WEBVTT 00:00:00.240 --> 00:00:02.766 All right. Good morning everybody to ... 00:00:02.766 --> 00:00:04.559 the last session of Area 6. 00:00:05.040 --> 00:00:06.880 My name is Severin Habisreutinger. 00:00:06.880 --> 00:00:09.800 I'm the external projects manager at Oxford pv. 00:00:09.800 --> 00:00:15.360 I will be chairing this session together with my lovely co-chair, 00:00:15.360 --> 00:00:17.280 Dr. Noel from the University of Oxford. 00:00:18.800 --> 00:00:22.428 The original co chair, Sam Teale actually cannot be ... 00:00:22.428 --> 00:00:24.959 here for a very interesting reason. 00:00:25.200 --> 00:00:29.587 He is currently racing a horse that is not a metaphor, 00:00:29.587 --> 00:00:33.774 but there is a literal marathon in Wales right now, 00:00:33.774 --> 00:00:38.759 man versus horse that he is currently running or is about to. 00:00:39.080 --> 00:00:41.030 Unfortunately, we don't have a live stream ... 00:00:41.030 --> 00:00:43.305 because that would have been very entertaining to ... 00:00:43.305 --> 00:00:45.959 every once in a while check in with Sam to see how he's doing. 00:00:47.640 --> 00:00:49.782 Other than that, just a quick reminder that if ... 00:00:49.782 --> 00:00:51.980 you have questions, always use the microphone ... 00:00:51.980 --> 00:00:53.558 because they are being recorded. 00:00:54.640 --> 00:00:57.154 Other than that, I would like to introduce our ... 00:00:57.154 --> 00:01:00.396 first speaker of the session, which is Malcolm Abbott from ... 00:01:00.396 --> 00:01:03.506 Australia from PV Lighthouse, who is going to tell us all ... 00:01:03.506 --> 00:01:06.616 about the spectral correction in yield forecast for two ... 00:01:06.616 --> 00:01:09.197 terminal perovskite silicon tandem modules. 00:01:09.200 --> 00:01:10.640 Malcolm, whenever you're ready. 00:01:10.720 --> 00:01:12.520 Okay, thank you for that introduction. 00:01:12.520 --> 00:01:15.276 I don't know quite how I can compete with man versus ... 00:01:15.276 --> 00:01:17.376 horse, but maybe spectral correction ... 00:01:17.376 --> 00:01:19.279 in yield forecasts is second best. 00:01:20.120 --> 00:01:22.523 So thanks everyone for being here on the final morning of ... 00:01:22.523 --> 00:01:23.239 the conference. 00:01:23.800 --> 00:01:26.558 For those of you who don't know PV Lighthouse, 00:01:26.558 --> 00:01:29.969 we're an Australian company and we provide software and ... 00:01:29.969 --> 00:01:31.638 expertise to the industry. 00:01:32.280 --> 00:01:35.027 We've been doing that for over 10 years now and at ... 00:01:35.027 --> 00:01:39.079 previous IEEE conferences we've been presenting in the system space. 00:01:40.040 --> 00:01:43.232 In particular last year we were talking about spectral ... 00:01:43.232 --> 00:01:45.799 correction of crystalline silicon panels. 00:01:46.200 --> 00:01:49.457 But with the emergence of all the press releases around ... 00:01:49.457 --> 00:01:52.218 tandems pushing towards commercialization, 00:01:52.218 --> 00:01:55.404 we thought it would be interesting and useful to start ... 00:01:55.404 --> 00:01:59.157 integrating that into our yield forecasting software as well. 00:01:59.960 --> 00:02:03.312 So there are a lot of reasons why you might want to solve ... 00:02:03.312 --> 00:02:05.919 the energy yield of tandem technologies. 00:02:05.920 --> 00:02:07.758 I mean, we talk a lot about standard ... 00:02:07.758 --> 00:02:09.976 test conditions, but once you've had your ... 00:02:09.976 --> 00:02:11.624 moment of fame on the IV tester, 00:02:11.624 --> 00:02:14.793 ultimately you go out into the field where you do experience ... 00:02:14.793 --> 00:02:15.997 a range of conditions. 00:02:16.400 --> 00:02:18.160 So it can be useful from that perspective. 