WEBVTT Kind: captions Language: en-US 00:00:02.100 --> 00:00:05.420 Hello, everybody. Let’s get started. 00:00:05.420 --> 00:00:08.320 Thank you all for coming out to this week’s Earthquake 00:00:08.331 --> 00:00:11.660 Science Center Seminar. Today we have our very own 00:00:11.660 --> 00:00:18.710 Rob Skoumal, but before we get started, I have a few announcements. 00:00:18.710 --> 00:00:26.560 So we have an ESC Coffee Hour today here in Menlo Park, and then tomorrow 00:00:26.560 --> 00:00:31.710 in Moffett Field, both at 2:00 p.m. And then tomorrow, we have a public 00:00:31.710 --> 00:00:35.390 lecture on the Loma Prieta earthquake anniversary. 00:00:35.390 --> 00:00:40.960 So there will be one talk tomorrow at noon and then one at 7:00 p.m. 00:00:40.960 --> 00:00:47.130 here in Menlo Park. And then Susan and others are working 00:00:47.130 --> 00:00:53.510 very hard at getting the webstreaming and seminar all set up for Moffett Field, 00:00:53.510 --> 00:00:58.530 but next week, our seminar will still be here in Menlo. 00:00:58.530 --> 00:01:06.220 Okay. And so today, we have Rob. Most of you know him. 00:01:06.220 --> 00:01:11.660 He got his B.S. in marine science at Eckerd College and his M.S. 00:01:11.660 --> 00:01:15.110 and Ph.D. in geology at Miami University. 00:01:15.110 --> 00:01:19.840 And he has been a Mendenhall postdoc with us since 2016. 00:01:19.840 --> 00:01:22.120 So take it away, Rob. 00:01:22.120 --> 00:01:23.800 - Thanks a lot. And thanks, everyone, 00:01:23.800 --> 00:01:27.420 for making the journey up to Menlo today. 00:01:27.420 --> 00:01:31.119 So I’d like to talk about some of our current and ongoing research looking at 00:01:31.119 --> 00:01:34.030 induced earthquakes in the south-central United States. 00:01:34.030 --> 00:01:36.830 So the work that I’ll be talking about has benefited greatly from the 00:01:36.830 --> 00:01:40.900 contributions and discussions with pretty much our entire induced 00:01:40.900 --> 00:01:45.780 seismicity project team, including Andy, Elizabeth, Ole, Art, and Justin, 00:01:45.780 --> 00:01:48.280 as well as our friends at Miami University, 00:01:48.280 --> 00:01:51.710 Mike Brudzinski, Shannon Fasola, and Rosie Ries. 00:01:51.710 --> 00:01:55.610 Okay. So just a very quick outline of the stuff that I’ll be talking about today. 00:01:55.610 --> 00:01:59.720 First, I’ll be talking about kind of our efforts to characterize the 00:01:59.720 --> 00:02:02.760 earthquakes in the subsurface. So kind of improving detections 00:02:02.760 --> 00:02:05.500 and locations of these events. 00:02:05.500 --> 00:02:07.780 Then using those earthquakes, we’re going to start to map 00:02:07.780 --> 00:02:12.370 the seismogenic faults that are largely responsible for these earthquakes 00:02:12.370 --> 00:02:16.310 in the central United States. Then I’ll talk about how we can 00:02:16.310 --> 00:02:20.490 use those faults to actually estimate SHmax and other things 00:02:20.490 --> 00:02:24.470 that we can learn from those types of analyses. 00:02:24.470 --> 00:02:26.790 Once we’ve kind of characterized the earthquakes, then we hope to 00:02:26.790 --> 00:02:29.690 learn about them. So I’ll be talking about how we 00:02:29.690 --> 00:02:32.830 can distinguish different types of injection-induced earthquakes. 00:02:32.830 --> 00:02:36.080 So most of the earthquakes in the United States are induced by wastewater 00:02:36.080 --> 00:02:38.410 disposal, but hydraulic fracturing- induced earthquakes are probably 00:02:38.410 --> 00:02:43.239 kind of the secondary most common source. So distinguishing those is 00:02:43.240 --> 00:02:47.180 very important for our understanding of what’s actually going on. 00:02:47.180 --> 00:02:50.420 Then finally, I’ll talk about the industrial and geologic factors 00:02:50.420 --> 00:02:54.600 that control whether or not a well will induce earthquakes, 00:02:54.600 --> 00:02:57.340 which I think is a very important question. 00:02:57.349 --> 00:03:01.480 Okay. So to start with, I’ll talk about Oklahoma. 00:03:01.480 --> 00:03:06.060 So what we did was took this – all the cataloged earthquakes from the 00:03:06.060 --> 00:03:10.019 Oklahoma Geological Survey and the ComCat catalogs and ran it through a 00:03:10.019 --> 00:03:14.600 large-scale template-matching routine. So this took about 20,000 cataloged 00:03:14.600 --> 00:03:20.100 events and turned this into about a quarter million newly detected events. 00:03:20.100 --> 00:03:25.099 Okay, so with these, I took about a million manually selected phase picks 00:03:25.099 --> 00:03:27.769 on those cataloged events, and I cross-correlated those phase picks 00:03:27.769 --> 00:03:31.410 against all of our new detections in order to propagate those manual phase 00:03:31.410 --> 00:03:34.740 picks throughout our entire catalog. Once we had phase picks for our 00:03:34.740 --> 00:03:38.000 catalog, we then did a double-difference relocation 00:03:38.000 --> 00:03:42.080 using the correlation coefficients and differential phase arrival times 00:03:42.080 --> 00:03:45.370 in order to – the relative locations for these events. 00:03:45.370 --> 00:03:48.640 For that, I used GrowClust. It’s very similar to HypoDD. 00:03:48.640 --> 00:03:50.240 I don’t know if you’ve used that before. 00:03:50.240 --> 00:03:54.230 Just GrowClust has some algorithmic advantages that allows us to relocate 00:03:54.230 --> 00:03:59.300 catalogs that are a bit larger with greater ease. 00:04:00.330 --> 00:04:03.800 So an example of the relocations. So we did the relocations for 00:04:03.800 --> 00:04:06.069 the entire state of Oklahoma. This is just kind of a zoomed-in 00:04:06.069 --> 00:04:09.900 view around the magnitude 5 Cushing sequence. 00:04:09.900 --> 00:04:14.739 So, on the left here is the OGS catalog. So it’s kind of a distributed 00:04:14.739 --> 00:04:17.470 cluster of earthquakes. Now, previously, Martin Schoenball 00:04:17.470 --> 00:04:21.739 and Bill Ellsworth took that OGS catalog, did automatic phase picks, 00:04:21.739 --> 00:04:23.639 and then relocated with HypoDD. 00:04:23.639 --> 00:04:25.960 So their results in the – are in the middle. 00:04:25.960 --> 00:04:27.830 And then our results – again, using the template matching, 00:04:27.830 --> 00:04:30.840 using the manual phase picks, and GrowClust – are on the right. 00:04:30.840 --> 00:04:34.700 So in general, our catalog is pretty similar to what Martin and Bill found. 00:04:34.700 --> 00:04:37.320 Just we have an order of magnitude more events. 00:04:37.320 --> 00:04:40.920 All right, so with this, we can see additional fault complexity that’s 00:04:40.930 --> 00:04:43.250 going on, as well as we can see additional faults that 00:04:43.250 --> 00:04:45.949 could not have been identified before. 00:04:45.949 --> 00:04:49.479 All right, so fairly similar to other catalogs, just, you know, 00:04:49.479 --> 00:04:52.770 it’s a little bit better. So what we’d like to do is take 00:04:52.770 --> 00:04:57.099 this catalog and start to map these seismogenic faults that are responsible. 00:04:57.099 --> 00:05:01.210 Because the vast majority of earthquakes – induced earthquakes 00:05:01.210 --> 00:05:05.470 have occurred along mapped – faults that were previously unmapped. 00:05:05.470 --> 00:05:08.069 So if we’re interested in just this area, we could probably identify these 00:05:08.069 --> 00:05:11.009 faults by hand relatively easily. What we want to do is identify the 00:05:11.009 --> 00:05:15.280 faults over a much larger area, and that becomes a much more onerous task. 00:05:15.280 --> 00:05:19.139 So, to help us with that, we’re going to turn to an algorithm, right? 00:05:19.139 --> 00:05:22.219 So I’ll kind of briefly run through just kind of the – give an overview 00:05:22.220 --> 00:05:23.900 of how this algorithm works. 00:05:23.900 --> 00:05:29.280 And I’ll be using this cartoon map of seismicity to try to explain that. 00:05:29.760 --> 00:05:33.280 So our algorithm has two primary steps. The first step is that we cluster 00:05:33.289 --> 00:05:37.039 the seismicity. And we do this in a density-based approach. 00:05:37.039 --> 00:05:39.199 This approach requires two different parameters. We need to know 00:05:39.200 --> 00:05:43.220 some sort of distance metric, and we use that to define the radius of a circle. 00:05:43.220 --> 00:05:46.680 We then draw out those circles around all of our earthquakes. 00:05:46.680 --> 00:05:48.620 If an earthquake falls within the radius of another circle – 00:05:48.620 --> 00:05:51.499 or, another event, those events become kind of linked together, 00:05:51.500 --> 00:05:55.280 and we form these chains of earthquakes. 00:05:55.280 --> 00:05:59.000 Then we need to know the minimum number of events to define a cluster. 00:05:59.000 --> 00:06:03.540 All right, so if a cluster has fewer than those events, we discard that cluster. 00:06:03.540 --> 00:06:08.460 Okay. So every type of clustering algorithm is limited in the extent 00:06:08.460 --> 00:06:12.089 that the result is always dependent on your input parameters. 00:06:12.089 --> 00:06:16.499 Right, so in this case, we’re also limited by those input parameters. 