ML19219A174
| ML19219A174 | |
| Person / Time | |
|---|---|
| Issue date: | 08/07/2019 |
| From: | NRC/OCM |
| To: | |
| K. Hamburger 415-2022 | |
| Shared Package | |
| ML19219A167 | List: |
| References | |
| NRC-0431 | |
| Download: ML19219A174 (239) | |
Text
Official Transcript of Proceedings NUCLEAR REGULATORY COMMISSION
Title:
High Energy Arcing Fault Comment Resolution and Modeling Docket Number: (n/a)
Location: Rockville, Maryland Date: Wednesday, July 24, 2019 Work Order No.: NRC-0431 Pages 1-238 NEAL R. GROSS AND CO., INC.
Court Reporters and Transcribers 1323 Rhode Island Avenue, N.W.
Washington, D.C. 20005 (202) 234-4433
1 1 UNITED STATES OF AMERICA 2 NUCLEAR REGULATORY COMMISSION 3 + + + + +
4 HIGH ENERGY ARCING FAULT COMMENT RESOLUTION 5 AND MODELING 6 + + + + +
7 WEDNESDAY, 8 JULY 24, 2019 9 + + + + +
10 ROCKVILLE, MARYLAND 11 + + + + +
12 The Commission met in Room T2D30, Two 13 White Flint North, 11555 Rockville Pike, at 9:00 a.m.,
14 Mark Thaggard, Deputy Director, Division of Risk 15 Analysis, presiding.
16 17 NRC STAFF PRESENT:
18 MARK THAGGARD, Deputy Director, Division of Risk 19 Assessment 20 MARK HENRY SALLEY, Branch Chief, Fire and External 21 Hazard Analysis Branch, Division of Risk 22 Assessment 23 MICHAEL CHEOK, Office of Research 24 MIKE FRANOVICH, Office of Research 25 RAY FURSTENAU, Office of Research NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
2 1 KENNETH HAMBURGER, Division of Risk Assessment 2 JS HYSLOP, Office of Research 3 NICHOLAS MELLY, Project Lead 4 GABRIEL TAYLOR, NRC 5 TOM AIRD, NRC 6 MEHDI REISI-FARD, NRC 7
8 ALSO PRESENT:
9 VICTORIA ANDERSON, Nuclear Energy Institute 10 JANA BERGMAN, Curtiss-Wright Corp.
11 PAUL CLEM, Sandia National Laboratories 12 JASON FLOYD, Jensen Hughes 13 GEORGE GELLRICH, Exelon 14 ASHLEY LINDEMAN, Electric Power Research Institute 15 CHRIS LAFLEUR, Sandia National Laboratories 16 DANE LOVELACE, Jensen Hughes 17 DENIS SHUMAKER, Public Service Enterprise Group 18 MARKO RANDELOVIC, Electric Power Research Institute 19 RON REYNOLDS, Exelon 20 ROBERT RISHEL, Duke Energy 21 SCOTT SANDBORN, Sandia National Laboratories 22 KELLI VOELSING, Electric Power Research Institute 23 24 25 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
3 1 C-O-N-T-E-N-T-S 2 Welcome and Opening Remarks . . . . . . . . . . . 4 3 Project Background . . . . . . . . . . . . . . . 8 4 Comment Resolution . . . . . . . . . . . . . . . 14 5 Public Questions/Comments . . . . . . . . . . . . 29 6 Working Group Update . . . . . . . . . . . . . . 58 7 EPRI Presentation . . . . . . . . . . . . . . . 104 8 Public Comment . . . . . . . . . . . . . . . . 134 9 Overview of HEAF hazard modeling . . . . . . . 136 10 Public Comment . . . . . . . . . . . . . . . . 230 11 Closing Remarks . . . . . . . . . . . . . . . . 236 12 Adjourn . . . . . . . . . . . . . . . . . . . . 238 13 14 15 16 17 18 19 20 21 22 23 24 25 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
4 1 P R O C E E D I N G S 2 9:01 a.m.
3 MR. HAMBURGER: Okay. Good morning. It's 4 9:02. We can get started. Most of you know the drill 5 already, but I'll go through it again for everyone who 6 is new here.
7 Security, please keep your badges on you 8 while you're in the building. The only place --
9 unfortunately, we didn't get the auditorium. The only 10 place you can go without an escort is across the 11 elevator lobby to the bathrooms. Men's are on the 12 left, women's on the right.
13 If you need to go anywhere else, even down 14 out of the building, we need an NRC escort for you.
15 So we have plenty of NRC staff in the room. Just grab 16 one of us, and we'll be happy to escort you down.
17 The morning meeting is going to run until 18 approximately 12:30. At that point, any available NRC 19 staff will escort everyone down into the lobby.
20 You're welcome to use our NRC cafeteria for lunch, or 21 if you'd like to go off campus, that's fine, too.
22 My phone number is up here on the screen.
23 That's my office number, but it will ring my cell 24 phone. So, if you need help getting back in the 25 building, we'll have people shuttling you up here, but NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
5 1 if you get stuck, feel free to give me a call.
2 We have a slightly revised agenda for this 3 morning, just shuffled some things around and made 4 some more time for one of our presentations. But the 5 overall day looks about the same. Our morning meeting 6 will go till 12:30. And the afternoon meeting will 7 start at 1:30.
8 We are broadcasting the meeting by 9 webinar. And we do have people dialed into that. So, 10 when you speak, please do so into a microphone. If 11 you're at the table, make sure your table mic is on.
12 If you're in the audience, I would ask that you use 13 the standing mic next to the inconveniently placed 14 column. If you don't use the mic, the webinar people 15 can't hear you and our court reporter can't get you on 16 the audio feed.
17 Along with that, if you can remember, 18 please try to state your name before you say anything 19 so that the court reporter can record who's speaking 20 and the webinar folks know who's talking.
21 If you have any issues, come grab me at 22 the break. And I think our Deputy Division Director, 23 Mark Thaggard, would like to say a few words.
24 MR. THAGGARD: Okay, okay. Good morning, 25 everyone. As Kenny mentioned, I'm Mark Thaggard. I'm NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
6 1 the Deputy Director, Division of Risk Analysis, here 2 in the Office of Research.
3 I want to welcome everybody to this 4 meeting. You may recall at the last meeting we had on 5 this subject a couple of months ago we promised to try 6 and have more public engagement. So this meeting is 7 part of that effort.
8 As Kenny mentioned, our plan today is to 9 have two, actually two separate meetings. The meeting 10 this morning is going to focus on two areas. One, 11 we're going to walk through how we have attempted to 12 address the comments that we received.
13 I just wanted to note that, you know, 14 while we may agree to disagree on, in terms of whether 15 or not we've appropriately addressed all the comments, 16 I hope that one of the things that come out of this is 17 a clear indication that we put a tremendous amount of 18 effort in terms of trying to resolve the comments.
19 And if we, if there's disagreement in 20 terms of whether we've adequately addressed the 21 comment, hopefully you'll see that we have a strong 22 basis for, that we, our proposed resolution, that 23 there's a strong basis for it.
24 The second focus of this morning's meeting 25 is to hear from the working group in terms of give NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
7 1 them the opportunity to give us a briefing on their 2 activities. The working group has put a lot of effort 3 into this project. They've put a lot of hours into 4 it. And I think that should come out as they go 5 through their presentations this morning.
6 This afternoon is focused on the computer 7 modeling that we're proposing to do in terms of 8 developing the zone of influence. And so, if that's 9 not your area of interest, that's what's going to, the 10 focus of this afternoon's meeting. So you're going to 11 be forewarned on that.
12 We've attempted to provide all our 13 briefing materials in advance. So hopefully you've 14 had an opportunity to look at those. And we can have 15 a lively, engaging discussion.
16 I do ask that people be mindful of the 17 time so that we can hear from everyone, that we can 18 give everybody an opportunity to speak. So be mindful 19 of the time.
20 So I just wanted to say those opening 21 statements. Again, I appreciate you all coming, those 22 of you who have made it here and those that are just 23 by webinar.
24 Before I turn it over to Mr. Salley, I 25 would like to say that both Mike Cheok and I, we're NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
8 1 going to be here throughout the meeting. So, if 2 there's some question that you need to have with us, 3 we'll be around.
4 So, with that, I turn it over to Mark 5 Salley.
6 MR. SALLEY: Thanks, Mark. I'm Mark Henry 7 Salley. I'm the Branch Chief for the Fire and 8 External Hazards Analysis Branch. I'm just going to 9 take five minutes here and kind of a little background 10 and key this meeting up as we get going.
11 First slide, Kenny. And with this 12 project, there's a lot of interest. And we understand 13 that. So we've had a lot of public interaction. If 14 we go back to the May 11, 2017 Commission meeting, I 15 believe it was Scot Greenlee from Exelon at the table 16 with the Commission and said that the industry really 17 wanted to work with the NRC on this project. And we 18 took that to heart.
19 We've made as much as transparent and as 20 public as possible. Our full scale test plan and our 21 small scale plans were both announced in the Federal 22 Register for public comment.
23 We've had a number of different meetings.
24 A very important one was in April 2018. We had a 25 public workshop, if you remember. And we had a lot of NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
9 1 speakers there. We had the NFPA. We had some folks 2 from DNVGL. And there's enough information there that 3 we've actually put it in a conference proceeding.
4 It's NUREG/CP-0311.
5 We've also had public meetings this year 6 in January and March and, of course, today's meeting.
7 And again, with the transparency, it's not just the 8 public meetings. But we're also making these 9 available for webinars for people who couldn't make it 10 here.
11 Next slide, please. Just quickly talk 12 about the project and the plan, I think it's important 13 that you understand who some of the players are and 14 what their areas of expertise is.
15 We're working with the National Standards 16 of, National Institute of Standards and Technology for 17 a lot of the measurement signs. That's very 18 important. If you're familiar, I mean, they've been 19 doing this for a very long time, you know, to develop, 20 verify, and utilize measurements to quantify the 21 behavior of fire.
22 We can go back as far as 1928 when Simon 23 Enberg (phonetic) here in D.C. did the first full 24 scale fire test. It's almost 100 years ago. So we 25 bring that depth of experimental measurement from NIST NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
10 1 in their fire research group. Dr. Tony Putorti, of 2 course, is leading that up. Many of you have seen him 3 on the testing.
4 We also have our partners at Sandia 5 National Laboratories. A lot of experience comes with 6 Sandia, experts in fire measurement. We've got to 7 talk to Dr. Chris LaFleur. She's going to be speaking 8 to some of that today.
9 We've also got their experience in 10 electrical measurements, Dr. Paul Clem, who's going to 11 do some presentations on that.
12 One who's not here, but if anybody 13 remembers the past RIC this year, we had Dr. Anthony 14 Tanbakuchi, if I said that properly. But he did a 15 fascinating presentation at the RIC on using video as 16 data and how to measure particle transport and to give 17 some heat fluxes. And again, we're bringing that 18 technology to the testing program from Sandia.
19 Sandia is pretty impressive. And it's 20 interesting. They're obviously a Department of Energy 21 laboratory. And they do a lot of things outside of 22 our area. But when we tend to reach to them and ask 23 questions, what we find is a lot of the problems that 24 we're facing they've already addressed through a 25 Department of Energy program.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
11 1 So it's just amazing that we can bring 2 that level of technology. And Dr. Scott Sandborn here 3 runs the program. So he's here if you want any 4 questions.
5 It's also important the laboratory, KEMA, 6 has been a fantastic partner to work with, Frank Cielo 7 at KEMA, and very understanding, very cordial working 8 with us. He understands that we're, we have a safety 9 mission and that he is very supportive of that. So 10 KEMA has been a great partner to work with.
11 And finally, our electrical contractor, 12 it's one thing to be an electrician. But when you 13 start working with medium and high voltage, it's a 14 different area altogether. And the folks at BSI have 15 just been fantastic to work with as we put these test 16 programs together.
17 So that's just a quick overview of the 18 players and the project plan and who we've assembled 19 to actually do this work.
20 Next slide, please. You know, a second 21 part of it is, okay, so we get the experiments, we do 22 the testing, we get the data, putting it all together 23 and what does it mean to us and how do we use it.
24 Going back to the comment at the 25 Commission meeting, working with industry on this, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
12 1 obviously we have a memorandum of understanding where 2 we work with EPRI on a number of fire risk and fire 3 research programs. And that seemed to be the natural 4 fit here to work with this working group.
5 You know, Lee Iacocca, if you're familiar, 6 I've used this quote before. But Lee Iacocca, he's, 7 of course, the father of the Ford Mustang and saved 8 Chrysler from one of their early bankruptcies. But he 9 had a pretty interesting quote on projects.
10 And he said, you know, the things is when 11 you're going to run a project what you want to do is 12 first thing is hire people smarter than you. Second 13 thing is give them clear direction. And the third 14 thing is get out of their way and let them do their 15 job. And that's kind of how I'm approaching the 16 working group.
17 We've put some really good people together 18 with this group. I mean, EPRI has Tom Short. Anybody 19 who's done any arc flash work is familiar with Tom.
20 Ken Fleischer is here. He has a lot of industry 21 experience from his years in Florida. Dane Lovelace 22 from Jensen Hughes is a part of it from EPRI.
23 Shannon Lovvern from my old stomping 24 grounds down at Brown's Ferry, he brings the present 25 to the group. Marko and, of course, Ashley, staff NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
13 1 with EPRI, are also on the panel.
2 On the NRC side, we have our lead, Nick 3 Melly, who you're going to hear from a lot today, is 4 the lead on this project. We also have a senior fire 5 protection engineer, Gabe Taylor.
6 Dr. JS Hyslop, who's a member of it, and 7 JS is a very important piece because he takes it back 8 to NUREG/CR-6850 and the original origins of how 9 Appendix M and how this all came to be back in the 10 2000 timeframe.
11 One of our senior electrical engineers is 12 Kenn Miller, who's also a member of the team. Chris 13 LaFleur from Sandia is a member of the group.
14 And we're also bringing on a lot of 15 support. Some of our youngest and brightest minds 16 like Kenny Hamburger are supporting the group. And, 17 of course, Tom Koshy is bringing a lot of it. So 18 we've got a very solid group of people with a very 19 focused mission.
20 And I think it's important to point out 21 the time that these folks have invested in it. We've 22 had two meetings this year. Each one was three days 23 face to face working. And folks had to reschedule 24 vacations and reprioritize their work activities and 25 just some of them had to fly here. We helped them at NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
14 1 Jensen Hughes to get together.
2 But those meetings were hugely beneficial.
3 There was a lot of good interaction, a lot of good 4 technical discussion.
5 Also, there's been 21, at least 21 working 6 group meetings where we do webinar and phone meetings.
7 And typically, that's every Thursday from 3:00 to 8 5:00. So this group has a lot of time and a lot of 9 energy invested in this.
10 And I try to listen in to some of them.
11 And I must say that there's some really good technical 12 exchange, some very good discussions about what 13 they're doing and the questions they're asking. I 14 think it's going to overall improve the quality of the 15 program.
16 One area that we've -- I guess there's 17 been so much activity there that one area we need to 18 improve on a little bit more is communication. And 19 Kenny is going to speak to that later today. He and 20 Marko are going to put some discussion on we're going 21 to improve the communications.
22 You know, at the end of the day, I think 23 the group is focused. I mean, it's all about safety.
24 At the end of the day, I think they're all focused on 25 that and when working forward.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
15 1 So that's kind of the primer for today.
2 And it's the high level overview. And with that, 3 Kenny, I'll turn it over to you. And we can get 4 started.
5 MR. HAMBURGER: So I have a presentation 6 here on some of the comment resolution work we've been 7 doing. Over the last several public meetings, 8 stakeholders have raised issues with comments that 9 they feel have not been completely or adequately 10 resolved. We've also had some issues that have come 11 up since those public meetings. So my goal is to 12 address those and discuss how they are being 13 dispositioned.
14 I actually pulled the transcripts from all 15 of our past public meetings and went through them to 16 pull out those issues that were raised as having not 17 been completely or adequately dispositioned. So 18 that's where these issues come from.
19 We do have a public meeting, a public 20 comment period this morning. So, if there are issues 21 that you feel should have made this list and aren't 22 here, I would invite you to raise those issues at that 23 time and we'll be happy to discuss them.
24 A lot of the things that I'm going to talk 25 about here are actually, there's some overlap with the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
16 1 next presentation on the working group. And that's 2 because many of these issues have been dispositioned 3 through the efforts of the working group. So Nick and 4 Marko are going to cover some of these issues in more 5 detail when they get to their presentation.
6 One of the comments we've received is 7 regarding our 8-second test that's in our current test 8 plan. The comments were that the 8-second tests are 9 not realistic and don't reflect operating experience 10 and don't reflect plant configuration.
11 So, as part of the working group's 12 efforts, they have reviewed all of the HEAF events 13 from the US operating experience. And Nick is going 14 to talk about the specifics of those OpE events. But 15 we do see 8-second or longer HEAF events in both 16 medium and low voltage equipment.
17 But more importantly, and I think this 18 gets really to the heart of the concern that was 19 raised by the stakeholders, is that our goal is not to 20 take the results of an 8-second test and just apply it 21 generically in a bounding manner to any enclosure in 22 the plant.
23 That's the one-size model, the one-size-24 fits-all model that we have now. And that's 25 specifically what we're trying to enhance. We're NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
17 1 trying to increase the realism. We're trying to 2 increase the resolution of that model. And again, 3 Nick and Marko are going to go into detail on this.
4 But I want to call out Kenneth Fleischer, 5 who did some excellent work. He did a comprehensive 6 review of all plants' one line electrical distribution 7 diagrams and has basically broken it up into zones.
8 And the working group has created an event 9 tree that will, takes into account where the component 10 is in the electrical distribution system, what 11 protective equipment is available at that point in the 12 electrical distribution system, and is the basis for 13 developing a maximum credible fault duration. And 14 that's going to be one of the bases for the hazard 15 model that the working group is going to develop.
16 So, again, we're not taking the most 17 conservative, the longest duration tests and just 18 applying it across the board. This, the results of 19 these tests are going to be applied where they're 20 applicable and where they're appropriate.
21 And if I say anything incorrect or in 22 insufficient detail, Nick and Marko, you should feel 23 free to jump in.
24 This slide might look familiar. This was 25 presented at our last public meeting. The comment we NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
18 1 received was that our arcing location, which pull up 2 the pointer here --
3 MR. MELLY: Yeah, and one additional piece 4 of information is we are kind of discussing some of 5 these at a very high level right now. We will be 6 discussing them in greater detail as we get forward 7 through today.
8 The working group has taken a look at a 9 lot of these comments and the resolution and tried to 10 come up with a very comprehensive way of dealing with 11 them. So we'll be discussing a lot of these issues as 12 we move forward today.
13 So, if there are any lingering questions 14 after Kenny's comment resolution, feel free to ask 15 right now, however, do acknowledge that a lot of this 16 will be discussed in greater detail as we move 17 forward.
18 MR. HAMBURGER: So, in our first set of 19 tests as part of our phase two testing, we initiated 20 the arcs in the highlighted location where it says 21 primary cable connections. This was done in 22 accordance with the IEEE guide for medium voltage arc 23 fault testing at the further point away from the power 24 supply across all three phases.
25 The comment that we received was that the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
19 1 majority of the medium voltage events from OpE occurs 2 in the supply configuration and at the main bus bars 3 or the breaker stabs.
4 So there's a related concern here, which 5 is that a fault that is initiated at the breaker stabs 6 may not propagate through the tortuous path to involve 7 the aluminum that's present in the cabinet.
8 So both those concerns have been raised 9 both in terms of the prevalence of where these arcs 10 have started, as well as whether or not it will 11 actually migrate to the aluminum presence in the 12 cabinets.
13 So this has been discussed extensively by 14 the working group. The NRC is essentially in 15 agreement that investigating the supply configuration 16 switch gear is worthwhile.
17 So we have incorporated that as part of 18 our tests to be conducted in spring 2020. So we're 19 calling those our supplementary tests. And we are 20 going to test switch gear in the supply configuration 21 and on the main bus bars.
22 In regard to the operating experience that 23 shows the majority of the faults are on the breaker 24 stabs, the working group is in the process of 25 reviewing all of the, re-reviewing all of the data NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
20 1 from our phase one testing where we had nine tests 2 that were, we had the arcs initiated on the breaker 3 stabs.
4 And just from our preliminary review, the 5 members of the working group did think that there was 6 enough data on those breaker stab initiated faults to 7 use that moving forward for our modeling and for 8 estimating migration within the enclosure.
9 MR. RANDELOVIC: Kenny, just to be clear, 10 the data still has to be reviewed before we confirm 11 that we have enough information to, you know, to 12 inform the PRA guidance.
13 MR. HAMBURGER: Right. So we looked at 14 those tests. But we have not done a comprehensive 15 review of the data from those tests.
16 MR. RANDELOVIC: Right. So just to be 17 clear.
18 MR. HAMBURGER: Yes. Thank you. So, in 19 terms of that second point where the concern is that 20 a fault on the breaker may not migrate to the aluminum 21 in the cabinet, it's a legitimate concern.
22 But ultimately, the goal of our research 23 program here is to answer the Pre-GI, which is what is 24 the impact of aluminum. So, if the aluminum is not 25 involved, it doesn't particularly provide us with any NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
21 1 useful answers.
2 However, recognizing that this is a 3 legitimate point, the working group's discussions have 4 focused around possibly handling that with a split 5 fraction ZOI to address the zone of influence for the 6 aluminum components if they're involved versus the 7 breaker if the aluminum is not involved.
8 This one's an easy one. So we received a 9 comment that, from the OpE on low voltage HEAF events.
10 The events have all occurred in the load center supply 11 cubicle, which was not part of our initial test plan.
12 The NRC is in agreement. So we have modified the 13 equipment that we've procured for the test plan to 14 include the supply cubicles.
15 So I already talked about one set of our 16 supplementary tests that we have slated for spring of 17 2020, and that's the supply configuration switch gear 18 tests. We also have the generator decay curve tests.
19 And those were tests that were added to the test plan 20 at the request of our stakeholders.
21 So, again, we're planning to do those. We 22 have contracts in place, correct? We have contracts 23 in place to perform those tests in spring of 2020.
24 So this was one of the comments we 25 received in several forms and in several meetings, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
22 1 which is regarding the modeling. So the NRC and the 2 working group are looking at the use of modeling to 3 help extend the range of applicability for the test 4 data.
5 We can only test so many different 6 configurations and so many different parameters in 7 these full scale tests. So we are looking towards the 8 modeling to help the working group evaluate alternate 9 configurations, evaluate the hazard sensitivity to 10 various parameters, and estimate conditions that maybe 11 can't be directly measured because of instrumentation 12 limitations.
13 So I'm not going to address the modeling 14 here. We have an entire afternoon dedicated towards 15 the modeling approach, data needs, modeling inputs, 16 V&V. So I would ask that if there are any issues 17 related to the modeling that we hold those until the 18 afternoon presentations.
19 Measuring the conductivity of the cloud 20 has been an ongoing topic of discussion amongst the 21 working group. And it is not completely resolved at 22 this time.
23 The working group is exploring a number of 24 different measurement techniques that Nick and Marko 25 are going to speak to in slightly more detail. This NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
23 1 is going to remain a focus of the working group 2 discussions as we move towards testing.
3 EPRI had initially suggested a mock switch 4 gear that they designed. But ultimately and at the 5 current time, they felt that it has not been validated 6 to be representative of plant equipment. And there is 7 not currently any funding to build the mock switch 8 gear units. So we have not moved forward with that.
9 MR. MELLY: This is Nick Melly, Office of 10 Research. Just to add on to that discussion, we will 11 be adding more detail.
12 We have an alternative way that we will be 13 able to measure much of the same effects that that 14 mock switch gear will be looking at in a more dynamic 15 measurement technique. And we'll be covering that 16 this afternoon as part of the modeling. It's based on 17 a standard. And we think that we'll be able to move 18 that forward.
19 One important thing to note is that we 20 have a lot of ongoing activities. The working group 21 is meeting weekly. And we are discussing all of these 22 issues. The conductivity that I'm going to be 23 speaking about was actually the main topic of the 24 working group meeting last Thursday.
25 So these are real time developments that NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
24 1 we are incorporating into the modeling, and we, 2 incorporating into the testing and modeling that we 3 hope will be successful.
4 MR. HAMBURGER: One of the questions we 5 received was how are we isolating the impact of 6 aluminum. This is being done in a number of ways.
7 For starters, we are, what we're testing is aluminum, 8 and that's the data that the working group is going to 9 use moving forward to create the hazard model and the 10 zone of influence model.
11 But to understand the specific impact of 12 aluminum as compared to copper, that is probably a 13 great use of our modeling techniques that you're going 14 to hear about this afternoon.
15 We've also done some small scale testing 16 to identify the properties of the aluminum. And 17 subject to verifying that those small scale results 18 are scalable to large scale, that is a potential way 19 that we can isolate the impact of aluminum.
20 This was a recent comment during one of 21 our working group meetings. But it's also been asked 22 by some of our international partners and other people 23 as we've moved along with our phase two testing, which 24 is why did we remove the heat release rate calorimetry 25 hood that we use during the phase one testing.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
25 1 There's a couple reasons that we've not 2 included that in the test plan for phase two. The 3 first reason is that the hood does not capture the 4 initial blast.
5 So, if you've seen videos from our phase 6 one testing, the initial pressure pushes all of that, 7 those pilots of combustion out of the capture area for 8 the hood. So the hood does not capture that initial 9 blast. What it does capture is the ensuing fire if 10 there is one.
11 There is a lot of heat release rate data 12 available from our phase one testing. And what we 13 observe there is consistent with the current guidance 14 for postulating ensuing fires. Does that -- would you 15 agree with that? So --
16 MR. MELLY: For a subset of the tests, 17 yes.
18 MR. HAMBURGER: Okay. So we don't see a 19 whole lot of value in continuing to collect this heat 20 release rate data. Less important than the value of 21 the data is the logistics. But they're worth 22 addressing here. The location where we are planning 23 to initiate the arcs for our low voltage enclosure and 24 bus ducts, we are expecting the arc to move in the 25 vertical direction.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
26 1 So our test stands, I think I actually 2 have a picture of that test stand I can show you. The 3 test stand is designed to capture the thermal energy 4 moving in the vertical direction.
5 So we have -- so our test stands are 6 actually almost 14 feet high. And this precludes the 7 use of the hood. The hood won't fit over the top of 8 those racks. It's not wide enough. It's not tall 9 enough. And I'm not aware of another portable hood 10 that we could use.
11 MR. MELLY: Yeah, and one of the main 12 reasons, again, for not including the hood in this 13 design is that we feel that there is a large amount of 14 heat release rate data from all the other electrical 15 enclosure tests that we have performed. There is as 16 well data from heat release rate for the high energy 17 arcing fault test that JNRA has performed.
18 So, logistically, putting a hood up that 19 we felt didn't capture the initial blast and only 20 captures the after-effects of the high energy arcing 21 fault was less important for the modeling techniques 22 that we're going to be following in order to evaluate 23 the source term of the high energy arcing fault.
24 So we're going to try and piece together 25 all the other information that we have, keeping in NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
27 1 mind that the KEMA facility is an electrical testing 2 facility and not a fire test facility.
3 These portable hoods are on the smaller 4 end of the scale. And they're usually overwhelmed by 5 the initial blast of the high energy arcing fault, as 6 well as having a lot of challenges with being in an 7 open air environment. We're losing a lot of the smoke 8 due to crosswinds.
9 And even having an ensuing fire lasting 10 for a long time in an outdoor test environment causes 11 some concerns environmentally with folks at the EPA.
12 We're not in a controlled environment like 13 NIST with smoke scrubbers and things like that. So, 14 having these longer duration fires is a challenge in 15 this, in the single test program.
16 MR. HAMBURGER: So I apologize. I can't 17 pull up my instrumentation rack diagrams. The file 18 doesn't want to open.
19 But they're about 14 feet high and about 20 6 feet wide. So they're really, with the hood that we 21 used in phase one, there's really no place to put that 22 in the test cell.
23 And there's also a lot of complications 24 that come along with the use of those oxygen 25 consumption calorimetry hoods. It takes about 100 man NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
28 1 hours1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> to construct and deconstruct that hood. We have 2 to transport compressed gas cylinders, sensitive gas 3 analysis equipment. And there's frequent calibration 4 burn.
5 So it's not as simple as just throwing the 6 hood in there. So, if we're not going to get data 7 that we feel is valuable out of it, we've decided not 8 to include that in the test plan.
9 So this is a big one. And there's a bunch 10 of people working on this as we speak. And that is 11 communication with our stakeholders. This is a large 12 and growing program. There are a lot of pieces to it.
13 And they come together in many ways.
14 So we've been asked to be much more clear 15 about what those pieces are, how they fit together, 16 the deliberations and details of the working group's 17 conversations and how they're dispositioning each task 18 and subtask.
19 So, to that end, we are working on a 20 website to improve communication in real time with all 21 of the interested stakeholders. And this is going to 22 contain detailed data of all the tests and research 23 program components, how they fit together, the inputs 24 and outputs to each component, where the data flows 25 from component to component.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
29 1 Unfortunately, due to personnel 2 availability issues in the last couple weeks both from 3 EPRI, NRC, and Sandia, we do not have that ready at 4 the moment. But you should expect to see that in the 5 next two to three weeks.
6 So we have taken that comment to heart.
7 And we are diligently working to put together a 8 comprehensive plan that details everything that's been 9 asked for.
10 So that is the end. And again, Nick and 11 Marko are going to cover some of these issues in much 12 more detail during their presentation.
13 But if you have questions that you'd like 14 to ask about comments that you feel were not 15 dispositioned or you'd like to bring up now, I would 16 invite you to come to the mic stand and ask away.
17 MR. GELLRICH: Ask away we will. George 18 Gellrich from Exelon.
19 You know, first off, you know, I just want 20 to acknowledge that this is a phenomena that has to be 21 understood by the NRC and the industry. You know, we 22 do agree we have to understand the significance and 23 safety comes first.
24 I heard that you talked about, that you've 25 looked at the sites and now understand the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
30 1 configuration of the sites insofar as generators, 2 aluminum location, you know, is there single breaker 3 isolation from the safety bus.
4 So, with that data, like what's it telling 5 you about the number of plants susceptible? Is it all 6 of them, some of them? Do we know that?
7 MR. MELLY: So we're going to discuss that 8 in a little bit in the upcoming presentation. We've 9 essentially tried to break it down into zones. And 10 we're going to feed that through our model.
