The article is devoted to the mismatch normalization as one of the aspects of the objective function formulation for automated history matching of hydrocarbon reservoir models. To carry out history matching, it is necessary to define the objective function that describes the discrepancy between simulation results and observed data. Objective function guides the optimization algorithm to move in the right direction across the model parameter space in the search for solutions. The aim of the work is to compare the influence of the normalization method on history matching results using the example of a synthetic model and a model of a real oil field. The novelty of this study is in the introduction of a behind-the-casing crossflow simulated in one of the production wells of the synthetic model. The crossflow acts as noise in the well observed data. The results of this research allowed to understand which method of the objective function normalization is more effective for history matching and why.
Materials and methods
The main methods in this study are numerical reservoir simulation, system analysis and computational experiment using a synthetic reservoir model and a sector model of a real oil field located in Siberia.
reservoir simulation, oil field, numerical model, objective function, mismatch normalization, automated history matching, optimization algorithms, history matching quality