Reservoir Characterization with Limited Sample Data using Geostatistics

Abstract

The primary objective of this dissertation was to develop a systematic method to characterize the reservoir with the limited available data. The motivation behind the study was characterization of CO2 pilot area in the Hall Gurney Field, Lansing Kansas City Formation. The main tool of the study was geostatistics, since only geostatistics can incorporate data from variety of sources to estimate reservoir properties. Three different subjects in geostatistical methods were studied, analyzed and improved. The first part investigates the accuracy of different geostatistical methods as a function of the available sample data. The effect of number and type of samples on conventional and stochastical methods was studied using a synthetic reservoir. The second part of the research focuses on developing a systematic geostatistical method to characterize a reservoir in the case of very limited sample data. The objective in this part was the use of dynamic data, such as data from pressure transient analysis, in geostatistical methods. In the literature review of this part emphasis is given to those works involving the incorporation of well-test data and the use of simulated annealing to incorporate different type of static and dynamic data. The second part outlines a systematic procedure to estimate the reservoir properties for a CO2 pilot area in the Lansing Kansas City formation. The third part of the thesis discusses the multiple-point geostatistics and presents an improvement in reservoir characterization using training image construction. Similarity distance function is used to find the most consistent and similar pattern for to the existing data. This part of thesis presents a mathematical improvement to the existing similarity functions.