11 In terms of our, which plants are 12 susceptible and which plants aren't, it really comes 13 down to a lot of underlying factors, whether they have 14 failures in their system or whether the design of the 15 systems explicitly. Whether a generator circuit 16 breaker is installed in the plant design can have a 17 large effect of whether we see these generator fed 18 faults persisting for extended durations, which is 19 really what we're worried about in terms of the HEAF, 20 one of the primary parameters.
21 So I can't give you a number off the top 22 of my head as to 30 plants are at risk versus 70 23 plants are not. But we're trying to take into account 24 the plant design in the way that we will be doing the 25 modeling, allowing for those contractors or anyone who NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
31 1 is postulating HEAF scenario to incorporate their 2 unique plant system into the potential outcome or the 3 zone of influence with an associated HEAF event.
4 MR. GELLRICH: Okay. My concern would be 5 is I'm hearing this third hand but, you know, I'm 6 getting an understanding that the number of actual 7 plants affected would be small, maybe less than 20.
8 And it seems to me that the research 9 you're doing should first understand what the 10 configurations are for the plants that really are 11 susceptible and then develop the research plan to 12 understand, you know, for those specific 13 configurations what the, you know, what research needs 14 to be done or what the plants could do to mitigate the 15 situation.
16 To, you know, to go forth as you are right 17 now and cast a wide net, you're spending a lot of 18 research money that may not produce valuable results.
19 You know, you talked about areas such as this testing 20 is worthwhile where we think this is legitimate.
21 Well, you should really focus your research on exactly 22 what you know and what you don't know.
23 So, you know, I'm very concerned that 24 you're really putting the cart before the horse --
25 MR. MELLY: When you say susceptible, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
32 1 you're referring to the presence of aluminum?
2 MR. GELLRICH: Not just, presence of 3 aluminum is one aspect. But the configuration of the 4 plant is a whole other one. Like many plants, the 5 generator, you know, goes out to the grid, but the 6 safety buses are all fed from an AUX transformer or 7 start-up transformer.
8 Those I think would be taken off the 9 table. I don't know because I haven't seen the data 10 that you guys have pulled together as to what plants 11 are susceptible and which ones aren't.
12 MR. MELLY: So we're going to be trying to 13 incorporate that into the design of the methodology.
14 I wouldn't say that those situations are not 15 susceptible to having this extended duration high 16 energy arc fault.
17 MR. HAMBURGER: Do you want me to pull up 18 the zone --
19 MR. RANDELOVIC: There is an EPRI white 20 paper that we just issued today. We went through 105 21 plants in the U.S. And we classified all of those 22 plants into eight different categories. The 23 categories are based on, you know, susceptibility to 24 HEAF, long duration HEAFs, and to the impact in the 25 class one buses.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
33 1 So we make a difference in that white 2 paper. It was issued today, the difference between 3 generator fed designs and to the more complicated and 4 the more I would say less prone designs to the long 5 duration HEAFs.
6 MR. GELLRICH: Right --
7 MR. RANDELOVIC: So there is a group of 8 eight different designs --
9 MR. GELLRICH: Okay. So that, I think 10 that data would be critical to understand what 11 research you would want to do.
12 MR. MELLY: Well, absolutely. And we are 13 trying to incorporate all that.
14 One important distinction also to make is 15 that we're not just talking about the susceptibility 16 to a safety bus having a high energy arcing fault.
17 There are some cases potentially out there where a 18 non-safety bus can still have a high energy arcing 19 fault where it is a risk significant contributor at 20 your plant if it damages safety systems in the zone of 21 influence or in the area of a hot gas layer 22 contribution. So we're trying to look at all the 23 pieces together in order to inform our modeling.
24 MR. GELLRICH: Yeah, and I think making 25 statements like you think that that's non-safety NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
34 1 related bus has a high risk significance, you know, my 2 experience would say that that probability is pretty 3 low.
4 MR. MELLY: But it's going to be all plant 5 dependent, scenario dependent depending on where that 6 bus is, what room that bus is in, and what is nearby.
7 One of the other distinctions to make that 8 we learned this actually at the last working group 9 meeting was that so far we don't have a very robust 10 understanding of which plants have aluminum 11 susceptibility, which plants do not. We have --
12 MR. GELLRICH: Exactly.
13 MR. MELLY: But we have an informal survey 14 done by NEI which showed that there was a fairly large 15 contribution of aluminum in the plants.
16 But speaking with one of our experts on 17 the working group panel, he said it was very difficult 18 for him even to ascertain whether he had aluminum in 19 his main switch gear supply. He had outdated specs 20 for his cabinet itself, the 4160 volt cabinet.
21 However, when he dug through the information, it was 22 found that he did have aluminum in that bus work for 23 the main supply.
24 So, if you asked ten people in his plant, 25 nine of them might have said, no, we don't have NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
35 1 aluminum in this cabinet. However, by digging in and 2 doing the work, they found out they actually did. So 3 that is one issue --
4 (Simultaneous speaking.)
5 MR. GELLRICH: -- and understand 6 specifically. We should hold ourselves accountable as 7 an industry to really understand the configurations 8 and then base the testing on that configuration.
9 I think putting the research before we 10 understand what we have is exactly what we did with 11 GS191. Okay. GS191, with strainers, apply to all 12 plants. And then the NRC did research. And then each 13 plant had to do their own research to validate what 14 aspects of that research applied to their specific 15 sites and which ones didn't. And it was an immense 16 waste of resources.
17 And, you know, my point is I think we're 18 following that pattern again. And we should step 19 back, understand what the configurations are. We owe 20 it to you to understand that configuration and then 21 figure out what realistic testing we need to do.
22 I'm concerned that the availability and 23 test facility is putting time pressure on all of us.
24 And we're going to do this research, and we're going 25 to step back, and we're going to be in a position of NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
36 1 not understanding what the real applicability is to 2 all the sites. And then the sites are going to have 3 to go off and do their research on their own and 4 recreate research, which is not where we want to be.
5 MR. HAMBURGER: So I think to some extent 6 we've done that. I mean, the working group has looked 7 at every single plant and broken them down by 8 susceptibility and what the HEAF would look like in 9 various locations in that plant. Are you suggesting 10 looking at other aspects of the configuration?
11 MR. GELLRICH: No. I'm hearing we just 12 finished this paper today.
13 MR. HAMBURGER: Okay.
14 MR. GELLRICH: So how can we be doing 15 testing if we haven't even looked at the results in 16 the paper?
17 MR. TAYLOR: This is Gabe Taylor, NRC 18 Office of Research. So I'm not, I haven't looked at 19 what EPRI published today. But January last year they 20 did publish two white papers. And in one of those 21 white papers, they did do an evaluation that broke 22 down the electrical distribution system into different 23 zones.
24 Now, I'm guessing the difference between 25 what was recently released and that paper was they NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
37 1 focused on safety-related equipment in the -- no?
2 Ashley is going to correct me.
3 MS. LINDEMAN: This is Ashley Lindeman 4 from EPRI. So, just to back up one step, the 5 consequential events that we've seen in the operating 6 history are on the non-safety switch gear. And they 7 have been due to a generator fed fault.
8 With that said, the EPRI white paper that 9 was published, Rev 0, had around 60 sites. But we got 10 105 sites, so pretty much everything. And we 11 classified them from high susceptibility to really 12 lower or not applicable.
13 So, in the lowest susceptibility, there 14 are 63 plants. And those are plants that feed from 15 the station, the AUX transformer, you know, from the 16 grid. So it's not running from the house loads.
17 So right away, you know, 60 percent of the 18 plants aren't really susceptible to the generator fed 19 fault. An additional 11 have circuit breakers, 20 generator circuit breakers that may interrupt the 21 generator fed fault.
22 So now we're left with roughly 30 sites.
23 And six of those sites I think or less have safety 24 buses that may come from the house. So the population 25 really gets smaller and smaller. So I just wanted to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
38 1 provide some context for everyone.
2 MS. VOELSING: Ashley, when you started 3 you said that the consequential events that we've 4 seen.
5 MS. LINDEMAN: Yes.
6 MS. VOELSING: You mean consequential in 7 terms of having long durations --
8 MS. LINDEMAN: Oh, sorry, yes. When I say 9 consequential, I mean from a damage perspective or the 10 HEAF may have damaged components outside of the 11 component origin, not necessarily from a risk 12 perspective. So --
13 MR. MELLY: Yeah, and Ken just put 14 something on the screen here. This is kind of a 15 preview of some of the information we were going to 16 cover in the next presentation is we are trying to 17 take these actively into consideration of the 18 methodology that we are going to develop for the high 19 energy arcing faults.
20 We're going to try and take into things 21 like whether the plant has the generator circuit 22 breaker, what the design of the plant itself is, what 23 it's fed by, whether it's the unit auxiliary 24 transformer or the site auxiliary.
25 We're taking into account things like the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
39 1 circuit breaker, whether it's apply or load. And 2 we're trying to let the actual in-plant design inform 3 what our overall duration potential is, as well as 4 then how that feeds to the overall zone of influence.
5 So, for some of these plants that are less 6 susceptible or for some of these plants that don't 7 even have aluminum, this might not be an issue that is 8 of a grave concern. However, we are trying to let the 9 methodology lead us to that discussion and that 10 decision rather than bringing that to the forefront.
11 MS. VOELSING: Yes, so this is Kelli 12 Voelsing from EPRI. If I could, I don't want to put 13 words in anybody's mouth, but kind of summarize what 14 I think I'm hearing, which is, you know, Ashley has 15 presented that, you know, we're talking less than 30 16 plants that have a bus susceptible to this. And of 17 those, you know, not all of those are safety-related 18 buses and, therefore, may be less risk significant in 19 the PRA.
20 And, you know, as George pointed out, of 21 those, you know, 10, 20, whatever it is number of 22 plants that might be susceptible to this, we haven't 23 validated which of those have aluminum and which of 24 them don't.
25 And we absolutely agree that from a, you NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
40 1 know, a methodology, a PRA standpoint, a technical 2 approach, that, you know, understanding that and 3 building all that into the methodology is the right 4 thing to do and we support that.
5 I think, you know, maybe what George is 6 saying is, you know, how generic really is this on the 7 front end. You know, I think we could do some more 8 work as an industry to answer that question.
9 MR. MELLY: Yeah, the only clarification 10 I want to make to this discussion that we're having is 11 we're throwing out numbers like 30, 40, 20 plants are 12 susceptible to generator fed faults.
13 We are as a working group also taking into 14 account things like circuit breaker failure or 15 protection scheme failure into effect as well.
16 So a plant that might not seem susceptible 17 can still have a lower probability of having 18 additional failures, which might then open the door 19 for having this potential. But we're trying to deal 20 with it in the probabilistic way by looking at what 21 that rate of failure is and incorporating that into 22 the methodology.
23 That may address the issue explicitly.
24 And it may lower the potential risk enough that it 25 might not be an issue for a specific site. But we're NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
41 1 hoping that the methodology and the zone of influence 2 and the hazard modeling will actually inform that 3 decision.
4 MR. RANDELOVIC: So the model that we are 5 building right now is basically very similar to 6 internal events, PRA, where you have an event and you 7 have, you know, different protection systems and you 8 are crediting different protection systems to mitigate 9 those events. And then based on the success or 10 failure of those, of the protection system, you have 11 different end states.
12 In this case, we are doing exactly the 13 same. You are crediting, fully crediting the 14 protection scheme based on the specific designs that 15 you have in your plant.
16 So it's fairly complex in the integrated 17 methodology that we will provide that will basically 18 provide insights on, you know, bridging the gap 19 between the OE, what we are seeing in the OE, and 20 incorporating the design of the plants --
21 MR. MELLY: Yeah, that's --
22 MR. RANDELOVIC: -- in the methodology.
23 MR. MELLY: That's one very important 24 aspect. In this specific approach that we're doing, 25 we're actually pulling in insights from the operating NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
42 1 event experience and the fire events database. So 2 we're trying to merge the plant design, the OpE of 3 what we've seen, and allow that to inform the result 4 and how we're going to be performing our scenario 5 modeling.
6 MR. HAMBURGER: So I think we may have 7 just addressed this. But we did have a question on 8 the webinar from Tian Fenglin, which is there are very 9 few aluminum buses in the plant so why are we 10 researching aluminum, if you want to add anything to 11 that.
12 MR. MELLY: So, based on the informal NEI 13 survey, we've seen that there are aluminum concerns in 14 medium voltage switch gear, low voltage switch gear, 15 and the bus ducts themselves, as well as bus ducts 16 enclosures.
17 So the generic issue process is triggered 18 when it is two or more plants. And we have identified 19 two or more plants that do have aluminum concerns or 20 the potential of concern for aluminum interaction 21 within a high energy arcing fault environment.
22 MR. HAMBURGER: Go ahead.
23 MR. RISHEL: Bob Rishel from Duke Energy.
24 So, on the comment about the vertical effect of the 25 high energy arc fault, so my OpE is all, all our NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
43 1 faults I have seen or had in our plants have been 2 horizontal events. Very little damage has ever 3 occurred in the vertical direction. And so I was 4 wondering why we're putting so much effort into that.
5 It seems to be --
6 MR. HAMBURGER: Are you referring --
7 MR. RISHEL: It seems to be technically 8 wrong.
9 MR. HAMBURGER: Are you referring to bus 10 ducts as well or are you just referring to switch 11 gears and load centers?
12 MR. RISHEL: -- switch gear, you know, 13 cabinet, high energy arc faults.
14 MR. HAMBURGER: Okay.
15 MR. RISHEL: If you look at breaker stab 16 interactions, that's a horizontal event.
17 MR. HAMBURGER: Okay. So you're not 18 referring to the bus ducts. You're okay with --
19 MR. RISHEL: Not referring to the bus 20 ducts.
21 MR. HAMBURGER: Okay. Great. So --
22 MR. RISHEL: But your testing 23 configurations, like cabinet high energy arc faults.
24 MR. HAMBURGER: That's correct. Okay. I 25 just want to clarify the scope of your question.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
44 1 MR. RISHEL: And I would point out also in 2 another separate comment on your fault propagation, 3 having been intimately familiar with the Robinson 4 events, the Robinson event was stopped by the 5 transformer itself.
6 So, if you're talking about protection, 7 don't forget that those transformers are not 8 infinitely capable of propagating faults. They do 9 have a point at which they will stop.
10 MR. MELLY: Right. And in that Robinson 11 event, that point as discussed in a previous slide, we 12 were talking around, on the order of 8 to 12 seconds.
13 MR. RISHEL: Correct. But there was no 14 vertical damage by the way. And --
15 MR. MELLY: Yes.
16 MR. RISHEL: -- additionally, the way that 17 that fault stopped was that the transformer failed.
18 MR. MELLY: Yes. And we are trying to 19 take these into consideration in our modeling 20 approaches. We're looking at the OpE as to whether it 21 occurred on a breaker stab versus a different point 22 within the cabinet. And we're going to be trying to 23 incorporate that potentially with a split fraction, as 24 Kenny mentioned, as to where we believe that occurred.
25 However, from testing, we do see that it NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
45 1 is possible to orient the arc in the vertical 2 direction depending on where the arc is initiated 3 within the cabinet, what the magnetic field is in the 4 cabinet itself, and where that arc initiation point 5 is.
6 But you are correct in that we see the 7 OpE, a lot of it is at the breaker stab. And we're 8 trying to take that into account when we move forward 9 with the methodology. So, hopefully, we will be 10 addressing that issue.
11 MR. HAMBURGER: Gabe, do you want to say 12 anything about where we're initiating the arc for low 13 voltage and why we're expecting it to be a vertical?
14 MR. TAYLOR: Sure. So, for the test we 15 have planned, we have a piece at Westinghouse DS 16 series gear. And the gear that we procured, it has 17 vertical bus bars and horizontal bus bars on the main 18 buses. And the verticals are aluminum. And per the 19 procurement document, the horizontals are copper.
20 So, because of the vertical aluminum bus 21 bars, we are planning to initiate them at the top, 22 because that's the predicted, expected migration point 23 of the arc. So, for that case and that piece of 24 equipment, we planned on initiating it there.
25 MR. RISHEL: But I would just question, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
46 1 you know, so most of the cabinets I have seen, the 2 connections are towards the bottom or maybe the middle 3 not, there's no connections towards the top of the 4 cabinet. The bus bars come in or the connections come 5 in the top and typically go down. And the connections 6 are, the stabs or whatever are towards the bottom or 7 middle at the worst.
8 MR. TAYLOR: So --
9 MR. RISHEL: So why initiate the fault at 10 the top?
11 MR. TAYLOR: So, for the -- you're right.
12 For the incoming and outgoing cable connections to a 13 low voltage gear, typically the connection points are 14 going to be at the lower portion of the gear.
15 However, internally where you have the 16 drop, the runbacks to the breakers, they're going to, 17 for feeders, they're going to be at any height where 18 you have a feeder breaker. For the supply, they're 19 typically going to be somewhere in the middle or lower 20 portion.
21 So, depending on how the power flow is 22 within the breaker and where the arc is initiated at, 23 that's going to predicate where the arc is going to 24 migrate to. It's either going to stay where the arc 25 initiates or the magnetic forces are going to push it NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
47 1 away from the incoming power supply.
2 So, based on the design of the equipment 3 that we procured, all that type of information is kind 4 of influencing us that the arc's going to migrate to 5 the top of that cabinet.
6 MR. HAMBURGER: And I just want to add one 7 more thing. Although we are now placing two 8 instrumentation racks above the cabinet because we're 9 expecting that thermal energy to be directed upwards, 10 we are still measuring at horizontal stands. So we're 11 not neglecting the horizontals.
12 MR. RISHEL: Actually I'm just worried 13 we're going to draw the wrong conclusions about the 14 industry and electrical cabinets if our testing 15 emphasizes vertical direction versus horizontal 16 direction.
17 MR. MELLY: So we're not saying that the 18 energy, we're not saying that it's going to be 19 vertical in every single case. We are just -- we take 20 a look at the cabinet that we did procure. And we are 21 taking our best estimate as to where we think that arc 22 will have the highest impact on our instrumentation.
23 And we believe right now with the cabinet 24 design that it will go in the vertical direction. So 25 we're placing multiple instrumentation test stands NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
48 1 there. However, we will be placing them in the 2 horizontal direction as well if the arc actually moves 3 in the horizontal direction away from the cabinet.
4 Now, that being said, just because this 5 specific test we believe it's going to go vertical, we 6 are not going to be influencing the methodology to say 7 that all high energy arcing faults will show the 8 highest risk in a vertical direction. That is not the 9 intention of linking the modeling to the testing.
10 MR. HAMBURGER: It's essentially an --
11 MR. RISHEL: So why do it?
12 MR. HAMBURGER: Because --
13 MR. RISHEL: Why spend money on that?
14 MR. MELLY: Because we want to make sure 15 that we have adequately measured the source term. And 16 we believe in this test the ejecta or that plume that 17 we're going to see the highest impact or, is going to 18 go in the vertical direction.
19 So, by measuring the source term 20 adequately, we can inform the modeling so that we can 21 know what the source term would look like in any 22 direction of the cabinet because it is probabilistic 23 as to where that arc will initiate.
24 MR. HAMBURGER: It's essentially an 25 experimental concern. We have a limited number of NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
49 1 test stands. We have to put them where we think we're 2 going to measure the most valuable data.
3 MR. MELLY: So, and for instance, the test 4 that we just performed in 2018, all of the, most of 5 the ejecta was in the horizontal direction away from 6 the incoming power supply in the cabinet.
7 We had our two test stands in the 8 horizontal direction at three-foot, six-foot. But 9 that single test where it shot in that direction at 10 the back of the cabinet is just for that experimental 11 test 7. It allowed us to measure the source term of 12 that HEAF.
13 But we're now not saying that all medium 14 voltage is going to go to the back of the cabinet in 15 one direction. We're allowing the modeling to say 16 where in that cabinet we'll expect to this zone of 17 influence and what could potentially be in that zone 18 of influence. So we're not trying to --
19 MR. RISHEL: Shouldn't the direction be 20 largely by the cabinet construction --
21 MR. MELLY: Configuration dependent.
22 MR. RISHEL: -- configuration and also two 23 things? One is how strong is the short, right?
24 MR. MELLY: Yes.
25 MR. RISHEL: What is the capacity of that NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
50 1 short, right?
2 MR. MELLY: Yes.
3 MR. RISHEL: You know, some are incidental 4 contact where there's just fractions of inches 5 connection, which is enough to start the event, and 6 some could be more. So it seems to me that's pretty 7 important input.
8 MR. MELLY: I agree. And that's why we 9 are altering the current, the duration of the tests, 10 and the voltage so that we have a full picture of the 11 important parameters that are going to impact the 12 energy release of that event.
13 MR. RANDELOVIC: And we are discussing 14 about procuring the vertical racking designs versus 15 horizontal racking designs --
16 MR. MELLY: Yes.
17 MR. RANDELOVIC: -- which is also 18 different configurations and we have to be able to 19 provide additional insights of how --
20 MR. MELLY: Yes.
21 (Simultaneous speaking.)
22 MR. MELLY: That is a very important 23 aspect that configuration is something we're trying to 24 take into account. Not only are we testing the, where 25 we tested in 2018 at the rear of the cabinet. We're NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
51 1 going to be testing the supply configuration, which 2 has a different enclosure around it, a different set-3 up of the bus bars themselves. We're testing the low 4 voltage in this vertical.
5 But we're not allowing the tests, 6 experimental set-up, to influence what we're going to 7 be coming up with in terms of the generic model. So 8 we're trying to take these into account as we move 9 forward.
10 MR. HAMBURGER: Mike, did you want to say 11 anything? Hit the button.
12 MR. CHEOK: So we hear all these comments.
13 And I think it's important. I think what we also need 14 to keep in mind is that we are now in the pre-generic 15 issue phase. So what we are doing is we are not 16 saying that this is a generic issue. We have seen 17 that an issue could be of safety significance, could 18 be or maybe.
19 We are not saying that it applies to all 20 plants. As a matter of fact, we know that it doesn't 21 apply to all plants.
22 We are just trying to determine -- so this 23 is an issue, and we will take risk into effect. We 24 will determine what the test parameters are as a 25 different, on a different track.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
52 1 We are also looking at the frequencies of 2 the different configurations and the frequencies of 3 initiation given the different configurations. And we 4 are also, you know, taking into account and working 5 with our counterparts in EPRI on what mitigative 6 actions that the industry is doing to lower the 7 frequencies.
8 And, you know, all those will be taken 9 into account at the next phase. I mean, the next 10 phase is given the fact that this is potentially what 11 we see out there with the different configurations and 12 what are the frequencies and if plants have these 13 configurations what are they doing to lower the 14 frequencies in terms of the source term, in terms of 15 the targets.
16 I think what we are trying to do here is 17 get enough data to inform ourselves whether this is a 18 generic issue or not.
19 MR. GELLRICH: Yeah, Mike, George 20 Gellrich. Just my comment on that would be that, you 21 know, from our perspective it would be better in the 22 pre-generic phase of this is that we should understand 23 what configurations are out there first and then do 24 the research versus, you know, going off, hey, we 25 think this might be a configuration or it might go NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
53 1 horizontal or this or that.
2 I mean, it's, you're casting a very wide 3 net. And quite frankly, I don't know whether you have 4 to or not. You might be able to use the research that 5 was done by EPRI and yourselves and determine, hey, 6 these are the specific configurations we really need, 7 these are the plants that show the issue, and deal 8 with those.
9 I just think we're coming up with out-of-10 the-box ideas and we don't know whether they're 11 applicable or not.
12 And what the result will be is you'll 13 publish it, and it will get everybody spun up around 14 the safety significance of this issue. And it won't 15 be realistic. When you apply it to the plants, you'll 16 find out, oh, some, number one, it doesn't apply. And 17 then there's going to be a handful of plants that have 18 to do something. So that would be my opinion.
19 Thanks.
20 MR. CHEOK: Thank you. So I think I just 21 wanted to make one comment on that. We will not 22 publish the results of the tests without context.
23 That's something we will not do.
24 We will publish the results. You know, we 25 will continue to work with all our stakeholders as to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
54 1 properly quantify the risk significance. And the risk 2 has to be, as you all say, plant specific. And it 3 will not apply to all plants.
4 And, you know, so we will have, we will 5 not -- we understand that, you know, the data by 6 itself is not the results of what a HEAF would look 7 like. We understand that the risk incorporates and 8 encompasses a lot of different elements.
9 And we will take into account all those 10 elements before we even publish any results or make 11 any kind of recommendation as to whether this is a 12 generic issue. And even then, you know, we will have 13 all our results characterized in the risk-informed 14 way.
15 And so the other thing I was kind of 16 listening to was, you know, the placing of the 17 different detectors, whether it's vertical or 18 horizontal, is not that, relatively speaking, not that 19 expensive to do. It's the test itself that causes the 20 expense. Whether we put additional detectors, 21 vertically or horizontally or something, that doesn't 22 add that much more to the cost.
23 So, if we were going to perform a test, it 24 would be more beneficial for us let's just say to, you 25 know, get all the data we need. And a lot of the data NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
55 1 may not be applicable. And we will be sure to state 2 that.
3 But if we perform a test and not take all, 4 not get all the data we can get, it might be an 5 opportunity lost if we have to re-perform a test 6 again.
7 MR. HAMBURGER: Okay. Tom, do we have 8 anything on the webinar? No? Okay. If there are no 9 more questions, at this point we go to break here 10 unless you want to start.
11 MR. MELLY: I just want to add one more 12 clarification that I just thought of in terms of the 13 directionality of the zone of influence as we've been 14 talking about.
15 So, in the 6850 methodology, it's three-16 foot, five-foot, three-foot horizontal, five-foot 17 vertical. The working group thoughts initially during 18 our discussion was that that difference in the three-19 foot, five-foot, having an increased vertical zone of 20 influence due to these HEAFs, was largely due to the 21 fact that that methodology was created because it was 22 an investigation of post-event experience.
23 So you're seeing not only the damage from 24 the initial blast. But you're seeing the larger 25 thermal damage from the ensuing fire. And it's very NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
56 1 difficult to look at an accident investigation after 2 the fact or root cause investigation and separate 3 those two out and say that this was the thermal fire 4 after versus the blast at the very start.
5 Our initial thoughts are that there 6 shouldn't be an initial difference in the horizontal 7 versus vertical direction. So we're not talking about 8 potentially three-foot, five-foot or things like that.
9 We're thinking of these things actively, but maybe we 10 don't have a difference in the vertical towards 11 horizontal.
12 And we're taking all that into account in 13 the way that we're going to be looking at the data, as 14 well as performing the methodology and modeling as we 15 move forward.
16 MR. RISHEL: So, on the zone of influence, 17 so to speak, so, you know, I'll go back to the 18 Robinson event.
19 So we actually sent Sandia the cables that 20 were in the zone of influence, actually three inches 21 above the top of the cabinet. And Sandia sent us back 22 a letter that says those cables are not faulted, that 23 they're perfectly capable of continued service.
24 And there was a little heat damage on the 25 outer jacket of the cables. But that would tend to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
57 1 say that there's, in that case, almost no vertical 2 impact.
3 And in fact, there was a panel two and a 4 half feet away from the faulted cabinet that had 5 splatter on it and had a plastic gauge cover on it.
6 And that plastic gauge cover was not melted. It was, 7 and it showed some damage. And it had some splatter.
8 It wasn't melted. The cabinet was undamaged.
9 There was a construction fence some 10 distance away that was melted. So, when the hot 11 material deposited on, essentially, you know, like a 12 coke bottle, construction was great, that thing was 13 gone.
14 But everything else, there was no damage 15 other than the inside of the cabinet was totally 16 burned out. And the next cabinet over also showed 17 damage.
18 So, when we talk about ZOI, we need to be 19 very careful about it, because it isn't three-foot, 20 one-foot cone of death. It's probably almost no cone 21 of death frankly.
22 MR. MELLY: And I hope that the working 23 group efforts and our efforts to increase the realism, 24 as well as put forward the methodology, will address 25 those types of issues.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
58 1 MR. HAMBURGER: Yeah, I think everybody is 2 in agreement that the one-size-fits-all model is not 3 the state of the art. And we're working to get away 4 from that.
5 Okay. Let's take a break until 10:30. I 6 forgot to put the sign-in sheet up on the podium over 7 there. So, when you come back in at 10:30, if you 8 could do me a favor and please sign in so we have an 9 accounting of who attended today.
10 If you're NRC staff and you wouldn't mind 11 escorting people up and down from the lobby so they 12 can get some coffee, I would appreciate that.
13 (Whereupon, the above-entitled matter went 14 off the record at 10:09 a.m. and resumed at 10:27 15 a.m.)
16 MR. HAMBURGER: So, Nick and Marko are 17 going to provide an update on the Working Group status 18 and their activities. And we'll be expanding on some 19 of the topics that were addressed in the first 20 presentation.
21 If you didn't get a chance to sign in, on 22 your way back in please just do that before lunch, if 23 you don't mind.
24 So, Nick, go ahead.
25 MR. MELLY: All right. So, we wanted to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
59 1 give a kind of high-level overview of some of the 2 activity of the Working Group. As Mark said, we do 3 have a lot of activity going on. We have our weekly 4 meetings as well as we've had several in-person 5 meetings to discuss how all of these pieces are going 6 to come together. We are going to be discussing the 7 mission of the Working Group, some of the PRA modeling 8 approaches that we are going to be investigating, as 9 well as updating. We're going to look at some of the 10 lessons learned from the operational experience 11 review. We're going to be discussing some of the 12 testing approaches, as well as discussing our planned 13 project plan that Kenny has alluded to previously.
14 So, out of this information we have 15 discussed previously as part of the Working Group, the 16 mission/charter of this Working Group is to improve 17 the understanding, the risk from high energy arcing 18 faults in nuclear power plants. The goal of our group 19 is to understand some of the key factors. That 20 includes the occurrence of high energy arcing faults 21 or the frequency, as well as the severity. So, in 22 that regard, the zone of influence.
23 And we want to advance the high energy 24 arcing fault PRA modeling. We are going to be using 25 the experimental data, operating experience, and NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
60 1 engineering judgment in coordination with the 2 frequency to get a more robust model. We are not 3 talking about creating a one-size-fits-all. We are 4 trying to push forward, so that we can actually 5 understand plant insights, OpE, the specific 6 characteristics that are going to affect a specific 7 high energy arcing fault scenario, so that we can 8 tailor that to get a unique zone of influence for that 9 particular place within your plant.
10 We're going to also, then, be talking 11 about analyzing the plant impact and risk 12 implications. That gets towards the pilot plants that 13 are going to be part of the Working Group activities 14 into further understanding what the risk from these 15 events are. That is part of the generic issue 16 assessment plant currently, to perform pilots.
