AMERICAN STATISTICAL ASSOCIATION SPRING 2007 MEETING OF THE AMERICAN STATISTICAL ASSOCIATION COMMITTEE ON ENERGY STATISTICS WITH THE ENERGY INFORMATION ADMINISTRATION DAY 1 of 2 Washington, D.C. Thursday, April 19, 2007 2 1 PARTICIPANTS: 2 COMMITTEE ON ENERGY STATISTICS: 3 NAGARAJ K. NEERCHAL Department of Mathematics and Statistics 4 University of Maryland 5 DEREK BINGHAM Simon Fraser University 6 EDWARD A. BLAIR 7 University of Houston 8 CUTLER CLEVELAND Center for Energy and Environmental Studies 9 JAE EDMONDS 10 Pacific Northwest National Laboratory 11 MOSHE FEDER Research Triangle Institute International 12 BARBARA FORSYTH 13 Westat 14 WALTER W. HILL St. Mary's College of Maryland 15 NEHA KHANNA 16 Department of Economics Binghamton University 17 EDWARD KOKKELENBERG 18 19 PRESENTERS: 20 PHILLIP TSENG HOWARD BRADSHER-FREDRICK 21 ROBERT SCHNAPP BOB RUTCHIK 22 RUEY-PYNG LU MARIE LARIVIERE BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 3 1 OTHER PARTICIPANTS: 2 WILLIAM (BILL) WEINIG ALETHEA JOHNSON 3 ANTHONY RADICH CAROL BLUMBERG 4 MARK GIELECKI LAWRENCE STROUD 5 LOUISE GUEY-LEE GUY CARUSO 6 HOWARD GRUENSPECHT NANCY KIRKENDALL 7 BOB SMITH TANCRED LIDDERDALE 8 MIKE CONNER SUSAN HARRIS 9 WILLIAM (BILL) HELKIE STEVE EBEL 10 11 12 13 14 15 16 17 18 19 20 21 22 BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 4 1 P R O C E E D I N G S 2 (8:15 a.m.) 3 MR. NEERCHAL: I think we should 4 get started. First of all, welcome you all 5 to the ASA Committee on Energy Statistics. I 6 want to emphasize that this is an ASA 7 Committee, it is not an EIA Committee, which 8 is an advisory body to EIA. And EIA is free 9 to choose to take that advice whenever it 10 thinks it is appropriate. 11 I'm Nagaraj Neerchal; I'm the chair 12 of this Committee. This is my first chair 13 job here, I think. But I had some beautiful 14 experiences -- 15 MR. CLEVELAND: And you're doing a 16 great job. 17 MR. NEERCHAL: Pardon me? 18 MR. CLEVELAND: You're doing a 19 great job. 20 MR. NEERCHAL: Thank you. It's the 21 easiest job in the Committee, actually, 22 because you don't need to do any discussions BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 5 1 and -- 2 (Laughter) 3 MS. FORSYTH: I think we should 4 change that this year. 5 MR. NEERCHAL: So -- 6 MS. KIRKENDALL: I actually thought 7 we stuck you with the discussion. 8 MR. NEERCHAL: I think you may have 9 it, yes. The meeting is open to public, 10 public comments are welcome. Time will be 11 set aside at the end of each session inviting 12 public to make their comments. And there are 13 two microphones there. We would like you to 14 come to the microphone and make the comments 15 so that the transcription picks it up. And 16 whenever you have something to say, please 17 announce your name and affiliation, and then 18 make your comments. 19 All attendees -- I think that 20 includes everybody, including the EIA 21 employees, should sign the register in the 22 hall as you're entering in, and put down your BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 6 1 e-mail address. I think if you're new to 2 this room and this floor, the restrooms are 3 at the two ends. So you go out of the room 4 and either turn right or left. And there's a 5 drinking fountain on both sides. 6 We have Chrissy Campbell (?) and 7 Donna Arrington (?), two people from the ASA 8 meetings department here helping us. And if 9 you have any questions regarding the expense 10 reimbursements or any logistics issues, 11 please contact them, they'll be helping you. 12 So we do have a new transcriber by 13 the name of Christine, but she is asking me 14 to remind you that if any are speaking, 15 please bend over and this is not a mouse pad. 16 COURT REPORTER: Just project. 17 MR. CLEVELAND: We can't take these 18 with us? 19 SPEAKER: Actually, you can. 20 MR. CLEVELAND: I thought it was a 21 perk of some sort of -- 22 MS. KIRKENDALL: Yeah, it's a perk. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 7 1 Alethia thought that they would be a nice 2 token. And so -- they are a perk actually. 3 MR. NEERCHAL: But make sure that 4 you're close to the microphone, not too far 5 away from it. No, don't lean back and talk. 6 We have two new members: Ed Blair 7 from University of Houston, Ed Kokkelenberg 8 from Binghamton University, and John Weyant, 9 who is not here yet. When he comes, we'll 10 give him a warm welcome. I would like us all 11 to welcome our new members. Thank you for 12 agreeing to -- 13 (Applause) 14 MR. NEERCHAL: We have two people 15 helping from the EIA side this time, I think: 16 Bill and Alethia. And I think some of you 17 already corresponded. 18 So the EIA type of questions, 19 something you can direct to them, and 20 Alethia, when Bill was away on vacation -- 21 (Laughter) 22 MR. NEERCHAL: On his surgery, BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 8 1 Alethia took care of the EIA liaison duties 2 and did a fantastic job. Thank you very 3 much. And they're responsible for putting 4 together this program and things like that 5 and making -- make sure that the chair looks 6 good and stuff like that, so it's a very 7 important job. 8 And we have Lawrence, who is 9 helping us with -- right, yeah, helping us 10 with the audiovisual thing. And the people 11 who are talking, they will get a microphone 12 if you want, or if they can stand at the 13 lectern and speak, whatever they want to. 14 SPEAKER: Okay. 15 MR. NEERCHAL: And once again, I'm 16 Nagaraj Neerchal, and I'm the chair. And if 17 you don't know where -- who to go to, just 18 come to me, I'll find out the person who can 19 help you. I won't help you necessarily. 20 All right, so I think we should 21 probably start with the introduction around 22 the room, including the audience members. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 9 1 I'm Nagaraj Neerchal once again. And I'm 2 from UMBC. I'm the chair of the Math and 3 Stat Department since July. 4 MS. KIRKENDALL: I'm Nancy 5 Kirkendall; I'm the Director of the 6 Statistics and Methods Group, and 7 orchestrater of this event. 8 MR. CARUSO: Guy Caruso, 9 Administrator of the Energy Information 10 Administration. 11 MR. KOKKELENBERG: I'm Edward 12 Kokkelenberg, Professor at SUNY Binghamton. 13 MS. KHANNA: Neha Khanna, State 14 University of New York, Department of 15 Economics. 16 MR. CLEVELAND: Cutler Cleveland, 17 Washington University, Professor in the 18 Department of Geography and Environment. 19 MR. HILL: I'm Walter Hill, St. 20 Mary's College of Maryland; currently the 21 chair of the Political Science department. 22 MS. FORSYTH: Barbara Forsyth, I'm BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 10 1 in Westat for another week. 2 (Laughter) 3 MR. BLAIR: Ed Blair. I'm at the 4 University of Houston. 5 MR. BINGHAM: Derek Bingham, Simon 6 Fraser University, Department of Actuarial 7 Science and Statistics. 8 MR. STROUD: Lawrence Stroud, 9 Statistics and Methods Group, EIA. 10 MR. HODGE: I'm Tyler Hodge with 11 the EIA short-term energy outlook team. 12 MR. EBEL: I'm Steve Ebel with 13 Lyondell Chemical, Houston. 14 MR. WEINIG: Steve, you'll need to 15 restate. 16 MR. EBEL: Steve Ebel with Lyondell 17 Chemical, Houston. 18 MS. BLESSING: Colleen Blessing, 19 EIA. 20 MS. MILLER: Renee Miller, EIA. 21 MS. BLUMBERG: Carol Blumberg, 22 Office of the SEIA. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 11 1 MR. DELEY: John Paul Deley, EIA's 2 records officer. 3 MR. BOURNAZIAN: Jake Bournazian, 4 EIA. 5 MS. WEIR: Paula Weir, Office of 6 Oil and Gas. 7 MS. LENT: Janice Lent, Statistics 8 and Methods Group, EIA. 9 MR. CONTI: John Conti, Director of 10 Integrated Analysis and Forecasting. 11 MR. RUTCHIK: Bob Rutchik, 12 Statistics and Methods group, EIA. 13 MR. BROENE: Tom Broene (?) SMG, 14 EIA. 15 MS. O'BRIEN: Christina O'Brien, 16 EIA. 17 MR. RASMUSSEN: Eric Rasmussen, EIA 18 Office of the Administrator. 19 MR. TSENG: Phillip Tseng, SMG, 20 EIA. 21 MR. BRADSHER-FREDRICK: Howard 22 Bradsher-Fredrick, SMG, EIA. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 12 1 MR. WEINIG: Bill Weinig, SMG, EIA. 2 MR. NEERCHAL: Thanks. And I think 3 without wasting too much time, I think that 4 we should get on with our agenda. I just 5 want to announce that -- this is a mandatory 6 announcement -- this is an ASA Committee, but 7 Nancy has the authority to adjourn this 8 meeting at any time she feels -- 9 MS. KIRKENDALL: So you all behave. 10 (Laughter) 11 MR. NEERCHAL: She feels that you 12 don't behave. So we can start with our 13 traditional session with Guy. 14 MR. CARUSO: Thanks, Nagaraj. 15 Welcome. Good morning, everyone, and 16 congratulations on your assumption of the 17 chairmanship. 18 Welcome to Ed and Ed. That's easy. 19 I know you're going to enjoy your time on the 20 committee, and we really appreciate all of 21 your time because I know you're all very 22 busy. And so this time is valuable, and we BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 13 1 value your advice. 2 I'm going to bring you all 3 up-to-date on some of the things that have 4 been going on in the EIA since our last 5 meeting in October, especially those that are 6 directly affecting some of the work we do 7 that you have given us advice on. And in 8 particular, looking forward to some of the 9 things that we are going to be doing in the 10 next one or two years that I know the 11 committee will have a chance to give us some 12 advice on. 13 So we really look forward to 14 talking to you these two days, and then 15 throughout the next couple of years about 16 some of these initiatives that are either 17 beginning now or will begin in fiscal '08. 18 Then I want to start with the budget, which 19 is already, I think, a significant departure 20 of our -- all of my previous reports to you. 21 We usually put the budget last 22 because it was usually bad news, but this BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 14 1 year it's good news, so I'm going to start 2 off with that and talk about some other of 3 the things that are going on here. 4 In the last six weeks or so, we 5 finally got a budget for Fiscal Year '07, and 6 the good news is that EIA was singled out in 7 the Department of Energy budget for an 8 increase above what most other organizations 9 were getting, which was a continuation of 10 their FY '06 budget in a continuing 11 resolution. 