00:02:18.320 --> 00:02:20.240 But there's other points to make here as well. 00:02:20.800 --> 00:02:23.573 If we are heading towards utility scale, 00:02:23.573 --> 00:02:27.407 then yield forecasts are actually required and they're ... 00:02:27.407 --> 00:02:29.038 used at various stages. 00:02:29.680 --> 00:02:32.961 So we need to make sure that by the time that's happening, 00:02:32.961 --> 00:02:37.359 we have software in place that can do this with confidence for those people. 00:02:37.760 --> 00:02:40.223 And I think in terms of integrating tandems, 00:02:40.223 --> 00:02:42.952 it gets quite interesting because obviously the ... 00:02:42.952 --> 00:02:46.081 spectral dependence is quite a lot stronger and how that ... 00:02:46.081 --> 00:02:48.278 then mixes in with the thermal effects. 00:02:48.840 --> 00:02:50.920 So we've started putting that into our software. 00:02:50.920 --> 00:02:53.768 But part of the reason we wanted to come into this sort ... 00:02:53.768 --> 00:02:56.681 of a session was to say to people that if you are working ... 00:02:56.681 --> 00:02:58.623 on this and characterizing things, 00:02:58.623 --> 00:03:01.471 to keep this in mind as well, because developing more ... 00:03:01.471 --> 00:03:03.478 models would be really appreciated. 00:03:03.880 --> 00:03:06.580 And particularly for us, what we're looking for is ... 00:03:06.580 --> 00:03:09.794 things that can solve quickly, because our yield solves all ... 00:03:09.794 --> 00:03:11.722 have to happen less than 15 minutes, 00:03:11.722 --> 00:03:14.358 and typically we're talking about a few minutes. 00:03:14.680 --> 00:03:17.320 So we need stuff that's quick to solve. 00:03:18.440 --> 00:03:20.200 And there's some other interesting things that come ... 00:03:20.200 --> 00:03:22.040 up once you start integrating this into your software. 00:03:22.040 --> 00:03:24.200 Like what happens in the case of bifacial. 00:03:24.440 --> 00:03:27.480 We spend a lot of time solving that for people. 00:03:27.480 --> 00:03:29.240 And so how does that work for tandems? 00:03:29.320 --> 00:03:31.160 How are they going to interact with inverters? 00:03:31.800 --> 00:03:33.720 Because the inverter is what's setting the operating point. 00:03:34.200 --> 00:03:37.032 And do you want to start optimizing the design of the ... 00:03:37.032 --> 00:03:38.679 cell and module for the field? 00:03:39.560 --> 00:03:41.438 Now, last point to make here is that ... 00:03:41.438 --> 00:03:44.677 there is actually a de facto standard yield software that's ... 00:03:44.677 --> 00:03:45.519 used out there. 00:03:45.520 --> 00:03:48.300 It's PVsyst, and love it or hate it, 00:03:48.300 --> 00:03:52.519 it is required at times that you have a PVsyst report. 00:03:53.080 --> 00:03:57.500 And so another thing that we look at in this paper is "can it ... 00:03:57.500 --> 00:04:01.920 be used for tandems and how can we calibrate it to do that?" 00:04:03.520 --> 00:04:07.120 So PV Lighthouse is well known for the SunSolve software. 00:04:07.200 --> 00:04:09.680 It can solve something as simple as just a wafer. 00:04:10.240 --> 00:04:11.686 You can start adding thin films, 00:04:11.686 --> 00:04:13.239 metal fingers, that sort of thing. 00:04:13.240 --> 00:04:14.640 Create a cell in a module. 00:04:14.800 --> 00:04:18.280 It's used by all the major cell and module manufacturers. 00:04:18.280 --> 00:04:20.982 I think we calculated something like 60% 00:04:20.982 --> 00:04:24.479 of all modules in 2024 were from SunSolve customers. 00:04:24.640 --> 00:04:27.680 And there's been over 100 academic papers so citing it. 00:04:27.920 --> 00:04:30.480 So it's pretty well validated at this point. 00:04:30.880 --> 00:04:33.920 But you can actually connect that all the way down to the system level. 00:04:34.240 --> 00:04:37.044 And down there we have major developers and IEs ... 