WEBVTT 00:00:00.000 --> 00:00:02.970 align:middle line:90% [AUDIO LOGO] 00:00:02.970 --> 00:00:18.673 align:middle line:90% 00:00:18.673 --> 00:00:20.090 align:middle line:84% RICHARD DE NEUFVILLE: So I'm going 00:00:20.090 --> 00:00:26.240 align:middle line:84% to explore this basic trade-off between big versus small 00:00:26.240 --> 00:00:31.710 align:middle line:84% from an economic point of view in this short presentation. 00:00:31.710 --> 00:00:36.480 align:middle line:84% And it's all based on an old analysis from a simple case. 00:00:36.480 --> 00:00:41.400 align:middle line:84% And the real situation is more complex. 00:00:41.400 --> 00:00:47.310 align:middle line:84% But this analysis leads you to some simple, intuitive 00:00:47.310 --> 00:00:50.500 align:middle line:84% understanding of the trade-offs, which is what I want to use. 00:00:50.500 --> 00:00:55.320 align:middle line:84% And by the way, this was built upon the experience 00:00:55.320 --> 00:01:04.849 align:middle line:84% of building steel plants in India back half a century ago. 00:01:04.849 --> 00:01:08.140 align:middle line:84% So here's the picture when it all comes out to. 00:01:08.140 --> 00:01:09.790 align:middle line:84% Let me explain the picture first, then 00:01:09.790 --> 00:01:12.080 align:middle line:84% let's talk about the explanations. 00:01:12.080 --> 00:01:16.250 align:middle line:84% So I have on the horizontal axis is this alpha, 00:01:16.250 --> 00:01:17.810 align:middle line:90% which I've been talking about. 00:01:17.810 --> 00:01:19.700 align:middle line:84% And he ran the numbers because it's 00:01:19.700 --> 00:01:25.345 align:middle line:84% a simple analysis actually, which I'm not going to repeat. 00:01:25.345 --> 00:01:26.530 align:middle line:90% You can look it up. 00:01:26.530 --> 00:01:30.760 align:middle line:84% But it's not a crucial to the understanding. 00:01:30.760 --> 00:01:35.620 align:middle line:84% So that from our purposes, the economy of scale factors 00:01:35.620 --> 00:01:39.970 align:middle line:84% is in the range between 0.6 and 100. 00:01:39.970 --> 00:01:44.660 align:middle line:84% And so this is the economies of scale on the horizontal axis. 00:01:44.660 --> 00:01:49.390 align:middle line:84% On the vertical axis is how often, 00:01:49.390 --> 00:01:52.240 align:middle line:84% if you're going to have modules, how often should we 00:01:52.240 --> 00:01:55.270 align:middle line:90% repeat them to be efficient. 00:01:55.270 --> 00:01:57.610 align:middle line:84% And it's based upon the simple analysis 00:01:57.610 --> 00:01:59.710 align:middle line:90% that it's a linear growth. 00:01:59.710 --> 00:02:03.640 align:middle line:84% But again, having made that analysis, 00:02:03.640 --> 00:02:05.140 align:middle line:84% it gives you the trade-offs, which 00:02:05.140 --> 00:02:06.307 align:middle line:90% I'm going to be looking for. 00:02:06.307 --> 00:02:10.030 align:middle line:84% So here is the cycle time, which also translates in the size 00:02:10.030 --> 00:02:12.310 align:middle line:84% because, if you have 10% growth a year, 00:02:12.310 --> 00:02:14.380 align:middle line:84% if you're supposed to build it every eight years, 00:02:14.380 --> 00:02:16.130 align:middle line:90% it's talking about 80% growth. 00:02:16.130 --> 00:02:19.540 align:middle line:84% So the size of the plant, of the module, basically, 00:02:19.540 --> 00:02:20.830 align:middle line:90% is, in this way. 00:02:20.830 --> 00:02:24.010 align:middle line:84% And this is the economy of scale. 00:02:24.010 --> 00:02:27.940 align:middle line:90% And it shows the-- 00:02:27.940 --> 00:02:29.900 align:middle line:90% that's the setup for it. 00:02:29.900 --> 00:02:33.290 align:middle line:84% And now let me explain what's happening. 