WEBVTT 00:00:00.000 --> 00:00:01.931 align:middle line:90% [SQUEAKING] 00:00:01.931 --> 00:00:03.362 align:middle line:90% [RUSTLING] 00:00:03.362 --> 00:00:05.270 align:middle line:90% [CLICKING] 00:00:05.270 --> 00:00:10.820 align:middle line:90% 00:00:10.820 --> 00:00:15.560 align:middle line:84% PROFESSOR: OK, so let me just contextualize where we are. 00:00:15.560 --> 00:00:19.280 align:middle line:84% So I showed you this, we developed these screening curves 00:00:19.280 --> 00:00:21.080 align:middle line:90% in the previous class. 00:00:21.080 --> 00:00:23.160 align:middle line:84% And for dispatchable technologies, 00:00:23.160 --> 00:00:28.940 align:middle line:84% they are a way of figuring out what is the optimal technology. 00:00:28.940 --> 00:00:32.720 align:middle line:84% The first key lesson here is that it's never one. 00:00:32.720 --> 00:00:35.520 align:middle line:84% It depends on how frequently you use it. 00:00:35.520 --> 00:00:37.420 align:middle line:84% If you're building a peaker plant, 00:00:37.420 --> 00:00:39.820 align:middle line:84% you're going to use a technology with low capital cost, 00:00:39.820 --> 00:00:43.420 align:middle line:84% but higher fuel cost, or higher variable costs. 00:00:43.420 --> 00:00:44.400 align:middle line:90% So it's steeper. 00:00:44.400 --> 00:00:46.720 align:middle line:84% That's like this brown line here. 00:00:46.720 --> 00:00:49.180 align:middle line:84% If you're building something that's on all the time, 00:00:49.180 --> 00:00:52.560 align:middle line:84% then you're going to use something that is 8,000 hours 00:00:52.560 --> 00:00:54.040 align:middle line:90% a year, for example. 00:00:54.040 --> 00:00:56.340 align:middle line:84% You're going to use something over here, 00:00:56.340 --> 00:01:00.560 align:middle line:84% which has maybe a higher capital cost, nuclear to high capital 00:01:00.560 --> 00:01:02.300 align:middle line:90% costs, but very low fuel costs. 00:01:02.300 --> 00:01:05.560 align:middle line:84% And so it turns out, when we take the whole suite of things 00:01:05.560 --> 00:01:08.920 align:middle line:84% into consideration, that there is 00:01:08.920 --> 00:01:12.360 align:middle line:84% a role, if you will, for high capital cost technologies, 00:01:12.360 --> 00:01:14.680 align:middle line:90% like nuclear. 00:01:14.680 --> 00:01:17.380 align:middle line:84% Geothermal kind of eats his lunch. 00:01:17.380 --> 00:01:19.500 align:middle line:84% But it's not geothermal everywhere. 00:01:19.500 --> 00:01:23.760 align:middle line:84% It's not geothermal in the southwestern United States. 00:01:23.760 --> 00:01:28.160 align:middle line:84% But we see that here, these are US prices. 00:01:28.160 --> 00:01:31.340 align:middle line:84% If we set the social cost of carbon to $100-- 00:01:31.340 --> 00:01:32.053 align:middle line:90% Yes. 00:01:32.053 --> 00:01:34.220 align:middle line:84% AUDIENCE: So could you remind me what the y-axis is? 00:01:34.220 --> 00:01:40.040 align:middle line:84% PROFESSOR: Oh, the y-axis is dollars per kilowatt of-- 00:01:40.040 --> 00:01:43.600 align:middle line:84% I think this is either per kilowatt 00:01:43.600 --> 00:01:46.960 align:middle line:90% or per megawatt of capacity. 00:01:46.960 --> 00:01:49.560 align:middle line:84% I don't know what I multiplied it by. 00:01:49.560 --> 00:01:51.520 align:middle line:84% So this is the amount you basically 00:01:51.520 --> 00:01:54.400 align:middle line:84% have to pay to have that capacity, even 00:01:54.400 --> 00:01:57.240 align:middle line:84% if you don't generate any electricity from that plant. 00:01:57.240 --> 00:02:00.770 align:middle line:84% And then this is the amount that it takes, 00:02:00.770 --> 00:02:04.510 align:middle line:84% if you run the plant for 4,000 hours a year, is here. 00:02:04.510 --> 00:02:08.250 align:middle line:84% So this is the capital charge that you pay every year. 00:02:08.250 --> 00:02:09.250 align:middle line:90% We calculated it. 00:02:09.250 --> 00:02:12.110 align:middle line:84% And then this is essentially the variable cost is the slope. 00:02:12.110 --> 00:02:17.230 align:middle line:84% And that's determining your net price. 00:02:17.230 --> 00:02:19.730 align:middle line:84% So we wind up with all these lines. 00:02:19.730 --> 00:02:21.223 align:middle line:84% And there are certain technologies 00:02:21.223 --> 00:02:22.890 align:middle line:84% that are optimized for different things. 00:02:22.890 --> 00:02:25.590 align:middle line:84% And we talked about the idea of base load. 00:02:25.590 --> 00:02:28.830 align:middle line:84% And base load is something that just means 00:02:28.830 --> 00:02:31.630 align:middle line:90% it operates most of the time. 00:02:31.630 --> 00:02:33.310 align:middle line:84% And we reminded ourselves that base load 00:02:33.310 --> 00:02:34.950 align:middle line:90% isn't a special feature. 00:02:34.950 --> 00:02:37.390 align:middle line:84% It's actually kind of a limitation. 00:02:37.390 --> 00:02:39.130 align:middle line:84% There are technologies, like geothermal, 00:02:39.130 --> 00:02:42.790 align:middle line:84% like nuclear, that only make sense, meaning they're only low 00:02:42.790 --> 00:02:48.210 align:middle line:84% down on the plot, when operated in a base load type capacity. 00:02:48.210 --> 00:02:51.830 align:middle line:84% But there's nothing special, technically, about base load 00:02:51.830 --> 00:02:54.630 align:middle line:90% otherwise. 00:02:54.630 --> 00:02:58.078 align:middle line:90% So what we found is this. 00:02:58.078 --> 00:02:59.370 align:middle line:90% Everyone's good with this plot? 00:02:59.370 --> 00:03:00.930 align:middle line:84% Any other questions we might need to refresh? 00:03:00.930 --> 00:03:01.670 align:middle line:90% Yep. 00:03:01.670 --> 00:03:03.337 align:middle line:84% AUDIENCE: Could you just show this graph 00:03:03.337 --> 00:03:05.250 align:middle line:90% with no social cost of carbon? 00:03:05.250 --> 00:03:06.510 align:middle line:90% PROFESSOR: Yes. 00:03:06.510 --> 00:03:08.710 align:middle line:90% AUDIENCE: Thank you. 00:03:08.710 --> 00:03:10.970 align:middle line:84% PROFESSOR: But that's less exciting. 00:03:10.970 --> 00:03:14.350 align:middle line:90% 00:03:14.350 --> 00:03:15.690 align:middle line:90% OK, there it is at zero. 00:03:15.690 --> 00:03:19.230 align:middle line:90% 00:03:19.230 --> 00:03:23.990 align:middle line:84% So, does anyone know what the current EPA prediction is? 00:03:23.990 --> 00:03:27.470 align:middle line:90% Social cost of carbon? 00:03:27.470 --> 00:03:29.910 align:middle line:84% So during the Biden administration, 00:03:29.910 --> 00:03:33.390 align:middle line:90% they revised their methodology. 00:03:33.390 --> 00:03:38.230 align:middle line:84% Initially, the way it was done, it was a higher discount factor. 00:03:38.230 --> 00:03:40.870 align:middle line:90% So making the social cost lower. 00:03:40.870 --> 00:03:44.950 align:middle line:84% And they considered only the damages in the United States, 00:03:44.950 --> 00:03:47.150 align:middle line:84% because it was a federal entity looking 00:03:47.150 --> 00:03:49.270 align:middle line:84% at the welfare of the United States. 00:03:49.270 --> 00:03:53.550 align:middle line:84% And in the Biden administration, they revised that procedure. 00:03:53.550 --> 00:03:55.300 align:middle line:84% And they said, we should take into account 00:03:55.300 --> 00:04:00.260 align:middle line:84% the damages done to the entire world in our calculation. 00:04:00.260 --> 00:04:04.980 align:middle line:84% And they also lowered the discount factor. 00:04:04.980 --> 00:04:08.420 align:middle line:84% And I think they went from a $22-- 00:04:08.420 --> 00:04:10.420 align:middle line:84% we should go look this up-- $22 to something 00:04:10.420 --> 00:04:13.280 align:middle line:90% like $30 for the near term. 00:04:13.280 --> 00:04:18.579 align:middle line:84% And then this number slowly rises over time. 00:04:18.579 --> 00:04:20.380 align:middle line:84% In like 100 years, I think it reaches 00:04:20.380 --> 00:04:22.540 align:middle line:90% like $100, something like that. 00:04:22.540 --> 00:04:23.620 align:middle line:90% Yeah. 00:04:23.620 --> 00:04:25.380 align:middle line:90% AUDIENCE: 21 to 33. 00:04:25.380 --> 00:04:26.640 align:middle line:90% PROFESSOR: 21 to 33. 00:04:26.640 --> 00:04:28.620 align:middle line:90% Yeah, thank you. 00:04:28.620 --> 00:04:33.860 align:middle line:84% It's great to have just in time fact checking. 00:04:33.860 --> 00:04:39.580 align:middle line:84% So let's just say we're way in the future, 00:04:39.580 --> 00:04:41.580 align:middle line:84% things are much worse than we thought. 00:04:41.580 --> 00:04:42.960 align:middle line:90% And so it happens earlier. 00:04:42.960 --> 00:04:49.620 align:middle line:84% So there's 35, and let's go to 100, 00:04:49.620 --> 00:04:53.900 align:middle line:84% like hundreds, something like that. 00:04:53.900 --> 00:04:58.580 align:middle line:84% So what we see is that if we don't have access to geothermal, 00:04:58.580 --> 00:05:00.520 align:middle line:90% definitely a role for nuclear. 00:05:00.520 --> 00:05:04.720 align:middle line:84% And this was the story in the '90s and early 2000s, 00:05:04.720 --> 00:05:07.860 align:middle line:84% when people like your professors and me were getting into nuclear 00:05:07.860 --> 00:05:11.660 align:middle line:84% power, this is how we looked at the problem. 00:05:11.660 --> 00:05:16.900 align:middle line:84% But then something happened in the 2010s, 00:05:16.900 --> 00:05:20.940 align:middle line:84% and that was solar and wind just got cheaper and cheaper 00:05:20.940 --> 00:05:23.800 align:middle line:84% and cheaper and cheaper and cheaper and ate our lunch. 00:05:23.800 --> 00:05:27.580 align:middle line:90% 00:05:27.580 --> 00:05:29.520 align:middle line:84% So I just showed you results that said, 00:05:29.520 --> 00:05:32.540 align:middle line:84% if you allow solar and wind into the equation, 00:05:32.540 --> 00:05:34.580 align:middle line:84% things sound really bad for nuclear. 00:05:34.580 --> 00:05:38.180 align:middle line:84% And it's just like it was a technology that was essential, 00:05:38.180 --> 00:05:41.420 align:middle line:84% but solar and wind, like I said, ate our lunch. 00:05:41.420 --> 00:05:44.020 align:middle line:84% But there are still places in the world where 00:05:44.020 --> 00:05:48.660 align:middle line:84% solar and wind are not very good, where there's 00:05:48.660 --> 00:05:52.130 align:middle line:84% either high latitude, or high amounts of cloud cover, 00:05:52.130 --> 00:05:55.210 align:middle line:90% or very low available land. 00:05:55.210 --> 00:06:02.050 align:middle line:84% And in those places, the situation looks more like this. 00:06:02.050 --> 00:06:07.010 align:middle line:84% The other part of the story is that all of this 00:06:07.010 --> 00:06:09.720 align:middle line:84% is predicated on what I'm calling nuclear fast, which 00:06:09.720 --> 00:06:11.970 align:middle line:84% is to say that we can build these reactors essentially 00:06:11.970 --> 00:06:15.810 align:middle line:84% at the cost and at the timescales that the vendor says 00:06:15.810 --> 00:06:17.370 align:middle line:90% we can do it. 00:06:17.370 --> 00:06:21.610 align:middle line:84% In the United States, we didn't build any reactors after 00:06:21.610 --> 00:06:26.090 align:middle line:84% the '80s, early '80s, the last reactor was ordered in the '70s. 00:06:26.090 --> 00:06:31.130 align:middle line:84% And then we built two, and they were extremely expensive and not 00:06:31.130 --> 00:06:33.990 align:middle line:84% actually that long in their construction duration, 00:06:33.990 --> 00:06:35.490 align:middle line:84% but they were extremely expensive. 00:06:35.490 --> 00:06:39.150 align:middle line:84% So if nuclear is actually closer to this, 00:06:39.150 --> 00:06:41.690 align:middle line:84% which I call it empirical cost, which 00:06:41.690 --> 00:06:46.690 align:middle line:84% is the average of the Western countries in the last decade, 00:06:46.690 --> 00:06:49.130 align:middle line:84% then it actually doesn't have a role. 00:06:49.130 --> 00:06:50.970 align:middle line:84% So the other part of the story is 00:06:50.970 --> 00:06:55.610 align:middle line:84% that we do have to figure out how to get nuclear cheaper. 00:06:55.610 --> 00:06:58.810 align:middle line:90% Now, if you go back to-- 00:06:58.810 --> 00:07:03.290 align:middle line:90% 00:07:03.290 --> 00:07:06.530 align:middle line:90% let me just push Play. 00:07:06.530 --> 00:07:08.330 align:middle line:84% If you go back to this, you see that there 00:07:08.330 --> 00:07:10.538 align:middle line:84% are plenty of predictions that nuclear would be cheap 00:07:10.538 --> 00:07:13.010 align:middle line:84% and somehow it wound up expensive. 00:07:13.010 --> 00:07:18.290 align:middle line:84% And part of the variability in these predictions 00:07:18.290 --> 00:07:22.650 align:middle line:84% is because they come from Oklo and TerraPower, and companies 00:07:22.650 --> 00:07:24.150 align:middle line:84% that are trying to sell the reactor, 00:07:24.150 --> 00:07:26.192 align:middle line:84% and they've never built it, and they had no idea. 00:07:26.192 --> 00:07:30.290 align:middle line:84% And they're making very highly optimistic predictions 00:07:30.290 --> 00:07:31.590 align:middle line:90% about what things will cost. 00:07:31.590 --> 00:07:34.007 align:middle line:84% And they're going to get all this discount because they're 00:07:34.007 --> 00:07:35.450 align:middle line:90% going to order so many things. 00:07:35.450 --> 00:07:38.210 align:middle line:84% This is exactly what CFS is currently 00:07:38.210 --> 00:07:39.910 align:middle line:90% doing with its cost estimates. 00:07:39.910 --> 00:07:42.070 align:middle line:84% I sat down with them and they're like, really? 00:07:42.070 --> 00:07:44.050 align:middle line:84% You're going to get that major discount? 00:07:44.050 --> 00:07:45.290 align:middle line:90% Oh yeah, we think we can get. 00:07:45.290 --> 00:07:47.410 align:middle line:90% OK, then. 00:07:47.410 --> 00:07:49.660 align:middle line:90% So that's to be expected. 00:07:49.660 --> 00:07:52.220 align:middle line:84% These are sort of now more realistic numbers. 00:07:52.220 --> 00:07:54.860 align:middle line:84% We're going to talk about this one a little bit later today, 00:07:54.860 --> 00:07:56.700 align:middle line:84% The Future of Nuclear Power report. 00:07:56.700 --> 00:07:58.140 align:middle line:90% This is run by [INAUDIBLE]. 00:07:58.140 --> 00:08:02.760 align:middle line:90% They're median number of 7,520. 00:08:02.760 --> 00:08:07.480 align:middle line:84% And then this is currently what the Department of Energy 00:08:07.480 --> 00:08:08.055 align:middle line:90% is saying-- 00:08:08.055 --> 00:08:08.680 align:middle line:90% I guess-- wait. 00:08:08.680 --> 00:08:10.760 align:middle line:84% Yeah, this is the median number, 7,520, 00:08:10.760 --> 00:08:16.480 align:middle line:84% something in this kind of range of around 7,500, 7,700. 00:08:16.480 --> 00:08:17.300 align:middle line:90% What is that? 00:08:17.300 --> 00:08:20.000 align:middle line:90% What does that mean? 00:08:20.000 --> 00:08:22.600 align:middle line:84% Now, just to remind you, these are overnight costs. 00:08:22.600 --> 00:08:25.400 align:middle line:90% So that finance is excluded. 00:08:25.400 --> 00:08:28.880 align:middle line:84% The effect of construction delays is excluded. 00:08:28.880 --> 00:08:31.080 align:middle line:84% And so what we're really trying to measure 00:08:31.080 --> 00:08:35.600 align:middle line:84% is just the inherent cost of the technology 00:08:35.600 --> 00:08:39.600 align:middle line:84% and indirectly a measure of the complexity of the technology. 00:08:39.600 --> 00:08:42.620 align:middle line:84% Or to put it another way, the cost of the inputs, 00:08:42.620 --> 00:08:44.880 align:middle line:84% the amount of labor and the amount of goods 00:08:44.880 --> 00:08:48.640 align:middle line:90% required to build this out. 00:08:48.640 --> 00:08:50.320 align:middle line:84% And so where do these numbers come 00:08:50.320 --> 00:08:54.360 align:middle line:84% from if we didn't build any plants in the US, 00:08:54.360 --> 00:08:56.640 align:middle line:90% until we built this? 00:08:56.640 --> 00:08:58.680 align:middle line:84% They came mainly by looking at what was 00:08:58.680 --> 00:09:00.840 align:middle line:90% happening in other countries. 