00:04:37.044 --> 00:04:40.639 using it to answer complicated questions on a system level. 00:04:41.600 --> 00:04:44.046 The most relevant thing maybe to go into today, 00:04:44.046 --> 00:04:47.119 because to describe how that all works would take too long. 00:04:47.280 --> 00:04:49.200 It accounts for the wavelength dependence. 00:04:49.280 --> 00:04:51.040 So it's in the materials in the cell. 00:04:51.520 --> 00:04:53.440 It's in things like the albedo in the system. 00:04:54.240 --> 00:04:56.927 And when we solve the yield, we actually synthesize the ... 00:04:56.927 --> 00:04:59.439 spectra for direct and diffuse light at every hour. 00:04:59.680 --> 00:05:02.974 So you can make changes on the thickness of your thin film ... 00:05:02.974 --> 00:05:06.479 and see how that's going to play out in terms of energy yield. 00:05:06.720 --> 00:05:08.821 And that's what we're going to be looking at a little bit today ... 00:05:08.821 --> 00:05:09.439 with the tandems. 00:05:10.320 --> 00:05:12.440 Now, we don't make tandem cells. 00:05:12.440 --> 00:05:14.160 We had to go and borrow one from the literature. 00:05:15.280 --> 00:05:18.560 Kaust use SunSolve, they had a model for their tandem. 00:05:18.720 --> 00:05:20.640 So this is what we're going to use as an example. 00:05:21.720 --> 00:05:24.457 You can see there where they've matched the output of ... 00:05:24.457 --> 00:05:26.199 SunSolve to their measurements. 00:05:27.000 --> 00:05:28.440 So that's the cell that we'll be using. 00:05:29.560 --> 00:05:32.468 Now one of the interesting things is how we going to ... 00:05:32.468 --> 00:05:34.199 handle the thermal side of it. 00:05:34.760 --> 00:05:37.370 So for this first version of it, what we've done is we've ... 00:05:37.370 --> 00:05:39.639 treated it just like we do crystalline silicon. 00:05:39.880 --> 00:05:41.924 So we start with an optical solve, 00:05:41.924 --> 00:05:44.845 we do that at a single temperature and then we're ... 00:05:44.845 --> 00:05:48.277 going to correct the light generated current for the top ... 00:05:48.277 --> 00:05:51.709 and bottom cell based on the operating temperature that ... 00:05:51.709 --> 00:05:52.877 we've calculated. 00:05:53.680 --> 00:05:56.494 So to get that, we've fit the coefficients to ... 00:05:56.494 --> 00:05:58.319 this measurement data here. 00:05:58.480 --> 00:06:03.760 And that was measured, in the actual stack. 00:06:04.000 --> 00:06:06.640 So you've got the top cell, that's pretty clear how that's changing. 00:06:06.800 --> 00:06:09.826 But then for the crystalline silicon there's two ... 00:06:09.826 --> 00:06:13.295 components to it because the current increases with the ... 00:06:13.295 --> 00:06:14.919 increasing temperature. 00:06:15.000 --> 00:06:17.103 But then there's also an additional amount of current ... 00:06:17.103 --> 00:06:18.599 that's coming through that top layer. 00:06:19.320 --> 00:06:22.840 This was captured here and we add that into the coefficient. 00:06:22.840 --> 00:06:25.465 So these temperature coefficients for the Jsc are ... 00:06:25.465 --> 00:06:28.840 different to what we would usually use for crystalline silicon. 00:06:29.720 --> 00:06:30.720 So that's where we're at. 00:06:30.720 --> 00:06:33.400 And this is an example of where we would like some better models. 00:06:33.400 --> 00:06:35.200 And you can have a look at the paper as well. 00:06:35.200 --> 00:06:39.480 For things like the voltage. Obviously this is for front side illumination. 00:06:39.640 --> 00:06:41.880 So what happens from the rear side is a bit of a problem. 00:06:42.280 --> 00:06:43.720 It's not wavelength dependent yet. 00:06:43.720 --> 00:06:45.160 So there's a lot more we could do here. 00:06:45.960 --> 00:06:50.040 And if anyone has any ideas, they're welcome to contact us. 00:06:50.680 --> 00:06:53.103 We are then going to put that cell into a module. 