Quantitative Prediction of Braided Sandbodies Based on Probability Fusion and Multi-Point Geostatistics

Predicting the spatial distribution of braided fluvial facies reservoirs is paramount significance for oil and gas exploration development. Given that seismic materials enjoy an advantage in dense sampling, many methods have been proposed to predict reservoir based on different attributes. Nevertheless, attributes sensitivities reservoirs, informational redundancy between them makes it difficult combine effectively. Regarding modeling, multi-point geostatistics represents characteristics Despite this, very build high-quality training images. Hence, this paper proposes a three-step method predicting probability fusion geostatistics. Firstly, similar statistical data modern sedimentation field paleo-outcrops were processed under guidance pattern construct images suitable target stratum research area. Secondly, each linear combination selected was demarcated calculate principal component value work out elementary conditional probability. Lastly, PR integration approach employed all probabilities joint Then combined with model through We illustrated detailed workflow our new by applying modeling case Bohai Bay Basin, East China. The reduced error prediction results 32% 46% respectively, water content 36.5% 60.3%. This potentially effective technique characterize other fields same geological background.