DEMAND RESPONSE: HOW TO REACH THE OTHER SIDE

Eric Hirst is an energy consultant based in Oak Ridge, TN.

Electricity customers can be active in the wholesale market. But getting them to participate and creating a vibrant marketplace require several steps.

Everyone's excited about it. Virtually every state and federal government agency with an interest in energy issues has supported improved integration of wholesale and retail electricity markets. When the U.S. Federal Energy Regulatory Commission (FERC) and the U.S. Department of Energy cosponsored a Conference on Demand Response, almost 400 people attended. FERC's new notice of proposed rulemaking on a standard market design emphasizes the importance of demand response to healthy wholesale power markets.

Such improved integration—getting customers to respond to generation shortfalls by reducing demand—requires at least some retail consumers to face dynamic prices that vary from hour to hour or to participate in programs for reducing their electricity consumption when prices are very high. These price-responsive demand actions would improve bulk-power reliability, economic efficiency, and environmental quality.

In spite of all this enthusiasm, few retail customers currently face electricity prices that accurately reflect wholesale power costs. For example, the emergency and economic demand-reduction programs run by the PJM Interconnection in 2001 signed up 220 megawatts (MW) of load reduction and achieved a maximum load reduction of 62 MW on August 9, 2001, just over 0.1 percent of PJM's typical summer peak load.  The Emergency Demand Response Program, run by the New York Independent System Operator (ISO), achieved an average load reduction of 355 MW during summer 2001, a savings of 1.1 percent. These programs have achieved only modest results to date, in part because they are so new and in part because of many obstacles—customer acceptance, regulation, infrastructure, and technology.

Desire and Ability
Consumers do not respond to dynamic prices for two reasons:

• They have no motivation to do so because, in most cases, the price they pay for electricity is time-invariant.
• They have no means to do so because they are not informed of changes in wholesale electricity prices, and the meters that record their electricity use do not store data at the hourly level.

But even if customers had the opportunity to face dynamic prices and had interval meters and the necessary communications systems, they still might choose not to participate in such programs. Most consumers do not want to face volatile prices because they equate them with higher bills. They also generally do not recognize that high prices during a few hours a year are more than offset by low prices during the rest of the year, resulting in a lower annual electricity bill. In addition, consumers may not recognize the opportunities they have to shift consumption from high-priced to low-priced periods.

Customer education, therefore, is critical. Customers need information on how dynamic pricing works and how they might benefit from such programs before they will be willing to participate in them. At a more fundamental level, it might be worthwhile to educate consumers on how electrical systems operate and how wholesale electricity markets function. An understanding of operations and markets might help consumers realize that electricity consumption and production costs are quite volatile. As a consequence, a fixed price for electricity provides insurance to the consumer against both quantity and price risk, insurance for which customers must pay.

That becomes a challenge for regulators, who should recognize electricity's commodity and risk-premium components that currently protect consumers from uncertainties about the timing and amount of their electricity use and the volatility of wholesale electricity prices.

Electricity consumers differ in ways that affect their interest in and ability to respond to time-varying electricity use. Some of the key characteristics include amount of electricity use, load shape (how load varies from hour to hour, day to day, and season to season), flexibility of operations (including the speed and ease with which consumers can modify loads), and automated control of some electricity-using equipment.

The greater their electricity use, the more likely customers will be to devote time and effort to understanding their options to reduce electricity costs. The hour-to-hour patterns of electricity consumption for different kinds of equipment affect the benefits of modifying consumption. For example, electricity use for air conditioning is highly correlated with hourly prices, while water-heating electricity use is largely uncorrelated with prices. This difference suggests that reductions in air-conditioning use are likely to yield larger dollar savings per kilowatt of load reduction.

Where customers are flexible, the cost of participating in demand-response programs is low. For example, municipal water-pumping systems (which account for about 3-4 percent of total U.S. electricity consumption) typically have tanks, reservoirs, or lakes to store water for later distribution to consumers. These facilities permit the water-treatment system to interrupt pumping operations for up to a few hours at a time. During such periods, gravity will ensure sufficient water flow and the appropriate pressure to consumers.

If control of a particular process or piece of equipment can be automated, so that it does not require manual intervention to respond to time-varying prices, participation is likely to be much higher. For example, households are unlikely to participate in programs that require them to turn off manually their electric water heaters at certain times. However, many utilities operate direct-load-control programs that send a radio signal to the heater to switch it on and off automatically.