00:06:53.103 --> 00:06:54.159 We've built that up. 00:06:54.160 --> 00:06:56.120 That's the SunSolve interface there. 00:06:56.200 --> 00:06:58.120 We've added EVA and glass layers. 00:06:58.280 --> 00:07:00.994 And really what I've done is I've just swapped out a PERC ... 00:07:00.994 --> 00:07:03.000 cell for a two terminal tandem cell here. 00:07:03.160 --> 00:07:05.880 So it's 144 half cut cells. 00:07:06.280 --> 00:07:08.360 I think that panel was about 700 W. 00:07:10.200 --> 00:07:11.560 And then we put that into a system. 00:07:12.440 --> 00:07:16.440 SunSolve can create quite complicated 3D scenes. 00:07:17.080 --> 00:07:19.497 But we don't need that here because we're just really ... 00:07:19.497 --> 00:07:21.639 focusing in on the cell and module technology. 00:07:22.200 --> 00:07:25.208 But what we've done is mounted it on a single axis ... 00:07:25.208 --> 00:07:29.720 tracker and then we've scaled that out to create an entire array of them. 00:07:29.960 --> 00:07:32.640 We had to match the voltage to the inverter. 00:07:32.640 --> 00:07:34.890 So there's 10 modules in the string and then there's a ... 00:07:34.890 --> 00:07:37.243 whole bunch of strings in parallel connected to quite a ... 00:07:37.243 --> 00:07:37.959 large inverter. 00:07:38.280 --> 00:07:40.560 We've also, for the paper, 00:07:40.560 --> 00:07:42.840 used a DC to AC ratio of 1.3. 00:07:43.000 --> 00:07:45.663 That's where we've oversized the DC side so that on certain ... 00:07:45.663 --> 00:07:47.239 days there's going to be clipping. 00:07:47.400 --> 00:07:50.206 And we did that because we wanted to have a look at how ... 00:07:50.206 --> 00:07:53.077 that changes the spectral effects before and after, 00:07:53.077 --> 00:07:54.839 basically on the DC and AC sides. 00:07:55.080 --> 00:07:57.160 So how do we calculate the spectral correction? 00:07:57.240 --> 00:07:58.922 Well, we start with the first ... 00:07:58.922 --> 00:08:02.287 SunSolve simulation where we synthesize the spectra and ... 00:08:02.287 --> 00:08:05.302 then we run the simulation again where we always use ... 00:08:05.302 --> 00:08:07.335 AM1.5g. We combine those two to get ... 00:08:07.335 --> 00:08:08.877 our spectral correction. 00:08:09.360 --> 00:08:12.003 So this is a factor that you'd put out the front of the power ... 00:08:12.003 --> 00:08:13.599 had you not accounted for spectra. 00:08:13.680 --> 00:08:15.520 And it gives you what the proper power would be. 00:08:16.160 --> 00:08:18.738 So if it goes below one, you would have overestimated ... 00:08:18.738 --> 00:08:20.319 your power had you not done this. 00:08:20.880 --> 00:08:22.960 And we do that at every hour of the day. 00:08:23.040 --> 00:08:26.240 The paper looks at eight sites. 00:08:26.240 --> 00:08:27.840 We'll just have a look at a couple of them in this talk. 00:08:28.080 --> 00:08:32.040 So we've got the PERC cell from last year's paper on the ... 00:08:32.040 --> 00:08:34.240 left, the tandem on the right. 00:08:34.480 --> 00:08:36.103 We've got two different sites here, 00:08:36.103 --> 00:08:38.159 Golden at the top and Beijing down the bottom. 00:08:38.400 --> 00:08:42.361 And we like using those two sites because Golden tends to ... 00:08:42.361 --> 00:08:45.732 have quite a flat response through the seasons, 00:08:45.732 --> 00:08:47.839 whereas Beijing varies a lot. 00:08:47.840 --> 00:08:50.640 And it just comes down to how the local atmospheric ... 00:08:50.640 --> 00:08:53.640 conditions either line up or don't with changes in the ... 00:08:53.640 --> 00:08:55.639 position of the sun and the air mass. 00:08:57.000 --> 00:09:00.569 The spectral changes for the tandem, 00:09:00.569 --> 00:09:04.599 it's more dependent, so it's much bigger. 