부산대학교 도서관

The relevance of the research is caused by the need to select from a variety of parameters of a petroleum reservoir model the most important ones. By adjusting these parameters, the optimization algorithm should find solutions that ensure the correct reproduction of historical data by the model. History matching is a time-consuming and inevitable step in numerical modeling of hydrocarbon reservoirs, which confirms the suitability of the model for predictions. Limiting the number of tuning parameters to the most significant allows the model to be successfully history matched in practice within a reasonable time frame. The main aim of the work is to determine a set of significant history matching parameters based on the study of the degree and nature of their influence on the convergence of key performance indicators of the wells and the field. Object: automated history matching of a petroleum reservoir model to the data of a reference case, which is an analogue of historical indicators of real field development. Methods: numerical geological modelling and reservoir simulation, computational experiment, sensitivity analysis. Results. The authors have developed the classification of history matching parameters according to the degree of influence and the degree of uncertainty based on the law of fluid flow in a porous medium. Based on multiple computational experiments on the synthetic reservoir model built during the work we studied the degree and nature of the influence of various history matching parameters on the convergence of field development indicators. The use of two types of sensitivity analysis made it possible to characterize both the individual influence of the parameters and the combined influence taking into account the variation of the remaining parameters. Understanding the mutual influence of the parameters is important, since during history matching all tuning parameters are varied simultaneously. The conclusions of the numerical experiments are consistent with the analytical approach. The implemented comprehensive approach makes it possible to form a reasonable set of tuning parameters of the model, ranked according to the degree of importance for history matching.