other obstacles that may arise. While a range of barriers have hindered the uptake of DR programs, the global experience points to a suite of policies, programs, and regulatory approaches that may be leveraged to overcome them. Table 3 provides a prioritized list of both barriers and ways to address them. DR PROGRAM CHECKLIST While the design and implementation of a DR program have many common features, the specifics are necessarily unique to each utility and proposed DR program. It is therefore not possible to draw a universal roadmap. However, the steps required to select, design, and implement a DR program are common, and the following checklist has been developed to guide utilities and regulators on the common steps needed. Selecting and Implementing Demand Response Programs 13 Table 3. Policy and Regulatory Actions to Address Common DR Implementation Barriers Policy, Regulatory, Barriers Responsible and Utility Actions Addressed Party Financial Start with demand response (DR) Inability to provide appropriate Regulator, measures that do not require time- time-differentiated DR program utility, differentiated price signals, and/or a large price signals government number of participants 》 Subsidized rates limit the ability 》 Fixed bill credits for direct load control to provide meaningful time- (DLC) for central air conditioners differentiated price signals 》 Interruptible contract (INT) measure 》 Demand meters and/or advanced for industrial customers metering infrastructure (AMI) not deployed to all customers Implement DR program(s) for a few large Utility lacks the requisite Regulator, customers technologies such as billing utility Explore funding sources for the upfront software, AMI, a demand response costs of DR programs management system/distributed energy resources management 》 Grants from external sources system 》 Nonutility government sources (e.g., environmental or transportation budgets) 》 Dedicated DR program rate surcharge (rate rider) 》 Vendor financing programs 》 Transition toward cost-based rates Ensure rates send economically efficient Time-differentiated rates may Regulator, price signals that reflect marginal costs further erode a utility’s financial utility 》 On-peak rates not priced higher than situation by lowering revenues the marginal cost 》 Off-peak rates not priced lower than the marginal cost Ensure rates appropriately reflect costs Upfront program costs expected to Regulator 》 Utility capacity planning should result in a net cost or rate increases incorporate projected DR load reductions 》 In rate making, upfront costs should be amortized over economic life 》 Update rates to reflect the savings generated by a program due to avoided costs 14 Selecting and Implementing Demand Response Programs Design regulatory incentives to mitigate Investor-owned utilities may Regulator lost earnings opportunities caused by experience reduced future earnings the reduction of the asset base. Example opportunities if DR programs reduce incentive measures: capacity investments 》 Shared savings. Utility retains a share of DR program savings (balance of savings is applied to reduce rates) 》 Cost capitalization. Utility earns a return on capitalized DR program costs 》 Bonus equity return. Utility earns a bonus rate of return on DR program investments Data and Information Seek knowledge transfer, training from Lack of DR program procedures and Regulator, external sources tools for utilities; utility staff lacks utility, 》 Internal training programs for utilities the skills or bandwidth to implement government 》 Engage DR consultants to adapt DR programs program approaches and tools from 》 Technical regulations; metering other utilities code; DR road map; DR 》 Access to information and approaches regulations; control room from multilateral development banks/ operational protocols; customized donors tools (i.e., load research, load 》 Interaction with international peer reduction estimation); DR utilities for sharing procedures, tools, program tariffs; integration and methodologies with the wholesale market 》 Government-sponsored training (if applicable); program cost- efforts and credentialling programs benefit analysis; measurement 》 Local universities and verification, software, and installation Implement customer education Difficulty enrolling participants; Utility, campaigns customers reluctant to participate aggregator 》 Educate stakeholders on DR’s benefits 》 Concerns over stifling business and promote participation; develop operations and growth; potential tools and guides (e.g., online bill participants uncertain about DR savings calculators); conduct customer price-based programs’ impacts surveys to understand the level of on electricity bills; perception that compensation necessary to encourage compensation will be inadequate; participation lack of understanding of DR’s benefits; customer fatigue due to frequent events Selecting and Implementing Demand Response Programs 15 Program Goal 》 Identify key grid issues (e.g., peak load reduction, distribution deferral, VRE integration) and the quantity of demand reduction and/or load shift necessary to alleviate constraint(s): ■ For example, if peak demand reductions are desired, analyze the top (~100) highest net peak loads9 (megawatts, MW) in the annual load duration curve. The data can then be used to identify an appropriate number and duration of DR events, the definition of the peak period, and the potential peak MW reduction that could be obtained from a DR program. A steep load duration curve generally suggests targeting a small number of coincident peak demand reductions. A shallower curve supports a rate design such as TOU, which can target more hours without creating participant fatigue. ■ Targeting DR in less than 1 percent of annual hours (~80 hours) will help reduce customer response fatigue in the context of a peak load management program. Target Customer Class(es) 》 Using hourly class load shapes, determine the degree to which each customer class contributes to the identified peaks. Analyze the expected load reduction. 