00:06:16.499 --> 00:06:20.629 Our solution to try to limit that limitation is to use a whole bunch 00:06:20.629 --> 00:06:23.490 of different parameters. We follow the general scheme of, 00:06:23.490 --> 00:06:26.800 we consider large distances with large cluster sizes first. 00:06:26.800 --> 00:06:30.540 And then we iteratively consider smaller and smaller clusters. 00:06:30.540 --> 00:06:33.420 All right. Now, between each iteration, we’re going to take these clusters 00:06:33.420 --> 00:06:36.680 that we identified and try to identify faults that are within them. 00:06:36.680 --> 00:06:39.980 All right, so let’s imagine we had that cluster in the lower left there. 00:06:39.980 --> 00:06:42.680 What we’re going to do is then use random sample consensus. 00:06:42.680 --> 00:06:44.960 With this approach, we randomly selected two points – 00:06:44.960 --> 00:06:49.199 or, two earthquakes in that cluster. We draw a rectangle around them. 00:06:49.199 --> 00:06:51.969 We then see how many other earthquakes fall within that rectangle, 00:06:51.969 --> 00:06:54.580 and we define those earthquakes as inliers. 00:06:54.580 --> 00:06:57.979 All right, in order to determine which model works best, 00:06:57.979 --> 00:07:00.889 we look for the model that has the largest number of inliers. 00:07:00.889 --> 00:07:03.490 So this initial model, right – hopefully you can see it. 00:07:03.490 --> 00:07:06.449 You know, this does not represent the seismicity very well. 00:07:06.449 --> 00:07:09.059 But what we’re going to do is, we’re going to run through all possible 00:07:09.059 --> 00:07:14.050 combinations to try to evaluate and see what kind of fault plane 00:07:14.050 --> 00:07:16.479 will represent the seismicity the best. 00:07:16.479 --> 00:07:19.460 Once we identify that best case, we run some quality control checks. 00:07:19.460 --> 00:07:23.789 If it passes those checks, those faults become associated with this kind of 00:07:23.789 --> 00:07:28.069 fault and are sort of removed from this catalog. 00:07:28.069 --> 00:07:30.729 We then apply – oops. We then apply random 00:07:30.729 --> 00:07:33.560 sample consensus iteratively. So you can see that there’s these 00:07:33.560 --> 00:07:38.240 kind of five earthquakes that are still left. We’ll apply random sample 00:07:38.240 --> 00:07:41.499 consensus on those events, and maybe we’ll identify a fault like this. 00:07:41.499 --> 00:07:43.339 Once we’ve gone through all those earthquakes, 00:07:43.339 --> 00:07:46.289 we’ll return to that clustering. We’ll use those smaller set of 00:07:46.289 --> 00:07:50.009 parameters. Maybe we’ll identify that cluster on the top right there. 00:07:50.009 --> 00:07:51.849 Then we’ll repeat the random sample consensus. 00:07:51.849 --> 00:07:55.610 All right, so this is obviously a very elementary view of what’s going on. 00:07:55.610 --> 00:07:58.220 There’s some nuances that I’m not describing. 00:07:58.220 --> 00:08:00.699 And this also only shows it in two dimensions. All right, 00:08:00.700 --> 00:08:04.120 but hopefully this kind of gives you an idea of what we’re doing here. 00:08:05.260 --> 00:08:09.259 So here’s an application. So this is that Cushing sequence that I showed earlier. 00:08:09.260 --> 00:08:12.760 This was our map of relocated events in black. And, on top of this, 00:08:12.760 --> 00:08:17.440 I’m going to show our automatically detected faults by these red lines. 00:08:17.449 --> 00:08:19.989 So would you have picked slightly different faults? 00:08:19.989 --> 00:08:22.889 All right, I hope so. All right, I know there’s some faults that I’d like to tweak. 00:08:22.889 --> 00:08:25.909 You know, there’s some that – you know, is that even a fault, right? 00:08:25.909 --> 00:08:29.789 But, in general, these look like faults that we might pick. 00:08:29.789 --> 00:08:33.860 Right, not just for this area, but for the entire state. 00:08:33.860 --> 00:08:36.880 We can do other things to try to, you know, improve our confidence 00:08:36.889 --> 00:08:38.300 in these events. We can compare them to 00:08:38.300 --> 00:08:42.409 focal mechanisms, which, in general, agree pretty well. 00:08:42.409 --> 00:08:45.220 We can also start to look at the orientations of these faults. 00:08:45.220 --> 00:08:47.060 Right, so when we look at the orientations, I’m going to 00:08:47.060 --> 00:08:50.810 start to zoom out and look at larger areas. 00:08:50.810 --> 00:08:53.819 So here are three rose diagrams showing the orientations of 00:08:53.819 --> 00:08:56.750 our automatically detected faults, with each diagram representing 00:08:56.750 --> 00:08:59.500 about a 400-square-kilometer area. 00:08:59.500 --> 00:09:02.519 Now, in the center of each of these rose – or, the center of 00:09:02.519 --> 00:09:06.709 each of these areas, there’s an SHmax measurement, which was determined 00:09:06.709 --> 00:09:11.620 from drilling-induced tensile fractures. All right, so if we estimate the optimal 00:09:11.620 --> 00:09:15.800 orientation from that SHmax, shown in kind of pink here, we see 00:09:15.800 --> 00:09:20.800 there’s a good agreement between the optimal orientations and our faults. 00:09:20.800 --> 00:09:23.040 Now, this kind of raises an interesting question. 00:09:23.040 --> 00:09:27.130 If the vast majority of our faults are optimally oriented to the original stress 00:09:27.130 --> 00:09:33.829 field, can we use our fault orientations to estimate the stress orientations? 00:09:33.829 --> 00:09:37.519 And I would argue that we could. All right, so here’s an example 00:09:37.519 --> 00:09:41.360 of a different area. So this is the SHmax measurement for 00:09:41.360 --> 00:09:43.770 one of those areas that I showed earlier. 00:09:43.770 --> 00:09:48.220 And around that are earthquakes and the faults that we identified. 00:09:48.220 --> 00:09:50.310 So what I’d like to do is compare our fault orientations – 00:09:50.310 --> 00:09:54.900 all right, those red lines against this SHmax measurement. 00:09:54.900 --> 00:09:57.500 All right, so to do that, we’ll be doing these little geographical bins 00:09:57.500 --> 00:10:01.120 and kind of running them over our fault areas to look at what types 00:10:01.120 --> 00:10:04.870 of orientations we have. And those results are shown here by these colors. 00:10:04.870 --> 00:10:09.279 All right, so the yellow areas represent that our faults are 30 degrees away 00:10:09.279 --> 00:10:13.120 from the SHmax measurement. Right, so assuming a coefficient 00:10:13.120 --> 00:10:17.139 of friction of 0.6 and that these are vertical strike-slip faults, 00:10:17.139 --> 00:10:21.430 30 degrees would be optimally oriented. So this is an example from central 00:10:21.430 --> 00:10:26.620 Oklahoma, where pretty much the stress is all kind of in this east-west direction 00:10:26.620 --> 00:10:32.199 within 5 degrees or so, which is kind of on the order of our uncertainty. 00:10:32.199 --> 00:10:35.069 But there are areas of Oklahoma that have a little bit 00:10:35.069 --> 00:10:40.700 more exciting stress regimes. So this is an area in northern Oklahoma. 00:10:40.700 --> 00:10:43.620 In this area, there are actually three stress measurements, kind of going 00:10:43.620 --> 00:10:49.810 clockwise from the top left here, this is 79 degrees, 68 degrees, and 59 degrees. 00:10:49.810 --> 00:10:52.010 So we’re going to apply the same kind of methodology, but this time, we’re 00:10:52.010 --> 00:10:56.620 going to be using that top left SHmax measurement as a reference point. 00:10:56.620 --> 00:10:59.519 So when we do that, we see that the faults over there are 30 degrees 00:10:59.519 --> 00:11:02.779 away from that. But, as you move away from that measurement, for every 00:11:02.779 --> 00:11:06.709 1 kilometer you travel, there’s a 1 degree rotation in the stress field. 00:11:06.709 --> 00:11:10.670 So that may not be very large, depending on where you work 00:11:10.670 --> 00:11:14.820 in the field. In California, they have some pretty extreme rotations. 00:11:14.820 --> 00:11:18.360 But, by Oklahoma standards, you know, this is – this is pretty exciting stuff. 00:11:18.360 --> 00:11:22.730 A 1-degree rotation on a kilometer basis is pretty large. 00:11:22.730 --> 00:11:24.839 So our measurements are in great agreement with these 00:11:24.839 --> 00:11:29.189 other actual measurements. Also, you can start to see that we can, 00:11:29.189 --> 00:11:34.199 you know, estimate SHmax in areas that do not have measurements. 00:11:34.199 --> 00:11:38.089 So this is an interesting area because it’s right next to this Nemaha Fault. 00:11:38.089 --> 00:11:40.240 So we can – and we’ll start to look at questions. 00:11:40.240 --> 00:11:44.100 You know, is this rotation related to that ridge? 00:11:44.100 --> 00:11:46.920 You know, I think that there’s a lot that we can do now that we have 00:11:46.920 --> 00:11:51.139 these orientations. All right, but just as a reminder, all right, 00:11:51.140 --> 00:11:54.540 our stress measurements come only from earthquake locations. 00:11:54.540 --> 00:11:57.200 We don’t know anything about the mechanisms or any kind of 00:11:57.200 --> 00:12:00.399 borehole measurements. All of this information is gleaned 00:12:00.399 --> 00:12:04.300 just from the earthquake locations and are automatically determined faults. 