17 And one of the issues that we previously 18 did bring up of trying to do a better job of 19 understanding those plants that could be at risk, 20 those plants that do have aluminum, I would highly 21 recommend that the industry does take a more 22 standardized or robust measure of trying to figure out 23 which plants do have aluminum, where it is, what 24 configuration it is in, because that will only help us 25 in terms of selecting pilots and performing those NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
61 1 pilots as we move forward. So, if there is an 2 industry effort or a renewed industry effort to get 3 better data associated with where we do see aluminum 4 and where these sensitivities could be, we welcome 5 that type of work.
6 As I said, we do hold weekly meetings.
7 These are more involved. We pushed them forward from 8 biweekly, one hour, to weekly, two-hour meetings 9 because we've seen that we just have so much material 10 to cover and we need to keep active progress in order 11 to push our testing forward, as well as to understand 12 the data that we are collecting.
13 We have discussed the project members. We 14 have Ken Fleischer, Dane Lovelace, Shannon Lovvern, 15 Tom Short, Marko, and Ashley. We have Dr. Hyslop, Dr.
16 Chris LaFleur, myself, Kenn Miller, and Gabe Taylor.
17 And that's just the main team members.
18 In terms of support, we have a lot of 19 other members who are doing some data analysis at 20 Jensen Hughes. We have Ken Hamburger. We are pulling 21 in all the resources that we need because this is an 22 unevolved process. We're putting a lot of resources 23 towards this to solve the issue.
24 MR. RANDELOVIC: Now the next couple of 25 slides, we will be discussing some of the Working NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
62 1 Group activities.
2 The first one that we would like to bring 3 right now is the HEAF methodology report. Basically, 4 the report will contain the details and the test 5 results, the operating experience, insights. And that 6 report will be basically used to implement the HEAF 7 PRA methodology for the plants.
8 So, we have initiated this report by 9 performing initially the extensive review of all of 10 the OE in the U.S. Out of that extensive review of 11 the OE, we have gathered a lot of insights regarding 12 the fault durations, the fault locations. Is it in 13 the supply or load configuration? The accident 14 sequences that we are getting the insights for to 15 inform our decision trees and event trees.
16 So, that is the first kind of section of 17 the report. That data from the operating experience 18 is not only used to develop the methodology, but also 19 to inform the testing.
20 Then, there's another section which is the 21 HEAF frequency. So, we reviewed the 30 events. We 22 sat down for hours and days looking at every single 23 event and classified this as Bin 16.A, 16.B, 16.1, 24 16.2, and like whether it is an arc flash, arc blast, 25 or HEAF. So, we have been through all of that effort, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
63 1 and we believe that at this point we have a fairly 2 good knowledge regarding those events and how those 3 events will be used for the frequency calculations.
4 And lastly, the risk model development.
5 So, as I said, initially, we had Version 1 of the EPRI 6 white paper where we had the different designs of the 7 electrical distribution systems in the plants in the 8 U.S. Revision 2 has been issued today. And so, we 9 are going to use the design, specific designs from the 10 plant with the operating experience to come up with 11 the realistic model that moves away from one-size-12 fits-all and, basically, incorporates and integrates 13 the insights from the OE and the protection scheme 14 from the electrical distribution systems.
15 MR. MELLY: This is kind of a snapshot of 16 some of that work that we performed. We did spend a 17 significant amount of time trying to understand these 18 events. And you can see up here on the screen, we did 19 an extensive review, and we're trying to classify 20 those. We're trying to understand the duration, which 21 is one of our primary parameters.
22 And for some of these events, we found 23 that it was very easy to actually come up with a 24 defined duration from the event report itself or 25 potentially the LER, or even going back and collecting NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
64 1 more information. So, we have been able to nail down 2 the duration for some of these, and hopefully, that 3 will feed into our modeling approach.
4 You see here that we have some bolded 5 event durations which are from 4 to 8 seconds. Some 6 of these are alluding to the fact that these are 7 generator-fed faults. And that 4-to-8-second duration 8 is an estimation.
9 We have a current Working Group action 10 item to try and pin that down a little bit better, in 11 order to reduce our uncertainty associated with that 12 time duration, but it is fairly difficult to 13 understand these generator-fed faults and what the 14 actual duration was because they're from a long time 15 ago. They may not have had digital fault recorders in 16 place, and it is going to be generator-specific. So, 17 we are trying to pin that down in order to understand 18 the OpE and what the OpE is showing us, and how we're 19 going to incorporate that into our model.
20 One of the important distinctions that we 21 need to make here that has been brought up in previous 22 Working Group meetings, previous public meetings, is 23 that we are not focusing on these millisecond fault 24 occurrences. They're not part of our current HEAF 25 frequency bin. If we're talking about an arc fault NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
65 1 only persisting on the order of cycles, that's not 2 going to be included in the frequency for how we're 3 treating high energy arc faults. Those are types of 4 events where their circuit protection does work as 5 designed, and the arc is immediately extinguished, 6 causing very little damage. That is not what we're 7 talking about in terms of HEAF and is not what we are 8 going to be modeling or testing. One of the important 9 aspects is that the OpE did show more generator faults 10 than we had typically thought were out there just from 11 our extensive review.
12 What we have shown here on the screen here 13 is how that fault duration, the current breakdown of 14 what that duration looks like. We see several unknown 15 events, but, then, on the X-axis, you are looking at 16 the duration of fault in terms of seconds. And you do 17 see that we have a large item pegged at that 4-second 18 estimated duration. That's identifying these 19 generator-fed faults.
20 And we do hope that the methodology that 21 we're going to present will capture those plants that 22 are more susceptible to generator-fed faults versus 23 plants that are not susceptible or less likely to be 24 susceptible. And we're hoping that this kind of 25 distribution potential for the duration will allow us NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
66 1 to pinpoint where susceptibilities in the plants are 2 in terms of which cabinets or which lineups can 3 actually have this extended-duration high energy 4 arcing fault event.
5 MR. RANDELOVIC: Just to add something, so 6 because we have these fairly large uncertainties on 7 the generator-fed faults, I took an action to go back 8 to some of the plants that experienced the generator-9 fed faults, asking for specific duration. I only 10 found one plant that provided some additional data and 11 confirming that the fault was between 4 and 6 seconds.
12 The remaining plants, unfortunately, didn't have 13 enough details or information to supplement this 14 assessment.
15 MR. MELLY: And the current estimation is 16 based on some information that was gleaned from the 17 event that occurred at Maanshan, where some of the NRC 18 electrical experts said that a generator-fed fault in 19 that condition could have persisted somewhere 4 to 8 20 seconds. So, we hope to do a little bit more work in 21 that area to pinpoint that, if we can't get more 22 information from the plants.
23 MR. RANDELOVIC: Do you want to expand a 24 little bit the estimate, how we came up with the 25 estimated durations for the other events? Two NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
67 1 seconds, 4 seconds. So, we basically went into every 2 single event and looked. We had Ken Fleischer and 3 Dane, electrical experts, looking at what type of 4 system, protection systems, performed to mitigate a 5 fault. And based on that knowledge, and based on the 6 selective coordination and the protection scheme 7 designs, they were able to provide kind of an 8 estimated, the maximum estimated duration, given the 9 sequence and the sequence of events for a given HEAF 10 event.
11 MR. MELLY: Moving forward again, as part 12 of this, we were looking at low voltage. And part of 13 our question here was, what do we test for low 14 voltage?
15 In the April Working Group of 2018, it was 16 said that these 8-second tests for low voltage are 17 unrealistic because we have so many levels of 18 protection before we can actually have this 8-second 19 fault. What we have identified from the OpE is that 20 there are several cases out there where we can have a 21 prolonged-duration event with the low voltage if we're 22 at a certain point of the protection scheme.
23 So, we specifically are trying to link our 24 testing parameters in terms of current and voltage to 25 those areas where we might see an extended-duration NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
68 1 event. For the upcoming tests, we're going to be 2 performing at 13.5 kA for these low-voltage tests 3 because it was seen from the operating experience and 4 from actual plant configuration that we could run into 5 a case where you can have between, I believe it was 8 6 kV, 13.5 kV, that could last for an extended duration 7 up to 8 seconds, looking at the specific configuration 8 of that breaker design. So, we are trying to take 9 into account these situations.
10 Another piece of information that we hope 11 to be able to glean from, if we go back to the low 12 voltage, from the extended-duration event is, we can 13 only perform the medium-voltage tests at the KEMA 14 facility up to 4 seconds. The generator is simply not 15 large enough to perform anything larger than 4 seconds 16 on medium voltage. But, for low voltage, we can go up 17 to 8-10 seconds, just based on the generator size.
18 So, we are hoping that some of the low-19 voltage information at extended durations, we'll be 20 able to inform some of the modeling for medium 21 voltage. Again, that's an extrapolation. It will not 22 be straightforward, and that's going to be an action 23 item for the Working Group to even evaluate where we 24 can make that potential extrapolation.
25 So, those are the two reasons why we do NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
69 1 see a benefit for testing some of the low-voltage 2 cases at 8 seconds, which is our maximum duration of 3 our test series.
4 Next slide.
5 MR. RANDELOVIC: So, for the medium-6 voltage tests, EPRI, following our assessments of the 7 OE, made a comment back in January regarding the 8 location of the fault. In the previous run of tests, 9 we have located the arc wire at the buzz bars. And 10 looking at the OE, basically, the majority of the 11 medium-voltage switchgear HEAFs occurred at the 12 breaker slabs, where you basically don't have the 13 aluminum. So, the majority occurs at the breaker 14 slabs. Some of the events, we have the fault 15 initiated in the main buzz bars; in only one event on 16 the back bars.
17 So, we have brought those insights to the 18 Working Group, and we have discussed on the path 19 forward for the next round of tests. For the breaker 20 slabs location, we have decided to look at the OECD 21 data. It looks like there were some tests performed 22 at the breaker slabs. However, we still have to 23 evaluate the amount of data that could be used to 24 define the zone of influence. And there's an action 25 item for Energy Research to send to the Working Group NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
70 1 and send the other data. And then, we need to 2 evaluate if we have enough information to come up with 3 a realistic zone of influence. If not, then we will 4 have to decide how to proceed.
5 So, the testing for the medium-voltage 6 tests in 2020, the Working Group has decided to 7 proceed with the tests, placing a wire on the main 8 buzz bars to replicate several of the OE events. And 9 we are also looking at changing the configuration of 10 the design of the switchgear, basically the horizontal 11 racking versus the vertical racking design of the 12 breakers to assess the impact of the volume and the 13 configuration difference between those two. And I 14 believe this has been added in the 2020 test plan.
15 MR. MELLY: Yes, I will have a slide kind 16 of detailing what that looks like in terms of our test 17 matrix.
18 MR. RANDELOVIC: So, if we have sufficient 19 data from OECD testing, we will be testing with two 20 different configurations of the switchgear design, 21 locating the wire at the different locations to 22 replicate the OE. And we believe that that will 23 provide sufficient information to come up with a model 24 that replicates the operating experience.
25 MR. MELLY: Yes. And one important thing NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
71 1 to note, also, is that we are trying to maintain that 2 1-to-1 parameter investigation as part of this test 3 plan, so we can really investigate the differences in 4 the damage or in the differences in the data by only 5 varying one parameter.
6 So, as we talk about testing on the supply 7 side configuration, as it where here, we're going to 8 be doing the supply side -- we're going to be doing 9 the decrement curves also in the supply side 10 configuration, so that we can compare them as we move 11 forward.
12 And this is also one area where the 13 Working Group is actively improving the way that we 14 will be conducting the test. If we go back to the 15 slide, that blue arrow, at the last Working Group, in 16 terms of understanding how the power is going to be 17 flowing through the cabinet, may change depending on 18 Working Group recommendations. We will be still 19 initiating the arc in the supply side configuration.
20 However, we are currently discussing options to best 21 reflect the realism of what that cabinet will look 22 like when it is arced and how it will relate to 23 operating experience, and how it will arc actually in 24 a real plant. So, that is one area where we are 25 currently still hammering out the details of where NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
72 1 we're going to be initiating the arc and what that 2 cabinet will look like in the KEMA test facility.
3 So, as we discussed previously, one of the 4 important aspects of this is frequency. As part of 5 this Working Group, we're looking at all the OpE.
6 We're evaluating the energy arcing fault frequency as 7 well. We will be coming up with new frequency values 8 apart from those in 2169, NUREG-2169 currently.
9 Our main focus here is that we're going to 10 be looking -- we're trying to understand the high 11 energy arcing faults events that have occurred and 12 what we're calling high energy arcing fault events as 13 we move forward. There is some clear distinction 14 there that we -- on the screen here you see Arc Fault 15 Class 1, Arc Fault Class 2, and Arc Fault Class 3.
16 That is trying to get in to count the difference 17 between what we're calling arc flash, arc blast, and 18 HEAF.
19 The arc flash event is typically things 20 that are going to be captured in the Bin 15 electrical 21 fire bin, because these are where arc fault protection 22 schemes work. These are cycle-type events, 23 millisecond-type events. These are not the high 24 energy arcing faults. It's a quick pop; the event is 25 over. We don't have a sustained arc.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
73 1 In the arc blast event, we do see that 2 there might be a slightly extended duration, on the 3 order of 30-40 cycles, typically less than a second in 4 duration, and there's no ensuing fire after this 5 event. That is one important distinction to make with 6 the arc blast and what distinguishes that from the 7 high energy arcing fault events.
8 These high energy arcing fault events, 9 Class 3 are what are going to make up the Bin 16.1, 10 .2, .A., .B frequency and how we're going to relate 11 the modeling. There may be some end-states in our 12 trees that we showed previously to look at the blast.
13 However, it will have a unique modeling approach to 14 basically show that there is not an ensuing fire 15 associated with these events.
16 The classic event that we could use an 17 example for that case is the event that occurred in 18 2017 at Turkey Point. We did see that we had an arc 19 fault. We see pictures of it here on the screen. It 20 was, roughly, a half-second event. We saw some 21 pressure damage over here on the right. That is the 22 door from Turkey Point. However, there was no ensuing 23 fire associated with that event. The question 24 becomes, is that a HEAF? How do we deal with that?
25 We don't want to model this huge, ensuing NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
74 1 fire event, a rapid ignition, hot gas layer-type 2 concerns, if that's not what the OpE is showing us.
3 So, the Working Group is actively taking these into 4 consideration, so that the frequency is a true 5 reflection of how we will model it in terms of 6 scenario modeling.
7 Just visually, you can see the depiction 8 of these, the differences in the fault classes on the 9 screen from the quick, pop-type event where we're 10 going to blow a panel open; we'll see some scorching 11 versus, potentially, the arc glasses one level up, we 12 see some potential pressure increase damage, versus 13 what classically would be defined as HEAF large damage 14 state; could be potential pressure questions as well, 15 and this zone of influence.
16 One important distinction that I don't 17 think we've been discussing is we're not just looking 18 at electrical cabinets. We are also dealing with the 19 bus ducts as well as the bus duct enclosures, and the 20 impact that those can have on high energy arcing 21 faults.
22 So, the Working Group's charter is both 23 electrical enclosures, low voltage, medium voltage, 24 and the bus ducts. This upcoming test series does 25 have five tests on bus ducts in various NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
75 1 configurations.
2 Marko, anything to add on that?
3 MR. RANDELOVIC: Next slide.
4 MR. MELLY: Next slide.
5 MR. RANDELOVIC: What we currently have in 6 the 6850 is 3x3x5. Take that zone of influence, and 7 you are applying to it every single switchgear in the 8 plant. No matter what kind of protection you have, 9 you are still applying zone of event, zone of 10 influence, in every single switchgear.
11 So, what we have been doing, we have been 12 reviewing 150 power plants. We wrote a white paper on 13 this, and we grouped eight different designs, as we 14 explained already several times, how susceptible they 15 are to the long-duration faults and what would be the 16 response to HEAFs, based on in which zone those HEAFs 17 occur.
18 In addition to reviewing and incorporating 19 into more the plant design, as I said, we are also 20 incorporating the insights from the OE and how we are 21 going to split fractions, loads versus breaker, lead 22 versus supply. And then, as I said, you know, we have 23 the majority of those events occurring at the breaker 24 step. So, we will still have to divide and split 25 additional, have additional split fractions in our NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
76 1 event trees.
2 So, this is definitely moving away from 3 one-size-fits-all. It does incorporate the OE 4 insights and credits the protection scheme directly in 5 the event trees.
6 MR. MELLY: Yes, and we also took the 7 operating experience. In our last Working Group 8 meeting, we went through every single event and tried 9 to link it with one of these zones. Where in the 10 plant did it occur? Which zone can we call that? And 11 how does that allow us to investigate the event?
12 Also, this picture on the screen is just 13 we're using it for visualization purposes. It's not 14 every plant. We have different pictures for 15 potentially different plants and different designs.
16 You can flip to 1E versus non-1E in some cases. Some 17 plants will have a generator circuit breaker. This is 18 just used as an example of how we're going to present 19 the information in our report.
20 MR. RANDELOVIC: So, if you have a HEAF, 21 this, the Class 1 and switchgear, you have one, two, 22 three, four breakers that are currently not credited.
23 We're just using the zone of influence, some zone of 24 influence, in doing the zone of influence in the PRA 25 model.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
77 1 Now the more breakers you have, of course, 2 the less likelihood of having a long-duration HEAF 3 there is. And that's going to be reflected if you go 4 to the next slide.
5 MR. MELLY: Yes, and we can see on -- can 6 we go back one second? We can see on this slide that 7 we can now identify in your plant where you might have 8 a susceptibility to having these long-duration, 9 generator-fed events based on your plant design. And 10 we're going to allow the modeling to show that.
11 Next slide.
12 MR. RANDELOVIC: And so, this is an 13 example event tree. Basically, it's exactly the same 14 as the long-event PRA. You have an event and you are 15 modeling your protection scheme throughout the fault 16 trees.
17 And here, the end-state will be the 18 duration of the HEAF. And you can see in this fault 19 tree that you are creating different breakers in your 20 protection scheme to reflect the duration of the zone 21 of influence of the HEAF, depending where you are in 22 the design of the plant.
23 How we are incorporating here the 24 operating experience, for example, here you have 25 supply and load. We are seeing that 80 percent of the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
78 1 HEAFs occurred in supply, and only 20 percent occurred 2 on the load configuration. So, this is a complete 3 integration of the operating experience and plant 4 side.
5 MR. MELLY: Yes, and also, we have shown 6 here -- we haven't shown the values that we're going 7 to be putting in here, and these duration values that 8 we have currently on this table are only placeholders.
9 They're "for example" purposes only. We've not 10 actually done the work to decide which durations will 11 link with what part of your plant in your in, and 12 then, the zone of influence, what will be affected.
13 This is all preliminary. We're just making sure that 14 we have the structure in place, so that we can 15 incorporate unique plant design.
16 MR. RANDELOVIC: So, the testing is going 17 to come into play here. The testing is going to 18 inform those numbers. So, we would have a table with 19 the zone of influence for copper and for the zone of 20 influence for aluminum. Because we will be testing 21 two different switchgear designs, vertical versus 22 horizontal, we may even have a further split for data 23 for the zone of influence to see if there is any 24 difference.
25 MR. MELLY: And I will note that we've NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
79 1 discussed that the Working Group is actively working 2 on these types of things. At our first Working Group 3 meeting, we had a version of that preliminary table 4 that we just showed, and it completely changed on the 5 second Working Group meeting. And we feel like it is 6 now more robust. But when we do have further Working 7 Group meetings, I do expect that we're going to make 8 tweaks to that table and that structure in order to 9 make sure that we adequately have enough hooks in the 10 model to capture all the unique plant aspects.
11 All right. Now what we have on the screen 12 here is kind of the overview of our measurement 13 techniques, so what we will be measuring in this 14 upcoming test series. We're going to be looking at 15 the temperature and heat flux at multiple distances 16 away from the arc location. We are hoping that that 17 will aid in the dynamic zone of influence creation as 18 well as some of the modeling approaches that we will 19 hear about later today. And the hope is that that can 20 link to some of the planned work at Sandia in terms of 21 the fragility and the criteria testing that we will, 22 again, discuss in the afternoon.
23 We're also going to be measuring pressure 24 internal to the cabinet. What's mentioned here is 25 that it's improved measurement techniques. We are now NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
80 1 using the fiber optic cables in this phase 2 testing.
2 We're seeing a lot less noise and a lot more pure 3 pressure measurements from internal to the cabinet.
4 And as a Working Group, we're looking at 5 the potential to measure the impact on room pressure, 6 how the internal cabinet pressure could, then, feed to 7 potential overpressurization of a larger space in 8 terms of an electrical enclosure room, a switchgear 9 room, like we saw at the Turkey Point event.
10 We're looking at the damage zone, both 11 visually as well as the furthest extent of damage that 12 we saw, both the thermal, the fire damage and smoke, 13 and the physical, if we throw any cabinet doors or 14 have shrapnel, in terms of effects like those.
15 We will also be measuring conductivity, 16 which we'll be discussing in a few more slides greater 17 detail, using Sandia measurement techniques as well as 18 other options. And we'll go into a little bit more 19 detail of what we mean in terms of how we're measuring 20 that conductivity.
21 As part of the first test series, we saw 22 that aluminum did short out some of the KEMA 23 components and their incoming power supply. That was 24 a surprise to us in the first series of testing. So, 25 we have given a lot of additional effort in this phase NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
81 1 2 testing program to try to understand what, if any, 2 the smoke impacts have on conductivity in terms of 3 surface deposition, as well as potential arc or the 4 breakdowns for it to bear in terms of creating a 5 secondary arc in a secondary power supply component 6 within the room. We will discuss how we're looking at 7 that, and there are some pros and cons to some of the 8 methodologies.
9 Next slide.
10 MR. RANDELOVIC: We're getting the 11 conductivity. We just need to acknowledge that there 12 was no OE event.
13 MR. MELLY: Yes.
14 MR. RANDELOVIC: There was no operating 15 event where we actually saw the combustion cloud 16 causing a flashover.
17 MR. MELLY: Yes.
18 MR. RANDELOVIC: So, this investigation is 19 only because of the KEMA event, right?
20 MR. MELLY: Yes. The KEMA event and, 21 anecdotally, some questions that were raised from some 22 of the foreign OpE in terms of the event that occurred 23 at Onagawa. It was postulated that there were two 24 arcs, and it's unclear whether it was the same power 25 supply that had the two arcs. So, we're looking into NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
82 1 that to try to assess whether we actually can induce 2 an arc from this cloud of aluminum. We haven't seen 3 it in OpE right now, but it is a question that has 4 been raised.
5 Also, one of the things that we did not 6 bring up on this slide is we are also going to be 7 measuring EMI effects. Again, we've seen no failures 8 from HEAF OpE in terms of EMI effects. However, it 9 was a question that was raised from the NRC research 10 staff and DEE, that if we will be performing these 11 high energy arcing fault tests, it would be a shame 12 not to collect relevant EMI data. So, we will be 13 adding probes to collect relevant data. However, the 14 current plan for the Working Group is we have no 15 action item to incorporate that into any PRA damage 16 models for HEAF unless we see surprising or unexpected 17 results from those measurements.
18 We're also going to be looking at the mass 19 of the vaporized material in terms of how much 20 aluminum is vaporized during the test, trying to 21 correlate that to the duration of the event itself, as 22 well as some potential chemical energy release values.
23 That's going to be linked with the validation of some 24 of the computer models as well as the theory equations 25 for the vaporized material, as well as trying to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
83 1 evaluate the approximate energy release, the total 2 energy release from the high energy arcing fault as a 3 subset of the electrical energy inputted as well as 4 some of the chemical energy from the oxidation of 5 aluminum.
6 We have cable sample materials put in 7 place on our test stands. We're also going to be 8 looking at some of the byproduct in terms of 9 conductivity as well. We're going to be using some 10 carbon tape, and we used aerogel in 2018. For the 11 2019, we're going to be still using the carbon tape as 12 well as a silicon/quartz-type measurement device, 13 simple because the carbon tape did give us good 14 results; the aerogel is a robust material. However, 15 we found that the carbon tape gave us enough or 16 adequate information, so we didn't need to add the 17 aerogels. And the carbon -- or the silicon/quartz 18 I'll get to in a further slide, as to what exactly 19 we're looking at in terms of why that's been added.
20 Again, this is trying to answer the 21 question of what is potential conductivity on that 22 aluminum that could be potentially deposited on 23 surfaces. So, we see this white cloud being deposited 24 on equipment or on adjacent walls, and we're trying to 25 understand what effect that could have for electrical NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
84 1 components.
2 Some of the preliminary information is 3 that the material further away from the test sample, 4 that deposited material is all aluminum oxide. And if 5 that is the case, then, hopefully, this test device 6 will answer the question final, whether that even is 7 a concern. If it's aluminum oxide and non-conductive, 8 it may not cause failure on adjacent components from 9 the deposition, and we will have the answer as to 10 whether it needs to be incorporated into the PRA 11 failure model.
12 Again, we will not measuring the heat 13 release rate in this phase 2 of testing, based on the 14 lesson learned in phase 1.
15 Next slide.
16 So, in terms of the surface conductivity 17 measurements, we do have fairly good experience with 18 this type of measurement. It's a passive measurement.
19 We're going to be taking interdigitated resistivity 20 measurement structures, essentially, looking at what 21 the resistance on this material is prior to the test, 22 and then, what the resistance is after the test, when 23 we have deposited aluminum or aluminum oxide or 24 byproducts on this surface. It will help us evaluate 25 any impact of the deposition of material in terms of NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
85 1 electrical failure. We'll get into some of the pros 2 and cons of that in a later slide.
3 MR. RANDELOVIC: So, I would like just to 4 explain a little bit the issue with this measurement.
5 So, this idea was brought to the Working Group maybe 6 three weeks ago.
7 MR. MELLY: Yes.
8 MR. RANDELOVIC: And so, the Working Group 9 members have been asking how the data that is going to 10 be collected is going to be used for the PRA model.
11 So, how do we ensure that we collect the data that is 12 prototypical to the plant conditions, actual plant 13 conditions? I think the NRC has an action item to 14 provide that information to the Working Group, so that 15 we can determine if this is a valid approach for the 16 conductivity measurement.
17 MR. MELLY: Yes, and that's an active, 18 ongoing Working Group action item. I believe Gabe has 19 taken the lead on that one. If he wasn't sitting 20 here, that's what he would be doing upstairs.
21 So, again, we're trying to get to the 22 potential for failure of electronic equipment if it is 23 exposed to the arc ejecta through this type of device.
24 Next slide.
25 We're also trying to take a more active NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
86 1 measurement of the conductivity using a Sandia product 2 as well. This is a mesh design. We have a mesh 3 design on these air conductivity measurements for the 4 EMI rejection. This is not the device that is going 5 to be measuring EMI. This is more looking at the 6 conductivity of the air itself during the test.
7 Again, this is a more recent development in terms of 8 testing parameters, and it is, again, an action item 9 of the Working Group to evaluate how the measurement 10 that we will be receiving will relate to the 11 functional failure or potential functional failure of 12 components within actual plant design.
13 Next slide.
14 And we're going to be discussing these, I 15 believe, later in terms of modeling, and we have 16 additional backup slides going into pretty good detail 17 on what these devices are and how they are going to be 18 used.
19 One of the other main aspects that Marko 20 is going to cover is this flashover concern in terms 21 of, can we induce an arc in a secondary piece of 22 equipment? Marko, do you want to discuss this?
23 MR. RANDELOVIC: So, following 2018 24 testing and before the January meeting with the 25 Working Group, EPRI has been raising some concerns NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
87 1 regarding the conductivity measurement, and especially 2 the static conductivity measurement.
3 So, for the January meeting with the 4 Working Group, we took an action on the EPRI side to 5 come up with a concept that could be used to measure 6 the flashover that would be more prototypical to what 7 you would expect in the plant. We brought the concept 8 to the Working Group in January, and the Working Group 9 really liked the concept.
10 So, what EPRI did, we came back from the 11 Working Group and spent some time designing a mock 12 switchgear test unit that, basically, preserves some 13 specific characteristics of the switchgear in the 14 plant, given the restrictions that it has to be small; 15 it can be powered by KEMA, and we can't use the DAC 16 from Sandia.
17 So, we came up with the design, a detailed 18 design, of the mock switchgear test unit that, as I 19 said, represents a typical switchgear with respect to 20 voltage, bus bar spacing, and standoff insulators to 21 ground. It's portable, reusable, and doesn't require 22 excessive power.
23 We have brought this to the Working Group, 24 once we finalized the design, and the Working Group 25 really liked the concept. It has been verified by NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
88 1 NRC's electrical experts, EPRI's electrical experts.
2 However, we have not brought this concept to the 3 switchgear manufacturer to really have the last checks 4 to ensure that, you know, this type of test equipment 5 would actually replicate really the flashover in the 6 switchgear.
7 MR. MELLY: Right, and this was one of the 8 pieces that the Working Group did discuss in our last 9 meeting in March. And it was a benefit to have this 10 type of equipment. It was found out last week that we 11 wouldn't actually have or EPRI couldn't provide this 12 piece of equipment for the upcoming test series.
13 However, we have tried to work with Sandia in order to 14 get data that is active data, rather than a binary 15 bounding result of, yes-no, we have flashover, in 16 terms of trying to answer this type of question with 17 the air breakdown strength, another active measurement 18 device that we're going to be putting in the test 19 cell.
20 So, while this unit will not be in this 21 upcoming test series, Sandia is confident that we can 22 get relevant data to inform this flashover effect, and 23 we're actively sharing that information with the 24 Working Group. Again, this is the last week's 25 discussion, that this would not be coming to the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
89 1 upcoming tests due to the last step not being 2 performed on verification with switchgear 3 manufacturers or funding issues.
4 MR. RANDELOVIC: Funding and, also, the 5 tests are in August.
6 MR. MELLY: Yes.
7 MR. RANDELOVIC: So, it takes eight weeks 8 to manufacture those components.
9 MR. MELLY: Right.
10 MR. RANDELOVIC: DCI, I have been 11 discussing it with DCI. It takes about eight weeks.
12 MR. MELLY: Yes.
13 MR. RANDELOVIC: First, given that we 14 started to design it in February-March, and it took 15 some time to actually collect all that information and 16 come up with some kind of design, while the last step 17 was not performed, but, then, the problem is, you 18 know, if the test is in August, even if they performed 19 the last check, and even if they want to buy the 20 equipment, it takes about eight weeks to even deliver 21 this test equipment.
