WEBVTT 00:00:00.000 --> 00:00:02.445 align:middle line:90% [SQUEAKING] 00:00:02.445 --> 00:00:03.423 align:middle line:90% [RUSTLING] 00:00:03.423 --> 00:00:07.335 align:middle line:90% [CLICKING] 00:00:07.335 --> 00:00:18.100 align:middle line:90% 00:00:18.100 --> 00:00:20.710 align:middle line:84% RICHARD DE NUEFVILLE: Talks about a perspective LNG plant 00:00:20.710 --> 00:00:24.800 align:middle line:84% that I worked on with these colleagues, Cardin, 00:00:24.800 --> 00:00:26.680 align:middle line:90% Bourani, his thesis student-- 00:00:26.680 --> 00:00:33.220 align:middle line:84% Cardin and his thesis student, Bourani, and others 00:00:33.220 --> 00:00:37.240 align:middle line:84% with the Keppel Offshore and Marine Corporation, 00:00:37.240 --> 00:00:42.460 align:middle line:84% which is a big shipbuilding platformbuilding, 00:00:42.460 --> 00:00:48.700 align:middle line:84% metal bending government company in Singapore. 00:00:48.700 --> 00:00:51.430 align:middle line:84% And they had this as a perspective case. 00:00:51.430 --> 00:00:54.910 align:middle line:84% And they had a commission, a design. 00:00:54.910 --> 00:00:56.890 align:middle line:84% They were working on it for some oil company, 00:00:56.890 --> 00:01:01.090 align:middle line:84% and we took this on as a way of thinking about can one 00:01:01.090 --> 00:01:03.500 align:middle line:84% do better, as I think indeed we can. 00:01:03.500 --> 00:01:06.260 align:middle line:90% And that's what this is about. 00:01:06.260 --> 00:01:16.820 align:middle line:84% So it is, first of all, a parametric analysis, that is, 00:01:16.820 --> 00:01:19.280 align:middle line:84% looks at a significant range of factors. 00:01:19.280 --> 00:01:21.140 align:middle line:84% Instead of saying, this is the condition, 00:01:21.140 --> 00:01:24.500 align:middle line:84% it says, well, what if the economies of scale were longer? 00:01:24.500 --> 00:01:26.570 align:middle line:84% What if the uncertainty was different? 00:01:26.570 --> 00:01:29.340 align:middle line:84% What if learning was a different scale? 00:01:29.340 --> 00:01:32.270 align:middle line:84% So that it's designed to illustrate 00:01:32.270 --> 00:01:35.390 align:middle line:84% a range of possible conditions of how you might do it, 00:01:35.390 --> 00:01:38.880 align:middle line:84% rather than trying to specify a particular one. 00:01:38.880 --> 00:01:42.440 align:middle line:84% So I'd say design is a learning experience. 00:01:42.440 --> 00:01:44.930 align:middle line:90% I hope you'll find it so. 00:01:44.930 --> 00:01:50.510 align:middle line:84% So the general idea is to do MITEX 00:01:50.510 --> 00:01:54.740 align:middle line:84% if you want, that homework in a much more detailed 00:01:54.740 --> 00:01:57.180 align:middle line:90% and complicated situation. 00:01:57.180 --> 00:02:01.310 align:middle line:84% And to what conclusions can we talk about in terms 00:02:01.310 --> 00:02:02.810 align:middle line:90% of preferred strategies? 00:02:02.810 --> 00:02:05.810 align:middle line:84% Notice again that I'm talking about preferred strategies 00:02:05.810 --> 00:02:09.620 align:middle line:84% because there's tradeoffs about the cost and the risk. 00:02:09.620 --> 00:02:12.530 align:middle line:84% There are tradeoffs about the political extent. 00:02:12.530 --> 00:02:14.330 align:middle line:84% And there's a variety of other factors 00:02:14.330 --> 00:02:15.380 align:middle line:90% that are involved in it. 00:02:15.380 --> 00:02:20.610 align:middle line:84% But here the focus is on the economic aspects. 00:02:20.610 --> 00:02:24.140 align:middle line:84% So the key conclusion is a flexible 00:02:24.140 --> 00:02:27.630 align:middle line:84% can provide clear economic benefits. 