- Jef:In this short presentation, I will present an overview of how to model petrophysical properties using sequential Gaussian simulation. Let's consider first the purpose of such modeling. Here we see a simple example which illustrates the effective area stages in such modeling. First, we need to model the structural framework of which we then build a structural grid. This grid needs to be filled with petrophysical properties because these high resolution petrophysical properties will then later be used for upscaling and for flow simulation. Note that the modeling of such petrophysical properties is typically done within each facies, because each facies, whether litho-facies or sedimentary facies, has a very distinct population of porosity and permeability. What is required for sequential Gaussian simulation is evidently a stratigraphic grid, in which we can fill these various properties. This would require modeling the structure first. Next we need to have permeability and porosity data, obtained from wells. Either as cores or well logs. And that may require some interpretation and processing. Next, we need to decide on what the histogram of the petrophysical properties is, within the entire reservoir. This is an important step because we may have, we may need to be aware of biases that do occur when we extract such data from wells. Wells could be preferentially drilled in high pay zones. Next, we also need to have the varlogram of petrophysical properties. That varlogram needs to be calculated and modeling before we can run sequential Gaussian simulation. Here is a simple example of how we would execute such sequential Gaussian simulation. It is done with a public domain of software called SGeMS, which you can freely download and try out. But it's essentially the same in other commercial software such as sgems. What we have here is a grid, and with that grid, inside this grid, we have wells that are drilled preferentially in certain areas of the reservoir. And along these well bores, we have measurements from well logs of porosity. The aim here is to build a 3D model of porosity, constrained to this well of data. And reflecting, of course, the histogram and the varlogram information that we will provide. The input in most of the software is essentially the same. So I'll use here the SGeMS input which is rather simple. The first thing as I stated before, is that you need to specify stratigraphic grid. The name of the property that you're simulating. And how many realizations you would like to simulate. Secondly of course, we need to provide the well data, along which we have measured porosity from cores of wells. Next, we also need to specify what is our target histogram. As I mentioned before, this is a difficult, or at least a precarious decision. And in this case, I decided not to use the histogram extractor from the well data, but histrogram from some other information. Either from analog information or a histogram that you have constructed yourself by doing some kind of a bias correction. Next of course, we need to specify the variogram. And in this case, in many softwares, we can load the existing variogram model that has been modeled previously. Such model would then specify target effects. The type of variogram, the nesting, the various ranges and angles. After having generated realizations such as one realization shown here, it is very important to do a quality control. Such quality control would consist of calculating realization to the variogram in various directions, to show or at least to check whether this variogram is properly reproduced. In this case, this is nicely done. Because as you know, there's the horizontal variogram has a larger range of 40, where the vertical variogram has a smaller range of five, as we have specified in the input. We also need to check whether the histogram is reproduced. And as you notice here, the histrogram is by model. And, you can also check some summary statistics. By making slices through the model, we would like of course, also to check whether we have constrained properly to the well data. And this is nicely done here as is shown. You have your two wells. You take a section through the 3D model and you can clearly see that where areas of high porosity occur in the wells, we also notice area of high porosity in the generated reservoir model. So, that's pretty much it. In summary, sequential Gaussian simulation allows for very fast modeling of 3D petrophysical properties. Based on the data provided, such as histogram, variogram and well data. It is also very important to do a quality control to verify that the input constraints that have been provided are indeed respected.