22 MR. MELLY: Right. So, we have identified 23 an alternative to answer this flashover question. We 24 shared it last week with the Working Group. This is 25 a device we currently have available from Sandia, and NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
90 1 we believe it is now linked, our alternative is linked 2 with a standard. It's slightly altered to facilitate 3 the test environment that we're going to be in, and it 4 will answer, hopefully, the same question as this mock 5 switchgear box with more data that we can post-6 processing use.
7 Again, active area of discussion with the 8 Working Group. We just presented it last week. We're 9 working on this test plan.
10 Do we have additional slides to that 11 effect? Hold off, yes. And we will have additional 12 slides as to our alternative to this mock switchgear 13 in the afternoon.
14 MR. STONE: Hi. This is Jeff Stone from 15 Exelon. A quick question.
16 Are you designing any of this test to see 17 if there are potentials to reduce the scope of this 18 review? For example, if you find that medium or small 19 or low-voltage cabinets don't have a large difference 20 in zone of influence for aluminum versus a non-21 aluminum breaker, to try to limit the scope that we're 22 working on, or are we just going to continue forward?
23 MR. MELLY: There is no current plan to 24 reduce the scope because the test matrix is designed 25 as the minimum number of tests that we needed to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
91 1 answer the question. Now, if we see a huge -- if we 2 see something different as we start running these low-3 voltage tests, and we do not see a damage shape at 4 all, it could inform potentially future testing, as 5 not part of the GI. But in terms of reducing the 6 scope, I would say not for the pieces of equipment 7 because we're only testing five bus ducts and four 8 low-voltage equipment, and then, four additional tests 9 on medium-voltage equipment. So, there's not very 10 much room to reduce the scope.
11 However, if we're talking about actively 12 learning from our results, I would say yes. For 13 instance, if we see that we do not have any impact 14 from this switchgear test unit or the holdoff strength 15 in open air, we might be done with doing that test.
16 We've evaluated that it has no impact, and we don't 17 need to perform additional tests.
18 On the other hand, if we see that we see 19 this huge impact on open air, we're now investigating 20 putting our device inside an electrical enclosure or 21 inside a box, so we can limit the open air and get 22 more close to realism.
23 MR. STONE: So, you have high confidence 24 that aluminum is going to make a dramatic difference 25 in low- and medium-voltage cabinets right now?
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
92 1 MR. MELLY: Based on our current 2 understanding from phase 1 of tests, I would say yes.
3 The low-voltage tests that were performed in phase 1 4 on copper, it was my thought that I could drastically 5 reduce the zone of influence, based on the fact that 6 it was low voltage. However, one of the last tests 7 that we ran, the low-voltage piece of gear with 8 aluminum in it had a much larger zone of influence 9 than we anticipated. So, it's still an unknown at 10 this point.
11 MR. STONE: Was it larger than the copper?
12 MR. MELLY: Yes.
13 MR. STONE: Thank you.
14 MR. TAYLOR: I think just one point to add 15 to that. We don't know at this point how big of an 16 increase the hazard would be with aluminum versus 17 copper. And the second piece of that, even if it is 18 an increase to the hazard, we still don't know how big 19 of an increase on the risk it has. So, that's all 20 things and actions that have to be done as part of the 21 GI program to really assess the risk impact on the 22 plants before it would move forward to the regulatory 23 stage or it gets kicked out of the GI program, and it 24 gets resolved in an update to fire PRA guidance.
25 MR. HAMBURGER: In terms of the question NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
93 1 about high confidence that there is a difference 2 between aluminum and copper in low-voltage gear, we're 3 conducting four low-voltage tests with aluminum. So, 4 just in terms of minimum number of tests to 5 investigate that question, I think we're probably in 6 the realm of a reasonable number of tests to answer 7 that question of is there a difference.
8 MR. RISHEL: A question. Bob Rishel from 9 Duke Energy.
10 Just on the test configuration and the OE 11 that is informing that, I just question, what does the 12 short construction look like? Are we trying to 13 replicate foreign material being captured in the 14 cabinet and that's causing shorts? Or are we assuming 15 bus bars come loose and form a hard contact? So, what 16 is the short we're trying to replicate?
17 MR. MELLY: So, it would be any and all.
18 From, basically, the OpE, we've seen that you can have 19 foreign material enter the cabinet. We've seen fallen 20 pieces of conductive material within the cabinet. We 21 see breakdown at the breaker itself. We see high-22 resistance contacts initiating an arc. The OE is 23 trying to reflect the frequency that is making up Bin 24 16, which includes all of that.
25 Now, in the testing space, we are NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
94 1 initiating the arc using a three-phase connection 2 associated with the IEEE standard. So, it's a copper-3 aluminum wire, copper-tin wire that is placed between 4 the three phases of power to initiate the arc. It, 5 essentially, provides the initial ionized material to 6 create the arc in the cabinet, which, then, initiates 7 the arc. And whether it has enough energy to persist 8 is dictated by the power supply.
9 MR. RISHEL: So, we're going to take that 10 data and we're going to try to translate that into a 11 zone of influence for that. And actually, it's that 12 specific configuration or it's that specific amount of 13 energy, right? Each high energy arc fault has a 14 unique footprint, and it is based upon how much energy 15 can be released before the event is terminated, either 16 through relaying or just the material is consumed.
17 MR. MELLY: Uh-hum.
18 MR. RISHEL: And so, I've got a lot of 19 concern over how that testing is, with one standard 20 configuration, how that's going to be propagated 21 throughout all other possibilities, and especially if 22 you take the arc flashover. So, that depends on how 23 much aluminum you want to throw up in the air. I 24 mean, worst case, you could make an aluminum smelter 25 and put a lot of aluminum in the air. And you're NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
95 1 probably going to get some kind of effect out of that 2 versus, you know, something much less significant, 3 which my guess is -- we haven't seen any of the specs 4 from EMI or arc flashover. So, I would guess, from a 5 thoroughness, you're investigating it, but I'm not 6 sure what value it is at the end of the day to the 7 industry. But I am concerned about the test setup and 8 how we're going to propagate that to be generically 9 applied.
10 MR. HAMBURGER: Okay. I think some of 11 that will, hopefully, be addressed when we discuss the 12 modeling approaches, but your point is noted.
13 MR. RANDELOVIC: Okay. I think we can go 14 to the next slide.
15 MR. MELLY: We can go to the next slide.
16 MR. RANDELOVIC: It's the limitations 17 of --
18 MR. MELLY: Yes, and so, this is some of 19 the pros and cons in terms of the conductivity 20 measurements. In terms of the surface conductivity, 21 it is a known measurement technique. It can measure 22 the holdoff, breakdown, in addition to the surface 23 resistance. The limitation, it is a passive design.
24 We can only test it after, not actively during the 25 test. It does not measure air conductivity, and this NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
96 1 does require failure criteria of components, which 2 will potentially require additional testing or 3 engineering judgment.
4 And what I mean by that is, if I say that 5 I have a breakdown strength or a surface resistance of 6 "X" amount, how does that link with whether I have a 7 failure criteria of a component within my plant? So, 8 that is a limitation of that type of surface 9 measurement that we'll be conducting.
10 The air conductivity measurement probes, 11 they are an active instrumentation device. So we will 12 know what's going on during the arc itself. The 13 limitation is we can only deploy a limited number in 14 various locations throughout the test environment.
15 MR. RANDELOVIC: It would also require 16 some kind of scaling analysis --
17 MR. MELLY: Yes.
18 MR. RANDELOVIC: -- to come up with the 19 criteria specific to the actual plant equipment.
20 MR. MELLY: Yes.
21 MR. RANDELOVIC: So, I would take the 22 third bullet from the first one and add that bullet to 23 the second item.
24 MR. MELLY: Yes.
25 MR. RANDELOVIC: It still requires some NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
97 1 kind of data manipulation or scaling analysis to take 2 the data from the measurement data and understand how 3 to apply that data to the actual equipment. For now, 4 we don't have that answer yet.
5 MR. MELLY: Yes. The mock switchgear, it 6 is a close simulation of plant equipment, or it was 7 designed to be. It has not been verified, as was 8 mentioned, yet with switchgear manufacturers.
9 The limitation is that it was a bounding 10 result. It is a yes-or-no answer, binary. Do we see 11 arcing; do we not see arcing?
12 Some of the devices that we will be 13 discussing in terms of air breakdown will provide more 14 of an active measurement that we can make evaluations 15 on, depending on which system you were in. We'll 16 discuss that later.
17 Next slide.
18 So, the phase 2 tests that we're going to 19 be performing in the August timeframe, September of 20 this year, are listed here on the screen as our test 21 matrix. We will be performing the 480-volt aluminum 22 bus bar tests.
23 And on this screen we show 15 kA. This 24 came out of the April 2018 Working Group. This value 25 has been adjusted to 13.5, based on the Working Group NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
98 1 assessment and looking at plant design. So, we're 2 going to be testing at the 13.5 kV, or kA, at 2 3 seconds and 8 seconds. And we'll be testing at the 25 4 kA setpoint for 2 seconds and 8 seconds as well.
5 Again, the 25 kA there will potentially inform any 6 failure of protection, but, more likely, it will 7 represent some extrapolation to medium voltage, which 8 will be the Working Group activity to see if it's even 9 applicable to do that extrapolation.
10 Next slide.
11 Additionally, we will be doing the bus 12 duct testing. This is going to be performed at 4160, 13 25 kA. We have our duration parameters varied from 2 14 to 4 seconds, and we're going to be varying here the 15 material properties more so. We're going to have 16 copper bus with an aluminum rounded enclosure, 17 aluminum bus with a steel enclosure, and aluminum bus 18 with an aluminum enclosure.
19 All of these configurations have been seen 20 present as part of that NEI survey, and they are out 21 there in the plants. So, we want to understand the 22 differences in the energy release in terms of where 23 that aluminum is. Is it in the buswork, in the 24 enclosure, both, or neither, would be part of the OECD 25 testing.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
99 1 Next slide.
2 So, as part of this test plan 3 reevaluation, we are using new information when it 4 becomes available to identify and reevaluate the 5 objectives of the test plan. Some of the changes that 6 have been proposed, and that we are taking into 7 account as we move forward for those September tests, 8 are the arc location, as well as the equipment design.
9 Those are the tests in the spring 2020.
10 We're going to be taking into account the 11 decrement curves and incorporating some of this 12 generator-fed aspect into the testing to understand, 13 if there is an 8-second test that is generator-fed on 14 the decrement, what is the reduced amount of energy 15 that is seen in the test environment as the generator 16 spins down?
17 Also, we have been evaluating the arc 18 current. Do we see enough of a difference in that 19 parameter or can we reduce the variation in arc 20 current, so we can test more important or, as deemed 21 more important by the Working Group, parameters of 22 interest? So, these are all things that are being 23 discussed after and during the tests that are 24 performed. As well as duration, we want to ensure 25 that it matches OpE.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
100 1 So, as part of the supplementary tests 2 that we've been calling them for spring 2020, which 3 are direct results of some of the discussions we've 4 had with the public engagement, we're looking at doing 5 the supply side configuration for both copper and 6 aluminum. And as Marko mentioned, we're not only 7 talking about the vertical lift breakers, as we've 8 done in the past, but also now we're looking at the 9 potential to be horizontal, draw out the breakers.
10 Again, we have two upcoming tests planned 11 to look at the decrement curve and to look at the 12 static, as we have done in the past. So, we'll have 13 one-to-one comparisons of the ability to look at what 14 the impact of this decrement curve has on the test 15 results.
16 These are the tests that are planned for 17 the spring of 2020, to answer a lot of the questions 18 that have come up in these public meetings. Again, 19 this is an active area for the Working Group. We are 20 taking all recommendations into account in terms of 21 procurement, test design, conditions, applicability to 22 realism. We're taking that not only for these 23 upcoming 2019 tests, but also the 2020 tests.
24 MR. RANDELOVIC: The last slide, as Kenny 25 was saying earlier, we are working on the project NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
101 1 plan, basically, how these pieces fit together. It's 2 a complex project. There's so many moving pieces.
3 And unless you are inside the Working Group, it's kind 4 of hard to understand what is going on. So, this 5 project plan is going to capture the discussions that 6 we have at the Working Group level, but also how all 7 of these four different pieces actually fit together.
8 We're also going to have like kind of a 9 Gantt chart providing the schedule for different 10 activities.
11 MR. MELLY: Yes. We've had several 12 meetings where we've tried to at a high level discuss 13 what we're going to be doing and some questions or 14 concerns or comments were raised. My reaction was, I 15 wish they were just in the Working Group meeting 16 because we discussed that for three hours and had 17 back-to-back conversation. And it's something we're 18 thinking about; we're actively trying to incorporate, 19 and we hope that this project plan will allow any 20 interested party to dig into an area where they have 21 questions, concerns, or try to further their 22 understanding of what we're doing.
23 MR. HAMBURGER: Okay. So, the rest of our 24 morning meeting is dedicated to public comment. I 25 know we have a presentation from EPRI. Before we do NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
102 1 that, do we have any questions or comments about the 2 Working Group status update?
3 MR. MELLY: And also, I'd invite anyone, 4 if there is any line or multiple questions, if you 5 want to come up to the table and just use the 6 microphone at the table, feel free to do that as well, 7 rather than standing in line. So, either at the 8 microphone or at the table will be fine.
9 MR. FLOYD: Hello. I'm Jason Floyd, 10 Jensen Hughes.
11 It was good to see that you're going to 12 make some effort to make conductivity measurements, 13 but if we do observe an effect of conductivity, either 14 due to deposition or in the gas phase, that's going to 15 be highly dependent upon either the mass deposited or 16 the aerosol concentration. I didn't notice you 17 discussing making either galvanometric measurements of 18 deposition or maybe use of some kind of galvanometric 19 observation measurements of aerosol concentrations in 20 the gas. Because in an actual HEAF in a facility, at 21 those concentrations it's going to be very dependent 22 upon configuration, ventilation, actions taken by 23 plant personnel in the long term maybe to purge 24 aerosols from rooms. And so, we need to have an 25 understanding of that mass, not just the effect.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
103 1 MR. MELLY: We are going to be hoping to 2 evaluate that in terms of the breakdowns measurement 3 device by looking at both open air and in an enclosure 4 to see the effect there.
5 In terms of the surface deposition, do we 6 have further discussion on that later in the 7 afternoon, Chris?
8 MS. LAFLEUR: Minor, because it was added 9 after the 30-day distribution slide.
10 MR. TAYLOR: Yes, so as far as the 11 galvanometric measurement for the surface deposition, 12 there weren't plans right now. We can go and look and 13 see if we can take the sample holders and take a 14 measurement of the deposits.
15 As far as the air concentrations, we'll 16 have to get back with the experts that run our 17 spectroscopy devices and our imaging equipment at 18 Sandia to see if they are able to capture that type of 19 information. So, I think it's a good point and we 20 need to look into that further.
21 MR. HAMBURGER: Any other questions from 22 the room?
23 (No response.)
24 No questions from the webinar.
25 Okay. If you think of any, you're welcome NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
104 1 to ask them after EPRI presents their material.
2 Kelli, are you ready?
3 MS. VOELSING: Yes.
4 MR. HAMBURGER: Okay. Would you like to 5 come to the front or just sit at the table?
6 MS. VOELSING: So, I think green means 7 it's on?
8 MR. HAMBURGER: Yes.
9 MS. VOELSING: Okay. First of all, my 10 name is Kelli Voelsing. I'm the program Manager at 11 EPRI for the Risk and Safety Management Program, where 12 our fire PRA research activities fall.
13 Thanks for the opportunity to speak today.
14 And although there's a lot of people in the room and, 15 obviously, a lot of interest in this topic, I'm at 16 least glad that probably my comments and feedback will 17 get less interest than another person participating in 18 a public meeting in D.C. down at the Capitol today.
19 (Laughter.)
20 So, I wanted to set some context for my 21 comments in this presentation. We've heard about 22 EPRI's participation in the Working Group, and Marko 23 clearly was participating in the Working Group. And 24 so, why in the public comment period does EPRI have 25 additional comments to make on the HEAF activities?
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
105 1 So, I'll try to provide a little context for that.
2 First of all, our participation in the 3 Working Group under the Memorandum of Understanding --
4 you know, we've clearly been commenting and 5 identifying technical issues related to NRC Research's 6 plan for the next phase of testing. I am very 7 encouraged to see that there is a lot of that 8 technical interaction happening, that the information 9 is being exchanged, that updates and enhancements to 10 the test plan are being made, and that that 11 relationship and that feedback is working very well.
12 We've also served as a conduit to provide 13 industry OE and industry survey-type information to 14 NRC Research to inform testing. We heard about the 15 electrical plant schematics of all the plants in the 16 U.S. as well as our review of OE, and how that has 17 contributed to the test plan. So, I think that's 18 pretty much been covered.
19 I will say that, to date, our engagement 20 has been commenting on the technical aspects of the 21 plan as presented to us. So, we did not participate 22 in the PIRT. We did not participate in the OECD 23 testing. We don't have access to any of that data.
24 EPRI was not engaged in the scoping or initial 25 planning or drafting of initial test plans for various NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
106 1 activities involved in the overall plan for HEAF.
2 And I think we focused a lot on the next 3 phase of testing being planned to start in late August 4 or September of this year, and we see that we still 5 have some major open issues that the Working Group is 6 working on. So, they've been very focused on getting 7 to resolution on those issues. And therefore, some of 8 the broader questions we have about the overall plan 9 and where we're going have not yet been part of the 10 scope of activities being resolved by the Working 11 Group. So, that is kind of what I'm hoping to focus 12 in this presentation, is some of those things that the 13 Working Group hasn't yet directly engaged on.
14 I was taking notes during the 15 presentation, and I have counted up here, you know, 16 maybe 8-10 times where we heard, "Well, that will be 17 a good use of modeling." We hope the modeling will 18 help with this. We're trying to take that into 19 account. We're hoping to address this in the 20 methodology in modeling. We're hoping some of the 21 information collected can be used with the modeling.
22 So, we're putting a lot, you know, we're kind of 23 putting a lot of faith in that, to take the data 24 that's being collected and resolve the issue. And so, 25 that's kind of what my presentation here is focused NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
107 1 on.
2 Go to the next slide.
3 I do think it's important to clarify 4 EPRI's engagement in this activity. EPRI is an 5 independent, objective, not-for-profit, scientific 6 organization chartered in the public interest. And 7 so, for us, clearly, our objective in engaging in the 8 work is identifying the input, the OE, the data, the 9 methods, the tools necessary to support realistic 10 evaluation of any change in nuclear plant risk related 11 to HEAFs involving aluminum. We obviously have that 12 public sector mission. Obviously, for us, nuclear 13 safety is first as well.
14 But we do want to be independent and 15 objective and make sure that the testing and the 16 evaluation of that potential change in risk is 17 representative of actual plant conditions and OE and 18 configurations. And so, although, clearly, the NRC 19 has some process activities and timelines and other 20 criteria that they have to meet in their program 21 related to the pre-GI process and other regulatory 22 processes that they may be involved or may enter into, 23 for EPRI it's about getting the right answer.
24 I put a little background information, 25 which, for this in the rooms, is going to be everybody NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
108 1 knows, but perhaps in the public slides this may be 2 important for the context of some of my other 3 comments. And I think you have seen most of this in 4 the previous presentation by the Working Group.
5 So, in order for us to evaluate risk to 6 nuclear safety, that risk has two pieces: frequency 7 and consequences. The frequency piece, as Marko and 8 Nick talked about, is being worked in the Working 9 Group, that parsing of the Bin 16 events to more 10 accurately reflect flashes, blasts, HEAF events, and 11 where the susceptibility in the plant exists, based on 12 electrical distribution systems, based on types of 13 components and applications.
14 That was a task under the EPRI Fire PRA 15 Research Plan, and that technical work, we've had 16 meetings. The technical work is ongoing, and we're 17 making good progress in that area.
18 And then, of course, the consequences, as 19 we've talked about, we're all hoping that, coming out 20 of this activity, that you have zones of influence 21 that align with the end-states of the event trees for 22 the frequency work. So, then, you can in the PRA 23 method apply whatever frequency you come up with for 24 your component with a relative zone or a related zone 25 of influence for that activity.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
109 1 So, I'm very encouraged. I think there 2 was good alignment across everybody here that our goal 3 was to move from the one-size-fits-all modeling 4 approach to one where the frequency is informed by 5 what we know, and that those end-states align with 6 representative zones of influence. So, I think we're 7 all aligned on the goals.
8 Next slide.
9 Zone of influence. Very generically, the 10 volume and which plant SSCs or targets would be 11 expected to fail as a result of the event -- in this 12 case, HEAF -- I think it is important to recognize 13 that a zone of influence does not mean that we had 14 deposition on the wall or that we had bare bus bars 15 that experienced damage. A zone of influence, by 16 definition, has to include an understanding of the 17 fragility of potential targets. We need to know what 18 parameters influenced the failure of that target; at 19 what intensity those failures occur; how are the 20 factors to be related; what methodology turns a heat 21 rate or a flux or a peak temperature or a pressure 22 into a failure? What is the model that supports that?
23 And where in the plant can these factors realistically 24 be expected to exist and to cause a failure? We've 25 talked about the open space measurements versus what NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
110 1 a subcomponent might experience inside a cabinet.
2 So, I think in saying we've observed a 3 large zone of influence in our previous testing, we 4 need to be careful and recognize that we haven't yet 5 defined the fragility of any of these components. And 6 what we observed was perhaps a deposition, but not 7 necessarily a zone of influence at this point.
8 So, with that as background, a few issues 9 that I would like to address today. First, I've kind 10 of already alluded to the defining of fragility.
11 Second, clarification of the mod/sim approach or the 12 modeling approach and the quantification of the tools 13 for that application. I know we're going to talk more 14 about that this afternoon.
15 We've already beat the next one to death:
16 utilizing the full range of OE to represent the 17 distribution or the probabilistic spectrum of events.
18 Availability and insights from previous test results, 19 and I won't spend time covering the conductivity and 20 EMI open issues from the phase 2 that the Working 21 Group is still working on. So, this is the content of 22 the remainder of my comments.
23 Okay. With respect to determination of 24 fragility and a zone of influence, to date, as we've 25 heard, the parameters for defining fragility for NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
111 1 relevant components have yet to be defined. We 2 understand that this is a future activity of how 3 Research plans to address the HEAF issue, that they 4 will identify the parameters that relate to fragility; 5 determine the models and the relationships of how 6 those parameters are correlated to an actual failure, 7 and then, demonstrate the ability of the modeling and 8 simulation approach to produce results that both align 9 with the experimental data that's collected as well as 10 be translated into fragilities and a zone of 11 influence.
12 And as those activities are still yet to 13 be determined, a future phase of the work, we 14 understand that if that approach is not successful, 15 that the backup would be to fall back on utilizing 16 more expert judgment to inform the process.
17 So, our concern with this being a future 18 activity that still I think we're all committed to, I 19 think we think it's the right thing to do, but our 20 concern is, without yet knowing whether that's going 21 to be successful, what that's going to look like, what 22 defines fragility, our concern is, what assurance do 23 we have that we're collecting the necessary 24 parameters? We think we've got the right parameters 25 of interest that are being collected, but without NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
112 1 defining how you're going to model the fragility, how 2 do you know that?
3 If the data being collected are the right 4 parameters of interest, it's yet to be defined what is 5 the method that will relate those measurements to a 6 fragility determination. And the collection of data 7 in open space, I think we all know, does not directly 8 relate to the value that a given parameter might be 9 experienced in a nuclear plant configuration -- so, 10 inside a cabinet, inside a conduit, behind a barrier.
11 And we need to understand how that's going to be dealt 12 with when translating the measurements collected in 13 open space into a zone of influence.
14 With respect to the modeling-simulation 15 approach, first of all, we are very encouraged that 16 everyone's committed to not using the one-size-fits-17 all approach; that we're committed to having relevant 18 zones of influence that are informed by what we know 19 about plant design; what we know about the function 20 and the location of these components in the plant and 21 what we know from OE. So, that's all a good thing, 22 and we all recognize that our test budgets are not 23 infinite. And so, having a modeling and simulation 24 capability that can help interpolate or extrapolate 25 between those measurements, test other configurations, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
113 1 is obviously, you know, would be of great benefit.
2 NRC Research and Sandia are proposing to 3 use the coupling of area and FUEGO codes to model the 4 HEAF events and to find a zone of influence. Our 5 understanding, obviously, these codes have been used.
6 They've been used in nuclear weapons applications.
7 They've been used, and, of course, they have their own 8 V&V support, which justifies the physics in the code 9 and that the code is giving the expected results; and 10 that it's being compared against a variety of data; 11 and that, for the purposes for which the code was 12 designed, that it is capable of doing what it's 13 supposed to do.
14 But what we haven't yet had time to work 15 on or have seen is the capability and qualification of 16 the codes for this specific application. So, the code 17 is just, you know, the math, but getting the right 18 answer is all in the model that you set up. And you 19 need to define how you're going to set up that model; 20 how you're going to do the coupling of the codes; how 21 you're going to benchmark against relevant 22 experimental data for this particular approach, and 23 prove that what you're modeling is providing you 24 reasonable answers.
25 So, we haven't had time to do that work NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
114 1 yet. To a certain extent, we already have a lot of 2 data available or, presumably, there should be a lot 3 of data available. The OECD testing was completed 4 several years ago. Phase 1 testing on medium-voltage 5 equipment was completed last year in September. And 6 so, if we had the modeling and simulation approach 7 better defined, we would be able to compare that to 8 the test data that we already have; see how it's 9 performing. Are we collecting the right inputs? Are 10 we getting the expected outputs? Do we know how to 11 translate that into a zone of influence or fragility?
12 And based on that analysis, do we need to make any 13 tweaks to what we're collecting or what we're doing in 14 the phase 2 testing?
15 And again, I think we all think the 16 modeling and simulation approach would be a huge 17 benefit. Definitely having different zones of 18 influence for different locations and different 19 applications is much more realistic in terms of our 20 PRA modeling. But we've said, well, if it's 21 unsuccessful, then we can fall back on expert judgment 22 or test data.
23 And our concern there is, if you don't 24 have the modeling and simulation approach to 25 interpolate between 4 seconds and 8 seconds, or NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
115 1 between 5 feet and 10 feet, or between this type of 2 enclosure and that type of enclosure, then perhaps we 3 needed to have different targets in the test field for 4 the phase 2 testing to collect different relevant 5 measurements that would be more representative of in 6 situ plant equipment.
7 Realistic use of OE and data. You know, 8 I am very encouraged by the discussion in the room, 9 and I very much thank NRC Research and the Working 10 Group for taking the feedback on this activity. It's 11 obvious that you've heard the message.
12 So, I think we all understand that testing 13 is not infinite and that it is expensive; and that if 14 -- I don't know -- if the large majority of time all 15 you get is a spark, and then, the event terminates, 16 well, testing that doesn't really give you any 17 experimental data because you didn't have an arc fault 18 to collect any data. So, we absolutely understand 19 that.
20 But the recognition that testing outliers 21 or bounding configurations does not represent the most 22 probable event or the distribution of the event in the 23 plant, I think is a concern from the stakeholders 24 because of some past experience. And so, I think 25 that's why we're asking for better understanding of NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
116 1 how the mod/sim, how the calculation or the definition 2 of fragility and the calculation of that zone of 3 influence, the translation from measured parameters 4 into that zone of influence -- having more detail 5 about exactly how that's going to work and what the 6 preliminary results we have, what are the insights we 7 have based on the testing we've already done, having 8 more detail about that would help to reduce 9 uncertainty about how we're going to approach it. And 10 that would probably alleviate the concerns about 11 taking a very bounding configuration and applying that 12 all over the PRA.
13 And finally, on the availability of the 14 results, we do have not a lot of data from the tests 15 that have already been conducted more than a year ago.
16 So, some results from the phase 1 testing are not yet 17 available. We understand that, apparently, detailed 18 analysis and data scrubbing is required. We have some 19 descriptions of the small-scale testing, and we 20 understand that they weren't intended to be 21 representative of a HEAF, that Sandia needs to build 22 up piece by piece to the full-scale, large HEAF 23 condition. But where we are with that small-scale 24 testing is not yet representative of a HEAF in terms 25 of the geometries and the arc, arc lengths, and other NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
117 1 factors.
2 And then, we only have very limited 3 material characterization from the small-scale 4 testing, as Nick alluded to, but our understanding is 5 that the majority of the ejecta that was measured thus 6 far was either aluminum oxide or molten metal with an 7 aluminum oxide kind of coating, which would be an 8 insulative and non-conductive material. So, 9 understanding whether that's the actual condition in 10 the cloud would be important to knowing whether this 11 conductivity and flashover is really an event that we 12 need to be concerned about, since we haven't really 13 seen that in OE, and the bare bus bars on the wall 14 damage at KEMA is probably not representative of any 15 condition that exists in a nuclear plant. We don't 16 have bare, uninsulated bus bars exposed in such a 17 manner.
18 So, I guess, just from a project 19 management, not a nuclear risk, but from a project 20 risk perspective, the question is, is it prudent to 21 understand the insights and the utility from previous 22 data prior to proceeding with additional testing? I 23 think it's a broader concern that we want to make sure 24 that, before we spend more money and charge forward 25 with more testing, that we've really evaluated the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
118 1 data that we have from previous results, defined how 2 we're trying to use that with the modeling and 3 simulation approach. And are we getting results that 4 we think we can use or do we need to make tweaks to 5 the plan?
6 Phase 2 testing, they already talked about 7 that at length. We talked about that before or that 8 the Working Group is still working to resolve the 9 conductivity issues, and that the EMI was added very 10 recently. And we don't see that in OE. So, I 11 understand it was a request from others, other 12 departments within the NRC. But if there's any 13 thought that somehow that translates into a fragility 14 or a zone of influence relative to HEAF, we would like 15 to have a more detailed review of that technical 16 basis, because we have not observed that in the OE in 17 any of the international or domestic events that have 18 been reviewed.
19 So, I'll try to be quick to make sure we 20 leave time for public comment here or questions from 21 others other than me before the conclusion of the 22 meeting.
23 I think the graphic there in the upper 24 right corner is pretty representative. There's a lot.
25 As we've talked about, this is a complex project.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
119 1 There are a lot of moving pieces. There's the plant-2 specific aspects of configuration and layout, presence 3 of aluminum, the frequency work that's being led under 4 the EPRI fire PRA tests, definition of fragility, 5 mod/sim, test results data. All of those pieces 6 ultimately have to come together to answer the 7 question. And really, I guess the overriding concern 8 is we don't want to get too far out ahead with one of 9 those roads when, ultimately, they all have to work 10 together to solve the problem and get the right 11 answer.
