These are a few semi-organized thoughts that I hope might help people start thinking about the issue. I’d welcome responses or other takes on the problem before the workshop if anyone feels like it. Any format would be fine, including simply redlined comments in this document.
Cost and benefits of Demand Response (DR)
Demand response is in many respects in the position of conservation in the 1970s. That is, there is increasing evidence that DR is a worthwhile resource to pursue, but we are only beginning to learn how much may be available at what costs and how much benefit the power system receives from a particular DR action.
Costs: While we haven’t yet got the base of experience to construct supply curves of DR similar to the conservation supply curves we’re familiar with, I think estimating the costs of DR programs (putting aside pricing strategies for the moment, see below) is conceptually fairly simple. That is, the cost of DR to the power system is a fair approximation of the cost of DR to society, since voluntary participants of DR programs are presumably are no worse off for their participation. In fact, the cost to the power system most likely somewhat overstates the cost to society, since some participants are probably at least somewhat better off for their participation. The bottom line is that we need more experience in running DR programs to better understand power system costs of DR, but there aren’t any huge conceptual problems.
Benefits: The estimation of benefits poses more issues. I think most folks would agree (let me know if I’m wrong about this) that the appropriate principle to use is that of avoided cost. That is, the benefit of DR is the cost of serving the incremental load that is avoided by DR, evaluated at a specific time and place. That cost varies with the balance between supply and demand and will (usually) be highest at peak demand hours. Even with agreement on the principle, however, there a number of issues to be addressed before we have a clear methodology for estimation of benefits.
Short run/long run
In the short run: we simply operate the power system we’ve got. The avoided cost will be the variable operating cost of the incremental generator (if generating capacity is adequate) or the cost of curtailment (if the generating capacity is inadequate).
For the purpose of this discussion, I’m assuming we’re planning to build and operate a system that meets some standard of adequacy, and the choices we’re considering are between alternate means of meeting that standard. Consequently, I’m putting aside the problems of estimating costs of curtailment, and focusing on the costs of the incremental generator. The Council’s data suggest that the variable operating cost of incremental generation from our regional power system probably doesn’t go higher than about $80/MWh (e.g. 13,000 Btus/kWh and $6/million Btus for natural gas yields $78/MWh).
In the longer run: more costs are variable and fewer costs are fixed. We have the time to make adjustments to our power system, e.g. building or contracting for peaking capacity if necessary. Currently, such contracts appear to be available at costs that are one half or less of the fixed cost of building a new peaking generator (in the range of $25-45/kW-yr, compared to $70-100/kW-yr for a newly-built peaker). These added costs raise the cost of meeting peak loads by amounts inversely proportional to the number of hours the generator is dispatched (e.g. $25/kW-yr used for 100 hours/yr costs $250/MWh for the contract) making a total cost that can be $300/MWh or more.
I would argue that $25-45/kW-yr capacity contracts won’t be available forever, at least if we leave new construction to the market, and that eventually costs of capacity contracts will approach the cost of new builds. I think others might argue differently about the possible persistence of e.g. $25-45/kW-yr capacity contracts, and we should certainly have that discussion.
It could be that some utilities find themselves continually in a position to take advantage of others’ overbuilding (by choice or by mandate), but it seems to me that the incremental cost to the power system as a whole must include the full cost of new builds. The full cost of incremental power from new builds used 100 hours/yr would run in the neighborhood of $900/MWh (e.g. $85/kW-yr for 100 hours yields $850/MWh for fixed costs plus $50/MWh or more for variable costs).
The bottom line is the long run perspective leads to much higher avoided costs than the short run perspective. The short run perspective on avoided cost is perfectly valid for many decisions, but the Council is tasked with long run planning, and we focus on long run costs. I would expect utilities’ long term planning to take a similar long run perspective.
How many hours/year are we calling on DR?
The illustrative calculations above assume the incremental plant or contract (potentially displaced by DR) is dispatched 100 hours/year. There’s nothing magic about that assumption — I use it because it’s a round number and it illustrates the high cost of building generation that only runs a few hours a year. Our system has a lot of load that is served 100 hours or less/yr (depending on which years are included in the analysis, 6,000 to 7,000 MW). The critical question is "How much will an incremental generator be dispatched (see CA and Hydro below) to meet an increment in peak load?"
Non-generation avoided costs
Bonneville and others have been paying more attention to the transmission and distribution cost of serving incremental load. These costs are very site-specific and hard to pin down but in some cases can be quite significant. Targeted DR could avoid or defer significant T&D capacity costs in those cases. Counting only avoided generation costs in the benefits of DR understates those benefits by a little in some cases and a lot in others.
Sharing peak generating capacity with other regions (e.g. CA)
PNW utilities have shared capacity with California over the intertie for decades. This sharing has certainly reduced the total amount of generating capacity required to serve the two regions’ diverse loads, but how does that affect the avoided cost of DR? Even if we add the hourly loads of the two regions and construct a combined load duration curve (equivalent to assuming the intertie is never full), there are over 6,000 MW of load that need to be served less than 100 hours/year. As above, the critical question is "How much will the incremental generator be dispatched to meet an increment in peak load?"
Serving peak loads with hydro
The hydroelectric system has provided the Pacific Northwest with a large peaking capacity for many years. But as loads have grown beyond the energy capability of the hydro system we’ve added thermal generation, and we’ve constrained some operations of the hydro system to meet fish and wildlife needs. As a result, our power system is becoming more like power systems everywhere else in the nation. Where are we at the margin? Peaking generators have been built in the Northwest in the past few years, so their costs are a reasonable starting place for calculation of avoided costs of DR. However, at least at the Council we don’t feel we have been able to model the interaction of thermal peakers with the hydro system as well as we’d like. I’d welcome ideas in this area.
What about pricing strategies?
In an economist’s perfect world we could provide efficient price signals to consumers and they would respond appropriately and the question of cost-effective DR just about goes away, since however they responded would be "cost-effective." The only cost I can think of to the power system of that strategy would be for metering.
This issue is kind of academic for the present, since a comprehensive pricing strategy looks pretty unlikely to be gain much acceptance anytime soon. The Council’s Fifth Power Plan proposes a discussion of ways to make pricing strategies more acceptable, and I hope that’s a subject we can take up in a future workshop.