Ken Corum, Council, July 22, 2005

This issue has come up more than once, and I don’t feel like I understand it very well.  When I’ve talked to utility people about it I’ve generally felt frustrated because I didn’t digest the new information I got fast enough to think of the follow-on questions that I later wished I’d asked.  I thought if I wrote everything down it might force me to see the issues, and frame my questions, more clearly. 

I know that I don’t really understand what traders do, so I’m open to instruction by anybody who does.  Remedying that ignorance could go a long way toward making everything more sensible.  But I also suspect that traders and planners have different perspectives, and that if we better understand what the differences are, we’ll have a better understanding of the potential value of DR.

My understanding is that traders generally think that DR is expensive, and too expensive to play a very significant part in resource plans.  Yet Pacificorp’s planning analysis concludes that construction of new generators is deferred by access to DR, and total PVRR is reduced — that is, DR is less expensive than the alternatives.  How should we interpret these two views?

1. One take could be that we (The PNW and the west coast as a whole) are currently somewhat surplus, so that the cost of meeting peak hours to the power system is the variable cost of operation of the marginal generator, which should drive the spot market price of peak power.  Compared to these costs, most DR is pretty expensive, and nobody would advocate calling on lots of DR under these circumstances.  (Pilot programs, where we’re testing and refining the resource for potential significant deployment in the future, still make sense.) 

This explanation reduces to “traders trade in the short run, planners plan in the long run” and the costs avoided by DR are different in the short run and long run.  There is no real disagreement between the two perspectives.  The planner’s response to “DR is too expensive.” is “Yes, it is now, but that changes when we consider building new generators in the long run, which increases avoided costs.”

2. Another take is traders know about the long run, and have price curves that enable them to acquire power 5-10 years in the future, and those curves indicate that DR that costs more than e.g. $150/MWh is too expensive.  The question that occurs to me in that case is, “Who is going to build the new generation to provide peak power at those prices?”  A peaking generator has to run more than 400 hours a year to cover its total costs out of sales at $150/MWh.  But there are several thousand MW of PNW load that have to be served less than 400 hours/year.  The generators to serve that increment of load won’t be paid for out of energy market revenues, but a price curve that doesn’t go over $150/MWh implies that they get built (and paid for) somehow — how?

Perhaps the costs of those seldom-used generators don’t get covered entirely by sales into the energy market, but also partly by a separate mechanism such as a capacity credit.  If so, to say that DR is too expensive if it costs more than $150/MWh is incorrect, because (in the long run at least) DR avoids not only the market cost of energy; it also avoids the cost of the capacity credit or whatever other mechanism helps pay for the incremental generator.  We can understand how an energy trader might not treat the capacity credit as a cost, but the planner must include it in long run costs.  In this case DR that looks too expensive based on energy market savings is cost effective when all the avoided costs are counted.

In such a case, DR should be acquired as long as its all-in costs are less than those of a new generator, and in the short run the DR should be dispatched against market prices like a generator, based on its variable costs.  The acquisition of this DR could result in higher peak market prices than if peak demand is served entirely by generation, but total costs (including capacity credits or whatever) will be less.

3. A third take is another variation of “DR avoids more than what traders trade.”  This is speculative, because as I said above, I don’t really understand what traders do, but if, for example, traders deal in blocks of on-peak and off-peak energy, system operators also have to shape those blocks to match the hourly demand that consumers actually bring to the system.  That shaping has a cost, either an explicit cost or an opportunity cost.  The total cost of meeting the highest hour’s demand out of each on-peak block of energy will be significantly higher than the market quote for the block as a whole.  Using DR can avoid these highest hour’s costs, raising the cost at which DR becomes non-cost-effective. 

I’m probably wrong in the specifics here, but the important question is: “Are the costs that traders see good measures of the total costs that can be avoided by DR?”

I’m sure there are holes in my understanding and/or logic here, and I’d appreciate someone pointing them out.  If we can at least get this issue clearly defined, I’m thinking it’s worth talking about at the next DR workshop, because I’m sure that Pacificorp is not the only utility where DR has been questioned by traders.