?                     Market prices in regions, in particular the differences among prices, will give a ?dual? representation of the state of the system.

?                     To predict when prices between regions are likely to be different (when congestion is likely to occur), we need statistical information relating congestion to other variables, such as temperature or loads.

?                     Transmission congestion can then be modeled using a distribution of price differences that are correlated with the other variables.

            Does that make sense to anyone? Schilmoeller asked. It does if you can translate this outcome into actual results, rather than just words, one participant replied. What I'm hearing, then, is that we may want to refer at least a portion of this question to transmission experts, Schilmoeller said.

Following the meeting, Kurt Granat, PacifiCorp Transmission Engineer, provided insights into the expected relationship of congestion and other parameters.  He pointed out that scheduled loads, rather than actual flows, are a better indicator of congestion for economic modeling.  He agreed that loads, temperature, and East-side hydro generation should be among the best predictors of congestion.  Kurt also suggested that 90 percent of capacity should be used to define congestion, because there are several reasons why operators would avoid maximum loads on a path or intertie.

5. Representation of Resource Diversity.

            Schilmoeller began this portion of the presentation by saying that this is a section of the analysis he has thought less about, so he will be seeking the input of the group on this question. He touched on the benefits of resource diversity he has identified so far. Can you explain exactly what you mean by ?resource diversity?? one participant asked. What I meant when I prepared this section was distributed generation, not diversity, Schilmoeller said. You may want to change the title of this section, then, another participant suggested; Schilmoeller agreed. Various participants noted that ?diversity? can mean a variety of things ? resource size, fuel source, ownership, geographic location and other factors.

            In response to another question, Schilmoeller said the point of this segment of the analysis is to capture the reduced risk associated with distributed generation in the portfolio model ? basically, we want to quantify that risk mitigation capability, he said. Another participant suggested that it is important that this section of the analysis be as flexible as possible; it needs to include the ability to capture additional statistical distributions representing physical phenomena, fixed cost adjustments and future regulatory change, he said. Schilmoeller agreed that this would be a useful additional capability, adding that by the next meeting of this group, he hopes to develop a wizard to simplify the use of this tool. Jim Litchfield added that additional slides on specific distributed generating resources would be a useful addition to this section. We'll add that, so that we can talk about the specific representations for each of those resources, Schilmoeller replied.

6. Influence Diagram of Effects.

            Here we wanted to talk about the significance of the effects of, and the relationships between, the different independent variables we'll be looking at in the this analysis, Schilmoeller explained: resource outages, transmission congestion, hydro generation, resource margin, the market price of electricity, DSI loads, aluminum prices, fuel prices, non-DSI loads and temperature. Litchfield said that, in his view, hydro generation is not an independent variable; the system includes storage capacity, obviously, so generation is human-controlled. Bonneville has discretionary hydro generating capacity at any given moment, he observed. However, hydro generation output increases or decreases in response to outside variables, Schilmoeller replied.

            One proposal was to add a bubble for GDP, which would drive load, fuel prices, and so forth.  Terry Morlan suggested that there were enough other factors confounding the relationship among these that any influence of GDP is small.  Moreover, neither he nor anyone he knows has an explanatory model for relating GDP to these other factors.  GDP doesn't contribute enough to the model to offset the additional computational burden of including it.

            The group devoted a few minutes of discussion the relationship between loads and electricity prices.  Schilmoeller pointed out that Terry Morlan's Aluminum industry model will be incorporated into this modeling work, and it provides an explicit relationship between DSI loads, electricity prices, and aluminum prices.  There are both short-term and long-term load responses to electricity price, by both the Non-DSI and the DSI loads, several participants pointed out.

            Another factor influencing load is precipitation, one participant observed.  If precipitation goes up then the irrigation loads decrease substantially.

            Market prices are going to drive resource acquisitions, another participant noted ? there needs to be another bubble here, leading back to resources. In response, Schilmoeller said it would be possible to incorporate a price-driven resource addition function. A key function of this analysis is to tell us whether Resource Portfolio A is better than Resource Portfolio B -- an influence diagram, Litchfield observed. Shouldn't the resource selection and addition be part of this influence diagram?  This diagram is intended to reflect exogenous variables, Schilmoeller replied -- basically, all we're talking about is correlation factors rather than cause and effect (e.g., prices incent capacity addition). The prices that we are representing in the model are equilibrium prices, which incorporate all of the effects of capacity addition, etc.

7. Statistical Results for Natural Gas Prices, Electricity Prices, Load, Temperature, Aluminum Prices, Hydro, Transmission Congestion.

            Schilmoeller described some of the statistical data available for use in the model ? historical dailies, temperature through the year and the region, aluminum prices, hydro generation, electricity prices, hourly transmission and cut-plane congestion data and natural gas prices back to 1997. I have also uploaded all of the California ISO reports on forced outages dating back to 2000, Schilmoeller said, adding that he is open to any other sources of relevant statistical data the other SAAC participants might suggest.

            One thing that I have observed, Schilmoeller stated, is a correlation between electricity prices and gas prices, contrary to what several other researchers have observed.  Terry Morlan suggested that the Winter of 2000-2001 were removed from the data, the correlation may vanish.

            One thing we've talked about is any changes of state that occurs, Schilmoeller said ? there was the complete shift in the underlying relationship between variables that occurred during the recent California energy crisis, for example. It would be very interesting if we can capture that sort of rare and catastrophic event here in the model, he said. You might try creating a flat round-the-clock energy price, Jim noted; Schilmoeller agreed that this might be useful. Phil Sher added that the basis he used to calculate heating degree-days was 55 to 65.

 8. Next SAAC Meeting Date.

            The next meeting of the System Analysis Advisory Committee was set for Thursday, January 16. The agenda at this meeting will focus on remaining work on statistics, incentives for new generation, review of the risk management problems during 2000-2001 (what worked and what did not), and initial optimization for the region, using all mechanisms. Meeting summary prepared by Jeff Kuechle, NWPPC Contractor.