00:12:04.300 --> 00:12:06.699 Which I think is – which is pretty cool, right? 00:12:06.699 --> 00:12:10.100 So even if we have a fairly limited but high-quality data set, 00:12:10.100 --> 00:12:12.580 I think we can still do some pretty cool science. 00:12:12.580 --> 00:12:16.740 Okay. So what I’ve shown so far are just, you know, two-dimensional 00:12:16.740 --> 00:12:20.420 fault lines that represent these vertical strike-slip faults. 00:12:20.420 --> 00:12:23.460 But obviously not all earthquakes are vertical strike-slip faults. 00:12:23.460 --> 00:12:26.720 So, to represent those, we need to look at fault planes. 00:12:26.720 --> 00:12:29.760 All right, so this is an application of the same kind of algorithm that 00:12:29.779 --> 00:12:31.939 I showed before, but we’ve scaled it up one dimension, 00:12:31.940 --> 00:12:34.880 so now we can tackle three-dimensional fault planes. 00:12:34.880 --> 00:12:39.260 All right, and so we applied this to Elizabeth’s Prague catalog. 00:12:39.280 --> 00:12:44.140 So, following the magnitude 5.7 in Prague, a local network was deployed. 00:12:44.140 --> 00:12:47.360 And Elizabeth spent a lot of time relocating the events, 00:12:47.360 --> 00:12:50.800 and her catalog are shown by these black dots. 00:12:50.800 --> 00:12:53.560 All right, now our fault planes are represented by these lines. 00:12:53.569 --> 00:12:57.069 It’s pretty difficult to represent planes with lines. 00:12:57.069 --> 00:13:01.029 So I have this little animation. All right, so in this view, they’re the 00:13:01.029 --> 00:13:05.009 same kind of map view, but when I run this animation, all right, it’ll be going 00:13:05.009 --> 00:13:09.100 from kind of the map view into profile view, and then it’ll rotate around. 00:13:09.100 --> 00:13:12.459 All right, so I think we can see more complexity with this type of analysis. 00:13:12.459 --> 00:13:15.420 There’s certainly still a lot of advancements that we want to make 00:13:15.420 --> 00:13:19.509 with this algorithm, but we can go very quickly from just having 00:13:19.509 --> 00:13:24.180 a catalog of events to start to identify these fault planes very rapidly. 00:13:24.180 --> 00:13:27.180 So this is – this is highly scalable. We can, you know, identify planes 00:13:27.180 --> 00:13:31.459 from millions of earthquakes within about a minute or so. 00:13:31.460 --> 00:13:34.040 So I think it’s a pretty promising technique. 00:13:34.040 --> 00:13:37.380 All right. Now, for the – for this actual Prague – oops. 00:13:37.380 --> 00:13:40.420 For this Prague sequence – and Elizabeth found a pretty 00:13:40.439 --> 00:13:43.790 interesting observation. All right, so Elizabeth had to look 00:13:43.790 --> 00:13:46.779 at shear wave splitting to estimate the SHmax orientations. 00:13:46.779 --> 00:13:49.959 The idea that the upper crust is composed of these numerous fluid-filled 00:13:49.959 --> 00:13:54.879 cracks that will be preferentially oriented in the direction of SHmax. 00:13:54.879 --> 00:13:58.970 All right, so her results are shown by these little green lines. 00:13:58.970 --> 00:14:02.370 And she found that, in general, the SHmax is fairly constant – 00:14:02.370 --> 00:14:04.230 this standard east-west direction. 00:14:04.230 --> 00:14:08.279 Now, if we compare our fault planes against those orientations, 00:14:08.279 --> 00:14:12.110 we see that this kind of primary fault that hosted the magnitude 5.7 00:14:12.110 --> 00:14:16.779 is optimally oriented to those – to those stresses estimates. 00:14:16.779 --> 00:14:19.499 But there are other faults that are poorly oriented. 00:14:19.499 --> 00:14:24.560 Right, they’re pointed 30 degrees away from the optimal orientation. 00:14:24.560 --> 00:14:28.660 All right, so I think this – you know, is a good reminder that, although 00:14:28.660 --> 00:14:32.559 we may be able to identify the faults and determine the stress state such that 00:14:32.559 --> 00:14:35.660 we could maybe determine the likelihood of certain faults that 00:14:35.660 --> 00:14:40.810 would slip, we can still have these faults that aren’t poorly – gosh – that are 00:14:40.810 --> 00:14:45.209 poorly oriented that are still close to failure that can still induce earthquakes. 00:14:45.209 --> 00:14:48.209 All right, so we just – I think this is a great data set, and we can do 00:14:48.209 --> 00:14:50.290 a lot of cool things with it. We just have to be careful in how 00:14:50.290 --> 00:14:54.430 we interpret those results and how other people interpret those results. 00:14:54.430 --> 00:14:58.060 Okay. So most of the earthquakes in Oklahoma and in the central 00:14:58.060 --> 00:15:00.970 United States are induced by wastewater disposal, but hydraulic 00:15:00.970 --> 00:15:06.710 fracturing is probably the second most likely cause of earthquakes. 00:15:06.710 --> 00:15:09.680 So a couple of years ago in my seminar, I talked about kind of our plans to 00:15:09.680 --> 00:15:13.029 look at this in Oklahoma, and I’m talking about the results. 00:15:13.029 --> 00:15:16.029 It turned out that there’s a lot more [laughs] hydraulic fracturing- 00:15:16.029 --> 00:15:17.790 induced earthquakes than we thought. 00:15:17.790 --> 00:15:24.180 So the project kind of became a pretty large beast by the end. 00:15:24.180 --> 00:15:27.490 So what this is showing – each one of these boxes is a region that we found 00:15:27.490 --> 00:15:31.290 to be dominated by hydraulic fracturing-induced seismicity. 00:15:31.290 --> 00:15:34.130 On the right, this represents each of these geographical areas, 00:15:34.130 --> 00:15:37.220 and it shows the percentage of earthquakes in those areas 00:15:37.220 --> 00:15:39.839 that we’re able to attribute directly to hydraulic fracturing. 00:15:39.839 --> 00:15:43.829 So, in many of these areas, more than 80% of the earthquakes 00:15:43.829 --> 00:15:47.879 we could tie to individual wells. Which I think is a pretty remarkable 00:15:47.879 --> 00:15:51.910 feat considering all the uncertainties in the industry data. 00:15:51.910 --> 00:15:54.990 There are, you know, certainly misreported wells 00:15:54.990 --> 00:15:58.680 and wells that are missing. So to get – still to get associations 00:15:58.680 --> 00:16:02.050 of 80-plus percent I think is pretty neat. 00:16:02.050 --> 00:16:05.839 So you may notice that there’s this big blob of seismicity in kind of 00:16:05.839 --> 00:16:09.329 central northern Oklahoma that we do not have hydraulic fracturing 00:16:09.329 --> 00:16:12.430 induced earthquakes. So this is largely because there’s just 00:16:12.430 --> 00:16:16.139 so many wastewater disposal-induced earthquakes, it kind of obscures the 00:16:16.139 --> 00:16:19.379 signal of hydraulic fracturing. All right, so in kind of even the 00:16:19.379 --> 00:16:23.660 best-case scenarios for the wells, we usually just have a start day and an end 00:16:23.660 --> 00:16:26.959 day of the stimulation with really no knowledge of what occurs in between. 00:16:26.959 --> 00:16:33.089 So it makes it pretty difficult to, you know, drive statistically significant 00:16:33.089 --> 00:16:36.779 correlations between earthquakes and those wells with just those data. 00:16:36.779 --> 00:16:38.630 So it’s expected that we do have hydraulic fracturing-induced 00:16:38.630 --> 00:16:43.290 earthquakes within that big blob. Just they’re kind of hidden up. 00:16:43.290 --> 00:16:46.769 So, just from our results, we found that there were 274 wells that 00:16:46.769 --> 00:16:51.339 had induced earthquakes. And kind of after this time 00:16:51.339 --> 00:16:56.139 window of our study, there have been larger – kind of magnitude 3.7. 00:16:56.139 --> 00:16:59.319 But so they’re – the magnitudes are fairly small from hydraulic 00:16:59.319 --> 00:17:01.029 fracturing-induced seismicity in the United States. 00:17:01.029 --> 00:17:04.450 So the largest earthquake from hydraulic fracturing that we’ve – 00:17:04.450 --> 00:17:06.810 that we’re confident about in the United States is 00:17:06.810 --> 00:17:10.510 a magnitude 4.0 in Texas. So fairly small. 00:17:10.510 --> 00:17:14.510 Now, I’m not trying to downplay the significance or the risk or hazard 00:17:14.510 --> 00:17:17.199 of hydraulic fracturing-induced earthquakes. Internationally, 00:17:17.199 --> 00:17:21.539 they have had larger cases. So, in Canada – western Canada, 00:17:21.539 --> 00:17:24.639 they’ve had earthquakes up in the mid-magnitude 4s. 00:17:24.639 --> 00:17:29.140 And overseas in China, there have been deaths related to hydraulic fracturing. 00:17:29.140 --> 00:17:32.240 So the past December, there was a magnitude 5.2 that killed a couple 00:17:32.240 --> 00:17:34.779 people that was induced by hydraulic fracturing. 00:17:34.779 --> 00:17:38.150 And more recently, this past June, there was a magnitude 5.8 00:17:38.150 --> 00:17:40.730 that killed about a dozen people. And it’s also been suggested 00:17:40.730 --> 00:17:42.799 that earthquake was also induced by hydraulic fracturing. 00:17:42.799 --> 00:17:45.809 But in the United States, we’re pretty fortunate that the 00:17:45.809 --> 00:17:48.720 earthquakes from hydraulic fracturing tend to be quite small. 