Previous Lesson Complete and Continue  

  1. Background and Introduction

Lesson content locked

Enroll in Course to Unlock
If you're already enrolled, you'll need to login.

Transcript

- First thank you to everybody for dialing in today. I know it's first thing Monday morning, so I appreciate you taking the time out of your schedule. Hopefully I can keep you going through your morning coffee or whatever you, uh, maybe your morning pastry or whatever you have going this morning. I wanna talk today about some lessons we've learned in core analysis, mostly in the Permian basin, but these apply not just to the Permian but core analysis in general. And really a comparison study that we did and originally presented at the 2017 symposium between Retort, GRI, and Routine Methodologies. I also wanna just thank my co-workers up front. They've been really instrumental in getting this together and really a critical part of the story. Tyler Croft, Brian Driskill, Brian Tepper, they're all great resources and have been really helpful throughout this. So first I have to show this slide. This is, you know, the legal slide. Just don't trust what I say, I think is the overall summary. So just as a background, what I'm gonna do first is I'll go through a little bit of background and maybe a little bit of philosophical discussion around core analysis, and then I'll really get into the topic of the paper. And, as mentioned, the paper's also, I think it's uploaded into the handout section of the webinar. So if you wanna download the paper and read through it, I strongly encourage that. There's a lot of information in the paper. And at the same time, a lot of this stuff is designed to either, you know, provoke questions or really stimulate discussion. And I'd really encourage you to either submit questions, you can submit questions using the webinar, or just email me separately, however works best. And I'm really happy to discuss any of these topics. So as a background, you know, I've been working, I've been at Shell as mentioned for about five years now. I've been working for a few years now in the Permian Basin, which I'm very excited to do, I love working with Permian. I think it's a very thriving, exciting basin to work. And Shell is focused primarily in Loving and Ward Counties. If people aren't aware of our acreage position, we're really in the heart of the Delaware Basin. And really the goal of this study is when you start comparing area to area, what we started observing is systematic differences in shifts between core analysis. And then you trying to reconcile, are these truly geological shifts? Are we seeing difference in shifts because of core analysis technique or core analysis vendor? And then how to you get everything on an apples-to-apples comparison? How do you really trust what you're seeing? Because ultimately, I think, we can all agree the goal of petrol business is to keep pushing forward, and helping the operator working for the service, you know, the service they're providing. And ultimately to deliver better wells. And so if we don't become predictive, and develop this predictive power, I think, you know, we don't really carve the niche of our business. So into this philosophical discussion, the first thing... And I presented these slides last week and hopefully my math is right throughout these, but this is what I call the fundamental problem of upscaling. So this is, as you step through, you know, from petrology-level analysis into SENs and thin-sections, all the way through into the challenge of today in unconventionals, which is trying to develop petrophysics, to develop a, say, a two-mile lateral. So we're really getting into large reservoir scale petrophysics that really is very, very new to the industry. And so the first thing I have on here is a 3-D, you know, if I did a 3-D FIB-SEM, on the left hand side that's a very small volume. And I had to do a hundred billion of these to get to get up to a core plug resolution. Which is a significant upscaling problem. And then, that's equivalent to if I took my doormat at home and I chopped it into three pieces, and then I drop it somewhere inside the Greater Houston Metropolitan Area, and try and use it to characterize the entire area. So there's probably places that that will work, and I can probably get representative samples, but there's also probably things that I'm going to miss. And if I take that to the next step, I go from core plugs, so I need a 100 billion SEMs to get to a core plug, and then I need, you know, say, 1500 core plugs to really get to a single log measurement. Which is a significant upscaling effect again. And then to get to the near-wellbore reservoir at a one-mile lateral I need two point four, you know, 16,000 log measurements, and then if I get to the stimulated volume after I frack, and I'm trying to characterize what my stimulated rock volume is, petrophysically, I need another 2500 of those. So I have this huge order of magnitude in upscaling as I move through different types of measurement, and that just really speaks to some of the issue here in core analysis and trying to get this right the first time. So onto this, which is the core dilemma and what I have here is a breakdown of, you know, and people can debate this, but this is a breakdown of a single piece of rock. So if I took a single core plug, you know, and this is gonna be, this is gonna vary based on, you know, where I am, geographically along with geologically. But just some of the key components that I think we need to be able to provide in order to be useful. And I have this statement up top which is, "As petrophysicists, we need to provide predictive, "reliable, and prepetable quantifications of the rock "in order to move the needle for the business." And I think that's a really critical, just, way to think around these. We're trying to perform rock