12 So EIA was given another 13 $5 million, which for us is very important 14 going from 85 to about 90, and even a bit 15 above what the administration had requested 16 for EIA. So I think that reflects the hard 17 work and the quality of work that the EIA 18 staff has done over these years to earn 19 Congress's trust in EIA. 20 So that's why I particularly think 21 that's good news. And a number of the things 22 that that will allow EIA to do this year have BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 15 1 been activities that you have given us advice 2 on, including a number of -- redesigning a 3 number of key oil and gas surveys, which 4 during a period of declining real budgets, I 5 think we all agree had suffered in terms of 6 quality that we wanted to maintain. And 7 certainly, the statistical accuracy that we 8 have gotten the advice from this committee 9 on. 10 We're also going to be able to 11 increase the modeling capabilities in both 12 the oil and gas area as well. 13 And then the mandated requirement 14 that was in the Energy Policy Act of 2005 15 that EIA start collecting, on a monthly 16 basis, biofuel data. They had mandated that 17 in that Act of 2005, but there had been no 18 monies appropriated for it in Fiscal '06. 19 And if we had not gotten this additional 20 money embedded in that $5 million increase, 21 we wouldn't have been able to do it. But we 22 now will do that, and we're in the process of BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 16 1 planning for these new surveys, and I'm sure 2 we'll be seeking your advice on that. 3 And then there were two petroleum 4 marketing surveys that -- under the lower 5 numbered $85 million budget we would not have 6 been able to continue, and we will now 7 continue them. Then, one of the things we're 8 going to definitely -- be on the agenda this 9 session, and I think future agendas, is 10 enhancing the activities of NEMS that you all 11 have been familiar with over the years. 12 Especially -- it was designed in 13 the early '90s, the first Energy Outlook 14 using the current version of NEMS was I think 15 about '93 or so. 16 So it was definitely time, and it 17 is time for an enhancement. And I know John 18 Conti and his team are going to really be 19 working hard over the next couple of years to 20 do that. And then -- mentioned the 21 improvement in the quality and the 22 reliability of the oil data, and we're going BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 17 1 to certainly be working hard on that. And 2 with the additional marketing on the 3 reinstatement of the two marketing surveys. 4 And then, we are also looking to 5 improve -- the cost structure of the way we 6 collect oil data has suffered somewhat from 7 being -- or having had the Legacy systems 8 which are less efficient. 9 And that will certainly change, we 10 think, during Fiscal '07 and '08. The Weekly 11 Ethanol balance is increasingly important. 12 As many of you are aware, the amount of 13 ethanol in our gasoline pool is increasing 14 steadily. 15 And we expect it will double, and 16 perhaps triple, even within the next five 17 years. So the need to have better handle on 18 ethanol production, the blending into 19 gasoline, and what that all means for the 20 supply and demand for gasoline is critically 21 important, especially as we enter into the 22 driving season. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 18 1 And of course, it's gotten even 2 more attention recently as the President has 3 announced a goal of 35 billion gallons of 4 alternative fuels, including ethanol, by 5 2017. Even better news is the request that 6 went up in early February for Fiscal Year 7 '08, the administration is requesting over 8 $105 million for EIA in Fiscal '08 which is 9 more than a 17 percent increase over '07. 10 And that would allow us to continue 11 to do the improvements in the oil and gas 12 surveys that I mentioned, would provide even 13 more money for the enhancement of NEMS, as 14 well as the normal cost of doing business, 15 which is mainly salaries and benefits for EIA 16 employees and contractors as well as the 17 other fixed costs that we have to pay here as 18 part of the Department of Energy. 19 One of the things that's really 20 ramped up even in the last two to three 21 months is with the seating of the new 22 Congress, has been more service requests BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 19 1 from Congress for EIA. And they've 2 concentrated mainly on issues related to 3 global climate change. 4 We've done one report; we completed 5 one report for Senator Bingaman and five 6 other senators on the energy market and 7 economic impacts -- to reduce greenhouse gas 8 emissions using a cap-and-trade system with a 9 safety valve. And that was published I think 10 in about mid-January. 11 We are now working on another 12 request from Senators McCain and Lieberman on 13 a bill that would impose even more stricter 14 controls on carbon. We have requests in from 15 Senators Inhofe, Kerry and Snow. So I could 16 go on and on because we have probably about 17 10 outstanding requests that we're either 18 working on right now or in discussions with 19 the Hill staff that involve in some way 20 greenhouse gas emissions. 21 And so all the more 22 important -- some of the advice we've gotten BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 20 1 from you about the modeling and dealing with 2 these issues, and how important it is that we 3 do improve the modeling system that we have 4 now. 5 But we also expect that we'll be 6 doing some work on yet another attempt to 7 pass a bill on a Renewable Portfolio Standard 8 that Senator Bingaman tried very hard to get 9 into the Energy Policy Act of '05. He has 10 now, in co-authorship with Senator Domenici, 11 submitted another bill that would have a 12 Renewable Portfolio Standard, I believe, at 13 20 percent. 14 And the last time they tried it, 15 that was 10 percent. The percentage could be 16 15. But in any case, we do think that we'll 17 be getting a request to work on Renewable 18 Portfolio Standards as well. And there are a 19 number of other studies that have been 20 completed or are underway that are listed 21 here. 22 Main point is that this is a very BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 21 1 active new Congress, and EIA, as I pointed 2 out in the budget story, has gained the 3 respect and trust of many members, and that's 4 definitely good news. Of course, what that 5 does mean is we do get -- getting asked to do 6 a lot more. And many times, it comes without 7 specific funding. 8 So it clearly is a question of 9 priority-setting. It makes it very difficult 10 for -- particularly the long-term forecasting 11 area, because most of these requests are what 12 if we pass a bill; what if we change the law 13 on a particular thing, how does that change 14 the outlook, and therefore -- and NEMS is the 15 tool for answering many of those questions. 16 MR. CLEVELAND: Guy? 17 MR. CARUSO: Yes. 18 MR. CLEVELAND: Do you have to 19 produce a report for every request? And how 20 do you set priorities in terms of which ones 21 get done first, and how you roll them. 22 MR. CARUSO: That's an excellent BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 22 1 question. The answer is no, we don't have to 2 accept every request. And in the past, what 3 we've done is never say no directly, but 4 negotiate. So what we've tried to do is we 5 limit the scope of a request, because if it's 6 so open-ended it's going to take enormous 7 resources, we certainly let them know that 8 it's possible, but it may be six months 9 before -- so if you really want this in a 10 timely fashion, we need to focus the 11 assumptions and what you're targeting that to 12 do. 13 And in other cases, we try to get 14 similar requests, to combine them. So we 15 might have one from Kerry and Snow, which we 16 do, which is very similar to what we're doing 17 already for McCain and Lieberman or Inhofe, 18 or some -- so it's really mostly negotiation. 19 There are some organizations that 20 limit requests to chairmen or -- of either 21 committees or subcommittees or ranking 22 members. We don't have such a strict a rule BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 23 1 as that. But it's trying to match -- setting 2 priorities based on what resources we have, 3 and what impact that would have on other core 4 work. 5 So as I said, the brunt of most of 6 these requests fall on the Forecasting 7 office, John Conti's office. So you run the 8 risk that you actually could delay the AEO 9 publication or other key activities. So it's 10 a balancing act, but no hard-and-fast rules. 11 Wonder why it's not going. All 12 right. Okay, there we go. 13 We have finally completed all the 14 program guidelines for the new voluntary 15 reporting on greenhouse gas redesign, and we 16 expect that the new reporting forms will be 17 approved by OMB by next month. 18 And we do actually expect to begin 19 internet data collection by this fall. And 20 this been a long process, as Jae Edmunds 21 knows very well. I think it's been at least 22 two years in just getting approval of these BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 24 1 guidelines. 2 One of the areas that you have 3 worked very hard on and given us excellent 4 advice in collaboration on is the Form 914, 5 which is now the official natural gas 6 production data collection as of January of 7 '07, which was in the March issue of Natural 8 Gas Monthly. And it's been performing 9 extremely well in terms of the degree of 10 reporting error, reducing the amount of 11 error. 