Engaging customers will require more than TOU pricing.
Lester B. Lave is a university professor and Higgins Professor of Economics and co-director of the Electricity Industry Center at Carnegie Mellon University. Email him at: firstname.lastname@example.org
Imagine a setback thermostat programmed at the factory that the consumer couldn’t modify. Who would want this device? You could give the customer a big enough discount to get her to accept the device, but she would be happier and you could save about as much energy if the customer could decide on the temperature and time settings.
Similarly, most of the in-home displays and dynamic-pricing structures entering the market today lack the most important feature necessary to change customer behavior. Namely, they don’t give customers control over their energy consumption in response to real-time price signals. Instead they tend to force changes in consumption according to a schedule that might or might not reflect true market prices at any given moment.
An electronic energy manager with a display conceptually is different from the smart meters and other devices in the field. It focuses on empowering the customer to decide how much electricity to use under specified conditions. It does not oblige the customer to reduce electricity use, but only does what the customer tells it to do.
The energy manager is similar to the best devices in that it empowers the consumer by providing better information. Unlike the other devices, it gives a wider range of choices as to how to respond to high prices. And like the best devices, it can respond immediately to price signals from the utility.
An underlying philosophy supporting the regulation of electric and gas utilities is that the utility should be required to supply sufficient electricity and gas to meet customer demand under all circumstances. The supply and delivery of energy through regulation must be met in a manner that mimics the provision of such services in a competitive market; prices must be reasonable and fair, supplies must be efficiently produced and delivered, and all customers must have access to such services. If this philosophy had been accompanied by real-time pricing that reflected the cost of the most expensive electricity generator dispatched, and customers were able to respond to price signals, it would have produced an efficient outcome. Instead, most customers served by a regulated utility pay a flat price, with energy priced the same at 2 a.m., when the locational marginal price is close to zero or even negative, and at 5 p.m. on the August afternoon when the year’s peak demand occurs and the locational marginal price is $1.00 per kWh. One result is that demand is sharply peaked, with 15 percent of the capacity in PJM used less than 100 hours per year, because customers are not even aware of the higher peak costs. Time-of-use studies show that customers want to buy less electricity at 18 cents/kWh than they do at 12 cents/kWh. Even critical-peak prices don’t get into the range that reflects the locational marginal price during the peak hours.
A 2008 study 1 showed in PJM that the daily bid schedule increases sharply even on days when demand is far below the system peak. Thus, reducing de-mand at peak periods on most days would lower the market clearing price of power significantly.
Utilities and their regulators have been reluctant to embrace real-time pricing because they fear that customers could not cope with this price volatility. However, customers cope with volatile prices for commodities such as gasoline, fresh produce, and even airline tickets and hotel rooms. One difference is that the volatile prices for these other commodities do not change each hour. (Many restaurants, airlines, and movie theatres do have time-of-use or seasonal pricing, and London, Rome, Stockholm, Singapore, and Milan have variable tolls for vehicles that depend on levels of congestion).
Most pundits reject the notion that residential, or even small commercial and industrial customers, would be willing or able to embrace real-time pricing that would require regularly checking prices and then responding by changing their thermostats or getting up at 2 a.m. to do the laundry. Concern for the constraints on consumers’ attention and effort has led to the use of approximations to realize the savings potential associated with real-time pricing. The simplest approximation is time-of-use pricing, with two periods: peak and off peak. Experiments have shown that customers can understand this simple distinction and reduce their use during the peak periods. However, the PJM study and other analyses show that time-of-use pricing delivers only a fraction of the potential benefit that real-time pricing could produce.2
A slight variant of time-of-use pricing, adding critical-peak pricing, can increase savings significantly. Consumers are advised, perhaps a day in advance, that a critical situation will exist, during which prices will be considerably higher. An experiment in Anaheim, Calif., in 2005 found that customers reduced demand, with a price elasticity greater than expected from previous econometric studies.3 Given this experience, critical peak pricing could be used on a larger scale as advanced meters are more widely employed.4
One of the Anaheim study’s key findings is that, once empowered, customers wanted more control over their electricity use and better information about how to lower their energy bills. A system that manages a customer’s energy usage without that person’s consent probably won’t be approved of by many customers. To be successful, a system must engage the customers—it must have their knowledge and consent.
Nonetheless, critical-peak pricing still only is an approximation of real-time pricing. Critical-peak pricing can handle anticipated high demand, but not the unanticipated demand of a weather change or a supply shortfall due to a generator or transmission line tripping. These unanticipated situations represent a surprisingly large proportion of the benefit of real-time pricing (see endnote 2).
Electronic Energy Manager
A utility could implement real-time pricing in a way that reduces the burden on customers to respond to price changes, and in particular to high price volatility. This would require providing individual consumers with an electronic energy manager combined with a smart meter and a display that they can program to respond to price changes and personalize to their own home energy policy. The utility would send real-time pricing information to the energy manager, which then would execute the customer’s chosen policy, with no further customer attention needed. For example, a consumer might specify that, at 20 cents/kWh, the thermostat be shifted by 3 degrees. At 30 cents, a 6 degree change might be specified. At $1/kWh, the customer might want all appliances turned off, except a few essential lights, clocks, and entertainment equipment. Such a device, capable of receiving price signals from the utility and communicating with appliances throughout a residence or business, is well within the state-of-the-art of electronics. It could be coupled with an advanced meter with two-way communication for an estimated total installed cost of perhaps $300 to $400.