00:09:04.680 --> 00:09:06.600 It's like four times there [compared to the PERC]. 00:09:06.920 --> 00:09:10.525 You can see the factors if you look at it from an annual ... 00:09:10.525 --> 00:09:11.999 energy point of view. 00:09:12.000 --> 00:09:14.036 In Beijing, we're talking about a 5% 00:09:14.036 --> 00:09:17.159 error we would have had in the yield had we not done this. 00:09:17.960 --> 00:09:20.917 And then there's also quite a bit of seasonal variation. 00:09:20.917 --> 00:09:23.308 That can be important because sometimes what ... 00:09:23.308 --> 00:09:26.076 we're doing with yield assessment is we're actually ... 00:09:26.076 --> 00:09:29.159 comparing it to measurements at a particular time of year. 00:09:29.400 --> 00:09:31.861 And so if you're only looking at a couple of weeks or a ... 00:09:31.861 --> 00:09:34.035 couple of days, you really need to account for ... 00:09:34.035 --> 00:09:36.438 these sorts of spectral changes through the year. 00:09:37.880 --> 00:09:41.690 In terms of bifaciality, we solve the bifacial gain for ... 00:09:41.690 --> 00:09:45.831 the system by running another simulation where we cover the ... 00:09:45.831 --> 00:09:49.226 backside of the panel and you can see for the PERC, 00:09:49.226 --> 00:09:53.118 in this case it's quite significant 6.7% bifacial gain. 00:09:53.600 --> 00:09:56.360 There was still some gain for the two terminal tandem. 00:09:56.360 --> 00:09:58.300 I mean, you need the conditions to be ... 00:09:58.300 --> 00:10:01.533 such that the top cell is not limiting what extra is coming in ... 00:10:01.533 --> 00:10:03.279 the backside, but it is reduced. 00:10:04.800 --> 00:10:08.079 Now you can try to optimise for that by thickening up the ... 00:10:08.079 --> 00:10:11.145 perovskite on the top, but it will be tricky because ... 00:10:11.145 --> 00:10:14.638 there is a wide range of rear irradiances that are going on. 00:10:14.840 --> 00:10:18.570 And you'd really have to do it for each site to find where the ... 00:10:18.570 --> 00:10:19.559 best design is. 00:10:19.800 --> 00:10:21.896 When we run that sort of optimization, 00:10:21.896 --> 00:10:24.844 we typically find that what would be best for STC is not ... 00:10:24.844 --> 00:10:28.119 really best for what would happen in the actual site itself. 00:10:29.880 --> 00:10:32.120 Then we have the inverter clipping. 00:10:32.440 --> 00:10:34.360 So here we're looking at three different days. 00:10:35.480 --> 00:10:39.849 The blue is the AC output of the inverter and the red is the ... 00:10:39.849 --> 00:10:42.079 DC side coming off the array. 00:10:42.720 --> 00:10:44.886 And you can see for that first sunny day where you get this ... 00:10:44.886 --> 00:10:46.868 whole part of the day where you're exceeding what's ... 00:10:46.868 --> 00:10:48.158 happening and it's clipping off. 00:10:49.760 --> 00:10:52.331 Sometimes as a cell engineer that's really disappointing ... 00:10:52.331 --> 00:10:54.902 because you've spent all this effort to get extra power out ... 00:10:54.902 --> 00:10:56.318 and then it's gone on those days. 00:10:56.640 --> 00:10:58.715 But you can see that what's happening is when you come ... 00:10:58.715 --> 00:11:00.319 down to the cloudy day, it still matters. 00:11:00.720 --> 00:11:02.272 Now, the effect on the spectral ... 00:11:02.272 --> 00:11:04.959 correction is that it becomes one during these times. 00:11:05.440 --> 00:11:07.597 It really doesn't matter if you were accounting for it or not ... 00:11:07.597 --> 00:11:08.719 because you've clipped it off. 00:11:09.720 --> 00:11:12.640 So you need to think about whether you're assessing this on the DC 00:11:12.640 --> 00:11:15.842 side or whether you want to get onto the AC side and have ... 00:11:15.842 --> 00:11:18.119 a look at what it meant in that context. 