12 And we're pleased with the way 13 that's actually turned out. We now have a 14 much shorter time frame -- 60 days instead of 15 120. So I wanted to just bring you 16 up-to-date on that, and you had a number of 17 sessions with John Wood and others on that, 18 and that's worked out very well. 19 EIA was part of the formation of 20 the Department of Energy, and EIA was 21 actually put together -- officially put 22 together in 1977. And obviously, this makes BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 25 1 this 30 years. So the good news is we're 2 going to have a great anniversary celebration 3 later this year. And we certainly want this 4 committee, which has been a big part of that 5 history, to be involved in that. 6 And we're going to schedule a 7 session that will focus on ASA's 8 contributions to EIA, including a -- we're 9 looking at doing a two-day conference, which 10 will include a session with ASA. 11 The bad news is something I've 12 talked about in the past -- and that is a lot 13 of people, a lot of staff members were here 14 in 1977 when we began. And therefore, in the 15 federal government, 30 years is a magic 16 number for being able to retire at 55 years 17 old. 18 So we're projecting about 30 19 retirements this year, this calendar year. 20 And when you combine that with normal 21 attrition and the fact that we're a little 22 below our FTE ceiling, we're projecting 50 BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 26 1 new staff members to be hired in '07. The 2 last time I checked, we had about 15, so we 3 got about 35 net more to go. 4 So we've talked about this in the 5 past about how much we appreciate your advice 6 in terms of students or other colleagues that 7 might have appropriate candidates for -- not 8 only statisticians, mathematicians, 9 economists, and operations research. 10 So we again are working very hard 11 to get up to full -- to be able to 12 effectively use that, those monies that we're 13 getting. And just specifically on the ASA, 14 John Paul Deley is our kind of person who is 15 handling our history of EIA. And he is 16 putting together an index of all papers of 17 all the ASA meetings all the way back to '79. 18 And we would also be open to having 19 a web exhibit on the history page directly 20 related to ASA, this ASA committee, so more 21 on that later. 22 I think we're actually going to BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 27 1 have a little discussion about that with you 2 all during the next day or so. 3 Then, the historic -- you're here 4 at a historic moment. This last week, for 5 the first time in 20 years, all of EIA, 6 except for the Dallas Field Office, is now in 7 Forrestal Building. And Nancy and her team, 8 which you're most familiar with, have moved 9 down here near -- guess it's two weeks ago, 10 and then the last of the Electricity folks 11 came last week. 12 We now have 525 EIA federal 13 employees and contractors in this building, 14 and mostly on the second floor but also on 15 the first and the basement. 16 And I think that -- we all believe 17 that that's going to improve our internal 18 communications. And we're already benefiting 19 from seeing the smiling faces of Bill Weinig 20 and Alethia every day. And I personally 21 appreciate that. So any questions that I can 22 answer about this or any other things that's BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 28 1 concerning EIA and its direction? 2 If not, Nagaraj, once again, 3 welcome to EIA. And we really value the 4 advice we're going to get from you in the 5 next day or so. 6 MR. NEERCHAL: Thanks, Guy. 7 Nancy? 8 MS. KIRKENDALL: Well, I'll join my 9 welcome to Guy's for the committee. We hope 10 we have an interesting series of talks, and 11 hope we get lots of good advice. 12 In terms of SMG updates, as you 13 heard, we just moved down from the building 14 up the street. So we're now all down here in 15 the same building. This certainly will 16 facilitate discussions with people. Now 17 there's no more excuse not to just go drop 18 over and bug somebody about something. 19 In SMG, one of the reasons why 20 Alethia is taking on more of a responsibility 21 for the ASA meetings is that Bill thinks he 22 might actually retire soon. And so he has BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 29 1 been absolutely wonderful in trying to 2 prepare us for this transition. And for this 3 year, he and Alethia are sharing 4 responsibilities for the meeting, so she's 5 going to becoming more and more involved in 6 managing the meetings. 7 So thanks to both Bill and Alethia, 8 especially to Bill, for his continuing to 9 support the committee by making sure that 10 there's a smooth transition. 11 Another announcement for SMG is 12 that Janice Lent has joined us as our senior 13 mathematical statistician. She comes to us 14 from the Bureau of Transportation Statistics 15 and from the Bureau of Labor Statistics. One 16 of the areas that she has a lot of expertise 17 in is development indicies, and so we'll see 18 how that proceeds here in EIA. 19 So this is just a summary of some 20 of the things we talked about at the fall 21 meeting. And as you remember, Lawrence 22 Stroud and Phillip Tseng gave a talk about BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 30 1 using models to detect outliers in refinery 2 data. 3 The Committee suggested we needed 4 to be a little more careful in handling the 5 correlation in the data series, but also 6 suggested that we should look at regional 7 data. So we do plan to pursue this. We just 8 haven't had time over the last few months. 9 On suggestion from the external 10 study team, I'm not sure we went through a 11 lot of detail about all the recommendations 12 that we've received. But one of the 13 recommendations was that we should get some 14 of our high visibility weekly data series 15 classified as principal economic indicators. 16 And we have actually moved on that. 17 We sent in an application to OMB to make our 18 weekly natural gas storage report a principal 19 economic indicator. So we don't know the 20 outcome yet. OMB has promised to let us know 21 in May or June. They just haven't had time 22 to think about it. They said they liked our BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 31 1 application but they haven't thought about it 2 deeply. So that's some progress on one 3 recommendation. 4 They also recommended that we 5 should provide researcher access, 6 particularly to our commercial building 7 survey data and to our residential energy 8 consumption survey data at protected sites. 9 I think they had in mind particularly using 10 the Census Bureau's research data centers. 11 But any way of getting better 12 access to the data is I think what they 13 really intended. So Jake Bournazian will 14 talk to you tomorrow about some of the 15 options we're considering on that. 16 The other thing they talked about 17 is that we needed to have more collaboration 18 with academia. So one of the things that 19 Ruey-Pyng Lu is going to talk about is that 20 we have joined the National Institute of 21 Statistical Sciences as an affiliate member. 22 And we'll be operating a sort of a research BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 32 1 grants program for students, and possibly for 2 new researchers to try to get people to work, 3 researchers to work, on problems that we're 4 interested in. 5 And so Ruey-Pyng is going to talk a 6 little bit about that arrangement, and he's 7 going to go over four draft project 8 descriptions to try to get your input. What 9 we'd really like is to encourage people to do 10 work that we're really interested in. And so 11 you need to have something that's a short 12 description that people might read, but 13 detailed enough so that they'll know whether 14 or not they're interested in the topic and 15 what they need to do. 16 So there's kind of a little balance 17 there. And I'm sure we'll have good advice 18 on that from you. So that's a session 19 tomorrow morning. 20 Tomorrow morning? No, it's today. 21 MR. NEERCHAL: Afternoon. 22 MS. KIRKENDALL: It's today; I hear BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 33 1 it from the back, okay. 2 And then other recommendations from 3 the external study team, we'll be considering 4 in our strategic planning process. And 5 that's going to begin soon. I think we meet 6 next week, in fact. 7 Ruey-Pyng last time also talked 8 about estimating weekly stocks of other oils, 9 and I think Moshe recommended that we try 10 using an unobserved components model with 11 monthly seasonality. So Ruey-Pyng did that, 12 and he found that it was better than anything 13 else he had tried, so there's a little 14 success story. 15 And now we need to work with the 16 program office to show them what we've done 17 and see if they'll pick it up and use it. 18 Last time, George Lady talked about the EIA 19 quality project, and he's starting out 20 focusing on forecasting, and in particular, 21 he is starting out with NEMS. 22 He's continuing to work on this BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 34 1 project and he'll talk to you tomorrow 2 morning, to give you a better update. The 3 challenge that we haven't gotten to yet is 4 the how to do something similar. He has 5 ideas on how to handle NEMS in reasonable 6 ways. 7 We think steel (?) is different, 8 but there are other ways of handling it, but 9 then how do you something consistent on the 10 data side, because if you are going to have 11 an annual quality report, we ought to address 12 all of the areas. So that's work ongoing and 13 things to think about. 14 So in the spring meeting, I think 15 Stan Freedman talked about the functional 16 requirements document for the EIA's internet 17 data collection. We threw that out as IDC. 18 The functional requirements document was 19 adopted within EIA, and we're now in the 20 process of determining how our surveys, how 21 to actually implement it. 22 There are a couple of things being BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 35 1 considered. We found out that the National 2 Agricultural Statistic Service has a software 3 tool that they have developed already, and 4 they are willing to share it with us. 5 It may not be exactly what we 6 wanted, but it does a lot of the things that 7 we were interested in doing. So there's 8 always a debate -- do you adopt an existing 9 system and then adapt it to your needs or do 10 you build a new system? 