Reliability Is A Supply and Demand Issue

A blackout is a supply failure, so naturally people look for supply-side solutions: more transmission lines, high-tech system monitoring devices, building power plants closer to population centers, better grid planning and testing procedures. Few people consider how effectively demand response and active retail markets can help reduce strains on the grid and forestall future grid and plant construction. We can, and should, use market-based retail pricing to communicate customer demand into the grid, and one reason to do so is that it would enhance reliability.

First, to give credit where credit is due, note that the Task Force final report does, in fact, mention “demand response.” Twice. On page 149, in the discussion of recommendation 13 (DOE should expand its research programs on reliability-related tools and technologies), the report: (1) cites “demand response initiatives to slow or halt voltage collapse” as one aspect of research into ways to prevent cascading power outages; and (2) urges the “study of obstacles to the economic deployment of demand response capability and distributed generation.” In addition, with a little creativity, a role for demand response could be read into a few other recommendations.

For the most part, however, the report is about a supply failure and supply-side proposals. For 99+ percent of the blackout report, end-use consumers are simply the “load.” A passive burden that the supply side must go to great lengths to serve.

Consumers are the sleeping giant of electric reliability. The North American Electricity Reliability Council (NERC) divides reliability into two categories: security of operating reserves and adequacy of installed reserves. Security is more of a short-run operational issue, while adequacy relates to planning for system growth. Demand responsiveness can contribute to both kinds of reliability.

Retail electric choice puts more control in the hands of consumers and empowers them to make intelligent energy choices. Consumers could choose anything from a fixed price that incorporates an insurance premium to full real-time pricing, in which the customer bears the financial risk of price volatility, but could see electricity bills fall by shifting or reducing use. (See Eric Hirst, “The Financial and Physical Insurance Benefits of Price-Responsive Demand,” The Electricity Journal, May 2002.)

Dynamic pricing harnesses the dramatic improvements in information technology of the past decade to provide price signals that reflect variations in the actual costs of providing electricity at different times of the day. These same technological developments also give consumers a tool for managing their energy use. They can set electricity monitors to increase air conditioning temperatures if prices go above a certain amount, for example, or can shift manufacturing schedules to minimize electricity use during peak hours. Right now, with almost all U.S. consumers paying average prices (even many industrial and commercial consumers), consumers have little incentive to manage their consumption and shift it away from peak hours during the day.

That inelastic demand leads to more capital investment in power plants and transmission lines than would occur if consumers could make choices based on their preferences. Reducing peak use contributes to greater operational security, as fewer reserves are necessary to maintain reliability, and eases stress on adequacy planning, as the need for system expansion to support ever greater system peak loads is diminished.

Both historical experience and laboratory experiments show that electricity customers do respond to price changes, and that both suppliers and customers are better off from doing so. (On experience: Google electricity and “elasticity of demand”; on Experiments, see “Rassenti, Smith, and Wilson.”) This option does not currently exist for most customers in most places.

Another approach to enabling consumers to contribute directly to reliability comes from efforts to turn demand response into a tool that transmission system operators can call on in their efforts to keep supply and demand constantly in balance. Research by Brandon Kirby and John Kueck, Oak Ridge National Laboratory, showed that a significant portion of the California Independent System Operator’s spinning reserve requirement could be supplied from the California Department of Water Resources pumping load. The CDWR could stop pumps for brief intervals in response to specific short-term transmission system needs. Another scheme would enable controllable air conditioning units to be cycled off for brief periods when the system is stressed. (See reports here. The research was supported in part by the DOE’s current research program in transmission reliability.)

Consumer demand is the sleeping giant of electric reliability, and retail rate regulation is what put the giant to sleep. Consumers, it is time to wake up! Retail pricing is a crucial component of a healthy, dynamic electricity industry. Offering consumers service choices in a range of prices would make diverse consumers better off, and bolster system reliability. This approach, though, is a novel value proposition in this historically regulated, vertically integrated industry with its “one size fits all” approach of regulated, fixed, average rates. Customers, regulators, and the electricity industry itself will increasingly recognize the range of value propositions that the electricity industry can profitably present to consumers, and we’ll all be better of for it.