Because of these differences in customer characteristics, retail providers should (and do) design different programs for different types of customers. Residential and small commercial customers might be best served with the traditional direct-load control and time-of-use programs utilities have run for years. Large commercial and industrial customers, on the other hand, might take advantage of more sophisticated program offerings, including the opportunity to sell short-term load reductions as contingency reserves for reliability purposes and real-time pricing that varies from hour to hour. (See the sidebar, "Reforming Pricing in Retail Markets.")


Surveys of customers participating in demand response show they prefer programs that

• are simple to understand and sign up for,
• permit aggregation of small loads,
• are announced well in advance of implementation,
• provide a public-relations benefit to the customer (e.g., for helping to avert a regionwide blackout),

are voluntary, without penalties for failure to reduce load when called upon to do so,

• provide ample advance notice of any consumer action that must be taken, and
• pay well for the load reductions.

But utilities still have a dilemma. With wholesale electricity prices currently low, consumers see little incentive to participate in these programs. On the other hand, when prices are high, it is too late to design and implement such programs. One possible solution is for retail suppliers to accept some risk and guarantee a certain number of high-priced hours to participating customers. This guarantee would ensure that customers receive some benefit from the investment in metering technology and avoid too much price volatility. In essence, the retail providers would act like long-term insurers.

The Regulatory Risk
A fundamental obstacle to greater use of demand-side resources is uncertainty about future government regulations and market design. Until the wholesale market rules for energy, transmission congestion, and ancillary services are stable, suppliers and consumers will be unwilling to invest time and money to manage demand. FERC's proposed standard market design will be of enormous value here. The rules concerning price caps and other forms of market-power mitigation also need stability. The California electricity crisis and the regulatory responses to those problems have made many participants wary of long-term commitments to programs that may be abruptly modified or canceled.

The rules that state regulators have established to govern retail competition are, perhaps inadvertently, limiting customer participation in dynamic-pricing programs. In some states, public utility commissions (PUCs) have mandated rate discounts, imposed rate freezes, and established predetermined load profiles as part of the transition to retail competition.

Standard-offer (or provider-of-last-resort) service is especially troublesome for demand-response programs. In particular, these services typically require the distribution utility to offer a discounted and time-invariant electricity price. As such, they can block competition by

suppressing customer awareness of efficient price signals (as customers remain ignorant about the dynamics of pricing and never consider demand management);

• hindering development of forward markets that would otherwise be used for customer hedging;
• discouraging new retail providers from offering risk-management services as value-added products;
• keeping utilities from becoming true wires companies; and
• potentially threatening the financial viability of incumbent utilities.

The key to resolving these problems is explicit PUC recognition that providing fixed-price electricity includes an insurance policy as well as the electricity commodity. These risk-management costs must be included in the customer's rates and reflected in the service provider's earnings. PUCs should be sure the standard-offer service is a fair deal, for both customers and the provider—but not too good a deal for customers. A deal that is "too good to believe" today will have to be paid for later! To that extent, regulators should encourage retail providers to offer risk-management products to help customers deal with dynamic pricing.

When PUCs impose rate caps on the local utilities, the utilities lose money if they run innovative load-reduction programs and pay for the associated metering infrastructure. To the extent the utility recovers fixed transmission, distribution, and customer-service costs through a volumetric charge (i.e., on a cents-per-kilowatt-hour basis), its revenues and earnings will decline if customers reduce their electricity use. PUCs may want to modify their pricing policies to encourage such programs.

Metering and Billing
PUC indecision on metering, billing, and access to meter data may slow the adoption of the infrastructure technologies necessary for dynamic pricing. If utilities do install advanced meter-reading systems, how will they recover costs if customers switch to a different energy supplier? In the meantime, what entities have access to customer-meter data? Advanced metering can occur with either a regulated monopoly or a competitive market, but it is unlikely to occur until regulators decide on the framework for such infrastructure issues.

The same is true for the computer systems required for billing and settlements. Obviously, more sophisticated software is required to bill customers with a time-varying price or for ad hoc load reductions than for electricity consumption at a time-invariant price. Unresolved issues—recovery of software and computer development and implementation costs and whether these services should remain with the local utilities—may inhibit development of these systems.