11 And so that is going on. As you 12 heard from Guy, the greenhouse gas survey is 13 supposed to go live on that system this fall, 14 and the electric power surveys are going to 15 be new beginning in January of 2008. 16 So they need to be in the new 17 system in January. So there are going to 18 have to be some decisions soon and we will 19 have to march ahead. So I'm sure in the fall 20 meeting, we'll able to tell you even more. 21 We talked about how modeling can 22 suggest data needs, and you suggested to BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 36 1 engage the energy modeling forum -- but one 2 way we've done that is to invite John Weyant 3 to join us. He's not here right now, but he 4 has agreed to join us. 5 The other thing was that you 6 suggested that we have better outreach, more 7 users' conferences and things, and we have 8 been having an annual NEMS conference for 9 many years, and that's slowly becoming 10 broader. I think in the last couple of 11 years, it was the forecasting conference 12 rather than just the NEMS conference. 13 This year, we actually had sessions 14 on electric power, and I think next year is 15 going to be even broader. So we are moving 16 towards an EIA-wide conference. This is 17 usually held in the spring in March. 18 We had a session on the 2006 MECS, 19 and I think you recommended incentives, 20 pre-mailing to respondents, and a variety of 21 other things. We didn't consider incentives 22 for that project. It's a little hard for BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 37 1 surveys of establishments to come up with 2 good incentives. 3 They did do a pre-mailing, and this 4 is partially because for the first time, 5 they're using Census Taker, which is the 6 Census Bureau's internet data collection 7 system. And it sounds like they've had -- I 8 mean, they're still in the middle of the data 9 collection process. But it sounds like they 10 have had very good response to the use of 11 Census Taker, and so hopefully that will help 12 response. 13 The other thing they did was to 14 have a short form for the survey, it's a very 15 large survey. And some very small 16 establishments can fill out a short form, and 17 they're hoping that that will help the 18 response rates among the smaller companies. 19 And finally, you've heard about the 20 30th anniversary. One thing I would like to 21 point out that's historic about this very 22 meeting is that Ed Kokkelenberg has joined BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 38 1 us. This is the first time we've had -- 2 MR. KOKKELENBERG: Am I historic? 3 MS. KIRKENDALL: Yes, it's 4 historic. This is the first time we've had a 5 former ASA Energy Committee member come back 6 for another round of appointments. And so he 7 served for six years -- was it in the early 8 '90s? 9 MR. KOKKELENBERG: I think so. 10 Maybe in the mid-90s -- 11 MS. KIRKENDALL: Just so it was 12 long enough ago that everybody has forgotten 13 that he did it. 14 MR. KOKKELENBERG: I remember that 15 NEMS was starting up then. 16 MS. KIRKENDALL: So it was during 17 the NEMS -- so there's a chance that in 18 another 10 or 20 years, you can come back and 19 be on the Committee again. Hopefully, it 20 will be a good enough experience you would 21 like to do that. 22 Then the last one is just -- this BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 39 1 was the relationship between crude oil prices 2 and natural gas prices was a paper given by 3 Jose Villar in Natural Gas Division. We have 4 done just a little bit more. There was Andy 5 Kydes, who works in the Office of Integrated 6 Analysis and Forecasting, had an intern who 7 looked at this project in the summer, and she 8 added coal prices, I think, to the model, and 9 was looking what to do with the trend term. 10 But on the other hand, she was just 11 a summer intern and so the work isn't really 12 complete. So we're still very interested in 13 this topic. The topic in general is 14 co-integration between data series and how 15 does that affect your forecasts. 16 So this work is still of great 17 interest to us, and it is one of the project 18 descriptions that we have put in for the 19 National Institute of Statistical Sciences' 20 student grants program, so maybe you can help 21 us write that up. And maybe you have a 22 student or two who might be interested in it, BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 40 1 that would be even better. 2 That's the end of my talk. 3 So thank you. 4 MR. CARUSO: Nancy just reminded me 5 when she said about the work that the intern 6 did -- we also as part of our recruitment 7 enhancement with having to hire so many new 8 people, had a very expansive interns program 9 last year. 10 We had I think 24 interns. So 11 that's another area where if you have 12 students or acquaintances that you know that 13 might be interested, we're very open to that, 14 and it's about the right time of the year to 15 start hearing from them. 16 Thank you, Nancy. 17 MR. NEERCHAL: Thank you, Nancy, 18 and Guy. Questions for either Nancy or Guy? 19 MR. CLEVELAND: It's kind of moving 20 ahead to the -- this National Institute 21 thing, but on this point, I think we really 22 need to do more to advertise these BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 41 1 initiatives, like who knows about this intern 2 program. 3 This should be on that front page. 4 Now accepting applications for interns and 5 developing a communication network, which I 6 think we'll take about as part of the -- this 7 research part, I think, where you're going 8 to -- where you have list servs and 9 newsletters and e-mail lists, that you target 10 when these initiatives come out, that people 11 should be able to sign up for the list serv 12 on your website and things like that. 13 We'll talk more about that later, 14 but I think there's a lot you can do to 15 enhance your outreach effort. 16 MS. KIRKENDALL: Great. 17 MR. NEERCHAL: Any other questions, 18 comments? Thank you. Let's move on. 19 I think since we will not have 20 another discussion I think as you have said 21 we're going to -- the time is going to work 22 out I think, so John is -- so we'll go to the BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 42 1 next session here. Phillip Tseng on Modeling 2 Regional Electricity Generation. 3 MR. TSENG: Good morning, 4 Mr. Caruso, Nancy, and members of the ASA 5 Energy Committee, and ladies and gentlemen. 6 Today, I will take about the modeling work I 7 have been working on for a long time. It's 8 Modeling Regional Electricity Generation. 9 Basically, I will cover a few 10 topics. I will talk about how we started, 11 why we started, and objectives of the 12 efforts, and the data sources, methodology, 13 and some of the preliminary findings, and I 14 have some questions on fuel switching, 15 because the methodology I used kind of bypass 16 explicit price representation. And I have 17 some other possible applications of the 18 model. If I have time, I'll probably go over 19 it. 20 We started the project about two, 21 three years ago, and the objective was to 22 increase regional details to the short-term BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 43 1 projections and analysis. And also, we 2 wanted to just understand what regional 3 markets are doing. 4 And we developed a methodology. I 5 think I gave a presentation with Dave 6 Costello, who passed away, unfortunately. 7 And we used a methodology -- it was a little 8 bit awkward to use, and it takes a lot of 9 effort to actually calibrate a model and to 10 generate satisfactory results. 11 And part of the reason we kind of 12 took so long to complete the model to this 13 stage is, we lost a valuable co-worker, and 14 as a result, we're kind of in transition for 15 a while. 16 Now to iterate the objectives. 17 As we move along, we learn more 18 about data. We thought, we think we can do 19 more with the datasets we have. 20 And so again the first objective is 21 project power generations from power plants, 22 burning fossil fuels, and we have regional BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 44 1 representations. And the second objective is 2 actually using the model to conduct scenario 3 analysis, and hopefully provide insights into 4 potential power flow problems. 5 We focus on fossil plants, because 6 in the short run, dispatching patterns of 7 other types of power plants are much less 8 responsive to changes to demand. Like 9 renewable, for example, we don't really have 10 control over when solar will dispatch or wind 11 will dispatch. 12 And nuclear power plants is more or 13 less baseload, and it has to around almost 14 all the time. Operators may be able to ramp 15 up the generation a little bit or ramp down a 16 little bit, but in general, nuclear is a 17 baseload power plant. 18 So only fossil fuel plants are used 19 to meet changes in demand patterns. The data 20 sources we have -- basically the EIA-826, in 21 the sales and revenue by end-use sectors. 22 And we use the EIA sales numbers and we have BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 45 1 the power control area data with a daily 2 24-hour load, and then from that we create 3 the demand curves -- average daily 24-hour 4 load demand curves. 5 And on the supply side, we use net 6 generation by generator -- I think we talked 7 about last time that data record is about 8 8,000 or 9,000 records, and we have the 9 generating capacity by fuel type. It's an 10 annual number, which covers about 20,000 11 generators. And from that net generation and 12 generation capacity, we create supply curves. 