00:11:18.360 --> 00:11:21.068 SunSolve lets you string this all the way through to the ... 00:11:21.068 --> 00:11:22.599 inverter level and take a look. 00:11:22.760 --> 00:11:25.182 And maybe you'd be optimizing now for different ... 00:11:25.182 --> 00:11:26.999 times of day and cloudy conditions. 00:11:27.880 --> 00:11:31.080 So I'm just going to finish up by talking about PVsyst. 00:11:31.560 --> 00:11:35.082 So a lot of software that's out there requires spectral ... 00:11:35.082 --> 00:11:38.681 correction through a kind of model like you can see there ... 00:11:38.681 --> 00:11:40.978 in the middle, where you just have an ... 00:11:40.978 --> 00:11:44.194 equation that links to the things like air mass and ... 00:11:44.194 --> 00:11:45.878 precipitate water vapor. 00:11:46.360 --> 00:11:48.520 And then it has these coefficients out the front. 00:11:49.240 --> 00:11:52.056 So that's in the software right now and it has default ... 00:11:52.056 --> 00:11:55.000 coefficients you can see those in the purple line here. 00:11:55.160 --> 00:11:57.759 So what we're plotting is the spectral correction factor ... 00:11:57.759 --> 00:11:58.679 versus the air mass. 00:11:59.080 --> 00:12:01.341 And the yellow dots are what came off that SunSolve ... 00:12:01.341 --> 00:12:03.279 simulation we were looking at for Beijing. 00:12:03.680 --> 00:12:07.117 What we're doing to sort of calibrate this model is we're ... 00:12:07.117 --> 00:12:10.115 then refitting those coefficients to what we've ... 00:12:10.115 --> 00:12:12.528 simulated, and then we can put that into ... 00:12:12.528 --> 00:12:14.868 PVsyst, and that starts to account for ... 00:12:14.868 --> 00:12:16.477 the spectral correction. 00:12:16.560 --> 00:12:19.520 So that allows you to then use PVsyst more accurately. 00:12:20.480 --> 00:12:22.320 And SunSolve is able to do all of that. 00:12:23.760 --> 00:12:25.671 It is sort of hard to follow this one, 00:12:25.671 --> 00:12:27.519 but it's an example of how it works. 00:12:27.600 --> 00:12:29.984 On the left, we've got the assumption that ... 00:12:29.984 --> 00:12:32.505 there's AM1.5g, and we're always comparing ... 00:12:32.505 --> 00:12:35.639 on the bottom axis there to where we've synthesized it. 00:12:35.800 --> 00:12:37.611 So we have an error in the annual yield of 5% 00:12:37.611 --> 00:12:39.319 there when we haven't corrected for it. 00:12:39.640 --> 00:12:42.840 And then we've applied that First Solar correction just to those dots. 00:12:43.160 --> 00:12:44.880 We haven't run PVsyst here. 00:12:44.880 --> 00:12:46.788 There's other differences between SunSolve and ... 00:12:46.788 --> 00:12:47.951 PVsyst, but we've applied the ... 00:12:47.951 --> 00:12:49.719 correction factor to bring that back across. 00:12:50.440 --> 00:12:54.074 And it got all the way to zero, which is somewhat surprising ... 00:12:54.074 --> 00:12:57.559 because this side actually has a lot of aerosols as well. 00:12:58.000 --> 00:13:02.491 But we find that we can usually get it close to zero ... 00:13:02.491 --> 00:13:05.119 through this way of doing it. 00:13:05.440 --> 00:13:08.448 So this is at least an initial way that we can use PVsyst ... 00:13:08.448 --> 00:13:10.319 for these sorts of technologies. 00:13:11.040 --> 00:13:14.455 So just to finish up, spectral effects are significant ... 00:13:14.455 --> 00:13:16.533 for two terminal tandem systems, 00:13:16.533 --> 00:13:20.171 and you can use SunSolve to explore that and quantify it for ... 00:13:20.171 --> 00:13:22.398 different technologies and sites. 00:13:23.120 --> 00:13:27.400 On both the DC and AC side, it's computing the synthetic spectra. 00:13:27.640 --> 00:13:30.100 And it's probably worth pointing out that is also an ... 00:13:30.100 --> 00:13:32.732 area of research still coming up with better and better ... 