00:17:48.720 --> 00:17:53.900 Now I think one of the reasons for these relatively small events is that both the 00:17:53.900 --> 00:17:58.320 industry and operators are aware of these events, and they’re interested in 00:17:58.320 --> 00:18:03.230 learning what’s going on and how to reduce the likelihood of those events. 00:18:03.230 --> 00:18:06.600 So fairly recently, the Oklahoma Corporation Commission, 00:18:06.600 --> 00:18:12.121 which is the body that’s in charge of regulations in Oklahoma, put in some 00:18:12.121 --> 00:18:15.230 regulations targeting the SCOOP and STACK play in kind of central and 00:18:15.230 --> 00:18:19.100 western Oklahoma, which is where a lot of induced earthquakes have occurred. 00:18:19.100 --> 00:18:23.190 So, from these regulations – they kind of follow a stoplight pattern, 00:18:23.190 --> 00:18:26.610 such that, if a well induces a magnitude 2, the operator 00:18:26.610 --> 00:18:29.180 has to implement a predetermined mitigation plan. 00:18:29.180 --> 00:18:33.039 If a magnitude 3 is induced, they have to pause operations for a few hours. 00:18:33.039 --> 00:18:36.190 And they induce a magnitude 3.5, they have to suspend operations and 00:18:36.190 --> 00:18:38.960 wait for the Corporation Commission to give them approval to resume. 00:18:38.960 --> 00:18:43.180 All right, so if these regulations had been in effect, you know, 00:18:43.180 --> 00:18:46.570 during our study window, you know, hundreds of wells would have 00:18:46.570 --> 00:18:50.910 triggered various levels of this. But the red light of the magnitude 3.5 00:18:50.910 --> 00:18:53.330 would have only been reached a couple times. 00:18:53.330 --> 00:18:58.330 So, after these regulations went into effect, there have been earthquakes 00:18:58.330 --> 00:19:03.240 that have been induced, and the wells have been suspended. 00:19:03.240 --> 00:19:08.000 So these are – from what we can tell, these regulations have been successful 00:19:08.000 --> 00:19:12.559 at at least limiting or reducing the number of hydraulic 00:19:12.559 --> 00:19:13.870 fracturing-induced earthquakes. 00:19:13.870 --> 00:19:16.980 Now, it’s interesting because these regulations only apply 00:19:16.980 --> 00:19:20.940 to the SCOOP and STACK play. We have this kind of Arkoma Basin 00:19:20.940 --> 00:19:23.490 in the southeastern Oklahoma, where we have, you know, 00:19:23.490 --> 00:19:25.309 very prominent cases of hydraulic fracturing-induced seismicity. 00:19:25.309 --> 00:19:29.220 The regulations do not apply to these cases. 00:19:29.220 --> 00:19:31.960 So even if they induce a magnitude 3.5, these – 00:19:31.960 --> 00:19:36.740 any wells sort of – this down there, that would not be affected. 00:19:36.740 --> 00:19:39.620 Okay, so I mentioned before that, you know, all we really have 00:19:39.620 --> 00:19:42.590 is a start day and an end day of these stimulation windows. 00:19:42.590 --> 00:19:47.450 The exception is that we have information about two different wells. 00:19:47.450 --> 00:19:50.870 So, in this Manning-Irene well – this is a horizontal well, 00:19:50.870 --> 00:19:54.510 and the operator voluntarily provided us with the stimulation times. 00:19:54.510 --> 00:19:57.750 So we don’t know, you know, the volumes or rates, but we know 00:19:57.750 --> 00:20:00.440 when they were stimulating in kind of a minute-by-minute resolution. 00:20:00.440 --> 00:20:02.860 All right, so with that, we can make some pretty interesting 00:20:02.860 --> 00:20:05.320 temporal correlations, and that’s very important. 00:20:05.320 --> 00:20:08.840 Just to the northwest, there’s the Eagleton Well Pad. 00:20:08.840 --> 00:20:12.049 So this is a vertical well. And we were – we were able 00:20:12.049 --> 00:20:14.700 to determine injection rates because those rates were published 00:20:14.700 --> 00:20:17.660 in an Open-File report. So I was able to digitize those data, 00:20:17.660 --> 00:20:22.930 and then we were interested in creating a poroelastic model for those earthquakes. 00:20:22.930 --> 00:20:27.559 All right, so our earthquake catalog – these gray symbols are the initial 00:20:27.559 --> 00:20:31.570 earthquake locations. And the colored symbols are improved locations. 00:20:31.570 --> 00:20:33.990 So southern Oklahoma, the geology is pretty complex. 00:20:33.990 --> 00:20:37.090 There’s not a lot that’s known about the velocity model, 00:20:37.090 --> 00:20:41.220 so it’s definitely a struggle to locate the earthquakes. 00:20:41.220 --> 00:20:43.380 So our earthquake uncertainties are still kind of on the order of about 00:20:43.380 --> 00:20:46.270 a kilometer or so, which is about the distance away from the well. 00:20:46.270 --> 00:20:48.820 All right, so kind of – but we’re still pretty confident the earthquakes 00:20:48.820 --> 00:20:51.440 are within a kilometer of these stimulations. 00:20:51.440 --> 00:20:56.679 So what we’d like to do is take this – there it is – take this Eagleton Well Pad 00:20:56.679 --> 00:21:01.149 and start to model to see what actually happened with these earthquakes. 00:21:01.149 --> 00:21:04.050 So here are the actual injection parameters that I was able to digitize. 00:21:04.050 --> 00:21:06.350 All right, so there are four stimulations that were done. 00:21:06.350 --> 00:21:09.390 So in a vertical well – so this first stimulation was the deepest. 00:21:09.390 --> 00:21:11.860 And then each sequential stimulation represented by 00:21:11.860 --> 00:21:15.000 these lines were slightly shallower. 00:21:15.000 --> 00:21:19.110 Our earthquakes represented by – represented by these gray boxes, 00:21:19.110 --> 00:21:21.370 with the largest magnitude occurring in this first stimulation 00:21:21.370 --> 00:21:22.930 with about a magnitude 3.2. 00:21:22.930 --> 00:21:25.830 All right, so when we stimulate, we had a bunch of earthquakes 00:21:25.830 --> 00:21:28.510 following the stimulation. Kind of get this – kind of the aftershock 00:21:28.510 --> 00:21:33.429 decay following a stimulation. So this is a fairly typical example of 00:21:33.429 --> 00:21:37.211 hydraulic fracturing-induced seismicity, where the earthquakes and the 00:21:37.211 --> 00:21:41.730 aftershocks only last a few days or so. We do have examples where aftershocks 00:21:41.730 --> 00:21:45.610 from hydraulic fracturing can last months. But those are fairly rare. 00:21:45.610 --> 00:21:49.480 Okay, so what we’d like to do is take these injection parameters, 00:21:49.480 --> 00:21:53.510 take these earthquakes, and estimate some geologic parameters and provided 00:21:53.510 --> 00:21:58.470 these to Andy Barbour, who was able to put this into a relatively simple axial 00:21:58.470 --> 00:22:01.929 symmetrical model to estimate the poroelastic and pore pressure effects. 00:22:01.929 --> 00:22:04.600 All right, so those results are shown here. 00:22:04.600 --> 00:22:08.070 So this is just showing the estimates in the Lower Viola, which is the 00:22:08.070 --> 00:22:13.890 formation that being stimulated. These yellow dots are our earthquakes. 00:22:13.890 --> 00:22:17.770 And then these – the black dashed line is the model poroelastic effects. 00:22:17.770 --> 00:22:21.309 So we observed that there’s a pretty good correlation between 00:22:21.309 --> 00:22:25.190 the onset of our earthquakes and the poroelastic stress increases. 00:22:25.190 --> 00:22:28.570 And we also noticed that the pore pressure effects were 00:22:28.570 --> 00:22:31.940 only significant within 100 meters or so of the well. 00:22:31.940 --> 00:22:34.880 All right, so if these earthquakes were induced by pore pressure, 00:22:34.880 --> 00:22:37.669 the earthquakes would be very close to the well itself. 00:22:37.669 --> 00:22:41.600 Whereas, if it was poroelastic effects, the earthquakes could be, you know, 00:22:41.600 --> 00:22:45.150 further way, on the order of hundreds of meters away. 00:22:45.150 --> 00:22:49.789 So we do have other examples in the Appalachian Basin, where, 00:22:49.789 --> 00:22:51.950 following the stimulation, within minutes, you know, 00:22:51.950 --> 00:22:54.529 we’ll have earthquakes about 700 meters away. 00:22:54.529 --> 00:22:57.419 So, in those cases, we do believe that poroelastic effects are the 00:22:57.419 --> 00:23:01.330 primary cause of those seismicity, but we can’t rule out pore pressure 00:23:01.330 --> 00:23:04.540 in this – in this particular case. 00:23:04.540 --> 00:23:08.380 Okay, so up until now, I think a lot of the stuff that I’ve been talking about 00:23:08.380 --> 00:23:11.330 is sort of reactionary, right? So people caused earthquakes. 00:23:11.330 --> 00:23:13.700 We want to learn about those earthquakes as much as we can. 00:23:13.700 --> 00:23:17.830 The end goal of what I’d like to get to is to hopefully start to understand 00:23:17.830 --> 00:23:21.149 why these earthquakes were induced, such that we can tell operators and 00:23:21.149 --> 00:23:25.610 tell regulators, you know, these are the parameters that largely drive whether 00:23:25.610 --> 00:23:30.260 or not the well will induce earthquakes. All right, now this starts to get pretty 00:23:30.260 --> 00:23:33.970 complicated because there are [chuckles] very rarely independent parameters. 00:23:33.970 --> 00:23:36.169 Right, a lot of these parameters depend on one another. 