13 We will figure out a way to kind of identify 14 the supply curves, and then we use the demand 15 and supply representation to solve for 16 generation. 17 For the electricity region, we have 18 13 regions -- just for the lower 48, and we 19 modeled like Hawaii and Alaska separately, 20 because those two states are kind of 21 independent of the lower 48 states. And we 22 have -- for the Eastern Regions, we have nine BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 46 1 regions -- number 1 through 7, number 11, 12. 2 And for the Western Region, we have Mountain 3 and -- Pacific, excluding California -- and 4 California, so at the West End, we have three 5 regions, and Texas is a region by itself. 6 So this is the map which shows the 7 regions. This is the census region, and we 8 have Pacific, California, and Mountain in the 9 West, and we have nine regions in the East, 10 and Texas, we make it as one region. So 11 that's the basically three trading blocks, 12 and we allow trading, say, between the 13 Mountain region, Pacific, and California, and 14 then to east nine regions trading as one 15 trading block, and Texas is itself. 16 The challenges for us was 17 to -- yes? 18 MR. EDMONDS: Just to make sure I 19 understand -- when you're modeling, are you 20 trying to predict the demand and supply of 21 electricity at each hour of each day of the 22 year in each of these regions? Or BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 47 1 what -- can you just kind of get me on to the 2 time. 3 MR. TSENG: For the modeling 4 effort, the first step is to use the existing 5 historical data on demand side and supply 6 side, and then see if we can dispatch power 7 patterns for fossil fuel matching the 8 historical pattern. So the demand numbers 9 are historical numbers. For projection, then 10 we project monthly demand, and from the 11 monthly demand, we convert it to average 12 daily load, and then we solved (?) the model 13 is based on the average daily load. 14 So each month, we have a typical 15 month, a typical day, average day, and then 16 we have to convert the average day to 24-hour 17 load. And then we solve it hourly, so the 18 model is solved hourly, but with the 19 projection horizon is normally 24 to 36 20 months. So when we look at 36 months -- 21 MR. EDMONDS: It's 36 times that -- 22 MR. TSENG: Yeah. Each month we BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 48 1 solve 24-hour timestamps, so we get a typical 2 average day and we convert it to monthly. 3 MR. EDMONDS: All right. 4 MR. TSENG: So the projection of 5 demand is based on weather and income. So 6 what we're doing here is basically for the 7 reference case or the base case, is seeing 8 development methodology -- see if we can 9 actually replicate historical dispatching 10 patterns for fossil plants. 11 And also the challenge is we need 12 to provide a framework, because we want to 13 understand if that thing is beyond control 14 like power plant outages -- if somehow we 15 lose one or two nuclear power plants, that 16 could be 2,000 megawatts of power plant 17 generating capacity in the summer, what 18 happens? What happens to the power flow and 19 what happens to the generation capacity and 20 capability in meeting the demand? 21 So those are the kind of things 22 relevant to the modeling. And another thing BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 49 1 is when we do simulation analysis, sometimes 2 as -- actually, one very important aspect is 3 what happens if summer demand is higher than 4 normal, and what kind of power plants would 5 be used to meet the demand? 6 And what happens to the demand for 7 natural gas? Because a few years ago when 8 natural gas price was very high, relatively 9 much higher than even the crude oil prices in 10 BTU terms, some of the gas combined cycle 11 power plants, very efficient state-of-the-art 12 gas power plants, were shut down because of 13 the high prices. So there are a lot of 14 interactions between the electricity sector 15 and the natural gas market, and we wanted to 16 get some insight in terms of the interactions 17 in one sector and its impact on the other 18 one. 19 That's part of reason why we want 20 to build a model which can provide more 21 consistent information on demand for fossil 22 fuels. So that's the purpose of building a BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 50 1 framework, so that the model should have a 2 capability to do those kind of analysis. 3 And of course for the users who run 4 the model, the short-term model is run on the 5 EViews. That's a software that many people 6 use for econometric simulation modeling. And 7 so the modeling framework has to be on 8 EViews, because it's easier to integrate the 9 electricity module into the framework of the 10 overall short-term regional forecast. 11 So with the next few slides I will 12 talk about the methodology, how do I create 13 the supply curve, because the generation 14 model is about how fossil fuel plants are 15 dispatched. And the question is, how can I 16 have a representation capture the supply 17 which the supply curve would incorporate the 18 capacity constraint? 19 So when we run the model, the model 20 solution would not go beyond available 21 capacity. So that's one of the issues in the 22 short run. In the long run, we can always BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 51 1 add new power plants. So what we did 2 actually is we used most up-to-date data, in 3 this case, it's year 2005, to project 4 generation for the next 24 months. And so 5 power plant owners -- normally, power plant 6 owners should dispatch power based on the 7 least cost power plant first. 8 If you'll look at a system, 9 however, there are some other factors, like 10 causing the accessibility of transmission 11 lines -- if some generators have to sign some 12 kind of a contract with the transmission 13 operators, so there are some contractual 14 issues. 15 And they send to the load centers 16 and transmission and distribution losses, and 17 also availability of fuels, because we look 18 at data and we see some power plants even 19 though it's a dual-fuelled, but even when we 20 see the prices, we see some power plants 21 would still use more expensive fuels sometime 22 in the year, just because of there's not BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 52 1 enough certain type of fuel. 2 For example, the gas and oil units, 3 the oil storage capacity will limit how much 4 a dual fuel units can burn residual fuel, for 5 example, even though residual fuel may be 6 cheaper sometimes in the year. 7 But once that cheaper oil is used, 8 it may take some time or logistic to bring 9 those fuel into the storage, and so we still 10 see some natural gas used. So those are the 11 consideration for our estimation of the 12 supply curve. 13 So we actually look at a system 14 cost instead of the variable cost of the 15 power generation to represent the supply 16 curve. So the construction of short-term 17 supply curves for coal and natural gas, 18 residual fuel, and diesel fuel uses observed 19 dispatching patterns instead of -- yes? 20 MS. KHANNA: Just a quick question. 21 Can you clarify what you mean by variable 22 cost? BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 53 1 MR. TSENG: Variable cost is -- for 2 example, you have a coal power plant, and 3 each power plant has certain conversion 4 efficiency. Coal is probably like 5 35 percent, for example, and so one unit of 6 coal, one BTU of coal, and we can get certain 7 units of electricity -- and then from that, 8 we can estimate a variable cost of generating 9 electricity from a coal power plant. 10 MR. LIDDERDALE: It's fuel cost. 11 MS. KHANNA: Fuel cost? 12 MR. TSENG: Yeah, basically it's a 13 fuel cost, but fuel cost is converted to unit 14 electricity cost when we incorporate 15 efficiency loss. 16 MS. KHANNA: Okay. 17 MR. TSENG: But system cost, then 18 there's transmission charge tariff, and then 19 also TMB loss. So from the end users' point 20 of view, end user only see the charge for the 21 end used electricity, but there's usually the 22 transmission loss of 5 to 7 percent. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 54 1 And so someone is paying for that 2 loss as well, so the end user probably would 3 prefer the cheapest electricity first. And 4 so when we look at this is -- since the 5 spatial consideration is one element in the 6 siting and building of infrastructure, so 7 sometimes power plants with higher variable 8 cost may be dispatched earlier because it's 9 closer to the load center. 10 So when we estimate a short-run 11 supply curve, we simply use observed pattern, 12 because we see -- what if this power plant is 13 dispatched first, then it must be cheaper, 14 for whatever reason. It could include the 15 cost, also could include institutional 16 factors, because sometimes, from what I have 17 heard, a big generator may have better 18 relationship with the transmission line 19 operators because of the volume, and so in 20 order to increase the utilization rates of 21 transmission line, those generators may have 22 some benefits in getting to the assigned, BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 55 1 allotted capability of moving their products. 2 And that's one area. 3 And so the way I did it to 4 calculate the dispatching pattern is, we 5 assign -- first, we calculate the utilization 6 rate of all power plants, and if power plants 7 are utilized more, then they must be cheaper 8 in the system sense -- the cost of delivering 9 electricity to end users must be cheaper. 10 That's why those power plants are used more. 11 And I assigned the power plants 12 into 10 categories, basically 10 bins, and I 13 will have little bit more detailed when I 14 move down. And then I choose two functional 15 forms to fit a supply curve, and one for each 16 of the four types of fossil fuels: Coal, 17 gas, diesel, and residual fuel. 18 And then for each demand level, I 19 adjust for the non-fossil plants, and then I 20 can solve the model. 