8 thoughts on “Reliability Is A Supply and Demand Issue

  1. Hmm. If consumers are offered pricing plans that reflect their marginal demand at any time (and, thus, gives them an incentive to cut their demand , just how much system overcapacity is that going to generate?

    There is a whole generation of peaker plants that are going to be wiped off the map. They will simply never run, being designed to meet a static, average demand.

    There might be some transition problems to the new system.

  2. I agree that there is great POTENTIAL value in enabling consumers to respond to peak demand conditions by shifting their demand from on-peak to off-peak periods. However, I think this discussion is premature with regard to the Northeast blackout, for several reasons.

    First, consumer response to situations in which they could benefit by shedding load requires notification. The logical source of such notification in this case would have been First Energy. However, First Energy was apparently unaware (or unwilling to admit) that it had a problem and thus would likely not have been in a position to provide such notification. This capability must be put in place first.

    Second, consumer response will not be adequate to make any significant difference unless consumers see a benefit to them which exceeds their perception of the cost of response. The history of consumer demand response programs has been problematic, largely because the perceived benefit was lower than the perceived cost. In most cases, this was the case because the consumer benefit offered was unrelated to ( and significantly lower than)the real cost of continuing to meet their demand.

    Third, consumer response must be able to occur rapidly enough to make a difference. While this is technically possible, it certainly is not the case today. One explanation of why the blackout spread as far and as fast as it did is that the safety systems employed at many power generating stations operated to protect these generators faster than the safeties on the transmission system acted to protect the grid. For a consumer load shedding protocol to respond meaningfully, it would not only have to be automated, but it would also have to detect and respond to a problem on the grid faster than either the systems protecting the grid or the systems protecting the generators. This would require a degree of integration and coordination which cannot realistically be achieved until the integration and coordination of the grid and generator safety systems has been achieved.

    Fourth, the conditions which would trigger consumer response must be clearly defined. The typical definition for existing programs is very high peak demand. However, the grid was not at a very high level of demand at the time of the blackout. Therefore, the parameters of existing consumer demand response programs would not have triggered load shedding prior to the blackout.

    Finally, as long as utilities are both generating and distributing electricity, there will be significant reluctance on the part of utilities to shed consumer load. This would likely result in delaying consumer response until it was too late to make a difference.

    It certainly is reasonable for consumer response to be considered as part of the long term solution going forward. However, its implementation as part of the long term solution must wait until the remainder of the system issues are addressed and resolved. That does not mean that the implementation of consumer demand response programs should be delayed; it means only that they cannot be relied upon as a part of the long term solution until the far greater issues are addressed and resolved.

  3. Ed:

    How would “modulation” strike you as a verb, compared to “shedding”?  Consumers already strike deals with power companies for cheaper rates vs. off-peak, controllable or interruptible service.  Utilities have been in control of many people’s air conditioners for at least a decade.  It does not take much (if any) new technology to allow a utility to control all the major loads in a home or business and use them to regulate load on a minute-by-minute basis.

    What I’d like to know is if a grid requires central command and control to be stable, or if things can be handled by local sensing of load conditions by measurement of voltage and phase.  Could many small generators keep a system up and running just by keeping the local conditions in good shape, or would this lead to instability and failure?  So far as I know this has not been determined.

  4. E-P,

    Most electric loads cannot be modulated, either locally or remotely, except by reducing voltage (brownouts). Unfortunately, brownouts can damage motors through overheating, thus causing economic damage to their owners. Therefore, loads are shed (turned off).

    I am aware of the existing technology for turning loads off during peak periods and the existing utility programs offering rate reductions in exchange for the right and ability to switch specific loads on and off to manage peak demand. These programs have achieved only limited success because they have typically offered only minor savings to the target consumers.

    Curtailable and interruptible contracts for large commercial and industrial customers have been more successful because the customers were able to achieve significant savings and also to make alternative arrangements to meet their needs during interruptions.

    The issue here is that First Energy would not have had the information necessary to trigger the load shedding. In addition, the load on the system at the time of the failure was not high enough to have triggered a load shedding program anyway.