00:09:00.840 --> 00:09:04.300 align:middle line:84% And people saying, well, if that country can do it, 00:09:04.300 --> 00:09:05.720 align:middle line:90% then we can do it. 00:09:05.720 --> 00:09:09.920 align:middle line:84% So I want to talk about what other countries have 00:09:09.920 --> 00:09:16.400 align:middle line:84% done to see if we can glean some additional phenomenological 00:09:16.400 --> 00:09:20.360 align:middle line:84% insight and what lessons we have. 00:09:20.360 --> 00:09:28.360 align:middle line:84% So this for reference is construction duration 00:09:28.360 --> 00:09:30.600 align:middle line:90% for all US reactors. 00:09:30.600 --> 00:09:35.080 align:middle line:84% And what you see is that it took longer and longer 00:09:35.080 --> 00:09:36.300 align:middle line:90% to build our reactors. 00:09:36.300 --> 00:09:38.140 align:middle line:84% This explains some cost increase, 00:09:38.140 --> 00:09:40.948 align:middle line:90% but not in the overnight cost. 00:09:40.948 --> 00:09:43.240 align:middle line:84% Most of these reactors were supposed to take about half 00:09:43.240 --> 00:09:45.830 align:middle line:84% the time that they actually took. 00:09:45.830 --> 00:09:47.950 align:middle line:90% And those estimates were bad. 00:09:47.950 --> 00:09:50.950 align:middle line:84% And the reason they were bad is because 00:09:50.950 --> 00:09:53.310 align:middle line:84% of safety-related considerations. 00:09:53.310 --> 00:09:55.590 align:middle line:84% So for example, they would pour concrete 00:09:55.590 --> 00:09:58.872 align:middle line:84% and the concrete wouldn't be as strong as it was supposed to be. 00:09:58.872 --> 00:10:00.830 align:middle line:84% They'd have to rip it out and put new concrete. 00:10:00.830 --> 00:10:04.470 align:middle line:84% Or there were issues with improperly installed rebar 00:10:04.470 --> 00:10:05.330 align:middle line:90% in the basement. 00:10:05.330 --> 00:10:07.750 align:middle line:84% So then they had to rip it all out. 00:10:07.750 --> 00:10:11.610 align:middle line:84% And so this suggests, these are, if you will, 00:10:11.610 --> 00:10:14.190 align:middle line:84% measures of the complexity of the technology 00:10:14.190 --> 00:10:16.510 align:middle line:84% and all the safety versions associated 00:10:16.510 --> 00:10:19.510 align:middle line:90% with a complex technology. 00:10:19.510 --> 00:10:24.390 align:middle line:84% So duration is a really useful measure. 00:10:24.390 --> 00:10:26.270 align:middle line:84% And the reason it's a useful measure 00:10:26.270 --> 00:10:30.470 align:middle line:84% is it's directly comparable between different companies. 00:10:30.470 --> 00:10:35.220 align:middle line:84% If someone takes five years to build a reactor, 00:10:35.220 --> 00:10:37.470 align:middle line:84% and someone else takes 10 years to build the reactors, 00:10:37.470 --> 00:10:40.030 align:middle line:84% and those reactors are comparable, 00:10:40.030 --> 00:10:44.850 align:middle line:84% then we have some sense of country-specific variation. 00:10:44.850 --> 00:10:46.750 align:middle line:84% But if you look at dollars, it turns out 00:10:46.750 --> 00:10:51.430 align:middle line:84% there's so many confounding factors about how they count, 00:10:51.430 --> 00:10:54.810 align:middle line:84% who pays what, and do they include the price of the land? 00:10:54.810 --> 00:10:56.910 align:middle line:84% Do they include the price of the engineering? 00:10:56.910 --> 00:10:59.590 align:middle line:90% And a lot of that is opaque. 00:10:59.590 --> 00:11:01.750 align:middle line:90% They don't really issue numbers. 00:11:01.750 --> 00:11:04.430 align:middle line:84% They don't tell you what goes into that number. 00:11:04.430 --> 00:11:08.150 align:middle line:84% So I will give you both the dollar estimates, 00:11:08.150 --> 00:11:10.070 align:middle line:84% and I will give you duration estimates 00:11:10.070 --> 00:11:14.030 align:middle line:84% so that you can have some better apples to apples comparison. 00:11:14.030 --> 00:11:14.530 align:middle line:90% All right. 00:11:14.530 --> 00:11:16.070 align:middle line:84% So let's just look around and see 00:11:16.070 --> 00:11:18.110 align:middle line:90% what other countries have done. 00:11:18.110 --> 00:11:22.070 align:middle line:84% I think the country that everyone learns about first 00:11:22.070 --> 00:11:25.236 align:middle line:90% as being a success is 00:11:25.236 --> 00:11:26.040 align:middle line:90% AUDIENCE: France. 00:11:26.040 --> 00:11:26.790 align:middle line:90% PROFESSOR: France. 00:11:26.790 --> 00:11:28.430 align:middle line:90% Yeah. 00:11:28.430 --> 00:11:28.930 align:middle line:90% All right. 00:11:28.930 --> 00:11:31.350 align:middle line:90% So let's look at France. 00:11:31.350 --> 00:11:31.930 align:middle line:90% All right. 00:11:31.930 --> 00:11:37.590 align:middle line:84% So I'm misbehaving a little bit by putting the DOE cost 00:11:37.590 --> 00:11:38.890 align:middle line:90% prediction on this plot. 00:11:38.890 --> 00:11:42.780 align:middle line:84% Because French numbers are not done 00:11:42.780 --> 00:11:45.520 align:middle line:84% under the same kind of accounting as American numbers. 00:11:45.520 --> 00:11:48.480 align:middle line:90% They have a state-owned utility. 00:11:48.480 --> 00:11:50.480 align:middle line:84% They have a state-owned reactor company. 00:11:50.480 --> 00:11:51.600 align:middle line:90% They have a lot of sites. 00:11:51.600 --> 00:11:53.180 align:middle line:84% They recycle their reuse of land. 00:11:53.180 --> 00:11:55.580 align:middle line:84% Like, they're not apples to apples, 00:11:55.580 --> 00:11:58.980 align:middle line:84% but it gives you some rough estimate. 00:11:58.980 --> 00:12:03.340 align:middle line:84% So the officially reported cost data for France stayed low 00:12:03.340 --> 00:12:06.100 align:middle line:84% for quite a long time in the early fleet, 00:12:06.100 --> 00:12:11.580 align:middle line:90% up until maybe the '90s. 00:12:11.580 --> 00:12:13.640 align:middle line:84% And then between the '90s and the 2000s, 00:12:13.640 --> 00:12:15.680 align:middle line:90% there was a little uptick. 00:12:15.680 --> 00:12:18.260 align:middle line:84% And part of the problem with having that data 00:12:18.260 --> 00:12:20.320 align:middle line:84% point on the plot is it compresses this. 00:12:20.320 --> 00:12:25.300 align:middle line:84% It's hard to see, but it's basically a 50% cost increase 00:12:25.300 --> 00:12:27.820 align:middle line:84% between the '90s and the 2000s in France. 00:12:27.820 --> 00:12:32.960 align:middle line:84% Per unit, kilowatt hour, and inflation adjusted, right? 00:12:32.960 --> 00:12:37.020 align:middle line:84% These are all overnight costs that are inflation adjusted. 00:12:37.020 --> 00:12:39.020 align:middle line:90% So something is going on. 00:12:39.020 --> 00:12:41.900 align:middle line:84% And the most recent plant, Flamanville, 00:12:41.900 --> 00:12:44.900 align:middle line:90% is just outrageously expensive. 00:12:44.900 --> 00:12:47.180 align:middle line:84% Five times more expensive than anything 00:12:47.180 --> 00:12:50.980 align:middle line:84% the French have built previously. 00:12:50.980 --> 00:12:54.020 align:middle line:84% And twice as expensive as the DOE estimate. 00:12:54.020 --> 00:12:58.140 align:middle line:84% And part of this is that Flamanville is a first-of-a-kind 00:12:58.140 --> 00:13:02.500 align:middle line:84% reactor that was never be built. It's a gen III reactor, 00:13:02.500 --> 00:13:06.800 align:middle line:84% AP1000 Gen III+ reactor with a lot of extra safety features. 00:13:06.800 --> 00:13:11.380 align:middle line:84% And there was a lot of challenges 00:13:11.380 --> 00:13:15.540 align:middle line:84% in getting that reactor to be built directly. 00:13:15.540 --> 00:13:20.740 align:middle line:84% Giant lawsuits, issues related to the casting of the head, 00:13:20.740 --> 00:13:21.880 align:middle line:90% the pressure vessel. 00:13:21.880 --> 00:13:26.180 align:middle line:84% It had carbon precipitates in it that may limit its life. 00:13:26.180 --> 00:13:27.820 align:middle line:90% It was a big mess. 00:13:27.820 --> 00:13:31.580 align:middle line:84% They also had the first, I think, entirely virtual control 00:13:31.580 --> 00:13:35.860 align:middle line:84% system, and it took a long time to certify all the software. 00:13:35.860 --> 00:13:37.260 align:middle line:84% And so there were a lot of issues 00:13:37.260 --> 00:13:39.540 align:middle line:90% associated with that reactor. 00:13:39.540 --> 00:13:40.500 align:middle line:90% All right. 00:13:40.500 --> 00:13:43.360 align:middle line:84% The more interesting question is, why were these passed? 00:13:43.360 --> 00:13:46.980 align:middle line:84% Why were the earlier reactors cheap? 00:13:46.980 --> 00:13:48.900 align:middle line:84% In part, it seems to have something 00:13:48.900 --> 00:13:51.580 align:middle line:90% to do with standardization. 00:13:51.580 --> 00:13:55.160 align:middle line:84% There's more standardization in France than there is in the US. 00:13:55.160 --> 00:13:58.160 align:middle line:84% We have [INAUDIBLE], which is running at full copies. 00:13:58.160 --> 00:14:00.040 align:middle line:90% So they kind of knew what to do. 00:14:00.040 --> 00:14:03.580 align:middle line:84% And the US, even though we had a notional standard design, 00:14:03.580 --> 00:14:05.700 align:middle line:84% like the Westinghouse four-loop, every time 00:14:05.700 --> 00:14:09.100 align:middle line:84% it got built, it was different because things 00:14:09.100 --> 00:14:11.980 align:middle line:84% had to be adjusted for the site, and the owner wanted 00:14:11.980 --> 00:14:14.540 align:middle line:84% to connect it with a different turbine 00:14:14.540 --> 00:14:16.160 align:middle line:84% than the one that was designed for. 00:14:16.160 --> 00:14:17.940 align:middle line:90% And all of these other issues. 00:14:17.940 --> 00:14:23.740 align:middle line:84% So they have some advantages on that front. 00:14:23.740 --> 00:14:27.180 align:middle line:84% And if you look at the engineering cost 00:14:27.180 --> 00:14:30.820 align:middle line:84% for mapping out where everything goes, where all the pipes go, 00:14:30.820 --> 00:14:34.780 align:middle line:84% before the reactor gets built, France's average engineering 00:14:34.780 --> 00:14:37.730 align:middle line:84% cost is about 15% of the total cost. 00:14:37.730 --> 00:14:41.270 align:middle line:84% The US average engineering cost is about 25% of the total cost. 00:14:41.270 --> 00:14:44.490 align:middle line:84% So you can see there's a little bit of cost savings right there. 00:14:44.490 --> 00:14:49.570 align:middle line:84% 10% just from this standardization. 00:14:49.570 --> 00:14:55.750 align:middle line:84% So given the wonderful prices for these reactors, which, 00:14:55.750 --> 00:14:58.860 align:middle line:84% again, are not really apples to apples comparisons to what 00:14:58.860 --> 00:15:04.050 align:middle line:84% the US has, we might ask, what is the cost of electricity 00:15:04.050 --> 00:15:06.490 align:middle line:90% in France? 00:15:06.490 --> 00:15:10.930 align:middle line:84% According to the French government, 00:15:10.930 --> 00:15:13.850 align:middle line:84% the price is about $70 per megawatt, 00:15:13.850 --> 00:15:15.310 align:middle line:90% if we convert it to dollars. 00:15:15.310 --> 00:15:17.650 align:middle line:90% So about $70 per megawatt hour. 00:15:17.650 --> 00:15:20.290 align:middle line:84% Which puts the price of nuclear in France 00:15:20.290 --> 00:15:27.130 align:middle line:84% in between the price of, say, wind and offshore wind, if you 00:15:27.130 --> 00:15:29.790 align:middle line:90% remember those things, right? 00:15:29.790 --> 00:15:33.510 align:middle line:84% So it's still more expensive than wind, regular wind, 00:15:33.510 --> 00:15:35.290 align:middle line:90% onshore wind and solar. 00:15:35.290 --> 00:15:38.970 align:middle line:84% But it's not the most expensive thing. 00:15:38.970 --> 00:15:40.730 align:middle line:90% And it is dispatchable. 00:15:40.730 --> 00:15:45.370 align:middle line:84% So it fits into that role where you 00:15:45.370 --> 00:15:48.970 align:middle line:84% have some kind of low-carbon dispatchable power 00:15:48.970 --> 00:15:52.850 align:middle line:84% source that is useful for setting of variability. 00:15:52.850 --> 00:15:54.930 align:middle line:84% So I think it would still make sense in France, 00:15:54.930 --> 00:15:56.930 align:middle line:90% if those numbers are true. 00:15:56.930 --> 00:15:58.630 align:middle line:84% The construction durations, however, 00:15:58.630 --> 00:16:01.810 align:middle line:84% tell a little bit of a different story. 00:16:01.810 --> 00:16:04.890 align:middle line:84% If France was building all these reactors, 00:16:04.890 --> 00:16:06.930 align:middle line:84% and they were all cheap, and they 00:16:06.930 --> 00:16:09.810 align:middle line:84% were learning on how to build them cheaply, 00:16:09.810 --> 00:16:13.010 align:middle line:84% why did it take more and more and more and more 00:16:13.010 --> 00:16:15.590 align:middle line:84% time to build each reactor as it went along? 00:16:15.590 --> 00:16:19.650 align:middle line:90% 00:16:19.650 --> 00:16:24.570 align:middle line:84% So far, I don't have a really convincing answer 00:16:24.570 --> 00:16:26.490 align:middle line:90% to the details. 00:16:26.490 --> 00:16:29.830 align:middle line:84% But there is a correlation that I want to talk about. 00:16:29.830 --> 00:16:32.050 align:middle line:90% I can't prove this is the cause. 00:16:32.050 --> 00:16:36.200 align:middle line:84% And the correlation is that it turns out 00:16:36.200 --> 00:16:39.960 align:middle line:84% France's reactors are not very reliable. 00:16:39.960 --> 00:16:43.120 align:middle line:84% So at any given moment, about one in four 00:16:43.120 --> 00:16:45.680 align:middle line:84% [INAUDIBLE] reactors is down for repair. 00:16:45.680 --> 00:16:52.720 align:middle line:84% So they have about a 74.5% uptime availability-- sorry. 00:16:52.720 --> 00:16:57.020 align:middle line:84% This is probably where the noise is coming up. 00:16:57.020 --> 00:17:03.280 align:middle line:84% 74.5% availability over their fleet. 00:17:03.280 --> 00:17:06.760 align:middle line:90% Historically up until 2024. 00:17:06.760 --> 00:17:09.000 align:middle line:90% So that's pretty low. 00:17:09.000 --> 00:17:11.720 align:middle line:84% The US availability is about 99%. 00:17:11.720 --> 00:17:16.000 align:middle line:90% And so what is causing that? 00:17:16.000 --> 00:17:19.599 align:middle line:84% Well, they have all these unplanned outages. 00:17:19.599 --> 00:17:24.040 align:middle line:84% And they've had some major issues, like stress corrosion 00:17:24.040 --> 00:17:26.800 align:middle line:84% cracking, but they just have other maintenance stuff 00:17:26.800 --> 00:17:29.680 align:middle line:84% that they do that we don't have to do. 00:17:29.680 --> 00:17:33.440 align:middle line:84% And so the kind of intuitive takeaway 00:17:33.440 --> 00:17:36.340 align:middle line:84% is that as they've built these reactors, 00:17:36.340 --> 00:17:39.240 align:middle line:84% they've realized they need to build them to higher standards. 00:17:39.240 --> 00:17:41.360 align:middle line:84% And they were cheaper, but that's 00:17:41.360 --> 00:17:43.240 align:middle line:84% why they kind of took longer and longer 00:17:43.240 --> 00:17:46.200 align:middle line:84% is they're trying to get that reliability up there. 00:17:46.200 --> 00:17:53.240 align:middle line:84% So to put it another way, you pay either upfront 00:17:53.240 --> 00:17:55.660 align:middle line:84% or you pay later to do the maintenance. 00:17:55.660 --> 00:17:59.440 align:middle line:84% And when you pay later, you have to pay that capital charge 00:17:59.440 --> 00:18:02.560 align:middle line:84% on the reactor every year, even though the reactor is not 00:18:02.560 --> 00:18:03.700 align:middle line:90% producing electricity. 00:18:03.700 --> 00:18:08.305 align:middle line:84% And so that drives up the cost of electricity, 00:18:08.305 --> 00:18:09.680 align:middle line:84% the effective cost of electricity 00:18:09.680 --> 00:18:11.760 align:middle line:90% on the fleet-wide average. 