WEBVTT 00:00.000 --> 00:06.000 All right, thank you. 00:06.000 --> 00:10.000 All right, so I'm Mark Barnes from Los Alamos National Laboratory. 00:10.000 --> 00:14.000 I guess first off, I apologize as I'm a bit of a cold. 00:14.000 --> 00:17.000 So I'm going to attempt speaking for the people in the back, 00:17.000 --> 00:20.000 but I probably won't succeed. 00:20.000 --> 00:23.000 Anyways, I'm very glad to be here. 00:23.000 --> 00:26.000 Obviously things are in flux right now, 00:26.000 --> 00:29.000 but I was able to get my travel approved. 00:29.000 --> 00:32.000 Yeah, I'm so I'm very excited to be here. 00:32.000 --> 00:37.000 So Los Alamos is a reasonably long history with open source. 00:37.000 --> 00:40.000 We have a lot of university academic collaborations, 00:40.000 --> 00:45.000 and so releasing a research software under a premise of license. 00:45.000 --> 00:51.000 We found it to be very helpful to foster collaborations and bring in students. 00:51.000 --> 00:57.000 So it's been a great value to us and a great value to the academic community as well. 00:57.000 --> 00:59.000 So I'll be talking about here. 00:59.000 --> 01:03.000 I know there's been a good bit of energy analysis software. 01:03.000 --> 01:08.000 This is a little bit more specialized than some of the other packages we've seen. 01:08.000 --> 01:15.000 So this is a add-on package that's specifically focusing on the analysis and mitigation 01:15.000 --> 01:20.000 of geomagnetic events. 01:20.000 --> 01:23.000 So why do we care about them? 01:24.000 --> 01:30.000 As we've probably noticed in kind of more casual reporting, 01:30.000 --> 01:35.000 there's a great variation in terms of predicted severity of events 01:35.000 --> 01:38.000 and it kind of varies so I'm okay not much will happen 01:38.000 --> 01:42.000 to one hemisphere of the world is going to go back 01:42.000 --> 01:45.000 into the pre-industrial age. 01:45.000 --> 01:50.000 So how do we predict this? 01:51.000 --> 01:56.000 It will an electrified transportation network be completely knocked out. 01:56.000 --> 01:59.000 Will your phone explode in your face? 01:59.000 --> 02:05.000 I'll attempt to answer some of this so let's dive in. 02:05.000 --> 02:12.000 Okay so I apologize that half of this presentation is background, 02:12.000 --> 02:15.000 but it was a current in level events. 02:15.000 --> 02:22.000 This was a space weather event named after the British scientist Richard Carrington 02:22.000 --> 02:28.000 back in 1859 and this was the largest observed geomagnetic event 02:28.000 --> 02:34.000 which fortunately occurred before there were continental scale interconnected power systems. 02:34.000 --> 02:41.000 We just had telegraph lines which was enough to observe the effects 02:41.000 --> 02:47.000 but sufficiently simple to not have a major impact. 02:47.000 --> 02:49.000 Anyway so how do these work? 02:49.000 --> 02:53.000 So what we have is charged particles coming off of the sun. 02:53.000 --> 02:59.000 They get trapped in the earth's ionosphere or magnetosphere. 02:59.000 --> 03:05.000 And they form a not current flowing above the surface of the earth through the electric jet. 03:05.000 --> 03:09.000 So this induces a magnetic field 03:09.000 --> 03:12.000 and an electric field on the surface of the earth. 03:12.000 --> 03:17.000 And this will couple into electric transmission lines. 03:17.000 --> 03:20.000 Now what do these fields look like? 03:20.000 --> 03:26.000 So here's a measured waveform of round electric field 03:26.000 --> 03:34.000 from the Quebec March 1989 event which was one of the more severe events of the last century. 03:34.000 --> 03:42.000 And so what we notice is the time scales on the earth about one second to one day in terms of variation. 03:42.000 --> 03:47.000 And the amplitude if you're looking carefully at the y-axis. 03:47.000 --> 03:50.000 So this is on the order of volts per kilometer. 