Finally, PUCs need to decide who will pay for these infrastructure costs. The individual customers participating in dynamic-pricing programs could pay for the metering and communications associated with their facility. Alternatively, all retail customers could pay for them if these programs benefit all customers, not just program participants. Whoever pays, regulators should require that interval meters be installed for larger customers. As the costs and performance of metering and communications systems improve, regulators can lower the minimum size for mandatory interval meters.

The actual determination of rates is another hurdle. Utility rates today are based on customer-class load shapes. High-load-factor (e.g., large industrial) customers are the ones most likely to leave these rate classes to participate in price-responsive demand programs (a form of self-selection). The load shape of the remaining members of the class will then worsen. Absent a rate case, this change raises the utility's costs to serve the rate class with no corresponding increase in revenues.

Using predetermined load profiles rather than hourly metering to bill customers further inhibits adoption of price-responsive demand. If customer meters are read only monthly, retail providers will have no knowledge of the dynamics of electricity use and, therefore, no ability either to charge or to reward customers appropriately. In a similar fashion, customers will have no incentive to respond to time-varying wholesale prices.

In addition to making interval meters a requirement for retail electric service (at least for larger users), PUCs can encourage adoption of dynamic pricing and voluntary load-reduction programs regardless of whether retail markets remain regulated or are open to competition. Indeed, many of the price-responsive demand programs are run by utilities under state regulation. A prerequisite for these programs, however, is a competitive wholesale market with visible hourly prices.

Federal Regulators
At the federal level, FERC's low price caps in the independent system operator (ISO) markets will suppress customer participation in voluntary load-reduction programs. The limited experience to date suggests that consumers will not participate in such programs unless the price exceeds about $250 per megawatt-hour (MWH). Indeed, the California ISO proposed a performance payment of $500/MWH plus a reservation payment of $20,000 per MW per month for its summer 2001 Demand Relief Program, well above the $250/MWH cap in place at that time. The New York ISO's emergency demand response program paid participants an average of $514/MWH of load reductions during summer 2001. More generally, the greater the risk and magnitude of price spikes, the greater the incentives for price-responsive demand.

A critical issue is the potential conflict between state and federal regulation of price-responsive demand programs. Although FERC regulates wholesale markets and the ISOs that operate them, it has no jurisdiction over retail activities. State PUCs, on the other hand, have authority over sales and service to retail customers but limited jurisdiction over wholesale markets.

Shareholder-owned utilities, regulated by both FERC and the states in which they operate, are caught in the middle. The ISOs, acting under FERC authority, might implement programs that impose costs on utilities—such as those for metering, billing, program administration, and loss of revenues. The latter can occur when fixed costs (e.g., for transmission and distribution systems and for stranded costs) are recovered through volumetric charges, as they often are. In addition, the retail price caps under which many utilities operate provide gaming opportunities for customers—they can choose to face market prices when it is advantageous for them and then regain the protection of the state-imposed price cap when that is the preferred option. Thus, utilities may be required by FERC to implement programs that increase their costs and reduce their revenue. These costs, however, can be recovered only with approval from the state regulator.

The solution here is not to ban ISO programs but rather to ensure that the wholesale and retail efforts to better manage retail demand are coordinated. The goal should be to offer a comprehensive set of programs aimed at all customer classes with a minimum of duplication and confusion.

Finally, all retail-service providers should be allowed to offer price-responsive demand programs. And no providers (including the local utility) should be forced to incur costs that are not fully recovered. In particular, PUCs should permit regulated utilities to recover all reasonable program costs through rates.

Can the ISO Adapt?
In some respects, ISOs are logical candidates to promote demand-side participation: They run many of the day-ahead, hour-ahead, and real-time markets, anyway. Because they are responsible for making these markets efficient and competitive, they are also interested in expanding the range and diversity of participants.

On the other hand, ISOs, experienced in the wholesale market, are structured to deal with only a few market participants, each of which is responsible for hundreds (or at least tens) of MW of demand or supply. ISOs are not set up to interact with thousands (or millions) of customers. Perhaps the issue is whether the manager of wholesale markets has a responsibility to ensure easy access for retail customers and their providers.

