1.                  We need a way to quickly estimate the dispatch factor for thermal generation, so that we can calculate variable cost.

2.                  Should have certain basic properties

3.                  If average monthly prices ($/Mwh) for gas are about the same as average monthly prices for electricity, the dispatch factor should be about 50 percent

4.                  If average monthly prices ($/Mwh) for gas are well above the average monthly prices for electricity, but there is a good deal of uncertainty in the prices, the plant should dispatch, albeit a small amount.

5.                  If average monthly prices ($/Mwh) for gas are well below the average monthly price for electricity, but there is a good deal of uncertainty about the prices, the plant should run close to, but not quite 100% capacity factor (disregarding maintenance and forced outage)

            So the resources you?re assuming here do not affect the price of electricity? Michael McCoy asked. Correct, Schilmoeller replied ? we're modeling an open system, rather than a closed system. We will use tools like AURORA and GENESYS to inform us about what those cost factors might be, he said. So you?re assuming that there is little correlation between the price of thermal generation and the price of electricity? a BPA representative asked. Correct, Schilmoeller replied. So there is some residual amount of generation being left for the daily market? the BPA representative asked. The model assumes that power plants are dispatched for spot prices, Schilmoeller replied. We can represent those types of commitments, but there will also be contracts for managing financial risk, he said. And this model can address thermal resources other than gas, such as coal? another participant asked. Yes, Schilmoeller replied ? we should be able to disaggregate those plants into what is committable and what is not. You don't break that out, currently? the participant asked. Correct, said Schilmoeller.

            Next, Schilmoeller put up a sample analysis in the form of a graph, titled ?Typical Dispatchable? ? example of 1 MW single-cycle combustion turbine (no dispatch constraints), Natural gas price: $3.33/MBTU, Heat rate: 9,000 BTU/kWH) etc. He then proceeded onward through a series of slides illustrating other aspects of how the portfolio model deals with thermal generation, including the Price Duration Curve, variability viewed as Cumulative Probability Density Function (CDF), and various potential simplifications to this facet of the analysis.

            Some have suggested that ramping periods have higher volatility than peak periods, said McCoy ? that creates a small issue you might want to give some thought to. Good suggestion, said Schilmoeller.

            The group devoted a few minutes of discussion to the technical nuances of Simplifications 1 and 2, with Schilmoeller fielding a variety of detailed questions and comments. Ultimately, he said he welcomes such discussion, and invited the other SAAC participants to contact him outside of today's meeting if they have further concerns.

            In response to a question from Oliver, Schilmoeller reiterated that the portfolio model is intended to provide only a snapshot in time, not a time-sequence of prices. Next, he put up slides titled ?Almost There...? ?Option Delta: and ?The Payoff (equation)?

             However, said Schilmoeller, two issues remain: gas prices are not constant (the strike price X is not fixed). Also, most of what we may think we know about future price uncertainty might be expressed in terms of average monthly prices. The solution, he said, is to

1.                  Use a European ?spread? option instead of a standard European call option. Also,

2.                  Try to estimate the volatility of the hourly spread from the monthly volatilities and correlations.

            After a break, Schilmoeller returned to the topic of Geometric Brownian Motion (GBM), discussing the probability distribution it generates with respect to future pricing. The group devoted a few minutes of discussion to the technical nuances of this portion of the model.

            Schilmoeller then moved on to the topics of ?European Spread Option? and ?Hourly Volatiles from Monthly.? He finished this portion of the agenda with an example capacity factor calculation employing the principals laid out over the course of the morning.

            Essentially, he said, we want a dispatch that gives us a capacity factor of about 50%; you?re never going to dispatch 100% of the time. We commandeered some of the algebra in the Black-Scholes model, setting r at zero, for example. We'll dig into this further at our next meeting, when we will be discussing electricity prices, Schilmoeller said. Does everyone understand the logic that went into this model? he asked. So far, was the reply.

5. Metrics.

            This will be more of a free-form discussion, Schilmoeller said; essentially, I would like the help of this group in developing a risk metric for the region. In my view, this metric should be minimum total power cost, driven by a CVaR conversion.

            Schilmoeller touched on stakeholder issues, including a list of proposed stakeholders and a proposed stakeholder perspective. He then moved on to a list of potential metric candidates, including value at risk (VaR), standard deviation, expected shortfall, conditional VaR (CVaR), Van Neumann utility functions and block maxima. Next, Schilmoeller touched on a series of graphs illustrating how these various metrics work.

            The group devoted a few minutes of discussion to these metrics, debating their coherence, distributions and other factors. Schilmoeller said that, in his opinion, CVaR and block maxima offer the most potential upside; the people at Crystal Ball have told him that it should be possible to incorporate CVaR into the Crystal Ball computer tool.

            What about seasonal exchanges, and streamflow ?trenches?? asked McCoy. Selling deep in a money call is another way to sell off the upside ? I agree, said Schilmoeller. I'm concerned that a lot of this gets back to the point that we have random prices here, said the BPA representative ? many of these CVaR assumptions make sense only if you assume random prices. It is a bit confusing, another participant noted. The group also discussed the current status of the energy market, and its impact on the feasibility of this approach. One participant noted that the energy trading market is currently at its lowest ebb in years; however, two years from now, the market will almost certainly be substantially different.

            Schilmoeller noted that this entire presentation is available via the SAAR website, including the final slide from this section, titled ?Conclusions.?

 6. Representations In the Portfolio Model ? Plan Issues, Price Responsive Demand.

            This topic was not explicitly discussed at today's meeting.

7. Next SAAC Meeting Date.

            The next meeting of the System Analysis Advisory Committee was set for Friday, November 22, beginning at 9:30; the main topics discussed at this meeting will be electricity  prices and hydrogeneration. Schilmoeller asked that, if any of the other SAAC participants know of good sources of hourly electricity prices, that they provide that information to him. Meeting summary prepared by Jeff Kuechle, NWPPC contractor.