基于支持向量机的致密油藏水平井体积压裂初期产能预测

水平井产能预测和评价是致密油藏合理有效开发的关键环节.由于体积压裂后,缝网分布复杂且相互干扰,传统经验公式法和数学解析法在低渗-致密油藏的产能预测普遍误差较大.本文基于大庆油田M2区块20口水平井体积压裂的钻遇、压裂和试油数据,首先采用皮尔森系数、斯皮尔曼系数和肯德尔系数进行了主控因素分析,筛选和评价了7个主要参数.在...

Support vector machine-based initial productivity prediction for SRV of horizontal wells in tight oil reservoirs

The productivity prediction and evaluation of horizontal wells is a key element in the reasonable and effective development of tight oil reservoirs.Due to the complex distribution and mutual interference of fracture network after stimulated reservoir volume(SRV),the traditional empirical formula method and mathematical analysis method generally have large errors in the productivity prediction of low-permeability tight oil reservoirs.Based on the drilling encountering,fracturing and oil testing data of 20 horizontal wells with SRV stimulation in Block M2 of Daqing Oilfield,this paper firstly analyzes the main controlling factors using Pearson coefficient,Spearman coefficient and Kendall coefficient,and selects and evaluates 7 main parameters.On this basis,through the standardization of sample data,analysis preprocessing of principal components and the optimization of cross-validation methods,a support vector machine-based initial productivity prediction model is established for SRV of horizontal wells in tight oil reservoirs.The results show that when the number of wells is small in the early stage of development,the prediction effect of SVM leaving P cross-validation is the best,and the average relative error is only 5.4%.As the number of subsequent production wells increases,it is recommended to adopt the 10-fold method with better accuracy and faster calculation speed.According to the data volume in different development stages,the prediction model can be adjusted at any time to provide important references for the optimization of on-site operation parameters and productivity evaluation of SRV horizontal wells in different development stages of the target block.