Multiple-Point Facies Simulation Conditioned to Probability Models: An Example from a Fluvial Reservoir Located in Bohai Bay Basin, China

摘要

Correct conditioning of geological models to well data is important as it has a major impact on facies simulation. In this paper, we performed a simulation of lithofacies using a multiple-point facies (MPF) method conditioned to probability models, which applied successfully to a subsurface reservoir modeling in the Bohai Bay Basin, China. Probability models, from four lithofacies proportion maps transformed from Average Negative Amplitude cube to lateral trends and Vertical proportion curves to vertical constrains, are simulated from probability logs of lithofacies generated from drilling data by using the compositional kriging method. These probability models even give an exact percentage for every lithofacies probability and then improve the accuracy of the model grid description. Later, different parameter settings lead to the generation of a set of three-dimensional models. Several tests show that the model generated by the MPF method with strong conditioning to probability models can more accurately describe the distribution of sand bodies and interbedded intervals, and provide the best match for both the lithofacies proportion of well logs and the actual production performance. The error in estimating the amount of geological reserves is less than 5%. More than 90% of the wells show an excellent match with the water, with an error of less than 5%. The approach proposed in this paper provides a systematic method for reservoir evaluation, especially in offshore fields with widely spaced wells such as the Bohai Bay Basin.