- [Gary] We took Well A here, and we ran our inelastic capture and gamma ray spectroscopy measurements, and the blue curves are the logs. And the measurements that we compared to were handheld XRF, and we compared the elements, and the elements we compared were aluminum, calcium, iron, potassium, magnesium, manganese, sodium, silicon, sulfur, and titanium. We had a summer intern that we put down in the basement and put him down there all summer and had him scan the core every three inches, so each of these red dots is a three-inch scan of that. And you can see that we have a really nice fit to the elements from the XRF to the log data that we could feel pretty good about. From this, we put that in again to help develop the mineralogy model we ultimately came up with. Also in this same Well A, this well was part of a consortium well with one of the core service companies, and of the 360 roughly cores that we took in this, they only selected 26 of those to do XRF on. When we compare the XRF to the cores of the different things, it looks like we have a pretty good fit. But when we compare that to the density of measurements that we have every three inches, I came to realize that 26 was not a statistically valid number of points to really give me the confidence that I needed to benchmark the mineralogy. You'll see what we did here is, so we compared this with, in two ways, with FTIR and also DRIFTS. These are the FTIR measurements. Again, we compared the elements, but the samples here came from each three-foot section of core that we cut and harvested at the well site. And what we did is take a hammer, and we knocked off an end chip on the end of each of these barrels that we cut in three-foot sections, and we put that into a plastic bag and labeled it. We had a research project going with a company, and they took this, those samples, and they selected 63 of them that had enough sample that we could do both the DRIFTS and the FTIR on. On the FTIR measurements here, as we go depth-wise, each of these red dots took eight hours to do. And so, but we have a really nice fit of our elements, looking at the FTIR here, and now, here is a comparison of the DRIFTS measurements with the FTIR. Now we're breakin' it down, we're looking at the clays, the chlorite, illite, montmorillonite, and we're seeing how well we fit with that. The red dots are from the FTIR, and the black are from the DRIFTS measurement. Here's the quartz and feldspars, and this is the FTIR minerals. Carbonates, we have both DRIFTS and FTIR. And then over here, the nice about the DRIFTS measurement is it also makes a kerogen content, or a total organic carbon content measurement, and so, we see that compared to the log measurements here also. The logging total we made was a pulse neutron type of inelastic and capture spectroscopy and from that, we're able to calculate a organic carbon content to use. And then, for the DRIFTS, we're comparing the quartz, feldspar, and micas to the measurement. After all this was done, we finally had this final mineralogy model that we have here, and this what we populated throughout all the different petrophysics models that we use from this point on. Now, let's look at comparing NMR porosity and permeability versus core. The reason for running NMR is that it is a mineralogy and matrix independent total porosity. We ran this NMR with the short echo spacing of .2 milliseconds. We combined it with an enhanced processing mode. This allowed us to capture fast-relaxing T2s for quantifying small porous spaces and clay bound fluids and yielded an accurate NMR total porosity to compare to core. Over here on the right hand side, these are all the core plugs that we took compared to grain density and compared our grain density from the logs we made. The black curve here is NMR total porosity, and the red dots are the Dean Stark of hole plugs. The green dots are GRI crushed rock analysis in the shells, and you see some gold-colored dots in here. These were retort analysis, and there's some blue dots scattered through here that were NMR porosities. But there's a fairly good fit of porosity through here. The permeability up here in the middle Bakken, we see a pretty good fit of permeability from NMR to the core, but down here in this lower part, you see quite a bit of scatter, and what we've since found is that a lot of these three forks cores after we Dean Stark 'em, they try to split, and so, we get a permeability that's affected by the fracture in the core, so we have quite a bit of scatter in the measurement there. Next, I wanna talk about the analysis techniques that we used for the three forks and the middle Bakken, and one of the things that we had run here was a triaxial reefs induction log. We were trying to figure out, can we apply the models that were developed for shaley sand analysis to this type of interpretation approach. If we look at the core over here from the three forks one section, we see this brown or tan-colored rock, and this is actually our billow silkstone, and this is our reservoir rock. We were able to compare this to say this is the sand portion of this model that we would normally use for shaley sand anal. This green, or apple-colored green, or gray-colored rock you see here is the dullow mudstone, and that's the non-reservoir rock. We have these layers of the reservoir and non-reservoir rock in here, but these green-colored rocks, we can compare that to the shale in the thin-bed models that we use for shaley sand. Using the vertical and horizontal resistivities and puttin' it into that shaley sand model, we can solve for the fraction of shale and the fraction of sand, and ultimately, we're gonna solve here for the resistivity of the sand, and this is what we're gonna use to calculate water saturation. To see how this actually fit into that type of model, we get a cross plot of RvRh, and this is what we call the Klein plot. Some people call it the butterfly plot, but the data fit in here real nicely, so you can see here fraction of shale, fraction of sand, and so, we felt that this was a good model to use. The next part of the analysis that was typical is that in the Bakken shales, we had this total organic carbon content, and so from our spectroscopy measurement, we were able to calculate a total organic carbon content from the log. We benchmarked it against the DRIFTS measurements that we had, and also, these blue dots in here are actually rock eval measurements using paralysis. And so, we have a nice fit of our logged core data. Once we know what this total organic carbon