00:13:32.732 --> 00:13:35.078 models for recreating that from the atmosphere. 00:13:35.480 --> 00:13:38.335 And so you can do that for any system and location, 00:13:38.335 --> 00:13:41.399 and you can also use it to inform PVsyst simulations. 00:13:42.280 --> 00:13:44.068 So with that, I'd like to thank ARENA, 00:13:44.068 --> 00:13:45.799 who are now funding some of our work. 00:13:46.280 --> 00:13:48.453 And if you want to learn more about SunSolve, 00:13:48.453 --> 00:13:50.039 we have the sunsolve.info site. 00:13:50.600 --> 00:13:52.920 We post stuff on a blog and videos there as well. 00:13:53.000 --> 00:13:54.360 Thank you for listening. 00:13:59.760 --> 00:14:01.280 Excellent. Questions. 00:14:04.240 --> 00:14:09.720 [Nils from Total] Hi Mal, thank you. 00:14:09.720 --> 00:14:10.960 Lots of stuff in there. 00:14:11.680 --> 00:14:14.168 A naive question, because I didn't capture all of ... 00:14:14.168 --> 00:14:15.199 what you presented. 00:14:15.440 --> 00:14:19.286 Were you suggesting that, at least to some degree of ... 00:14:19.286 --> 00:14:22.238 accuracy, one can represent the effect ... 00:14:22.238 --> 00:14:25.905 of a tandem outdoor performance by just applying ... 00:14:25.905 --> 00:14:30.198 a certain spectral mismatch factor in a single junction. 00:14:31.160 --> 00:14:33.000 Namely PVsyst simulation. 00:14:34.360 --> 00:14:35.788 Yeah, that's right. 00:14:35.788 --> 00:14:39.400 So PVsyst often deals with effects by just creating ... 00:14:39.400 --> 00:14:41.080 factors, loss factors. 00:14:42.040 --> 00:14:44.747 And when CdTe and First Solar came along, 00:14:44.747 --> 00:14:48.648 they needed a way to account for the fact that it really uses ... 00:14:48.648 --> 00:14:50.479 the spectrum differently. 00:14:51.040 --> 00:14:52.960 And so that was what was developed for that. 00:14:53.600 --> 00:14:59.600 And I think as a first pass to get it into PVsyst, this is good. 00:15:00.560 --> 00:15:03.850 I think that that model and that equation probably could ... 00:15:03.850 --> 00:15:07.280 be better and we would need to explore it for more examples. 00:15:07.280 --> 00:15:09.251 We've just sort of done one or two, 00:15:09.251 --> 00:15:12.349 but it seems to us like right now that would be the best ... 00:15:12.349 --> 00:15:13.334 way to do it. Yes. 00:15:13.334 --> 00:15:14.038 Okay, cool. 00:15:14.040 --> 00:15:15.975 I mean, there still has to be pan files ... 00:15:15.975 --> 00:15:18.159 and things like that created. Yeah, sure. 00:15:18.160 --> 00:15:21.280 So this is a whole area of work that needs to be done, I think. 00:15:21.920 --> 00:15:22.560 Thank you. 00:15:25.840 --> 00:15:26.480 Thanks, Mal. 00:15:27.040 --> 00:15:29.200 [Natasha from Q-Cells] You used a header junction bottom cell. 00:15:29.200 --> 00:15:31.743 Like how much do you expect it to shift at all if, 00:15:31.743 --> 00:15:33.959 like you have a topcon, like bottom cell? 00:15:33.960 --> 00:15:37.520 Or is it kind of in the noise with the shifting band gap at the top cell? 00:15:38.400 --> 00:15:40.320 Yeah, I mean, that's. 00:15:40.720 --> 00:15:41.880 How much will it like. 00:15:41.880 --> 00:15:44.800 We haven't really explored changing the top cell and the bottom cell. 00:15:46.000 --> 00:15:47.811 For us, it's about our customers ... 00:15:47.811 --> 00:15:50.159 bringing in whatever they happen to have. 00:15:51.040 --> 00:15:53.440 So I really couldn't answer that actually. 00:15:54.160 --> 00:15:56.643 We just have those measurements and there's not ... 00:15:56.643 --> 00:15:59.002 a lot of these sorts of measurements I've seen ... 00:15:59.002 --> 00:16:01.919 where you get EQE measured as a function of temperature. 00:16:03.120 --> 00:16:05.072 I think in general, I find the perovskite tandem ... 00:16:05.072 --> 00:16:05.999 area really exciting. 