This paper outlines details of a method that includes uncertainty in deterministic simulations. The method uses the probability distributions of reservoir and geological properties to obtain a likelihood distribution of the output variables. The Monte Carlo sampling (mcs) method is often applied, but it is impractical for simulation studies because the number of computations becomes excessive. This paper uses an analytical thermal screening model to compare various variance-reduction methods with MCS simulations. The Latin hypercube sampling (LHS) method is then selected to conduct numerical simulations. The methods allow computation of the partial correlation coefficients, which, as demonstrated, can be used to determine the sensitivity of process variables.

Keywords:
output variable, stochastic modeling, modeling & simulation, upstream oil & gas, correlation coefficient, water production, spe reservoir engineering, water leg, partial correlation coefficient, screening model
Subjects:
Evaluation of uncertainties, Flow in porous media, Improved and Enhanced Recovery, Reservoir Fluid Dynamics, Reservoir Simulation, Thermal methods
Copyright 1992, Society of Petroleum Engineers
You can access this article if you purchase or spend a download.