关键词: 辫状河, 夹层, 成因类型, 井震结合, 夹层预测

Abstract:  Prediction of interlayers in braided river reservoir is a key component in reservoir description. Interlayer predictions have been largely done on the interlayer developmental pattern and channel bar architecture using one single prediction method. Here, based on literature survey, and focusing on the complex interlayers in the braided river reservoir of P filed, South Sudan, we propose an interlayer prediction method on the basis of multiinformation association following genetic analysis of different sand and shale morphologies developed under different depositing conditions. First, in areas where well densities were large, the skeleton profiles and triangular cross sections were established; and scales of different interlayer were confirmed by use of well log and seismic data with high vertical resolution and strong horizontal continuity. Next, based on the lithofacies model, stable barrier model was built by using the deterministic modeling method, constrained by the barrier thickness map. Then, channel and channel bars were simulated in the sand facies by using stochastic method, constrained by sedimentary microfacies. The interlayer model was also established by using stochastic method, constrained by genetic analysis of various types of interlayers. It was established that there were four types of interlayers in the study area, and a fine geological model reflecting their spatial distributions was built on the basis of multiinformation association. Our study found that the scales of interlayers with thickness greater than 2 m can be verified by well and seismic integration. Thus, the scales of interlayers can be quantitatively determined for different layer types and directions, which lays the foundation for predictive model building. Lithofacies probability model based on Kriging interpolation increases lithofacies model accuracy to 94%. The deterministic modeling method, constrained by stable barrier thickness map, can build accurate barrier model for sand layers and sublayers. This objectbased modeling method combines deterministic and stochastic approaches to set parameters such as form, quality, scale and trendfor different interlayers to build a fine geological model of braided river reservoir. Our method may also be applicable for predicting interlayers in other related reservoirs.

Key words: braided river, interlayer, genetic type, well and seismic integration, interlayers prediction