》 Using Figure 2, identify candidate DR program measures capable of addressing program goals. Remember that TOU or CPP rates can be easily extended across several customer classes, although the load reduction per customer will be small relative to industrial customer DLC or INT measures. 》 Select the customer classes to be targeted, which are influenced by the metering and billing capabilities available in each customer class, utility- and customer-sited technologies, implementation timelines, and the desired load change. These will vary by utility and the requirements will be different for each DR measure. Perform a gap analysis to identify administrative and technical barriers to implementing candidate DR measure(s) in the target customer classes; determine any utility billing system, metering, and/or other technologies required to implement the program: ■ Whether the necessary metering is in place to enable billing time-differentiated tariffs and/or the calculation of baseline usage. Calculation of the baselines for PTR and DSB measures requires historical, hourly usage data. Such data will be available if a utility has the requisite metering and data systems already in place. Further, even if historical data are available to calculate customer baselines, the formulas utilized may not accurately estimate the counterfactual load (or adjusted baselines) during events that is needed to determine the actual load reductions (ISGW 2015). Customers with prepaid electric meters are not precluded from participating in DR programs. ■ What billing system updates may be necessary to bill and compensate program participants. A small number of participants can be billed manually. Billing system updates will be required for large numbers of participants. 9. The net peak load is peak demand net of any front-of-meter and behind-the-meter renewables-based generation. 16 Selecting and Implementing Demand Response Programs ■ What event communication systems (DRMS, DERMS) may be necessary. While a small number of customers can be notified of events manually, in the absence of automated load control, many participants will require an event communications system (i.e., mass emails, text messaging). ■ What additional technologies may need to be acquired. A supervisory control and data acquisition (SCADA) system enables a utility to manage its distribution system, including control and regulation equipment to reliably deliver power to customers. Implementing DR programs at scale may require advanced metering infrastructure (AMI), DRMS or DERMS, and management and/or geographic information systems. Further, ADR and DLC programs may require incremental investments in utility- and/or customer-side technologies. For example, a utility or aggregator may control residential air conditioner loads via pagers, a Zigbee radio frequency signal from the utility meter, or a customer’s Wi-Fi connection via a smart thermostat or a Wi-Fi–enabled air conditioner. Such programs may not require complex billing system modifications—customers can be compensated for participation via a fixed monthly bill credit. 》 Given the utility’s unique technical and administrative context, determine the candidate DR program best suited to deliver the desired results and target customer classes: ■ A “no regrets” initial program should be relatively easy to implement and be likely to deliver large benefits. With success, other programs can be deployed incrementally. For example, an industrial customer INT program would likely deliver the highest load reductions per customer and could initially be implemented on a manual basis. Funding and Approvals 》 Determine whether a program should be implemented as a pilot or rolled out at scale. 》 Begin engaging with relevant entities to obtain funding for any upfront program expenditures. 》 Develop a timeline and budget for acquiring and implementing the necessary technologies and billing system changes. 》 Engage in outreach and knowledge transfer activities with utilities that have implemented similar programs. 》 Conduct outreach to the regulator to determine the regulatory requirements and the process for obtaining approval for the DR program. Determine Participant Compensation 》 Quantify utility avoided costs during the period(s) of grid constraints. Avoided costs may include energy and generation capacity costs, avoided transmission and/or distribution capacity costs, ancillary services costs, losses, and VRE procurement impacts. Avoided costs are also utilized in the calculation of regulatory cost-effectiveness tests (Appendix A). 》 Determine annual participant compensation. This should be high enough to encourage participation, but should not exceed the annual utility avoided costs net of annual program implementation costs. Consider customer surveys to determine adequate compensation levels. Selecting and Implementing Demand Response Programs 17 Regulatory 》 Determine and develop the data necessary for any regulatory approval, tariffs, and quantifications (e.g., cost-effectiveness tests). 》 Verify that DR activities, national energy/climate goals, and/or other policy goals are properly aligned. 》 Engage with regulators to mitigate potential disincentives for a utility to pursue DR. For example, the New York utility, Consolidated Edison, was allowed to earn performance incentives, shared savings incentives, and a rate of return on the costs borne as part of the Brooklyn-Queens Demand Management program (Utility Dive 2019). Regular utility engagement with regulators can foster a robust program regulatory process. 