00:02:33.290 --> 00:02:35.690 align:middle line:90% It's the same graph as before. 00:02:35.690 --> 00:02:38.420 align:middle line:84% So it says if I have higher discount rates-- 00:02:38.420 --> 00:02:41.770 align:middle line:90% so this is from 10% to 15%-- 00:02:41.770 --> 00:02:45.190 align:middle line:90% higher discount rates-- sorry-- 00:02:45.190 --> 00:02:47.620 align:middle line:84% higher discount rates mean that you 00:02:47.620 --> 00:02:50.050 align:middle line:84% ought to build smaller, which says, again, 00:02:50.050 --> 00:02:53.740 align:middle line:84% that, in fact, if you're looking at it from here to here, 00:02:53.740 --> 00:03:00.250 align:middle line:84% you go from say 70% factor for economies of scale, 00:03:00.250 --> 00:03:04.250 align:middle line:84% you go from about 7 down to about 4 and bit. 00:03:04.250 --> 00:03:10.510 align:middle line:84% So you make it basically about 40% 00:03:10.510 --> 00:03:13.580 align:middle line:84% smaller is your size of your module. 00:03:13.580 --> 00:03:16.970 align:middle line:84% And on the other hand, if you have greater economies 00:03:16.970 --> 00:03:18.380 align:middle line:90% of scale, you build it larger. 00:03:18.380 --> 00:03:19.820 align:middle line:84% If you have no economies of scale, 00:03:19.820 --> 00:03:23.660 align:middle line:84% there's no benefit to building more than the existing demand. 00:03:23.660 --> 00:03:27.230 align:middle line:84% So you have the two factors that it's 00:03:27.230 --> 00:03:31.160 align:middle line:84% a trade-off between economies of scale and discount rate. 00:03:31.160 --> 00:03:35.190 align:middle line:84% Greater discount rate, build smaller. 00:03:35.190 --> 00:03:37.530 align:middle line:90% Greater economies of scale-- 00:03:37.530 --> 00:03:40.690 align:middle line:84% sorry-- greater economies of scale, 00:03:40.690 --> 00:03:42.450 align:middle line:84% which means smaller numbers here-- 00:03:42.450 --> 00:03:46.080 align:middle line:84% there's a little bit of cognitive dissonance. 00:03:46.080 --> 00:03:50.580 align:middle line:84% Smaller numbers mean greater economies of scale. 00:03:50.580 --> 00:03:52.080 align:middle line:90% You build larger. 00:03:52.080 --> 00:03:54.210 align:middle line:84% That's the basic trade-off, which I 00:03:54.210 --> 00:03:57.570 align:middle line:90% was trying to express earlier. 00:03:57.570 --> 00:04:01.890 align:middle line:84% So that is a graphical representation 00:04:01.890 --> 00:04:05.320 align:middle line:84% of the kind of thing that can happen. 00:04:05.320 --> 00:04:10.490 align:middle line:84% Now well, how sensitive is all this? 00:04:10.490 --> 00:04:13.850 align:middle line:84% So here's the thing that happens. 00:04:13.850 --> 00:04:18.160 align:middle line:84% So here is now the total cost of the plant, 00:04:18.160 --> 00:04:24.286 align:middle line:84% and the cycle time is how big you make it considering 00:04:24.286 --> 00:04:27.790 align:middle line:90% a constant growth per year. 00:04:27.790 --> 00:04:30.890 align:middle line:84% If I go from 4 to 8, I double the size. 00:04:30.890 --> 00:04:35.350 align:middle line:84% So the response is, the observation 00:04:35.350 --> 00:04:36.880 align:middle line:90% is the optimum is flat. 00:04:36.880 --> 00:04:40.000 align:middle line:90% What do I mean by that? 00:04:40.000 --> 00:04:44.500 align:middle line:84% If we're here, since these things can't be predicted that 00:04:44.500 --> 00:04:48.790 align:middle line:84% precisely, that going from say a size of based 00:04:48.790 --> 00:04:52.900 align:middle line:84% on four times the growth rate to up to about here, about 10 00:04:52.900 --> 00:04:55.390 align:middle line:84% or 12 times the growth rate, it doesn't 00:04:55.390 --> 00:04:56.870 align:middle line:90% make that much difference. 