As a practical matter, existing ISOs have introduced pilot demand-side programs during the past few years. They operated small experimental programs for summer 2000, which they then refined and expanded in 2001.

At a minimum, ISOs must ensure that all customer loads can participate in regional transmission organization (RTO) markets. Just as ISOs have developed rules and procedures to deal with the idiosyncrasies of different types of generation (different ramp rates, speeds with which generators can change direction, minimum runtimes, and abilities to operate at less than full output) so must RTOs develop rules that accommodate differences between the loads and generators (as well as among different types of loads) that might participate in their markets.

ISOs cannot simply require that loads conform to the existing supply-side rules—those rules were developed with generators, and only generators, in mind. For example, some loads may be able to reduce their output quickly and therefore provide contingency reserves. But part of the contract is to restore those loads so that the ISO can respond to the next contingency, and some loads may not be able to do that within the ISO-prescribed time (say, 10 minutes). Rather than prohibiting those loads from the reserve markets, ISOs should decide if a longer period of reduced load may be acceptable. Perhaps a customer who cannot increase consumption for at least two hours should still be permitted to participate in contingency-reserve markets but receive a lower price than those who can more rapidly restart their loads.

ISOs also must ensure that their scheduling and dispatching software can accept bids from customer loads as well as supply resources—not all ISOs have software that can do that. In some cases, the number of retail loads that can be analyzed is limited. In other cases, demand bids are treated after the program optimizes across the supply bids, resulting in a situation in which the demand bids cannot set the market clearing price.

In addition, flaws in market design and implementation can yield prices that are economically inefficient and that discourage demand participation in wholesale energy markets. According to a report analyzing New England market prices during high-load hours in summer 2001, "...New England market rules and/or actions of the ISO have artificially depressed market prices and created attendant economic distortions during periods of high demand this summer." These price distortions would have suppressed any demand response that might otherwise have occurred. The California ISO has also experienced persistent problems with its intrahour market, leading to interval and hourly prices that often do not reflect actual operations.

ISOs should review their metering and telecommunication requirements. The typical requirement is for metering and telecommunication of generator output to the ISO control center every four to six seconds. Such a high frequency may make sense for large generating units, but it may not be necessary for small loads. System operators monitor the output from large generators so frequently because the failure of any one unit must be compensated for immediately. Because almost all loads are tiny compared to generators, the statistical averaging across loads greatly reduces the need to monitor the consumption of individual loads. Also, it is inherently more reliable to turn something off (e.g., interrupt a load) than to turn something on (e.g., start a combustion turbine) to provide contingency reserves.

Moreover, ISOs need to be sure that their metering and telecommunications requirements are consistent with the RTO's reliability responsibilities. The California ISO, in its effort to encourage load participation in its ancillary-service markets, relaxed the frequency with which individual loads must record and report their consumption from once every four seconds to once a minute. Because these loads are required to respond within 10 minutes of an ISO request, it is hard to see why the ISO would need to know the consumption levels of these loads more often than once a minute. Although the ISO needs to be able to measure the performance of loads in delivering services to the ISO, it may not need to make these measurements in real time.

A New World
Perhaps the greatest cultural barrier to price-responsive demand is the belief that electricity costs and prices should remain time-invariant. Again, consumers, suppliers, and regulators need to recognize the risk-management component of electricity pricing. A related barrier is the lack of experience with market-based prices that vary from hour to hour and with retail-customer programs that encourage economic responses to price changes.

In addition, the system operators (today's vertically integrated utilities and ISOs and tomorrow's RTOs) have traditionally focused on the supply side and ignored the demand side of the equation (by assuming, in essence, that demand is completely price inelastic). That is, they maintain reliability by managing generation and transmission assets to meet fixed customer demands. However, customer loads can participate in bulk-power operations to maintain reliability in commercial markets. System operators need to broaden their thinking to accommodate the unique characteristics of customer loads, just as they have done for generating units. Ancillary services should be defined in terms of their functions, not with reference to the generators that traditionally provided the services. System operators must recognize the reliability benefits of using large numbers of small loads that can respond quickly.

The retail providers should conduct additional market research to understand what customers want from their electricity supplier and how customers might respond to different products and services. More important, retail providers need to offer a variety of price and risk options—customers differ substantially in the magnitude of their electricity consumption, load shape, storage capabilities, automation of electricity-using processes, and other factors that affect their interest in and ability to participate in dynamic-pricing programs. The retail providers and PUCs need to educate consumers about electricity, its production, costs, and alternative pricing and risk-management strategies.