00:02:27.630 --> 00:02:29.630 align:middle line:90% I don't say must. 00:02:29.630 --> 00:02:33.480 align:middle line:84% But it's pretty clear that they can provide it. 00:02:33.480 --> 00:02:39.290 align:middle line:84% These are pushed by discount rate effects and learning 00:02:39.290 --> 00:02:42.200 align:middle line:84% effects, discount rate because if you invest later on, 00:02:42.200 --> 00:02:44.810 align:middle line:84% it costs less in terms of present value. 00:02:44.810 --> 00:02:48.230 align:middle line:84% Learning, in fact, is that if you build several modules, 00:02:48.230 --> 00:02:50.360 align:middle line:84% the next time you do it, you'll avoid mistakes 00:02:50.360 --> 00:02:52.830 align:middle line:90% or improve design and so forth. 00:02:52.830 --> 00:02:59.720 align:middle line:84% And this comes with building often, but not necessarily, 00:02:59.720 --> 00:03:02.120 align:middle line:84% on some kind of modularity of design. 00:03:02.120 --> 00:03:04.580 align:middle line:84% And these factors, discount rate, 00:03:04.580 --> 00:03:06.980 align:middle line:84% learning effects, module design, work 00:03:06.980 --> 00:03:12.980 align:middle line:84% against the economies of scale, which encourage people to build 00:03:12.980 --> 00:03:15.450 align:middle line:84% for all future capacity right up front, 00:03:15.450 --> 00:03:21.230 align:middle line:84% which was part of the task of the MITEX example 00:03:21.230 --> 00:03:23.960 align:middle line:84% that you worked through, was that many people 00:03:23.960 --> 00:03:27.783 align:middle line:84% had the normal reflex was, oh, that's what we need. 00:03:27.783 --> 00:03:30.200 align:middle line:84% We're going to build for it right at the beginning and not 00:03:30.200 --> 00:03:31.320 align:middle line:90% mess around with it. 00:03:31.320 --> 00:03:34.760 align:middle line:84% But it turns out that common reaction 00:03:34.760 --> 00:03:38.360 align:middle line:90% is often anti-economic. 00:03:38.360 --> 00:03:41.960 align:middle line:84% And also the actual expected value of the CAPEX, 00:03:41.960 --> 00:03:43.880 align:middle line:84% of the project under uncertainty, 00:03:43.880 --> 00:03:47.550 align:middle line:84% is less than deterministic estimation. 00:03:47.550 --> 00:03:50.640 align:middle line:90% So that's where we're going. 00:03:50.640 --> 00:03:54.720 align:middle line:84% So I'm now giving you, taking from it, 00:03:54.720 --> 00:03:58.160 align:middle line:84% our presentation, their presentation, of the analysis 00:03:58.160 --> 00:04:01.900 align:middle line:84% has given for a particular project 00:04:01.900 --> 00:04:04.110 align:middle line:84% at a particular conference and so forth. 00:04:04.110 --> 00:04:13.850 align:middle line:84% And let's go, so the motivation, looking 00:04:13.850 --> 00:04:16.819 align:middle line:84% through design alternatives, the methodology, the results. 00:04:16.819 --> 00:04:19.700 align:middle line:90% 00:04:19.700 --> 00:04:24.070 align:middle line:84% So in location, we're talking about the state of Victoria, 00:04:24.070 --> 00:04:26.930 align:middle line:84% which is this bit here of Australia. 00:04:26.930 --> 00:04:29.520 align:middle line:84% This is where Sydney is, New South Wales. 00:04:29.520 --> 00:04:32.770 align:middle line:84% This is where Brisbane and the Gold Coast is. 00:04:32.770 --> 00:04:35.980 align:middle line:84% This is where the big mineral area 00:04:35.980 --> 00:04:39.640 align:middle line:84% is, or some of the big mineral on Western Australia, 00:04:39.640 --> 00:04:44.110 align:middle line:84% and Perth, and this is the great nothing, 00:04:44.110 --> 00:04:47.840 align:middle line:90% if I can put it that way. 00:04:47.840 --> 00:04:49.840 align:middle line:84% And so we're I don't want to-- there's 00:04:49.840 --> 00:04:51.820 align:middle line:90% Tasmania, Tassie, down here. 00:04:51.820 --> 00:04:54.970 align:middle line:90% Here is Melbourne. 00:04:54.970 --> 00:04:59.990 align:middle line:84% And there are various sites for distribution. 