》 Evaluate program cost-effectiveness. Evaluations may be conducted prior to program implementation, as well as after implementation to consider results and identify areas requiring changes. 》 Determine whether the utility or a third party will manage the DR program. If a third-party program manager or aggregator is envisioned for at-scale program management, work with the regulator and the wholesale market to ensure necessary regulations and revenue opportunities are in place. 》 Develop program tariffs for regulatory approval. 》 Support the regulator in its evaluation of programs and tariff materials. Develop data necessary to calculate DR program costs. Implementation Planning 》 Develop a program implementation plan. 》 Conduct customer education and outreach regarding the importance of mitigating targeted grid issues, the impacts of electricity usage, and the benefits of DR. Alleviate any potential customer fears surrounding participation. Outreach to customers can yield insights that inform appropriate compensation levels. It can also foster high levels of participation as the program is implemented. Educated customers may also participate in other utility offerings and will be better positioned to manage their electricity use and bills. 》 Consider developing an online bill calculator to demonstrate bill savings opportunities to customers. 》 Conduct any outreach with similar utilities and other entities specializing in relevant program aspects to foster knowledge transfer and obtain program development input. Utilities implementing a DR program for the first time should seek data and information from a wide variety of external and internal sources, such as peer utilities, DR experts, multilateral development banks and donors, as well as governments and local universities. 18 Selecting and Implementing Demand Response Programs 》 Issue request(s) for proposals for any third-party support and/or to obtain the necessary hardware and software and/or professional services to administer DR programs. 》 Conduct outreach to promote customer participation and recruit customers. Develop a plan for program monitoring and evaluation, including periodic regulatory review of program cost effectiveness. Implementation 》 Procure the necessary technologies and third-party services. 》 Once regulatory program approval has been obtained, commence the necessary billing system updates so that participant bills using new tariffs can be autogenerated at full program participation. 》 Implement the program(s) and carry out measurement and verification to monitor the degree to which desired load reductions have been achieved. 》 Conduct periodic evaluations of programs to assess their effectiveness (including cost- effectiveness) and incorporate findings and recommendations to improve efficacy. 》 DR programs should operate within the cohesive planning efforts of the utility system. Projected load reductions should be incorporated into the capacity planning of utilities, with cost savings ultimately reflected in electricity rates. Develop a plan for integrating anticipated load reductions into the resource planning process. CONCLUSIONS The electricity sector is undergoing significant changes, and developing countries are not immune. While efforts to decarbonize electricity supply will translate into higher shares of VRE in utility generation portfolios, electrification of energy demand, distributed resource penetration, and the ever-growing and complex power needs of industry, households, and transportation will also have to be managed. Further, new loads, from IT data centers to new artificial intelligence–enabled devices, may be on the horizon. Therefore, DR needs to be considered as a valuable strategy to address these challenges, particularly in developing countries, where growth in electricity consumption is high and DR can be much more cost-effective than supply-side solutions. (Appendix C includes more details on the findings and lessons learned from the programs analyzed for this guidance note.) Key takeaways include: 》 DR has been successfully, and cost-effectively, implemented in developed and developing countries, with measurable benefits. 》 Appropriately designed DR programs are highly cost-effective and can offer substantial benefits to grid operators in developing countries. Reduced utility costs can translate into lower tariffs for all customers. 》 DR programs have also been shown to address emerging issues, driven by VRE, EV charging, and DER, on a cost-effective basis and can benefit both utilities and customers. Selecting and Implementing Demand Response Programs 19 》 Well-designed DR programs can be economic, even if retail rates are subsidized. Cost-reflective rates, while desirable, are not a precondition to and do not preclude the implementation of DR programs. 》 While technology can help implement larger DR programs, smaller programs and pilots can be implemented with little or no additional incremental technologies, making the programs potentially less costly for utilities. 》 Load reduction is more certain when the penalty for noncompliance is strong or when direct load management is carried out. Examples of such DR programs are INT, DLC, and ADR. While the future of the electricity sector may continue to be volatile for years to come, increased coordination among energy suppliers, distributors, and customers using programs such as DR will be an important advancement and can yield mutual benefits, including lower costs for all. ADDITIONAL RESOURCES 》 FERC (Federal Energy Regulatory Commission). 2021. National Action Plan on Demand Response. Washington, DC: FERC. https://www.energy.gov/sites/prod/files/oeprod/DocumentsandMedia/ FERC_NAPDR_-_final.pdf. 