00:23:36.169 --> 00:23:41.000 So we generated large odds ratio tables, which are pretty dull. 00:23:41.000 --> 00:23:44.450 So I won’t bore you with that, but what I’d like to do is try to qualitatively walk 00:23:44.450 --> 00:23:49.940 you through kind of four of the most intuitive parameters that we’ve found. 00:23:49.940 --> 00:23:54.120 All right, so the first parameter that we’d like to look at is the formation, right? 00:23:54.120 --> 00:23:59.340 So for this, we’ll be looking at our catalogs, both in Oklahoma and Texas, 00:23:59.340 --> 00:24:02.360 to try to understand what factors are important. 00:24:02.360 --> 00:24:07.139 All right, so this is sort of a geologic figure of Oklahoma. 00:24:07.139 --> 00:24:10.340 All right, so this – these are showing the formations with the shallow 00:24:10.340 --> 00:24:12.980 formations on top, deeper formations on the bottom. 00:24:12.980 --> 00:24:15.710 And the formations are colored by the formation type. 00:24:15.710 --> 00:24:17.980 All right, so yellow are sandstones. Blue are carbonates. 00:24:17.980 --> 00:24:21.570 And the gray is a shale. All right. 00:24:21.570 --> 00:24:25.529 And these little brackets represent the percentage of wells that were 00:24:25.529 --> 00:24:28.929 stimulated in those formations that we were able to induce – 00:24:28.929 --> 00:24:31.179 that we associated with induced earthquakes. 00:24:31.179 --> 00:24:34.840 So we don’t really see a correlation between the formation type and 00:24:34.840 --> 00:24:37.149 the percentage of earthquakes. 00:24:37.149 --> 00:24:41.860 But you may start to notice that we do see a correlation with depth, right? 00:24:41.860 --> 00:24:44.760 So we don’t really see a correlation directly with formation, right, 00:24:44.760 --> 00:24:47.500 but we do see a correlation with depth. All right, so this figure on the right 00:24:47.500 --> 00:24:51.059 is showing the depth of the wells and the probability that those 00:24:51.059 --> 00:24:54.419 wells induce earthquakes. So both in the SCOOP and STACK, 00:24:54.419 --> 00:24:57.559 which is kind of in the western part of Oklahoma, and that Arkoma Basin, 00:24:57.559 --> 00:25:02.299 which is in the southeast, we see a very significant statistical correlation 00:25:02.299 --> 00:25:06.660 between the depths of these wells and the probabilities that 00:25:06.660 --> 00:25:07.660 they induce earthquakes. 00:25:07.660 --> 00:25:11.039 Now, there’s a couple explanations for why depth matters. 00:25:11.039 --> 00:25:15.519 One, it could just be that the wells are closer to the seismogenic faults. 00:25:15.519 --> 00:25:19.150 But it could also be a factor of the over-pressurization. 00:25:19.150 --> 00:25:22.970 So previously, up in Canada, Dave Eaton and Ryan Schultz 00:25:22.970 --> 00:25:25.610 found that shales that were over-pressurized had a much 00:25:25.610 --> 00:25:29.060 higher likelihood of being associated with induced earthquakes. 00:25:29.060 --> 00:25:33.840 In Oklahoma, we don’t have as good of pressure data. 00:25:33.840 --> 00:25:37.240 This is pretty much a cross-section of the SCOOP and STACK plays 00:25:37.240 --> 00:25:40.880 showing the pressure gradient. This is really all the pressure data 00:25:40.880 --> 00:25:44.230 that we have in the state. But, from this study that was done 00:25:44.230 --> 00:25:49.549 a couple decades ago, we do see a great correlation between the areas that are 00:25:49.549 --> 00:25:53.309 over-pressurized and the occurrence of induced earthquakes. 00:25:53.309 --> 00:25:59.070 So, on this pressure gradient – so the red have higher over-pressurizations, 00:25:59.070 --> 00:26:03.269 and the cooler colors have lower pressures – we kind of plotted this 00:26:03.269 --> 00:26:05.850 Woodford Shale formation. So, as you can see, you know, 00:26:05.850 --> 00:26:08.269 at these deeper formations, it’s going to be higher – 00:26:08.269 --> 00:26:13.040 more over-pressurized and could be more likely to induce earthquakes. 00:26:13.040 --> 00:26:14.040 Okay. 00:26:14.040 --> 00:26:16.760 Another factor that we can look at is the volume of wells. 00:26:16.760 --> 00:26:20.260 All right, so here’s a figure of this. All right, so this figure shows the 00:26:20.260 --> 00:26:26.110 volume at an individual stimulated well. So we see a – or, so Ryan Schultz 00:26:26.110 --> 00:26:29.101 and others found a correlation between injection volume and the probability 00:26:29.101 --> 00:26:31.889 that those wells induced earthquakes up in the Duverney in Canada. 00:26:31.889 --> 00:26:35.590 From our previous work, found a correlation between volume and 00:26:35.590 --> 00:26:38.659 earthquakes in the Appalachian Basin. And it’s not shown here because it’s 00:26:38.659 --> 00:26:42.309 in review, but we also see a correlation between volume and earthquakes 00:26:42.309 --> 00:26:44.320 in the Eagle Ford Shale of Texas. 00:26:44.320 --> 00:26:47.929 You might notice that we do not see a correlation in the Arkoma Basin 00:26:47.929 --> 00:26:50.130 and SCOOP and STACK. All right, so this is where things 00:26:50.130 --> 00:26:53.210 start to get a little bit muddy. Because this is where different 00:26:53.210 --> 00:26:58.660 parameters start being influenced. All right, so in Oklahoma, a lot of 00:26:58.660 --> 00:27:02.090 these wells are just vertical wells that are hydraulically stimulated. 00:27:02.090 --> 00:27:07.529 And these are spatiotemporally isolated from other wells that might be nearby. 00:27:07.529 --> 00:27:12.490 However, in the Duvernay, in the Appalachian Basin, and Eagle Ford in 00:27:12.490 --> 00:27:14.289 Texas, a lot of these are horizontal well, right, with multiple laterals in 00:27:14.289 --> 00:27:19.049 close proxy to each other, but all associated with a single well pad. 00:27:19.049 --> 00:27:24.539 All right, so although the amount of fluid injected at an individual well is 00:27:24.539 --> 00:27:30.570 similar in Oklahoma to these other areas, I think we need to start to consider 00:27:30.570 --> 00:27:34.139 the influence of injection from nearby wells as well, right? 00:27:34.139 --> 00:27:38.159 So it’s really – we need to consider the injection into a volume of area 00:27:38.159 --> 00:27:41.100 rather than individual wells, right? But, for simplicity, I’m just going to 00:27:41.100 --> 00:27:45.820 say that volume can sometimes play a role in induced earthquakes. 00:27:45.820 --> 00:27:49.950 All right. This kind of leads me into the final topic that I want to talk about, 00:27:49.950 --> 00:27:53.039 which is the influence of operations on seismicity. 00:27:53.039 --> 00:27:58.389 All right, so, in this kind of cartoon view, let’s just pretend that this operator 00:27:58.389 --> 00:28:00.889 had two horizontal wells that they wanted to stimulate. 00:28:00.889 --> 00:28:03.179 So this is a map view. So we have the toe of the well 00:28:03.179 --> 00:28:05.269 at the very top and the heel at the bottom. 00:28:05.269 --> 00:28:07.659 And these triangles are trying to represent the stimulations. 00:28:07.659 --> 00:28:12.049 So this operator may decide to first stimulate this Well #1, and then they 00:28:12.049 --> 00:28:15.840 may switch over to Well #2. Right, so immediately after the 00:28:15.840 --> 00:28:18.330 stimulation, they go to Well #2. After they finish stimulation at Well #2, 00:28:18.330 --> 00:28:21.740 they may go back to #1 and kind of go in a zipper pattern. 00:28:21.740 --> 00:28:25.960 All right, so this is called a zipper frack. There’s variations of this. 00:28:25.960 --> 00:28:28.649 For example, they could stimulate multiple stimulations 00:28:28.649 --> 00:28:33.200 at the same time at different wells. And we call that a simulfrack. 00:28:33.200 --> 00:28:37.789 But, in general, all the wells kind of in this family of where you’re 00:28:37.789 --> 00:28:40.179 stimulating multiple wells at the same time we’re going to 00:28:40.179 --> 00:28:43.539 call simultaneous stimulations, okay? 00:28:43.539 --> 00:28:48.419 Now, there’s other types of strategies that the operator could decide to use. 00:28:48.419 --> 00:28:51.769 For example, in this same kind of cartoon view, the operator may 00:28:51.769 --> 00:28:55.970 just decide to stimulate one well. All right, when they finish that well, 00:28:55.970 --> 00:28:59.399 they may then move on to the second well, and then stimulate that way. 00:28:59.399 --> 00:29:01.519 Right, because these wells are sequentially stimulated, 00:29:01.519 --> 00:29:04.340 we’re going to call those sequentially stimulated wells. 00:29:04.340 --> 00:29:07.700 All right, and then finally, we can have a well that’s all by itself. 00:29:07.700 --> 00:29:10.100 All right, they may just stimulate like that. 00:29:10.100 --> 00:29:14.220 And these are little isolated well, and it’s sad because it’s all alone. 00:29:14.220 --> 00:29:18.840 So, with just this kind of cartoon view, right, could you make a guess as to 00:29:18.840 --> 00:29:22.570 what type of well would be more likely to induce earthquakes? 