00:04:59.990 --> 00:05:03.880 align:middle line:84% So the plan was to be here, and their sites for distribution 00:05:03.880 --> 00:05:07.360 align:middle line:84% of the product, which I'll describe in a moment, 00:05:07.360 --> 00:05:09.595 align:middle line:90% at these sites one through five. 00:05:09.595 --> 00:05:12.590 align:middle line:90% 00:05:12.590 --> 00:05:20.870 align:middle line:84% So the whole idea is that you get LNG, liquid natural gas, 00:05:20.870 --> 00:05:25.050 align:middle line:84% from the offshore used to do a certain amount of processing 00:05:25.050 --> 00:05:29.090 align:middle line:84% and separate out the good stuff, get rid of the bad stuff. 00:05:29.090 --> 00:05:33.170 align:middle line:84% You somehow get it to land, and then you liquefy it. 00:05:33.170 --> 00:05:37.400 align:middle line:84% And it can be used for a variety of purposes. 00:05:37.400 --> 00:05:40.820 align:middle line:84% But the idea here in Australia was 00:05:40.820 --> 00:05:44.480 align:middle line:84% the project was to market it for the transportation sector, 00:05:44.480 --> 00:05:50.840 align:middle line:84% that it would be cleaner than diesel fuel and so forth, 00:05:50.840 --> 00:05:53.930 align:middle line:84% and it'd be in not only an environmental benefit, 00:05:53.930 --> 00:05:59.210 align:middle line:84% but it would be a good business for the oil companies, which 00:05:59.210 --> 00:06:01.880 align:middle line:84% unfortunately in the past, have often simply burned 00:06:01.880 --> 00:06:04.790 align:middle line:84% off the gas, flared it, which is neither 00:06:04.790 --> 00:06:09.950 align:middle line:84% profitable nor environmentally desirable. 00:06:09.950 --> 00:06:16.580 align:middle line:84% So it's viewed, in this case, as a transportation issue. 00:06:16.580 --> 00:06:23.390 align:middle line:84% In Europe, they use LNG, for home heating a lot and also 00:06:23.390 --> 00:06:24.710 align:middle line:90% for power plants. 00:06:24.710 --> 00:06:31.760 align:middle line:84% But they don't particularly need heat in Australia and South 00:06:31.760 --> 00:06:32.600 align:middle line:90% Australia. 00:06:32.600 --> 00:06:35.530 align:middle line:90% 00:06:35.530 --> 00:06:39.880 align:middle line:84% So there are three versions of the system 00:06:39.880 --> 00:06:42.370 align:middle line:90% I want to sketch out for you. 00:06:42.370 --> 00:06:47.230 align:middle line:84% The original design that Keppel Offshore and Marine 00:06:47.230 --> 00:06:51.670 align:middle line:84% had developed was to have one big plant with the idea 00:06:51.670 --> 00:06:56.890 align:middle line:84% then that the liquid product would be distributed 00:06:56.890 --> 00:06:59.830 align:middle line:84% to various sites, the five sites that I 00:06:59.830 --> 00:07:04.630 align:middle line:84% mentioned, and from there they would use the fuel tanker 00:07:04.630 --> 00:07:09.850 align:middle line:84% trucks to deliver goods and whatever they were doing, 00:07:09.850 --> 00:07:13.750 align:middle line:84% so one big plant, economies of scale, very much 00:07:13.750 --> 00:07:17.680 align:middle line:84% similar to the big plant for the water, 00:07:17.680 --> 00:07:20.020 align:middle line:84% except that it was not modular in any way. 00:07:20.020 --> 00:07:22.450 align:middle line:90% It was one big thing. 00:07:22.450 --> 00:07:26.640 align:middle line:84% So the first phase that we looked at, 00:07:26.640 --> 00:07:31.140 align:middle line:84% Cardin, me, and other, was to have modular plants. 00:07:31.140 --> 00:07:34.590 align:middle line:84% We said instead of having the whole size at once, 00:07:34.590 --> 00:07:40.110 align:middle line:84% we're going to say, what happens if we had modules of 25 tons, 00:07:40.110 --> 00:07:44.970 align:middle line:84% 50 tons, 75, 100 tons, and so forth? 