Back

Biography

Evgeniy Kovalevsky has graduated from the Moscow Physical-Technical Institute and he has completed his Postgraduate study in the Institute of Oceanology of the Russian Academy of Sciences. Further, he has worked in the Department of Geophysics in the Institute of Marine Engineering Geology, Riga. He has also worked in the Central Geophysical Expedition in Moscow, dealing with development of DV-Geo geological modeling system. His area of interests includes geological models of natural hydrocarbon reservoirs, fuzzy geological models, geostatistics, hydrodynamic modeling, abiogenic theory of petroleum origin and has published more than 30 scientific works.

Abstract

An interpolation of well data require taking into account the following: (1) A geological environment is of a categorical nature and it is thus impossible for the quantitative properties to be interpolated among different categories. The quality of property interpolation in such an environment is controlled by comparing histograms. (2) The interpolated values should represent the actual variability of a geological environment, where the term actual means meeting a certain criterion. This requirement can be achieved only by means of multiple stochastic realizations. Where a variogram is used as a mentioned criterion, the quality of interpolation is controlled by comparing variograms. (3) It is geostatistical techniques that are mostly applied to compute stochastic realizations. In all these methods it is assumed that the geological environment is statistically uniform. Is a real geological environment statistically uniform? Certainly, it is not. So, in stochastic methods, the key to a successful implementation is dividing the interpolated parameter into deterministic (non-random) and random components. (4) The deterministic features of initial well data (trends, categories, anomalous zones, etc.) are not obvious and if not specially marked, will be erased in the realizations. Stochastic realizations are sterile with respect to the deterministic features and for this reason are not preferred by geologists. (5) In order to overcome this sterility, a number of non-classical (heuristic) geostatistical methods have been developed. The object modeling technique generates realizations which include sand bodies of a particular shape. Being able to manage the shape of individual bodies, the geologist however cannot manage the configuration of the set thereof. (6) To enable managing such combinations, multiple-point statistics (MPS) has been introduced and employed. Its heuristics consists in using a training image. Advantages and disadvantages of different geostatistical methods for borehole data interpolation are discussed.