Geostatistical integration of core and well log data for high-resolution reservoir modeling
Burch, Katrina M.
Metadata[+] Show full item record
Analyzing data derived from well logging and core plugs to understand the heterogeneity of porosity in geologic formations is paramount in petrological studies. The well-log data and core-plug data are integrated in order to generate an accurate model describing the porosity distribution; however these data exist at different scales and resolution. This difference necessitates scaling of one or both sets of the data to aid in integration. The present study established a geostatistical scaling (GS) model combining mean, variance, skewness, kurtosis and standard deviation with a misfit algorithm and sequential Gaussian simulation to integrate porosity data in conjunction with correlating the depth of core-plug data within the well-log data through a scaling process. The GS model examined well-log porosity data from a Permian-age formation in the Hugoton Embayment in Kansas and well log data from a Cretaceous-age formation in the GyeongSang Basin in The Republic of Korea. Synthetic core-plug porosity data was generated from well-log data with random number generation. The GS model requires basic histograms and variogram models for scaling the computerized tomography (CT) plug data to well log scale as well as integrating the data in a sequential Gaussian simulation. Variance-based statistics were calculated within specific intervals, based on the CT plug size, then a best fit for depth correlation determined. A new correlation algorithm, named the multiplicative inverse misfit correlation method (MIMC), was formulated for accurate depth correlation. This associated depth then constrained the well log porosity data at reservoir- or field-scale to interpolate higher-resolution porosity distributions. Results for all the wells showed the MIMC method accurately identified the depth from which the CT plug data originated. The porosity from the CT plug data was applied in a sequential Gaussian co-simulation, after kriging the well log data. This culminated in a greater refinement in determining the higher porosities distributions than the interpolation of solely the well log data. These results validate the proposed high-resolution model for integrating data and correlating depths in reservoir characterization.
Table of Contents
Introduction -- Geostatistical framework -- Formulation of the geostatistical scaling model -- Applications of the model: case studies -- Discussion of results -- Conclusion -- Appendix