00:04:56.870 --> 00:05:00.730 align:middle line:84% So it makes a great deal of difference 00:05:00.730 --> 00:05:03.070 align:middle line:84% that you want to build some time in advance. 00:05:03.070 --> 00:05:09.130 align:middle line:84% But the optimal is not that significant. 00:05:09.130 --> 00:05:11.950 align:middle line:84% But you don't want to be building for 15 or 20 00:05:11.950 --> 00:05:15.280 align:middle line:84% years in advance because then it really starts hitting you. 00:05:15.280 --> 00:05:18.250 align:middle line:84% Now this was calculated for a particular set of values. 00:05:18.250 --> 00:05:22.990 align:middle line:90% And it changes at other times. 00:05:22.990 --> 00:05:28.930 align:middle line:84% But the notion is, well, if you look at a relatively small 00:05:28.930 --> 00:05:32.590 align:middle line:84% unit, you can have it better, not for 15 years, 00:05:32.590 --> 00:05:35.610 align:middle line:90% not for 20 years. 00:05:35.610 --> 00:05:38.510 align:middle line:84% So the implications here are, which 00:05:38.510 --> 00:05:41.000 align:middle line:84% I wanted to bring out through this, 00:05:41.000 --> 00:05:45.980 align:middle line:84% is that, for industries with economies of scale 00:05:45.980 --> 00:05:48.960 align:middle line:84% such as electric power, chemical processes, 00:05:48.960 --> 00:05:52.760 align:middle line:84% and so forth, and assuming that you have a steady growth 00:05:52.760 --> 00:05:57.310 align:middle line:84% indefinitely, small plants, those very small ones 00:05:57.310 --> 00:05:59.840 align:middle line:90% are uneconomical. 00:05:59.840 --> 00:06:02.090 align:middle line:84% The optimal size is associated with growth 00:06:02.090 --> 00:06:06.150 align:middle line:84% from between about 5 and 10 years of growth. 00:06:06.150 --> 00:06:11.120 align:middle line:84% And they're not especially sensitive to the higher end, 00:06:11.120 --> 00:06:15.620 align:middle line:84% and that's good because the forecasts are not accurate, so 00:06:15.620 --> 00:06:17.240 align:middle line:90% modest size. 00:06:17.240 --> 00:06:20.000 align:middle line:84% But now, this analysis assumes that growth is going 00:06:20.000 --> 00:06:23.800 align:middle line:90% to continue on indefinitely. 00:06:23.800 --> 00:06:27.490 align:middle line:84% But what happens if the growth stops, decreases? 00:06:27.490 --> 00:06:29.240 align:middle line:90% What's the risk? 00:06:29.240 --> 00:06:33.030 align:middle line:84% Well, you risk building capacity that's not needed. 00:06:33.030 --> 00:06:34.870 align:middle line:90% What can you do about it? 00:06:34.870 --> 00:06:36.630 align:middle line:84% Well, you build smaller, so there's 00:06:36.630 --> 00:06:38.580 align:middle line:90% less reliance on the future. 00:06:38.580 --> 00:06:43.440 align:middle line:84% This is similar but not the same as Eli's comments 00:06:43.440 --> 00:06:47.160 align:middle line:84% that the oil company, I think it was that he was working with, 00:06:47.160 --> 00:06:52.320 align:middle line:84% built two plants so as not to be vulnerable in case something 00:06:52.320 --> 00:06:55.530 align:middle line:84% happened, in this case, not that the capacity would stop 00:06:55.530 --> 00:06:59.970 align:middle line:84% but because they wanted to be able to be assured that they 00:06:59.970 --> 00:07:02.820 align:middle line:84% could meet their client's needs or their market share 00:07:02.820 --> 00:07:07.670 align:middle line:84% and, therefore, wanted to have them smaller. 