PRODUCTION FORECAST FOR BAZHEN FORMATION RESERVOIRS ON THE BASIS OF STATISTICAL ANALYSES AND MACHINE LEARNING TECHNIQUES

Abstract

The paper reviews actual production results and production forecasting for horizontal wells completed using multistage hydraulic fracturing in Bazhenov formation of Palyan area, Krasnoleninskoye oilfield (HMAO). The key feature of these reservoirs is high degree of uncertainty in geological and geomechanical properties. The primary purpose of this study is to provide both identification of correlation between well productivity index (rates, cumulative production) and general geological-technological factors and software development for production forecasting of Bazhen formation wells. In the course of the works, more than 2000 probable correlations between geological and geomechanical formation properties as well as between technological factors and well performance have been analyzed. Main key complex parameters (length of horizontal section, number of frac stages, average tonnage of proppant and fracturing fluid back-production) that govern producing well performance have been determined in this analysis. The approach for evaluation of forecast well performance (initial rate, cumulative oil production) is provided on the basis of obtained statistical correlations and machine learning techniques. Approaches described in this paper summarize the first results of the phase «Select» in progress of pilot project on Palyan area, Krasnoleninskoye oilfield.