00:18:11.760 --> 00:18:15.500 align:middle line:84% The other problem with this strategy of build it cheap, 00:18:15.500 --> 00:18:19.880 align:middle line:84% fix it later is every time something 00:18:19.880 --> 00:18:24.240 align:middle line:84% fails, you have a potential safety-related issue, right? 00:18:24.240 --> 00:18:28.960 align:middle line:84% It becomes a precipitating event for a scram or some kind 00:18:28.960 --> 00:18:31.650 align:middle line:84% of valve failure or something or other. 00:18:31.650 --> 00:18:34.150 align:middle line:84% There isn't a strict one-to-one correspondence 00:18:34.150 --> 00:18:38.790 align:middle line:84% between unplanned outages and safety issues, 00:18:38.790 --> 00:18:42.550 align:middle line:84% but you can see that there would be some kind of correlation 00:18:42.550 --> 00:18:44.310 align:middle line:90% between those two factors. 00:18:44.310 --> 00:18:48.210 align:middle line:84% So arguably, France's fleet, while cheap, 00:18:48.210 --> 00:18:52.710 align:middle line:90% is maybe not super safe. 00:18:52.710 --> 00:18:55.230 align:middle line:84% And it turns out this kind of correlation between safety 00:18:55.230 --> 00:18:57.550 align:middle line:84% and cost actually shows up over and over, as you're 00:18:57.550 --> 00:18:59.110 align:middle line:90% going to see in a moment. 00:18:59.110 --> 00:19:04.630 align:middle line:84% So another big success story is Japan. 00:19:04.630 --> 00:19:08.990 align:middle line:84% Cost and duration in Japan were exceedingly well controlled. 00:19:08.990 --> 00:19:11.490 align:middle line:84% They were building the reactors really, really fast. 00:19:11.490 --> 00:19:14.430 align:middle line:84% On average, they didn't have any increase 00:19:14.430 --> 00:19:19.190 align:middle line:84% in the duration of construction, and they were about four years 00:19:19.190 --> 00:19:22.790 align:middle line:84% to build a reactor, compared to our predicted number 00:19:22.790 --> 00:19:26.550 align:middle line:84% of around six years and actual number of around 10 years. 00:19:26.550 --> 00:19:28.570 align:middle line:84% And we can just go look at the French numbers. 00:19:28.570 --> 00:19:29.470 align:middle line:90% Just go back. 00:19:29.470 --> 00:19:31.750 align:middle line:84% You see, they were like six in the beginning 00:19:31.750 --> 00:19:34.590 align:middle line:84% and they went up to 10 to 12 later on, which is 00:19:34.590 --> 00:19:37.550 align:middle line:90% kind of what we see in the US. 00:19:37.550 --> 00:19:41.230 align:middle line:84% So how was it that Japan was able to keep 00:19:41.230 --> 00:19:44.870 align:middle line:90% these numbers so low? 00:19:44.870 --> 00:19:47.870 align:middle line:84% And you'll see that all these reactors are just 00:19:47.870 --> 00:19:50.990 align:middle line:84% being constantly built. Prices were not nearly as 00:19:50.990 --> 00:19:52.490 align:middle line:90% low as the French reactors. 00:19:52.490 --> 00:19:57.710 align:middle line:84% They're closer to what the MIT numbers were based on or closer 00:19:57.710 --> 00:19:58.840 align:middle line:90% to the MIT predictions. 00:19:58.840 --> 00:20:00.590 align:middle line:84% And the reason is that this is the numbers 00:20:00.590 --> 00:20:03.190 align:middle line:90% they were based on initially. 00:20:03.190 --> 00:20:06.350 align:middle line:84% And then you see, of course, they stopped building reactors 00:20:06.350 --> 00:20:08.090 align:middle line:90% after the Fukushima incident. 00:20:08.090 --> 00:20:11.350 align:middle line:90% 00:20:11.350 --> 00:20:14.590 align:middle line:84% Another interesting just upfront observation 00:20:14.590 --> 00:20:16.270 align:middle line:84% is most of these reactors in Japan 00:20:16.270 --> 00:20:19.310 align:middle line:84% were actually US designs being just 00:20:19.310 --> 00:20:20.890 align:middle line:90% being built by the Japanese. 00:20:20.890 --> 00:20:24.010 align:middle line:84% So a lot of people in the United States were saying, well, 00:20:24.010 --> 00:20:27.900 align:middle line:84% these are US reactors that we can build here. 00:20:27.900 --> 00:20:30.820 align:middle line:84% So presumably we can build them at the same speed 00:20:30.820 --> 00:20:32.120 align:middle line:90% and at the same cost. 00:20:32.120 --> 00:20:33.700 align:middle line:84% And that was the assumption that went 00:20:33.700 --> 00:20:41.100 align:middle line:84% into a lot of the estimates, like these estimates here. 00:20:41.100 --> 00:20:41.920 align:middle line:90% All right. 00:20:41.920 --> 00:20:47.460 align:middle line:90% 00:20:47.460 --> 00:20:49.180 align:middle line:90% What actually happened? 00:20:49.180 --> 00:20:53.460 align:middle line:84% Well, it turns out they also had major safety issues. 00:20:53.460 --> 00:20:57.840 align:middle line:84% So I guess I can read it to you or you can read it. 00:20:57.840 --> 00:21:00.673 align:middle line:84% Maybe you just read it and raise your hand 00:21:00.673 --> 00:21:01.840 align:middle line:90% when you're done reading it. 00:21:01.840 --> 00:21:06.140 align:middle line:90% 00:21:06.140 --> 00:21:08.860 align:middle line:90% Good? 00:21:08.860 --> 00:21:09.440 align:middle line:90% Yeah. 00:21:09.440 --> 00:21:13.580 align:middle line:84% So it turns out that basically the regulator 00:21:13.580 --> 00:21:22.580 align:middle line:84% was completely neutered and it had no power in Japan. 00:21:22.580 --> 00:21:29.780 align:middle line:84% And this prevents all kinds of delays during construction. 00:21:29.780 --> 00:21:32.700 align:middle line:84% You just build a reactor and you just get what you get, 00:21:32.700 --> 00:21:37.100 align:middle line:90% and it's fine. 00:21:37.100 --> 00:21:39.700 align:middle line:84% Even with all of this, it's interesting to note 00:21:39.700 --> 00:21:42.660 align:middle line:84% that TEPCO, the company that built most of these 00:21:42.660 --> 00:21:45.660 align:middle line:84% and owned the Fukushima plant, still 00:21:45.660 --> 00:21:51.192 align:middle line:84% kept losing money because the plants were just not profitable 00:21:51.192 --> 00:21:53.400 align:middle line:84% with all the maintenance that they were having to do. 00:21:53.400 --> 00:21:56.980 align:middle line:84% Not surprising because they were so shoddily built, 00:21:56.980 --> 00:22:00.500 align:middle line:84% and that was leading to this downward spiral 00:22:00.500 --> 00:22:05.980 align:middle line:84% where they would try to cut further cost to try to make 00:22:05.980 --> 00:22:07.600 align:middle line:90% the company profitable again. 00:22:07.600 --> 00:22:11.420 align:middle line:90% 00:22:11.420 --> 00:22:13.700 align:middle line:84% And just as an example of that, when the Fukushima 00:22:13.700 --> 00:22:18.700 align:middle line:84% accident occurred, TEPCO, the company who owned the plant, 00:22:18.700 --> 00:22:23.100 align:middle line:84% did not even have a diagram of the internal pipework 00:22:23.100 --> 00:22:27.250 align:middle line:84% of the plant for the emergency responders. 00:22:27.250 --> 00:22:32.190 align:middle line:84% They had an original draft of what they thought it might be, 00:22:32.190 --> 00:22:34.970 align:middle line:84% but no one had actually kept track of what was actually 00:22:34.970 --> 00:22:36.290 align:middle line:90% built in the plant. 00:22:36.290 --> 00:22:40.210 align:middle line:84% It was really cutting every possible corner. 00:22:40.210 --> 00:22:44.210 align:middle line:84% And so it's true that Japan's reactor fleet 00:22:44.210 --> 00:22:48.770 align:middle line:84% is kind of an economic miracle, but maybe it's 00:22:48.770 --> 00:22:52.330 align:middle line:90% just too much miracle. 00:22:52.330 --> 00:22:56.490 align:middle line:84% And it turns out there's no free lunch. 00:22:56.490 --> 00:23:01.290 align:middle line:84% They didn't just get out of jail with one Fukushima accident 00:23:01.290 --> 00:23:02.770 align:middle line:90% card. 00:23:02.770 --> 00:23:04.890 align:middle line:90% They continued to pay the price. 00:23:04.890 --> 00:23:09.330 align:middle line:84% So we are now 14 years after the Fukushima accident, 00:23:09.330 --> 00:23:14.130 align:middle line:84% and most of the reactors are still shut down. 00:23:14.130 --> 00:23:18.970 align:middle line:84% And they are paying for all of these reactors, 00:23:18.970 --> 00:23:20.770 align:middle line:84% and they're not producing electricity. 00:23:20.770 --> 00:23:22.170 align:middle line:84% And why are they still shut down, 00:23:22.170 --> 00:23:24.370 align:middle line:84% is that they are working on fixing the safety 00:23:24.370 --> 00:23:28.690 align:middle line:84% problems associated that they've discovered 00:23:28.690 --> 00:23:32.430 align:middle line:84% after these reactors were shut down. 00:23:32.430 --> 00:23:35.632 align:middle line:84% So when you factor in all of this shutdown time, 00:23:35.632 --> 00:23:37.590 align:middle line:84% even the ones that are operational, by the way, 00:23:37.590 --> 00:23:40.590 align:middle line:84% are only operating about 60% of the time. 00:23:40.590 --> 00:23:42.650 align:middle line:90% They're that unreliable. 00:23:42.650 --> 00:23:48.450 align:middle line:84% So when you factor all of this in, their price per kilowatt 00:23:48.450 --> 00:23:52.410 align:middle line:84% of capacity, effective price per kilowatt of capacity, 00:23:52.410 --> 00:23:57.010 align:middle line:84% it's not $5,000 per kilowatt, or $7,000 per kilowatt, 00:23:57.010 --> 00:24:00.330 align:middle line:84% or $19,000 per kilowatt, which is what it is for Vogtle. 00:24:00.330 --> 00:24:04.450 align:middle line:90% It's $44,000 per kilowatt. 00:24:04.450 --> 00:24:08.210 align:middle line:84% So it's really easy to look at this historical data 00:24:08.210 --> 00:24:10.370 align:middle line:84% and say, look, we could do what Japan did. 00:24:10.370 --> 00:24:12.370 align:middle line:84% But unless you look at the whole picture 00:24:12.370 --> 00:24:14.810 align:middle line:84% about their availability, the safety, maintenance issues 00:24:14.810 --> 00:24:17.770 align:middle line:84% and so on, you don't get an understanding 00:24:17.770 --> 00:24:19.210 align:middle line:90% what really happened. 00:24:19.210 --> 00:24:22.080 align:middle line:84% And now they have probably some of the most expensive 00:24:22.080 --> 00:24:26.280 align:middle line:84% nuclear power on the planet, in real terms. 00:24:26.280 --> 00:24:27.940 align:middle line:84% So it did not save money in the end. 00:24:27.940 --> 00:24:29.960 align:middle line:84% It actually cost them a lot of money. 00:24:29.960 --> 00:24:32.240 align:middle line:84% And that just goes to say that building fast and cheap 00:24:32.240 --> 00:24:37.120 align:middle line:84% probably isn't the way to do this. 00:24:37.120 --> 00:24:40.480 align:middle line:84% There was a similar trend happening in Korea. 00:24:40.480 --> 00:24:44.760 align:middle line:84% Total investment costs and overnight costs in Korea 00:24:44.760 --> 00:24:47.760 align:middle line:84% were trending downward for many years. 00:24:47.760 --> 00:24:53.920 align:middle line:84% Even construction durations were going very slightly downwards. 00:24:53.920 --> 00:24:58.820 align:middle line:84% And then there was a turnaround, as you can see here, 00:24:58.820 --> 00:25:03.200 align:middle line:84% right here at the Fukushima accident, things turned around. 00:25:03.200 --> 00:25:05.120 align:middle line:90% And I had a conversation. 00:25:05.120 --> 00:25:09.540 align:middle line:84% The former chairman of Korea's version of the NRC, 00:25:09.540 --> 00:25:12.600 align:middle line:84% their safety regulator, is a friend of mine. 00:25:12.600 --> 00:25:18.040 align:middle line:84% And he was telling me that they had all these unsafe labor 00:25:18.040 --> 00:25:23.840 align:middle line:84% standards for workers building these reactors. 00:25:23.840 --> 00:25:25.540 align:middle line:90% So let's see. 00:25:25.540 --> 00:25:28.040 align:middle line:90% Do I have some number here? 00:25:28.040 --> 00:25:34.840 align:middle line:84% So it turns out, they passed a law, 00:25:34.840 --> 00:25:39.240 align:middle line:84% after somewhere around here, that said that workers cannot 00:25:39.240 --> 00:25:43.320 align:middle line:84% work for more than 56 hours a week on the reactor. 00:25:43.320 --> 00:25:45.680 align:middle line:84% So before that, their construction workers were 00:25:45.680 --> 00:25:47.840 align:middle line:90% working 80-hour weeks. 00:25:47.840 --> 00:25:51.080 align:middle line:84% Can you imagine being a construction worker working 00:25:51.080 --> 00:25:54.280 align:middle line:84% an 80-hour week, how exhausted you would be? 00:25:54.280 --> 00:25:58.040 align:middle line:84% How prone to making mistakes you might be? 00:25:58.040 --> 00:26:03.640 align:middle line:84% And this is how they were making this happen. 00:26:03.640 --> 00:26:05.260 align:middle line:84% There were also a bunch of scandals. 00:26:05.260 --> 00:26:15.000 align:middle line:84% So in 2003, it was revealed that some 8,000 parts in at least six 00:26:15.000 --> 00:26:19.080 align:middle line:84% different reactors supplied between 2002 and 2013, 00:26:19.080 --> 00:26:21.760 align:middle line:84% which come with safety certificates, 00:26:21.760 --> 00:26:24.400 align:middle line:84% in this country we call them the N-Stamp, 00:26:24.400 --> 00:26:26.660 align:middle line:84% they were completely falsified safety certificates. 00:26:26.660 --> 00:26:29.040 align:middle line:90% They were just fake. 00:26:29.040 --> 00:26:30.600 align:middle line:84% So all of those parts had to go back 00:26:30.600 --> 00:26:34.100 align:middle line:90% and be inspected and checked. 00:26:34.100 --> 00:26:35.480 align:middle line:84% A bunch of reactors got shut down 00:26:35.480 --> 00:26:38.920 align:middle line:84% and some people went to jail because of this. 00:26:38.920 --> 00:26:46.320 align:middle line:84% So now because of these fake parts, 00:26:46.320 --> 00:26:49.880 align:middle line:84% the reactors are undergoing retrofits which cost roughly 7% 00:26:49.880 --> 00:26:53.080 align:middle line:84% of the total construction cost of a reactor 00:26:53.080 --> 00:26:55.840 align:middle line:90% every five years or so. 00:26:55.840 --> 00:26:59.160 align:middle line:84% So they have a very chunky maintenance bill 00:26:59.160 --> 00:27:02.720 align:middle line:84% on top of their operating costs because of this. 00:27:02.720 --> 00:27:05.500 align:middle line:84% And they also have very low reliability. 00:27:05.500 --> 00:27:07.840 align:middle line:84% About 23% of the time their reactors 00:27:07.840 --> 00:27:10.780 align:middle line:84% are being shut down for maintenance and service. 00:27:10.780 --> 00:27:15.320 align:middle line:84% Again, in the US, only 1% of the time. 00:27:15.320 --> 00:27:19.190 align:middle line:84% So then there were safety standard increases 00:27:19.190 --> 00:27:22.630 align:middle line:84% for the designs after both the Fukushima and Gyeongju 00:27:22.630 --> 00:27:26.050 align:middle line:84% earthquakes that they implemented. 00:27:26.050 --> 00:27:29.350 align:middle line:84% And all of this has basically driven the price up 00:27:29.350 --> 00:27:33.470 align:middle line:84% and the construction up, construction duration up. 00:27:33.470 --> 00:27:36.930 align:middle line:84% They've stopped publishing the prices for their reactors. 00:27:36.930 --> 00:27:39.330 align:middle line:84% So we don't know how much they actually cost. 00:27:39.330 --> 00:27:41.710 align:middle line:84% But we do know when the reactor turns on. 00:27:41.710 --> 00:27:47.950 align:middle line:84% So you can see there's more data here in the duration time. 00:27:47.950 --> 00:27:50.110 align:middle line:84% And what you see is essentially, Korea 00:27:50.110 --> 00:27:53.390 align:middle line:84% went from being something similar to Japan, 00:27:53.390 --> 00:27:55.910 align:middle line:84% of an economic miracle, to having 00:27:55.910 --> 00:27:58.270 align:middle line:84% construction durations of about 10 years, just 00:27:58.270 --> 00:27:59.830 align:middle line:90% like the US and the French. 00:27:59.830 --> 00:28:02.710 align:middle line:84% And their costs are turning upwards. 00:28:02.710 --> 00:28:06.470 align:middle line:84% And we'll see where they come out, 00:28:06.470 --> 00:28:09.750 align:middle line:84% if they ever publish their data again. 