    Large customers with critical loads have already determined that local generation can support local demand and consumption, and have installed systems to carry their critical loads during grid outages. Your question regarding local support of the grid will likely not be answered until the utilities make it worthwhile for customers to use their capacity for this purpose and provide assurance that the massive swings possible on the grid will not damage the much lower capacity local systems.

  5. E-P,

    Most electric loads cannot be modulated, either locally or remotely, except by reducing voltage (brownouts). Unfortunately, brownouts can damage motors through overheating, thus causing economic damage to their owners. Therefore, loads are shed (turned off).

    I am aware of the existing technology for turning loads off during peak periods and the existing utility programs offering rate reductions in exchange for the right and ability to switch specific loads on and off to manage peak demand. These programs have achieved only limited success because they have typically offered only minor savings to the target consumers.

    Curtailable and interruptible contracts for large commercial and industrial customers have been more successful because the customers were able to achieve significant savings and also to make alternative arrangements to meet their needs during interruptions.

    The issue here is that First Energy would not have had the information necessary to trigger the load shedding. In addition, the load on the system at the time of the failure was not high enough to have triggered a load shedding program anyway.

    Large customers with critical loads have already determined that local generation can support local demand and consumption, and have installed systems to carry their critical loads during grid outages. Your question regarding local support of the grid will likely not be answered until the utilities make it worthwhile for customers to use their capacity for this purpose and provide assurance that the massive swings possible on the grid will not damage the much lower capacity local systems.

  6. In response to E-P’s last question:

    Research has been done looking at what might be called distributed dispatch. Google the term “Context Dependent Network Agent” to get some idea of what has been done.

  7. Mike:  Thanks for the pointer, I’ll try to read up on that when time permits (I have been very short of time this past month, no time to read much or blog anything).

    Ed:  In the context of the full load of the grid, being able to switch individual loads of 10 KW or less amounts to infinitesimal modulation.  Being able to cut the duty cycle of air conditioners on command (cut off compressors which have been running more than X% of their recent cycle time, delay the starting of all others) could be used to gain immediate relief from a supply-deficiency condition.  Being able to use all electric water heaters as dump loads is a possibility for handling brief oversupply conditions; if we can push the control systems down as far as domestic refrigerators, being able to cycle them on on command could accomplish the same end at the rate of perhaps 300 W/household (roughly 30% of average demand).  As I noted elsewhere, if we have a substantial number of electric or plug-in hybrid vehicles to use as storage buffers the possibilities multiply; the modulation capacity can go into the tens of gigawatts for a typical metropolitan area.

    There are all kinds of things we can do, the issues revolve around deciding what to do and what means to use to do it.  Compared to the prospects being held out for demand-side management two decades ago, it’s obvious that we have barely begun.  And I’m with Mike:  these measures will become attractive to consumers when consumers pay the true cost of peak power and reap the benefits of DSM, and not one minute sooner.

  8. Mike:  Thanks for the pointer, I’ll try to read up on that when time permits (I have been very short of time this past month, no time to read much or blog anything).

    Ed:  In the context of the full load of the grid, being able to switch individual loads of 10 KW or less amounts to infinitesimal modulation.  Being able to cut the duty cycle of air conditioners on command (cut off compressors which have been running more than X% of their recent cycle time, delay the starting of all others) could be used to gain immediate relief from a supply-deficiency condition.  Being able to use all electric water heaters as dump loads is a possibility for handling brief oversupply conditions; if we can push the control systems down as far as domestic refrigerators, being able to cycle them on on command could accomplish the same end at the rate of perhaps 300 W/household (roughly 30% of average demand).  As I noted elsewhere, if we have a substantial number of electric or plug-in hybrid vehicles to use as storage buffers the possibilities multiply; the modulation capacity can go into the tens of gigawatts for a typical metropolitan area.

    There are all kinds of things we can do, the issues revolve around deciding what to do and what means to use to do it.  Compared to the prospects being held out for demand-side management two decades ago, it’s obvious that we have barely begun.  And I’m with Mike:  these measures will become attractive to consumers when consumers pay the true cost of peak power and reap the benefits of DSM, and not one minute sooner.

Comments are closed.