03:50.000 --> 03:52.000 So this is one thing that's good, right? 03:52.000 --> 03:57.000 So even for your largest fablet, a field on the order of volts per kilometer 03:57.000 --> 04:01.000 is not going to make your sound sound galaxy explode in your face. 04:01.000 --> 04:06.000 But it can still have the fact suddenly electric power grid. 04:06.000 --> 04:09.000 So what are those? 04:09.000 --> 04:12.000 So over here we have a large power transformer. 04:12.000 --> 04:17.000 These are very big, very heavy and very expensive. 04:17.000 --> 04:22.000 And for this reason, we operate them right on the edge of saturation. 04:22.000 --> 04:29.000 So that means that even for a not enormous additional current flowing through the transformer, 04:29.000 --> 04:34.000 we can put it into half cycle saturation like we see on the figure here. 04:34.000 --> 04:37.000 So what's this mean? 04:37.000 --> 04:43.000 Well, this means that our transformer's exciting current has these big spikes in it. 04:43.000 --> 04:46.000 And these will introduce harmonics in the power system. 04:46.000 --> 04:48.000 Those can cause trouble. 04:48.000 --> 04:56.000 But the most significant impact is that there's a fundamental component of that current that's out of phase with the voltage. 04:56.000 --> 05:03.000 And this increases your so-called reactive power losses. 05:03.000 --> 05:07.000 And these can impact voltage stability on the grid. 05:07.000 --> 05:16.000 Also we can experience transformer overheating as a result of local hotspots. 05:16.000 --> 05:21.000 Now let's go into how does computing impact power systems? 05:21.000 --> 05:24.000 So we've got a number of different analysis types. 05:24.000 --> 05:28.000 Transient solvers, steady state solvers, dynamic solvers. 05:28.000 --> 05:34.000 But for a mitigation problem where we care about solving this in terms of optimization, 05:34.000 --> 05:39.000 typically we're limited to a steady state solver. 05:39.000 --> 05:41.000 So modeling GIC is in this context. 05:41.000 --> 05:44.000 It's a little bit tricky. 05:44.000 --> 05:49.000 This motivates our workflow that we hear here, 05:49.000 --> 05:55.000 that's suggested by the north national electric reliability corporation. 05:55.000 --> 05:58.000 So we have a number of different steps. 05:58.000 --> 06:01.000 We apply earth connectivity model, put into a coupling model. 06:01.000 --> 06:05.000 Apply DC solve, apply our transformer saturation model. 06:05.000 --> 06:10.000 Apply our AC solve and then perform mitigation measures. 06:10.000 --> 06:13.000 Now what does this look like in terms of circuits? 06:13.000 --> 06:17.000 So in love here we have a positive sequence model, 06:17.000 --> 06:22.000 where we represent a transmission system that's approximately balanced. 06:22.000 --> 06:28.000 As one line, this doesn't work for quasi-DC current. 06:28.000 --> 06:31.000 So what we have is the model on the left here, 06:31.000 --> 06:35.000 which is approximately equal to the real portion of a zero sequence, 06:35.000 --> 06:41.000 not work representation that's commonly used in full studies. 06:42.000 --> 06:46.000 Now in terms of what we do about GICs, 06:46.000 --> 06:51.000 so a common mitigation strategy that suggested is transformer neutral blockers. 06:51.000 --> 06:56.000 These don't impact the power carrying capacity of a power grid. 06:56.000 --> 07:00.000 However, they could impact ground fault reeling, 07:00.000 --> 07:03.000 which means that they're active devices and they're somewhat expensive. 07:03.000 --> 07:05.000 So we can't put them everywhere. 