00:28:09.750 --> 00:28:15.910 align:middle line:84% Korea does publish predicted pre-construction costs, still, 00:28:15.910 --> 00:28:18.470 align:middle line:90% for new reactor projects. 00:28:18.470 --> 00:28:22.910 align:middle line:84% And their current predicted price is $9 billion a reactor, 00:28:22.910 --> 00:28:25.970 align:middle line:84% which is slightly above what the US 00:28:25.970 --> 00:28:29.070 align:middle line:90% DOE is predicting per reactor. 00:28:29.070 --> 00:28:33.890 align:middle line:84% So they've learned that essentially they cannot do it 00:28:33.890 --> 00:28:35.270 align:middle line:90% so cheaply. 00:28:35.270 --> 00:28:39.990 align:middle line:84% Basically, the Korean industry has turned into the US industry. 00:28:39.990 --> 00:28:40.830 align:middle line:90% All right. 00:28:40.830 --> 00:28:45.110 align:middle line:84% India is a very interesting and unusual country. 00:28:45.110 --> 00:28:48.590 align:middle line:84% The cost of reactors in India has been systematically coming 00:28:48.590 --> 00:28:51.150 align:middle line:90% down. 00:28:51.150 --> 00:28:54.510 align:middle line:84% And we will say, what is going on there? 00:28:54.510 --> 00:28:57.270 align:middle line:84% Is this evidence that India is the only country that 00:28:57.270 --> 00:29:00.270 align:middle line:84% can actually figure out how to learn? 00:29:00.270 --> 00:29:02.510 align:middle line:84% We'll talk about that in a moment. 00:29:02.510 --> 00:29:04.970 align:middle line:84% However, construction durations are going up. 00:29:04.970 --> 00:29:09.150 align:middle line:84% And this is, if they were really learning, 00:29:09.150 --> 00:29:11.270 align:middle line:90% why do the durations go up? 00:29:11.270 --> 00:29:14.020 align:middle line:84% Also, you'll notice that they start at around eight 00:29:14.020 --> 00:29:16.720 align:middle line:90% and they go up towards 14 years. 00:29:16.720 --> 00:29:19.460 align:middle line:84% They actually take quite a long time to build. 00:29:19.460 --> 00:29:22.040 align:middle line:84% So India is a bit of a strange case. 00:29:22.040 --> 00:29:24.860 align:middle line:90% 00:29:24.860 --> 00:29:29.360 align:middle line:84% The nuclear power in India is financed only partly by loans, 00:29:29.360 --> 00:29:32.340 align:middle line:84% and until recently, at least half of the cost of the reactor 00:29:32.340 --> 00:29:36.700 align:middle line:84% was just paid every year out of the central government's budget. 00:29:36.700 --> 00:29:38.360 align:middle line:84% So they don't have to borrow money. 00:29:38.360 --> 00:29:40.375 align:middle line:84% They don't have to worry about finance charges. 00:29:40.375 --> 00:29:42.000 align:middle line:84% And one of the effects of that, though, 00:29:42.000 --> 00:29:44.540 align:middle line:84% is that they have a certain budget of what they can spend 00:29:44.540 --> 00:29:46.580 align:middle line:84% this year, and that's all the construction 00:29:46.580 --> 00:29:48.300 align:middle line:90% they can afford to do. 00:29:48.300 --> 00:29:51.940 align:middle line:84% And so the construction duration is 00:29:51.940 --> 00:29:57.300 align:middle line:84% going to go very methodically and plod along slowly, as 00:29:57.300 --> 00:29:59.612 align:middle line:84% opposed to in the US, where people are taking loans 00:29:59.612 --> 00:30:01.320 align:middle line:84% for billions of dollars, paying interest, 00:30:01.320 --> 00:30:03.940 align:middle line:84% and they better get that damn plant built as fast as possible. 00:30:03.940 --> 00:30:08.500 align:middle line:84% So it's a different dynamic, and that probably results 00:30:08.500 --> 00:30:14.240 align:middle line:84% in basically fewer errors during construction and better safety. 00:30:14.240 --> 00:30:17.580 align:middle line:84% And indeed, India's fleet is one of the most reliable 00:30:17.580 --> 00:30:18.580 align:middle line:90% in the world. 00:30:18.580 --> 00:30:21.200 align:middle line:84% They're down only 4.4% of the time, 00:30:21.200 --> 00:30:25.280 align:middle line:84% compared to the 23%, 25% that you see in other countries. 00:30:25.280 --> 00:30:27.780 align:middle line:90% 40% for Japan now. 00:30:27.780 --> 00:30:32.860 align:middle line:84% So this slower method of construction 00:30:32.860 --> 00:30:37.100 align:middle line:84% appears anecdotally to be related 00:30:37.100 --> 00:30:43.780 align:middle line:84% to this high availability, high safety record 00:30:43.780 --> 00:30:46.820 align:middle line:90% that India has at the moment. 00:30:46.820 --> 00:30:50.040 align:middle line:84% So just if we step back and look at all the countries, 00:30:50.040 --> 00:30:55.020 align:middle line:84% we see that all the major nuclear countries-- 00:30:55.020 --> 00:30:59.500 align:middle line:84% USA, France, Japan-- costs have been going up. 00:30:59.500 --> 00:31:02.220 align:middle line:84% India, the costs have been coming down. 00:31:02.220 --> 00:31:06.940 align:middle line:84% Korea had costs going down for a while, but they've turned 00:31:06.940 --> 00:31:11.470 align:middle line:84% and their new predicted costs are way up here. $9 billion 00:31:11.470 --> 00:31:12.610 align:middle line:90% a reactor. 00:31:12.610 --> 00:31:15.890 align:middle line:90% 00:31:15.890 --> 00:31:17.810 align:middle line:84% There are still residual differences 00:31:17.810 --> 00:31:22.370 align:middle line:84% between these countries that are not 00:31:22.370 --> 00:31:24.270 align:middle line:84% revealed in the conversation we just had. 00:31:24.270 --> 00:31:26.850 align:middle line:84% For example, labor productivity in Asia 00:31:26.850 --> 00:31:29.850 align:middle line:84% is much higher than it is in the US. 00:31:29.850 --> 00:31:34.450 align:middle line:84% There, employees don't have as many safety standards. 00:31:34.450 --> 00:31:38.090 align:middle line:90% They get more done, basically. 00:31:38.090 --> 00:31:39.950 align:middle line:84% You can look at that in two ways. 00:31:39.950 --> 00:31:43.850 align:middle line:84% You can say, oh, they're more efficient 00:31:43.850 --> 00:31:45.230 align:middle line:90% and we should do the same. 00:31:45.230 --> 00:31:47.930 align:middle line:84% Or you could say, well, they're exploiting their workers 00:31:47.930 --> 00:31:49.630 align:middle line:90% and we would never do that. 00:31:49.630 --> 00:31:52.570 align:middle line:84% And so this is not something that we can replicate. 00:31:52.570 --> 00:31:54.010 align:middle line:90% And that's up to you. 00:31:54.010 --> 00:31:56.650 align:middle line:84% But that's one of the reasons why labor costs in the US 00:31:56.650 --> 00:31:59.890 align:middle line:90% are as high as they are. 00:31:59.890 --> 00:32:01.530 align:middle line:84% One insight that appears to emerge 00:32:01.530 --> 00:32:05.230 align:middle line:84% is this correspondence between, as I've mentioned, 00:32:05.230 --> 00:32:06.670 align:middle line:90% nuclear costs and availability. 00:32:06.670 --> 00:32:09.170 align:middle line:84% Cheaper programs have more outages 00:32:09.170 --> 00:32:11.090 align:middle line:84% in the long run, more failures, and that 00:32:11.090 --> 00:32:12.787 align:middle line:90% implies more safety risk. 00:32:12.787 --> 00:32:14.370 align:middle line:84% And it's very visible in the countries 00:32:14.370 --> 00:32:17.770 align:middle line:84% that had scandals, like Korea and Japan, 00:32:17.770 --> 00:32:19.290 align:middle line:90% where people went to jail. 00:32:19.290 --> 00:32:22.410 align:middle line:84% We also saw a correlation between reactor generation 00:32:22.410 --> 00:32:24.330 align:middle line:90% and costs. 00:32:24.330 --> 00:32:27.600 align:middle line:84% Both Vogtle, which is the AP1000-- 00:32:27.600 --> 00:32:28.368 align:middle line:90% yeah. 00:32:28.368 --> 00:32:29.410 align:middle line:90% AUDIENCE: You can finish. 00:32:29.410 --> 00:32:30.827 align:middle line:84% PROFESSOR: OK, let me finish this. 00:32:30.827 --> 00:32:32.930 align:middle line:84% Both Vogtle, which is the AP1000, 00:32:32.930 --> 00:32:36.230 align:middle line:84% which is the Gen III+ reactor, and EPR2, 00:32:36.230 --> 00:32:41.110 align:middle line:84% which is the French reactor, are both new Gen III+ reactors, 00:32:41.110 --> 00:32:44.890 align:middle line:84% and they both have outrageously high cost compared to everything 00:32:44.890 --> 00:32:46.710 align:middle line:90% that we've seen historically. 00:32:46.710 --> 00:32:50.310 align:middle line:84% And whether those are first-of-a-kind costs or not, 00:32:50.310 --> 00:32:52.390 align:middle line:84% or are they something related to the technology, 00:32:52.390 --> 00:32:54.570 align:middle line:84% we won't know unless we build a bunch of them. 00:32:54.570 --> 00:32:56.590 align:middle line:84% But there does seem to be something going on, 00:32:56.590 --> 00:32:58.690 align:middle line:84% because the first copies of older reactors 00:32:58.690 --> 00:33:01.690 align:middle line:90% were not nearly as expensive. 00:33:01.690 --> 00:33:07.010 align:middle line:84% So an important question for us as a class will be, 00:33:07.010 --> 00:33:10.280 align:middle line:90% is there too much safety? 00:33:10.280 --> 00:33:15.680 align:middle line:84% So we saw that when they tried to cut costs, 00:33:15.680 --> 00:33:19.420 align:middle line:84% they paid later on with reliability 00:33:19.420 --> 00:33:22.360 align:middle line:84% that potentially leads to more safety issues down the road. 00:33:22.360 --> 00:33:25.160 align:middle line:84% But then when we designed really safe reactors, 00:33:25.160 --> 00:33:28.640 align:middle line:84% like the French and the American reactor, 00:33:28.640 --> 00:33:30.720 align:middle line:84% the price went through the roofs. 00:33:30.720 --> 00:33:36.040 align:middle line:84% So we need to do a calculation to estimate 00:33:36.040 --> 00:33:38.360 align:middle line:84% the true safety of nuclear power in order 00:33:38.360 --> 00:33:41.820 align:middle line:84% to make a decision about what is the right technology to build, 00:33:41.820 --> 00:33:44.400 align:middle line:84% in order to make a decision about whether nuclear 00:33:44.400 --> 00:33:46.920 align:middle line:84% has a future to address climate change. 00:33:46.920 --> 00:33:48.460 align:middle line:90% And so we will do that. 00:33:48.460 --> 00:33:48.960 align:middle line:90% Yeah. 00:33:48.960 --> 00:33:50.335 align:middle line:84% AUDIENCE: What's your perspective 00:33:50.335 --> 00:33:54.260 align:middle line:84% on engineering safety versus operations safety? 00:33:54.260 --> 00:33:57.200 align:middle line:84% So for commercial reactors in the US, 00:33:57.200 --> 00:33:59.340 align:middle line:84% we typically rely on more engineering safety. 00:33:59.340 --> 00:34:01.743 align:middle line:84% But for naval reactors, the theory 00:34:01.743 --> 00:34:03.160 align:middle line:84% is basically you just have a bunch 00:34:03.160 --> 00:34:05.720 align:middle line:84% of very overqualified reactor operators 00:34:05.720 --> 00:34:07.300 align:middle line:90% who are always on shift. 00:34:07.300 --> 00:34:09.420 align:middle line:84% PROFESSOR: Oh, I think there's lots of engineering 00:34:09.420 --> 00:34:10.659 align:middle line:90% safety in naval reactors. 00:34:10.659 --> 00:34:11.860 align:middle line:84% AUDIENCE: There is a lot, don't get me wrong. 00:34:11.860 --> 00:34:13.318 align:middle line:84% PROFESSOR: Yeah, no, there's a lot. 00:34:13.318 --> 00:34:20.679 align:middle line:84% But we do have a better safety culture in the US industry 00:34:20.679 --> 00:34:22.920 align:middle line:90% than in other countries. 00:34:22.920 --> 00:34:26.100 align:middle line:84% And that's driven by a whole lot of things. 00:34:26.100 --> 00:34:29.360 align:middle line:90% Liability drives that. 00:34:29.360 --> 00:34:32.679 align:middle line:84% It's driven by the supporting structure set up 00:34:32.679 --> 00:34:44.520 align:middle line:84% by INPO, which is Institute of Nuclear Power Operators, INPO. 00:34:44.520 --> 00:34:48.040 align:middle line:84% It says, essentially, if there are small issues at your plant, 00:34:48.040 --> 00:34:51.679 align:middle line:84% you can share this information with others 00:34:51.679 --> 00:34:54.639 align:middle line:90% and without it becoming public. 00:34:54.639 --> 00:34:58.120 align:middle line:84% And so people learn from each other without being embarrassed, 00:34:58.120 --> 00:34:59.840 align:middle line:84% which is not something that necessarily 00:34:59.840 --> 00:35:01.140 align:middle line:90% happens in other countries. 00:35:01.140 --> 00:35:05.590 align:middle line:84% And it also turns out that the insurance that the plant 00:35:05.590 --> 00:35:08.710 align:middle line:84% operators have to purchase, the price of that insurance 00:35:08.710 --> 00:35:11.950 align:middle line:84% is related to how many safety incidents they have. 00:35:11.950 --> 00:35:14.870 align:middle line:84% So that drives them to have a lot of pressure 00:35:14.870 --> 00:35:17.390 align:middle line:90% for safe operation of the plant. 00:35:17.390 --> 00:35:20.310 align:middle line:84% For state-owned enterprises, where 00:35:20.310 --> 00:35:24.410 align:middle line:84% it's just person showing up to their government job, 00:35:24.410 --> 00:35:28.530 align:middle line:84% and they don't give an F, that might not be the case. 00:35:28.530 --> 00:35:31.770 align:middle line:84% So questions, questions, questions? 00:35:31.770 --> 00:35:32.470 align:middle line:90% Yeah. 00:35:32.470 --> 00:35:34.803 align:middle line:84% AUDIENCE: Are there any insights on the other countries, 00:35:34.803 --> 00:35:37.397 align:middle line:84% like China and Russia and all that, or is the data just-- 00:35:37.397 --> 00:35:38.730 align:middle line:90% PROFESSOR: There's no good data. 00:35:38.730 --> 00:35:42.590 align:middle line:84% And that's largely because the plants in those countries 00:35:42.590 --> 00:35:46.770 align:middle line:84% are built entirely outside of the free market system. 00:35:46.770 --> 00:35:53.710 align:middle line:84% So they have Institute of Machine Building 44 00:35:53.710 --> 00:35:55.410 align:middle line:84% making this part for the reactor. 00:35:55.410 --> 00:35:59.370 align:middle line:84% But that Institute of Machine Building would exist anyway. 00:35:59.370 --> 00:36:02.590 align:middle line:84% And its budgetary costs show up somewhere 00:36:02.590 --> 00:36:04.910 align:middle line:84% that is unrelated to the nuclear plant. 00:36:04.910 --> 00:36:07.570 align:middle line:84% And so there's no good way to actually price it. 00:36:07.570 --> 00:36:10.410 align:middle line:84% China does publish prices for their reactors, 00:36:10.410 --> 00:36:13.970 align:middle line:84% but you basically just can't trust them. 00:36:13.970 --> 00:36:17.030 align:middle line:84% It's not even clear they really know what they cost. 00:36:17.030 --> 00:36:20.910 align:middle line:90% So, yeah. 00:36:20.910 --> 00:36:23.430 align:middle line:84% AUDIENCE: It's like approximately 20-year gap 00:36:23.430 --> 00:36:27.150 align:middle line:84% between us building a reactor and also France that plays 00:36:27.150 --> 00:36:29.533 align:middle line:84% into that trend of the heavily increased costs? 00:36:29.533 --> 00:36:30.450 align:middle line:90% PROFESSOR: Yeah, I do. 00:36:30.450 --> 00:36:34.590 align:middle line:84% I think there's definitely some component of that. 00:36:34.590 --> 00:36:40.150 align:middle line:84% We have certainly lost a lot of industrial capacity in the US 00:36:40.150 --> 00:36:43.830 align:middle line:84% and probably in the West as things have been offshored. 00:36:43.830 --> 00:36:48.230 align:middle line:84% Korea has fantastic industrial capacity. 00:36:48.230 --> 00:36:51.350 align:middle line:84% Again, their predicted prices are higher. 00:36:51.350 --> 00:36:54.390 align:middle line:84% So I do think it's a combination of all of these things. 00:36:54.390 --> 00:36:55.750 align:middle line:90% Yeah. 00:36:55.750 --> 00:36:57.970 align:middle line:84% AUDIENCE: Is the Indian technology different? 00:36:57.970 --> 00:36:59.570 align:middle line:90% Are they also using US designs? 00:36:59.570 --> 00:37:02.290 align:middle line:84% PROFESSOR: No, they use completely their own designs. 