To get this information, the electricity industry should support market research to learn which customers can benefit from which types of demand-response programs and how best to educate customers about these opportunities.

Finally, not everyone benefits when loads respond to prices. In particular, generators lose money when customer demand drops in the face of high prices. Because generator earnings are sensitive to price spikes, generator owners might object to RTO efforts to accommodate price-sensitive loads.

For example, one observer of the Electric Reliability Council of Texas believes that ERCOT has been slow to encourage demand-side participation in its markets because the owners of generation and the power marketers dominate the board and committees. These market participants lose money if customers are more price responsive.

It Comes Down to the Meter
All the technical components necessary for dynamic-pricing and voluntary load-reduction programs exist and have been applied in various settings. However, the market penetration of these technologies has been very limited because of their high capital costs (which in turn relate to their limited market penetration) and the customer, utility, and regulatory barriers.

The technological barriers include lack of interval meters, lack of digital communications systems, and limited use of sophisticated end-use energy management and control systems. This is a key issue, no matter how much the technology has become incrementally more accessible. (See the sidebar, "The Oldest Prophecy in the Energy Industry.")

Unfortunately, the industry has not evolved to the point that standardized off-the-shelf equipment and communication packages are readily available. It seems that every program custom-designs its own infrastructure. To the extent that complete systems involve components from various manufacturers (e.g., meters, communication systems, and data-analysis software), the industry may need to develop standards to ensure that the various components can work well with each other, regardless of who manufactures what. Thus, large-scale deployment of price-responsive demand requires technologies that can manage hundreds of thousands of retail customers, encompass several functions, and integrate systems from different vendors and service providers. (See the sidebar, "The Regional Negawatt Hub.")

Although the evidence on the capital and ongoing costs of these systems is sketchy, there is substantial opportunity for cost reductions. In particular, as more utilities and retail providers offer such programs and the number of installations increases, the cost per customer should decline. This will permit the cost-effective application of these programs to smaller and smaller electricity consumers.

The metering, communications, and analysis tasks are not particularly challenging. Cell phones, pagers, and calculators perform more demanding duties. If the requirements for these services can be standardized, mass-produced electronics can likely dramatically reduce the cost and increase the performance of advanced metering. This would facilitate real-time market response for even the smallest load.

Making Price-Responsive Demand Work
Virtually all participants in bulk-power operations and wholesale electricity markets recognize the importance of a robust demand side. Increasing the amount of retail load that responds to time-varying wholesale electricity prices and operating conditions (especially emergencies) will improve economic efficiency, lower consumer costs, improve reliability, and reduce the adverse environmental effects of electricity production and use.

But getting to this desirable end state is complicated by the many regulatory, customer, cultural, and technology barriers to the adoption of such programs. While FERC's standard market design is an important step forward, more needs to be done.

Regulators have the bulk of the work. First, they should recognize that traditional electricity pricing has two components: the electricity commodity and a risk premium that protects consumers from uncertainties about the timing and amount of their electricity use and the volatility of wholesale electricity prices.

Then, because the traditional fixed-price for electricity includes critical risk-management services, regulators should define the default service for all retail customers as day-ahead hourly pricing. Because most residential and small commercial customers will want to face prices that change from hour to hour, regulators should encourage retail providers to offer risk-management products that will help customers deal with dynamic pricing.

Regulators also should require that interval meters be installed for larger customers. As the costs and performance of metering and communications systems improve, regulators can lower the minimum size (in kilowatts) for mandatory interval meters.

FERC needs to ensure that demand can participate fully in all energy, congestion management, and ancillary service markets. At the same time, ISOs and RTOs need to recognize the unique characteristics of loads, just as they already do for generators. Moreover, FERC and the ISOs need to decide whether the current ISO load-reduction programs are transitional (intended to jump-start retail markets) or permanent.

Vendors need to develop automation technologies that make it simpler for customers to respond to dynamic pricing. And the electricity industry needs to conduct market research to learn who can benefit from which types of demand-response programs and how to best educate customers about these opportunities.

© 2008 Edison Electric Institute