00:29:22.570 --> 00:29:27.120 I’m sure you can, right, but if you think simultaneous wells, 00:29:27.120 --> 00:29:31.289 you would be correct, right? So we found that simultaneous 00:29:31.289 --> 00:29:34.899 stimulated wells were more than three times as likely than isolated wells 00:29:34.899 --> 00:29:37.690 to induce earthquakes. At the same time, simultaneous wells 00:29:37.690 --> 00:29:40.780 were more than twice than likely than sequential wells to induce earthquakes. 00:29:40.780 --> 00:29:43.860 All right, so I think this has to do with the reservoir pressures. 00:29:43.860 --> 00:29:47.210 Right, so when you’re doing multiple stimulations in close spatiotemporal 00:29:47.210 --> 00:29:50.149 proximity to each other, the reservoir pressures will be 00:29:50.149 --> 00:29:52.090 elevated for a longer period of time. 00:29:52.090 --> 00:29:55.350 All right, we’re going to compare that to isolated wells that are stimulated, 00:29:55.350 --> 00:29:59.750 you know, kind of sequentially with further timing between the stimulations. 00:29:59.750 --> 00:30:02.529 That gives the time for the reservoir and the pressures 00:30:02.529 --> 00:30:05.540 and fluids to kind of diffuse out. 00:30:05.540 --> 00:30:09.880 So this idea is kind of corroborated by these other findings that we found. 00:30:09.880 --> 00:30:14.769 So most unconventional oil wells are hydraulically fractured with 00:30:14.769 --> 00:30:20.650 a frack prop composed largely of water. So these wells are called kind of a 00:30:20.650 --> 00:30:24.500 slickwater proppant, and we found that these were 1-1/2 times more likely 00:30:24.500 --> 00:30:27.509 than when a gelling additive was used. 00:30:27.509 --> 00:30:32.399 All right, so previously, models have determined that this gel increases the 00:30:32.399 --> 00:30:36.960 viscosity of the material such that it’ll limit the extent fluids and 00:30:36.960 --> 00:30:40.070 stresses are able to propagate. Right, so it’s kind of limited to a – 00:30:40.070 --> 00:30:45.850 to a smaller region. As I kind of showed before, 00:30:45.850 --> 00:30:49.539 wells that have multiple laterals that are stimulated in close proximity to each 00:30:49.539 --> 00:30:53.200 other, the more laterals you have, the higher probability that you 00:30:53.200 --> 00:30:56.259 induce earthquakes. And similarly, wells that inject 00:30:56.259 --> 00:30:59.730 at a much higher rate also have an increased likelihood. 00:30:59.730 --> 00:31:04.659 Okay, so kind of the downside is that we don’t have any actual pressure 00:31:04.659 --> 00:31:07.590 measurements in these reservoirs. All right, so industry definitely 00:31:07.590 --> 00:31:10.899 has those data, but they’re not publicly available. 00:31:10.899 --> 00:31:14.299 But from these observations, all right, I think it’s fairly reasonable to make 00:31:14.299 --> 00:31:17.940 the conclusion that stimulation strategies that cause higher reservoir 00:31:17.940 --> 00:31:22.769 pressures for longer periods of time could lead to increase in the likelihood 00:31:22.769 --> 00:31:24.370 of hydraulic fracturing- induced seismicity. 00:31:24.370 --> 00:31:28.460 So I think this is a fairly common-sense statement, but I think 00:31:28.460 --> 00:31:31.769 it’s an important one to make. Because this is really the first time 00:31:31.769 --> 00:31:35.059 that we’re even able to investigate these type of parameters on this 00:31:35.059 --> 00:31:38.710 kind of large regional scale. And, you know, this is not a model. 00:31:38.710 --> 00:31:41.159 This is looking at real-world parameters to see what 00:31:41.159 --> 00:31:45.210 actually happened from these types of operations. 00:31:45.210 --> 00:31:48.230 So certainly there are – this is a very complicated issue. 00:31:48.230 --> 00:31:51.250 There’s a lot of questions left to answer. 00:31:51.250 --> 00:31:52.990 But hopefully, you know, we’re headed in the right direction 00:31:52.990 --> 00:31:56.879 with this type of work. Okay. 00:31:56.879 --> 00:31:58.779 So the final thing that I’d like to talk about – oops. 00:31:58.779 --> 00:32:02.429 Oh, yeah. So the stimulation type is also induced. All right. 00:32:02.429 --> 00:32:04.940 So the final thing I’d like to talk about is our efforts 00:32:04.940 --> 00:32:07.710 in the Delaware Basin in Texas. 00:32:07.710 --> 00:32:10.309 So I like to think of the Delaware Basin as a baby Oklahoma. 00:32:10.309 --> 00:32:12.940 All right, we’re not quite sure what it’ll grow up into, 00:32:12.940 --> 00:32:17.700 but it’s a sketchy little baby. You know, we need to keep an eye on it. 00:32:17.700 --> 00:32:23.169 Okay, so this is a map of western Texas. The upper left here is New Mexico. 00:32:23.169 --> 00:32:24.799 The lower left is Mexico. 00:32:24.799 --> 00:32:28.140 And this region largely encompasses the Permian Basin. 00:32:28.140 --> 00:32:30.960 So the Permian Basin has been in the news a lot because it’s undergoing 00:32:30.960 --> 00:32:36.009 a lot of hydrocarbon development. But this basin also a history 00:32:36.009 --> 00:32:39.380 of documented cases of induced seismicity. 00:32:39.380 --> 00:32:42.580 So Cliff Frohlich has done a great job to characterize induced earthquakes 00:32:42.580 --> 00:32:45.129 over the past century or so. And he’s found a lot of 00:32:45.129 --> 00:32:49.720 production-related earthquakes in the Permian Basin. 00:32:49.720 --> 00:32:57.769 There’s also examples in Snyder. In the ’70s and ’80s, earthquakes were 00:32:57.769 --> 00:33:00.659 induced by water flooding, which is an enhanced oil recovery technique. 00:33:00.659 --> 00:33:03.059 And about 10 years ago, there were some more earthquakes 00:33:03.059 --> 00:33:06.850 in Snyder induced by CO2 injection. 00:33:06.850 --> 00:33:10.179 For the west, there’s the Dagger Draw field that was 00:33:10.179 --> 00:33:14.760 induced by wastewater disposal. But really, in the past five years or so, 00:33:14.760 --> 00:33:18.490 in this – kind of this southern portion of the Delaware Basin, there’s been 00:33:18.490 --> 00:33:20.100 a large increase in the number of earthquakes. 00:33:20.100 --> 00:33:23.880 All right, so this is kind of our study area that I will be focusing on. 00:33:23.880 --> 00:33:28.379 So if we zoom into this area, these black dots are representative 00:33:28.379 --> 00:33:31.259 of our earthquakes. The orange dots are the 00:33:31.259 --> 00:33:34.850 hydraulic fractured wells. And the blue are wastewater disposal wells. All right, 00:33:34.850 --> 00:33:39.809 so there’s a tremendous amount of industry operations in this area. 00:33:39.809 --> 00:33:42.299 So I won’t be talking about it, but we’ve done analysis of the 00:33:42.299 --> 00:33:44.149 hydraulic fracturing-induced earthquakes and kind of about 00:33:44.149 --> 00:33:46.919 5% of the earthquakes, represented by the green stars, 00:33:46.919 --> 00:33:50.580 are induced by hydraulic fracturing. Okay. 00:33:50.580 --> 00:33:53.399 Okay, so much like Oklahoma, we’ve done template matching. 00:33:53.399 --> 00:33:57.379 So this is showing our regional template matching. 00:33:57.379 --> 00:33:59.640 So we took a regional catalog and correlated events 00:33:59.640 --> 00:34:01.490 over the past decade or so. 00:34:01.490 --> 00:34:07.460 We also have access to the TexNet data. All right, so these are more local 00:34:07.460 --> 00:34:11.040 station deployments in this area. And from that, we can get a much 00:34:11.040 --> 00:34:14.550 better catalog, but we’re limited in time. Then also I’ll talk about this later, 00:34:14.550 --> 00:34:18.290 but we also are interested in improving the catalog in this northwest area. 00:34:18.290 --> 00:34:23.270 But in general, we’ve identified about 40,000 events, but the largest of which 00:34:23.270 --> 00:34:26.570 is kind of in the mid-magnitude 3 range. All right, so we have a lot of 00:34:26.570 --> 00:34:30.180 earthquakes, but the earthquake magnitudes are still fairly small. 00:34:30.180 --> 00:34:33.810 If we look at the depths of these operations – so on the left 00:34:33.810 --> 00:34:37.350 are the wastewater disposal. The middle is hydraulic fracturing, 00:34:37.350 --> 00:34:40.790 and the right are the cataloged earthquake locations. 00:34:40.790 --> 00:34:42.230 This is a little bit different than Oklahoma. 00:34:42.230 --> 00:34:44.680 So, in Oklahoma, most of the wastewater disposal is into the 00:34:44.680 --> 00:34:47.610 Arbuckle, which is right above the basement. 00:34:47.610 --> 00:34:51.870 But here in the Delaware Basin, a lot of the injection is into this 00:34:51.870 --> 00:34:55.000 Delaware Mountain Group, which is a sandstone formation, 00:34:55.000 --> 00:34:58.010 which has been used a lot for conventional oil exploration, 00:34:58.010 --> 00:35:04.440 and now it’s being used as a reservoir to dispose of the wastewater. 00:35:04.440 --> 00:35:08.630 The earthquakes – or the hydraulic fracturing that’s occurred is largely 00:35:08.630 --> 00:35:10.910 done in the Wolfcamp or Bone Spring Formations, 00:35:10.910 --> 00:35:12.720 which is below the Delaware Mountain Group. 00:35:12.720 --> 00:35:17.140 And interestingly, a lot of the earthquake hypocenters, 00:35:17.140 --> 00:35:20.060 those reported depths are below these shales. 00:35:20.060 --> 00:35:26.700 So, if these earthquake locations are reliable, it raises some questions. 