00:37:02.290 --> 00:37:06.770 align:middle line:84% Yeah, it's a fascinating history for a nuclear program. 00:37:06.770 --> 00:37:10.470 align:middle line:84% The original design was they stole a CANDU design 00:37:10.470 --> 00:37:13.390 align:middle line:84% from the Canadians, and then they just 00:37:13.390 --> 00:37:17.990 align:middle line:84% couldn't build it because they didn't have the industrial base. 00:37:17.990 --> 00:37:24.430 align:middle line:84% So they said, OK, well, if it has some fancy, let's say, root 00:37:24.430 --> 00:37:26.518 align:middle line:84% style pump or some kind of fancy pump, 00:37:26.518 --> 00:37:28.310 align:middle line:84% we'll just replace it with a propeller pump 00:37:28.310 --> 00:37:30.070 align:middle line:84% because we can make a propeller pump. 00:37:30.070 --> 00:37:31.987 align:middle line:84% And then they just systematically went through 00:37:31.987 --> 00:37:33.030 align:middle line:90% and changed things. 00:37:33.030 --> 00:37:36.510 align:middle line:84% And they have a really weird regulatory process also. 00:37:36.510 --> 00:37:38.270 align:middle line:84% They have these people who are kind 00:37:38.270 --> 00:37:42.350 align:middle line:90% of gods in their little area. 00:37:42.350 --> 00:37:46.430 align:middle line:84% So they'd be like the god of pump safety, 00:37:46.430 --> 00:37:50.830 align:middle line:84% and whatever that person says, goes. 00:37:50.830 --> 00:37:54.030 align:middle line:84% But the result is that there's these high concentrations 00:37:54.030 --> 00:37:59.970 align:middle line:84% of expertise in different areas in the regulatory structure. 00:37:59.970 --> 00:38:02.420 align:middle line:84% It's a really interesting system, 00:38:02.420 --> 00:38:04.420 align:middle line:84% completely different from the rest of the world. 00:38:04.420 --> 00:38:05.180 align:middle line:90% Yeah. 00:38:05.180 --> 00:38:05.820 align:middle line:90% Yeah. 00:38:05.820 --> 00:38:08.040 align:middle line:84% AUDIENCE: So how does someone become the god of-- 00:38:08.040 --> 00:38:09.580 align:middle line:90% PROFESSOR: I have no idea. 00:38:09.580 --> 00:38:10.620 align:middle line:90% Yeah. 00:38:10.620 --> 00:38:12.440 align:middle line:90% Yeah. 00:38:12.440 --> 00:38:15.900 align:middle line:84% It's not at all like with computer codes 00:38:15.900 --> 00:38:17.160 align:middle line:90% in teams and stuff. 00:38:17.160 --> 00:38:19.200 align:middle line:90% It's just like-- yeah. 00:38:19.200 --> 00:38:22.660 align:middle line:84% AUDIENCE: They seem to be experts, by training or-- 00:38:22.660 --> 00:38:24.800 align:middle line:90% it's like political appointee? 00:38:24.800 --> 00:38:26.360 align:middle line:84% PROFESSOR: No, no, no, no, no, no. 00:38:26.360 --> 00:38:26.620 align:middle line:90% Yeah. 00:38:26.620 --> 00:38:26.920 align:middle line:90% Yeah. 00:38:26.920 --> 00:38:28.337 align:middle line:84% No, they are people who have spent 00:38:28.337 --> 00:38:30.900 align:middle line:84% their life, being the Indian expert on pump 00:38:30.900 --> 00:38:32.080 align:middle line:90% design for reactors. 00:38:32.080 --> 00:38:34.580 align:middle line:90% And then they become-- 00:38:34.580 --> 00:38:37.400 align:middle line:84% yeah, it's a very interesting kind of model. 00:38:37.400 --> 00:38:41.500 align:middle line:90% 00:38:41.500 --> 00:38:42.000 align:middle line:90% All right. 00:38:42.000 --> 00:38:45.440 align:middle line:84% So we do need to figure out what the situation. 00:38:45.440 --> 00:38:48.580 align:middle line:84% Any other questions on this or other countries? 00:38:48.580 --> 00:38:49.212 align:middle line:90% Yeah. 00:38:49.212 --> 00:38:50.920 align:middle line:84% AUDIENCE: You may be about to cover this, 00:38:50.920 --> 00:38:54.580 align:middle line:84% but how does nuclear accident rates 00:38:54.580 --> 00:38:56.980 align:middle line:90% compare to other energy sources? 00:38:56.980 --> 00:38:59.780 align:middle line:84% Is there a socially acceptable accident rate 00:38:59.780 --> 00:39:03.860 align:middle line:84% when it comes to coal spills or oil spills? 00:39:03.860 --> 00:39:07.238 align:middle line:84% PROFESSOR: We will cover that, but weeks from now. 00:39:07.238 --> 00:39:07.780 align:middle line:90% AUDIENCE: OK. 00:39:07.780 --> 00:39:10.260 align:middle line:84% PROFESSOR: Yeah, we will get to that. 00:39:10.260 --> 00:39:13.300 align:middle line:84% So yeah, we do need to do that calculation of what 00:39:13.300 --> 00:39:14.920 align:middle line:90% is acceptably safe. 00:39:14.920 --> 00:39:18.140 align:middle line:84% And that is going to be based on a comparative analysis 00:39:18.140 --> 00:39:20.500 align:middle line:90% with other technologies. 00:39:20.500 --> 00:39:24.100 align:middle line:84% So yeah, and we'll do the deaths per kilowatt or terawatt hour 00:39:24.100 --> 00:39:27.860 align:middle line:84% calculation, which is floating around on what-- yeah. 00:39:27.860 --> 00:39:30.000 align:middle line:84% AUDIENCE: You touched on it for India. 00:39:30.000 --> 00:39:33.078 align:middle line:84% Could you discuss what the regulatory structure looks 00:39:33.078 --> 00:39:34.620 align:middle line:84% like in France, and Japan, and Korea, 00:39:34.620 --> 00:39:36.490 align:middle line:84% and how that might contrast with the US? 00:39:36.490 --> 00:39:37.240 align:middle line:90% PROFESSOR: Oh, OK. 00:39:37.240 --> 00:39:42.720 align:middle line:84% So if you're really interested in this, go read on DSpace. 00:39:42.720 --> 00:39:44.540 align:middle line:90% Everyone know about DSpace? 00:39:44.540 --> 00:39:48.500 align:middle line:84% DSpace is an MIT website where all your theses are stored 00:39:48.500 --> 00:39:51.060 align:middle line:84% and people can put papers up there. 00:39:51.060 --> 00:39:54.220 align:middle line:84% You'll probably find on DSpace a PhD dissertation 00:39:54.220 --> 00:39:57.610 align:middle line:84% by Aditi Verma, who did a comparative study 00:39:57.610 --> 00:40:00.930 align:middle line:84% of the French, Indian, and US regulatory histories 00:40:00.930 --> 00:40:02.610 align:middle line:90% and how they came about. 00:40:02.610 --> 00:40:05.830 align:middle line:90% The French regulator has-- 00:40:05.830 --> 00:40:08.730 align:middle line:90% 00:40:08.730 --> 00:40:13.690 align:middle line:84% they're kind of like the NRC, but they work more 00:40:13.690 --> 00:40:16.050 align:middle line:90% intimately with the designers. 00:40:16.050 --> 00:40:18.350 align:middle line:84% So in the US, you design your reactor, 00:40:18.350 --> 00:40:20.610 align:middle line:84% you're a private company, you design your reactor, 00:40:20.610 --> 00:40:23.490 align:middle line:84% and you pay $1 billion to the NRC, 00:40:23.490 --> 00:40:26.450 align:middle line:84% and they run all of their codes and they check 00:40:26.450 --> 00:40:28.530 align:middle line:90% to see if your reactor is safe. 00:40:28.530 --> 00:40:32.130 align:middle line:90% And it's like going for a test. 00:40:32.130 --> 00:40:37.530 align:middle line:84% And in the French system, the regulatory safety people 00:40:37.530 --> 00:40:41.930 align:middle line:84% work with the designers as they're designing the reactor. 00:40:41.930 --> 00:40:44.270 align:middle line:84% So well, we think this should be improved here. 00:40:44.270 --> 00:40:45.410 align:middle line:84% And then they say, well, but then we're 00:40:45.410 --> 00:40:46.230 align:middle line:90% going to have to do this. 00:40:46.230 --> 00:40:48.450 align:middle line:84% OK, well, then why don't we do this other thing? 00:40:48.450 --> 00:40:52.930 align:middle line:84% And there's like this back and forth all through the process. 00:40:52.930 --> 00:40:59.610 align:middle line:84% So it seems like a good way to more efficiently make trade-offs 00:40:59.610 --> 00:41:01.210 align:middle line:90% between safety and cost. 00:41:01.210 --> 00:41:07.010 align:middle line:84% But unfortunately, this number doesn't back it up. 00:41:07.010 --> 00:41:10.250 align:middle line:90% So it is what it is? 00:41:10.250 --> 00:41:10.930 align:middle line:90% Yeah. 00:41:10.930 --> 00:41:16.290 align:middle line:84% And then I think the Japanese and Korean regulatory 00:41:16.290 --> 00:41:18.270 align:middle line:84% structures, but I'm not 100% sure of this, 00:41:18.270 --> 00:41:21.730 align:middle line:84% but I think they're basically just modeled on the US NRC. 00:41:21.730 --> 00:41:27.810 align:middle line:84% But they are much disempowered relative to the NRC, 00:41:27.810 --> 00:41:32.570 align:middle line:84% or historically were disempowered. 00:41:32.570 --> 00:41:34.490 align:middle line:90% Further questions? 00:41:34.490 --> 00:41:35.930 align:middle line:90% OK. 00:41:35.930 --> 00:41:42.210 align:middle line:84% So a big part of this is, we would like to get the cost down. 00:41:42.210 --> 00:41:44.030 align:middle line:90% And people talk about learning. 00:41:44.030 --> 00:41:49.150 align:middle line:84% And there hasn't been a lot of evidence of learning here. 00:41:49.150 --> 00:41:55.400 align:middle line:84% Oh, I want to answer the mystery before I go on. 00:41:55.400 --> 00:41:57.960 align:middle line:90% OK. 00:41:57.960 --> 00:42:00.170 align:middle line:90% So it turns out, India-- 00:42:00.170 --> 00:42:01.420 align:middle line:90% this will come up in a moment. 00:42:01.420 --> 00:42:03.040 align:middle line:84% I'll just tell you what the answer is. 00:42:03.040 --> 00:42:07.000 align:middle line:90% Why did India's costs come down? 00:42:07.000 --> 00:42:10.080 align:middle line:84% Because their reactors got bigger. 00:42:10.080 --> 00:42:11.740 align:middle line:84% Dramatically bigger as they print. 00:42:11.740 --> 00:42:16.040 align:middle line:84% And so their cost per unit of electricity came down. 00:42:16.040 --> 00:42:18.080 align:middle line:84% And their times to build those reactors 00:42:18.080 --> 00:42:20.560 align:middle line:84% got longer because the reactors got bigger. 00:42:20.560 --> 00:42:22.620 align:middle line:84% So that turns out to be the answer. 00:42:22.620 --> 00:42:28.960 align:middle line:84% And the fact that wasn't obvious from these data 00:42:28.960 --> 00:42:32.180 align:middle line:90% is an important point. 00:42:32.180 --> 00:42:34.600 align:middle line:84% There are all these unobserved variables. 00:42:34.600 --> 00:42:38.920 align:middle line:84% And people looking at these data, trying to draw conclusions 00:42:38.920 --> 00:42:42.040 align:middle line:84% about what things cost, but without looking 00:42:42.040 --> 00:42:45.280 align:middle line:84% at all of the parameters, are often 00:42:45.280 --> 00:42:47.520 align:middle line:90% drawing the wrong conclusions. 00:42:47.520 --> 00:42:52.000 align:middle line:84% And so we are going to try to measure 00:42:52.000 --> 00:42:54.520 align:middle line:90% this in a more robust way. 00:42:54.520 --> 00:42:57.943 align:middle line:84% So let's look at how we would do this rigorously. 00:42:57.943 --> 00:42:59.360 align:middle line:84% How we would look at the potential 00:42:59.360 --> 00:43:03.580 align:middle line:84% for bringing the cost down through learning. 00:43:03.580 --> 00:43:07.000 align:middle line:84% So there are various ideas for how learning 00:43:07.000 --> 00:43:09.320 align:middle line:90% occurs when you build things. 00:43:09.320 --> 00:43:13.520 align:middle line:84% Probably all of you have heard of Moore's law. 00:43:13.520 --> 00:43:18.960 align:middle line:84% Moore's law basically says that the cost, beta is minus, 00:43:18.960 --> 00:43:21.480 align:middle line:84% comes down as a function of time. 00:43:21.480 --> 00:43:26.720 align:middle line:84% Or you could say that the number of transistors per unit cost 00:43:26.720 --> 00:43:29.680 align:middle line:84% goes up per, or something like that. 00:43:29.680 --> 00:43:33.300 align:middle line:84% The problem is that if you don't do anything, 00:43:33.300 --> 00:43:35.640 align:middle line:84% there's no reason for the cost to come down. 00:43:35.640 --> 00:43:37.380 align:middle line:84% If you don't practice making reactors, 00:43:37.380 --> 00:43:39.360 align:middle line:84% they're not going to magically get cheaper. 00:43:39.360 --> 00:43:44.300 align:middle line:84% So it's a poor correlation where time is correlated with output. 00:43:44.300 --> 00:43:47.140 align:middle line:90% But it's a bad measure. 00:43:47.140 --> 00:43:49.240 align:middle line:90% It's a bad model. 00:43:49.240 --> 00:43:54.987 align:middle line:84% Unfortunately, NREL, the National-- 00:43:54.987 --> 00:43:56.570 align:middle line:84% AUDIENCE: Renewable Energy Laboratory. 00:43:56.570 --> 00:43:58.870 align:middle line:84% PROFESSOR: Yeah, Renewable Energy Laboratory, 00:43:58.870 --> 00:44:03.070 align:middle line:84% has a giant cost model that they use 00:44:03.070 --> 00:44:06.350 align:middle line:84% to predict the future cost of all kinds of generation 00:44:06.350 --> 00:44:08.350 align:middle line:90% technology, including reactor. 00:44:08.350 --> 00:44:17.630 align:middle line:84% And in fact, if we go back to this number, this 2018 number, 00:44:17.630 --> 00:44:20.590 align:middle line:90% this came from NREL. 00:44:20.590 --> 00:44:22.050 align:middle line:90% That is NREL's number. 00:44:22.050 --> 00:44:24.190 align:middle line:90% That's where they got it. 00:44:24.190 --> 00:44:29.503 align:middle line:84% And the problem is that, if we don't build reactors, 00:44:29.503 --> 00:44:31.170 align:middle line:84% we shouldn't expect them to get cheaper. 00:44:31.170 --> 00:44:33.110 align:middle line:84% But the NREL model just magically lets 00:44:33.110 --> 00:44:35.750 align:middle line:90% them get cheaper with time. 00:44:35.750 --> 00:44:37.590 align:middle line:90% So we shouldn't do that. 00:44:37.590 --> 00:44:40.190 align:middle line:84% And we can't trust the annual data for that reason. 00:44:40.190 --> 00:44:42.310 align:middle line:84% So there are other cost models that 00:44:42.310 --> 00:44:45.510 align:middle line:84% are based on something called Wright's law. 00:44:45.510 --> 00:44:48.910 align:middle line:84% This was developed by TP Wright in 1936 00:44:48.910 --> 00:44:50.770 align:middle line:84% to predict the cost of airplanes. 00:44:50.770 --> 00:44:52.558 align:middle line:90% This is not the Wright brothers. 00:44:52.558 --> 00:44:54.350 align:middle line:84% This is a different Wright, but it was also 00:44:54.350 --> 00:44:56.590 align:middle line:90% associated with airplanes. 00:44:56.590 --> 00:45:00.110 align:middle line:84% And there have been a number of statistical analyzes 00:45:00.110 --> 00:45:03.270 align:middle line:84% of different cost models to see which cost models actually 00:45:03.270 --> 00:45:04.970 align:middle line:90% fit historical data best. 00:45:04.970 --> 00:45:10.910 align:middle line:84% And indeed, these cost models, this capacity-based cost models, 00:45:10.910 --> 00:45:13.770 align:middle line:84% or units built, or capacity kilowatts built, 00:45:13.770 --> 00:45:20.870 align:middle line:84% or hours of time spent building, are the best models. 00:45:20.870 --> 00:45:21.990 align:middle line:90% All right. 00:45:21.990 --> 00:45:24.670 align:middle line:84% The problem with just having a simple cost model that 00:45:24.670 --> 00:45:27.710 align:middle line:84% looks like this, the cost is number of units 00:45:27.710 --> 00:45:32.430 align:middle line:84% that I built to some exponent, which is to say, 00:45:32.430 --> 00:45:35.670 align:middle line:84% if the exponent were negative, we would be like, 00:45:35.670 --> 00:45:39.110 align:middle line:84% the cost comes down for every doubling by the same amount 00:45:39.110 --> 00:45:42.030 align:middle line:84% for every doubling or every tripling of capacity 00:45:42.030 --> 00:45:43.430 align:middle line:90% is logarithmic. 00:45:43.430 --> 00:45:48.100 align:middle line:84% Problem is that you have these, you can't really 00:45:48.100 --> 00:45:52.460 align:middle line:84% solve this equation because of something called collinearity. 00:45:52.460 --> 00:45:55.540 align:middle line:84% So let me just show you what I mean. 00:45:55.540 --> 00:46:00.100 align:middle line:84% This is the history of construction of US reactors. 