APPENDIX ENERGY MODELING FORUM SENIOR ADVISORY PANEL Mr. Charles J. Hitch (Chairman) Dr. Philip Abelson Editor, Science Dr. Harvey Brooks Professor, Harvard University Mr. David Cohen President, Common Cause Mr. Gordon R. Corey Vice Chairman, Commonwealth Edison Dr. Floyd L. Culler, Jr. President, Electric Power Research Institute Mr. Charles Di Bona President, American Petroleum Institute Mr. Herman M. Dieckamp President, General Public Utilities Service Corporation The Honorable John D. Dingell Member, United States House of Representatives The Honorable Joseph L. Fisher Member, United States House of Representatives The Honorable William P. Hobby Mr. Jack K. Horton Chairman, Southern California Edison Company Mr. W.F. Kieschnick, Jr. Vice Chairman, Atlantic Richfield Company Dr. Henry R. Linden President, Gas Research Institute Mr. Guy W. Nichols President, New England Electric System Mr. John F. O'Leary Deputy Secretary, United States Department of Energy Dr. Alan Pasternack Member, California Energy Commission Dr. John Sawhill President, New York University Dr. Chauncey Starr Vice Chairman, Electric Power Research Institute The Honorable Morris K. Udall Member, United States House of Representatives ENERGY MODELING FORUM Adam Borison Research Assistant John Lindsay Bower Research Assistant Patrick Coene Research Assistant Wendelin Dintersmith Administrative Assistant Darylin Druhe Secretary Mark Edgerton Research Associate Kathleen Favor Research Assistant Elizabeth Heck Secretary Charles Kolstad Research Assistant Douglas Logan Research Assistant Susan Missner Research Associate Zakia Rahman Research Assistant Dorothy Sheffield Research Associate Pamela Sherby Secretary Nancy Silvis Editorial Assistant James Sweeney Executive Director John Weyant Associate Director Thomas A. Wilson Research Assistant ELECTRIC LOAD FORECASTING: PROBING THE ISSUES WITH MODELS Bernard H. Cherry Vice President Working Group Three of the Energy Modeling Forum was convened in December of 1977 in order to evaluate the approaches utilized for the forecasting of electric energy demand in several utilities across the country. The final report of this activity was published in the spring of 1979. After the selection of load forecasting as a topic for the third working group, potential participants were identified. There was an attempt to get a cross section of participants in the utility sector in terms of the type of service area, geographical location, and the economy of the service area. There was no attempt to pick the best ten electric tility load forecasting approaches, but rather to randomly select a number of utilities across the country. Invitations were issued to about 25 utilities. Invitations were also forwarded to university odelers active in energy modeling and finally to a number of state and federal regulators active in the field. The ten models which were ltimately used in the load forecasting study are shown in Figure 1. The initial meeting of the group was aimed at trying to define a umber of issues: (1) what to evaluate in the study; (2) how to comare the results of these evaluations; and (3) how to select the cenarios for evaluation. There was substantial concern about the ltimate use of a comparison of various utility and university odels. There was concern particularly among the utility particiants of the perception of study results by one or more of their espective regulatory agencies. While this might be viewed as an nnecessary concern, those who are interested in the validation and omparison of models will have to come to grips with this very real oncern. he Working Group meetings (the first two in particular) resembled, sort of encounter session. There were very strong views and opinions eld by a large number of the participants in the sessions, and the esolutions and accommodation of some rather divergent points of view, ecame a very difficult and tedious task for all of us. However, it s to the credit of all of the members of the Working Group, that the ey issues were resolved. The difficulties which were faced in the arly meetings can be highlighted by the problems associated with he selection of a title for the study: Electric Load Forecasting: robing The Issues With Models. The word "Probing", in particular, as one which required a good deal of negotiation to gain ultimate greement. There was a resistance in using any words which conveyed more quantitative comparison than "Probing". Rejected were words like evaluating, comparing, and investigating. The next issue was to decide exactly how the evaluation could go forward. The key was the definition of a base or reference case for each modeler. It was ultimately agreed that the most efficient approach was to use each modeler's current planning scenario as a base case with certain adjustments relating to time of day pricing and appliance efficiency standard effects removed. This provided a relatively quick opportunity to run calculations based upon input which already included the best judgements of all of the modelers. In the course of the discussions, it was agreed that it would be extremely time consuming to start from scratch and construct an artificial set of data, and have each of the participating modelers attempt to run some standard computation. While it was recognized that there were deficiencies in the approach chosen, particularly relating to an explicit understanding of the judgements which went int the base data, the participants agreed that this was the most efficien way of proceeding. The selection of the scenarios to be evaluated beyond the reference was the subject of an entire meeting and after discussion of a large number of issues which could be important in the future, about a dozen were selected as being worthy of future evaluation. Of those, seven were viewed as being able to be evaluated by some of the models which were being investigated. The scenarios which were ultimately evaluated in the forecasting study are shown in Figure 2. Figure 3 indicates that while at least one model was able to evaluate every scenario, not all models were able to evaluate all scenarios. It should be noted in Figure 3 that some modelers did not evaluate all scenarios due to personnel availability and time constraints. Two major objectives were met in the EMF 3 Working Group. First, the experiments identified and illuminated key forecasting issues, and second, the interactions among the model developers and users improved the understanding of the model's capabilities and limitations. The second objective, in the author's view, was most important in this evaluation, inasmuch as there was little formal communication between model developers and users in various utilities prior to the convening of this working group. The major findings of the analysis and the discussions of the group are shown below: Future electricity consumption, given the assumptions Load shape, as well as peak demand and electricity con- |