Published: June 17 Petroleum Society of Canada. You can access this article if you purchase or spend a download. Sign in Don't already have an account? Personal Account. You could not be signed in. Please check your username and password and try again.
Sign In Reset password. Sign in via OpenAthens. Pay-Per-View Access. Buy This Article. Annual Article Package — Buy Downloads. View Your Downloads. We can choose to either decrease the top layer perm from to md , or we can increase the bottom layer perm from to There are a very large number of combinations of the 3 variables that will give In terms of what is allowed to vary in the base case, any conclusions of probable description or behavior that may be inferred from any chosen Px case, or from differences between any two chosen Px and Py cases, are virtually guaranteed to be wrong!
To efficiently compute probabilistic results from data uncertainties, the number of uncertain variables, generally equal in actual number to many times the numbers of gridblocks and wells, must be minimized. We must strive to build and evaluate as many of the fastest and coarsest and least detailed models that are sufficient as fast as we can, rather than the most detailed.
Detailed modeling is valid only at the fine scale, for subsequent upscaling to field-scale problems that can be practically solved. In our discrete upscaled numerical models, any surface or feature is justifiably represented by nothing more than large coarse-block average permeability and porosity and rock type distributions. Conforming the grid to detail or surfaces is counter-productive. All issues of behavior must be investigated with respect to the probabilistic results given by a large set of possible realizations.
That includes history matching and optimization. We can determine the uncertainty in our predictions only by basing them on many equally probable history matches or scenarios.
In optimization, questions that can not be represented by changes in the data sets history matches can not be answered by reservoir modeling.
The question of whether or not Option A is better than Option B is represented in all the realizations by some change in the data. The probability that A is better than B is given by the fraction of case A runs giving better results.
A simple component of probabilistic forecasting and optimization workflows for Sensor is provided by SensorPx.
The Px,y,z results are output in files casename. Exceedance P10 is a high, optimistic estimate, and exceedance P90 is pessimistic. Cumulative P10 is a low, pessimistic estimate, and cumulative P90 is a high estimate.
Both exceedance and cumulative probabilities are commonly used. The terms "at least" and "at most" appear in the above definitions because Pxi and Pyi values can be the same. Usually, if enough cases are run and if significant uncertainty exists, Pxi results will be continuously variable in x, and the terms "at least" and "at most" do not apply.
Runs for real cases in which the wells remain rate-limited at all times are very rare. Odeh, A. So how does that help us?
Well we can say things like: there are leaves bigger than the smallest leaves or there are leaves bigger than the medium size leaves. We can do the same exercise with the continuous frequency distribution in Figure 1 and we end up with the following continuous cumulative frequency distribution:. Although this looks terribly mathematical, its similar to the graph you have just produced with the leaf example.
You should be able to see that the shape described by the top of the green boxes in Figure 3 looks very similar to the shape of the red line in Figure 4.
Figure 4 looks smoother than Figure 3 because Figure 4 was created from the smooth continuous distribution in Figure 1. These estimates are usually termed the P90, P50 and P10 confidence levels. Using Figure 4, the estimate at the P90 confidence level is Its just the way the scale is presented it has been normalised to zero at the middle. This is NOT the same as the chance of that estimate occurring.
The chance of a single estimate occurring can be read off Figure 1. If we ask the question a different way: from Figure 4, what is more likely to occur more frequently - P90 or P50? To help you, you cant actually answer the question from the cumulative frequency distribution Figure 4 and you will need to jump from the cumulative frequency curve Figure 4 back to the frequency distribution Figure 1.
An easier way to understand the question would be to use the leaf example, assume P90 is the same as the small leaves and P50 is the same as the medium leaves.
So the question becomes: what is more likely to occur the small leaves or the medium leaves? So in a normal distribution, the P50 value is more likely to occur than the P90 value. In simple general terms, that is why P50 is sometimes also known as the best estimate because its the estimate that occurs more frequently.
An oil or gas estimate is calculated by multiplying together a number of parameters, for example:. Oil in place equals rock volume of the reservoir multiplied by porosity multiplied by oil saturation there are actually a lot more input variables but let us keep it simple for now.
Rock volume, porosity and oil saturation are measureable things. There is however uncertainty surrounding the measurement of those parameters. Integrated asset modelling in production forecasting.
Long term verses short term production forecast. Look backs and forecast verification. Material balance models in production forecasting. Probabilistic verses deterministic in production forecasting. Production forecasting activity scheduling.
Production forecasting analog methods. Production forecasting building blocks. Production forecasting decline curve analysis. Production forecasting expectations. Production forecasting flowchart.
Production forecasting frequently asked questions and examples. Production forecasting in the financial markets. Production forecasting principles and definition.
Production forecasting purpose.
0コメント