- [Voiceover] Now let's talk a little bit about porosity cores versus petrophysics because it's obviously important. When I was growing up in the early days it was petrophysics, it was always said, "Make sure that the petrophysics agrees with the core data." That is no longer I think a valid assumption because as we'll see, there is no real direct correlation. For example in porosity opinions vary, which in my opinion most core porosities are a measurement of what we would call C-E, effective, even maybe connected porosity only. There are no real service company standards in place and so procedures vary from one service company to the next. But as a general conclusion I find that core porosity equates pretty well with petrophysically defined effective porosity. I don't think it sees the shale porosity, I don't think it sees T-O-C. Okay for fluid saturations, obviously fluid saturations are made at the surface following decompression. And so in general you might anticipate that oil saturation would be reduced because of mud filtrate invasion in the coring process. Gas saturation could increase especially in volatile oil reservoirs because of solution gas coming out of solution as pressure is reduced. And water saturation might increase due to mud filtrate invasion, so you can't necessarily think that you're gonna get a direct correlation. Also really importantly in oil-based muds it's often suggested that the water saturation is representative of in situ conditions. That may or may not be true. We did a lot of work in the Texas Panhandle on this and found that in many cases it appeared that there was water being lost, a little bit of water from the makeup for the oil-based mud being lost to the formation and invading. Okay so grain density, core measurements obviously eliminate any influence of gas effect and the closest comparison probably with cores is gonna be in clean oil or particularly water bearing sands or carbonates. Permeability is, we talked about, is not an easy measurement. Again you have to be careful as to how it was measured. It might be quoted at you as air at ambient conditions, Klinkenberg extrapolation to infinite pressure, which is probably a better measure. And especially in tight gas sands which is not part of this course but for general knowledge, probably all of you already know, a huge effect on overburden pressure. It can reduce permeability by decades. Okay so the correlation with logs is not easy. What we do is experiment on the exponents in that porosity to get the best correlation we can with cores. Again now let's turn our attention a little bit to core shifting. It's obviously really important to shift the core data to agree with the logs. For sidewall cores it shouldn't be a problem, presuming that they have been measured on the same wireline so that you don't have a depth discrepancy. But for continuous whole core analysis it's obviously important. The depth registration is coming from the driller not from the wireline logger and it can be quite different in many cases. Also the problem with continuous coring, if you got less that 100 percent recovery, where have you lost that core? When I first joined the industry it was presumed that you lost it at the bottom, the core catchers didn't catch it. But also it could be, invented this term rubblizing. It could be that an incompetent part of the core has been lost and you don't know where that is. A little interesting story. To any of you that have done well-side work, you know what the hazard is when the core is cut if you cut the whole cores. It comes out of the core barrel and drivels all over the drill floor and you have to either know the driller very well to ask him to get it out gradually or you have to be very quick in putting the core into the right place. When I worked for Marathon I'd done a bunch of coring in England and East Africa and whatever the place, and went up and monitored a core that was being collected by an engineer. So we went up to the drill floor and I said, "Oh be careful." He said "Oh no no, the deal is that you let it all dribble out and then you puzzle it to go back as to the way it was." That sorta doesn't work terribly well. Now let's turn our attention a little bit to scale. Obviously core plug samples are small and logs are large. The difference in volume measurement between the two could be 1,000 or maybe in painful cases as much as 1,000,000 difference. So to assume that you're gonna get a correlation between cores and logs is maybe dreaming. It's remarkable in my estimation that is works as well as it does. Here is an interesting little plot giving depth of investigation on the horizontal axis and vertical resolution on the vertical axis. And the different logs, and you can see the micro-logs, micro-devices down in the one inch range right down the bottom left-hand corner. Whereas the induction log is in the top right-hand corner and it has a resolution of maybe four feet and a depth of investigation of maybe four feet. So huge differences, and all the other logs somewhere in between. So to expect correlations from one log to the next even is kinda dreaming. So here is a statement of that same graph that we saw just previously giving the approximate volume in cubic feet of a gamma ray log as compared with an induction log. Huge differences as you see. So what we do is to upscale. What we do is to put together an upscaling algorithm. And what it does is at any one level takes most of the information, the majority of the information from that level, but also considers adjacent levels at less and less weighting as you go around, go away. And then use that as a moving window. It's remarkable what it does, as we'll see. It's an enormous aid in depth shifting because the upscaled core is much more of a continuous core, and it also takes out, in quote marks, the noise. And we'll see that very well shown on the next slide. On the left hand slide is the raw data comparing core porosity with petrophysically-determined porosity. In the middle is shown the upscaled data, and you can see that all that noise has disappeared. It's now much more a continuous curve. And if you go right down to the bottom where the circle is, you can see that there's a clear mismatch of the cores with the logs. There's no way that you could see that on the raw data. Then the shift until that data agreed. Similarly, right at the top there's a bit of a mismatch. And again then the data shift, a much better match. But the interesting thing is that in between the, to me at least, the upscaled core data seems to be a better match than the shifted core data. So that means that we didn't quite do our homework correctly. In fact there's a differential shift that we should apply. But you can see the power of this. It's taking out all the spurious noise, making the curve a lot easier to interpret. There are all sorts of other applications. If any of you are doing equity work give us a call because upscaling core data, if core data is part of the requirement, can change your equity remarkably. We've proved it.