Evaluation of Geostatistical Techniques for Reservoir Characterization

Abstract One of the most important problems in reservoir management and EOR applications is to describe interwell heterogeneity for optimum well spacing and process design. The conventional approaches to build reservoir descriptions use interpolation schemes based on the geometric distance of the samples. In geostatistics, on the other hand, the interpolated values depend on geometric orientation and correlation. Geostatistics has been used in the mining industry for the past forty years and in hydrogeology for the past twenty years. However, its application has been limited in reservoir engineering because of skepticism due to a lack of needed data and the absence of comparisons with the conventional techniques. Conventional and geostatistical reservoir characterization techniques are compared in this paper. Field data from about 50 wells are used in this evaluation. Interwell properties are computed using conventional and geostatistical techniques, cross-validated, and compared. Histograms show there are no geographically distinct populations requiring separate modeling. Experimental variograms identify the scales of variability of the reservoir properties and the existence of preferential directions of maximum and minimum variability. Variogram models are fitted to the experimental correlations. The results of this work indicate that the success of any technique is a function of the geological environment, selection of proper reservoir parameters for that environment, an adequate number of samples, an appropriate sample spacing, and most importantly, the correct application of the technique. Geostatistical techniques have more potential to incorporate the underlying correlation present in the data in the final analysis.