- In the previous module, we looked at high-productivity heterogeneity in naturally fractured reservoirs and discussed the fact that it can make prediction of fractured reservoir behavior extremely challenging. Let's turn our attention now to exploring the causes of heterogeneity in naturally fractured reservoirs. It's usually a combination of a group of factors, all of which are basically geological factors. It's a function of well-fractured intersection probabilities. That is, how we sample the fracture system with our borehole. Fracture size is important. That is, the height, the length, and the aperture all have their own contributions to variability. A factor I'll call geologic heterogeneity. And this is enlargement of apertures, occlusion of apertures, or rock property-based variability in occurrence of fractures and their character. And, finally, I'll talk about fracture connectivity variability based on multiple fracture sets and their variability and, also, density of fractures in those sets. Let's begin by just getting a visual image in our heads about what fracture systems in nature look like. This is a photograph showing an outcropping of sandstone. You can see bedding that dips gently to the left. Fractures are roughly perpendicular to bedding. And we see some general characteristics of fracture sets in this outcrop. These fractures are known as joints because they look like the joints in a brick wall. The joints or fractures are formed parallel to other fractures in order to define the joint set. The joints are perpendicular to bedding. They commonly terminate at mechanical bed boundaries. And their more widely spaced in thicker beds. That was in a sandstone. Now, we're looking at some bedded carbonates. And this is a nice image because we get a 3-dimensional perspective on several joint sets. You can see we've got three main beds that are present here that are dipping gently off to the right. And there're several fracture systems present. I've highlighted one set here in red. There's a second set of fractures that are perpendicular to those, and I'm, highlighted a few of those in green. And, finally, there's a very sparse set also present showing up in blue here. These are joints, they're in sets of parallel fractures, they are perpendicular to bedding, they commonly terminate at mechanical bed boundaries, and they're more widely spaced in thicker beds. In other words, the fractures or joints in this layered carbonate sequence have essentially the same characteristics that we see for joints in the layered sandstone sequence that we considered previously. Now let's look at the impact that individual fractures can have on production from a well. The image on the left that you're looking at is resistivity-based image log from a well. The vertical scale here is two meters. The image is a 360-degree image around the interior of the borehole. It's been unwrapped and laid flat. When a well intersects a fracture or any planar feature, it will show up as a sinusoid in that image, and we see that here as a sinusoid. And dash line follows that sinusoid, and it's labeled fracture. Now, let's look also on the right side of the slide here. Let's compare. There's a well log here. It's a PLT, or Production Logging Tool, flow profile, and what it shows is cumulative production from the bottom of the well working our way up. Now, the scale is different here. This is about a 25 meter interval in the borehole, but there's an, I've outlined a rectangle here that just looks at a two meter interval, and it's the same two meter interval that we have the borehole image log for. And you see the cumulative production which is shown in red here jumps markedly right at that point, and that increase in production is about 2,000 barrels per day. What you see is representative of about maybe 3/4 of the production coming from that well. So, what that's telling us is that one fracture is responsible for 3/4 of the production coming from this well. If we had offset the well a few meters in any direction, we might have missed hitting that fracture. We couldn't count on it, of course, if we were just offset a little bit. And we would have had a very different result in production performance from that well. So, that's why, when we offset, truly do offset a well and drill the next well in the neighborhood, we may or may not hit a fracture there. Or we may hit several fractures all of which contribute a lot of flow into the well. Whatever we find, it's going to be a hit-or-miss affair. It's probabilistic in the basis of our sampling is probabilistic. We're not guaranteed any particular property of this fracture porosity system when we go in a drill a new well. And so, what that leads to is extreme well-to-well heterogeneity in production performance. Let's look a cartoon that explores the incidence of boreholes with fractures in the subsurface rock. As you can see, I've drawn three layers that contain fractures, and we're going to consider drilling into these layers. I'd like to simplify the story a little bit, but we'll just look at the 2-dimensional image here. So, we're trying to sample a single fracture set in this case. And the way that I've drawn the fractures is such that in the upper-most layer, Layer 1, the spacing of fractures is equal to the diameter of the borehole. In Layer 2, the spacing of the fractures is equal to two times the diameter of the borehole. And in Layer 3, the fractures are spaced four times the diameter of the borehole. So, we're maintaining a fracture spacing-layer thickness relationship in this image. Now, let's draw a well. And this first well that we drill, Well A, intersected a fracture in every layer. So, we conclude, "Well, we've got a high fracture frequency in this formation." So, we offset, drill a second well, and we don't encounter any fractures in this second well. So, it's a probabilistic sampling game. We could conclude that the fracture density varies between Location A and Location B, but that's fundamentally not what's happening. It's just that we've hit fractures in A, and, just by chance, we missed all the fractures in B. So that raises the question: should we use convention geostatistics to try to predict fracture distributions? Generally, people that have tried to use conventional geostatistics, that is, very ground-based geostatistics in order to model fracture systems, are usually unhappy with the results. I mentioned the representative elemental volume. Let's look at that concept a little more thoroughly here. The representative elemental volume, or REV, is the minimum volume over which a physical property can be measured and serve as an average value over a larger volume. So, it's really a sampling to get a representative value of a parameter. And as you see on the right-hand side of the slide in the lower right, this is a cross-section of fractured layers. The bedding is horizontal, and the fractures are near vertical. And above that is a cross-plotted X-Y plot that shows value of the parameter being measured versus sample volume size with volume increasing off to the right. Let's consider a parameter that relates to natural fracture systems. Let's think about fracture porosity here. So, if we could come in and measure fracture porosity over a very large volume in the case shown here in multiple layers, then we might get a stable value of the measured parameter, or fracture porosity. In other words, if we move this sample ellipse around, up or down or off to the side, we might get a value that would be about the same as what we're seeing here. How might we sample such a large subsurface volume? Maybe it would be using pressure transient tests, for example, which measures, commonly, a very large volume in the subsurface. It's not a very precise test, but it can be a very accurate test. Now, suppose we were sampling over a smaller volume. In this case, it might be more like a deviated well coming along a horizontal layer where we're just sampling in one layer. And in this case, maybe we'll measure a different value for that fracture porosity, but, again, it's a smaller volume but making a stable measurement, a stable representative measurement of fracture porosity. But when we come down to, let's say, a vertical borehole, and we're just sampling just a very small volume, maybe we're only taking a few cores out of this fracture reservoir, we see very high variability in our measured value from point to point so that, if we change our sample location just slightly, we can get wide swings in the value. And we're definitely below the representative elemental volume for this parameter. So, what we're gotta conclude here is that small changes in location lead to very big changes in property, and to assess the representative elemental volume, we need to integrate geologic and flow-related data. That is, we need to look at the fine details to better understand the character of the fractures, but we need to look at the larger-scale response of the reservoir, and we need to integrate those together. And that's really a challenge when it comes to people who are specialists in working with naturally fractured reservoirs, because we've got to be able to work with and understand both the geological data and the reservoir engineering data that relates to it.