00:16:06.000 --> 00:16:08.045 With all the extra characterization that has to go ... 00:16:08.045 --> 00:16:09.359 on, it's much more interesting. 00:16:10.000 --> 00:16:11.240 It needs measuring, though. 00:16:11.240 --> 00:16:12.119 Yeah, cool. Thanks. 00:16:12.200 --> 00:16:13.960 [Severin from OxfordPV] Maybe a quick follow up from me. 00:16:13.960 --> 00:16:15.880 Have you looked at four terminal tandems? 00:16:16.760 --> 00:16:17.800 No, we haven't yet. 00:16:18.440 --> 00:16:22.600 The way SunSolve works though is it's just a great big spice circuit. 00:16:23.320 --> 00:16:27.579 So I think that's almost one of the next things to do actually, 00:16:27.579 --> 00:16:30.919 is to wire it up and like, I'm not even quite sure. 00:16:30.920 --> 00:16:33.945 At what point does it stop being four terminals because ... 00:16:33.945 --> 00:16:36.839 the inverter has two and the modules tend to have two. 00:16:37.520 --> 00:16:40.718 So I guess we would then be simulating some of the voltage ... 00:16:40.718 --> 00:16:42.079 mismatch at that point. 00:16:42.400 --> 00:16:43.680 So certainly it can be done. 00:16:44.080 --> 00:16:46.175 We haven't quite put it in there yet, 00:16:46.175 --> 00:16:47.222 but yeah, we will, 00:16:47.222 --> 00:16:50.225 particularly once someone starts asking to see it. 00:16:50.225 --> 00:16:50.923 Yeah, nice. 00:16:50.923 --> 00:16:52.878 Nils is back. Sure. One final one. 00:16:53.840 --> 00:16:56.730 [Nils from Total] Yeah, I struggle a bit with the topic ... 00:16:56.730 --> 00:16:58.079 that I asked about first. 00:16:58.080 --> 00:16:59.820 So let's say you have a tandem, 00:16:59.820 --> 00:17:01.978 you know, what should be the spectrum ... 00:17:01.978 --> 00:17:05.598 mismatch factor to treat it like a one junction cell in PVsyst. 00:17:06.320 --> 00:17:09.263 But now you will have different temperatures and ... 00:17:09.263 --> 00:17:11.344 that should, in my understanding, 00:17:11.344 --> 00:17:14.215 change your spectrum mismatch factor due to the ... 00:17:14.215 --> 00:17:16.799 band gap effect that you were pointing out. 00:17:17.440 --> 00:17:20.984 So is that the point where the simplification breaks down ... 00:17:20.984 --> 00:17:24.159 and you should rather have a full blown simulation? 00:17:24.480 --> 00:17:26.705 Well, it's a good point, 00:17:26.705 --> 00:17:31.600 but we are also lining up maybe more than I made clear. 00:17:31.680 --> 00:17:34.560 So you would be using the same TMY data, for example. 00:17:35.440 --> 00:17:39.416 So PVsyst would be calculating the temperature of ... 00:17:39.416 --> 00:17:41.120 the cell at each hour. 00:17:41.280 --> 00:17:43.935 It would probably be very similar to what we'd done ... 00:17:43.935 --> 00:17:45.199 when we synthesized it. 00:17:45.760 --> 00:17:50.400 So how well it would then work for other TMY files, for example. 00:17:51.920 --> 00:17:54.073 Again we haven't explored it, so we could. 00:17:54.073 --> 00:17:55.119 Yeah, you're right. 00:17:55.120 --> 00:17:58.094 You could calibrate it in one site and see how well it works ... 00:17:58.094 --> 00:17:59.519 in lots of different sites. 00:18:00.880 --> 00:18:02.588 In our opinion, the thing to do is to ... 00:18:02.588 --> 00:18:04.414 synthesize the spectra and solve it, 00:18:04.414 --> 00:18:06.829 but that's probably not going to get into PVsyst, 00:18:06.829 --> 00:18:09.303 so there needs to be a way to put it in as best as we can ... 00:18:09.303 --> 00:18:10.834 and yeah, that's a good idea for ... 00:18:10.834 --> 00:18:12.718 something for someone to try, I think. 00:18:13.840 --> 00:18:14.870 All right, fantastic. 00:18:14.870 --> 00:18:16.279 Thank you, Malcolm. Thank you. 00:18:16.280 --> 00:18:18.241 That was an excellent presentation. 00:18:18.241 --> 00:18:19.759 Not lying. Another applaud.