12 So, we're just suggesting that defining 13 all of the necessary inputs and evaluating the 14 currently-available data before proceeding with 15 additional testing may be a prudent option to 16 consider.
17 Any questions for me?
18 MR. CHEOK: Thanks, Kelli. Good comments, 19 and I think we understand it. We have been working 20 together very well as part of the whole group, but we 21 understand your comment that EPRI wasn't involved at 22 the beginning in terms of the definitions, the tests, 23 and the bullets, and things like that. And so, those 24 are valid comments and we will take that into account.
25 So, I think when we are talking about NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
120 1 delaying the next set of tests, for example, there are 2 several things that we have to consider. One of them 3 is that this is a pretty generic issue. If the staff 4 or the industry had identified an issue that 5 potentially could be safety-significant -- potentially 6 -- we owe it to all our stakeholders to not drag the 7 issue out and just to see if it's really a safety 8 issue or something that we all can deal with in a lot 9 of different other ways. So, that's one part of the 10 schedule.
11 We have gone ahead and tried -- we have, 12 between my friend and myself, asked for an extension 13 of the current GI scheduling, so that we can 14 accommodate additional comments and industry future 15 test results, et cetera.
16 So, having said that, our next set of 17 tests, whether it's going to be spring of next year or 18 summer of next year, I hope and anticipate that it's 19 going to be the last set of tests that the NRC will be 20 carrying out in terms of the aluminum HEAF generic 21 issue. And I know, as you say, budgets are limited.
22 We need to do smart testing.
23 And you also mentioned that we have a lot 24 of "unknown unknowns," a lot of things that we, in 25 terms of the fragilities -- some of them I think we NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
121 1 can work out over the next six months or year.
2 However, a lot of the "unknown unknowns" will not be 3 known in the next year, and I don't think we can delay 4 a set of tests because there are "unknown unknowns".
5 I think we have to be prudent in terms of selecting 6 our parameters we have to be testing for, and then, we 7 will look at the results and parse out what is 8 important and what isn't important, and then, 9 determine from that set of data, while we are marrying 10 that with the frequencies of those particular 11 scenarios, to see if any further action is needed.
12 So, I think without the August tests we 13 will not be able to further validate our models.
14 That's one set of things that I think would be 15 important for us. Taking the tests that we have done 16 already are only -- for example, the facade test that 17 was done by OECD -- could skew the results simply 18 because the set of tests, as was pointed out before, 19 could be conservative. So, I think we have to be --
20 you know, we are trying to expand our database a 21 little bit more to accommodate what you've been 22 seeing.
23 MS. VOELSING: To be clear, I wasn't 24 advocating that we should draw conclusions based on 25 only the testing that's been done. I was advocating NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
122 1 that we would have more confidence that we are doing 2 everything we need to do in the future testing if we 3 had taken the data that we had and evaluated how is 4 the mod/sim performing against that data or what have 5 we learned from that data. And we haven't really done 6 any analysis of that data yet.
7 MR. CHEOK: Thanks. Thanks for the 8 clarification.
9 So, in terms of the fragility and the zone 10 of influence, we totally agree with you that the zone 11 of influence will depend on the fragility, and the 12 zone of influence is very plant-specific. And we will 13 not be able to carry out every single test to take 14 into account all the different configurations. Again, 15 we will have to be smart in terms of defining what we 16 need to get.
17 So, the next set of tests should provide 18 us with data that may have some obvious results. So, 19 if it shows that, hey, given this cloud of dust, for 20 example, everything gets oxidized, there is no effect.
21 Either that or you can have a test that says, hey, 22 look, given this test at 3 feet, you know, the 23 temperatures and the pressures, and everything else, 24 is so extreme that we don't think that any defined 25 fragility of equipment should matter.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
123 1 So, I think the next set of tests could 2 also tell us what could be the scope for us. You 3 talked a lot, some --
4 MS. VOELSING: We're just concerned 5 because we haven't made any of those kinds of reviews 6 or analysis based on all the work that's already been 7 done.
8 MR. CHEOK: Yes, and we will do that, 9 right? I mean, we talk about some of the fringe 10 results, and we may not, and you're right, we may not 11 be able to relate the fringe data results, for 12 example, to a fragility. And we will acknowledge 13 that, and that's absolutely correct.
14 I think, with the fringe results, we will 15 obviously use those results to define our next set of 16 tests. But, without knowing what the bounding 17 conditions are, where we can illuminate things or add 18 things into the scope, our next set of tests will be 19 a lot larger than it should be.
20 Again, we will try to make available, all 21 the data available as much as we can. You had 22 mentioned the fact that we are looking at some data, 23 and I think I addressed the comment earlier from 24 George.
25 So, we cannot just release data. We can, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
124 1 but we should not release data without putting context 2 to it. So, we do not want people to take data and 3 come up with incorrect assumptions or incorrect 4 results. I think we need to say, hey, look, let's 5 look at this data; let's put some context to this 6 data. And then, if we need be, we'll release all the 7 data, but we want to put some context to the data, so 8 that this data cannot or should not be misinterpreted 9 by all parties concerned.
10 MS. VOELSING: Yes, I think you heard that 11 in the previous comments.
12 MR. CHEOK: Uh-hum.
13 MS. VOELSING: And you guys addressed that 14 you're responding to those comments. There was a 15 request to kind of show what is the basis for -- you 16 know, this is the gap; it gets filled here; it gets 17 addressed here. This is where it's addressed in the 18 program. This is what we've learned. This is what 19 still remains open. And you're working on a project 20 plan to help make that all more clear.
21 So, I absolutely understand that and 22 appreciate your response to that. I guess my only 23 question is, we seem to have not done that yet for the 24 phase 1 test results that took place a year ago. So, 25 when are we going to be able to get that done?
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
125 1 MR. CHEOK: And, you know, I guess we can 2 discuss that later also.
3 But I think Jeff is standing at the 4 microphone now. I'll turn it over to --
5 MR. STONE: Yes, I just wanted to add some 6 context here from a utility perspective or from my 7 perspective as a utility member. It is that I'm 8 concerned and hopeful that we don't go down the same 9 path we did with some of the fire stuff years ago, 10 where we overestimated possibly some heat release 11 rates and, then, we didn't -- and along with that, we 12 had conservative fragilities. It sounds like we're 13 trying to get the right heat release rates here from 14 that.
15 But if we don't understand the 16 fragilities, and then, we make conservative 17 assumptions next year because we're in a hurry to get 18 this GI resolved, then we could end up with dramatic 19 overestimation of the impact of the HEAFs, if we don't 20 understand the fragilities. So, I was not clear why 21 we didn't attempt to look more at the fragilities in 22 this test set. But I would like some confidence from 23 our side that we don't step into a conservative 24 assessment of the fragilities because we don't have 25 the tests.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
126 1 MR. MELLY: So, part of the reason why the 2 fragilities aren't explicitly included or inherently 3 included at the KEMA facility is the difficulty with 4 doing the source term data, the electrical supply, 5 with the measurement techniques that are needed in 6 order to ascertain the fragility of equipment.
7 I would need to have an electrically-8 powered cable, for instance, in the test cell and 9 evaluating the shorting performance of the cable, the 10 potential for the cable to ignite, the flux and the 11 temperature inside that cable that it's exposed to, 12 the source term of the HEAF. And that's just for one 13 cable which is defining one piece of fragility 14 information.
15 KEMA is an electrical test laboratory, not 16 a fire test laboratory. All of that would need to be 17 designed and implemented at KEMA around the HEAF, 18 which is impractical for how the tests are run in both 19 a cost impact as well as the ability or space impact 20 to actually have that type of data married up with the 21 electrical data of the HEAF.
22 So, that's why you don't see electrically-23 powered cable trays in the test setup that we're 24 running. We're trying to get the data from the source 25 term of the high energy arcing fault, and then, apply NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
127 1 that to fragility data, rather than do the fragility 2 at the KEMA facility.
3 The way that KEMA works is we only have on 4 day to perform the tests because we're using that test 5 bay. So, if we took three days for test setup, we're 6 essentially paying for three tests, which would 7 largely balloon all the entire scope as well as cost 8 of the test program.
9 MR. STONE: So, I'm unclear how we --
10 there's no other way to evaluate the fragilities or do 11 tests in a realistic method for that? Because, 12 obviously, if we go conservative there, the 13 implications to us in the utilities could be pretty 14 dramatic from a cost perspective.
15 MR. MELLY: So, we do not plan on going 16 conservatively. That is not the current approach.
17 The plan is to marry the source term data that we are 18 collecting with the fragility data or with failure 19 criteria, but not explicitly testing it at the same 20 time at the KEMA facility. We understand that if you 21 go conservative, which is the scoping analysis -- if 22 I take out everything in a particular room from a high 23 energy arcing fault, it's going to have a huge risk 24 impact. Is it realistic? No. We understand that, 25 and our goal is not to go conservative with the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
128 1 fragility data.
2 MR. RANDELOVIC: So, Nick, maybe you 3 should be a little more explicit on how we are going 4 to get at the fragility. Like explain we are 5 collecting the test data. The test data is going to 6 be used by --
7 MR. MELLY: I believe we have --
8 MR. RANDELOVIC: We have case from Sandia.
9 MR. CHEOK: Chris, first, I just want to 10 address something Jeff said. So, we will not go 11 conservative -- will not. We will engage, once we 12 have all the tests, data, and results, we will engage 13 all our stakeholders with all our data, and we will 14 incorporate everything that we hear before we even 15 decide what the next steps are.
16 MR. HAMBURGER: Just before Chris says 17 anything, I want to give the webinar folks a minute to 18 get their questions together. If you're on the 19 webinar, we currently have the lines muted. But if 20 you do have a question for us, you can either type it 21 to Tom in the chat or you can use the "raise hand" 22 feature and we can unmute your line, so you can ask 23 the question. So, I'll give you guys a minute to do 24 that, if you would like to ask a question.
25 MS. LAFLEUR: Okay, thanks. I'm Chris NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
129 1 LaFleur from Sandia National Laboratories. Sorry 2 about my voice.
3 In terms of establishing what the failure 4 criteria for something like cables would be, right, 5 and that's going to be the first focus, it's a 6 function of the temperature and heat flux, but also 7 the time that a target cable is exposed to those harsh 8 conditions. And so, the reaction of the cable or the 9 response of the cable for the duration of that 10 exposure we can test separately in other facilities at 11 Sandia where we can expose an energized cable and a 12 cable that's being measured for the underjacket 13 temperature. We can expose it to short-duration, 14 high-heat fluxes or high temperatures, some of our 15 other facilities. And we can say, okay, "X" is the 16 temperature that that cable will fail in 2 seconds.
17 "Y" is the temperature that cable would fail in 8 18 seconds, right? And we can bound that based on the 19 expected duration of the HEAF events, and that's --
20 MR. STONE: Is that part of the plan?
21 MS. LAFLEUR: Yes, that is the plan that 22 we're currently formulating with the cooperation in 23 the Working Group to determine what exactly we need to 24 test, bounding in between, whatever, to determine what 25 the failure criteria for cables are.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
130 1 MS. VOELSING: Is there a timeline for 2 that?
3 MS. LAFLEUR: Yes.
4 MR. RANDELOVIC: Well, we just started.
5 MS. LAFLEUR: We're doing it, yes.
6 MR. RANDELOVIC: We just started three or 7 four weeks ago. We had a first phone call with Chris 8 and with Jason and Gabe. And I think we are just at 9 the beginning of a process, and I think it's going to 10 take some time to develop the methodology and the test 11 facility, the test facts --
12 MS. LAFLEUR: Yes.
13 MR. RANDELOVIC: -- whatever has to be 14 done.
15 MS. LAFLEUR: And cables are one thing 16 entirely different. The subject is other equipment in 17 enclosures. You know, what criteria would be 18 impacting them, right? Because the critical 19 components of those equipment are within a shield of 20 sheet metal, or whatever. So, we could make some 21 estimates of when that aluminum enclosure or steel 22 enclosure would be breached within a certain different 23 distance of different HEAF event energies. So, that's 24 what has been --
25 MR. HAMBURGER: I'll take a comment from NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
131 1 Bob, and then, I think we have one question from 2 Faramarz on the webinar. We'll unmute.
3 MR. RISHEL: So, I just want to caution 4 this is too important for conclusions to be drawn 5 prematurely because of cost and schedule. If it takes 6 longer, it takes longer, and if it costs more, it 7 costs more. And I don't think we should cut off 8 testing because it takes too long or too many days or 9 the lab's not available those days. If we need to do 10 it, then we need to do it.
11 Otherwise, you know, my fear is that the 12 utilities will be left with some information that is 13 incomplete and, essentially, potentially a generic 14 issue that they're going to have to go solve with 15 additional research on their own to either fill in the 16 blanks or -- so, I would caution to make sure we spend 17 the time and money needed to address it, not let cost 18 and schedule drive the answer.
19 MR. HAMBURGER: Faramarz, can you hear me?
20 MR. POURNIA: Yes, I can hear you. Just 21 so people know, that was Bob; that wasn't Faramarz, so 22 to make sure I'm clear from the comments that he's 23 made.
24 But, yes, just I want to say -- Faramarz 25 Pournia for the Nuclear Operating Company -- I do NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
132 1 appreciate NRC Research. The statement was made just 2 following Kelli's presentation that the public has 3 identified a safety concern, and the Research charter 4 is to go figure out if that's safety-significant or 5 not, and provide that feedback to the public. So, I 6 do really appreciate that, and I want to make sure I 7 emphasize that, because that kind of stuck with me.
8 I really listened to the discussion that 9 went on for a while in here. It started with George 10 actually. There appears to be so many issues that are 11 on the table, and some of us have been in an 12 environment of research and economic side of the 13 research. It seems, though, if we don't have concerns 14 that are addressed to the points that we should do, 15 and the extent of the testing that should be done, my 16 point was, would I really the meet the statement I 17 just heard from the individual from NRC Research that 18 the public has a safety concern? "Our objective is to 19 go to do clear, objective, and selective of the 20 operating plants to address the concern of the 21 public."
22 Because if I portrayed these tests not 23 reflective of what the stations are and more 24 realistic, my concern is, am I answering the public 25 question in a proper way? But I may be fanning an NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
133 1 anxiety over an issue that may not be reflective of 2 the true significance of this event.
3 So, with all that discussion that went on, 4 I think it would be understood that maybe a time of 5 doing a look-back and say, are there some anomalies in 6 here that perhaps -- to what the individual brought up 7 from NRC Research -- a clean time and go through the 8 process to make sure that we have the appropriate 9 layout to go to do this test is more warranted, rather 10 than just proceeding because you are driven by 11 schedules or other needs and ways that we just have to 12 do it.
13 So, thanks for your time.
14 MR. HAMBURGER: Okay. Thank you. Thank 15 you for that comment.
16 You have a written comment? Okay. Can 17 you remember it or?
18 So, there's a written question on the 19 webinar about the calibration of the models that we 20 intend to use. And I'm going to ask that that person, 21 during the webinar for the afternoon, because we have 22 an entire afternoon dedicated to exactly that subject 23 -- so, if you still have that question this afternoon, 24 then we will be happy to address any further 25 questions.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
134 1 Anyone else in the room have comments or 2 questions that they would like to make?
3 And again, if you're on the webinar, 4 please go ahead and use the "raise hand" feature or 5 the chat feature, and we can open up a line or relay 6 your question.
7 (No response.)
8 I want to remind everybody on the webinar 9 that there is a separate link and webinar session for 10 the afternoon meeting. So, when we finish here for 11 the morning session, we are going to terminate the 12 webinar. Please use the afternoon meeting webinar to 13 join us for this afternoon's session. And that is 14 available on our public meeting website.
15 And if anybody has any questions that 16 occur to them during the afternoon session, we do have 17 another public comment period scheduled for the last 18 part of the day. So, if it does come up, if it 19 strikes you while we're having our afternoon 20 discussions, you can feel free to ask any questions 21 then.
22 If you haven't signed in on our sign-in 23 roster, please do that before leaving for lunch, if 24 you don't mind. I'll have a new roster for the 25 afternoon meeting.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
135 1 And got another one maybe? We maybe have 2 a webinar question.
3 The afternoon session starts at 1:30. If 4 you're a visitor and you're going off-campus, I would 5 suggest trying to get back to the security booth by 6 about 1:20 to make sure all get back up here in time.
7 And we'll have NRC staff helping to escort people back 8 to this room.
9 I put my number up on the display again, 10 in case anybody has any trouble getting back in.
11 Tom, anything? No? Okay.
12 All right. Well, thank you all for your 13 participation this morning and I look forward to 14 seeing you back at 1:30, if you're participating in 15 the afternoon session.
16 (Whereupon, the above-entitled matter went 17 off the record for lunch at 12:14 p.m. and resumed at 18 1:33 p.m.)
19 MR. HAMBURGER: Okay, it's past 1:30, so 20 we'll get started. Welcome back. For those of you on 21 the webinar, unless you are one of our Sandia 22 participants, we have you muted currently. If you'd 23 like to say something, you can either use the raise 24 hand feature or send a chat message to Tom, who is 25 manning the webinar, and we can open up the phone NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
136 1 lines for you.
2 Same rules as this morning, we are 3 transcribing the session, so please state your name 4 before speaking. Please speak into the microphones, 5 and you'll need an NRC escort to go anywhere off the 6 floor and lobby level here.
7 MR. TAYLOR: All right, welcome, 8 everybody. My name is Dave Taylor. I'm with the 9 Office of Research here at the NRC, and this 10 presentation that I'll be going over this afternoon 11 basically provides an overview of the HEAF hazard 12 modeling.
13 So I'm going to be providing some 14 background information on how it's currently modeled 15 in the fire PRA, and then get into, when we're looking 16 at developing or improving the model, the different 17 approaches that we have or types, quickly going over 18 the instrumentation that we're using to help feed back 19 to the models and provide some form of validation to 20 the models.
21 Then I'll be going briefly over some of 22 the more popular existing models, and these are, they 23 really have been developed for arc flash hazards, 24 looking at personal protection, hazard to persons from 25 burns from an arc flash.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
137 1 But I'll briefly be going through some of 2 those models and then provide some comparisons to the 3 measurements that we made in the fall testing too, 4 those models, as well as I'll be talking about some 5 air voltage breakdown tests that we'll be running and 6 providing a little additional information on how that 7 information can be potentially used to confirm a model 8 that's existing that estimates the breakdown voltage 9 of air based on temperature and metal particulate 10 within the air vapor mixture.
11 So into the review of how we model it in 12 PRA, and there's really two different distinct 13 approaches when you look at electrical enclosures, how 14 they're modeled. The one on the right is the thermal 15 classical fire modeling that looks into, okay, you 16 have a fire inside of an electrical enclosure. How 17 does that fire grow and develop in such a state that 18 it could potentially damage other equipment that is 19 important to safe shut down?
20 And then on the right side of the diagram, 21 we have the high energy arc fault failure mode, and so 22 instead of the classical thermal slowly developing 23 fire, we have a more rapid, energetic release of 24 energy, a high energy release of energy to the 25 enclosure as well as surrounding components near the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
138 1 defaulted enclosure.
2 So the focus of the project that we're 3 underway in is looking at the high energy arcing fault 4 or the approach on the left, or, excuse me, right.
5 We did back to, you know, some of the 6 operating experience and how the methods were 7 developed. There's really the key event that happened 8 back in 2001 at San Onofre Nuclear Generating Station.
9 The NRC issued an information notice after 10 that event, and basically at the time frame when this 11 event happened, the NRC and EPRI were in the 12 development of the fire PRA methodology.
13 So because of that development of that 14 method, it was a great opportunity for those writers 15 of the method to go out and see these events and 16 understand the hazard from these types of events that 17 eventually got rolled into the model that we have in 18 the 6850 document.
19 Now, if you look in 6850, it was published 20 in 2005. In 2004, the NRC amended its fire protection 21 regulations to allow the use of performance-based 22 standard NFPA 805 for fire protection, and that 23 standard NFPA 805 was issued in 2001.
24 So since the Commission policy statement 25 of '95, the Agency has been moving towards a more risk NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
139 1 informed approach to a number of regulatory topics, 2 fire protection being one of them, so it was time to 3 have this method published, and it was developed with 4 the best available data and information at the time 5 with a joint team of experts to develop those methods.
6 So as I said earlier, they looked at that 7 San Onofre event. There was also an operational 8 experience assessment report that was put together 9 that looked into energetic faults that happened from 10 1986 forward to 2001, which also provided a lot of 11 information on the type of energetic failures that 12 have occurred in the plants.
13 If we look more specifically into how it's 14 modeled, we talked a lot about the Zone of Influence 15 this morning, but basically around the initiating 16 enclosure, you assume that any equipment inside a 17 three-foot horizontal or a five-foot vertical zone is 18 damaged and ignited.
19 And this was basically developed through 20 an expert solicitation process from the observed OpE.
21 There was no data that was used. However, it is the 22 model that has been accepted in a number of 23 applications that support the NFPA 805 transition.
24 So just another photo of the San's event 25 from 2001. It was a fairly significant thermal event NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
140 1 from the duration that it took to fully suppress, but 2 also we have some international operating experience 3 from Onagawa.
4 Back in 2011, they had a HEAF that damaged 5 a series of switchgear. The event also lasted for a 6 good portion of time. However, a lot of that duration 7 was due to difficulties for their fire brigade, off-8 site fire brigade, to respond to the event, but you 9 can kind of see from these photographs the extent of 10 damage to the initiating equipment and adjacent 11 electrical enclosures.
12 Subsequent to the issuance of 6850, we had 13 the Frequently Asked Question program that was 14 identifying a number of questions industry and our 15 stakeholders had on the application of the method.
16 One of those was FAQ 07-0035, and the resolution of 17 that frequently asked questions document in not only 18 an enclosure memo, but in supplement one to 6850.
19 And what this guidance provided was a 20 method for modeling high energy arc faults that occur 21 in bus ducts, so you can use that information to model 22 a HEAF that has occurred or would occur in a bus duct.
23 And here we're just showing a few examples 24 on the left and right of actual OpE showing the failed 25 bus duct at Diablo Canyon as well as at Columbia, and NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
141 1 the photograph in the center is showing you testing 2 that we did in Phase 1 of the OECD program where we 3 had a bus duct that was removed from the Zion Nuclear 4 Power Plant and tested in that program.
5 Now, the bus duct conductors themselves 6 are copper, but the enclosure is aluminum, and from 7 the observations that we've observed in the testing, 8 we saw that there was a lot of aluminum interaction 9 that created a more energetic event than we had seen 10 in other non-aluminum based experimental tests that 11 were performed.
12 So that's a quick overview of where we are 13 right now. Looking at those two models, the key point 14 to get there is it's one size fits all. You have a 15 model for electrical enclosures, and the zone of 16 influence of three-foot, five-foot, that matches up 17 whether you're in low voltage gear, medium voltage, 18 whether your available fault current is 10,000 amps or 19 30,000 amps.
20 It doesn't really get into a lot of the 21 details that really impact the amount of hazard that 22 you would deal with in the plant, so moving forward, 23 we'd like to improve what's currently being used and 24 available.
25 And one reason that, you know, I'm NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
142 1 presenting this information here is I think there has 2 been either miscommunication or misunderstanding from 3 some of our stakeholders that, you know, we're only 4 looking forward to working with the Sandia models and 5 developing new models.
6 At this point, we haven't made any 7 decision on what's the best approach? What's the best 8 model to accurately characterize the hazard?
9 So there are a lot of arc flash models out 10 there and we're looking to see if they have any 11 utility in the work that we're doing or if we can take 12 those models and make any modification to them to 13 support the hazards that we're trying to model.
14 So the next set of slides that I'll be 15 going back to, they were actually presented in an 16 April 2018 public workshop providing a high level 17 approach for how we're looking into doing the 18 modeling, and we kind of break it into three different 19 categories.
20 The first category would be the bounding 21 model, which would basically be what we currently use 22 in 6850 or the supplement for the bus ducts. You 23 basically have one model that captures everything.
24 You make it bounding to capture everything and you use 25 it.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
143 1 Evolution from that would be where you 2 take some plant information, whether it's the amount 3 of energy, the fault protection scheme, equipment 4 configuration, safety class, material, and you can use 5 that information and basically subdivide it into a 6 bunch of different bins, and each bin would have a 7 different ZOI or a different hazard associated with 8 that based on the potential for the equipment and the 9 category, so that would kind of be the second category 10 class of modeling.
11 And the third one, which we retermed 12 dynamic zone of influence, would basically be where 13 it's plant specific. You go out and you identify what 14 your available fault currents are, what your voltages 15 are, what your, your key parameters, whether it's the 16 bus bar gap spacing, component configuration, what 17 have you.
18 You feed that into the model, and then the 19 model outputs typically an incident energy or a heat 20 flux, and then you can take that information, and with 21 the target fragility information which Sandia and I 22 will be talking about later, you can connect the two 23 dots together such that if you know you have a 24 vulnerable target in your fire PRA scenario that's X 25 feet away, you run the model and you determine that NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
144 1 your heat flux or incident energy is Y, you know, 2 kilowatts per meter squared. Then you can look at 3 your target fragility and determine whether that 4 target would have failed or not.
5 So something like that's similar to what's 6 done in some of the arc flash models, and we'll go 7 through some of those later, but those kind of lays 8 out the three different categories of modeling that 9 we're looking into.
10 We're not tied into one particular, and if 11 I had to guess, we probably have a mix of some of 12 these categories for PRA applicability and ease of 13 application of the method.
14 So just a refresher again, bounding 15 assumes worst case, one size fits all. You need the 16 least amount of information to apply it and it's 17 fairly simple to apply. However, you get the least 18 realistic results for the majority of the cases, and 19 the costs for the application, at least the 20 application is low. The costs to your risk profile of 21 the plant might be high.
22 The second category, refined bounding ZOI, 23 this is where we subdivide the equipment into 24 different categories, whatever they may be, whether 25 it's equipment power, energy, protection schemes, so NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
145 1 on and so forth. It provides more realism. If you've 2 followed what we've done in the heat release rate 3 realm for electrical cabinets lately, that's kind of 4 the approach that we've taken.
5 We basically better characterized the 6 combustible loading in those cabinets and developed 7 more realistic and refined heat release rate profiles 8 for the different bins, so it's a similar analogous 9 approach to that, however it's a little more costly 10 for the development, and the application, you'll need 11 more information to apply it.
12 And then the final one is where you have 13 all of the detailed information and you're running, 14 you know, equations or correlations developed, the 15 critical heat flux or critical incident energy to your 16 fragility targets or your targets that have a specific 17 fragility, and then you're tying that fragility back 18 to the calculated fluxes or incident energies.
19 It has definitely the potential to provide 20 the most realistic results. However, it is very 21 timely and complex to apply and it may be costly, at 22 least in the application frame.
23 MR. STONE: Could I ask a quick question?
24 I think it's just a quick question. Are we looking --
25 like today, we are, for a lot of fires you have for a NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
146 1 specific cabinet, you might have three to four fires 2 based upon the heat release rates. Are you looking at 3 something similar for the HEAFs where you have a range 4 of fires or HEAFs that might occur for that particular 5 cabinet or source?
6 MR. TAYLOR: So there's a number of 7 parameters that influence the amount of energy that 8 goes into the HEAF, and I think the working group is 9 trying to identify key parameters, and the one that is 10 on everybody's mind is duration. That's the one that 11 has the biggest impact on the amount of energy 12 released.
13 So to get to your question, if you look at 14 the event tree that Nick put up earlier, one of the 15 intents of that event tree is to make it application 16 specific to the plant, and then when you're in 17 whatever the end state is, there would be a duration 18 associated with that. So because that's the most 19 important and the least known from the OpE, that event 20 tree helps us limit the scenario and the hazard for 21 your plant configuration.
22 So I think the short answer to your 23 question is no. We're not trying to provide a number 24 of different HEAFs that you have to analyze for your 25 component. We're trying to make it, at least on the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
147 1 PRA application, easy, easier to limit to one HEAF.
2 MR. STONE: That does make it easier, but 3 does it make it bounding because the, obviously any 4 particular source could have a single phase or a 5 three-phase fault, so are we looking at an average 6 heat release rate for a particular source again?
7 MR. MELLY: So I don't think that is 8 currently off the table. It was brought up at the 9 working group last meeting. However, the discussion 10 kind of focused around on what would the distribution 11 of durations look like and how could we potentially 12 incorporate that into the methodology?
13 So would we be doing typically a 98th 14 percentile fire into 75th? I don't think we've gotten 15 that far in the discussion. I don't think we're 16 leaning in that direction, but we might -- we're going 17 to be trying to inform the distribution of potential 18 event potentially using the rate as our main 19 parameter.
20 MR. STONE: Thank you.
21 MR. TAYLOR: Okay, so as far as the 22 status, and I kind of covered this in the opening, but 23 no approach right now is off the table. There are 24 some approaches we think have more benefit than 25 others, but everything is still on the table.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
148 1 We need to understand and evaluate, you 2 know, existing and new hazard models. So I'm going to 3 go over some of the existing coming up, and Chris and 4 Paul are going to talk about the new hazard models 5 that we're trying to develop that seem to be more 6 realistic and representative.
7 There's also kind of a balancing act 8 between, you know, how much detail do we need versus 9 -- how much detail do we want versus how much detail 10 do we need?
11 So we can make, you know, the most 12 scientifically advanced model that is very 13 computational intensive and needs a lot of information 14 to support it, but in the end, that's just going to 15 make it very difficult on the application engineer to 16 apply it, so we have to balance, you know, what our 17 needs are with what, you know, what the science can 18 take us to.
19 And I think a lot of this is being driven 20 also by the working group to help us form what's 21 important and what's not important so we can focus on 22 the important things for the test program as well as 23 for the risk assessment.
24 Just briefly what I'd like to do is cover 25 some of the measurements that we're making and how NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
149 1 they apply to the models, so measurements from the 2 phase, excuse me, from the fall testing if the 3 computer stops freezing.
4 But basically with NIST, we used a number 5 of different heat flux and incident energy gauges.
6 One of them was a slug calorimeter, the ASTM slug. My 7 computer froze. So the ASTM slug has been used in 8 Phase 1 of the program.
9 The tungsten slug has been developed by 10 NIST, and it was developed by them because we saw in 11 some of the Phase 1 testing that the actual copper 12 ASTM slug melted.
13 So because of that, we needed something 14 with a higher range, and NIST went and they developed 15 a new tungsten slug to provide us that heat flux 16 information at a much higher range, and we also have 17 plate thermometers.