》 Brattle (The Brattle Group, Inc.). 2019. The National Potential for Load Flexibility: Value and Market Potential through 2030. https://www.brattle.com/wp-content/uploads/2021/05/16639_ national_potential_for_load_flexibility_-_final.pdf REFERENCES Brattle (The Brattle Group, Inc.). 2018. PJM Cost of New Entry: Combustion Turbines and Combined-Cycle Plants with June 1, 2022 Online Date. https://www.brattle.com/wp-content/ uploads/2021/05/13896_20180420-pjm-2018-cost-of-new-entry-study.pdf. Christensen Associates. 2023. “Load Impact Evaluation: Base Interruptible Program.” Presentation, May 2, 2023. https://webtraining.cpuc.ca.gov/-/media/cpuc-website/divisions/energy-division/docu- ments/demand-response/demand-response-workshops/2023-load-impact-protocol-workshops/ drmec-presentation-bip_final_with_edits.pdf. CPUC (California Public Utilities Commission). 2001. California Standard Practice Manual: Economic Analysis of Demand-side Programs and Projects. California: CPUC. CPUC. 2023. “Decision 23-12-2005 Dated Decem ber 14, 2023, Directing Certain Investor-Owned Utilities’ Demand Response Programs, Pilots, and Budgets for the Years 2024–-2027.” https://docs. cpuc.ca.gov/PublishedDocs/Published/G000/M521/K486/521486520.PDF. Deshmukh, Ranjit, Girish Ghatikar, Rongxin Yin, G. Ganesh Das, and Sujay Kumar Saha. 2015. “Estimation of Potential and Value of Demand Response for Industrial and Commercial Consumers in Delhi.” Paper presented at the India Smart Grid Week Conference (ISGW 2015). https://eta-publi- cations.lbl.gov/sites/default/files/lbnl_6987e.pdf. 20 Selecting and Implementing Demand Response Programs FERC (Federal Energy Regulatory Commission). 2023. 2023 Assessment of Demand Response and Advanced Metering. Washington, DC: FERC. Gheorghiu, Iulia. 2019. “Green Mountain Power Pilots Tesla Batteries as Meters.” Utility Dive, May 2, 2019. https://www.utilitydive.com/news/green-mountain-power-pilots-tesla-batteries-as-me- ters/553873/. ICF (ICF International). 2015. Partnership for Growth, Ghana: TOU Tariff Analysis and Program Development: Final Report. Prepared for the United States Agency for International Development (USAID). https://pdf.usaid.gov/pdf_docs/PA00SX87.pdf. IEA (International Energy Agency). 2022. World Energy Outlook 2022. Paris: IEA. IEA. 2024. “Demand Response.” https://prod.iea.org/energy-system/energy-efficiency-and-demand/ demand-response. IRENA (International Renewable Energy Agency). 2021. Renewables Readiness Assessment: The Hashemite Kingdom of Jordan. Abu Dhabi: IRENA. M. E. Islam, M. M. Z. Khan, C. Nicolas and D. Chattopadhyay, "Planning for Direct Load Control and Energy Efficiency: A Case Study for Bangladesh," 2019 IEEE Power & Energy Society General Meeting (PESGM), Atlanta, GA, USA, 2019, pp. 1-5, doi: 10.1109/PESGM40551.2019.8973505 NREL (National Renewable Energy Laboratory). 2015. Demand Response in Bangalore: Implications for Electricity System Operations. Golden, CO: NREL. NREL. 2018. Potential Roles for Demand Response in High-Growth Electric Systems with Increasing Shares of Renewable Generation. Golden, CO: NREL. NREL. 2024. Determining and Unlocking Untapped Demand-Side Management Potential in South Africa: Demand Response at the Grid Edge. Golden, Co: NREL. ONS (Operador Nacional do Sistema Eletrico). 2022. Avaliação Do Mecanismo De Resposta Voluntária Da Demanda – Portaria MME Nº 22/2021. Rio de Janeiro: ONS. PSCM (Public Service Commission of Maryland). 2023. The EmPOWER Maryland Energy Efficiency Act Report of June 2023 with Data for Compliance Year 2022. Baltimore: PSCM. Utility Dive. 2019. “BQDM Program Demonstrates Benefits of Non-Traditional Utility Investments.” Utility Dive, March 11, 2019. https://www.utilitydive.com/news/ bqdm-program-demonstrates-benefits-of-non-traditional-utility-investments/550110/. Vermont Public Utility Commission. 2019. “PUC Case No. 19-3167-TF.” https://epuc.vermont.gov/. Vermont Public Utilities Commission. 2023. “PUC Case No. 23-1335-TF Final Order Approving Tariff Revisions.” https://epuc.vermont.gov/. Yin, Rongxin, Girish Ghatikar, Ranjit Deshmukh, and Aamir Hussain Khan. 2015. “Findings from an Advanced Demand Response Smart Grid Project to Improve Electricity Reliability in India.” Paper presented at the India Smart Grid Week Conference (ISGW 2015). https://international.lbl.gov/ publications/findings-advanced-demand-response. World Bank. 2023. The Economics of Electric Vehicles for Passenger Transportation. Sustainable Infrastructure Series. Washington, DC: World Bank. World Bank/ESMAP (Energy Sector Management Assistance Program). 2024. Harnessing the Potential of Flexible Demand Response in Emerging Markets: Lessons Learned and International Best Practices. Washington, DC: World Bank. Cost-Effectiveness Tests for DR Programs Selecting and Implementing Demand Response Programs 21 Glossary Term Definition Advanced metering Meters enabling two-way utility–customer communication, sophis- infrastructure ticated time-differentiated rate structures, and voltage and reactive power management. Aggregator An entity that aggregates the load of multiple customers as a single retail or wholesale resource. Avoided costs Utility marginal costs that can be avoided through reductions in customer usage. Avoided costs include capacity investments, financing costs, losses, variable operation and maintenance costs, fuel and other variable generation costs, and the costs of ancillary services. Coincident peak (CP) The maximum annual load (megawatts [MW]) of a utility. Distributed energy Customer-sited resources, including rooftop solar generation, resources (DERs) storage, electric vehicles, demand response (DR), and energy efficiency. Demand response Commonly, a temporary reduction in energy consumption (approx- imately 1 hour to 4 hours) in response