Production allocation model for thinly interbedded sandstone reservoirs based on grey relational generalized inverse distance weighting method

Owing to pronounced heterogeneity in thinly interbedded sandstone reservoirs, productivity is markedly different among individual sublayers. Conventional production-splitting methods are generally based on static parameters (e.g., sand-body thickness and permeability) or on extensive liquid-production-profile data. Consequently, the influence of multiple factors is not readily captured, and low prediction accuracy is often observed in reservoirs with complex production dynamics. To make production allocation in layered reservoirs more accurate and to reflect the nonlinear effects of several factors, we propose a new model that uses a generalized inverse-distance weighting interpolation scheme. A generalized distance" between wells is defined using dynamic and static parameters, and the integrated differences among static geological parameters (e.g., effective reservoir thickness, porosity, permeability, well spacing, and oil saturation) and dynamic production data (e.g., oil production, water production, bottomhole flowing pressure, and overall water cut) are accounted for within a multidimensional space. As a result, geological–engineering similarity among wells can be characterized with greater accuracy. The power index in the generalized distance controls how strongly differences are strengthened or weakened. The weights show how much each factor contributes. Key factors were first selected by grey relational analysis. Then, a genetic algorithm was used to optimize the power index and weights by minimizing the mismatch with the historical liquid production profile. Consequently, the optimal parameter combination can be automatically identified, and the accuracy and robustness of the production-splitting results are substantially improved. This method is applied to production-splitting practice in the VI oil group of the III layer in the Gasi Reservoir. A splitting accuracy of 90% is achieved, representing a significant improvement over the conventional method, and reliable technical support is provided for subsequent reservoir development and adjustment.