18 So those three gauges provide us basically 19 the thermal information or the thermal insult 20 information from the HEAF exposure. We've put a 21 number of those devices on an instrument rack that we 22 have positioned three to six feet away on the sides, 23 and the front, on top, so we can kind of characterize 24 any geometrical differences between the hazard.
25 And then also on Sandia's side, and I NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
150 1 don't have it on the slide, they're using some of 2 their advanced imaging equipment, thermal imaging 3 equipment, as well as high speed cameras to evaluate 4 the thermal impacts geometrically away from the test 5 equipment, as well as being able to use it to predict 6 measurements of particle velocities, particle sizes, 7 and composition.
8 On the composition side, we're also using, 9 well, in the past, we used aerogels, but we're using 10 black carbon tape to capture particles and do posttest 11 analysis to understand different information as far as 12 concentration, and particle sizes, and oxidation 13 levels.
14 And the last piece of information that we 15 include in our test racks are another device that's 16 provided by Sandia. This is a piece of PVC, and from 17 that, we can put it and expose it to the HEAF, and 18 then using spectroscopy methods, they can go back and 19 look at it and determine what the peak temperature was 20 at that location where that device was located.
21 On the electrical side, obviously KEMA is 22 measuring the voltage current profiles, so we have 23 that information as well as the time information that 24 will allow us to estimate what the energy was 25 delivered electrically to the device at the short NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
151 1 location.
2 We also have pressure measurements 3 internal to the cabinet. So in the fall, we took four 4 pressure measurements, and I do have some of the clean 5 data presented here. We're getting much cleaner data 6 than we did in Phase 1 because of some of the 7 improvements to the fiber optic system that's being 8 used at KEMA.
9 Other things that are important to the 10 models are physical dimension, so bus gap, how far the 11 phase is separated, as well as distances to where 12 we're actually making the measurements.
13 So before we do any testing on the 14 equipment, we spend about a week at the facility 15 documenting the equipment configuration sizes. We 16 have CAD drawings of the equipment that we test.
17 So in the end, we can understand not only 18 geometrically what portion of the cabinet or the 19 equipment was damaged, but also we weigh a lot of the 20 materials, the panels, the bus bars, so on and so 21 forth, so we know what the mass loss is, so that gets 22 into the measurements.
23 Now looking at the overview of existing 24 models, there's a number of models out there. There's 25 probably in excess of 20 models that are available to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
152 1 estimate incidental energy or heat flux, and what I've 2 tried to do here is focus on some of the more common 3 models that are used for, again, this is from the arc 4 flash personal safety standpoint.
5 So the first one comes from back in 1982 6 from the Dr. Lee model, and it basically provided a 7 simple configuration. It was theoretical in nature.
8 So you basically assumed an arc volume, 9 and then you knew your distance to your target so you 10 could extrapolate that geometry to your target, and 11 then using convention heat transfer mechanisms as well 12 as an assumption on the maximum arc power, he was able 13 to determine what the incident energy was at that 14 distance, and then using some research from Stoll and 15 Artz on the human skin and clothing fragility, he tied 16 that incident energy to second degree burns on humans.
17 So the approach in a lot of these models 18 is very similar, but this was kind of the first 19 advancement in understanding the other electrical 20 hazard as he termed it back in 1982.
21 So it was used in the 1584 2002 version.
22 However, they just updated that last year, and while 23 they do mention his work, they don't include it for 24 greater than 15 kV cases.
25 So in Lee's model, the outputs, basically NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
153 1 energy inputs are voltage, time, bolted fault current, 2 and distance to your arc point. So if you look at the 3 measurements that we're using, you can see for the 4 incident energy, we can use the ASTM slugs and the T-5 caps. For the voltage, time, and bolted fault 6 current, we can use the KEMA Daq.
7 Now, one thing that is slightly different 8 from his method and what KEMA has is KEMA is going to 9 report the arc current, which is going to be different 10 than your bolted fault current because the arc has a 11 certain impedance associated with it.
12 There are methods out there to convert 13 bolted fault to arcing fault, so I think we have that 14 captured based on those other methods, but the 15 measurement KEMA is making is arcing fault current and 16 not bolted vault, and then distance obviously through 17 physical measurement. We'll know what the distances 18 are between our machine component and we're taking the 19 measurements. So that was a theoretical model.
20 Moving on, there have been some other 21 semi-empirical models put together. The Wilkins Model 22 again uses incident energy as the output, which we can 23 get through the ASTM slugs and T-caps, but also it has 24 a lot more inputs to it.
25 So you have a variety of voltages, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
154 1 including system voltages and arc voltages, as well as 2 current and time, all of which can be developed or 3 collected from the KEMA data acquisition system.
4 Physical measurements include distance to the arc, 5 enclosure dimensions, and the gap between the 6 conductors.
7 And then the last piece is you need to 8 know your electrode voltage, so you can get that from 9 literature. Electrode voltage is basically the 10 voltage drop between, around your cathode or your 11 anode between the arc.
12 So Gammon took that model and basically 13 simplified it. So again, it has a lot of the 14 information that was just simplified from the previous 15 model, and again, we can use the devices that we're 16 using at KEMA to measure testing to support that model 17 evaluation.
18 Another simple, but fairly used model is 19 the model shown here. Basically it has two different 20 configurations whether you're in ambient air, open 21 air, or a box configuration. So it produces the max 22 incident energy as an output, and as input, we need 23 the currents, durations, and distance to the device 24 and target.
25 One of the most common approaches for NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
155 1 estimating incident energy is the IEEE approach. In 2 2002, they developed guidance which eventually got put 3 into a regulation for OSHA, and basically they used 4 200 tests to inform that model.
5 It is empirically statistically derived, 6 and what it does is basically provide a number of 7 boundaries for worker safety dependent on the type of 8 incident energy and the type of protection that 9 they're required to have to work on energized 10 equipment.
11 So the outputs, once again, it's incident 12 energy. Inputs are voltage current, time, gap 13 conductor, or conductor gap, which, you know, is used 14 as a surrogate for arc voltage, distance to the 15 target, as well as some coefficients based on some of 16 the experiments that's used that you can pull from 17 literature.
18 As I mentioned earlier, they just released 19 a new version. They included an additional 1,800 20 tests and they increased the number of configurations.
21 So in 2002, it was just opened or closed box. Now 22 they have vertical or horizontal closed and open 23 configurations.
24 It's a much more complex model. It was 25 statistically derived to match up with all of those NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
156 1 experiments, but if you look at what the range of that 2 model covers, it pretty much covers everything that 3 we've experienced inside the plant. So, you know, for 4 applicability, you know, it seems applicable for what 5 we're dealing with.
6 The output is incident energy. We can get 7 that from our slugs. The input is current, voltage, 8 duration, distance to the device or the target, 9 conductor gap, enclosure dimensions, as well as 10 equipment configuration, and that relates to the 11 orientation between the busses or the conductors and 12 the target has an impact on the incident energy 13 received.
14 Okay, so those are kind of the common --
15 do you have a question?
16 MR. STONE: Yeah, I apologize. I wanted 17 to go back to the distribution issue on the, like on 18 the types of heat release rates. The rest of our fire 19 PRAs include a distribution of the impact and we model 20 it. We can model it in a manner in that way.
21 I want to make sure that the approach that 22 we go for here is not a bounding approach that has 23 some sort of likelihood of failure if we're going to 24 use a distribution of just one failure, one frequency, 25 or the ability to use a distribution and have NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
157 1 different initiators for the same cabinet, for 2 example.
3 Because otherwise, I'm sure I'd want to go 4 to another, after all of these tests, to go to a 5 conservative method for what the frequency is and the 6 distribution. I'm a little concerned on that if you 7 follow my question. I may not be articulating it 8 well.
9 MR. TAYLOR: Let me just give a few things 10 on it and you probably can answer it better. I think 11 I understand the concern with the distribution, and I 12 think what Nick will tell us is that the duration is 13 kind of where we're going to address that.
14 MR. STONE: Yes.
15 MR. TAYLOR: All of these other parameters 16 like voltage, and current, and conductor spacing, we 17 can make that application specific, so we kind of 18 cover that there. The unknown, this uncertainty is 19 the duration of the events, but I think that's --
20 MR. MELLY: Yeah, that's kind of what I 21 was going to say, that we're not trying to give a one 22 size fits all. We are trying to cover the range of 23 applicability for these types of events. Some of the 24 complication comes in when we even look back all the 25 way to frequency in that the frequency of having one NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
158 1 of these HEAFs is already this large energy release.
2 So we're kind of hardwiring in the fact 3 that this is a very energetic event, whereas those in 4 the typical heat release rate mind frame where if I 5 have a low heat release rate of 10 or 20 kilowatts, 6 that's the lower end of the distribution.
7 MR. STONE: Yes.
8 MR. MELLY: Well, those events aren't even 9 included in the frequency at all, so there is some 10 correlation between how we define the frequency and 11 then how we're doing the modeling. So we are trying 12 to take in probabilistic effects in terms of duration.
13 However, we're mindful as well of what's constituting 14 that frequency at the forefront.
15 MR. STONE: I understand. I guess I was 16 a little bit concerned about that we were making -- we 17 were concerned it might become complex, and I would 18 rather spend more money and resources to get the 19 answer right, especially on a GI response or something 20 like that, than to try to go with a simplified 21 approach, at least to have the option once the methods 22 are developed.
23 MR. MELLY: Yeah, and that is an active 24 area for the working group where some of the concern 25 is that if we went with kind of the two to three type NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
159 1 scenario approach for a single cabinet, without a 2 really good way to handle the probability of having a 3 two-second, four-second, 10-second type fault, you 4 could have vastly different target sets that are 5 associated and damaged in each of those events.
6 MR. STONE: Okay.
7 MR. MELLY: So we're trying to deal with 8 how to make that assessment right now, and we think, 9 or at least we're leaning towards duration as being 10 the primary parameter that can drive our assessment, 11 which is why our table that we put up, the fault tree, 12 sorry, the event tree kind of style there will lead us 13 to a path of selecting a duration which will then link 14 with the zone of influence depending on our situation, 15 so we're hoping that it will be addressed through our 16 current framework.
17 MR. STONE: All right, thanks. As I said, 18 just to have the option of the more complex if there 19 is more data available to do a more realistic 20 assessment if that's possible would be --
21 MR. MELLY: Yeah.
22 MR. STONE: -- preferable.
23 MR. MELLY: And those are things that 24 we're absolutely taking into account, and we are also, 25 a lot of the working group is very application NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
160 1 oriented in that we're understanding what information 2 is easy to get at the plant.
3 If we tell you, "I need to know the gap 4 spacing. I need to know the manufacturer. I need to 5 know exactly what's in the cabinet," and then I go to 6 my plant personnel and they say, "Well, I don't have 7 that information."
8 MR. STONE: Yeah, I understand.
9 MR. MELLY: We're trying to, we're going 10 to try and link all of that together so we're not 11 giving a fully bounding approach. It may be useful to 12 have a bounding approach in terms of resources, but we 13 hope to have the ability to sharpen the pencil and get 14 more detail oriented information.
15 MR. STONE: Thank you.
16 MR. TAYLOR: Okay, the last piece, at 17 least on the thermal hazard modeling, was just a quick 18 comparison between some of the models to the data that 19 we had from last fall for the medium voltage 20 switchgear tests.
21 So this first slide here, I'm looking at 22 the ASTM slugs, which is basically a thin copper disc 23 thermal couple on the backside of it to take the 24 measurement, and comparing the 1584 2018 version to 25 our measurements.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
161 1 And the maximum overprediction between the 2 model and our actual measurements for all four tests 3 was almost 11 times what we observed in testing, so we 4 measured, you know, 550 kilojoules per meter squared 5 and calculated, you know, 6,100 kilojoules per meter 6 squared. On the minimum side, the minimum 7 overprediction was about a factor of two.
8 So the only note, there were, in one of 9 the experiments, there was some damage to our 10 instruments, so that may affect, if we had that 11 information, it may affect these results, but without 12 that, that was kind of the comparison between the 13 experimental and at least this model.
14 You can also look at those ASTM slugs and 15 compare it to the Lee model. We already know Lee is 16 going to be conservative because of his maximum power 17 assumption, and because of that, you can see it's 17 18 times max and three times minimum overprediction with 19 the model compared to our results. If we look at the 20 --
21 MR. HAMBURGER: I have a question from 22 Steve Turner. There's a question from Steve Turner on 23 the webinar about the 1584 model. He said, "You 24 mentioned the IEEE 1584 model has a minimum distance.
25 Is there a maximum distance for applicability?"
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
162 1 MR. TAYLOR: I'd have to look. I don't 2 know off the top of my head.
3 MR. HAMBURGER: Okay, thanks.
4 MR. TAYLOR: So for the -- there we go, 5 no, back one. So then if we look at the T-caps or 6 tungsten thermal capacity slugs that NIST developed 7 which have the higher range and we compare those to 8 the model, maximum overprediction of 26 times and a 9 minimum overprediction which was near, you know, 10 fairly close to what we measured.
11 So one thing with these slugs, because 12 they have a higher range on the lower end of the 13 exposures, they tend to underpredict the thermal 14 energy received, so that's why you can see some of 15 these maximum overpredictions being much higher than 16 the ASTM slugs. If we use the same T-cap data and 17 compare it to Lee's model, as expected it's an even 18 higher overprediction.
19 So, you know, the question that we get 20 asked, "But what can we use to assess the hazard?" and 21 we looked at existing models. And from what we -- I 22 put them here, but as well as what I've looked at from 23 other models, it seems like a lot of them are 24 overpredicting by quite a bit compared to what our 25 measurements are.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
163 1 So we're looking to better understand why 2 the overprediction is. Some of the reasoning is, 3 well, the tests that they ran were either in open air 4 or they were in a box, a cubicle with one end open.
5 So if you look at the OpE, you know, 6 there's only a few cases where you actually have an 7 open compartment, and typically a lot of these events 8 happen in a closed compartment, and you need to get 9 some burn through the enclosure to actually allow the 10 effluent or the energy to really exit the enclosure.
11 So that's one area that we're looking into 12 as possibly being a difference, but if you look at all 13 of these models, they follow a very similar approach, 14 and it's an approach that we're looking to use with 15 some of the CFD modeling that Sandia is doing.
16 That is we understand what your electrical 17 energy is, any contribution from the metal reaction, 18 and then you transfer that energy out to your targets, 19 and then once it's transferred out to your targets, 20 you understand what the fragility of your targets is 21 and you can see whether those fragilities are exceeded 22 or not.
23 So the last point I wanted to make is, you 24 know, there is some potential for, although they 25 overpredict, there is some potential for the existing NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
164 1 models to be modified to make them more applicable and 2 more in line with what we're seeing.
3 MS. VOELSING: This is Kelli Voelsing with 4 EPRI. I think you mentioned at the beginning that 5 most of these models were developed primarily, as it 6 says on the slide, for protecting humans and 7 developing PPE.
8 So, I mean, in that situation, it might be 9 desirable to have a conservative model that gives you 10 a higher than actual heat so that you're developing 11 your PPE to protect against that.
12 So maybe it's not surprising that these 13 models are conservative because it's good for them to 14 be conservative and bounding. Have you investigated 15 that?
16 MR. TAYLOR: I haven't looked into that 17 too much. I do know that some of the models in the 18 1584 2018 edition have so many parameters in them that 19 they had to high-fit the data.
20 So from those cases, it doesn't appear 21 that they tried to look at what the worst case data 22 was and fit a model to that. They used all of the 23 data, and through a very complicated statistical 24 regression model, they fit the model to that data.
25 But I think it's a valid point that on the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
165 1 personal protection side, you may want to, you know, 2 stay on the conservative side. I mean, that's one of 3 the logic that Lee used when he put it together was 4 that, as well as he couldn't predict what the phase 5 difference between current voltage could be to 6 estimate the power, so he assumed the worst case.
7 Regardless, I think all of these models 8 follow a similar form, and there are potentials that 9 -- there is a potential for us to make some 10 modifications to that form to more accurately 11 represent the hazard that we're seeing.
12 MR. MELLY: Yeah, another important note 13 here is that all of these tests were performed at very 14 short durations compared to what we are testing. The 15 other piece to that is while they may be beneficial to 16 be conservative for human protection, in the model 17 itself, we don't see any explicit safety factors that 18 are being applied.
19 So the question that we have right now is 20 why are they overpredicting by so much? Based on what 21 we've looked at, we don't see any safety factors put 22 in for conservatism.
23 MR. TAYLOR: The last topic I wanted to 24 cover before I turn it over to Sandia is that, you 25 know, the working group is continuing to progress NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
166 1 through, you know, measuring different information 2 they give back to the working group. So one of those 3 is air holdoff voltage, and we looked at using the 4 EPRI mock switchgear test unit to evaluate the arc 5 over effects.
6 Unfortunately, we couldn't procure or use 7 that in the upcoming tests. So one thing we did was 8 looked at Sandia and said, "Okay, is there any way to 9 measure air breakdown strength?" And basically they 10 came back to us with one approach that has a standard 11 associated with it, and we looked to doing that moving 12 forward.
13 So the whole reason why we're looking into 14 this is basically in some of the Phase 1 OECD test 15 results, it showed that when we ran an experiment, 16 some of the unenergized conductors that were in this 17 test cell, after the tests, were nonfunctional. So 18 you had a lot of surface deposition on those 19 conductors and it reduced the insulation between those 20 conductors so that it couldn't function and required 21 a significant cleanup or remedy to make those 22 functional.
23 So that's the surface piece, but also the 24 question arises, "Well, what if we have another piece 25 of equipment in the room? You have all of these hot NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
167 1 gasses and metal particulate that's being transported 2 to this other equipment. Are there any potentials for 3 the other equipment to arc over?"
4 So if you have another set of, you know, 5 6.9 kV gear in the room, can a HEAF in one lineup 6 affect the other lineup of gear? So we're looking at 7 measuring in the upcoming tests what the breakdown 8 strength is.
9 There is a model that Sandia came across 10 in their literature review that estimates the air 11 holdoff or breakdown strength based on temperature as 12 well as the concentration of metal vapor in the medium 13 or the air.
14 So the plot that's shown here is basically 15 showing ambient conditions where, you know, air has 16 basically a 30,000 volt per centimeter holdoff 17 strength.
18 As you add, you know, a certain 19 concentration of metal vapor to air, it decreases, 20 whether it's copper or aluminum. If you increase 21 temperature, it also decreases the holdoff strength, 22 so you can see gas density or temperature, as well as 23 metal vapor concentration has an impact on what your 24 holdoff strength is.
25 And also something worth noting is NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
168 1 aluminum ionizes more easily than copper, so that has 2 some of the impact of why the aluminum is showing a 3 lower holdoff strength than the copper.
4 So with this model, what we looked at 5 doing is running these holdoff strength measurements 6 and using the data that we received from the tests 7 that we performed this fall to look at this model and 8 how well it characterizes the concern.
9 So the approach is based off ASTM D2477, 10 test method for dielectric breakdown voltage and 11 strength of insulated gasses at commercial power 12 frequencies. However, if you look at that standard, 13 there are some things we needed to modify.
14 Because the HEAFs are such short duration 15 events, we needed to change how the voltage profiles 16 applied to the device, so we made adjustments that 17 allows us to rapidly increase the voltage a number of 18 times throughout the duration of the HEAF event such 19 that we can make multiple measurements during one 20 test.
21 We're also looking at using ultraviolet 22 illumination to ensure that there's consistent results 23 to standard step voltage approaches. So in the end, 24 it allows us to determine what the actual holdoff 25 voltage strength is and any confirmation of the model.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
169 1 It looks like this slide is missing a 2 photo on the top left. It was just a photo of the 3 actual probe. So basically if you look at the 4 illustration in the middle, the blue, basically you 5 have two spheres that are separated, metal spheres 6 that are separated a certain distance apart.
7 You apply a voltage between that. As your 8 air medium goes between those probes, if it has, you 9 know, low enough, or high enough conductance between 10 the probes, it causes a breakdown of the air and we 11 can measure that with the instrument's current voltage 12 instrument that's being used to capture it.
13 So we plan on using two units. The 14 evolution of the deployment of the units are going to 15 be iterative. So initially, we're going to put them 16 in open air, run a test, make sure that after the 17 test, make sure that they're in the cloud or the 18 effluent from the HEAF.
19 If they weren't, in the subsequent tests, 20 we're going to make adjustments to make sure that 21 they're in that cloud and run the test again. And 22 then with the data from those two tests, we're going 23 to be able to look at National Electric Code 24 requirements for holdoff strength and see if we exceed 25 those.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
170 1 If we do exceed those, then we're going to 2 put it in an enclosure and run tests again. If we 3 don't exceed those, well, then it shows that it's 4 probably not an issue.
5 But regardless, I think we're going to 6 continue to run these measurements throughout this 7 upcoming test campaign to make sure we can better 8 understand what any holdoff strength is in open air as 9 well as in an enclosed compartment.
10 Oh, there is the photo. So basically it's 11 just a photo of arcing between two of the spheres.
12 As far as where we're going to put it in 13 the test cell, probably anywhere between eight and 12 14 feet away from the test object, four to six feet above 15 the floor, typical locations where you're going to 16 have, you know, energized conductors within the room.
17 We're going to measure the temperature of 18 particle concentration and current voltage, and all of 19 it in the system and configurations shown on the 20 right.
21 So that's it for my presentation. Are 22 there any questions? Is there anything on the 23 webinar?
24 MR. FLOYD: Yeah, Jason Floyd from Jensen 25 Hughes, just sort of a question. You have the various NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
171 1 sort of simple empirical models and we've shown, okay, 2 a factor of two to, you know, whatever factor of 3 conservatism.
4 Is there plans or have you thought about 5 -- I mean, we have all of the information on plant 6 configuration that we've collected, and the voltages 7 and potential durations, of at least assessing what 8 the model says they exist now.
9 You know, where do we think the ZOI sits 10 gives us some confidence that, yeah, we've got a 11 problem we need to investigate, but if these 12 conservative models say that, you know, we're not 13 there, then, you know, that might cause us to rethink, 14 and also the check for, you know, instrumentation 15 placement in these tests?
16 MR. TAYLOR: So on the first piece, the 17 modeling, and have we done any work to kind of assess 18 where we're at with the current modeling, we haven't 19 done anything formally in the working group.
20 But what I have done kind of on the back 21 of the envelope is to look at some of the literature 22 that's available comparing copper versus aluminum, and 23 there is some literature that's been done that 24 basically it takes the input and output and relates it 25 by one parameter, I think they call it an effective NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
172 1 heat transfer coefficient, and basically it shows that 2 you have, you know, you can have about three times as 3 much energy from an aluminum versus a copper.
4 So with that information, I used 1584 as 5 well as the existing PRA guidance for the three-foot, 6 five-foot, and compared, you know, what could it be 7 with an aluminum assuming 6850 based on copper? And 8 from that, it took it to anywhere from six to 12 feet 9 from the three-foot.
10 MR. FLOYD: With the conversatisms that 11 we've seen, so.
12 MR. TAYLOR: Yeah, so assuming 6850 is 13 conservative, which, you know, it's meant to be a 14 bounding approach, we're not sure it captures 15 aluminum, but it's a bounding approach, and 1580, you 16 know, the heat transfer.
17 Now, it wasn't using all of the 18 information to extract that because you needed 19 fragility, right? So I used like a common fragility 20 as kind of a baseline. So I tied it to that baseline 21 fragility and then converted it for the aluminum.
22 So, yeah, I think the short answer is 23 there is some conservatism in that approach, but if 24 you're looking for kind of a what could it be? It's 25 somewhere in that window, and until we -- That's what NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
173 1 the working group is trying to do is, "Is it closer to 2 the three foot or is it closer to the 12 foot?"
3 MR. MELLY: Yeah, but in terms of have we 4 done the Lee approaches that have the 39 times 5 conservatism built into the output, and then linked 6 that with a theoretical fragility for what cables 7 would say, and done what our postulated zone of 8 influence would be? No, that has not been completed.
9 I don't know if that would be a useful 10 endeavor. It's definitely something we should try and 11 investigate, but with a conservatism anywhere from 17 12 to 39 times what we're seeing in testing, we may be 13 vastly overpredicting the zone of influence, and 14 again, we do have to link it with the fragility data 15 which we're incurring the stages to develop.
16 MR. FLOYD: Yeah, well, I'm just 17 suggesting that if you have some estimate on the bias 18 for your model, you can do a prediction and correct 19 it, and at least, you know, get some ballpark as to 20 where we think we are.
21 You know, make sure that we're thinking of 22 our test instrumentation and what we're thinking 23 about, and at least it's better to go in with some 24 ballpark and help guide your thinking than to sort of 25 go in blind.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
174 1 I mean, we've got data to at least get 2 some rough assessment, you know, for whatever it's 3 worth, you know, to sanity check what we're thinking 4 or testing.
5 MR. TAYLOR: Yeah, it's a good point, but 6 as far as, you know, at least our logic right now, 7 where we're putting our instruments for the instrument 8 racks, you know, we're tying that back to the 6850 9 model, so three foot, and then six foot, you know, 10 between that.
11 For some of these other measurements like 12 the conductivity measurements, we're looking more for 13 the far fields, not the -- outside of the thermal 14 effects, so that's why we're putting those farther 15 away, possibly back against the far wall in a test 16 enclosure, which would be dependent on where we can 17 put the actual test device, but, you know, you're 18 looking at 10 to 15-foot away from the object, test 19 object.
20 MR. HAMBURGER: We don't have any 21 questions on the webinar at the moment. Anyone else 22 in the room, a question or a comment?
23 Give us two minutes to pull up this 24 presentation. Our computers are --
25 All right, let's take a 15-minute break NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
175 1 while we get this sorted out. We'll come back at 2 2:45.
3 (Whereupon, the above-entitled matter went 4 off the record at 2:28 p.m. and resumed at 2:44 p.m.)
5 MR. HAMBURGER: Okay. It's 2:45. We have 6 sorted out our computer issues and, Gabe, you want to 7 --
8 MR. TAYLOR: Sure.
9 MR. HAMBURGER: -- introduce the Sandia 10 folks?
11 MR. TAYLOR: So, apologize for the little 12 glitch on the computer. Welcome back, everybody. So, 13 the next slide set will go through the HEAF modeling 14 effort that's being conducted by Sandia National 15 Laboratories, basically providing the modeling 16 approach and analysis.
17 So, why are we doing this?, is basically 18 a question arose you have this aluminum that showed in 19 Phase 1 to be an issue, what can we do better to model 20 it? So, with that, we started looking around, 21 initially, we were looking at going to one of the 22 universities, but that resource is no longer 23 available.
24 So, we identified that Sandia had a lot of 25 capability and existing models and expertise that NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
176 1 could be used to help address this issue and the 2 question, so we turned to them for their expertise to 3 help us develop that asset.
4 So, in front of me, or here with me today, 5 we have Dr. Chris LaFleur, from Sandia, as well as Dr.
6 Paul Clem. And they are going to provide that 7 information. Chris?
8 MS. LAFLEUR: Thanks. Yes, I'm Chris 9 LaFleur. I do fire risk studies at Sandia National 10 Laboratory, and Paul is in the electrical science and 11 experiments group. There's also two other members of 12 our team, it's a complex problem, as we've said, so 13 we've got some other resources on the line that you 14 may hear from today.
15 One is Matt Hopkins. He is our expert on 16 ARIA. He was involved in the development of it, so he 17 can speak to its applicability on environments that we 18 experience in a HEAF event.
19 I also have Caroline Winters on the line 20 and she is working on the small-scale experiments and 21 measure the spectral emissions from those arcs. So, 22 next slide.
23 So, our general overview of what we're 24 covering today, we have a few slides on the background 25 of why we're doing the physics modeling, Paul will NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
177 1 cover those. And then, we have our overall plan for 2 the arc modeling.
3 And then, a lot more details on the 4 approach of how that arc will be modeled and how that 5 will hand off into characterizing the thermal 6 environment and other hazardous environment that is 7 emitted from the HEAF.
8 We'll talk more about the next steps that 9 we're going to do for the modeling. Paul has some 10 results of our current tuning of the model and where 11 we're at. We have details on the Sandia National Lab 12 models, the ARIA and FUEGO models that can be coupled 13 together. And Matt will give that information.
14 And then, I have a few slides on where 15 we're at evaluating the target fragility and the 16 failure criteria. And that's where my voice is 17 leaving me. So, I'm going to let Paul talk about the 18 next few slides.
19 MR. CLEM: Okay. Great, thanks. So, I 20 think it's been pretty clear from what Gabe mentioned 21 that there's a big disconnect, in many cases, between 22 some of the empirical models or calibrated models for 23 arc energies and what's really measured in HEAF 24 events.
25 And on top of that, part of the reason we NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
178 1 were brought in, I believe, is to look at the effects 2 of aluminum and how could we begin to quantify why 3 aluminum and copper behave so differently in 4 switchgear or in busbars and other kinds of operating 5 equipment.
6 So, there's really a desire to have a 7 physically-based model to understand what is the 8 effect of aluminum during a HEAF? Is it oxidation, is 9 that the primary driver? Or are there other factors 10 that really affect the total radiant energy that comes 11 out of that?
12 And furthermore, can we make models that 13 are much less conservative going forward, that are 14 physically-based, they're not calibrated, they're not 15 based on these square fits, they're really based on 16 the physics of the model and really get much closer to 17 really what's going on?
18 So, that's really the goal of this, is to 19 have a physically-based, demonstrably correct model of 20 the energy that comes out of a given piece of 21 operating equipment. We'd like to do that simply 22 knowing the current going into the switchgear or the 23 equipment, the gap between electrodes, and the metal 24 composition.
25 So, have no other calibrated factors, no NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
179 1 other fits to the data, just simply look at the 2 physics basis for air conductivity, how does the 3 plasma develop, and what is the energy radiated out 4 from that plasma?
5 So, if you go to the next slide? And this 6 would let us look at a much wider variety of 7 components and situations. So, this isn't a problem 8 just for nuclear plants, we actually were brought in 9 earlier for another EPRI program, looking at 10 photovoltaic DC arc faults, where photovoltaic 11 facilities, basically a constant current source, and 12 if an arc fault develops, you can have a very 13 extensive DC arc fault and it was found that many of 14 the predictions from arc damage from DC arc faults 15 were also not well-grounded.
16 So, we began developing a model for 17 looking at DC arc faults, based on the physics, and it 18 turns out that this should overlap very closely with 19 AC arc faults, as far as the basic physics. So, 20 that's where this is going.