to a utility event notification. More recently, DR has expanded to include flexible load services. Demand response manage- Management information systems necessary to issue mass DR event ment system (DRMS); notifications, instruct load changes for a large number of customers, Distributed energy and/or geographically target customer load changes. resources management system (DERMS) Flexible load services Consumers increasing as well as decreasing loads to support the integration of variable renewable energy and electric vehicles. Price-based compensation Compensation based on the response to time-differentiated prices. May also be referred to as “implicit.” Quantity-based Compensation based on the achieved load reductions. May also be compensation referred to as “explicit.” Variable renewable energy Renewable energy from nondispatchable generation sources such as wind and solar photovoltaics, which produce intermittent output. 22 Selecting and Implementing Demand Response Programs Appendix A. DR Program Benefits and Cost-Effectiveness Metrics Table A.1 DR Program Benefits Improved grid reliability Employing demand response (DR) during grid emergencies and peak supply shortages can forestall grid outages and reduce network congestion. Reduced investments in Because DR can reduce the peak demand and/or alleviate distribution generation, transmission, constraints at a lower cost compared with investments in generation, and distribution capacity transmission, and distribution capacity, it can reduce investments resources in new capacity, which is important when capital is expensive or constrained. When system planning incorporates expected demand reductions, lower utility capital investments can be achieved, which can lead to lower electricity tariffs for all customers. Lower electricity bills for Limiting electricity use during DR events can result in a reduction of participating customers electricity bills for participating customers. Lower electricity rates Because the cost of DR programs is less than avoided costs, utilities can reduce their costs. This can allow all customers to benefit from more affordable electricity. Improved integration DR can harness load flexibility to cost-effectively integrate VRE via of variable renewable reduced ramping needs and reduced curtailment. DR can also enable energy (VRE) and electric shifting flexible electric vehicle charging loads to off-peak periods vehicles (valley filling), which can increase generator capacity factors, support VRE integration, and improve generators’ heat rate efficiency. Load flex- ibility can also mitigate VRE congestion on the distribution grid. Decreased costs of fuel Peak load reduction can mitigate the high costs of expensive, on-peak and purchased power fuel and/or imported power. In wholesale electricity markets, this can reduce the market clearing price applicable to all energy purchased in the relevant period. Economic growth DR-induced improvements in the provision of reliable electricity supply can support industrialization and economic growth through reduced outages and fewer involuntary load shedding events. Government borrowings that would otherwise be directed to capacity expansion can be utilized for other economic development priorities. Reduced curtailment of By shifting load to hours of VRE generation, flexible load services can VRE reduce the curtailment of VRE. Peak load management occurs in just a few hours and therefore yields limited climate benefit opportunities. Selecting and Implementing Demand Response Programs 23 Cost-Effectiveness Tests for DR Programs The cost-effectiveness of demand-side programs, including demand response (DR), can be quantified using standard tests that evaluate the programs’ impacts from the perspective of key stakeholders such as utilities, society, participating customers, and all ratepayers. These cost-effectiveness tests are discussed in this appendix. Commonly used tests include the participant cost test (PCT), the ratepayer impact measure (RIM), the utility cost test (UCT), and the total resource cost (TRC) tests (CPUC 2001): 》 The PCT measures a program’s expected impact from the perspective of the participants. In the PCT, bill savings for participants and any incentive received by them are benefits. Costs to participants include the capital and operating costs of a measure and any bill increases or penalties. 》 The RIM test measures the changes in utility revenues and avoided costs due to a program: in other words, it measures the expected cost shift from all ratepayers to participating customers. Avoided costs include any avoided environmental costs, such as costs of greenhouse gas emissions, which are actually borne by utilities (e.g., carbon allowance costs). Benefits include utility avoided costs, while costs comprise a program’s administration cost, the utility cost of a program’s hardware and software, the incentives paid to participants, and the bill savings for participants (or lost utility revenues). 》 The UCT measures the net costs of a program from the utility’s perspective. In the UCT, utility avoided costs are a benefit, while costs comprise a program’s administration costs, including the incentives paid to customers but excluding avoided bills for customers. 