Honghe oilfield HH156 area of Chang8 reservoir is a typical tight reservoir.As shown by the test da-ta,most industrial oil operation is related to fracture,but the distribution of fracture has significant heterogeneity. Thus,it increases the difficulty of petroleum exploration.Through core description and imaging logging of 15 wells, the fracture parameters are obtained.Well data and fracture factors are correlated,and the high correlated fracture factor is chosen using fuzzy logic technology.It can be seen that coherence cube is highly related to the fracture da-ta,while the distance of fault and tectonic curvature takes the second place.The first three factors are chosen to participate in nonlinear neural network training,and then the fracture density model is got.Furthermore,the DFN model and fracture property model are obtained using stochastic simulation method.It can not only predict the dis-tribution of fracture in this area but also provide the fracture model to numerical modeling and lay a basis for devel -opment program.%红河油田HH156井区长8油藏为典型的致密低渗透油藏,试油资料显示大多工业油流与裂缝发育相关,但裂缝的分布具有很强非均质性,为油气开发带来困难.通过研究区15口井岩心描述及成像测井资料获得了裂缝分布参数,应用模糊逻辑技术对井眼裂缝指示数据和各种裂缝要素进行相关性判别,筛选相关性高的裂缝要素.由筛选结果可以看出,相干属性和裂缝指示相关性最好,断层距离属性、构造曲率次之,本次选取相关性最好的前3种属性参与非线性神经网络的学习与培训,得出最终的裂缝密度分布模型,进而应用随机建模方法生成DFN模型及裂缝属性参数,不仅有效预测了该区裂缝分布,而且提供用于数值模拟的裂缝模型,为油藏方案实施提供依据.
展开▼