analysis here, want to be technically correct, but also to drill better and better wells, and really help the business to continue to perform well. That's both on the service side and the operating side. And you've got a lot of components here, you've got clay volume, clay bound water, organic components whether it's TOC, kerogen, bitumen, the porosity that you might have within that, so organic porosity. Obviously I've got a full mineral matrix, limestone, dolomite. And then my fluid components are filling up the rest of my matrix porosity which is water, gas and oil. And how that might vary from down-hole in situ, to then what I end up with at the surface, and then how do I test that using core analysis. And the other comment I'd add at the bottom here is with a finite budget, you get a handful of opportunities to get this. So we're not, you know, I don't think anyone's out there taking core on every single well in unconventionals. It's just not the way that things are done anymore. You really get a handful of chances to calibrate, develop the petrophysical model that you're interested in, and try and provide predictive power towards the business. And the other thing I just mentioned there, and this is not really the focus of today, but the significant fluid loss as I move from in-situ to the surface, and the example here I use is closer to the rocks I'm gonna show today which is Permian Basin rocks. So, in-situ we've got a significant amount of water in the matrix, and a significant amount of oil in the matrix porosity. And, you know, if I give an example, that's a 50-50 split, how does that change as I bring the sample to the surface. And what you typically observe is a significant amount of fluid loss and this is not unique to the Permian. This is typical across many, many unconventional plays and conventional plays themselves. So you end up with 30, 40, 50 percent gas-filled porosity. And then you trying to re-characterize what you do measure in the rock to what it might have been in-situ. So that's another topic, but just an uncertainty that gets brought into here. And then today, really, the focus is on everything except for the question mark on the right hand side. So trying to get the correct components of what I have left at the surface. And that's really what I'm gonna hone in on. So where are we comfortable and where do we have gaps? And these are the different analyses to try and get at the components that I mentioned in my little rock schematic. So you got mineral components, then you can do XRD, XRF. And we probably have good confidence in that. And then what I have on the last column over here is, and this is my personal opinion, right, this is my personal opinion of green light, yellow light, red light. So it's super-simplified. But how that might, the confidence I have in the analysis, and then how that might upscale to, say, a reservoir. Reservoir property. So we can do XRD, but we use a few grams. We might use five grams to do XRD. I'm sure it's very very accurate in those five or ten grams, but how does that upscale to reservoir scale? Similarly on some of these, you know, if you get into organic porosity, you know, and I mentioned the significant upscaling at SEMs, there's a huge amount of upscaling that has to take place there. And so, can you ever get a super-accurate reservoir representation of organic porosity? It's probably going to be challenging. And similarly with TOC. TOC is a very simple, relatively cheap, easy measurement to make, high reliability. But it's typically done on one gram of rock, if that. So you've got a significant upscaling just to get to core plug level, and then definitely back to reservoir scale. And the one I want to focus on today is none of the topics I just mentioned, but total porosity determination and then fluid saturation. So this is one that, I think, I think there's a sense that we have a better grip on as a community than we maybe do. I think there's still some significant uncertainty around reconstructing true saturations, and even accurately determining the saturations of what we end up with at the lab. And then, to me, and I've given this a green light on upscaling, I really do feel like, you know, you can take a few core plugs here and get closer to a representative scale of the reservoir. That you can actually use, let's say, reservoir modeling, and more integrated, a more integrated approach across the business. And again, back to delivering better and better wells. So let me set the stage and now I'm gonna get really more into the topic of the SPWLA presentation and the paper itself. So we observe this difference. As I mentioned before, this is a compilation of several wells between two vendors, vendor A and vendor B. And all I'm showing here is core porosity against core bulk density. So a very simple, typical cross plot, you might say, to QC core data or try and understand what you're looking at. You can see a significant shift, and this is just vendor separation. So you've got vendor A in orange, and that stays constant throughout this presentation. And then vendor B in green. And you see a 20 to 30 percent difference throughout the scale on total porosity. And again, these are unconventional samples, so you get up to, say, 10 percent porosity from one vendor, and the other vendor's telling you up to 13 or 14 percent porosity. So you're gonna get significant shifts. And then the next assumption of that is you're going into a significant shift in fluid volumes, fluid saturation, and how those fluids might flow to the surface. So how do you go area to area, and really compare if you can just see a trendology shift between vendors? It really makes life very very difficult. And you can't really also go and say, "I have a 30 percent in-place volume error." That's probably too significant for anyone to really stomach and be happy with.