》 The TRC test measures the net costs of a program to the utility’s service territory. In this test, costs are based on any measure-related capital and operating costs to participants plus a utility program’s technology and administrative costs. Utility avoided costs are considered benefits. The cost-effectiveness tests should be quantified from each perspective to ensure the programs have positive impacts on each stakeholder. In these cost-effectiveness tests, costs and benefits are quantified and present valued over the measure’s life. The ratio of the present values of benefits to the present values of costs is commonly referred to as the benefit-cost ratio. The result of the cost-effectiveness tests can be expressed in this form. A benefit-cost ratio of at least 1.0 for each of the tests indicates that the program is expected to be cost effective. Of course, if costs exceed benefits for any program, it should not be implemented. A hypothetical example is provided in Table A.2. The cost-effectiveness test results for the hypothetical example in Table A.2 show that the example is economical, as evaluated based on all the tests: in no case is the ratio less than 1.0. The benefit-cost ratio results in this example range from 1.2 for the TRC to 3.4 for the PCT . These results indicate that the program should be implemented since it will benefit utilities, ratepayers, program participants, and 24 Selecting and Implementing Demand Response Programs Table A.2 Example of a DR Program Cost-Effectiveness Test Line Program Metric Present Value ($, millions) A Equipment cost to participants 15 B Equipment cost to utilities 50 C Avoided costs for utilities 80 D Participation incentive to customers 25 E Avoided bills for customers 26 F Equipment cost to customers 15 PCT = participant cost test; RIM = ratepayer impact measure; TRC = total resource cost; UCT = utility cost test. Benefit-Cost Ratio Results PCT = (D+E) ÷ F = 3.4 TRC = A ÷ (B+F) = 1.2 RIM = C ÷ (E+D) = 1.6 UCT = C ÷D = 3.2 society. In addition to the above cost tests, DR program evaluations may also include program-specific metrics such as those summarized below: 》 Number of participating customers 》 Share of participating customers (%): [Number of participating customers] ÷ [Number of eligible participants]. For example, if eligible participants are limited to industrial customers, then the number of eligible participants equals the number of industrial customers. 》 Coincident peak (CP) demand savings: MW AC (alternating current) of customer demand savings occurring at the time of a utility’s peak load (utility CP MW) 》 Avoided CP capacity: [CP demand savings] * [Transmission and distribution (T&D) loss factor10] * [Reserve margin factor11] 》 Share of CP load reduction (%): [Avoided CP capacity] ÷ [Utility CP in MW] 》 Savings due to avoided costs ($): Utility avoided costs in the program year 》 Participant compensation ($): Participant compensation in the program year 》 Participant compensation ($ per kilowatt): [Participant compensation ($)] ÷ [CP demand savings] ÷ 1,000 》 Number of event hours in the program year 》 Number of program calls in the program year 10. T&D loss factor: [System production (megawatt hours [MWh])] ÷ [Retail sales (MWh)]. 11. Reserve margin factor: 1 plus system planning reserve margin. Selecting and Implementing Demand Response Programs 25 Appendix B. DR Program Case Study Examples for Developing and Developed Countries DR Measure Utility Name Country Program Description Results [customer class] [type] (state) ADR Tata Power Delhi India A smart-grid field study by TPDDL considered Lawrence Berkeley National Laboratory (LBNL) issued two [commercial, Distribution ADR with smart meters in New Delhi. The studies (Yin and others 2015; Deshmukh and others 2015), industrial] Limited (TPDDL) objective was to evaluate the technical capa- which described the analyses under the TPDDL study. [a public- bility, the potential for increased reliability, Depending on the methodology to calculate the baseline government joint and the readiness of commercial and indus- consumption, the maximum total demand reduction across all venture] trial buildings for the ADR program’s imple - 144 customers was 6.6 MW or 8.6 MW. For the 75th percentile, mentation. In total, 17 events, each lasting LBNL found a 2.2 MW reduction for both baseline methodolo- 0.5 to 1 hour, were called between noon and gies. With a system CP demand of 25.6 MW, a 2.2 MW demand 6 pm, and 144 commercial and industrial reduction equates to an 8.8 percent reduction in system customers participated, with data recorded at peak load (Yin and others 2015). During these hours, TPDDL’s 15-minute intervals. Two methodologies were avoided costs were at least ₹5.45/kWh (Deshmukh and others used to calculate counterfactual event loads. 2015) (6.5 US¢/kWh). TOU rates Electricity Ghana An ICF International study funded by the ICF’s recommended rate structure featured a 1.77:1 on-peak– [industrial] Company of United States Agency for International off-peak price ratio and a 5-hour on-peak period. With this Ghana Ltd. Development (ICF 2015) identified the structure, ICF estimated on-peak demand reductions of up to [government] implementation of a TOU rate structure for 56 MW under the proposed tariff, equal to 5.5 percent of the industrial customers as a measure capable industrial class peak in MW. of improving (1) forced load shedding and (2) ICF provided results from the UCT, which estimated net utility revenue collection via rates that better reflect benefits of $36.6 million, including annual program costs of the underlying costs. The industrial class about $0.3 million. was selected because these customers have smart meters and have the highest load per customer, offering the greatest potential for load modification. DSB Câmara de Brazil In 2021, Brazil faced both energy and capacity In September and October 2021, ONS accepted bids from Redução Voluntária Comercialização constraints and initiated a DSB program for eligible entities, which could provide load reduction services of da Demanda de Energia industrial customers and aggregators to over 5.9 GW (32.8 GWh) (ONS 2022). The accepted bids across [industrial, Elétrica (CCEE) provide day- and week-ahead offers to reduce these two months averaged approximately $1.41/kW, or $255/ aggregators] [nonprofit, load. The program lasted three months, from MWh. During this two-month period, CCEE reported savings private], and September 2021 through November 2021, on system service charges and short-term market prices of Operador but was no longer needed when loss of load R$42.1 million (equivalent to US$ 7.8 million), and the verified Nacional do probability returned to normal levels. load reduction during this period was 28.8 GWh (ONS 2022), Sistema Elétrico resulting in system savings of approximately US$ 270/MWh. (ONS) [nonprofit, The implied benefit-cost ratio between system cost savings private] ($270/MWh) and bid payments for cost reductions ($255/MWh) is approximately 1.06. Selecting and Implementing Demand Response Programs 26 DR Measure Utility Name Country Program Description Results [customer class] [type] (state) Managed EV STELCO Republic of The Maldives archipelago has heavily A World Bank study (World Bank 2023) found that in the charging [government] Maldives subsidized electricity rates since electricity Maldives, a 30 percent share of EVs by 2030 would shift peak [residential and generation costs approach US$0.50 per demand from afternoon hours to evening (6 pm to 9 pm) if EV commercial] kilowatt-hour. Fossil fuel used for transpor- charging is unmanaged. This increased peak demand would tation is modestly taxed. In this context, entail a 16 percent increase in generation capacity needs non-imposition of fossil fuel taxes and the relative to the status quo. However, an optimized EV charging provision of electricity rate subsidies have a regime would require only a 1.8 percent increase in generation combined impact of nearly $19,000 per million capacity relative to the status quo, and additionally would passenger vehicle-kilometers in favor of EV reduce stress on the distribution system. owners (World Bank 2023). A World Bank study analyzed the impact of optimized EV charging in this environment to better manage evening peaks. Supplemental Eskom [IOU] South Eskom, as a bulk grid operator, offers two The programs achieved up to 1.5 GW of avoided CP capacity DR; Africa dispatchable DR programs for industrial (NREL 2024), equivalent to 4.4 percent of avoided CP GW, instantaneous customers. These programs differ based or 15 percent of industrial customer contribution to CP GW DR [industrial] on the notification period for the dispatch (NREL 2024). of behind-the-meter generation (instan- taneously or 30 minutes in advance). Capacity-based compensation (ZAR/MW) for standby, and energy payment (ZAR/ MWh) for actual performance. Additionally, large industrial participants are exempt in early stages of load curtailment. INT base Southern California, SCE offers a base interruptible tariff to its In 2022, 343 customers were enrolled in the program, which interruptible California United large general service customers. Under was dispatched three times for a total of 7.5 hours. The program Edison (SCE) States this tariff, participating customers must demand reduction ranged from 391 MWh to 490 MWh per [IOU] receive service under a TOU rate schedule hour (roughly 2 percent of SCE peak demand) and averaged and commit to curtailing at least 15 percent 463 MWh per hour for a typical day event. The top 15 of maximum demand, or at least 100 kW, responders account for 54 percent of the total load impact during an event. Notification occurs no (Christensen Associates 2023). more than 30 minutes prior to an event. The 2024 SCE Base Interruptible program budget is approx- During nonevent periods, participating imately $68 million, comprising $1 million for administrative customers receive a monthly bill credit. expenses and $67 million in incentives (participating There is a high penalty for not achieving the customer bill credits). Assuming an average load reduction contracted load reduction during an event. of 463 MW, the average program cost is $147/kW of load Each event may not last longer than six reduction. The TRC test result for the program is 2.76 with hours. a 15-minute notice and 3.32 with a 30-minute notice (CPUC 2023). 27 Selecting and Implementing Demand Response Programs DR Measure Utility Name Country Program Description Results [customer class] [type] (state) Central Baltimore Gas Maryland, Participating customers have a switch or In 2022, there were roughly 373,000 residential DLC devices air-conditioning and Electric United programmable thermostat installed at their installed (PSCM 2023). Participating customers receive bill and heat pump Company [IOU] States properties to briefly curtail the usage of credits in the form of an installation incentive and monthly DLC central air-conditioning or an electric heat bill credits during the four summer months. The monthly bill [residential and pump when there are issues with system credits vary by utility and range from $50 to $100 depending commercial] reliability or high electricity prices during on the cycling level selected by a customer (50 percent, 75 critical peak hours (PSCM 2023). percent, 100 percent). The total DLC CP demand reduction capability in 2022 was 258 MW. Assuming an average summer month bill credit of $75, residential and commercial customers would have received bill credits of $112 million in 2022, equating to just under $434/kW of CP demand reduc- tion. This is a fraction of the $835–$938/kW cost of a new generation resource in the PJM region (Brattle 2018). ESS tariff Green Vermont, GMP subsidizes battery storage costs for As of 2023, the ESS program had: [residential] Mountain United customers in exchange for utility control 》 2,500 participants (1.1 percent of residential Power (GMP) States of storage devices in order to reduce peak customers). [IOU] load during events. Each customer partici- 》 22 MW of