21 So, we're working with Tom Short on DC arc 22 faults, open box experiments. In fact, we just 23 tested, up at Detroit Edison, last week, on switchgear 24 combiner boxes, recombiners, and invertors. So, next 25 slide.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
180 1 So, for the DC arc problem, a similar 2 problem, DC arc faults, as to some of the AC models, 3 the concern there is about what's the right PPE for a 4 given arc fault event?
5 And you can see, if you look at the 6 different models there, the green, navy blue, yellow, 7 and royal blue are the predicted incident energies 8 that are in the literature.
9 And it ranges from PPE 1, all the way to 10 PPE 4, depending on which model you believe. So, it 11 can be extremely conservative, factor of ten 12 conservative, in many cases, and the goal is to have 13 a much more accurate prediction.
14 So, if you look at the actual measured 15 data in red, and you're able to actually reproduce 16 that very closely with a model by Stokes and 17 Oppenlander, in the yellow curve.
18 And this is based on some previous work on 19 both DC and AC arc flashes and arc faults by Lowke.
20 So, we're basing our work on this Stokes and 21 Oppenlander model, which is really a physics-based 22 model, it's not a calibrated fit.
23 But the goal for the EPRI EERE PV work is 24 to make an arc physical model, where if you know the 25 DC current, you know the gap between any two terminals NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
181 1 in that operating equipment, you can predict the 2 energy transfer out of that HEAF.
3 That energy transfer is not just an 4 energy, it's actually composed of three different 5 kinds of energy. There's radiant energy transport, so 6 UV, visible, and IR radiation comes out. There's 7 convective energy transport into the air around the 8 arc itself. And there's thermal energy transport, 9 which is direct conductivity of heat into the 10 electrodes.
11 And those are not accounted for or 12 separated in the current models, there's just an 13 energy that's delivered to some point in space. We'd 14 really like to separate those out, it's critical to 15 understand what form that energy is in. Is it UV? Is 16 it visible? Is it air?
17 There's different speeds, obviously, of 18 light and of conduction through air. So, it really is 19 important, we believe, to include those in any kind of 20 damage assessment moving forward, and having a 21 physics-based model lets you get that data and look at 22 how that's going to be absorbed on equipment, on 23 wires, on other kinds of components, and then, look at 24 fragility criteria based on that real energy and how 25 it comes out.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
182 1 So, if you go to the next slide? So, our 2 goal is really to start that basis of the DC arc fault 3 model, and then, extend it to AC arc fault models.
4 So, there have been a number of international studies 5 looking at arc faults and some of the -- in AC 6 equipment.
7 One of these was by the CIGRE Group in 8 2014. And, again, the goal here is to develop an 9 energy input term that can be used, based on the 10 geometry and the current going into an arc fault, but 11 that can then be taken into any other model following 12 that. It could be used in FDS, it could be used in 13 GOTHIC, it could be used in OpenFOAM, FireFOAM, or 14 other kinds of CFD models.
15 But a note they made in their analysis is 16 that the current models look at energy input, but it 17 says the most complete approaches, where the arc could 18 be modeled using physical equations, describe the arc 19 roots, where the arc attaches to the electrodes, the 20 arc plasma column itself, the effect of magnetic 21 fields on the motion of the arc, and the transfer of 22 the energy from the arc plasma to surrounding gases 23 have never been applied to internal arcs to our 24 knowledge.
25 And that's the gap we're trying to fill NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
183 1 here. So, we're trying to make an accurate model that 2 really looks at the real physics of what's going on 3 with the arc.
4 So, again, we want to take the AC current 5 and electrode gap and electrode materials and be able 6 to predict the radiation, convection, and thermal 7 energy transport by conduction with these models. So, 8 next slide, please.
9 MS. LAFLEUR: Let me comment?
10 MR. CLEM: Sure.
11 MS. LAFLEUR: Okay. So, this is our broad 12 plan, overview for how we're going to go about the 13 modeling. And the first -- well, the vision is to get 14 a non-conservative, I think we all agree on that, non-15 conservative estimate of credible energy release 16 scenarios and the respective zones of influence for a 17 range of appropriate equipment in nuclear power 18 plants.
19 And the first thing we need to do is 20 develop a model that can characterize the arc, the 21 physics of the arc and what's emitted from the arc.
22 And we would like to do that in ARIA, which is a 23 Sandia model we'll talk a lot more about.
24 We will need to couple this to a model 25 that's capable of characterizing the affluent, the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
184 1 thermal plume from a HEAF. And we have a fire model 2 called FUEGO.
3 But as Paul was saying, we could use a 4 number of different models that would take the output 5 from ARIA, that source term, and predict the extent of 6 thermal heat flux and energy emissions to targets, 7 right?
8 Either predicting the breach of when an 9 enclosure, the source enclosure would breach. We 10 could also, in that other model, predict when a target 11 enclosure could breach, if it's subjected to a high 12 enough environment for long enough, right?
13 And we could use that to evaluate the 14 fragility. Evaluating the fragility is completely 15 separate from the source term, we're thinking of that 16 completely separately. Our source term models will 17 characterize the extent of a plume or an envelope of 18 hazardous conditions.
19 And then, we'll use the knowledge that we 20 get and work out in the working group for when cables 21 or other equipment fail, when exposed to that 22 hazardous condition and we'll be able to cross those 23 together.
24 Just like the existing conservative models 25 looked at the fragility of human skin, they know the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
185 1 heat flux of when you get a second-degree burn based 2 on how long you're exposed to that heat flux, very 3 similar model for looking at when cable or an 4 enclosure might breach.
5 As Paul said, we want the input parameters 6 to be the current, the type of material that's 7 involved in the HEAF, and the gap. We want to be able 8 to predict distances to critical temperatures or 9 fluxes or failure criteria. And that's going to allow 10 us to develop the zones of influence.
11 We'll run the model at Sandia and within 12 the work group. We don't intend the model to be run 13 by licensees or other people, it's not going to be a 14 commercial software that gets developed and 15 distributed for people to do.
16 What we'll generate is the look-up tables 17 for what the energy, incident energy, that's emitted 18 based on the duration of the arc and the type of 19 equipment.
20 Types of things that we need to measure, 21 in order to validate that model, are incident energy 22 and the three different types of energy that's being 23 emitted, as Paul said, the thermal energy, the radiant 24 energy -- what was the third one?
25 MR. CLEM: Conductive.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
186 1 MS. LAFLEUR: Conductive energy, there we 2 go, thermal field. And we'll do -- we need to measure 3 fragility parameters. Like Nick was saying this 4 morning, we can't do that in the KEMA facility, but we 5 can do that separately at Sandia, based on what 6 testing is identified as what we need to. Next slide.
7 So, first thing -- maybe you should talk --
8 MR. CLEM: Okay, sure.
9 MS. LAFLEUR: -- I think I'm killing people 10 with this, sorry.
11 MR. CLEM: Okay. So, I think we're as 12 eager as you are, probably, to have these models be 13 ready. But really, to get these models correct is an 14 iterative process, to build up the model and make sure 15 it's demonstrably correct, that as we make this, start 16 with a simple arc and move to a more complicated arc, 17 that we're capturing the correct physics.
18 So, the first parts we've put into the arc 19 model so far is the ability to look at a geometry, 20 look at the input current, and then predict such 21 things as the arc's temperature, the arc radius, the 22 radiant and convective heat transfer from the arc.
23 And from that, we can actually look at the 24 mass loss rate of the conductors, of the electrodes in 25 that arc. So, we currently prepared that basic part NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
187 1 of the model.
2 We've begun taking validation data for 3 that, some of it's been done at low currents, 100 to 4 300 amps. We're conducting kiloamp, and I think, 15 5 kiloamp tests this fall, the next couple months, on 6 open box tests, to help validate those initial models.
7 Ultimately, we have to include a much 8 broader suite of physics in this. We need to include 9 magnetic forces, buoyancy of the gas, as it gets very 10 hot, which changes the shape of the arc, and then, the 11 orientations of conductors. So, we can look at 12 vertical arcs, horizontal arcs, and then, parallel 13 arcs. And each of those are going to have a different 14 behavior, we need to look at each of those 15 independently.
16 But that's the goal, again, is to have a 17 non-conservative model, that's physically accurate, 18 but can be applied to multiple situations. And I 19 think the key output of this is you can look in your 20 simulation at various spacings, various metal 21 electrodes, and look at all kinds of different 22 scenarios, which we believe will be much more 23 effective than conducting experiments in the field for 24 every kind of orientation and situation that could be 25 present.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
188 1 And so, again, the goal is to have an 2 energy output for each of those that is accessible to 3 the community. So, if you go to the next slide, 4 please?
5 So, again, the goal of this is to be able 6 to support an energy source term that could then be 7 taken up by any CFD model and then, applied to help 8 identify failure criteria and zones of influence, so 9 you can look at the radiant heat absorption, you could 10 look at conductive heat transfer, and how that leads 11 to failure criteria for given parts, whether it's the 12 outer wall, switchgear, or whether it's a cable or 13 other kind of criteria.
14 So, this is supposed to inform the working 15 group and help provide this predictive capability and 16 provide energy inputs for looking at failure criteria.
17 So, again, we're -- the key thing we're 18 doing here is really replacing the open circuit 19 voltage that's used in a lot of the models there are 20 to date, the VoC with arc resistance, which is sort of 21 treating the arc as a circuit element, where the arc 22 has a known resistivity, it has a known radius, it has 23 a known length, and it's a function of time.
24 And from that, we can look at I squared R 25 delta T. And I squared R delta T is the energy NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
189 1 output, or the integral of I squared R delta T would 2 be the ultimate energy being released by a HEAF event.
3 So, we want to look at how the joule heating energy 4 input turns into an exothermic event.
5 Again, we can measure that in these 6 calibration tests by looking first on the supply side, 7 at the arc voltage that's measured as a function of 8 time during testing, monitoring the current input into 9 the system, and that gives us the arc resistance as a 10 function of time.
11 We can measure the radiant power with 12 black calorimeters. These are 99-plus percent 13 absorptive calorimeters in the UV, visible, and IR.
14 Or we can use calibrated thermal piles to separate out 15 the different fractions that are in UV, visible, and 16 IR.
17 We can measure electrode temperatures with 18 calibrated IR cameras. So, Sandia's thermography 19 group is involved with cameras that are calibrated to 20 3000 Kelvin to look directly at the different 21 components and calibrate their temperatures.
22 We can measure the arc temperature itself, 23 using spectroscopy. So, Caroline Winters will talk 24 a bit about this, but if you capture the emission 25 spectrum of the arc, you can determine the temperature NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
190 1 at the edge of the arc and look at how that compares 2 with our model.
3 You can also look at the temperature 4 inside the arc, through spectral emission lines of 5 aluminum and copper and other components and 6 demonstrate the arc model is getting each of these 7 parts correct. So, really, it's a plasma physics 8 problem being coupled to a CFD analysis, in the end.
9 And finally, there are arc dynamics 10 associated with multi-phase arcs, especially with 11 zero-crossings and other kinds of features. We have 12 high speed cameras, up to a million frames per second, 13 that can capture the real-time behavior of this and 14 correlate that with captured voltage and current 15 waveforms.
16 So, the goal is to really understand the 17 physical behavior of the arc, the electrical behavior 18 of the arc, and then, how that turns into an energy 19 output term. Go to the next slide, please.
20 So, what we're trying to do in this work 21 is, again, have a physically-based model with a small 22 number of inputs into that model. So, the inputs 23 we're currently looking at are the gap between 24 conductive components. What is the electrode itself 25 composed of, is it copper, aluminum, is it an alloy?
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
191 1 What is the input current, the short circuit input 2 current? And then, from these, we can then generate 3 the temperature that's developed.
4 Now, air normally, of course, is an 5 insulator at room temperature, but as you heat up air, 6 it begins to dissociate and ionize and the air can be 7 separated into ionized gas species and electrons and 8 conductivity begins to turn on in air, as a function 9 of temperature.
10 So, especially above 3000 or 5000 Kelvin, 11 air by itself has a significant conductivity, due to 12 the presence of electrons that are present, and you 13 can calculate the conductivity of air as it heats up.
14 So, if you know the arc temperature, you can calculate 15 the resistance of that arc itself.
16 So, air by itself is one species, however, 17 if your plasma is in contact with metal electrodes and 18 you're having electrode erosion, those metal 19 electrodes, copper or aluminum, are first melting and 20 then, likely vaporizing, if they're above 1000 degrees 21 C, to where, now, you have not just nitrogen and 22 oxygen in the air, but also aluminum and copper in the 23 air.
24 And these aluminum and copper also have 25 known conductivities, as a function of temperature.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
192 1 You can include these in these in these models and 2 have very accurate predictions of the conductivity of 3 air by itself or air with some volume fraction of 4 aluminum or copper, iron from the steel enclosures, 5 all these kinds of things.
6 So, if you know those conductivities, and 7 these have been calculated for standard metals, 8 aluminum, copper, gold, silver, iron, as well as air, 9 you can predict the conductivity of air and then, air 10 including metallic vapors.
11 If you have these, as well as the density 12 of air, thermal conductivity, and specific heat, you 13 can then look at the CFD heat transfer and treat this 14 arc as a living being that has both thermal 15 properties, as well as a circuit element that has a 16 known resistance.
17 So, there's been some nice work by a group 18 in France, Yves Cressault, looking at these kinds of 19 things, looking at two-meter long arcs, at one to 40 20 kiloamps, and what happens when the electrodes are 21 made of iron, copper, or aluminum, and what the 22 effects are that are measured in terms of air 23 conductivity, and then, the effects in terms of the 24 energy output from those arcs. So, if you go to the 25 next slide, please?
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
193 1 So, one example is the conductivity of air 2 is much higher when it has metal vapor included in it.
3 In addition, what's called the net emission 4 coefficient, that is the amount of energy that's 5 radiated outwards in the ultraviolet and visible and 6 infrared also increases when you have things like 7 aluminum and iron present in the vapor, as opposed to 8 copper.
9 So, what's shown here are, 100 percent is 10 your total VI or total joule heating going into the 11 system. What is the fraction of energy that comes out 12 of that arc fault as radiant energy, as optical 13 energy?
14 What's found is, if you look at steel or 15 aluminum, again, because these have a much higher net 16 emission coefficient of radiant energy, you're seeing 17 something like 60 percent to 30 percent more energy 18 coming out from an arc, simply due to the electrode 19 material that's present.
20 And so, some of the energy being measured 21 in arc faults, in aluminum, may be due to oxidation 22 and that needs to be considered. But simply the 23 presence of the electrode, of the copper or steel by 24 itself, is going to increase the output energy that's 25 radiated. And that radiant energy is really what you NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
194 1 measure on a calorimeter at some distance away.
2 The convective heat transfer in an open 3 chamber propagates very slowly, on the orders of maybe 4 ten centimeters, six to ten centimeters is what we see 5 so far, and quite slowly. If you look at the radiant 6 energy, that's instantaneous, the speed of light, 7 being transmitted out to far field objects.
8 So, simply the choice of electrode really 9 matters and this can be captured in these physics-10 based models and is not captured, as far as we can 11 tell right now, in calibrated models based on currents 12 and voltages in different systems.
13 So, if we take these models and then, 14 begin to look at modeling different systems, looking 15 at an arc between electrodes where the electrodes are 16 not interacting, to where the electrodes are composed 17 of aluminum, we can predict, as a function of input 18 current, from 100 amps or 200 amps up to 10 kiloamps 19 or above, what the output power should be.
20 Again, the output power times time is 21 going to give you the total energy from that arc fault 22 event. So, if you had non-interacting electrodes that 23 were not creating vapor versus electrodes that include 24 aluminum vapor, in this case, you see a significant 25 increase in energy output.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
195 1 And the current model we're using, based 2 on this Lowke model or Stokes and Oppenlander model, 3 show about a 50 percent increase in expected energy 4 output, in terms of power, due to the presence of 5 aluminum.
6 So, again, this is -- the radiation is 7 much increased with the presence of metallic vapors, 8 such as aluminum or iron, as opposed to the presence 9 of copper in an arc fault.
10 So, we've begun making these predictions.
11 We've begun taking validation data, or experimental 12 data to validate these. But it does agree with what's 13 been measured historically for aluminum electrodes.
14 So, if you go to the next slide, we're 15 taking this model by a researcher named Lowke from 16 Australia, that's from back in the late 1970s, that's 17 a simple theory of free-burning arcs.
18 And so, this is a simple assumption about 19 the way arcs behave, that allows you to treat them, 20 again, as a circuit element, as well as a thermal 21 element that's radiating energy.
22 And so, again, for a current input, and 23 simply knowing the air conductivity, specific heat, 24 density, and thermal conductivity as a function of 25 temperature, you can predict the temperature of the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
196 1 arc, you can predict the radius of the arc, you can 2 predict that arc resistance as a function of current, 3 and then, you can look at the energy balance that 4 comes out of that.
5 So, again, we have current feeding into an 6 arc of a given geometry and then, the output from that 7 is a combination of radiant energy, convective energy, 8 and conductive energy.
9 So, what we end up with are curves, as are 10 shown here, where, again, we can now look at different 11 atmospheres, we can look at air, air with aluminum, 12 air with copper, and really compare these quite 13 quickly, this runs on a single laptop, single 14 computer, and get rapid results for different 15 geometries, and then, begin to evaluate these in the 16 field.
17 If you go to the next slide, looking at 18 this in a little more detail, the black curve on top 19 shows what's expected for the total electrical energy 20 input into the system, so the joule heating, as a 21 function of input current.
22 So, we see the input power from roughly 23 three kilowatts per centimeter length of an arc, up to 24 10 to the fifth, 10,000 watts per centimeter of arc 25 length. That energy input is being dispensed of in NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
197 1 some way by radiative transport, conductive transport, 2 and convective transport.
3 And now, it turns out, as the current 4 increases, the balance of energy coming out of that 5 arc changes. At low arc currents, most of that arc 6 energy is thermal. It's conducted into the air or 7 it's conducted into the electrodes.
8 As you begin to go to 1000 amps and above, 9 there's a crossover and the radiation output 10 dominates, from a kiloamp up to tens of kiloamps, 11 which is the area we likely care about for these HEAF 12 events.
13 So, if you look at the ratio of those blue 14 curves, which is the radiated energy, to the orange 15 curves, which are the thermal transport energies, you 16 can look at what fraction of the energy is radiated.
17 So, again, what's sense is for pure air, 18 in the blue curve, that can be as low as ten percent, 19 and it goes up to around 80 percent, as you're 20 approaching ten kiloamps.
21 When you have the presence of aluminum, 22 you have a much higher radiated energy output in all 23 cases. So, again, you're going to measure that or 24 you're going to feel that with a calorimeter at some 25 distance for your zone of influence.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
198 1 So, these simple isothermal arc models 2 appear to be effective and lets us do quick 3 estimations. We're taking this basic model and 4 putting it into this full ARIA model, to look at more 5 complex situations, where we can go from the simple 6 model and then, include magnetic forces, include 7 buoyancy, include other effects that are really 8 important for actual arc modeling.
9 So, going to the next slide, we've taken 10 these initial models and begun to perform experiments, 11 both small-scale and large-scale experiments, to 12 evaluate these. We have a couple different test beds 13 at Sandia.
14 One of these is a short duration arc, 15 where it's a capacitive discharge system, it has 16 roughly 50 kilojoules of energy available. So, it can 17 perform short duration arcs from one to 100 18 milliseconds, but at high currents, a kiloamp to 160 19 kiloamps. So, we can look at multiple line cycles at 20 many tens of kiloamps to hundreds of kiloamps, for a 21 short duration.
22 We have a different setup that enables 23 long duration arcs, one to 120 second arcs, but at 24 lower currents, only 100 amps to one kiloamp. We 25 can't, at Sandia, go above a kiloamp. For that, we're NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
199 1 relying on large-scale testing at KEMA, in 2 collaboration with NRC and EPRI and NIST.
3 So, we have these two different 4 capabilities at Sandia, we're able to use these to at 5 least begin to benchmark these codes and look and see, 6 how effective are they?
7 So, if you go to the next slide, I have 8 some images and, ultimately, videos of different arcs, 9 if they'll come up. Unfortunately, the computer may 10 be slow.
11 For the first test we're doing, we're 12 looking at a very simple geometry. These are vertical 13 arcs. So, we have vertical arcs, again, of different 14 metals, aluminum or copper, or we can look at other 15 materials if they're of interest.
16 We're applying, in this case, from left to 17 right, 100 amps, 200 amps, sorry, 300 on the left, 18 200, 100 amp. And we'll make sustainable arcs that 19 last for 30 seconds or even up to two minutes with 20 these.
21 If you go to the next slide, let's see if 22 this plays. So, you should be able to play a movie at 23 the bottom there. Let's see.
24 MS. LAFLEUR: It says it's playing.
25 MR. CLEM: It says it's playing? Okay, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
200 1 it's thinking.
2 MS. LAFLEUR: There it is.
3 MR. CLEM: So, here, we show the onset of 4 the arc. So, we're triggering this arc with voltage.
5 We're then looking at a given arc length. We're 6 measuring the voltage and the current of these arcs 7 during these experiments, for, these are five-second 8 arcs, we've gone up to 30 seconds with these.
9 We're measuring many different factors 10 about these arcs experimentally. We're measuring the 11 arc temperature, using spectroscopy. We're measuring, 12 with black calorimeters, the radiant energy output.
13 We're measuring, with other calorimeters, inside a 14 closed chamber, the temperature rise in the chamber.
15 And we also can use, we're using infrared cameras, 16 calibrated infrared cameras, to calculate the 17 temperature of the electrode themselves, to look at 18 the conductive energy going into those busbars.
19 So, right now, we're looking at vertical 20 arcs, and we'll be looking soon at horizontal arcs, 21 and then, parallel arcs as well, at small-scale in 22 these experiments.
23 So, if you go to the next slide? What we 24 want to do with this is then begin to see how well we 25 agree or how poorly we agree with the outputs from our NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
201 1 modeling.
2 So, again, the curves on the left side 3 here, the black curve is the input electrical energy, 4 the dashed blue curve is what we predict for aluminum 5 electrodes, and the dark blue data points are what 6 we've measured so far, at 100 amps, 200 amps, and 300 7 amps.
8 And so, so far, the radiated energy 9 measured from these arcs is within 30 percent of the 10 predicted energy output from these arcs. So, we're 11 continuing this work -- if you hit the button one more 12 time?
13 Right now, we've been looking at low 14 current, but we'll be doing tests at higher current at 15 Sandia, up to a kiloamp, this summer at Sandia. And 16 then, at KEMA, an overlapping point at a kiloamp, and 17 then, another measurement I believe at 15 kiloamps 18 later this fall, to see how well we agree, as far as 19 these measurements.
20 So, a few of the takeaways so far is that, 21 initially, at least, it's within 10 to 30 percent, as 22 far as our predictions of radiant energy output, which 23 is really what would be propagated to other components 24 in the zone of influence, we believe. That's one 25 takeaway.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
202 1 Another is, we really believe we can take 2 out this open-circuit voltage that's used in a lot of 3 models and really substitute it by looking at this I 4 squared R of the arc times the time duration, delta T, 5 of the arc, to predict the energy output into a 6 system.
7 So, if we look at the wattage that's 8 measured on the right graph as a function of input 9 current, we see these scaled quadratically, with the 10 input current, so it goes as I squared, as we expect.
11 And again, we're able to measure a number 12 of different validation parameters, the arc voltage, 13 the arc resistance, the arc temperature, and again, 14 some of the arc dynamics of this, to start to validate 15 this model and show that it's accurate or show we have 16 at least confidence that it's predicting the right 17 things.
18 So, we have initial DC experiments, we'll 19 be moving in the future to AC experiments, again, on 20 an increasing current.
21 If you go to the next slide, we've begun 22 looking also at the effects of arcs on busbars. One 23 of the questions is, how arcs interact with busbars, 24 how material is evolved and liberated from busbars 25 during testing.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
203 1 So, we performed a number of tests last 2 year, at 480 volts, 4160, 6.9kV and 10kV. These are 3 very short arcs, around 100 milliseconds long, but we 4 wanted to look at things like mass loss of electrodes 5 for copper versus aluminum, and also volume loss of 6 copper versus aluminum, with these different arc 7 conditions and see if we could predict this with our 8 models, if it's consistent.
9 What we predicted, as shown in the upper-10 right, is that we'd expect a parabolic I squared 11 dependence of the mass loss of copper and aluminum 12 electrodes. And in fact, we'd actually predict the 13 same mass loss for copper and aluminum for these 14 electrodes from the simple model.
15 When we actually measure this, the data 16 points in the bottom-right, we do see this quadratic 17 dependence, both for aluminum and copper electrodes, 18 for these. So, we see an I squared R dependence of 19 mass loss from electrodes. And that indicates how 20 much vapor or how much molten material is being 21 generated by these arcs.
22 If you go to the next slide, we were a 23 little confused by why the masses were exactly the 24 same for copper and aluminum. I think we see, in the 25 field, that there's much more aluminum loss. It turns NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
204 1 out that you need to normalize the mass loss by the 2 density of aluminum versus copper.
3 Aluminum has roughly -- there's a few 4 different things that are different. One is the 5 aluminum melting point is roughly half that of copper.
6 So, 590 C versus 1090 C.
7 But, apparently, what's more important is 8 the density of aluminum is one-third that of copper.
9 And so, although your mass loss is the same, you're 10 losing triple the volume of aluminum compared to the 11 volume of copper.
12 And so, you're going to see much larger 13 generation of particulates, whether they're molten 14 droplets or vaporized aluminum that then reoxidizes, 15 there will be much more volume loss of aluminum 16 compared to copper.
17 And so, what we see here is that, again, 18 we predicted quadratic dependence of volume loss, in 19 the top curve, top-right, and the bottom-right is what 20 we actually measure, essentially the same dependence 21 of volume loss of aluminum with increasing current or, 22 in this case, voltage, because of the way our 23 experiment was run.
24 But we were able to predict the amount of 25 volume loss of material, the amount of material being NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
205 1 ejected into the zone of influence for these droplets.
2 So, another question, going beyond this, 3 is, what are those particles, can we begin to 4 characterize these? Can you press -- is there a movie 5 there? Or, it's not coming up yet?
6 Anyway, during these small-scale 7 experiments, we brought in high speed cameras that we 8 can bring in and look directly at the plasma 9 interface, with the arc roots, and look at the 10 generation of particles.
11 And there are clearly articles that are 12 tens of microns and even a nanoscale that are 13 generated during the arc itself. So, some particles 14 are melted metal that have been entrained in the edge 15 of the arc roots and ejected through an arc jet.
16 Other particles are melted and vaporized and then, 17 these subsequently reoxidize.
18 So, a couple questions we wanted to 19 answer, both at the small-scale and for field tests at 20 KEMA, are what kinds of particles are being evolved?
21 What are the size of these particles? What's the 22 degree of aluminum oxidation? And can we begin to 23 account for the exothermic nature of that oxidation, 24 the energy input of oxidation of aluminum?
25 We also want to see if there's a NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
206 1 difference of aluminum particle oxidation with 2 distance from the switchgear? Is it truly insulating 3 or is it conductive? And then, look at other sources 4 of energy, going into the arc fault.
5 So, if you go to the next slide, these are 6 some examples of particles that have been collected 7 during the KEMA experiments from last fall, but we see 8 similar particles in our small-scale tests, 9 essentially the same particle sizes.
10 We see two kinds of particles. One 11 particle size is on the order of two to 15 microns.
12 These are metallic particles of aluminum, that show 13 evidence that they have melted, and then resolidified 14 during flight, and then have been collected.
15 So, the middle of these particles are 16 dominantly aluminum metal, you can see the dendritic 17 structures showing where they resolidified, but on the 18 surface of these, you see sort of a white decoration.
19 This is a material that is insulating on the top 20 surface.
21 So, if you look at the top surface of 22 these particles, or if you look at just the background 23 of polymers or other things that are used to collect 24 particles in the field, we see that there's a dust, a 25 nanoscale dust everywhere in the test chamber.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
207 1 And that appears to be composed of very 2 small particles of aluminum oxide that are five to 30 3 nanometers in size. So, we believe that these have 4 been vaporized, so we made aluminum vapor. Aluminum 5 is not stable in oxygen. These, we believe, then, 6 have reoxidized and made nanoscale aluminum oxide 7 particles.
8 So, we see two kinds of particles. We 9 haven't yet quantified exactly what fraction is melted 10 and what fraction is oxidized, but we have the samples 11 in hand and can come to that conclusion and publicize 12 that, definitely by the end of the year, if not early 13 in the fall.
14 So, we can begin to quantify what kind of 15 oxidation energy input should be present and, 16 furthermore, what do these particles do? Are they 17 conductive or are they insulating?
18 We don't know the answer to that yet and 19 that's one of the goals for the test this fall, is to 20 collect the particles on a substrate, where we can 21 either use a surface probe or we can use energy tight 22 electrodes to look at, for these powder beds that are 23 collected, are they conductive or not between 24 electrodes? And if they are, if those collect on an 25 insulator, you may lose insulation resistance on that NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
208 1 insulator going forward.
2 If you go to the next slide, we've begun 3 to look at how oxidized these particles are. We see 4 a range of degrees of oxidation.
5 What we find, again, is that for these 6 nanoscale particles, that we believe have been 7 vaporized, if you look at the aluminum to oxygen peak 8 ratios that we collect by x-ray, these appear to be 9 essentially 100 percent oxidized. The aluminum to 10 oxygen ratios are three to two, so it looks like 11 Al2O3, aluminum oxide or sapphire.
12 If you look at the larger particles, these 13 particles have ranged from 25 percent to 73 percent 14 oxidized, in terms of the analysis we do. We still 15 have to run that more and understand more about that, 16 but we do believe that the core of the particle is 17 still metallic, but the outer skin appears to be 18 oxidized.
19 So, our goal is to be able to measure this 20 in the field, but also, ultimately, be able to predict 21 from these arcs what amount of metal is being melted, 22 what amount is being vaporized, and then, predict the 23 balance of energy that's coming out from radiation 24 effects, thermal effects, and then, this oxidation 25 effect as well.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
209 1 So, again, our first models have been 2 simple. They've been assuming a static arc, that's 3 not interacting heavily with the electrodes 4 themselves. So, the gap is not changing with time.
5 Ultimately, we know we're consuming material, that a 6 gap has to increase during the arc.
7 We also haven't included magnetic field 8 effects to cause this arc to move during, especially 9 at higher currents, a kiloamp and above, magnetic 10 field effects will dominate these arcs and we need to 11 include that.
12 So, we've conducted initial modeling runs, 13 using ARIA, which is our fluid code. Again, this is 14 based on this idealized Lowke model. The Lowke model 15 assumes, essentially, a cylindrical arc that has the 16 same temperature across the arc diameter.