00:07:44.970 --> 00:07:48.670 align:middle line:84% So what we could do about that was to phase it over time. 00:07:48.670 --> 00:07:52.090 align:middle line:84% So we start off small because the demand was small. 00:07:52.090 --> 00:07:56.760 align:middle line:84% And we'd add them, very similar to the MITEX homework 00:07:56.760 --> 00:08:01.230 align:middle line:84% two example is that, all right, what's the best way to do this? 00:08:01.230 --> 00:08:03.810 align:middle line:84% And we did this through a simulation analysis, 00:08:03.810 --> 00:08:06.960 align:middle line:90% as I'll point out. 00:08:06.960 --> 00:08:12.870 align:middle line:84% So we can play around flexibility as to size, one, 00:08:12.870 --> 00:08:18.332 align:middle line:84% and also as to timing because of the size. 00:08:18.332 --> 00:08:19.290 align:middle line:90% What is the right size? 00:08:19.290 --> 00:08:21.180 align:middle line:84% And when do we time the additions? 00:08:21.180 --> 00:08:22.950 align:middle line:84% But the third thing is we can also 00:08:22.950 --> 00:08:31.170 align:middle line:84% think about the flexible design so that we can locate the plant 00:08:31.170 --> 00:08:36.520 align:middle line:84% elsewhere, that as in many processes, electric power, 00:08:36.520 --> 00:08:38.909 align:middle line:84% for example, if you do everything 00:08:38.909 --> 00:08:41.640 align:middle line:84% in one large plant, power for example, 00:08:41.640 --> 00:08:44.130 align:middle line:84% you got to transmit it a long ways to where it's actually 00:08:44.130 --> 00:08:45.090 align:middle line:90% being used. 00:08:45.090 --> 00:08:50.000 align:middle line:84% Right now in New England, we are getting a fair amount of power 00:08:50.000 --> 00:08:53.120 align:middle line:84% from Northern Quebec, which is transmitted 00:08:53.120 --> 00:08:55.580 align:middle line:84% through high voltage lines all the way down to Boston 00:08:55.580 --> 00:08:57.500 align:middle line:90% or the Boston area. 00:08:57.500 --> 00:09:04.430 align:middle line:84% And that's maybe economical as compared to building a plant, 00:09:04.430 --> 00:09:05.720 align:middle line:90% but it costs a lot. 00:09:05.720 --> 00:09:09.560 align:middle line:84% So if you can avoid the transportation costs 00:09:09.560 --> 00:09:11.600 align:middle line:84% and do it locally, it might be better, 00:09:11.600 --> 00:09:14.870 align:middle line:84% so having a thought that maybe we'll 00:09:14.870 --> 00:09:18.050 align:middle line:84% have decentralized units of production. 00:09:18.050 --> 00:09:20.960 align:middle line:84% A standard issue, whether you're doing warehouses 00:09:20.960 --> 00:09:23.480 align:middle line:84% or electric power production or almost anything 00:09:23.480 --> 00:09:26.120 align:middle line:84% is not just how much capacity you have, 00:09:26.120 --> 00:09:28.350 align:middle line:84% but where do you have it at what time. 00:09:28.350 --> 00:09:34.310 align:middle line:84% So it's a generalizable example, applicable in thinking 00:09:34.310 --> 00:09:36.760 align:middle line:90% to many issues. 00:09:36.760 --> 00:09:40.410 align:middle line:84% So here's the schema for the design. 00:09:40.410 --> 00:09:43.740 align:middle line:84% And it's very similar to the-- the analysis. 00:09:43.740 --> 00:09:48.370 align:middle line:84% The schema is very similar to the garage case. 00:09:48.370 --> 00:09:51.300 align:middle line:84% So the first step one was to say, 00:09:51.300 --> 00:09:55.980 align:middle line:84% let's build the model of how this thing performs 00:09:55.980 --> 00:09:58.770 align:middle line:90% as one single large plant. 00:09:58.770 --> 00:10:03.900 align:middle line:84% And then because of the parametric one, 00:10:03.900 --> 00:10:08.040 align:middle line:84% we say let's suppose we know what the demand is going to be. 00:10:08.040 --> 00:10:11.100 align:middle line:84% We assume that, do a deterministic analysis, 00:10:11.100 --> 00:10:15.720 align:middle line:84% and think about what is the right size for the plant we 00:10:15.720 --> 00:10:17.520 align:middle line:90% ought to build. 