油井压裂效果预测方法研究

Research on methods for predicting oil well fracturing results

压裂措施效果和影响因素之间关系复杂,常规多元回归法又很难确定两者之间的定量关系,而利用人工神经网络可以解决此问题。在对已压裂井增油措施效果评价的基础上,建立了不同措施类型、不同工艺类型的样本库。样本库中考虑的主要因素为:全井射开有效厚度、压裂层地层系数、压前产液量、压前含水率、压裂层数、总加砂量。利用人工神经网络方法建立起压裂效果与这些影响因素的定量关系,建立压裂效果预测模型。矿场应用结果表明,该方法预测结果可靠性较高。

Based on the result evaluation of measures used to enhance oil production of fractured wells, a sample database with different types of measures and processes is established. Main factors taken into account in the database include the effective perforation thicknesses of whole wells, the formation coefficients of fractured layers, the pre - fracturing liquid production, the pre - fracturing water cut, the number of fracturing layers, and the total sand volume. Quantitative relationship is established between fracturing results and those factors mentioned above by employing the artificial neural network method. A model used to predict fracturing results is therefore established. Field application shows that the prediction results on the basis of the method are more reliable.