17 In the ARIA simulation, it's able to solve 18 the heat transfer exactly, so you actually get the 19 real heat distribution within the arc and, ultimately, 20 we can begin to include the copper aluminum 21 electrodes, the rates or electrode melting, removal of 22 those, and growth of the arc, as well as magnetic 23 forces in these.
24 So, for our first simulations with ARIA, 25 which is the more complicated code, this is a massive NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
210 1 parallel processor code that can run on 100,000 2 processes, run much more complicated geometries than 3 the Lowke model, we looked at a simple cylindrical 4 arc, looked to see if it could predict accurately the 5 temperatures of the arc and the kind of radiation we 6 should expect from that.
7 So, next slide shows some of the equations 8 of the ARIA model. I've got a lot more detailed 9 description of it, if it's of interest. But I didn't 10 really want to go into it, but just wanted to show 11 what we include.
12 We include tracking of all the chemical 13 constituents. We can include concentrations, 14 pressure, buoyancy, magnetic forces, joule heating, 15 and diffusion in these, and then, solve for the 16 temperature of the arc itself. And then, that can 17 then be used to generate the radiative heat transfer 18 and the thermal heat transfer out of the arc itself.
19 If you go to the next slide, and press 20 play, if it'll come up? What this is looking at is 21 the temperature of the arc, as a function of time.
22 So, this is from the center of the radius of the arc, 23 going out to a one centimeter boundary. And on the Y 24 axis, the peak temperature is 7000 Kelvin and the 25 outer temperature would be room temperature. So, this NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
211 1 is predicting the temperature profile of the arc, as 2 a function of time.
3 So, essentially, this would predict a 4 cylinder that has a 7000, roughly, Kelvin temperature 5 in the middle of the arc, decaying as a function of 6 radius to the outside of the arc, to the outer 7 boundary.
8 So, if you plot this another way, if you 9 go to the next slide, this gives you roughly a 10 parabolic temperature profile, where the center of the 11 arc should be around 6500 Kelvin, and the outer part 12 of the arc, again, should be room temperature.
13 There's a discontinuity at around the 14 outside of the arc, around 4000 Kelvin. So, actually, 15 there's the center of the arc, which is a very high 16 temperature, and there's an outer boundary of the arc, 17 which is around 4000 Kelvin, where the air interfaces.
18 So, we compare this to work in the 19 literature, we see that this agrees quite closely, the 20 same kinds of temperature profiles, the same 21 temperatures to within 300 degrees of what's been 22 predicted in the literature and also measured in the 23 literature for arcs.
24 If you go to the next slide, looking back 25 at some of the Lowke data, we see some of the same NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
212 1 kinds of data, same kinds of behavior, and again, 2 temperatures of around 6000 Kelvin for the center of 3 the arc temperature for this.
4 So, we're beginning to get validation data 5 compared to the literature. We also can go in 6 directly to our experiments and measure the 7 temperature of the arc itself using spectroscopy.
8 So, let's see, Caroline, are you online 9 right now?
10 MS. WINTERS: I am, can you hear me?
11 MR. CLEM: We can hear you. So, do you 12 want to talk through your slides here?
13 MS. WINTERS: Okay. Sure. So, to get a 14 better idea of the temperature in the arc and to be 15 able to compare the ARIA model results, not just to 16 prior simulations, but also to the small-scale 17 experiments we're seeing, we did a series of 18 spectroscopy experiments. So, if you could move to 19 the next slide?
20 MR. TAYLOR: We're on Slide 29.
21 MS. WINTERS: Okay. So, this is the 22 standard experimental schematic, in the top-left.
23 This is a top-down view, showing that the spectrometer 24 that was collecting this information was placed eight 25 inches away from the arc.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
213 1 Using a laser diode and a series of 2 calming instances, we were able to take light from the 3 center of the arc whenever we initiated it and get 4 measurements at a rate of 100 hertz. So, that's 5 taking a spectral profile every ten milliseconds for 6 the entire test.
7 And you can see on the right-hand side two 8 examples of the spectra that I'm talking about. So, 9 I call these optical emission spectroscopy. Using one 10 set line that's collected, it goes through grating or 11 impinges upon it, and that grating will spectrally 12 disperse it of the amassed photons with a given 13 wavelength. Or you can think about that wavelength as 14 equating to energy.
15 So, if you're at 300 nanometers, those 16 photons are going to be a higher energy than if you're 17 at 800 nanometers. And the benefit is that species 18 like copper or nitrogen, have very specific emission 19 wavelengths. So, you can see here, we've identified 20 them as being atomic copper, we also have atomic 21 nitrogen.
22 And then, you've got our spectra on the 23 bottom, that was taken with similar test parameters, 24 but without the metallic electrodes. One of the 25 benefits of copper is that we lose very little mass NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
214 1 during our run.
2 So, there, you can really see those strong 3 nitrogen, molecular nitrogen, as well as molecular 4 carbon bands, and see the difference between the 5 molecular features, which are a little bit more broad 6 and those sharp atomic lines.
7 So, what does this information tell us?
8 If you go to the next slide, I will give some 9 explanation.
10 If we look at how those strong atomic 11 features, especially right now for copper, if we look 12 and ratio the intensity, so how many photons show up 13 at 800 nanometers versus how many photons show up at, 14 say, 600 nanometers, we can get a relationship based 15 --
16 MR. CLEM: Caroline, we're losing you a 17 little bit on your phone.
18 MS. WINTERS: Okay. Is this any better?
19 MR. CLEM: Yes, it's better.
20 MS. WINTERS: Okay. So, by looking at the 21 amount of photons of a given energy, for, say, all of 22 the copper, atomic copper lines that we can identify, 23 we're able to generate measurements of temperature.
24 And we know that the copper vapor only 25 exists in the arc, because as it would transport NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
215 1 outside of the arc, it would rapidly cool down and it 2 wouldn't emit any more. So, that gives us our 3 measurement of our arc temperature, and that's what 4 you're seeing on the left-hand figure on Slide 30.
5 So, that's a time-resolved measurement, 6 throughout an entire four-second test, showing that 7 the arc temperature does vary a little bit, but not 8 much, it sticks around 6000 to 7000 Kelvin, with a 9 positive residual, so that you understand the error 10 associated with those temperature measurements.
11 Additionally, for certain spectra, we were 12 able to get a secondary measurement of the surrounding 13 air temperature. And so, that's shown in the right-14 hand figure. And that's shown for two different arc 15 currents.
16 So, so far, we've been limited in our 17 currents, although we're hoping, as Paul said, to go 18 up to a kiloamp. And so, in doing so, you can see 19 here, this is the difference between 100 and 150 amps, 20 but soon, we hope to have measurements at greater than 21 that, up to a kiloamp for comparison.
22 And because it's only a small difference 23 in current right now, we see only small -- we see very 24 little difference in temperature, except that the 25 surrounding air appears to be at a lower temperature, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
216 1 around 4000 Kelvin, and the center of the arc is 2 holding at 6-7000 Kelvin, which is in agreement both 3 with the ARIA model that we've run at Sandia, as well 4 as the prior literature, which predicts that it should 5 be between 6200 and 7200 Kelvin.
6 We'll be continuing these experiments, not 7 just for copper electrodes, but also for the aluminum 8 electrodes, and we're working on getting measurements 9 of the aluminum arc temperature, as well as the copper 10 temperature.
11 MR. CLEM: All right. So, these 12 measurements, so far, have been done on small-scale 13 experiments at Sandia and now that we've proven that 14 they work and they're valuable, we'll be using them in 15 onsite tests at KEMA this fall also, at kiloamp and 15 16 kiloamps, and I believe, on some of the other higher 17 current tests as well, maybe 25 kiloamps as well.
18 So, we think we can, again, in addition to 19 the calorimetry data and the electrical data for the 20 input, we can start to validate these models and get 21 more data about the actual arc temperatures in actual 22 HEAF events, in open box studies and potentially even 23 in enclosures, moving forward.
24 MR. MELLY: Yes, and we when say we're 25 going to be doing this testing up at KEMA, in this NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
217 1 upcoming round of testing, this is testing that will 2 be done in addition to the tests that were identified 3 on the test matrix.
4 These tests will be performed in an open 5 box, that we can actually visualize the arc itself.
6 These will not be the tests that we identified 7 previously on the matrix, they're additional open box 8 tests that are going to be solely used for validation 9 purposes.
10 MR. RANDELOVIC: Quick question.
11 MR. CLEM: Sure.
12 MR. RANDELOVIC: So, it's going to take 13 some time to validate the model. The small-scale --
14 MS. LAFLEUR: And it's iterative.
15 MR. RANDELOVIC: -- testing -- right.
16 MS. LAFLEUR: Yes.
17 MR. RANDELOVIC: The small-scale test, I 18 mean, all that work that you have done looks good, it 19 took some times. You're going to be performing 20 testing at KEMA, open box testing. How long do you 21 expect that effort to last? Are we going to have this 22 model ready before we start testing the actual 23 switchgear?
24 MS. LAFLEUR: In August, no.
25 MR. RANDELOVIC: Okay.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
218 1 MS. LAFLEUR: No.
2 MR. MELLY: It's going to be performed 3 concurrently.
4 MS. LAFLEUR: But we do have plans to 5 predict those tests. We'll just predict them after 6 the tests.
7 MR. RANDELOVIC: Okay.
8 MS. LAFLEUR: Yes.
9 MR. CLEM: I think we have a project plan 10 that includes, yes -- basically, it's going to be 11 predicting, if you give us a current, you tell us 12 based on a busbar, we predict blindly what should the 13 energy output be from that? What should the radiative 14 energy output be?
15 And we can predict it for future tests, we 16 can predict it for past tests where we haven't seen 17 the data. So, but that's the goal, is to really put 18 the model to the -- feet to the fire and see if it's 19 accurate.
20 But first, we're just building up the 21 belief that we can accurately predict the evolved 22 energy with copper busbars, aluminum busbars, on 23 increase in current and capture the atmospheric 24 chemistry correctly that gives rise to the resistance.
25 MR. RANDELOVIC: Okay. So, you'll be NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
219 1 working trying to set up the models, ARIA and FUEGO, 2 calculating. Have we -- do we have any experience 3 from the past, where these codes were coupled and we 4 had --
5 MS. LAFLEUR: Yes, and we can --
6 MR. RANDELOVIC: -- successful application?
7 MS. LAFLEUR: We can skip to Matt's slides 8 --
9 MR. RANDELOVIC: Okay.
10 MS. LAFLEUR: -- do you want -- can we skip 11 to, I think it's 36? Matt, are you able to speak?
12 MR. HOPKINS: I think so.
13 MS. LAFLEUR: Okay, we've got you. We're 14 just advancing our slides here, to the slide that 15 starts with Sierra Mechanics.
16 MR. HOPKINS: Okay. I'm only looking at 17 the slides through the online thing.
18 MR. CLEM: Okay.
19 MS. LAFLEUR: Okay.
20 MR. HOPKINS: That seems to be all right.
21 So, at Sandia, we have this Sierra Mechanics framework 22 and it's been under development, being used for about 23 20 years. Two of the elements in Sierra Mechanics are 24 this ARIA code and this FUEGO code.
25 ARIA is a multi-physics transport code, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
220 1 with free moving boundaries and various levels of 2 coupled solution techniques. So, we use it, 3 generally, for problems that have pretty tightly 4 coupled physics, complicated physics, and multi-phase 5 processes.
6 So, for example, solid electrode, molten 7 electrode, vaporized metal gas, and other gas where 8 you have temperature dependent physical parameters, 9 like conductivity, thermoconductivity and electrical 10 conductivity.
11 And FUEGO is a code we've used for 12 extensive fire modeling. So, if we want to determine 13 the effect of a box with material in it, while it sits 14 in a fire, FUEGO can simulate sort of the large-scale 15 convective transport, along with regular CFD as well.
16 And then, both of those codes have a sense 17 of radiation transport. So, we can track the 18 transport of all of thermal and radiative ablation of 19 material, entrainment of that material to change the 20 state of the gas, all coupled together.
21 And as, I think it was Gabe earlier 22 mentioned, or spoke in this direction, there are many 23 rabbit holes we could go into to get more even 24 detailed, along all of these axes. So, I think part 25 of what we're going to be developing is prioritizing NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
221 1 which of these particular phenomena are critical to 2 get correct and try to simplify where we don't need 3 it.
4 So, ARIA and FUEGO are the key ones and if 5 we go to the next slide, yes, I'd like to point out 6 that there's some question about where these codes 7 have been used in the past and what kind of coupled 8 problems they've been used on.
9 They have been coupled together, I don't 10 think any of that material is presented here. One of 11 our constraints here at Sandia is a lot of the 12 extensive validation work we have has not been made 13 available in the open literature. Depending on the 14 interest, we could try to find some of that material 15 and see if it can be released, but we haven't done 16 that extensively yet.
17 ARIA has been used -- so, I'm going to 18 explain some problems that I think are close to the 19 kinds of things we're talking about for HEAF modeling.
20 ARIA has been used for problems such as 21 laser welding. So, here, we have an intensive heat 22 source, if you will, applied to a metal and we do the 23 transient evolution of that metal, in terms of its 24 temperature, its phase change.
25 Similar questions related to punch-through NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
222 1 of the material. So, when does the radiation, thermal 2 environment, over some period of time, get to the 3 point that it actually melts and punches through a 4 material?
5 Both of these codes are massively 6 parallel, so they scale to the thousands and more than 7 thousands of cores. I think I've mentioned some of 8 the rest, and I don't know if anyone in this audience 9 wants to hear more details, I can provide it now or in 10 the future, but maybe skip some of the minutia here 11 and go to the next slide.
12 So, on the left, you're seeing, I think 13 that's an animation, that is a, I believe a metal box 14 containing an explosive material. And you see the 15 temperature release, as this front extends and you can 16 see the box deforming. So, this is just providing a 17 demonstration of the temperature dependence and the 18 free surface moving boundary portions of ARIA. Next 19 slide.
20 Here's another key part of the physics 21 that I think would be new, in terms of the types of 22 approaches that have been used in the past, radiation 23 transport.
24 It's not just determining the radiation 25 incident on a material, say for use in a fragility NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
223 1 model, it's also that the radiation can reradiate in 2 the gas environment that it's in, especially if 3 there's soot present.
4 So, computing the power in radiation from 5 an idealized arc still isn't the whole story, that 6 radiation, some of it anyway, is absorbed by the gas, 7 smoke, and soot environments and then, reradiates.
8 So, being able to track that may be important, so we 9 have those capabilities. Next slide.
10 MS. LAFLEUR: Yes, on this slide --
11 MR. HOPKINS: Yes.
12 MS. LAFLEUR: -- Calore was the previous 13 name of ARIA. Its name has evolved over time. Sorry, 14 Matt --
15 MR. HOPKINS: Sort of.
16 MS. LAFLEUR: Okay.
17 MR. HOPKINS: So, here, I think Paul 18 presented a slide that introduced this already, but 19 basically, the transport equations we include are 20 conservation of mass, conservation of momentum, 21 conservation of energy.
22 And we can do multi-species models, so we 23 can have, for example, air, air ions, electrons, and 24 aluminum as different species and they can each have 25 their own uplink to the rest of the system. Next NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
224 1 slide, please. That's somebody else now, I think.
2 Oh, no, that's still me.
3 So, one of the problems that was solved in 4 the past was a safety question in a battery storage 5 facility. So, if a fire broke out in this battery 6 storage facility, you'd like to understand the damage 7 to surrounding areas. So, that sounds awfully 8 familiar to what we're talking about here.
9 If you go to the next slide, so on the 10 left, before you hit go, on the left are three racks 11 of cells. And we've put a fire between the first and 12 second rack, so that fire is modeled as a source of 13 both radiative and thermal conductive energy.
14 And you can see, there's buoyancy effects, 15 because we have hot gas containing soot that have come 16 to the ceiling of this material. And you can see, on 17 the top-right, the internal flow patterns generated by 18 the non-constant temperature around, along the edges 19 of the cabinet.
20 And on the bottom-right is something that 21 I think is right in the middle of this question of how 22 do you take an energy source model and then, use that 23 to inform a fragility model?
24 On the bottom-right is actually the heat 25 flux experienced on one of these cabinets, due to this NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
225 1 fire. And the next slide, please. I think this is an 2 animation of the results.
3 And so, on the bottom-right, I don't 4 believe this particular model has an evolution of that 5 material, other than its temperature, but using ARIA 6 to model also the, quote, target, you can talk about 7 the phase change and eventually failure or blow-8 through of that material, in addition to using that 9 same set of capabilities to simulate the electrode 10 evolution, in terms of its phase change and loss 11 ablation. Next slide, please.
12 So, this is the same system, but now we've 13 changed the environment, and we have a cross wind 14 that's going at ten meters per second. So, this is 15 more to open up the set of questions you could ask if 16 you wanted to change environmental conditions. I 17 don't know that winds blowing nuclear reactors is a 18 concern. Next slide, please.
19 MS. LAFLEUR: I think that's his last 20 slide.
21 MR. HOPKINS: Okay.
22 MS. LAFLEUR: Yes, I think there's just a 23 question prompt at the end of yours, Matt.
24 MR. HOPKINS: Are there any questions?
25 MR. HAMBURGER: So, if anyone has questions NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
226 1 for Matt, they can ask them now. We do have a 2 question on the webinar about the spectroscopic 3 analysis methods.
4 MS. LAFLEUR: Okay. We'll take whichever 5 questions.
6 MR. HAMBURGER: Okay. I will read the 7 questions from Steve Turner, who is on the webinar.
8 And his question is this, for spectroscopy 9 measurements, can you see through the smoke in full-10 size tests?
11 In attempts to measure arc properties that 12 depends on sight, the smoke intervenes quickly, so for 13 spectroscopy methods and radiation heat measurements, 14 can you see through the smoke in long arc duration 15 full-scale tests?
16 In attempts to measure arc properties --
17 it looks like it got repeated. And he also asks, are 18 you validating to three-phase arcs?
19 MS. WINTERS: So, can you guys hear me?
20 MR. HAMBURGER: Yes, we can hear you.
21 MS. WINTERS: Okay. So, to his question, 22 the first is, we have done testing with the copper and 23 aluminum, and there are two specific sets of spectral 24 features. The first are sharp features, which is the 25 copper, atomic copper vapor lines that I pointed out.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
227 1 The second is a broad band emission 2 plateau, that I really didn't talk about, but that is 3 the smoke that he's describing. That is the mist from 4 that surrounding gas. And by fitting it to a gray 5 body approximation, that's how I was generating those 6 secondary temperature measurements that start around 7 4000 Kelvin.
8 To his point about the large-scale 9 testing, we have done and completed VC testing and 10 spectroscopic measurements in the field and we 11 certainly see the buildup of the smoke. And that 12 smoke, as the duration goes on, for about two to four 13 seconds, you see that becoming a larger component of 14 the spectral radiation that we get.
15 Now, those sharp spectral features will 16 still exist on top of it, and that's why it's 17 important to understand your baseline, when you're 18 trying to fit those, say copper vapor atomic lines for 19 temperature. And we do that through a series of 20 calibration with a Tungsten lamp, which acts as a 21 perfect repository.
22 An interesting thing that we didn't have 23 time to show on these slides, though, is that we also 24 completed a series of measurements with Tungsten. And 25 when we did that, we had a heavy amount of smoke, it NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
228 1 filled the box. And in that instance, it did 2 overwhelm spectral features signature and we were only 3 able to collect that gray body emission.
4 So, it's certainly something that we see 5 in the spectra and we can map out when it overwhelms 6 any sort of arc features you would expect.
7 MR. CLEM: So, I think, we're conducting 8 open box experiments, as Nick said, to look at the arc 9 temperature in three-phase arcs, in some KEMA tests.
10 In the full -- if you're inside a bus duct 11 or a switchgear, we would like -- it would be very 12 difficult to image the arc directly, for the full 13 duration of the arc fault, due to the smoke evolution.
14 But, yes, I think, like Caroline said, we can collect 15 spectra and then, see how much of that is the smoke, 16 how much of that is the arc itself.
17 We do have calibrated infrared cameras, 18 that were used in previous experiments at KEMA and 19 we'll use in the future. And that can image, of 20 course, the cloud, the sooty cloud that's emitted once 21 the breach occurs. But we may not be able to see 22 directly into the arc for the full duration of the 23 HEAF events. But it's a good question.
24 Second question was about, are we looking 25 at three-phase arcs? The experiments so far have been NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
229 1 DC arcs. We'll next be going to single-phase arcs, 2 and then, three-phase arcs. So, we'll be modeling 3 three-phase arcs, ultimately. There's been quite a 4 bit of work looking at how to model a three-phase arc, 5 an AC arc versus DC.
6 So, if you have a 60 hertz signal, you 7 have eight milliseconds of each polarity. For about 8 the first millisecond, the arc may not be at 9 equilibrium, but from the second millisecond to the 10 eighth millisecond, there's a stable arc resistance 11 and you can treat the arc as a circuit element for 12 those periods.
13 In a single-phase arc, people brought up, 14 several times, that at the zero-crossings, the arc may 15 extinguish and then, relight over and over again. So, 16 the single-phase arcs we actually believe are going to 17 be more complicated than the three-phase arcs to model 18 going forward. But we anticipate using ARIA to model 19 the three-phase arcs, ultimately.
20 MR. MELLY: Yes, and as Paul said, some of 21 the challenges with this are, the reason that we can't 22 take some of these measurements in the full-scale 23 tests that we'll be performing on the switchgear that 24 represent in-plant equipment is because we do not want 25 to alter the in-plant equipment that we'll be testing.
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
230 1 We do not want to open a back panel or 2 create a viewing port, because that can alter the test 3 itself. So, we're not doing that in part of the 4 large-scale matrix. We've added these additional 5 tests where we can have an open box arrangement, which 6 will feed back to the validation, without affecting 7 the larger test matrix tests.
8 MR. HAMBURGER: Thank you, Steve, for the 9 question, and if you have any followups, you can 10 either type them or we can unmute you. Do we have any 11 other questions for the Sandia folks about their 12 modeling work? Okay. Anything else on the webinar?
13 No?
14 Okay. It's almost 3:55, let's take 15 another 15-minute break, come back at ten after. And 16 then, we have a few more slides on the fragility 17 testing component, we'll have our public comment 18 period, any closing remarks, and then, we can adjourn 19 for the day. So, we'll take 15 minutes and come back 20 at ten past 4:00.
21 (Whereupon, the above-entitled matter went 22 off the record at 3:53 p.m. and resumed at 4:08 p.m.)
23 MR. HAMBURGER: Okay, just a few more 24 minutes of presentation about the target fragility and 25 failure criteria progress. And then, we'll open it up NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
231 1 for public comment.
2 MS. LAFLEUR: Okay. Thank you. This is 3 Chris LaFleur again.
4 So, one thing I wanted to stress is that 5 we are definitely treating the failure criteria 6 independent of the energy release prediction 7 calculation and measurement from the HEAF. That's 8 this incident energy or the emitted energy.
9 The diagram on the left, in the blue line 10 we're seeing like a typical hydrocarbon fire with a 11 typical growth curve that we model as part of the 12 traditional fire PRA. And the orange line there is 13 trying to compare the temperature or the heat release 14 rate profile of a HEAF event, and it's a much higher 15 magnitude, but a much shorter duration. And the 16 duration really matters as the target or the impact on 17 the target, because it has less time to absorb that 18 energy and be damaged.
19 And so, what we want to do is determine 20 the failure criteria's characteristics of specific 21 pieces of target equipment: cables, transformers, 22 switchgear, other cabinets, whatever. And the Work 23 Group will be determining what those target equipment 24 are that we need to evaluate or develop the failure 25 criteria for, as a Work Group. And those will take in NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
232 1 both the severity of the adverse environment as well 2 as the time that it's exposed to that. And that's the 3 vision.
4 The next slide is just a review of what 5 the current failure criteria is that we deal with in 6 the fire PRA all the time. And it's around 7 thermoplastic and thermoset cables, and there's both 8 the radiant heat criteria and a temperature criteria.
9 There is also some empirical data. It 10 doesn't go down to anything less than one minute. So, 11 obviously, in a HEAF event we're talking, the most we 12 were talking about is 8 seconds. So, there's clearly 13 a need to define that failure curve down below one 14 minute and see what those other temperatures or heat 15 fluxes would be when we get down to the seconds.
16 If a HEAF event causes an ensuing fire, 17 there would be some split fraction about that 18 generating of fire, and then, that would be modeled as 19 a regular hydrocarbon fire with a typical growth rate.
20 This is just for the actual heat and the arc effluent 21 event.
22 So, we are going through right now looking 23 at data from the current tests where we've got 24 temperature measurements and incident/injury 25 measurements at the 3-foot and the 6-foot range from NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
233 1 the OECD tests and the phase 1 testing. And we're 2 going to develop some needs for testing to happen in 3 four different types of cables, and then, we're also 4 going to, within the Work Group, determine what type 5 of failure criteria would be around enclosed 6 equipment.
7 Like would we base it on the breach 8 prediction for the target equipment's enclosure? And 9 we may say, if your target is breached, you have to 10 fail that. There will be something in the new PRA 11 method that will tell us how to treat those, and we'll 12 use the model to define the extend of that envelope of 13 whatever that criteria is, temperature "X" or heat 14 flux "Y", wherever that is. And then, you should be 15 able to have a site-specific evaluation of predicting 16 target failure around the HEAF event.
17 That's as far as we've gotten right now, 18 where it's an active, current, open action item with 19 the Working Group.
20 And I think that's the last slide I had.
21 MR. HAMBURGER: Okay. Any questions about 22 the approach we're taking towards fragility testing?
23 From anyone in the room?
24 (No response.)
25 The webinar?
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
234 1 (No response.)
2 Okay. Well, we have a public comment 3 period now. So, if anything does occur to you, you 4 can ask.
5 So, any other comments? About anything 6 we've discussed today for any of the presenters here 7 from the NRC, Sandia?
8 MR. AIRD: You have a comment.
9 MR. HAMBURGER: All right. So, we've got 10 another question from Steve Turner, and this is about 11 the IEEE 1584. It says, "You cover the IEEE 1584 12 empirical existing models and the more sophisticated 13 Sandia models. Did you look at the numerous CFD 14 models using accepted and verified codes different 15 than the A3.24 model, like ANSYS or ACE, developed 16 over the last 15 years by manufacturers and Kreppy 17 (phonetic)? That might be a good, middle-of-the-road 18 modeling approach. These should be considered. One 19 reason the most complete approaches have not been 20 done, as mentioned in slide 6, is that the CFD 21 approaches using an energy point source are deemed 22 adequate for manufacturers and safety evaluations."
23 I guess the question there is, have we 24 looked at what he's calling "middle-of-the-road" 25 models, like ANSYS and ACE?
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
235 1 MR. TAYLOR: So, no, we haven't looked at 2 ANSYS or ACE or CFD modeling.
3 MR. HAMBURGER: Okay.
4 MS. LAFLEUR: Are you familiar with it?
5 MR. CLEM: I guess I'm familiar with both 6 of those. I think those are both good codes as far as 7 fluid dynamics and modeling thermal heat transport, 8 possibly even fire analysis. I'm not aware that 9 either of those contains the physics right now to 10 address an AC arc fault and the energy released from 11 that as a function of time. If it does, it's worth 12 definitely looking into. I just haven't seen it.
13 There's just a lot of -- we're colleagues 14 with the CFD group, CFDRC, that developed that code, 15 and they do have plasma experts that work in similar 16 areas. But, yes, I haven't seen exactly them address 17 this problem per se.
18 MR. HAMBURGER: Okay. But he just said 19 there's a large body of work that applies to the ACR.
20 Okay. Okay, that's something we can look into.
21 Any other questions or comments?
22 (No response.)
23 Okay. And just a note. When I do receive 24 the transcript, I'm going to, as I've done in the 25 past, package it up with all of the updated meeting NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
236 1 materials and make that publicly available through our 2 ADAMS system. So, that will all be publicly 3 available, I'm hoping, within the next week or two.
4 Yes, so, Mark, do you have any closing 5 comments?
6 MR. THAGGARD: Yes. I would just like to 7 say, first of all, I want to thank those that managed 8 to hang around to the end here. I know this has been 9 a long day, but I think it's been fairly productive.
10 A couple of things I think came out of 11 this morning's meeting. I think there was an 12 acknowledgment that the Working Group has made a lot 13 of progress. They seem to be working well together.
14 Also, one of the things I took away from 15 this morning's part of the meeting is an 16 acknowledgment that I think most people will agree 17 that trying to develop, moving away from a one-fit-all 18 model is the right way to go. So, I think any effort 19 we can do in that area, I think there seemed to be 20 agreement with that.
21 I recognize, based on a lot of the 22 comments that we got this morning, that there is some 23 concern about, citizen concerns about us moving 24 forward with the testing that we're trying to do later 25 this summer. I think some of the discussion we got NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
237 1 this afternoon I think would communicate to me one of 2 the reasons that we need to move forward with that 3 testing is because we need that information. I mean, 4 part of this research effort is you've got to get 5 information to build -- you know, some information 6 builds on other information. So, we need to collect 7 the information.
8 This meeting, the discussion this 9 afternoon has focused on the mod/sim primarily. And 10 one of the reasons that we focused on that was because 11 we had gotten a lot of questions about that, I think, 12 at the prior public meeting. So, we wanted to focus 13 on that this afternoon.
14 I will say that one of the things -- I 15 think Kelli raised it this morning -- the recognition 16 of the importance of getting the frequency 17 information, I think we agree with that. The 18 discussion this afternoon focused on the mod/sim and 19 a lot of it on the consequences piece of it, but we 20 recognize, also, you've got to have the frequency.
21 And I think we see that the Working Group has actually 22 made a lot of progress in that area. So, I think that 23 is something positive we can take away from this.
24 Again, I would like to just say I 25 appreciate everybody taking the time to spend the day NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
238 1 with us and gave us some good feedback. And I 2 appreciate the people from Sandia coming here, taking 3 up the time, and putting a lot of effort into 4 preparing for this meeting.
5 With that, that is all I have. So, I 6 think we're done.
7 MR. HAMBURGER: All right. Thank you all 8 very much. Thank you all very much. Thank you to 9 those who participated by webinar as well.
10 And we'll publish the materials as soon as 11 we can.
12 And we'll have NRC staff escort you down.
13 Thanks.
14 (Whereupon, the above-entitled matter went 15 off the record at 4:19 p.m.)
16 17 18 19 20 21 22 23 24 25 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433