participating storage (3 percent of GMP’s pating in the ESS program leases two Tesla 750 MW peak demand), or 8.8 kW per customer on Energy Storage 2 systems with integrated average: inverter units and one Tesla Gateway 》 The average participant compensation was $18,055 (Gheorghiu 2019). Customers utilize the per system ($2,052 per participating kWa . remaining storage capacity not enrolled in 》 GMP’s analysis shows that each installation will result the program. GMP’s investment is treated in a projected positive lifetime net present value of as rate based. In approving the program, $2,749 per individual customer installation.a These the Vermont Public Utility Commission benefits accrue to ratepayers and are driven by determined that these tariffs benefit all reduced capacity and energy market savings, and ratepayers, and that the program supports Renewable Energy Standard Tier 3 benefits. carbon emission reductions and the devel- opment of Vermont’s small-scale battery storage market (Vermont Public Utility Commission 2019). a. PUC Case No. 23-1335-TF Final Order Approving Tariff Revisions—$23,555 upfront CAPEX minus $5,500 customer payment (Vermont Public Utilities Commission 2023). The federal investment tax credit (ITC) does not appear to have been incorporated. ADR = automated demand response; CP = coincident peak; DLC = direct load control; DR = demand response; DSB = demand-side bidding; ESS = energy storage system; EV = electric vehicle; GW = gigawatt; GWh = gigawatt-hour; INT = interruptible contract; IOU = investor-owned utility; kW = kilowatt; kWh = kilowatt-hour; MW = megawatt; MWh = megawatt-hour; TOU = time-of-use; TRC = total resource cost; UCT = utility cost test. 28 Selecting and Implementing Demand Response Programs Appendix C. Findings and Lessons Learned Analysis of the case studies in Appendix B supports the findings and lessons that are summarized below. Findings Insights Appropriately designed demand response Benefits include easing of capacity expansion investments, greater (DR) programs are cost-effective and can reliability, reduced system costs, and integration of variable renewable offer substantial benefits to grid operators in energy and new electric loads. developing countries. DR has been implemented cost-effectively In these countries, the implementation of peak load reduction in developing countries with measurable programs was driven by the high cost of serving a growing peak load, benefits. coupled with frequent load shedding events and reliability issues. The achieved load reductions in developing The case studies in Appendix B illustrate class peak demand reduc- countries are comparable to those experi - tions ranging from 4.4 percent to 8.8 percent of the system coincident enced in more mature markets. peak demand. These figures compare favorably to the 6.5 percent average peak demand reduction achieved in US wholesale markets (FERC 2023). Specific types of DR programs can be imple - Large commercial and industrial customers may already have neces- mented for a limited number of customers at sary metering and billing systems in place. Billing and event notifica- low cost. tion can be carried out manually for a limited number of participants. When a utility implements DR for the first Industrial customers have the highest usage per customer and time, industrial customers are often the therefore the highest potential load reductions per customer. These first customer class targeted to participate. customers usually already have metering capable of measuring hourly Targeting these types of customers is likely a or subhourly loads, and billing systems already suited to time-dif- “no regrets” initial offering. ferentiated rate structures. Industry may also be more capable of shifting production processes. Besides large nonresidential and industrial Factors contributing to this situation may include limited awareness classes, DR participation in developing coun - of potential benefits, weak administrative bandwidth, a lack of tries remains low. supporting customer- and utility-side technologies, or cultural and behavioral norms. There were no case study examples of resi - Air-conditioning is often an important contributor to peak loads, dential air conditioner control in developing and adoption is expected to increase in the future. This should be an countries. important target for future DR programs. Distributed energy resources can offer highly Case studies demonstrate that managed electric vehicle charging can cost-effective opportunities for flexible load reduce demand growth; residential air-conditioning direct load control programs. can deliver peak demand savings even in the absence of time-differen- tiated price signals; and behind-the-meter generation, such as storage, can cost-effectively reduce customer demand. Well-designed DR programs can be economic, When the DR program compensation is less than the avoided costs for even if retail rates are subsidized. utilities, DR can reduce utilities’ costs with no change to retail rates. Appendix A contains a discussion of how this can be achieved and quantified. Cost-reflective rates are not a precondition Subsidized rates may limit the types of programs that a utility can to and do not preclude the implementation implement. For example, subsidized electricity rates can diminish a of DR programs. utility’s ability to provide meaningful price signals and may limit the revenues available to invest in DR program infrastructure. Load reduction is more certain when the Examples of such DR programs are interruptible contract, direct load penalty for noncompliance is strong or when control, and automated demand response. direct load management is carried out Source: Original compilation. Selecting and Implementing Demand Response Programs 29