Energy arrow Demand Response arrow Archive

Strawman Proposal for Cost Effectiveness for Demand Response

This strawman proposal is a step in the development of recommendations for the evaluation of cost effectiveness of demand response (DR).  We hope that the strawman will help focus discussion on issues on which there is not yet agreement. 

Criteria for Cost Effectiveness

The basic criterion for cost effectiveness is, “Does inclusion of DR in a power system avoid more cost than it incurs, while not compromising system reliability[1] nor increasing system financial risk?”  In principle an improvement in system reliability or reduction in financial risk (without increasing system cost) would also be justification for DR.  With time to develop confidence in the analysis, the criterion for cost effectiveness may be expanded to include changes in system reliability and financial risk.

Ideally the determination of cost effectiveness of resources would be an implicit product of a comprehensive long term planning process.  The process would evaluate the system costs of all potential resource combinations, and all of the resources in the least-cost combination would be cost effective.  If such an analysis includes DR on an equivalent basis to other resources, the DR included in the least-cost resource portfolio is cost effective.  A comprehensive long term planning process is also a natural approach to a comparison of risks of alternative portfolios.  When possible, this is the preferred approach to determining the cost effectiveness of demand response.

Often, however, a demand response program is evaluated as an alternative to elements of a “conventional” resource portfolio, and its cost effectiveness must be estimated explicitly.  In practice that comparison usually amounts to weighing 1) the cost of achieving DR (e.g. administration, incentives and metering and communications equipment) against 2) the cost avoided by DR (e.g. reduced requirements for generation, transmission and distribution).  Before taking up issues specific to each of these costs, it’s worth making some points applicable to both categories.

Perspective

Our objective is to provide guidance to the Council, utility regulators, and utility planners.  For these parties, the appropriate perspective is that of society as a whole[2], over the long run.  Therefore, analysis must take into account the full cost of new facilities that will be needed over the long term.  The long run may well include periods of over- or under-capacity, so analysis also needs to take into account the short-term decisions involved in operating the power system (including participating in the wholesale energy market), if those conditions occur. 

We’re aware of points of debate about the differences between “societal costs” and “total resource costs.”  In the interest of streamlining discussion, this proposal assumes that total resource costs are a good approximation of societal costs for the time being. 

Level of Reliability

Our proposed approach is to compare alternate strategies that provide the same level of reliability.  This simplification avoids the difficulty of putting value on incremental changes in reliability.  However, it sets aside some possible interactions between DR and reliability, which are worth keeping in mind:  

  1. It may be that DR offers incremental increases in reliability at lower cost than conventional resources, and that taking this into account would change the chosen level of reliability. 
  2. It may also be that the acquisition of DR allows us to avoid involuntary curtailments in a short run situation when the power system’s conventional resources are inadequate.  This situation would be evidence of a failure of long term planning, but it has occurred (e.g. 2000-2001).  To the extent that DR has a shorter lead-time than conventional resources, it should receive credit in an evaluation that takes uncertainty into account.

Type of Demand Response Addressed

This proposal concentrates on one of the two major categories of demand response -- compensated reductions.  Compensated reductions include interruptible contracts and buyback programs, and the societal cost of achieving demand response by these approaches is fairly straightforward.  Customers who participate in these programs can be assumed to be no worse off for their participation, since they choose to participate in exchange for compensation.  In these situations, utilities’ costs of achieving demand response through compensated reductions are a good approximation[3] of societal costs.

The proposal does not attempt to deal with the cost effectiveness of pricing strategies.   Pricing strategies offer much promise, but estimating the cost of achieving demand response is not straightforward.  If customers are not compensated for higher bills and/or changed service, there is no convenient measure of that component of societal cost.  These difficulties don’t rule out pricing strategies as attractive approaches to demand response, but they do mean that the rationale for such strategies will be different than a simple cost effectiveness measure. 

Cost of Achieving Demand Response

Costs of DR will vary in amount and structure, depending on the unique circumstances of each program.  Without trying to anticipate the specifics of all DR program possibilities, some principles should be followed in estimating costs:

  1. When a program has non-DR benefits, their value should be taken into account.  In many cases, the most straightforward way to do that will be to subtract non-DR benefits from the programs costs, leaving DR’s net costs to be compared to the power system costs avoided by DR.  The cost of more sophisticated meters that permit air conditioning cycling programs is a good example--the meters also reduce billing cost and provide other benefits that must be recognized when estimating the net cost of AC cycling as a DR program.
  2. The structure of a program’s costs should be reflected in the analysis of DR’s effects on system operation.  For example, buyback programs tend to have low fixed costs and higher variable costs, which results in low costs in years when DR is less-needed due to mild weather or good precipitation for the hydroelectric system.  The cost structure of DR programs will help determine DR’s affect on risk (see below).

Cost Avoided by Demand Response

The simplest method of estimating avoided costs is the calculation of the fully-allocated cost of a new peaking generator that is dispatched a few hours a year.  A more comprehensive estimation should take into account the interaction of new peaking generators with the existing power system, wholesale energy market trade between our region and others, and various kinds of risk such as variable hydroelectric output and variation in wholesale prices of electricity.  In general the more comprehensive the analysis the better, but even if practical considerations limit the comprehensiveness of the analysis, there are other important principles to guide avoided costs estimation:

  1. As advocated above, the long term societal perspective should be adopted.
  2. The avoided cost estimates should reflect the specific qualities of the DR program analyzed.  This requires the analysis of each DR program’s impact on system resources and operation.  For example. AC cycling programs reduce loads in the summer and are probably dispatched as often as the contract with the customer allows.  In contrast, day-ahead buyback programs could be exercised at any time of year that prices or other circumstances warrant, but the amount of response on any given day will vary.

Risk

We’ve become increasingly concerned with risk in power system planning.  Changing resource portfolios by including DR affects both cost and risks.  A comprehensive evaluation of avoided cost of DR should take risk into account, then compare costs of portfolios that maintain risk at comparable levels (reliability another dimension of risk to be maintained).  Sources of risk to be taken into account, if possible:

  1. Uncertainty regarding underlying economic growth
  2. Variation in weather
  3. Variation in wholesale electricity prices
  4. Variation in generating fuel prices (e.g. natural gas)
  5. Variation in output of generating resources, as a result of plant outages or variation in precipitation
  6. Uncertainty regarding trading partners’ resource development plans

This is a Strawman

The point of this proposal is to stimulate and focus discussion.  Every component of the proposal is fair game for testing and discussion.

[1] Measures of reliability include loss-of-load probability (LOLP) and expected unserved energy (EUE).

[2] Utility planners have to consider both societal costs (to satisfy their regulators’ concerns) and the perspective of their utility as well.

[3] There may be some secondary effects not captured (e.g. employees of businesses that participate in a buy back program).