07:05.000 --> 07:09.000 And this motivates analysis and placement strategies 07:09.000 --> 07:14.000 in order to allocate them in a cost-effective way. 07:14.000 --> 07:17.000 And that motivates power models GMD. 07:17.000 --> 07:21.000 So this is a Julia jump package built on top of power models, 07:21.000 --> 07:23.000 infrastructure models, 07:23.000 --> 07:28.000 and includes a number of different analysis and mitigation formulations, 07:28.000 --> 07:31.000 as well as different realizations, 07:31.000 --> 07:34.000 which include our full AC polar equations, 07:34.000 --> 07:36.000 as well as semi-definite programming, 07:36.000 --> 07:38.000 and second order cone realizations. 07:38.000 --> 07:43.000 We also can interface into not optimization solvers 07:43.000 --> 07:45.000 through the jump interface, 07:45.000 --> 07:49.000 or we do have our own custom matrix based solvers 07:49.000 --> 07:51.000 when we're looking at getting a fast result, 07:51.000 --> 07:54.000 which I'm doing here. 07:54.000 --> 07:57.000 As far as what's our data model look like, 07:57.000 --> 08:00.000 so on the right we have our standard AC components, 08:00.000 --> 08:06.000 and then to that we add another set of so-called DC components. 08:06.000 --> 08:09.000 So we represent this as you know, 08:09.000 --> 08:12.000 one single interconnected network. 08:12.000 --> 08:16.000 What do we need in order to solve this problem? 08:16.000 --> 08:18.000 So there's some additional information 08:18.000 --> 08:21.000 that's needed besides our usual AC inputs, 08:21.000 --> 08:23.000 so this includes substation locations, 08:23.000 --> 08:25.000 transfer of winding configurations, 08:25.000 --> 08:28.000 and bus substation mappings. 08:28.000 --> 08:31.000 We originally using a custom file format, 08:31.000 --> 08:35.000 although we've now moved to a semi-standard, 08:36.000 --> 08:39.000 file format that's been developed under PSSC, 08:39.000 --> 08:43.000 and supported under other software packages. 08:43.000 --> 08:45.000 So what's this look like? 08:45.000 --> 08:49.000 Basically let's say CSV if format, 08:49.000 --> 08:52.000 which is relatively easy to parse. 08:52.000 --> 08:55.000 Okay, and now that I'm almost done, 08:55.000 --> 08:58.000 here we have the analysis I promised. 08:58.000 --> 09:01.000 So this is a 500 bus synthetic case, 09:01.000 --> 09:05.000 produced by the Texas A&M University, 09:05.000 --> 09:08.000 and we're applying a pretty hefty, 09:08.000 --> 09:11.000 10 volt per kilometer east-west field. 09:11.000 --> 09:12.000 And we use a screening criteria, 09:12.000 --> 09:15.000 like which of our transformers have a weighted neutral current 09:15.000 --> 09:18.000 over 100 amps? 09:18.000 --> 09:23.000 So this is what we got out of 131 total transformers. 09:23.000 --> 09:28.000 We only see that 13 have very high neutral currents. 09:29.000 --> 09:33.000 So this is actually kind of encouraging, 09:33.000 --> 09:39.000 and it's relatively similar to other studies. 09:39.000 --> 09:47.000 So while there's certainly a degree of exogenous uncertainty in here, 09:47.000 --> 09:52.000 so you don't want to take what I'm saying for grit, 09:52.000 --> 09:56.000 you know, completely trustworthy. 09:57.000 --> 10:00.000 Chances are, you know, the power grid is probably going to be fine, 10:00.000 --> 10:02.000 and if you want to buy a Tesla, 10:02.000 --> 10:05.000 you're probably safe doing that. 10:05.000 --> 10:10.000 Okay, so GitHub repo is here, 10:10.000 --> 10:16.000 and I guess I can answer any questions folks of. 10:17.000 --> 10:26.000 Thank you. 10:26.000 --> 10:35.000 Sure. 10:35.000 --> 10:38.000 Okay, yeah, so there's been a question. 10:38.000 --> 10:41.000 The question was, have transformers been tested 10:41.000 --> 10:44.000 over 100 amps to see what happens, 10:44.000 --> 10:48.000 and yes, there is some hardware testing and modeling. 10:48.000 --> 10:56.000 I believe that's done by Luis Martiel. 10:56.000 --> 11:05.000 Is any other questions? 11:05.000 --> 11:10.000 Cool, thank you.