00:46:00.100 --> 00:46:04.060 align:middle line:84% Starting in the 1950s, these were called small modular 00:46:04.060 --> 00:46:04.560 align:middle line:90% reactors-- 00:46:04.560 --> 00:46:07.940 align:middle line:84% I mean, these early reactors, they were basically 00:46:07.940 --> 00:46:10.500 align:middle line:84% the same size as our small modular reactors 00:46:10.500 --> 00:46:13.580 align:middle line:84% that we're talking about building today. 00:46:13.580 --> 00:46:17.140 align:middle line:84% And they got bigger and bigger and bigger. 00:46:17.140 --> 00:46:21.500 align:middle line:84% And they got bigger because you had better economies of scale. 00:46:21.500 --> 00:46:25.180 align:middle line:84% And to keep the cost per unit kilowatt down, 00:46:25.180 --> 00:46:26.860 align:middle line:90% we made the reactors large. 00:46:26.860 --> 00:46:28.340 align:middle line:90% But here's the problem. 00:46:28.340 --> 00:46:31.700 align:middle line:84% There's almost perfectly linear correlation 00:46:31.700 --> 00:46:34.900 align:middle line:84% between the number of reactors built 00:46:34.900 --> 00:46:37.060 align:middle line:90% and the size of the reactor. 00:46:37.060 --> 00:46:40.100 align:middle line:84% So if you look at the historical data 00:46:40.100 --> 00:46:44.020 align:middle line:84% and you attribute the cost reduction 00:46:44.020 --> 00:46:46.660 align:middle line:84% to the number of reactors built, you 00:46:46.660 --> 00:46:50.260 align:middle line:84% may accidentally be measuring the cost reduction associated 00:46:50.260 --> 00:46:52.620 align:middle line:90% with the change in reactor size. 00:46:52.620 --> 00:46:55.440 align:middle line:90% So you have to control for this. 00:46:55.440 --> 00:46:58.220 align:middle line:90% 00:46:58.220 --> 00:47:00.500 align:middle line:84% And there are a number of other variables 00:47:00.500 --> 00:47:03.820 align:middle line:84% that people should be controlling for. 00:47:03.820 --> 00:47:08.240 align:middle line:84% So this is what explains the India stuff. 00:47:08.240 --> 00:47:10.540 align:middle line:90% Here is the Indian reactor size. 00:47:10.540 --> 00:47:14.720 align:middle line:84% And you can see that the reactors are getting larger. 00:47:14.720 --> 00:47:19.660 align:middle line:84% And this basically accounts for the cost trend. 00:47:19.660 --> 00:47:22.435 align:middle line:90% OK. 00:47:22.435 --> 00:47:27.820 align:middle line:84% Also describes some of what was going on in Korea. 00:47:27.820 --> 00:47:29.740 align:middle line:84% The reactors were getting larger, 00:47:29.740 --> 00:47:31.360 align:middle line:90% the costs were coming down. 00:47:31.360 --> 00:47:34.100 align:middle line:84% And then the better safety standards 00:47:34.100 --> 00:47:35.700 align:middle line:90% pushed the cost back up. 00:47:35.700 --> 00:47:39.260 align:middle line:84% So it's kind of rough, because you 00:47:39.260 --> 00:47:42.760 align:middle line:84% have to remember that these are not points. 00:47:42.760 --> 00:47:45.690 align:middle line:84% These are the durations of construction 00:47:45.690 --> 00:47:47.430 align:middle line:90% are large and so on. 00:47:47.430 --> 00:47:50.390 align:middle line:84% So it's hard to see the correlation. 00:47:50.390 --> 00:47:52.910 align:middle line:90% But the problem. 00:47:52.910 --> 00:47:57.050 align:middle line:84% So organizations like NEI have published 00:47:57.050 --> 00:47:59.530 align:middle line:84% all kinds of statements about the ability 00:47:59.530 --> 00:48:04.330 align:middle line:84% to bring the cost down based on the slope of these curves, 00:48:04.330 --> 00:48:06.810 align:middle line:84% but they are ignoring the slope of these curves 00:48:06.810 --> 00:48:09.410 align:middle line:90% when they make those statements. 00:48:09.410 --> 00:48:15.490 align:middle line:84% So here's a generalized cost model. 00:48:15.490 --> 00:48:19.610 align:middle line:84% Let me just explain to you why it has the shape that it does. 00:48:19.610 --> 00:48:22.450 align:middle line:84% I guess I'll write right over here. 00:48:22.450 --> 00:48:28.130 align:middle line:84% So we have the idea that the cost 00:48:28.130 --> 00:48:32.690 align:middle line:84% is equal to something like A, some factor, some input 00:48:32.690 --> 00:48:37.530 align:middle line:84% B to some power alpha, some input C to some power gamma, 00:48:37.530 --> 00:48:39.130 align:middle line:84% and dot, dot, dot, dot, and so on. 00:48:39.130 --> 00:48:42.850 align:middle line:84% And this is called a Cobb-Douglas production 00:48:42.850 --> 00:48:43.730 align:middle line:90% function. 00:48:43.730 --> 00:48:47.170 align:middle line:84% And the reason why people like this shape of production 00:48:47.170 --> 00:48:51.330 align:middle line:84% is that you can take the first partial of cost 00:48:51.330 --> 00:48:53.330 align:middle line:90% with respect to any input. 00:48:53.330 --> 00:48:56.410 align:middle line:84% So let's say the partial with respect to B. 00:48:56.410 --> 00:48:59.730 align:middle line:84% And then, of course, we get B times alpha times B 00:48:59.730 --> 00:49:03.930 align:middle line:84% to the alpha minus 1, C gamma, dot, dot, dot, dot. 00:49:03.930 --> 00:49:06.610 align:middle line:84% And then we can do the following math. 00:49:06.610 --> 00:49:20.450 align:middle line:84% Partial cost, partial B times B over cost is equal to-- 00:49:20.450 --> 00:49:26.050 align:middle line:84% and we'll just write out-- the denominator is A, 00:49:26.050 --> 00:49:28.970 align:middle line:90% B alpha, C gamma, dot, dot, dot. 00:49:28.970 --> 00:49:31.530 align:middle line:90% And the numerator is-- 00:49:31.530 --> 00:49:34.010 align:middle line:90% let's write it as alpha-- 00:49:34.010 --> 00:49:38.410 align:middle line:84% numerator is B, times alpha, times A, 00:49:38.410 --> 00:49:43.330 align:middle line:84% times B to the alpha minus 1, C to the gamma, dot, dot, dot. 00:49:43.330 --> 00:49:49.530 align:middle line:84% And of course, this B joins up with this B to make this alpha. 00:49:49.530 --> 00:49:53.210 align:middle line:84% And then this whole base cancels this. 00:49:53.210 --> 00:49:55.730 align:middle line:84% And what we're left with is this equation 00:49:55.730 --> 00:49:57.210 align:middle line:90% where this is proportional. 00:49:57.210 --> 00:50:02.610 align:middle line:84% So now I just rearrange this side, and I get partial cost. 00:50:02.610 --> 00:50:04.410 align:middle line:84% Sorry, I'm blocking the board here. 00:50:04.410 --> 00:50:13.350 align:middle line:84% Over cost is equal to alpha times partial B over B. 00:50:13.350 --> 00:50:14.270 align:middle line:90% Did I do that right? 00:50:14.270 --> 00:50:16.810 align:middle line:90% Yeah. 00:50:16.810 --> 00:50:18.730 align:middle line:90% So what does that mean? 00:50:18.730 --> 00:50:20.330 align:middle line:90% How do we interpret that? 00:50:20.330 --> 00:50:28.410 align:middle line:84% That means a change in the cost of some input B, 00:50:28.410 --> 00:50:31.930 align:middle line:84% and say a 1% change in the cost of input B, 00:50:31.930 --> 00:50:33.830 align:middle line:84% will have an alpha percent change, 00:50:33.830 --> 00:50:37.210 align:middle line:84% a constant effective percent change 00:50:37.210 --> 00:50:40.800 align:middle line:84% in the change of cost in the reactor. 00:50:40.800 --> 00:50:45.800 align:middle line:84% And so that idea that there's this constant percent 00:50:45.800 --> 00:50:49.680 align:middle line:84% for percent change is a nice assumption 00:50:49.680 --> 00:50:54.180 align:middle line:84% that seems to more or less hold and is independent of scale. 00:50:54.180 --> 00:50:57.360 align:middle line:84% And so this is why people use this type of model. 00:50:57.360 --> 00:50:58.880 align:middle line:90% So there it is just written out. 00:50:58.880 --> 00:51:00.920 align:middle line:84% And there's the B, and the C, and the D, 00:51:00.920 --> 00:51:04.080 align:middle line:84% and the E. All these controls will be in here. 00:51:04.080 --> 00:51:06.520 align:middle line:84% And then there's the key one that we're 00:51:06.520 --> 00:51:09.040 align:middle line:90% interested in is experience. 00:51:09.040 --> 00:51:10.640 align:middle line:84% And another nice thing about this 00:51:10.640 --> 00:51:14.160 align:middle line:84% is you can take the log of the whole equation. 00:51:14.160 --> 00:51:15.740 align:middle line:90% So you take the log of cost. 00:51:15.740 --> 00:51:18.200 align:middle line:84% And then all of these exponents come down, right, 00:51:18.200 --> 00:51:20.000 align:middle line:84% when you take the log of something. 00:51:20.000 --> 00:51:23.520 align:middle line:90% So you get this format. 00:51:23.520 --> 00:51:25.200 align:middle line:84% And the nice thing about this format 00:51:25.200 --> 00:51:28.240 align:middle line:90% is it's a linear equation. 00:51:28.240 --> 00:51:32.440 align:middle line:84% And that means it can be solved with least squares regression. 00:51:32.440 --> 00:51:36.520 align:middle line:84% So this is how people essentially 00:51:36.520 --> 00:51:43.480 align:middle line:84% do the fitting to estimate these parameters. 00:51:43.480 --> 00:51:47.000 align:middle line:84% So what kind of controls are we interested in? 00:51:47.000 --> 00:51:50.220 align:middle line:84% So controls could be something like, 00:51:50.220 --> 00:51:54.360 align:middle line:84% is it the first of a kind for that reactor design? 00:51:54.360 --> 00:51:56.360 align:middle line:84% Is it the first reactor built at a site 00:51:56.360 --> 00:51:59.920 align:middle line:84% or is it the second or third reactor built at a site? 00:51:59.920 --> 00:52:01.840 align:middle line:90% What country was it built in? 00:52:01.840 --> 00:52:05.720 align:middle line:84% What labor cost is showing in that country? 00:52:05.720 --> 00:52:08.280 align:middle line:84% Was it built before Three Mile Island 00:52:08.280 --> 00:52:10.240 align:middle line:84% when we didn't care about reactor safety 00:52:10.240 --> 00:52:11.840 align:middle line:90% or after Three Mile Island? 00:52:11.840 --> 00:52:13.180 align:middle line:90% You have a binary variable? 00:52:13.180 --> 00:52:13.880 align:middle line:90% Yep. 00:52:13.880 --> 00:52:16.400 align:middle line:84% AUDIENCE: It's just a kind of weird question regarding 00:52:16.400 --> 00:52:17.060 align:middle line:90% the equation. 00:52:17.060 --> 00:52:19.300 align:middle line:84% What actually are we measuring for experience? 00:52:19.300 --> 00:52:23.608 align:middle line:84% Is that literal years spent in building reactors? 00:52:23.608 --> 00:52:25.400 align:middle line:84% PROFESSOR: You can do it in different ways? 00:52:25.400 --> 00:52:30.000 align:middle line:84% I think ideally it would be, man years in reactor construction, 00:52:30.000 --> 00:52:32.480 align:middle line:84% but these data are hard to come by. 00:52:32.480 --> 00:52:35.880 align:middle line:84% So the way most people do it is just by number 00:52:35.880 --> 00:52:38.505 align:middle line:90% of reactors built. Yeah. 00:52:38.505 --> 00:52:39.130 align:middle line:90% AUDIENCE: Cool. 00:52:39.130 --> 00:52:41.150 align:middle line:90% PROFESSOR: Yeah. 00:52:41.150 --> 00:52:42.910 align:middle line:84% Some people have proposed looking 00:52:42.910 --> 00:52:45.790 align:middle line:84% at kilowatt of capacity built, but I think 00:52:45.790 --> 00:52:49.610 align:middle line:90% that's probably sub-ideal. 00:52:49.610 --> 00:52:51.410 align:middle line:90% But you could do that also. 00:52:51.410 --> 00:52:54.590 align:middle line:90% 00:52:54.590 --> 00:52:55.130 align:middle line:90% All right. 00:52:55.130 --> 00:52:57.490 align:middle line:90% So if we control for-- oh, OK. 00:52:57.490 --> 00:52:59.650 align:middle line:84% So let me just give you some more terms of art. 00:52:59.650 --> 00:53:02.150 align:middle line:84% You can basically solve this equation. 00:53:02.150 --> 00:53:06.833 align:middle line:84% And then you get these betas out when you solve. 00:53:06.833 --> 00:53:08.250 align:middle line:84% When you do the linear regression, 00:53:08.250 --> 00:53:10.190 align:middle line:84% you get betas for all the inputs. 00:53:10.190 --> 00:53:13.510 align:middle line:84% And we have this thing called 2 to the beta 00:53:13.510 --> 00:53:16.310 align:middle line:90% is the progress ratio. 00:53:16.310 --> 00:53:20.430 align:middle line:84% Or we have the learning rate, which is 100% times 1 minus 2 00:53:20.430 --> 00:53:21.710 align:middle line:90% to the beta. 00:53:21.710 --> 00:53:24.630 align:middle line:84% And so the way to put this into words 00:53:24.630 --> 00:53:28.750 align:middle line:84% is, the cost reduction in percent terms 00:53:28.750 --> 00:53:31.910 align:middle line:84% for every doubling of experience, 00:53:31.910 --> 00:53:34.910 align:middle line:84% that is the way we talk about it. 00:53:34.910 --> 00:53:39.470 align:middle line:84% If we double the number of reactors built, a beta of 0.2 00:53:39.470 --> 00:53:42.670 align:middle line:84% means a 20% cost reduction for every doubling of reactors 00:53:42.670 --> 00:53:44.950 align:middle line:90% built. All right. 00:53:44.950 --> 00:53:49.270 align:middle line:84% So let's look at some historical estimates 00:53:49.270 --> 00:53:54.330 align:middle line:90% of learning and critique them. 00:53:54.330 --> 00:53:55.950 align:middle line:90% And then we'll do our own. 00:53:55.950 --> 00:53:59.990 align:middle line:90% So the early ones-- 00:53:59.990 --> 00:54:03.150 align:middle line:84% actually, these are not organized by time. 00:54:03.150 --> 00:54:06.390 align:middle line:84% The first two found that, indeed, 00:54:06.390 --> 00:54:09.030 align:middle line:84% if you build more reactors, the cost would come down. 00:54:09.030 --> 00:54:12.750 align:middle line:84% 4% per doubling or 12% per doubling. 00:54:12.750 --> 00:54:16.710 align:middle line:84% But they made no attempt to control for reactor size. 00:54:16.710 --> 00:54:19.190 align:middle line:84% So their models confused the fact 00:54:19.190 --> 00:54:23.830 align:middle line:84% that reactors got larger with the cost coming down. 00:54:23.830 --> 00:54:28.510 align:middle line:84% These second two focused on construction firms 00:54:28.510 --> 00:54:34.540 align:middle line:84% rather than on the cost to the utility. 00:54:34.540 --> 00:54:39.060 align:middle line:84% And they concluded that actually, construction firms 00:54:39.060 --> 00:54:44.020 align:middle line:90% were bringing their costs down. 00:54:44.020 --> 00:54:46.660 align:middle line:84% They were actually building reactors faster, 00:54:46.660 --> 00:54:48.820 align:middle line:84% but the costs were not coming down. 00:54:48.820 --> 00:54:53.220 align:middle line:84% And they concluded that might be because of what they called 00:54:53.220 --> 00:54:58.940 align:middle line:84% rent capturing, which is to say, they're now the company that 00:54:58.940 --> 00:55:00.740 align:middle line:90% builds reactors. 00:55:00.740 --> 00:55:02.977 align:middle line:84% And so they give you the first one at actual cost, 00:55:02.977 --> 00:55:04.560 align:middle line:84% but now they want to make some profit, 00:55:04.560 --> 00:55:09.260 align:middle line:84% and they're just squeezing the utilities for profit. 00:55:09.260 --> 00:55:11.323 align:middle line:90% That may be true. 00:55:11.323 --> 00:55:13.740 align:middle line:84% And there's some evidence that one of the big construction 00:55:13.740 --> 00:55:15.900 align:middle line:84% companies in the US, Fluor, has been doing 00:55:15.900 --> 00:55:19.700 align:middle line:90% that with the NuScale reactor. 00:55:19.700 --> 00:55:24.540 align:middle line:84% But the other explanation that would produce the same effect 00:55:24.540 --> 00:55:30.260 align:middle line:84% that these people observed is that, yeah, so 7% 00:55:30.260 --> 00:55:35.435 align:middle line:84% learning to firm to construction firms at minus 49% industry. 00:55:35.435 --> 00:55:37.060 align:middle line:84% The other thing that would explain this 00:55:37.060 --> 00:55:42.020 align:middle line:84% would simply be that there's a lot of extra regulatory crap 00:55:42.020 --> 00:55:43.140 align:middle line:90% going on. 00:55:43.140 --> 00:55:46.900 align:middle line:84% So while the construction is maybe becoming more efficient, 00:55:46.900 --> 00:55:51.060 align:middle line:84% there's more to do, build the same reactor because 00:55:51.060 --> 00:55:54.420 align:middle line:90% of regulatory pressure. 