00:07:07.670 --> 00:07:11.390 align:middle line:84% So the trade-offs that we're talking about here 00:07:11.390 --> 00:07:14.720 align:middle line:84% is between the higher cost per units for small plants 00:07:14.720 --> 00:07:18.350 align:middle line:84% when operating capacity versus saving by not 00:07:18.350 --> 00:07:20.480 align:middle line:90% paying for extra capacity. 00:07:20.480 --> 00:07:24.590 align:middle line:84% So there's a trade-off here that you 00:07:24.590 --> 00:07:27.680 align:middle line:84% need to think about very particular to the specifics 00:07:27.680 --> 00:07:32.360 align:middle line:84% of your situation between taking advantage of economies 00:07:32.360 --> 00:07:37.040 align:middle line:84% of scale, which is a typical driving force behind planning 00:07:37.040 --> 00:07:41.870 align:middle line:84% for a lot of industry, versus building it 00:07:41.870 --> 00:07:46.250 align:middle line:84% incrementally and smaller, which gives you advantages of it. 00:07:46.250 --> 00:07:52.340 align:middle line:84% And I will be talking later on about doing 00:07:52.340 --> 00:07:58.220 align:middle line:84% this in South Australia where they are building not only-- 00:07:58.220 --> 00:08:00.380 align:middle line:84% they had build a water supply plant, 00:08:00.380 --> 00:08:03.890 align:middle line:84% which was built at a terrific size for economies of scale. 00:08:03.890 --> 00:08:07.550 align:middle line:84% And they were thinking about doing the same thing for an LNG 00:08:07.550 --> 00:08:08.180 align:middle line:90% plant. 00:08:08.180 --> 00:08:13.250 align:middle line:84% And it illustrates the advantages of that trade-off 00:08:13.250 --> 00:08:19.160 align:middle line:84% and how you want to modulate between the capacity 00:08:19.160 --> 00:08:20.390 align:middle line:90% and when you implement it. 00:08:20.390 --> 00:08:22.920 align:middle line:90% 00:08:22.920 --> 00:08:25.560 align:middle line:84% So the takeaways from the discussion 00:08:25.560 --> 00:08:29.060 align:middle line:84% so far is, if there's no economy of scale, 00:08:29.060 --> 00:08:31.400 align:middle line:84% don't build in anticipation of demand. 00:08:31.400 --> 00:08:33.299 align:middle line:90% Expand as needed. 00:08:33.299 --> 00:08:37.880 align:middle line:84% If there are, then you might want 00:08:37.880 --> 00:08:40.070 align:middle line:84% to build somewhat in the future but not aggressively 00:08:40.070 --> 00:08:41.419 align:middle line:90% in the future. 00:08:41.419 --> 00:08:45.350 align:middle line:84% The more uncertainty there are, the small increments, 00:08:45.350 --> 00:08:49.550 align:middle line:84% when big capacity is not fully used or even needed, 00:08:49.550 --> 00:08:55.130 align:middle line:84% that is the risk of not needing it for whatever reason 00:08:55.130 --> 00:09:00.230 align:middle line:84% is a reason not to build as big as the engineering might 00:09:00.230 --> 00:09:00.980 align:middle line:90% like to have. 00:09:00.980 --> 00:09:03.660 align:middle line:90% 00:09:03.660 --> 00:09:07.150 align:middle line:84% So I want to now add in the effect of learning, 00:09:07.150 --> 00:09:18.460 align:middle line:84% which I mentioned emphasizes the effect of the discount rate. 00:09:18.460 --> 00:09:26.380 align:middle line:84% So here is the tail end of the Manne's analysis basically. 00:09:26.380 --> 00:09:30.120 align:middle line:84% So I've taken it from 0.6 to 0.95. 00:09:30.120 --> 00:09:35.350 align:middle line:84% There's basically a discrete representation. 