应用人工神经网络技术预测地层出砂

PREDICTION TO SAND PRODUCTION IN FORMATION BY ARTIFICAL NEURAL NETWORK TECHNOLOGY

范兴沃 李相方 童敏 胡才志 赵平

Fan Xingwo Li Xiangfang Tong Min Hu Chaizhi Zhao Ping

长庆石油勘探局, 陕西西安 710021 石油大学, 北京昌平 102249 胜利油田海洋开发公司, 山东东营 257062 辽河油田公司, 辽宁盘锦 124010

人工神经网络预测地层出砂是一种新型出砂预测方法。稠油油藏胶结疏松,成岩作用差,用传统出砂预测方法进行出砂预测与现场试验结果误差很大。结合辽河油田杜32区块和胜利油田埕北井区的现场情况,根据现场收集的地层、流体、测井解释、油井生产状况等资料,在分析影响地层出砂因素的基础上,应用BP网络模型对样本进行学习,预测地层出砂。结果表明,应用人工神经网络可将预测结果分为不出砂、轻微出砂、中等出砂、严重出砂4个等级,预测精度有很大提高。根据预测结果有针对性地选择防砂方法,取得了很好的防砂效果。

A new method to predict sand production in oil and gas by the Artificial Neural Network (ANN) is described in the paper. Viscous oil reservoirs are characterized by poor consolidation, poor diagenesis.Especially, the oil is highly viscous in Du32 of Liaohe Oilfield and Cheng Bei of Shengli Oilfield.There exists great deviation resulted from the sand production predicted by traditional methods in the test. Moreover, lack of the whole core samples in the two blocks, the method of core flow experiment to predict sand production is much too inaccurate.In order to settle the problem, the method of sand production predicted by ANN is applied in the two oilfields.According to the field data of the formation, fluid, log analysis and performance in oil and gas wells, based on the analy sis of sand production factors, BP Artificial Neural Network is used to predict sand production by studying samples.The application in this two fields shows the good results can be produced by the method of ANN.

10.3969/j.issn.1000-7393.2002.06.015