content is, then we can correct the porosity in those shale zones for the weight percent of total organic carbon to come up with a more accurate porosity. Now it's looking at the core to log water saturation using dielectric and triaxial induction. The first thing we did is we took the horizontal resistivity which would be our standard induction log type of resistivity, and we calculated water saturation from it, and the blue curve here is that water saturation. And you can see that there's not a very good fit in some areas with the core data. As we look up here at the triaxial resistivity log, and the red curve is the horizontal, and the blue curve is the vertical resistivity, you saw that Rv was greater than Rh in this section right here, which indicated this a laminated reservoir. Another indicator we had is that we were also able to calculate a RXO curve from the dielectric multifrequency component, and you can see that the fine resolution of showed that it was pretty laminated through here, too. Using the Rv and Rh into our thin bed analysis model, we take the R sand curve, and we calculate our water saturation using R sand, and you can see it, there's a much better fit through these laminated sections to the water saturation from cores. The last thing we did is we calculated water saturation from the dielectric measurement, and a dielectric has a one-inch vertical resolution, so we see a really fine water saturation curve here that we also compared to the core water saturation. With that said, we also noticed that down here in the lower part of the three forks formation of third and fourth benches, that the core water saturation was reading 40 to 60%, but the log saturation was reading 100%. And since we had run the NMR and dielectric, we calculated a water filled porosity from the dielectric, and we saw that it overlaid the water filled porosity from the NMR, I mean, the porosity from NMR. We felt very confident that these two logs were correct, and that this was a water filled volume of rock. Why were the core analysis reading so high in here? As it turned out, this started what has now turned into a 20-month investigation into the protocols for evaluating unconventional tight oil. We realized that gas shales had had a lot of research done on the analysis techniques with the Gas Research Institute and the GRI method, but we found out that there had not been hardly any research done on unconventional tight oil. The one thing that we found out that affected this the most was salinities in the Bakken petroleum system. We now feel that most of the salinity in that system is actually closer to 400,000 parts per million. But it ranges between 300,000 and 400,000 parts per million, which makes it a fairly unique reservoir around the world for having this high salinity. There is salt beds above and below the Bakken petroleum system that has fed salt into the system over time. And so, one day I just, we were sitting around discussing this, and I finally said, "You know, when we bring a core "to surface, what is the maximum salt that we can "actually hold in solution in ambient conditions?" When we got our chemistry and physics handbooks out, we found out that it was 240,000 parts per million. What happened to this additional salinity that was up here? As it turned out, my geologist went back and started looking at some SEM images, and it didn't take long before he started seeing these halite crystals that were plugging our pore spaces. What was happening when we ran the core to surface is this salt was precipitating out and plugging our pores within the rock. When we started doing full plug analysis, we found out that this gave us incomplete core cleaning and contributed to inaccurate core Sw porosity and permeability measurements. To prove this and to make us feel better that this is actually what was going on, we had a core that was just in, gettin' ready to be analyzed, and so, we took six fresh plugs from that and we Dean Starked 'em, and we Dean Starked 'em to what we thought was completion. Completion is where you don't see any additional discoloration of toluene or any additional water produced from that core for 24 hours, and you would call that the end of your test. Instead of callin' that the end of our test, what we did is we took those plugs out of the thimble, we crushed those plugs, and we returned that crushed material to a fresh clean Stark apparatus. What you see here is the discoloration from this crushed plug analysis, and we had a lot darker, so we relieved more oils out of that crushed rock, and we also recovered about 20% additional water extracted from that plug. It was then we found out that the GRI methods, or crushed rock methods, didn't use any type of post-Dean Stark flushing or methanol to remove salts out, so we started doing salt distillation after that. Here is a recent image of the middle Bakken reservoir plug, and this is one we sent to a researcher. He was complaining to us that we sent him some poor quality rock because he was having trouble injecting fluids into it. And we said, "No, we sent you "the best quality reservoir rock that we had." And so, we did an SEM image on it, and you can see, here's our pore throats, along the pore throats, we had this salt lining in this, and so, we had to develop protocols where he's going in now and under pressure injecting methanol in here to clean these salts out before he does any experiments. But this was a really validation that we were having salt plugging in the cores. Finally, now that we've gone over models developed, we realize that using these models that we could develop some type and look at some qualitative curves for understanding water's producibility in here. We tried a couple of different methods. In the original paper, we talked about the Jarvie oil saturation index, and since then,came up with the reservoir producibility index. Here you see both of those producibility indexes that are plotted against all of our different core data, our log data. This is kind of a composite of all the final stuff that we did. Lastly, 'cause we didn't talk about it too much, I just wanted to show you that using the multi-component dipole sonic that we did calculate 3D VTI where we were calculating dynamic and horizontal and vertical Young's modulus and Poisson's ratio, and then, we have static calculations from this. Of course, we're benchmarking all this now. We just started a project where we're going in and we're doing a whole bunch of plugs through here with a company to determine our stresses and everything.