00:09:35.350 --> 00:09:39.130 align:middle line:84% That's why they're bumps in these curves with no learning 00:09:39.130 --> 00:09:41.940 align:middle line:90% and at different interest rates. 00:09:41.940 --> 00:09:48.150 align:middle line:84% And I'm comparing this to another case with learning. 00:09:48.150 --> 00:09:51.640 align:middle line:84% And you have no learning, which is up here. 00:09:51.640 --> 00:09:53.640 align:middle line:84% And if you have higher amounts of learning, 00:09:53.640 --> 00:09:54.810 align:middle line:90% it really decreases it. 00:09:54.810 --> 00:10:00.720 align:middle line:84% So I can attend by this as an illustration of the idea 00:10:00.720 --> 00:10:06.720 align:middle line:84% that learning adds on to the effect of discount rates 00:10:06.720 --> 00:10:12.960 align:middle line:84% in terms of arguing against larger plants, which 00:10:12.960 --> 00:10:16.740 align:middle line:84% is all in the context that it's nice to have modules 00:10:16.740 --> 00:10:22.160 align:middle line:84% as a way of dealing with uncertainty. 00:10:22.160 --> 00:10:28.310 align:middle line:84% So the summary here is uncertainty is a main driver. 00:10:28.310 --> 00:10:31.430 align:middle line:84% Against are the pervasive economies of scale 00:10:31.430 --> 00:10:35.150 align:middle line:84% that impel designs for large capacity. 00:10:35.150 --> 00:10:39.420 align:middle line:84% And for are the discount rate and the learning, 00:10:39.420 --> 00:10:43.490 align:middle line:90% in addition to the uncertainty. 00:10:43.490 --> 00:10:46.720 align:middle line:84% So the big picture I want you to leave you with 00:10:46.720 --> 00:10:50.230 align:middle line:84% is that economies of scale are widespread. 00:10:50.230 --> 00:10:54.010 align:middle line:84% And in that way, they have been a mantra for engineering design 00:10:54.010 --> 00:10:56.470 align:middle line:90% to build big plans. 00:10:56.470 --> 00:11:02.260 align:middle line:84% And you see this very easily in the electric power 00:11:02.260 --> 00:11:04.780 align:middle line:84% industry where this has been the norm for a bigger 00:11:04.780 --> 00:11:09.010 align:middle line:84% plant as opposed to smaller plants, which you can then 00:11:09.010 --> 00:11:11.020 align:middle line:84% distribute around the landscape and don't have 00:11:11.020 --> 00:11:13.480 align:middle line:90% the same distribution costs. 00:11:13.480 --> 00:11:19.030 align:middle line:84% And it's also been the pattern of oil platforms of being 00:11:19.030 --> 00:11:22.030 align:middle line:84% large, as being more efficient, and so on as 00:11:22.030 --> 00:11:25.060 align:middle line:84% opposed to smaller ones, which reduce 00:11:25.060 --> 00:11:31.120 align:middle line:84% the risks or the uncertainties associated with the performance 00:11:31.120 --> 00:11:32.440 align:middle line:90% of the field. 00:11:32.440 --> 00:11:37.760 align:middle line:84% And in that context, smaller units may be better. 00:11:37.760 --> 00:11:44.630 align:middle line:84% So there's an argument for the investigation of smaller units. 00:11:44.630 --> 00:11:49.990 align:middle line:84% And our work with BP and subsequently with the one 00:11:49.990 --> 00:11:53.530 align:middle line:84% we did with Eli Dweck, in particular, 00:11:53.530 --> 00:11:59.290 align:middle line:84% was really interesting with BP because it 00:11:59.290 --> 00:12:05.710 align:middle line:84% was so against the idea of looking at smaller units 00:12:05.710 --> 00:12:11.820 align:middle line:84% but is proven actually quite productive for them. 00:12:11.820 --> 00:12:23.016 align:middle line:90%