00:55:54.420 --> 00:55:58.260 align:middle line:84% Zimmerman, Cantor and Hewlett both have 00:55:58.260 --> 00:56:03.620 align:middle line:90% really good statistics. 00:56:03.620 --> 00:56:08.420 align:middle line:84% I can't remember which one, maybe Cantor and Hewlett, 00:56:08.420 --> 00:56:13.740 align:middle line:84% one of these guys is on the board of the Fed. 00:56:13.740 --> 00:56:19.440 align:middle line:84% He's like a real econometrician and knows what they're doing. 00:56:19.440 --> 00:56:24.300 align:middle line:84% But they did not fit the traditional Cobb-Douglas style 00:56:24.300 --> 00:56:24.880 align:middle line:90% function. 00:56:24.880 --> 00:56:28.020 align:middle line:84% They instead found that the data was better 00:56:28.020 --> 00:56:31.370 align:middle line:84% fitted by a hyperbolic-learning function. 00:56:31.370 --> 00:56:37.410 align:middle line:84% So instead of a logarithmic decrease in cost, 00:56:37.410 --> 00:56:45.450 align:middle line:84% they have something that basically looks like this. 00:56:45.450 --> 00:56:48.330 align:middle line:84% And then the cost savings asymptotes to basically 00:56:48.330 --> 00:56:51.650 align:middle line:84% zero instead of continuously going down, down, down, 00:56:51.650 --> 00:56:53.392 align:middle line:90% and down. 00:56:53.392 --> 00:56:54.850 align:middle line:84% AUDIENCE: That makes sense, though? 00:56:54.850 --> 00:56:56.210 align:middle line:84% PROFESSOR: It does kind of make sense. 00:56:56.210 --> 00:56:56.970 align:middle line:84% AUDIENCE: Because you're going to reach 00:56:56.970 --> 00:56:58.178 align:middle line:90% a minimal cost at some point. 00:56:58.178 --> 00:57:00.012 align:middle line:84% PROFESSOR: Well, that's what people thought. 00:57:00.012 --> 00:57:01.490 align:middle line:84% But the whole learning literature 00:57:01.490 --> 00:57:04.810 align:middle line:90% actually shows the opposite. 00:57:04.810 --> 00:57:07.410 align:middle line:84% It's amazing how things just keep getting cheaper, more 00:57:07.410 --> 00:57:09.170 align:middle line:90% and more and more slowly. 00:57:09.170 --> 00:57:13.370 align:middle line:84% But they seem to get cheaper, almost like in perpetuity. 00:57:13.370 --> 00:57:15.810 align:middle line:84% But this says it very quickly after there's 00:57:15.810 --> 00:57:19.050 align:middle line:84% three or four reactors, there's no more learning to be had. 00:57:19.050 --> 00:57:22.730 align:middle line:84% And that model, I think, is a good model that describes 00:57:22.730 --> 00:57:25.210 align:middle line:84% the difference between first-of-a-kind 00:57:25.210 --> 00:57:26.810 align:middle line:90% and Nth-of-a-kind. 00:57:26.810 --> 00:57:31.330 align:middle line:84% The first time you build a reactor, oh, you didn't realize, 00:57:31.330 --> 00:57:34.510 align:middle line:84% this pipe is not going to fit through this doorway. 00:57:34.510 --> 00:57:36.290 align:middle line:90% And so you make some changes. 00:57:36.290 --> 00:57:39.010 align:middle line:84% But after a while, you debug it, and then there's 00:57:39.010 --> 00:57:41.690 align:middle line:84% no more construction based learning to be had. 00:57:41.690 --> 00:57:45.547 align:middle line:84% And so I think that is a good model for the first-of-a-kind, 00:57:45.547 --> 00:57:46.130 align:middle line:90% Nth-of-a-kind. 00:57:46.130 --> 00:57:48.088 align:middle line:84% The question is, though, that does not give you 00:57:48.088 --> 00:57:50.690 align:middle line:84% the long-term steady cost reduction that we would want 00:57:50.690 --> 00:57:52.690 align:middle line:90% to see to bring the cost down. 00:57:52.690 --> 00:57:57.450 align:middle line:84% So unfortunately, while these are very good models, 00:57:57.450 --> 00:57:59.850 align:middle line:84% they don't tell us anything about long-term learning 00:57:59.850 --> 00:58:00.630 align:middle line:90% potential. 00:58:00.630 --> 00:58:03.370 align:middle line:90% 00:58:03.370 --> 00:58:10.930 align:middle line:84% This Grubler and Escobar both focused on France. 00:58:10.930 --> 00:58:14.850 align:middle line:84% This guy is accused of using bad data. 00:58:14.850 --> 00:58:19.390 align:middle line:84% And so Escobar, Rangel, and Leveque-- 00:58:19.390 --> 00:58:21.090 align:middle line:90% or however you say this name-- 00:58:21.090 --> 00:58:24.250 align:middle line:84% redid the study using better quality data. 00:58:24.250 --> 00:58:28.300 align:middle line:90% 00:58:28.300 --> 00:58:31.240 align:middle line:84% So initially, Grubler was saying, 00:58:31.240 --> 00:58:35.960 align:middle line:84% prices going up by crazy amounts, negative learning. 00:58:35.960 --> 00:58:38.400 align:middle line:84% And they said, oh, this is not fair. 00:58:38.400 --> 00:58:43.280 align:middle line:84% And so they redid it, and they said, something in this range. 00:58:43.280 --> 00:58:46.840 align:middle line:84% And basically there was no statistical significance 00:58:46.840 --> 00:58:48.140 align:middle line:90% in their data set. 00:58:48.140 --> 00:58:51.400 align:middle line:90% So it's like approximately zero. 00:58:51.400 --> 00:58:56.160 align:middle line:84% And then Buongiorno, along with Eash-Gates, 00:58:56.160 --> 00:58:59.340 align:middle line:84% did this study of a few different kind of reactor types, 00:58:59.340 --> 00:59:01.400 align:middle line:84% looking at particular key components, 00:59:01.400 --> 00:59:04.160 align:middle line:84% and found, depending on what they 00:59:04.160 --> 00:59:08.200 align:middle line:84% were looking at, somewhere between minus 115% to minus 31% 00:59:08.200 --> 00:59:08.700 align:middle line:90% learning. 00:59:08.700 --> 00:59:12.080 align:middle line:84% So again, costs are going up and up. 00:59:12.080 --> 00:59:15.280 align:middle line:84% But basically, they did not do any of these controls 00:59:15.280 --> 00:59:16.880 align:middle line:90% that we talked about. 00:59:16.880 --> 00:59:19.960 align:middle line:84% They just looked at the prices and didn't control for things. 00:59:19.960 --> 00:59:21.420 align:middle line:90% So we need to do-- 00:59:21.420 --> 00:59:24.280 align:middle line:84% none of this, this is all the entire literature on cost, 00:59:24.280 --> 00:59:27.000 align:middle line:90% and it's all crap. 00:59:27.000 --> 00:59:28.780 align:middle line:90% So we need to do a better job. 00:59:28.780 --> 00:59:33.880 align:middle line:84% And so for this class, I have produced a regression 00:59:33.880 --> 00:59:36.240 align:middle line:90% with 428 reactors. 00:59:36.240 --> 00:59:40.040 align:middle line:90% Painfully data collected. 00:59:40.040 --> 00:59:41.700 align:middle line:84% The eight largest nuclear countries. 00:59:41.700 --> 00:59:43.760 align:middle line:84% Extensive controls on experimental reactors 00:59:43.760 --> 00:59:45.000 align:middle line:90% have been removed. 00:59:45.000 --> 00:59:46.860 align:middle line:84% Heteroskedastic and autocorrelation 00:59:46.860 --> 00:59:50.360 align:middle line:90% robust estimation model. 00:59:50.360 --> 00:59:54.720 align:middle line:84% And the various models are selected based on the Akaike 00:59:54.720 --> 00:59:57.040 align:middle line:90% information criterion. 00:59:57.040 --> 00:59:59.100 align:middle line:84% We can talk about that at some future point. 00:59:59.100 --> 01:00:01.280 align:middle line:84% But basically this is a very robust way 01:00:01.280 --> 01:00:05.920 align:middle line:84% of attempting to estimate the learning potential. 01:00:05.920 --> 01:00:09.140 align:middle line:84% And I will just show you some results. 01:00:09.140 --> 01:00:14.880 align:middle line:84% This is the screenshot of the output. 01:00:14.880 --> 01:00:18.840 align:middle line:84% In the end, only 341 reactors are 01:00:18.840 --> 01:00:25.670 align:middle line:84% in the model because of the, what, 428 we started with, yeah, 01:00:25.670 --> 01:00:29.110 align:middle line:84% we removed all reactors that were one-offs, 01:00:29.110 --> 01:00:32.402 align:middle line:84% for which that particular design was not built multiple times, 01:00:32.402 --> 01:00:34.110 align:middle line:84% and therefore, we can't measure learning. 01:00:34.110 --> 01:00:36.390 align:middle line:90% We do this by reactor design. 01:00:36.390 --> 01:00:40.630 align:middle line:90% 01:00:40.630 --> 01:00:42.450 align:middle line:84% I guess that's principally the cause. 01:00:42.450 --> 01:00:46.470 align:middle line:84% Experimental reactors are the same. 01:00:46.470 --> 01:00:49.810 align:middle line:84% Everything is statistically significant. 01:00:49.810 --> 01:00:53.870 align:middle line:90% 01:00:53.870 --> 01:00:55.870 align:middle line:90% What is this? 01:00:55.870 --> 01:01:00.930 align:middle line:84% Unless you're in Canada, so is Canadian, is German, is French, 01:01:00.930 --> 01:01:01.490 align:middle line:90% is British. 01:01:01.490 --> 01:01:05.510 align:middle line:84% So these are different controls for different countries. 01:01:05.510 --> 01:01:07.630 align:middle line:90% Log of capacity factor. 01:01:07.630 --> 01:01:09.590 align:middle line:90% First at a site. 01:01:09.590 --> 01:01:11.647 align:middle line:84% Number of reactors that you've completed, 01:01:11.647 --> 01:01:14.230 align:middle line:84% whether or not the reactor is built before or after Three Mile 01:01:14.230 --> 01:01:15.990 align:middle line:90% Island. 01:01:15.990 --> 01:01:17.870 align:middle line:84% Also we control for the total number 01:01:17.870 --> 01:01:20.990 align:middle line:84% of reactors in the project and the number of total projects 01:01:20.990 --> 01:01:24.870 align:middle line:84% that the country has built or licensed, 01:01:24.870 --> 01:01:27.550 align:middle line:84% so they did some separate measure. 01:01:27.550 --> 01:01:29.470 align:middle line:84% So just let me just put the results 01:01:29.470 --> 01:01:34.750 align:middle line:84% in a way that is easier to read the data table. 01:01:34.750 --> 01:01:39.590 align:middle line:84% So if your reactor was built before Three Mile Island, 01:01:39.590 --> 01:01:41.950 align:middle line:84% this says it will cost 40% of what it would 01:01:41.950 --> 01:01:44.270 align:middle line:90% cost after Three Mile Island. 01:01:44.270 --> 01:01:48.270 align:middle line:84% That's basically the cost of the safety improvements 01:01:48.270 --> 01:01:49.890 align:middle line:90% that evolves. 01:01:49.890 --> 01:01:52.030 align:middle line:84% Again, get back to this question. 01:01:52.030 --> 01:01:53.470 align:middle line:90% Is there too much safety? 01:01:53.470 --> 01:01:59.230 align:middle line:84% Because if we can drop the cost by that much, basically 50%, 01:01:59.230 --> 01:02:00.970 align:middle line:90% that would have a big impact. 01:02:00.970 --> 01:02:05.330 align:middle line:90% 01:02:05.330 --> 01:02:06.210 align:middle line:90% Reactor size. 01:02:06.210 --> 01:02:07.510 align:middle line:90% No surprise here. 01:02:07.510 --> 01:02:10.370 align:middle line:84% For every doubling of reactor size, 01:02:10.370 --> 01:02:13.710 align:middle line:90% there's a 20% cost reduction. 01:02:13.710 --> 01:02:17.150 align:middle line:84% So this is why reactors got big in the first place, 01:02:17.150 --> 01:02:20.070 align:middle line:90% as you saw in the data. 01:02:20.070 --> 01:02:21.330 align:middle line:90% But here's the problem. 01:02:21.330 --> 01:02:24.030 align:middle line:84% Now we're talking about making reactors small. 01:02:24.030 --> 01:02:27.110 align:middle line:90% So what that do to the cost? 01:02:27.110 --> 01:02:30.950 align:middle line:84% So we can just do a quick calculation. 01:02:30.950 --> 01:02:37.790 align:middle line:84% The NuScale reactor is about 77 megawatts electric. 01:02:37.790 --> 01:02:42.990 align:middle line:84% And the AP1000 is about 1,100 megawatts electric. 01:02:42.990 --> 01:02:47.830 align:middle line:84% So we can ask, how many doublings are there to go from-- 01:02:47.830 --> 01:02:52.070 align:middle line:84% inverse doublings-- to go from the AP1000, where we 01:02:52.070 --> 01:02:54.390 align:middle line:90% have cost data, to the NuScale? 01:02:54.390 --> 01:02:59.190 align:middle line:84% And so we can do this by taking this, log base 2, which 01:02:59.190 --> 01:03:02.990 align:middle line:84% will tell us how many doublings, which, of course, is equal to, 01:03:02.990 --> 01:03:04.310 align:middle line:90% what? 01:03:04.310 --> 01:03:10.350 align:middle line:84% Ln of 77 over 1,100 divided by ln of 2, 01:03:10.350 --> 01:03:16.390 align:middle line:90% which I believe is minus 3.84. 01:03:16.390 --> 01:03:22.100 align:middle line:84% So with minus 3.84 doublings, and so then we want to know, 01:03:22.100 --> 01:03:24.020 align:middle line:90% what is the change in cost? 01:03:24.020 --> 01:03:27.300 align:middle line:84% So the new cost will be 20% reduction. 01:03:27.300 --> 01:03:31.100 align:middle line:90% So 1 minus 0.2-- or 0.8-- 01:03:31.100 --> 01:03:34.940 align:middle line:90% to the minus 3.84. 01:03:34.940 --> 01:03:37.300 align:middle line:84% Because for every doubling, we have a 20% cost, 01:03:37.300 --> 01:03:39.540 align:middle line:84% and we have minus 3.84 doublings. 01:03:39.540 --> 01:03:43.900 align:middle line:90% And this turns out to be 2.35. 01:03:43.900 --> 01:03:47.100 align:middle line:84% So on a per kilowatt capacity basis, 01:03:47.100 --> 01:03:51.580 align:middle line:84% the fact that we've gone small, we would predict we'll increase 01:03:51.580 --> 01:03:55.420 align:middle line:84% the cost by a factor of 2 and 1/2, roughly. 01:03:55.420 --> 01:03:59.340 align:middle line:84% And so NuScale has to make up, these small modular reactor 01:03:59.340 --> 01:04:04.820 align:middle line:84% concepts, have to make up that loss of cost reduction 01:04:04.820 --> 01:04:05.980 align:middle line:90% somewhere else. 01:04:05.980 --> 01:04:10.180 align:middle line:84% And the place they hope to make it up is in factory fabrication. 01:04:10.180 --> 01:04:12.100 align:middle line:84% And we'll have a lecture on whether they're 01:04:12.100 --> 01:04:13.460 align:middle line:90% getting there or not. 01:04:13.460 --> 01:04:15.220 align:middle line:90% All right. 01:04:15.220 --> 01:04:18.880 align:middle line:90% So this comes out of the data. 01:04:18.880 --> 01:04:21.120 align:middle line:84% That's a 95% confidence interval. 01:04:21.120 --> 01:04:22.740 align:middle line:84% And you'll see that it's pretty good. 01:04:22.740 --> 01:04:26.260 align:middle line:90% Reasonably tight. 01:04:26.260 --> 01:04:28.120 align:middle line:84% If you build it first reactor at a site, 01:04:28.120 --> 01:04:31.380 align:middle line:84% it will cost you 12% more because you 01:04:31.380 --> 01:04:35.100 align:middle line:84% have to deal with this site purchase and issues like that. 01:04:35.100 --> 01:04:40.460 align:middle line:84% If your project size is larger, two reactors 01:04:40.460 --> 01:04:43.283 align:middle line:84% instead of one reactor, your cost comes down by 4%. 01:04:43.283 --> 01:04:45.200 align:middle line:84% If it's four reactors instead of two reactors, 01:04:45.200 --> 01:04:47.620 align:middle line:90% your cost comes down by 4%. 01:04:47.620 --> 01:04:49.780 align:middle line:84% So it's a little bit of savings, but it's 01:04:49.780 --> 01:04:51.980 align:middle line:84% one of the reasons you see that we never build 01:04:51.980 --> 01:04:55.700 align:middle line:90% one reactor at a time now. 01:04:55.700 --> 01:05:02.833 align:middle line:84% And this section is a weird phenomenon. 01:05:02.833 --> 01:05:04.000 align:middle line:90% It's not a weird phenomenon. 01:05:04.000 --> 01:05:05.500 align:middle line:84% It's kind of an expected phenomenon. 01:05:05.500 --> 01:05:07.580 align:middle line:84% Let me just explain what's going on. 01:05:07.580 --> 01:05:11.300 align:middle line:84% The model is attempting to apportion the variability 01:05:11.300 --> 01:05:17.610 align:middle line:84% that it sees between the number of licenses granted, or projects 01:05:17.610 --> 01:05:21.170 align:middle line:84% completed, if you will, and the number of reactors built. 01:05:21.170 --> 01:05:24.170 align:middle line:84% But there's an almost perfect one-to-one correlation 01:05:24.170 --> 01:05:25.810 align:middle line:90% between these things. 01:05:25.810 --> 01:05:31.490 align:middle line:84% So it doesn't know exactly how to apportion that variability. 