西南石油大学学报(自然科学版)

• 本期导读 • 上一篇    下一篇

新井试井曲线类型预测新方法研究

刘文东1 *,刘性全2   

  1. 1. 中国石油大庆油田有限责任公司第二采油厂第二作业区,黑龙江大庆163000
    2. 中国石油大庆油田有限责任公司第八采油厂,黑龙江大庆163000
  • 出版日期:2015-06-01 发布日期:2015-06-01
  • 通讯作者: 1. 中国石油大庆油田有限责任公司第二采油厂第二作业区,黑龙江大庆163000 2. 中国石油大庆油田有限责任公司第八采油厂,黑龙江大庆163000
  • 基金资助:

    国家自然科学优秀青年科学基金(51304164)。

A Study on the Prediction Method of Well-testing Curve Type of New Wells

Liu Wendong1*, Liu Xingquan2   

  1. 1. The Second Section of No.2 Oil Production Company of Daqing Oil-field Company Ltd.,Daqing,Heilongjiang 163000,China
    2. No.8 Oil Production Company of Daqing Oil-field Company Ltd.,Daqing,Heilongjiang 163000,China
  • Online:2015-06-01 Published:2015-06-01
  • Contact: 1. The Second Section of No.2 Oil Production Company of Daqing Oil-field Company Ltd.,Daqing,Heilongjiang 163000,China 2. No.8 Oil Production Company of Daqing Oil-field Company Ltd.,Daqing,Heilongjiang 163000,China

摘要:

在部署测试监测井点、制定合理的测试时间时,准确预测待测井的曲线类型具有重要的意义。从多组判别分
析基本原理出发,结合油田实际监测井点的试井曲线类型,以及各井的测井和生产动态资料,优选出与试井曲线类型
相关的井和储层参数,建立曲线类型预判模型。通过多次反复筛选参数并结合总错判率,最终建立了以储层厚度、含
水率、测井渗透率和产量为主要因素的八厂油田试井曲线类型预测模型,模型回代错判率为9.09%,现场应用符合率
为85.7%,可用于预判待测井曲线类型。研究成果为现场工作人员优化监测井点、制定合理测试时间、提升监测成果
质量等提供了科学的指导。

关键词: 试井, 曲线类型, 多组判别, 预测模型

Abstract:

When deploying monitoring well and setting appropriate time for well test,accurate prediction of well logging curve
types undoubtedly has important significance. Based on the principle of multiple discriminant analysis we combine the reality
well test curve types and the well logging and production performance data of oilfield monitored wells,we optimize the well and
reservoir parameters through repeated screening parameters. In the end,we established a model considering reservoir thickness,
moisture content,logging permeability and the oil production. The result of the error rates is 9.09% and the coincidence rate
is 85.7%,which means the model can be used in anticipation for logging curve type. This method provides a scientific and
reasonable guidance for workers to formulate reasonable testing time,optimize the monitoring well and improve the quality of
monitoring results.

Key words: well test, curve types, multi-group discriminant, prediction model