21 MR. KOKKELENBERG: Can I ask you a 22 question about that statement? BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 56 1 MR. TSENG: Yes. 2 MR. KOKKELENBERG: It says two 3 forms are selected to fit -- 4 MR. TSENG: Right. 5 MR. KOKKELENBERG: One for each, so 6 in other words, you have used two forms? 7 MR. TSENG: Okay. 8 MR. KOKKELENBERG: But not for 9 each. You only used one functional form for 10 each. 11 MR. TSENG: One kind for each, yes. 12 MR. KOKKELENBERG: Thank you. 13 MR. TSENG: Actually, I have 14 specific example here. So this is that the 15 way I did. Actually, I talked to my 16 co-worker, Dr. Hodge, he doesn't want to put 17 his name on the paper. But he actually put 18 in a lot of effort. For this part of the 19 work, he made very significant contribution. 20 What we did here is, first, we assigned the 21 bin numbers. For example, for coal here, if 22 capacity -- this is Texas -- and the capacity BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 57 1 numbers in megawatt, and for coal, we have 2 the bin number 1, is 4,623 megawatt, this is 3 the capacity. So what it means is you have 4 this much capacity with utilization rate of 5 more than 90 percent. So I put the numbers 6 in in bin number 1. 7 And in the next bin number 2 is 8 basically utilization rate between 80 and 90. 9 So that's the first column for coal. And 10 then here on the cumulative capacity, this is 11 how I estimate a supply curve. I simply say, 12 well, as we see that the cumulative capacity 13 tend to increase, and the maximum is 14 basically the capacity, total capacity of 15 coal in the region. And from that, I create 16 supply curves. 17 Before I estimate the supply 18 curves, I made some adjustments to the coal 19 power plants. And part of reason is, from 20 the data we have, we notice coal power plants 21 burn some natural gas, burn some oil. And 22 part of reason is coal operators, coal power BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 58 1 plant operators, probably have to do more 2 maintenance of the coal power plants than 3 natural gas or oil plant. 4 And for each time a coal power 5 plant is shut down, when it needs to restart, 6 and the way you restart a coal power plant, 7 you don't just light a match to the coal. 8 You need natural gas or oil. 9 And so we look in the data, we 10 assigned availability factor of 93 percent, 11 94 percent, and 96 percent to different 12 regions. Back to the functional form -- and 13 we actually select two types of functional 14 forms. And for the coal we use semi log 15 inverted functional form. And dependent 16 variable is coal cumulative capacity, and 17 independent variable is the bin number. 18 And for gas, diesel and residual 19 fuel, I selected the quadratic form; I used 20 semi log. Part of the reason is, if I used 21 the normal number, I get into negative 22 projected capacity, because the curve fit, BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 59 1 the quadratic forms, tend to go down a little 2 bit, and then move up. So the selection of 3 the functional form is basically is trial and 4 error, and I get very good fit when I use 5 these two functional forms. These are some 6 of the results. For Texas, I used July 2005 7 data, you can see the coal supply curve. The 8 R (?) squares above 0.9 -- yes? 9 MS. KHANNA: A quick question. How 10 many observations are these results based on, 11 the reverse -- 12 MR. TSENG: For this run, we used 13 10 bin numbers, so you should see number 1 14 through 10 on the X axis. 15 MS. KHANNA: But how many 16 observations in your estimation data set? 17 MR. TSENG: For each month, we have 18 only -- creation of the supply curve. We 19 have 10 bins -- 20 MS. KHANNA: Uh-huh. 21 MR. TSENG: But in terms of the 22 number of power plants and number of BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 60 1 generation numbers, I think for the U.S. 2 total, we have about 20,000 generators. We 3 have about 8,000 or 9,000 generation numbers. 4 Because for example, in Texas, if you have 5 two coal power plants, steam generators, then 6 the net generation number from the same power 7 plants is reported as one number. That's why 8 the net generation number will have a smaller 9 number of observations than the capacity. 10 And that's how we group the power plants into 11 10 bins, based on the utilization rate. 12 MR. CLEVELAND: But your curves 13 will only fill into 10 data points -- 14 MS. KHANNA: Ten data points, yes. 15 MR. CLEVELAND: Yeah. 16 MR. TSENG: Right, yeah. 17 MS. KHANNA: So it's not surprising 18 to get a really good fit. 19 SPEAKER: Yeah. 20 MR. CLEVELAND: So the axes here 21 are what? 22 MR. TSENG: Okay, the vertical -- BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 61 1 MR. CLEVELAND: Bins -- 2 MR. TSENG: The vertical axis is 3 that the natural log of capacity. 4 MR. CLEVELAND: Natural log of 5 capacity? Okay. 6 MR. TSENG: Yes. 7 MR. EDMONDS: Just to sort of see 8 if I understand this, what you've done is 9 you've created a dispatch ordering -- 10 MR. TSENG: Exactly. 11 MR. EDMONDS: Based on the 12 historical observation. 13 MR. TSENG: Yes. 14 MR. EDMONDS: So you look -- when 15 you did this, I guess, slide 11, if you had 16 plotted that, what you would've shown us was 17 that in bin 1, you've got -- those are just 18 running all the time. 19 MR. TSENG: Yes. 20 MR. EDMONDS: They're never off. 21 MR. TSENG: Almost, yes. 22 MR. EDMONDS: Right. And bin 2, BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 62 1 they're running almost as much. 2 MR. TSENG: Right. 3 MR. EDMONDS: And bin 3 and so 4 forth, until you get to the higher bins like 5 8, 9, and 10. 6 MR. TSENG: Right. 7 MR. EDMONDS: Where they're only 8 coming on to hit your peak. 9 MR. TSENG: Right. 10 MR. EDMONDS: And like V, if it's 11 July, the hottest hour of the day -- 12 MR. TSENG: Right. 13 MR. EDMONDS: Those come on, but 14 otherwise, they're not -- 15 MR. TSENG: Yes. 16 MR. EDMONDS: So basically what 17 you've done is essentially a load curve, and 18 then on the right is your cumulative load 19 curve or a dispatch curve -- 20 MR. TSENG: Right. 21 MR. EDMONDS: For these units. And 22 that's just based on observation. Basically BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 63 1 you're saying for whatever reason, this is 2 what we see -- the behavior of -- in this 3 case Texas. 4 MR. TSENG: Right. 5 MR. EDMONDS: Texas electric 6 utility operators. 7 MR. TSENG: Yes. 8 MR. EDMONDS: But you 9 actually -- and there's another calculation 10 you could've done from those data, and 11 probably did on the side, which is to take a 12 look at those units and actually ask what's 13 the economics of the -- if you were the 14 operator and you're just looking at the 15 performance of those units, would -- 16 MR. GRUENSPECHT: Economic 17 dispatch -- 18 MR. EDMONDS: Yeah, how would I -- 19 SPEAKER: And the whole idea of 20 this, I think, is to avoid the -- if in fact 21 the generation of the load were all located 22 at one point, and there were no transmission BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 64 1 constraints. 2 MR. TSENG: Right. 3 SPEAKER: The two should give you 4 the same result. 5 MR. TSENG: Right. 6 SPEAKER: But this is sort of an 7 effort to pick up -- 8 MR. EDMONDS: Well, this is -- 9 SPEAKER: Affect -- 10 MR. EDMONDS: Really did -- 11 SPEAKER: Affected -- 12 MR. TSENG: Exactly. 13 SPEAKER: Yeah. 14 SPEAKER: Right. 15 SPEAKER: So -- 16 MR. GRUENSPECHT: So implicitly, 17 the notion is if they really did something 18 that seems like it would be out of order, 19 based on the second calculation that you've 20 described, then that implies -- because 21 unlike in most parts of our modeling, we tend 22 not to assume rational economic behavior, we BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 65 1 don't assume that people buy the right 2 refrigerator or the right light bulb or the 3 right copying machine. 4 MR. EDMONDS: Right. 5 MR. GRUENSPECHT: Usually in the 6 electricity sector, we think -- I mean, 7 people running this thing pretty much 8 understand which is cheaper, which is more 9 expensive, so if they -- 10 SPEAKER: Right. 11 MR. GRUENSPECHT: Use the more 12 expensive. 13 MR. EDMONDS: There's a reason. 14 MR. GRUENSPECHT: There's a reason. 15 MR. EDMONDS: Right. 16 MR. GRUENSPECHT: And we wanted to 17 reflect that -- 18 MR. EDMONDS: Yeah. 19 MR. GRUENSPECHT: Yeah. 20 MR. EDMONDS: And the only thing I 21 was going to ask you is how do those stack 22 up? And presumably, there were some real BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 66 1 insight -- I presume you probably did that 2 stack -- 3 MR. TSENG: Actually -- 4 MR. EDMONDS: And there are 5 probably some real insights that came out 6 of -- 7 MR. TSENG: At the very beginning 8 of this project, we actually did a 9 scattergram with the vertical axis as to 10 variable cost and the horizontal as capacity 11 utilization rate, and we found some low cost 12 with low utilization rate power plants, and 13 some high costs with high utilization rate, 14 and so we -- initially, we started with the 15 cost -- minimization algorithm, and we 16 realized it -- we couldn't replicate a 17 historical generation pattern, and then we 18 stepped back and we said, well, maybe the 19 system cost is the one. 20 And for system costs, the only 21 thing -- the only easiest way to pick up that 22 is looking at utilization rate, because BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 67 1 that's almost like reviewed preference type. 2 MR. EDMONDS: Oh, yeah. 3 MR. TSENG: What we see is really 4 what happened. 5 MR. EDMONDS: Yeah. 6 MR. TSENG: And the question is can 7 we pick up the reviewed preference and then 8 replicate a historical pattern. This is the 9 test. And there's always trade-off -- when 10 we pick up something, we use a methodology 11 which uses information ignoring certain other 12 types of information, then we miss certain 13 things. 14 MR. EDMONDS: Right. 