Quantification of Variogram Range Uncertainties and Their Impact on Stochastic Property Distribution and Dynamic Model Variability

Geostatistical simulation techniques are the most common way to quantify reservoir uncertainties by generating multiple realizations, where each realization represents an equiprobable model. The geostatistical simulation algorithms do require various input parameters for uncertainty ranges and the common practice is to use scaling factors to the base value to compute multiple realizations ( Frykman and Deutsch, 2002 ). This may defeat the purpose of generating model variability, hence it is always preferred to quantify uncertainty ranges using representative data for more reliable equiprobable models.

). This article presents an experimental approach to measure and quantify variogram uncertainties and derive a best-fit variogram for spatial forward modeling of reservoir properties and subsequent integration in dynamic simulation models.

We demonstrate that the use of optimized variogram ranges improves the geological realism of generated property realizations, which ultimately contributes to a more efficient dynamic simulation model updating and calibration, referred to as the history matching process.