南堡35-2油田调驱优化设计及实施效果分析——应用基于物理模拟参数为学习样本的BP神经网络预测方法  

Profile Control Optimization and Application Analysis of Nanbao 35-2 Oilfield——Use the BP Neural Network Prediction Method Based on Core Flooding Parameters Sample

作　　者：刘文超[1] 卢祥国[1] 崔景盛[2] 刘进祥[1] 

机构地区：[1]东北石油大学,提高油气采收率教育部重点实验室,黑龙江大庆163318 [2]天津大港油田滨港石油科技有限公司,天津300000

出　　处：《油田化学》2012年第2期206-211,共6页Oilfield Chemistry

基　　金：国家重大专项课题"砾石充填防砂完井深部调剖(驱)技术研究"(项目编号2008ZX05024-004-007-004)

摘　　要：南堡35-2油田是海上稠油油田,具有原油黏度高、油层岩心渗透率高、胶结疏松和非均质性严重等特点,水驱开发过程中指进现象严重,难以获得较高的水驱采收率。本文利用仪器检测和物理模拟实验优选调驱剂。依据优选结果,利用物理模拟参数为学习样本的BP神经网络预测方法和经济评价方法,对调驱方案的平均浓度和段塞尺寸进行优化。室内实验分析和优化结果表明,对于该油田A21井,最佳段塞尺寸和平均质量浓度分别为0.1 PV和2200 mg/L。聚合物凝胶调驱矿场试验过程中,注入压力明显上升,增油效果显著。在调驱有效期内,在不考虑自然递减情况下受效油井增油27639.59 m3,取得了明显的增油降水效果,经济效益十分明显。Nanbao 35-2 was characteristic of the high viscosity, high permeability, serious heterogeneity, etc. During water flooding process, because of the serious fingering phenomenon, it was difficult to obtain high water fllooding recovery. In the paper, the profile control agents were optimized by the methods of the instrument testing and core flooding tests. And then the average concentration and slug size were optimized by using the BP neural network prediction method based on core flooding parameters sample and economic evaluation. Compared with laboratory analysis data and optimizafion results, for A21 well, the best slug size and average concentration were 0. 1 PV and 2200 rag/L, respectively. During the profile control process, injection pressure increased significantly. In the case without considering the natural decline rate, the oil wells incremental oil production was 27639.59 m3. The measurements achieved an obvious oil increment and water reduction. And economic benefits are very obvious.

关 键 词：海上稠油油田 物理模拟 BP神经网络 聚合物凝胶 增油效果 

分 类 号：TE357.46[石油与天然气工程—油气田开发工程]