00:10:17.520 --> 00:10:19.740 align:middle line:84% That, of course, depends upon the economies of scale 00:10:19.740 --> 00:10:22.110 align:middle line:84% so that depending upon the economy of the scale, 00:10:22.110 --> 00:10:25.270 align:middle line:84% you build it smaller and then expand it or not. 00:10:25.270 --> 00:10:27.690 align:middle line:84% This is the same kind of thing that man 00:10:27.690 --> 00:10:32.280 align:middle line:84% did in his very simple model that I showed you 00:10:32.280 --> 00:10:36.300 align:middle line:84% a couple of weeks ago, so deterministic model first. 00:10:36.300 --> 00:10:38.290 align:middle line:90% That's the base case. 00:10:38.290 --> 00:10:42.730 align:middle line:84% Secondly, we said we recognize this uncertainty, 00:10:42.730 --> 00:10:44.910 align:middle line:84% so we're going to think about this, 00:10:44.910 --> 00:10:48.000 align:middle line:84% and we're going to generate a lot of demand scenarios 00:10:48.000 --> 00:10:51.810 align:middle line:84% and see how the standard model performs under uncertainty 00:10:51.810 --> 00:10:54.870 align:middle line:84% or the deterministic model performance under uncertainty. 00:10:54.870 --> 00:10:58.590 align:middle line:84% And what you see is that once you recognize 00:10:58.590 --> 00:11:03.600 align:middle line:84% uncertainty, the value of the project changes, 00:11:03.600 --> 00:11:07.000 align:middle line:84% decreases because it's capacity limited in this case, 00:11:07.000 --> 00:11:11.590 align:middle line:84% and secondly, often the actual design changes. 00:11:11.590 --> 00:11:15.030 align:middle line:84% So this is very much like the garage case, 00:11:15.030 --> 00:11:17.490 align:middle line:84% except it's now applied into a much more complicated 00:11:17.490 --> 00:11:19.770 align:middle line:90% situation. 00:11:19.770 --> 00:11:24.870 align:middle line:84% Thirdly, we said given that we have the demand scenarios, 00:11:24.870 --> 00:11:28.710 align:middle line:84% let's think about having a flexible design that responded 00:11:28.710 --> 00:11:30.720 align:middle line:90% to the demand as it happened. 00:11:30.720 --> 00:11:34.590 align:middle line:84% So if it was growing well, transportation uses 00:11:34.590 --> 00:11:39.360 align:middle line:84% were ballooning and going way up, then we add more capacity. 00:11:39.360 --> 00:11:43.480 align:middle line:84% If people weren't using it, we would not do it. 00:11:43.480 --> 00:11:45.360 align:middle line:90% And so we now had a design-- 00:11:45.360 --> 00:11:46.390 align:middle line:90% two designs. 00:11:46.390 --> 00:11:50.670 align:middle line:84% One is thinking that we had flexible module designs, 00:11:50.670 --> 00:11:53.610 align:middle line:84% but we were going to do all the plant in one place. 00:11:53.610 --> 00:11:58.250 align:middle line:84% And the second one is to say, let's suppose it is possible 00:11:58.250 --> 00:12:01.470 align:middle line:84% and could be desirable to actually move 00:12:01.470 --> 00:12:03.880 align:middle line:90% some of the capacity to it. 00:12:03.880 --> 00:12:08.640 align:middle line:84% So we had a more limited version of only about the timing 00:12:08.640 --> 00:12:14.010 align:middle line:84% and a second version of design with timing and locational 00:12:14.010 --> 00:12:15.360 align:middle line:90% choices. 00:12:15.360 --> 00:12:18.840 align:middle line:84% And then we were thinking about overall calculating 00:12:18.840 --> 00:12:22.170 align:middle line:84% the value of flexibility, the economies of scale factor, 00:12:22.170 --> 00:12:25.280 align:middle line:90% and the sensitivity analysis. 00:12:25.280 --> 00:12:37.000 align:middle line:90%