15 MR. TSENG: Yes. 16 MR. EDMONDS: And so what you did 17 in -- when you got to figure 14, is you used 18 the historical behavior -- 19 MR. TSENG: Yes. 20 MR. EDMONDS: And then stepped 21 ahead one time period and did a comparison? 22 MR. TSENG: Okay -- BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 68 1 MR. EDMONDS: Essentially, if I 2 understand it, the green line -- 3 MR. TSENG: Yes. 4 MR. EDMONDS: Is fitting using your 5 bin system with the -- 6 MR. TSENG: Right. So the red line 7 is actually the table you saw. I calculated 8 cumulative capacity. 9 MR. EDMONDS: Exactly. 10 MR. TSENG: And then I fit the 11 curve. 12 MR. EDMONDS: Right. Okay, so it's 13 the same -- essentially, it's the curve -- 14 MR. TSENG: Yeah. 15 MR. EDMONDS: Fitted to the same 16 years of data? 17 MR. TSENG: Yes. 18 MR. GRUENSPECHT: So it's not out 19 of sample? 20 MR. EDMONDS: It's all in sample. 21 MR. TSENG: For the supply curve, 22 it's not out of sample, but when we do BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 69 1 projection and -- like for example, I use the 2 2005 data -- I have some results, and then I 3 can use the 2005 data to project 2004 or 2006 4 dispatching pattern, and that's out of 5 sample. Because I use the -- supply curve is 6 based on one year data, the monthly data to 7 create it -- yes, Ed? 8 MR. KOKKELENBERG: Basically, going 9 back to -- bin numbers are rather an 10 arbitrary assignment. You could've assigned 11 five bins or 5,000 bins. 12 And the assumption is when you put 13 it in, a fixed number of bins, that the steps 14 are equal -- 15 MR. TSENG: Right. 16 MR. KOKKELENBERG: In the way 17 you're using this in the model. 18 MR. TSENG: Yes. 19 MR. KOKKELENBERG: But the bin 20 number is basically a proxy for the history. 21 You look back and you say this one's used 22 90 percent, it goes in this bin, this one's BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 70 1 used 80 percent and it goes on a different 2 bin. 3 MR. TSENG: Right. 4 MR. KOKKELENBERG: So you have a 5 time series model -- in effect, although you 6 have filtered the data so that you don't get 7 all the nuances of the time series 8 variations. 9 MR. TSENG: Well -- 10 MR. KOKKELENBERG: So if you -- and 11 if you went to more bins, you would get more 12 of that detail -- 13 MR. TSENG: That's -- 14 MR. KOKKELENBERG: You're really 15 using 2004 data, because of the -- well, 16 maybe you're not -- 17 MR. TSENG: No -- 18 MR. KOKKELENBERG: Because of the 19 way the bins are picked off -- 20 MR. TSENG: Actually, it -- 21 MR. KOKKELENBERG: And so why don't 22 you just use a time series model? BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 71 1 MR. TSENG: Okay, good question. 2 The time series data is -- you have time 3 series data of very aggregated generation; 4 right? 5 MR. KOKKELENBERG: Yeah. 6 MR. TSENG: And -- 7 MR. KOKKELENBERG: But you must 8 know the capacity utilization rates to be 9 able to get them into a bin. 10 MR. TSENG: The generation number 11 is monthly data -- 12 MR. KOKKELENBERG: Right. 13 MR. TSENG: And the capacity is 14 annual number -- 15 MR. KOKKELENBERG: So you just -- 16 MR. TSENG: So when you run 17 regression analysis, you're basically -- is 18 holding capacity constant. 19 MR. KOKKELENBERG: Yeah. 20 MR. TSENG: And -- 21 MR. KOKKELENBERG: Well, you're 22 doing that throughout the model. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 72 1 MR. TSENG: No -- I'm holding 2 capacity -- 3 MR. KOKKELENBERG: But -- 4 MR. TSENG: I'm holding capacity 5 constant -- 6 MR. KOKKELENBERG: Right. 7 MR. TSENG: But a dispatching 8 pattern of each month is different. 9 MR. KOKKELENBERG: So why don't you 10 just use that dispatching pattern plant by 11 plant and generate a time series model, 12 because it would absorb all of these 13 things -- it gets away from the economic 14 issues that were raised just as well as this 15 one does, but it doesn't make the assumptions 16 that the bins are equidistant apart, as it 17 were, no matter how many bins you push into 18 it. 19 MR. TSENG: Okay. 20 MR. KOKKELENBERG: Anyway, that's 21 something for a -- you're in the middle of a 22 presentation, and we're interrupting you -- BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 73 1 MR. NEERCHAL: Could I suggest a -- 2 MR. TSENG: Yeah. 3 MR. NEERCHAL: If you can maybe -- 4 MR. TSENG: Yeah, okay. 5 MR. NEERCHAL: Next five minutes 6 and -- 7 MR. TSENG: Right, okay. 8 MR. NEERCHAL: Finish and then we 9 can come back to the discussion. 10 MR. TSENG: Right. Certainly. And 11 this is the natural gas supply curve. And we 12 have the diesel fuel supply curve. And so on 13 the supply side, we created supply curve 14 based on historical generation pattern, and 15 then we'll also look at the monthly load, and 16 we convert the monthly to average. So this 17 is just one example to let you see how it 18 looks. 19 On the left-hand side is the 20 monthly -- you look at the 2005 January 21 through December monthly electricity demand, 22 and for July, I convert the monthly to an BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 74 1 average daily load curve, and that's the 2 shape. And the interesting part is if we 3 want to capture the behavior of the 4 generators, the normal monthly time series 5 data -- actually, many people tried using 6 econometric approach, but just couldn't 7 forecast the variations in demand. 8 When demand changes, the 9 dispatching pattern of coal, especially 10 natural gas, is very different -- depends on 11 the shape of the load curve, so the monthly 12 data can never pick up this 24-hour load 13 shape to reflect the differences in the peak 14 load demand. So before I solve the model, I 15 simply make some adjustment. I remove the 16 non-fossil power plants from the load. 17 So you saw the load curve I removed 18 from the load, and I make some adjustments 19 for -- the hydro generation I assume is 20 proportional to the load curve. And for 21 hydro storage, we make adjustment to increase 22 the evening (?) demand. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 75 1 And to solve the model, the 2 solution is simple. We look at each slides, 3 like our number 1, or, you know, like say 4 1:00 in the morning, and we have this demand 5 number and we go up, the model simply traced 6 the demand to supply, and they get a bin 7 number. And then from the bin number, 8 automatically, the model would generate 9 generation numbers from oil, gas and coal. 10 So that's automatically generated. 11 So we have generation by fuel type simply by 12 equating the demand to total supply. This is 13 for a single region, and for trading block 14 like the eastern trading block, then we're 15 looking at aggregation of the supply of the 16 nine regions, and you have this trading block 17 total supply, and you have trading block 18 total demand. 19 And you equate supply to demand to 20 find out the trading block generation by 21 different fuel types. The preliminary 22 findings we have is the projections is BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 76 1 actually quite satisfactory. The out of 2 sample tests performed satisfactorily for 3 historical 2004 and preliminary 2006 data. 4 And the model is capable of 5 capturing seasonal fluctuations. The shape 6 of the load curves may affect dispatching of 7 oil-based generators, because oil-based 8 generators are more peak load demand, and 9 when we use an average daily load curve, we 10 probably can flatten the load curve a little 11 bit if we have like actual every -- actual 12 daily load curves and we just try to solve 13 it, we could see higher oil dispatching, but 14 that part -- we can actually do something to 15 adjust the model behavior or the dispatching 16 order for the oil-fired units. And -- but 17 that's not directly from the data. 18 So far, everything the model is 19 doing is straightforward and no adjustment, 20 and these are the results. For the eastern 21 region, the coal and natural gas generation, 22 the model can pick up the seasonality. And BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 77 1 actually, the 2005 month one through month 12 2 is based on historical dispatching pattern, 3 and in 2004 and 2006, the out of sample model 4 runs. 5 And for the western region, the 6 model also can pick up the fluctuations. For 7 example -- you can see in mid month of -- 8 midyear of 2006, we have a spike in demand 9 for natural gas, and the model actually picks 10 up the spike. And Texas basically followed 11 seasonality very well. Not only the 12 seasonality, also the change in the level of 13 demand, and because we -- that this model 14 actually -- because -- models -- the load 15 curve instead of the monthly time series 16 data, because when we use average data, it's 17 very hard to pick up the fluctuations in the 18 load demand, especially the peak load demand. 19 The next question people ask is how 20 does a model perform in terms of -- how do 21 you evaluate a performance? So I selected 22 two methods. One is the mean absolute BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 78 1 percent -- percentage error, and the 2 calculation is very simple. For each 3 observation, I calculated difference divided 4 by the historical number, take the absolute 5 value, and sum them and divided by the number 6 of observations. In this case, I used year 7 2005, so the h is equal to 12; I have 12 8 months data. I can calculate the MAPE. Now, 9 also I can look at the 12th month percentage 10 difference. Simply sum them up and see how 11 does it work. 12 Part of the reason I look at the 13 second approach is in the natural gas market, 14 what people look is what happens to the 15 injection to the storage, like in the summer 16 months, if natural gas storage injection is 17 below normal, and then you can see futures 18 market in December, January, moving up, 19 because -- the anticipation is there may be 20 shortage, so that's why I look at it -- the 21 cumulative -- this is one -- another measure 22 in terms of specific applications in the BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 79 1 market analysis. 