Method for exploiting a geological reservoir from a reservoir model matched by the computation of an analytical law of conditional distribution of uncertain parameters of the model
a reservoir model and model technology, applied in the oil industry, can solve the problems of significant computation time, and achieve the effects of reducing computation time, faster processing of results, and better approximation of objective functions
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Benefits of technology
Problems solved by technology
Method used
Image
Examples
example 1
[0117]To illustrate the method, an exemplary synthetic application is described. The reservoir considered contains gas, oil and water initially. It is sealed by faults on two sides, and by an aquifer on the other sides. Production is done by depletion from six wells (PRO1, PRO4, PRO5, PRO11, PRO12, PRO15) with an oil production flow rate of 150 m3 / day imposed on each producing well for 10 years. During this production period, the aggregate of oil (V) and the gas / oil ratio (R) are collected for each producing well, as is the total aggregate of oil in the reservoir. These values are known with a relative noise of 4%. These data are illustrated by the curves of FIG. 6. FIG. 6a) corresponds to the volume (V) of the aggregate of oil for each producing well as a function of the time t expressed in years. FIG. 6b) corresponds to the gas / oil ratio (R) for each producing well as a function of the time t expressed in years.
[0118]The distribution of the petrophysical properties (porosity, perm...
example 2
[0126]This example makes it possible to detail the proposed methodology on an analytical example, which enables us to validate and study in depth the possible analytical forms of the a posteriori law for different cases. All the examples discussed here have been processed with the Matlab® software (MathWorks, United States) and the “Dacefit” toolbox to construct the kriging.
[0127]For this example, the flow simulator f is replaced by a linear model f(ti, θ)=tT, θ where θ=(θ1 . . . θD) is a vector of dimension D.
[0128]To test the matching method of the method according to the invention on this linear model, the following procedure is repeated for D=1, D=2 and D=3:
[0129]1. the reference value of 0 is fixed
[0130]2. n=10 points are chosen in the interval [0,1]D according to a latin hypercube, these points are denoted ti, i=1, . . . , n and the observations yi=tiT θ+εi are generated by randomly generating Gaussian variables εi by fixing on a value σε2. On the one hand, this step makes it ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