00:35:26.700 --> 00:35:31.620 Now, if the injection is occurring shallower, right, is there some kind of 00:35:31.620 --> 00:35:34.210 permeated pathway through these shales that allows those 00:35:34.210 --> 00:35:37.630 fluids to reach these faults? All right, rather – or, I don’t know, 00:35:37.630 --> 00:35:41.130 is it just a stress transfer, you know, through those formations? 00:35:41.130 --> 00:35:44.890 Or possibly – I think the most likely explanation is that these earthquakes are 00:35:44.890 --> 00:35:48.080 actually a little bit shallower, and we should probably shift them up. 00:35:48.080 --> 00:35:54.490 All right, so just kind of make this point in that the earthquake hypocenter 00:35:54.490 --> 00:35:56.660 depths are going to be very important for our understanding 00:35:56.660 --> 00:35:59.300 of what’s going on in the Delaware Basin. 00:35:59.300 --> 00:36:04.190 Okay, so another comparison with Oklahoma. 00:36:04.190 --> 00:36:06.880 So some people have found that there are poroelastic effects that 00:36:06.880 --> 00:36:09.800 can induce earthquakes tens of kilometers away from wells. 00:36:09.800 --> 00:36:13.470 I think one of the best examples of this is done by Thomas Goebel and others, 00:36:13.470 --> 00:36:16.830 where they found that the Fairview and Woodward sequences in western 00:36:16.830 --> 00:36:20.330 Oklahoma were induced by injection wells that were 00:36:20.330 --> 00:36:24.180 kind of 30 to 40 kilometers away from those sequences. 00:36:24.180 --> 00:36:27.960 It should be noted that there are disposal wells that are within 00:36:27.960 --> 00:36:31.480 10 kilometers of these sequences. Just the volumes are much less. 00:36:31.480 --> 00:36:36.500 All right, so I think everyone in this room is aware of Art’s theory 00:36:36.500 --> 00:36:41.030 that the cumulative seismic moment from induced seismicity can be related 00:36:41.030 --> 00:36:44.540 and constrained by the cumulative injection volume. 00:36:44.540 --> 00:36:48.710 So this Fairview example that Thomas Goebel found, it did seem to 00:36:48.710 --> 00:36:52.640 follow along Art’s relationship, considering only wells that are 00:36:52.640 --> 00:36:53.640 within 10 kilometers. 00:36:53.640 --> 00:36:56.880 All right, so this a – kind of a [chuckles] controversial topic. 00:36:56.880 --> 00:36:59.770 I won’t get into debating whether or not these are poroelastic effects. 00:36:59.770 --> 00:37:03.630 I just more raise this point that observations have been made, 00:37:03.630 --> 00:37:06.420 but it’s not a super clear-cut example. 00:37:06.420 --> 00:37:10.110 All right, hopefully I can do that to help motivate our work. 00:37:10.110 --> 00:37:15.500 All right, so in this green box in the northwest region, you know, I looked 00:37:15.500 --> 00:37:18.180 for some geographical landmark in this area. There’s really nothing. 00:37:18.180 --> 00:37:21.930 So I just called it the northwest region. If we zoom in on this area shown on 00:37:21.930 --> 00:37:25.160 the right, we said that these earthquakes are about 25 kilometers away from 00:37:25.160 --> 00:37:28.560 the nearest injection well. All right, these injection wells are pretty 00:37:28.560 --> 00:37:33.090 shallow and relatively low volume. So pretty much the largest 00:37:33.090 --> 00:37:36.410 deep injection well is about 35 kilometers away. All right. 00:37:36.410 --> 00:37:39.180 So the earthquake uncertainties horizontally are kind of on the 00:37:39.180 --> 00:37:43.380 order of about a kilometer or so. So we have, you know, pretty good 00:37:43.380 --> 00:37:48.490 constraints that these earthquakes are at least, you know, a couple tens of 00:37:48.490 --> 00:37:50.780 kilometers away from the injection. 00:37:50.780 --> 00:37:53.920 All right, so I think that that’s a pretty interesting observation. 00:37:53.920 --> 00:37:57.840 So you may also notice that there’s a lot of wells in kind of this 00:37:57.840 --> 00:38:00.860 more northern part of the Delaware Basin. 00:38:00.860 --> 00:38:04.160 But there aren’t any earthquakes. All right, so this kind of same kind of 00:38:04.160 --> 00:38:07.750 data could be shown by these figures. So this is showing kind of the – 00:38:07.750 --> 00:38:11.120 kind of the spatial grid of wastewater disposal volumes. 00:38:11.120 --> 00:38:15.330 So you can see that there’s a lot of injection occurring throughout this area. 00:38:15.330 --> 00:38:17.520 But the earthquake – that looks kind of blue. [laughs] 00:38:17.520 --> 00:38:18.590 That’s supposed to be purple. 00:38:18.590 --> 00:38:24.030 But, yeah, a lot of the earthquake counts are all south in this basin. 00:38:24.030 --> 00:38:26.200 There aren’t really any earthquakes in this more northern part. 00:38:26.200 --> 00:38:29.930 And I think answering this question of why we have earthquakes in 00:38:29.930 --> 00:38:33.730 the southern part but not the northern part is going to be very important. 00:38:33.730 --> 00:38:39.660 All right, so we started to work and try to look at why this is. 00:38:39.660 --> 00:38:41.090 So I’ve been looking largely at the geology. 00:38:41.090 --> 00:38:44.050 All right, so this is showing the basement contours – 00:38:44.050 --> 00:38:46.060 all right, so the depths of the basement. 00:38:46.060 --> 00:38:48.980 And these lines are showing mapped faults in this area. 00:38:48.980 --> 00:38:51.690 So, to the east, we have this central basin platform. 00:38:51.690 --> 00:38:56.390 All right, I think I can make a fairly convincing argument that this basin – 00:38:56.390 --> 00:38:59.710 that this platform will inhibit the horizontal migration of fluids. 00:38:59.710 --> 00:39:02.630 But a real question is, you know, why do we have this kind of 00:39:02.630 --> 00:39:06.610 north-south division? So it’s – you know, potentially, 00:39:06.610 --> 00:39:08.750 this is just a spurious correlation. 00:39:08.750 --> 00:39:12.520 But we do see that the earthquakes stop right at this Grisham Fault. 00:39:12.520 --> 00:39:15.490 And along this fault, we see that there’s a rotation of the stress field. 00:39:15.490 --> 00:39:18.680 All right, so kind of east-west, north of this fault, and then 00:39:18.680 --> 00:39:23.960 just to the south, the stress rotates towards kind of a southeast direction. 00:39:23.960 --> 00:39:26.830 So this raises some interesting questions, right? 00:39:26.830 --> 00:39:30.870 Is there also a magnitude of stress change along this fault? 00:39:30.870 --> 00:39:36.360 Maybe, does this fault inhibit the flow of fluid across the fault? 00:39:36.360 --> 00:39:38.690 Maybe there are different reservoir pressures in these areas. 00:39:38.690 --> 00:39:41.700 I think there’s a lot of potential possibilities. 00:39:41.700 --> 00:39:45.490 But we’ve kind of reached the limits of what we can do 00:39:45.490 --> 00:39:49.230 observationally with the data that we have available. All right. 00:39:49.230 --> 00:39:52.910 So we’re working on a finite element model for this entire Delaware Basin to 00:39:52.910 --> 00:39:55.920 look at the influences that the structure will have on the seismicity. 00:39:55.920 --> 00:39:58.770 All right, so we want to look at the interplay of both the production and 00:39:58.770 --> 00:40:02.960 disposal to see – to see to what degree these both contribute. 00:40:02.960 --> 00:40:05.840 And to also see what influences these operations have on the 00:40:05.840 --> 00:40:08.450 pressures and stresses in this basin. 00:40:08.450 --> 00:40:12.960 All right, so if you have ideas of [chuckles] why the earthquakes will 00:40:12.960 --> 00:40:17.260 suddenly stop, you know, let me know. All right, but hopefully that’s 00:40:17.260 --> 00:40:20.260 a good question to leave you guys with. All right. 00:40:20.260 --> 00:40:24.060 So, in conclusion, this FaultID algorithm that I talked about, 00:40:24.060 --> 00:40:28.270 I think, is a – is a pretty powerful tool. I think there’s still plenty of room 00:40:28.270 --> 00:40:31.430 for improvement and advancements to this technique. 00:40:31.430 --> 00:40:34.860 But, so far, most of the earthquakes that we’ve found in Oklahoma have 00:40:34.860 --> 00:40:38.140 occurred along faults that are optimally oriented to the original stress field. 00:40:38.140 --> 00:40:40.680 And I would even argue that we could those stress orientations 00:40:40.680 --> 00:40:45.370 to estimate the SHmax in areas that we don’t have measurements. 00:40:45.370 --> 00:40:48.980 So hydraulic fracturing is quite a bit more common than we thought 00:40:48.980 --> 00:40:52.770 even just a few years ago. And there are various factors that will 00:40:52.770 --> 00:40:57.190 influence the likelihood of whether or not a well will induce earthquakes. 00:40:57.190 --> 00:40:59.820 So I think – largely due to stimulation depth of – 00:40:59.820 --> 00:41:04.230 you know, over-pressurization, as well as operational decisions 00:41:04.230 --> 00:41:05.670 made by the operators. 00:41:05.670 --> 00:41:08.840 And then finally, the seismicity rate in the Delaware Basin 00:41:08.840 --> 00:41:13.870 will likely continue as long as the operations are not altered. 