2 For the mean absolute percentage 3 arrow for the 2005, for the east is a little 4 bit high, but for the west is below 5 3 percent, and for the Texas is below 6 3 percent. And why include other sample 7 calculation, which means I use 24 months to 8 calculate the mean absolute percentage error 9 for the east; for natural gas, it increases. 10 But for other months still less than 11 5 percent. Yes. 12 SPEAKER: Do you know what happened 13 there with natural gas, why did you get 14 this -- that's -- they're a double or nearly 15 double. 16 MR. TSENG: Part of the reason 17 is -- in this model, I mentioned earlier, 18 oil-fired units like diesel and residual fuel 19 units mostly are peak loaders, and I think 20 the load shape we use is probably not picking 21 up the actual load, but we're looking at 22 average. When we used average, then some of BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 80 1 the peak load power plants would not be 2 picked up by the average load shape. 3 MR. EDMONDS: You might have had 4 some spikes for a few days in there that -- 5 MR. TSENG: Right, because 6 sometimes -- 7 MR. EDMONDS: A lot of gas, but on 8 the average, it's -- 9 SPEAKER: Right. 10 MR. TSENG: For example, weekday 11 and weekend peak load demand pattern is 12 different, because the weekday peak load 13 demand is probably high, because you have the 14 commercial end use. And for the weekend, you 15 don't have the commercial, so there may be a 16 few percentage difference. 17 MR. KOKKELENBERG: If you use the 18 median absolute percentage error, you might 19 find out if that's a good argument. 20 MR. TSENG: Right. So that's part 21 of the reason. For this exercise, because 22 the purpose is just to demonstrate this BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 81 1 methodology can work, but for -- the actual 2 use for other analysis, we can actually 3 calibrate the model a little bit more, 4 fine-tune the load curve, and also adjust a 5 bin number for the residual fuel power plants 6 and diesel fuel power plants so they'll be 7 used more. 8 That actually is modeling technique 9 which can be incorporated. For the 12th 10 month percentage difference, Texas actually 11 showed very good results here. And west is 12 very good, and east, and you have positive 13 numbers for both coal and natural gas and 14 that's the reason -- yes. 15 MR. FEDER: If I understand 16 correctly, your second performance measure is 17 not sensitive at all to any monthly or 18 seasonal inaccuracies in your predictions, 19 right? 20 MR. TSENG: That's basically 21 looking at -- 22 MR. FEDER: There's a cancellation BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 82 1 there, right? 2 MR. TSENG: For this one, there's 3 no cancellation, because it's absolute value. 4 MR. FEDER: No. The second one. 5 MR. TSENG: Yeah. For the second 6 one, there's cancellation, because the 7 purpose of looking at a second one is, if we 8 have a very hot summer, for example, in 2007, 9 if we have a very hot summer, and we expect 10 based on the model performance, you can see 11 that natural gas tend to spike up, and if we 12 can pick up the spike, then it implies demand 13 for natural gas will be higher than normal, 14 and then the seasonal injection to natural 15 gas storage in the north and northeast will 16 be below normal, and that has implication to 17 natural gas prices in the market. 18 And that's why we say -- it depends 19 on how we use the model. You actually have 20 different kind of criteria to evaluate a 21 performance of the model. 22 MR. LIDDERDALE: On the 12 month BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 83 1 percentage difference, wouldn't you expect 2 the coal air to offset -- 3 COURT REPORTER: Can you use the 4 microphone, please? 5 SPEAKER: You got to speak into the 6 mic. 7 MR. LIDDERDALE: Oh, I'm sorry. On 8 this 12-month percentage error, wouldn't you 9 expect the coal air to offset the gas air, 10 which is interesting that the east doesn't 11 have that offset? 12 MR. TSENG: The reason east doesn't 13 have the offset is because the residual 14 fuel -- in the east, I mentioned a few times, 15 we're not picking up the diesel fuel and 16 residual fuel generation very well, and part 17 of the reason is somehow, the ranking of 18 those power plants is in terms of numbers too 19 high, or the dispatching order is too low, 20 and that may require some offline adjustment. 21 So the model in issue here is -- I 22 think one is we need to do some adjustment to BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 84 1 the residual fuel and diesel fuel dispatching 2 order, because from the calculation, 3 straightforward calculation, we cannot let a 4 model pick up this thing, and -- but from 5 modelist's point of view, I can change the 6 dispatching order and try to make it fit the 7 historical pattern, so that's one approach. 8 The other thing is, since we're 9 using the ranking based on the observed 10 utilization rate, we kind of miss the price 11 information. So under normal relative price 12 market structure, the model will perform 13 reasonably well, but if we have very drastic 14 change, like sometimes -- a few years back, 15 natural gas price was very high, and we could 16 see more residual fuel, and in the future, we 17 could see the relationship change. 18 And so that part is not picked up 19 in these kind of modeling framework, and 20 we're actually looking into possible ways to 21 capture the fuel switch. And there are 22 several ways to do it. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 85 1 And we're still trying. And once 2 that's done, I think the model behavior will 3 be a lot more consistent and better. 4 So the questions for the committee 5 will be -- we 'd like to have some guidance 6 and help in that area in terms of exactly 7 what is the best way to model the fuel 8 switching. And we thought of several ways. 9 One is looking at one year -- for example, in 10 2001 or 2002, we had a year where natural gas 11 was actually a lot more expensive than oil, 12 and then in 2005, oil is a lot more expensive 13 than gas. So we could look at the 14 historically -- two years and then try to 15 fish out the relative price relationship and 16 see how the dispatching pattern changed. So 17 that's one way to do it. 18 The other one is simply look at the 19 relative price of natural gas and oil and 20 look at a total generation, just look at a 21 time series data and come up with some kind 22 of coefficient, and then we can actually BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 86 1 simply apply the coefficient to the 2 dispatching equation. And basically, it's 3 like shifting the curve according to the 4 relative price. So each month, if we see the 5 price changes is drastic, then we shift the 6 supply curve -- for example, if natural gas 7 price increases, then we shift the natural 8 gas price to the left, and then in the 9 meantime, we shift the oil and supply curves 10 to the right. That's when the natural gas 11 prices have shifted to the left. 12 Yes? 13 MR. LIDDERDALE: Can't you also 14 consider treating distillate fuel and 15 residual fuel as exogenous to the dispatch 16 process, do econometric estimation of those, 17 and so you're only left with dispatching coal 18 and natural gas? In other words, distillate 19 fuel is almost 20,000 barrels a day, it 20 doesn't change. Residual fuel is certainly 21 price-elastic, but you may not be able to 22 accomplish it within the dispatch model. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 87 1 MR. TSENG: That's another option, 2 obviously. And then the question will be how 3 do we separate the residual fuel supply from 4 the dual fuel units, because some of the 5 natural gas power plants can burn residual 6 fuel? And we need to separate that. So 7 there's a lot of data issue. So that's 8 basically my presentation. 9 I have some other applications. I 10 think I will stop it here. 11 MR. NEERCHAL: I think we have 12 designated ASA discussion. Shall we do that 13 first? And then open it -- 14 MR. FEDER: Can I ask a question to 15 clarify? 16 MR. NEERCHAL: Sure. And I -- 17 MR. FEDER: Although I shouldn't be 18 allowed to -- 19 MR. NEERCHAL: I want to remind you 20 to please use your microphones -- 21 MR. FEDER: Maybe I shouldn't be 22 allowed to since I came in late, but when you BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 88 1 (inaudible) your regression models, are they 2 separate for each region or not, and is there 3 any way one region affects the other regions? 4 MR. TSENG: In the trading block, 5 like say if we look at the western block, you 6 have three regions. One region -- basically, 7 when you have trade, one region will affects 8 the other region. 9 MR. FEDER: But you still have 10 separate regressions for free chains for -- 11 MR. TSENG: Yes, each region has 12 its own demand and supply curves. 13 MR. NEERCHAL: Shall we go to 14 the -- I think Neha, you're the -- 15 MS. KHANNA: I'll keep this 16 short -- 17 MR. FEDER: We have a very tough 18 chairman this year. 19 MS. KHANNA: In the interest of the 20 number of questions -- 21 MR. NEERCHAL: No, we'll have time, 22 no problem, okay. BETA COURT REPORTING www.betareporting.com 202-464-2400 800-522-2382 89 1 MS. KHANNA: I'll keep it really 2 short. I think I have two broad suggestions 3 or comments. One is, I personally don't like 4 the use of bin numbers in a regression 5 equation-type setting, because the question 6 that -- the comment that Ed made, which is 7 essentially you're using something ordinal to 8 make a cardinal decision or to make a 9 cardinal conclusion. And that's just not a 10 very exciting way of doing this. 11 But I like the underlying 12 assumption that if a plant has a high 13 utilization rate, it must be that it's -- got 14 a low cost, so that is a good proxy for your 15 system cost. Can you not use the utilization 16 rate directly? Why even put it into bin 17 numbers? You're losing a lot information 18 there by taking 20,000 generators and putting 19 them into 10 bins, right? 20 MR. TSENG: Okay.