From nilan@cat.syr.edu Mon Apr 18 15:39:08 1994 Date: Mon, 18 Apr 1994 15:16:17 -0400 (EDT) From: "Michael S. Nilan" To: paulc@npac Subject: some helpful NIMO text Paul: Here is some text relating to the "meta" lessons we learned from talking to Larry Eng at NIMO Planning, Bill Merritt and Wes Yeomans (who is no longer at NIMO) at NIMO Operations, the New York Power Pool and the New York Power Authority. Please remember that the way R & D works in NIMO is that they need an internal client (e.g., Planning or Operations or Nuclear) who will champion a particular project. It would appear as if this stability analysis project was handed to us either as a test of our ability to really find out what was wrong (which everyone seems to feel is the basic models and assumptions behind EMS) or else that stability analysis was an obvious choice for parallelism and no one at NMPC really thought about it. At any rate, our report should make the people in Planning and Operations look as rational as possible. It is still unclear to me who the internal client for this project actually was; at various times we were "pointed" towards both Planning and Operations. So: --------------------------------------------------------------------------- For Phase II, we were primarily involved in interpreting the results of our knowledge acquisition data from Phase I for the expert system and in examining NMPC's and other power parties' (NYPP, NYPA) evaluation of existing and emerging energy management systems (EMS). For example, EPRI is in the process of developing a new integrated EMS. We have a good feel for how that will be received by the New York-based parties, i.e., a mixed reaction. They all agree that a new EMS is appropriate and is needed but expressed reservations about the extent to which it will actually improve energy management decision making. We also met and/or talked with various individuals in NMPC and NYPP trying to get a more coherent picture of NMPC's needs and the larger electrical power picture in the Northeast. Richard also worked with Siva on the interface aspects of the prototype expert system. Operations people, in particular, see a great potential for an "on demand" stability analysis being useful in the tactical process of managing the network under fault conditions. For example, if they could compute the ACTUAL limit of an interface (as opposed to the conservative, hypothetical and static limits set by the NYPP), they would be better able to make decisions on how to respond to decreasing loads without hurting NMPC's profits. Estimates of how much this could effect the bottom line in NMPC ranged from "significant" to "astronomical." But, the problem is that the models and assumptions used by NYPP to run stability analyses differs considerably from the models and assumptions employed by operations (see the discussion on models and assumptions below). Overall, I think that the work done by IST and NPAC on the stability analysis is a good example of how parallel processing per se can increase the efficiency of certain data processing needs of NMPC and expert systems can reduce the amount of time required for interpretation of the output. As a demonstration then of our ability to improve certain activities, it has achieved its purpose - it is clear that we can perform this kind of work effectively. However, I do not believe that we had an adequate understanding of NMPC's needs before the stability analysis application was decided upon. As a result of working on this problem, I feel that there were other valuable insights gained as a result of our activity on this project. The primary reaction that we got from all of the people we talked to inside and outside NMPC was that speeding up the stability analysis to run in minutes rather than over night was a valuable improvement. However, the idea of an on-line contingency analysis was seen as a concern of more importance. Doubts were raised at NYPP in particular (but also by EPRI), about the extent to which contingency analysis was conducive to parallel analysis. Richard and I did begin collecting data on contingency analysis during Phase I of this project but suspended it when NMPC decided to focus exclusively on the stability analysis problem. So, we have a preliminary knowledge acquisition model on that decision process, but it is not really ready for sharing with a client. We need to look at the algorit hm more closely to see if parallelism will enhance its efficiency. The major comment by all the people we talked to was the weakness inherent in the models that are used to run the stability analysis (including contingency definition and selection). The "models" include capacities of all major power network components and assumptions about consequences of faults in those components. It would seem as if the models were established quite a while ago (e.g., fifteen years or longer) and differ from one power utility to another. No one we talked to expressed any optimism about the extent to which the models actually represent the capacities or tolerances of that actual network components. Also, the logic of selecting and ranking contingencies was seen as an activity that was based on archaic, serial processing models and had not really received a great deal of thought by anyone in the Northeast in recent history. So, the basic reaction of the utility people to our efforts with stability analysis is based more upon their lack of confidence in the underlying models and assumptions about EMS in general. It would seem that they are somewhat pessimistic because what is needed is to re-examine the underlying models BEFORE we start speeding up algorithms. Unfortunately, the models and assumptions used by each power utility are different from each other. It is clear that the knowledge about how to update the models and assumptions is not the problem but that getting everyone to do it is. I do not think we can advise NMPC to revise their models and assumptions in isolation. We could suggest however, that NMPC take the lead on trying to encourage the entire Northeast (if not the USA & Canada) to realize this objective. It is clear from our perspective that continuing with improvements in NMPC's existing EMS would be inappropriate unless at least NMPC reworks their models and assumptions. One impetus for the utilities of addressing this objective is the 1992 Energy Policy Act which gives independent power companies access to the power network. If the existing public utilities are going to be responsible (largely by default) for managing the networks, then it would be to their advantage to have a more coherent and valid set of models and assumptions not only for their management responsibilities but also for maximizing the profitability of their own power generation. I hope we can make the case that we demonstrated our abilities; that we have identified the "real" problem behind inefficient EMS functioning (i.e., conceptual and computing models and assumptions that need updating); and that together, NPAC, ECE and IST could facilitate NMPC in addressing these historical problems (i.e., the problems aren't a product of mismanagement or inappropriate system design but rather information and computing technology has advanced enough so that it simply needs to be done again). Our opinion is that Operations represents the best internal client for revamping the EMS but we would also need support from NMPC's corporate level in order to address the models and assumptions problem. One final note is that the folks at NYPP were actually very positive about lending support to whoever wanted to begin to address the models and assumptions - we just never got back to them because we couldn't get a letter of support from NMPC. --------------------------------------- So, I hope this helps in the final report. Let me know if you want more content, more details on any specific area, or other help in getting this done. Mike Michael S. Nilan School of Information Studies 4-206 Center for Science & Technology Syracuse University Syracuse, NY 13244-4100 Phone: (315) 443-2911 FAX: (315) 443-5806