01:05:31.490 --> 01:05:35.410 align:middle line:84% And this is a big, big uncertainty in how that does, 01:05:35.410 --> 01:05:38.090 align:middle line:84% they call it "horse racing" when you do regression. 01:05:38.090 --> 01:05:44.370 align:middle line:84% So basically, if you just want total number of everything's, 01:05:44.370 --> 01:05:48.570 align:middle line:84% licensed and built, we would add these two numbers together 01:05:48.570 --> 01:05:55.730 align:middle line:84% and we would get 1% as the learning per project license 01:05:55.730 --> 01:05:57.050 align:middle line:90% and completed. 01:05:57.050 --> 01:05:59.250 align:middle line:90% Does that make sense? 01:05:59.250 --> 01:06:00.530 align:middle line:90% OK. 01:06:00.530 --> 01:06:04.930 align:middle line:84% So that's interesting because a 1% learning 01:06:04.930 --> 01:06:08.490 align:middle line:84% per every doubling kind of is consistent with the data 01:06:08.490 --> 01:06:09.530 align:middle line:90% we've seen. 01:06:09.530 --> 01:06:10.830 align:middle line:90% The data is all over the place. 01:06:10.830 --> 01:06:12.350 align:middle line:90% We don't see dramatic learning. 01:06:12.350 --> 01:06:16.450 align:middle line:84% And what it says is that you can build a lot of reactors 01:06:16.450 --> 01:06:19.130 align:middle line:84% and they will come down in price, 01:06:19.130 --> 01:06:23.890 align:middle line:84% but it will happen very, very, very slowly. 01:06:23.890 --> 01:06:25.957 align:middle line:84% To get to the cost that we need, we're 01:06:25.957 --> 01:06:28.290 align:middle line:84% going to have to double and double and double and double 01:06:28.290 --> 01:06:28.790 align:middle line:90% and double. 01:06:28.790 --> 01:06:31.650 align:middle line:84% And we're going to go down 1%, go down 2%, go down 3%. 01:06:31.650 --> 01:06:35.370 align:middle line:90% It's going to be really slow. 01:06:35.370 --> 01:06:36.770 align:middle line:84% Building a lot of reactors is not 01:06:36.770 --> 01:06:39.710 align:middle line:84% going to be the answer to getting the price down. 01:06:39.710 --> 01:06:42.970 align:middle line:84% It's got to be something else, like reducing safety 01:06:42.970 --> 01:06:45.050 align:middle line:84% or switching how they're built from stick-built 01:06:45.050 --> 01:06:46.970 align:middle line:84% to factory-built, or something else. 01:06:46.970 --> 01:06:49.830 align:middle line:84% And so that's kind of the result of this. 01:06:49.830 --> 01:06:57.330 align:middle line:84% And that's also, if you will, consistent with all of this. 01:06:57.330 --> 01:07:00.410 align:middle line:84% Basically, the person who was the economist 01:07:00.410 --> 01:07:02.050 align:middle line:84% and really knew what he was doing 01:07:02.050 --> 01:07:03.890 align:middle line:84% said there's basically no learning. 01:07:03.890 --> 01:07:06.810 align:middle line:84% These people said there's basically very little learning. 01:07:06.810 --> 01:07:09.630 align:middle line:90% And it's consistent. 01:07:09.630 --> 01:07:11.650 align:middle line:84% We don't see any really good studies that 01:07:11.650 --> 01:07:14.600 align:middle line:90% show there's lots of learning. 01:07:14.600 --> 01:07:15.480 align:middle line:90% All right. 01:07:15.480 --> 01:07:23.560 align:middle line:84% So that's kind of the end of this. 01:07:23.560 --> 01:07:28.320 align:middle line:84% Will nuclear power be cheaper if we just build more? 01:07:28.320 --> 01:07:30.400 align:middle line:90% Yes, but insignificantly. 01:07:30.400 --> 01:07:32.740 align:middle line:84% So we have to fix it somewhere else. 01:07:32.740 --> 01:07:35.280 align:middle line:90% 01:07:35.280 --> 01:07:39.540 align:middle line:84% While I have this license and built story up here, 01:07:39.540 --> 01:07:42.040 align:middle line:84% I think I will use the next few minutes 01:07:42.040 --> 01:07:44.380 align:middle line:84% to tell you about the combined operating license. 01:07:44.380 --> 01:07:46.920 align:middle line:90% 01:07:46.920 --> 01:07:50.560 align:middle line:84% In the United States, we used to have a process where 01:07:50.560 --> 01:07:57.080 align:middle line:84% you would go to the NRC and get a construction permit, 01:07:57.080 --> 01:07:59.960 align:middle line:84% and then you would tell them what you were going to build. 01:07:59.960 --> 01:08:03.100 align:middle line:84% And then after you built it, and it will never 01:08:03.100 --> 01:08:04.600 align:middle line:84% be built in the way that you thought 01:08:04.600 --> 01:08:07.680 align:middle line:84% it was going to be built, they would come back 01:08:07.680 --> 01:08:12.440 align:middle line:84% and they would give you an operating license. 01:08:12.440 --> 01:08:14.960 align:middle line:84% The problem there is, I don't remember 01:08:14.960 --> 01:08:17.279 align:middle line:84% when it was, there was a shorthand nuclear plant built 01:08:17.279 --> 01:08:18.359 align:middle line:90% in New York. 01:08:18.359 --> 01:08:22.279 align:middle line:84% And they were doing this and they built the whole plant 01:08:22.279 --> 01:08:26.080 align:middle line:84% and they were going for their operating license. 01:08:26.080 --> 01:08:28.160 align:middle line:90% It was built in Long Island. 01:08:28.160 --> 01:08:31.640 align:middle line:84% And the local community, anti-nuclear activists 01:08:31.640 --> 01:08:33.279 align:middle line:90% got all activated. 01:08:33.279 --> 01:08:35.439 align:middle line:84% And they said, if there's a reactor accident 01:08:35.439 --> 01:08:37.540 align:middle line:84% and we're on the far side of Long Island, 01:08:37.540 --> 01:08:41.000 align:middle line:84% there's no way for us to evacuate except to go 01:08:41.000 --> 01:08:43.040 align:middle line:90% into the radioactive plume. 01:08:43.040 --> 01:08:48.040 align:middle line:84% And this led to a lot of public pressure 01:08:48.040 --> 01:08:52.319 align:middle line:84% to delay the issuance of the operating license. 01:08:52.319 --> 01:08:54.960 align:middle line:84% And eventually they just gave up. 01:08:54.960 --> 01:08:59.479 align:middle line:84% So there's a whole plant that was built, ready to go, 01:08:59.479 --> 01:09:02.399 align:middle line:90% never operated. 01:09:02.399 --> 01:09:06.200 align:middle line:84% So the industry said, never will we let this happen again. 01:09:06.200 --> 01:09:08.380 align:middle line:84% And they went to the NRC and they said, 01:09:08.380 --> 01:09:13.069 align:middle line:84% we need you to change the rules so that this can't happen. 01:09:13.069 --> 01:09:15.870 align:middle line:84% And so the NRC said, well, our obligation 01:09:15.870 --> 01:09:18.630 align:middle line:84% is to make sure that the plant that is built is a safe plant, 01:09:18.630 --> 01:09:22.430 align:middle line:84% and we need to know what has been built. 01:09:22.430 --> 01:09:24.310 align:middle line:84% And so they came up with this agreement 01:09:24.310 --> 01:09:28.470 align:middle line:84% called the COL, Combined Operating License. 01:09:28.470 --> 01:09:29.689 align:middle line:90% And it works like this. 01:09:29.689 --> 01:09:31.830 align:middle line:84% You tell us what you're going to build, 01:09:31.830 --> 01:09:34.630 align:middle line:84% and as long as you build exactly what you say, 01:09:34.630 --> 01:09:36.270 align:middle line:90% then you can operate it. 01:09:36.270 --> 01:09:42.529 align:middle line:84% But if you make any changes, you have to get a license variant 01:09:42.529 --> 01:09:44.149 align:middle line:90% exemption, basically. 01:09:44.149 --> 01:09:47.670 align:middle line:84% So that is how plants have been built since. 01:09:47.670 --> 01:09:50.310 align:middle line:84% But now what happens is, inevitably 01:09:50.310 --> 01:09:52.930 align:middle line:84% they have to change something during the construction, 01:09:52.930 --> 01:09:55.590 align:middle line:84% and then they have to go back and get NRC approval 01:09:55.590 --> 01:09:58.710 align:middle line:84% for that change as they're doing the construction, which then 01:09:58.710 --> 01:10:02.790 align:middle line:84% slows down the construction, which then causes the interest 01:10:02.790 --> 01:10:05.070 align:middle line:84% payments to accumulate, which increases 01:10:05.070 --> 01:10:07.350 align:middle line:90% the cost of nuclear power. 01:10:07.350 --> 01:10:12.110 align:middle line:84% And so this is what happened in part with the Vogtle design. 01:10:12.110 --> 01:10:17.550 align:middle line:84% There were something like of order 100 license modifications 01:10:17.550 --> 01:10:19.310 align:middle line:90% during the construction. 01:10:19.310 --> 01:10:22.750 align:middle line:84% And it's not a surprise that pervades everything and pushes 01:10:22.750 --> 01:10:23.830 align:middle line:90% up the price. 01:10:23.830 --> 01:10:27.630 align:middle line:84% So we have not yet figured out how 01:10:27.630 --> 01:10:30.310 align:middle line:84% to modify stick-built construction to make it 01:10:30.310 --> 01:10:31.490 align:middle line:90% so that this won't happen. 01:10:31.490 --> 01:10:31.990 align:middle line:90% Yeah. 01:10:31.990 --> 01:10:33.410 align:middle line:84% AUDIENCE: So given the obvious risk, 01:10:33.410 --> 01:10:35.118 align:middle line:84% I don't know that this ever would happen. 01:10:35.118 --> 01:10:37.910 align:middle line:84% But does the NRC still allow you to do the construction 01:10:37.910 --> 01:10:39.218 align:middle line:90% permit and then the-- 01:10:39.218 --> 01:10:41.010 align:middle line:84% PROFESSOR: I think it's still on the books. 01:10:41.010 --> 01:10:45.430 align:middle line:84% AUDIENCE: OK, I'm curious as to if there's any marginal savings 01:10:45.430 --> 01:10:47.950 align:middle line:84% there, if you have a community that's very down 01:10:47.950 --> 01:10:48.915 align:middle line:90% to have nuclear power. 01:10:48.915 --> 01:10:50.290 align:middle line:84% So you don't have to worry about. 01:10:50.290 --> 01:10:53.070 align:middle line:84% PROFESSOR: Yeah, so I think the sociologists who 01:10:53.070 --> 01:10:56.230 align:middle line:84% study nuclear energy would probably say, 01:10:56.230 --> 01:11:00.390 align:middle line:84% if you do more consent-based siting, where you get everyone 01:11:00.390 --> 01:11:03.910 align:middle line:84% to buy-in before you start building the plant, 01:11:03.910 --> 01:11:07.180 align:middle line:84% then you could go back to the other model 01:11:07.180 --> 01:11:08.760 align:middle line:84% and it would probably save you time. 01:11:08.760 --> 01:11:09.260 align:middle line:90% Yeah. 01:11:09.260 --> 01:11:12.460 align:middle line:84% So I mean, the way the new model works is, they 01:11:12.460 --> 01:11:15.500 align:middle line:90% get their construction permit. 01:11:15.500 --> 01:11:17.800 align:middle line:84% Nobody in the community really is paying attention. 01:11:17.800 --> 01:11:19.500 align:middle line:84% And suddenly there's a nuclear power plant being built 01:11:19.500 --> 01:11:21.280 align:middle line:84% and people are like, wait a second, we're opposed. 01:11:21.280 --> 01:11:22.655 align:middle line:84% You're like, oh, sorry, too late. 01:11:22.655 --> 01:11:24.780 align:middle line:84% You should have come to the meeting before-- 01:11:24.780 --> 01:11:28.640 align:middle line:84% now the COL has been granted, nothing you can do. 01:11:28.640 --> 01:11:32.920 align:middle line:84% And it's honestly not fair to those communities, 01:11:32.920 --> 01:11:36.060 align:middle line:84% but the better way would be just to do as you say. 01:11:36.060 --> 01:11:40.940 align:middle line:84% Do the consent-based siting first, and then build the plant 01:11:40.940 --> 01:11:42.800 align:middle line:90% under the old licensing model. 01:11:42.800 --> 01:11:43.425 align:middle line:90% Yeah. 01:11:43.425 --> 01:11:47.380 align:middle line:84% AUDIENCE: It sounds like the COL that 01:11:47.380 --> 01:11:50.300 align:middle line:90% reinvented the French system. 01:11:50.300 --> 01:11:53.020 align:middle line:84% Your company has to be in constant conversation 01:11:53.020 --> 01:11:53.740 align:middle line:90% with the NRC. 01:11:53.740 --> 01:11:55.560 align:middle line:90% You know, it would like. 01:11:55.560 --> 01:11:58.400 align:middle line:84% And they'll be like, I don't know, maybe try this instead. 01:11:58.400 --> 01:12:02.020 align:middle line:84% And then if you do that time after time for the whole period 01:12:02.020 --> 01:12:04.700 align:middle line:90% that you're building, is it-- 01:12:04.700 --> 01:12:09.000 align:middle line:84% I guess, do we know that the French system isn't the worst? 01:12:09.000 --> 01:12:11.820 align:middle line:84% Because maybe this is better long term 01:12:11.820 --> 01:12:16.180 align:middle line:84% if we actually have the NRC more involved at each stage 01:12:16.180 --> 01:12:18.040 align:middle line:90% rather than if they actually-- 01:12:18.040 --> 01:12:22.502 align:middle line:84% Well, these would be, I can't imagine a time where 01:12:22.502 --> 01:12:24.460 align:middle line:84% someone would actually just build a whole plant 01:12:24.460 --> 01:12:26.002 align:middle line:84% and then everything would be perfect, 01:12:26.002 --> 01:12:28.200 align:middle line:84% and at the end of the day, ah, great job. 01:12:28.200 --> 01:12:29.940 align:middle line:90% No notes. 01:12:29.940 --> 01:12:33.780 align:middle line:84% PROFESSOR: So your analogy is apt. 01:12:33.780 --> 01:12:36.580 align:middle line:84% It's not exactly the same as the French system 01:12:36.580 --> 01:12:40.260 align:middle line:84% because the French regulator doesn't work with them 01:12:40.260 --> 01:12:41.940 align:middle line:90% during construction. 01:12:41.940 --> 01:12:45.120 align:middle line:84% Well, they probably do, but it's more during design phase. 01:12:45.120 --> 01:12:47.540 align:middle line:84% And what's happening in the US is 01:12:47.540 --> 01:12:50.403 align:middle line:84% it's back and forth during construction, 01:12:50.403 --> 01:12:51.820 align:middle line:84% and that's where then the interest 01:12:51.820 --> 01:12:53.080 align:middle line:90% payments are accumulating. 01:12:53.080 --> 01:12:54.700 align:middle line:84% And then remember, the French reactors 01:12:54.700 --> 01:12:57.460 align:middle line:84% are being built by state-owned enterprises. 01:12:57.460 --> 01:12:59.680 align:middle line:90% So they notionally pay interest. 01:12:59.680 --> 01:13:01.940 align:middle line:84% But it's basically like their bond rate 01:13:01.940 --> 01:13:05.570 align:middle line:84% to the-- at the rate at which the state issues a bond. 01:13:05.570 --> 01:13:08.490 align:middle line:84% They don't have to pay the bank's profit. 01:13:08.490 --> 01:13:11.930 align:middle line:84% So delays in France do not cost anything 01:13:11.930 --> 01:13:13.770 align:middle line:84% like they cost in the United States. 01:13:13.770 --> 01:13:19.090 align:middle line:84% So there's all of these system-specific things 01:13:19.090 --> 01:13:19.950 align:middle line:90% that go in. 01:13:19.950 --> 01:13:23.050 align:middle line:84% Whether in the end you get a safer reactor or whatever, 01:13:23.050 --> 01:13:25.970 align:middle line:84% what is better, I don't think anyone has really studied it. 01:13:25.970 --> 01:13:30.030 align:middle line:84% It would be a wonderful PhD thesis to try to undertake. 01:13:30.030 --> 01:13:32.250 align:middle line:84% It would be hard to pull off, but yeah. 01:13:32.250 --> 01:13:34.810 align:middle line:90% 01:13:34.810 --> 01:13:35.310 align:middle line:90% All right. 01:13:35.310 --> 01:13:38.770 align:middle line:84% Any other questions before we wrap up? 01:13:38.770 --> 01:13:39.270 align:middle line:90% All right. 01:13:39.270 --> 01:13:41.530 align:middle line:84% So that's the international situation and the learning 01:13:41.530 --> 01:13:42.890 align:middle line:90% situation. 01:13:42.890 --> 01:13:45.810 align:middle line:90% And we will now move towards-- 01:13:45.810 --> 01:13:48.490 align:middle line:84% we're going to look briefly at whether small modular will 01:13:48.490 --> 01:13:49.890 align:middle line:90% fix the problem. 01:13:49.890 --> 01:13:52.610 align:middle line:84% We'll look at what wind and solar has done. 01:13:52.610 --> 01:13:54.930 align:middle line:90% And then we'll look at safety. 01:13:54.930 --> 01:13:56.680 align:middle line:90% All right. 01:13:56.680 --> 01:14:03.000 align:middle line:90%