A greater difficulty would be constructing a user interface that most electricity customers could use. The interface software would guide the average consumer through decisions on setting the energy manager, perhaps augmented by phone support. Some programmable thermostats talk to the customer, guiding decisions. A careful study of how people would interact with this device, combined with modern communication devices, could accomplish the task for most customers.
Like time-of-use and critical-peak pricing, this energy manager could efficiently and effortlessly lower peak demand. It has much else to contribute. If an unforeseen event occurred, the energy manager could react to it instantly—helping to maintain the stability and reliability of the grid. For example, weather forecasts might expect cloud cover that would keep the temperature below 90 degrees F. However, the weather front might stall and unexpected full sun might raise the temperature to 96 degrees, significantly increasing the demand for cooling. At the same time, an unexpected supply event, such as a generator or transmission line tripping, could produce an immediate shock to electricity grid. With real-time pricing, a customer would have an immediate incentive to respond to these unexpected events; the response could be large and fast, and each customer might use only 10 to 20 percent of load, if a large number of buildings had an energy manager acting on their behalf. By comparison, time-of-use pricing and day-ahead critical pricing couldn’t react similarly to the situation. At present, the utility must maintain spinning and non-spinning reserves to handle such a contingency, leading to costly amounts of spinning and standby reserve.
In contrast to time-of-use pricing, the energy manager would deal with actual real-time changes in demand by allowing customers to wash a load of clothes during the peak time of use, if the system’s actual demand and prices were less than expected.
A 2007 Pacific Northwest National Laboratory study5 employed a device that reacted to a drop in frequency to restore power quality. The experiment showed that the device could help to maintain power quality.
The Benefits of Control
Because the energy manager would react instantaneously to price or control signals, it could be used to bring load-side resources into the ancillary services markets. It could help to level demand, hold the power frequency close to 60 hertz and improve power quality more efficiently and effectively than making the adjustments on the generation supply side (see PNNL study). It might be possible to do away with spinning reserve from generators by providing this service at lower cost on the load or demand side. For example, if a utility’s grid experienced a sudden jump in demand or drop in supply due to a generator or transmission outage, the price would rise, leading energy managers to curtail demand. A 5-cent rise would prompt the most price-sensitive customers to curtail use by allowing temperature to change by a few degrees and curtailing other appliances. If more responses were needed, a 10-cent price increase would lead to further temperature changes for the price-sensitive customers and some temperature changes for the less-sensitive customers. This adjustment mechanism likely would reduce sufficient demand for long enough to start a gas turbine and get it synchronized with the grid, obviating the need for spinning reserves. Similarly, with a bit more sophistication, reactive power needs could be supplied by demand-side management of air conditioners, pool pumps, and air handlers. In order to provide these services, customers would be offered a payment for giving the utility the right to allow momentary, probably imperceptible, interruptions to these devices.
Perhaps the greatest benefit of managing demand in this manner is the resulting increase in energy-system reliability. In each of the most recent major cascading blackouts, demand exceeded generation by only a relatively small amount. If that amount could be shed on the consumer side, the utility would avoid a cascading blackout. At present, a utility could cut off a major customer with an interruptible contract or could black out a substation. But interruptible contracts call for warning periods, and blacking out a substation is an extremely costly solution since everyone served by the substation would lose power. The cost of all customers losing 10 to 20 percent of their power for a limited time via targeted curtailments would be significantly lower than the costs associated with having some customers lose all power for a longer time.
An ancillary benefit of these electronic energy managers is that they would support the introduction of more wind and solar power to the grid, as well as aggressive electricity reduction programs in California, New York, Vermont, and several other states. In today’s utility system, maintaining power quality with a substantial amount of generation from intermittent renewable resources would require fast-reacting storage, such as batteries, which would be expensive. The customer side of the meter offers much less expensive control with which to compensate for the intermittency of these growing generation sources. Finally, demand reduction, particularly at peak times, would be facilitated by these energy managers, since they would inform consumers and enable them to translate their desires into concrete action. By shifting demand from peak to off-peak, these energy managers would allow the installation of more baseload units. In contrast to gas turbines, the nuclear generators and new coal plants with carbon capture and sequestration can’t be cycled. To introduce these new units into the dispatch order, the load-duration curve must be flattened. Current demand is going in the opposite direction, since the peaks are rising relative to average load.
Company engineers prefer “direct load control” since they know what they have. In some cases, the customer doesn’t even perceive the momentary interruptions of the air conditioner and other appliances. The energy manager is better for two reasons. First, more customers will participate if they have control rather than ceding it to “big brother.” Second, the customers, not the utility, customize their responses. These two benefits make for happier customers and greater participation. The worry about loss of utility control is overblown. If the utility can query the energy managers, they know how the devices will react. Some customers may choose to override their energy manager, knowing there will be a penalty, but the individual power curtailment will be small because nearly all customers will participate.
1. Spees, Kathleen, and Lester Lave. “Impacts of Responsive Load in PJM: Load Shifting and Real Time Pricing,” The Energy Journal, Vol. 29, No. 2. 2008.
2. Borenstein, Severin. “The Long-Run Efficiency of Real-Time Electricity Pricing,” The Energy Journal. 2005; 26; 3. pp 93.
3. Wolak, Frank A. “Residential Customer Response to Real-Time Pricing: The Anaheim Critical-Peak Pricing Experiment,” May 24, 2006.
4. Energy Action Plan 2008 Update. February 2008. California Public Utility Commission and California Energy Commission.
5. Hammerstrom, D.J., et al, Pacific Northwest GridWise Testbed Demonstration Projects; Part II, Grid Friendly Appliance Project. October 2007, Pacific Northwest National Laboratory, PNNL-17079.