00:41:13.870 --> 00:41:16.230 But hopefully we can start to answer some of these questions 00:41:16.230 --> 00:41:28.010 before that baby Oklahoma grows up. All right, so with that, I’ll be 00:41:28.010 --> 00:41:31.934 happy to answer any questions. All right, thanks. 00:41:31.934 --> 00:41:32.934 [Applause] 00:41:32.934 --> 00:41:33.934 [Silence] 00:41:33.934 --> 00:41:34.934 - Okay. Do we have any questions? 00:41:34.934 --> 00:41:35.934 - Uh-oh. 00:41:35.934 --> 00:41:37.590 - [laughs] 00:41:37.590 --> 00:41:39.290 [Silence] 00:41:39.290 --> 00:41:46.100 - Nice talk, Rob. - [laughs] 00:41:46.100 --> 00:41:53.800 - Can we go back to the Delaware Basin where you’re plotting the depths 00:41:53.800 --> 00:41:59.800 sort of against injection … - Oh, yeah, yeah. 00:41:59.800 --> 00:42:08.350 - And here you said that you think that the depths need to be higher. 00:42:08.350 --> 00:42:10.230 And I guess … - It’s one possibility, yes. 00:42:10.230 --> 00:42:17.830 - I guess I don’t – I mean, I guess – can you justify that a little bit better? 00:42:17.830 --> 00:42:21.860 Why shouldn’t they be that deep? - Yep. So – I mean, so, if they are 00:42:21.860 --> 00:42:26.840 that deep, I mean, how do you explain why there were induced, right? 00:42:26.840 --> 00:42:29.870 So if it was – if it was induced by wastewater disposal, is there 00:42:29.870 --> 00:42:33.010 some permeability pathway through the shales? 00:42:33.010 --> 00:42:35.190 Is the stress being transmitted? 00:42:35.190 --> 00:42:38.820 - Why should this be any different than Oklahoma? 00:42:38.820 --> 00:42:43.190 - So, in that – in the Delaware Basin, we have these shales, right, 00:42:43.190 --> 00:42:46.880 which are relatively impermeable. So, in Oklahoma, it’s injecting into 00:42:46.880 --> 00:42:50.510 the basal sediments, right, into the Arbuckle, which is right above the 00:42:50.510 --> 00:42:54.900 basement, which is where a lot of these earthquakes and faults are located. 00:42:54.900 --> 00:42:57.630 But here, in the Delaware Basin, all right, we have those shales. 00:42:57.630 --> 00:43:01.090 All right, so is this acting as some impermeable layer? 00:43:01.090 --> 00:43:04.970 All right, are there the pathways that allows the fluids through? 00:43:04.970 --> 00:43:09.080 I mean, there’s questions. So preliminary work that’s not done 00:43:09.080 --> 00:43:13.000 by me that has not been published yet do indicate that those earthquakes 00:43:13.000 --> 00:43:17.420 tend to be shallower, even into these – kind of the shale formations, which is 00:43:17.420 --> 00:43:19.030 where they think a lot of the faults are. 00:43:19.030 --> 00:43:22.061 But I would say those are not published yet, so … 00:43:22.061 --> 00:43:27.530 - Any additional questions? 00:43:27.530 --> 00:43:36.660 - It was very interesting, Rob. 00:43:36.660 --> 00:43:42.070 My question concerns – I think you showed some faults that you defined 00:43:42.070 --> 00:43:50.490 in the Cushing area, and – early in your talk, in the first topic. 00:43:50.490 --> 00:43:52.550 - [inaudible] - Right. 00:43:52.550 --> 00:43:59.610 - Or the Prague? Or Cushing? - Cushing. 00:43:59.610 --> 00:44:01.070 - Yep. 00:44:01.070 --> 00:44:07.830 - Right. Now, I was wondering why we don’t see another fault to the 00:44:07.830 --> 00:44:13.700 south of the main one you showed toward the bottom of your figure there, 00:44:13.700 --> 00:44:19.790 which has kind of a conjugate orientation if the horizontal stresses 00:44:19.790 --> 00:44:25.180 are – the maximum horizontal stress is approximately east-west there. 00:44:25.180 --> 00:44:27.100 - So where are you … 00:44:27.100 --> 00:44:28.340 Like here? 00:44:28.340 --> 00:44:35.380 - Dan McNamara did a study that – on earthquakes that were, I think, in 2014, 00:44:35.380 --> 00:44:38.971 mostly, in the Cushing area before … - Oh, so … 00:44:38.971 --> 00:44:43.970 - … and he showed faults that those earthquakes … 00:44:43.970 --> 00:44:47.000 - Yes. So I think that – that’s actually just off scale – 00:44:47.000 --> 00:44:49.620 that’s just, I think, to the south. - Oh, it’s off the bottom? Okay. 00:44:49.620 --> 00:44:52.810 - Yeah. So it – yeah, it’s down here, but it’s off this map. 00:44:52.810 --> 00:44:55.500 - Yeah, south of Cushing, in fact. - Yeah. I think I know what 00:44:55.500 --> 00:44:57.350 you’re talking about, yeah. - Oh, okay. Thanks a lot. 00:44:57.350 --> 00:45:00.010 - Yes. Yeah. There are definitely earthquakes down there. 00:45:00.010 --> 00:45:07.160 - Okay. Thanks a lot. [laughs] - Yeah, sure. 00:45:07.160 --> 00:45:08.180 [Silence] 00:45:08.180 --> 00:45:14.310 - Hi, Rob. Nice talk. - Thanks, Steve. 00:45:14.310 --> 00:45:17.870 - Getting back to the question that Justin just asked, if you have – 00:45:17.870 --> 00:45:23.640 going back to that figure, if you have poroelastic effects, wouldn’t those – 00:45:23.640 --> 00:45:28.140 expected to change the stress below a relatively permeable shale there … 00:45:28.140 --> 00:45:30.580 - Yes. Yep. - So if you don’t believe in short-circuit 00:45:30.580 --> 00:45:32.100 fluid pathways through the shale, 00:45:32.100 --> 00:45:35.490 does this suggest that poroelastic effects might be important? 00:45:35.490 --> 00:45:38.010 If the earthquakes are really where they appear to be? 00:45:38.010 --> 00:45:40.330 - Yeah. So I tried to list that as one of the three possibilities. 00:45:40.330 --> 00:45:44.191 Right, either they have the pathways that stress is transmitted, or that 00:45:44.191 --> 00:45:47.250 the earthquakes are closer. All right, so it’s one of those three. 00:45:47.250 --> 00:45:50.050 And I’m just – I’m just listing those questions, right, out there. 00:45:50.050 --> 00:45:53.550 I’m not stating what’s the case. But I think it is – it is possible. 00:45:53.550 --> 00:45:54.890 - It’s a pretty short distance, and it’d be something to model. 00:45:54.890 --> 00:45:56.160 - Yep. So that’s what we’re working on, yep. 00:45:56.160 --> 00:45:57.410 - Yeah. And you’d also want to look at whether the shales – 00:45:57.410 --> 00:45:58.410 are they presumably seismogenic? Depending – you know, some shales 00:45:58.410 --> 00:45:59.410 are, some aren’t, so those are the kind of issues you could look at too. 00:45:59.410 --> 00:46:00.829 Another question I have had to do with the operational thing. 00:46:00.829 --> 00:46:02.220 - Yep. 00:46:02.220 --> 00:46:16.090 - Being an operational kind of guy, I latch onto these. 00:46:16.090 --> 00:46:18.940 So actually, no, it was the one – yeah, this one. This is good. 00:46:18.940 --> 00:46:26.520 So what is – so the fact that slickwater is at 1-1/2 times, 00:46:26.520 --> 00:46:30.310 I totally get the simultaneous thing. You’re building up fluid pressure over 00:46:30.310 --> 00:46:31.700 a larger volume – or poroelastic effects. What do you think the fact that gel 00:46:31.700 --> 00:46:34.340 is less likely to produce events than slickwater is? 00:46:34.340 --> 00:46:36.950 Is that – does that tell us it’s a fluid pressure effect? 00:46:36.950 --> 00:46:40.460 Does it rule out poroelasticity? Does it help discriminate between 00:46:40.460 --> 00:46:44.000 those two by looking at the viscosity and the kind of fractures you would 00:46:44.000 --> 00:46:46.910 stimulate and inflate with these different kinds of fluids? 00:46:46.910 --> 00:46:49.510 Because the gels, I presume, have huge viscosities. 00:46:49.510 --> 00:46:51.130 - Yeah. Very large. Yep. - Yeah. 00:46:51.130 --> 00:46:54.540 - So I’m only aware of a few modeling-type studies. 00:46:54.540 --> 00:46:56.590 I mean, so certainly the industry has done those analyses. 00:46:56.590 --> 00:46:59.610 They actually have measurements. But the stuff that’s public, 00:46:59.610 --> 00:47:05.570 I believe it’s just kind of models. But, from those highly viscous gels, 00:47:05.570 --> 00:47:10.100 it really limits the spatial extent that those stresses and fluids propagate. 00:47:10.100 --> 00:47:13.970 Right, it’s really just isolated, you know, close to the well itself 00:47:13.970 --> 00:47:18.180 and doesn’t really diffuse out as slickwater would. 00:47:18.180 --> 00:47:20.390 So that’s the – kind of our understanding of it. 00:47:20.390 --> 00:47:25.610 - I was going to also point out, I was reassured to see that the top one – 00:47:25.610 --> 00:47:27.530 simultaneously – three times more than isolated. 00:47:27.530 --> 00:47:30.410 Because that’s the guiding principle for enhanced geothermal system 00:47:30.410 --> 00:47:33.390 stimulations now. You’re more likely to get permeability 00:47:33.390 --> 00:47:35.800 creation through micro-earthquakes by injecting into multiple zones 00:47:35.800 --> 00:47:36.800 at one time. - Yes. 00:47:36.800 --> 00:47:40.010 - If you can. - So, I mean, that’s one of the reasons 00:47:40.010 --> 00:47:44.780 why industry does that, right, so … - Thanks. 00:47:44.780 --> 00:47:45.780 - Yep. 00:47:45.780 --> 00:47:48.150 - Do we have any other questions? 00:47:48.150 --> 00:47:55.340 Okay, great. Let’s thank Rob one more time. 00:47:55.340 --> 00:48:00.500 - All right. Thanks, guys. 00:48:00.500 --> 00:48:01.790 [Applause] 00:48:01.790 --> 00:48:03.080 [Silence] 00:48:03.080 --> 00:48:08.430 - And if you want to join us and Rob for lunch, please come up, 00:48:08.430 --> 00:48:09.820 and we can make a plan to do that.