Statistical Modeling of Oil Reservoir Life Cycle

Abstract

Abstract Traditional approaches to the problem of oil field life cycle partitioning onto the successive stages is mainly based on heuristic evaluations with no well-defined criteria. To solve this problem, we proposed a stage wise structuring based on the identification of the current oil production distribution in the individual stages and isolation of adjacent stages' boundary points defining the duration of each stage. Using statistical analysis of experimental data on development of a large number (more than twenty) of specific fields reduced into one "generalized deposit" followed The Savitzky–Golay smoothing filter shows that the life cycle of the field can be divided into four successive stages which are definitely described as logarithmically normal, exponential, Pareto and Weibull distributions Using non-linear logistic model for the cumulative oil production in the last fourth stage of the life cycle a computational procedure has been developed for assessing the initial recoverable reserves, as well as the method for estimating the maximum level of oil production has been proposed. The method is based on finding inflection point of the curve describing the dynamics of cumulative oil production, by examining second order differences. Based on the values of cumulative oil production in the later points of the final development stages oil production was predicted and the design parameters for the current development system were evaluated using adaptive Kalman filter in discrete time.