基于常规测井资料的储层流体识别  

Reservoir Logging Fluid Identification Based on Conventional Logging Data

作　　者：许鉴源 赵军龙 白倩 黄千玲 XU Jian-yuan;ZHAO Jun-long;BAI Qian;HUANG Qian-ling(Xi’an Shiyou University,Xi’an 710065,China;Shaanxi Key Laboratory of Petroleum Accumulation Geology,Xi’an 710065,China)

机构地区：[1]西安石油大学地球科学与工程学院,陕西西安710065 [2]陕西省油气成藏地质学重点实验室,陕西西安710065

出　　处：《河北地质大学学报》2023年第6期28-34,共7页Journal of Hebei Geo University

基　　金：陕西省自然科学基础研究计划(2019JM-359)。

摘　　要：由于M油田长3储层物性、流体性质等因素导致测井响应特征信息复杂,利用单一测井信息难以准确识别储层流体性质。为了提高识别研究区长3储层流体性质的准确性,基于大量文献梳理流体识别分析方法,结合现有实际测井资料,开展多方法流体识别应用比较,常用判别方法难以快速准确识别流体性质。因此本文从可以进行多信息结合分析且人为干扰少的随机森林原理出发,将其应用于研究区储层流体性质识别,经检验该方法取得了最佳识别效果,准确率达90.47%,为智能算法解决地质问题提供了思路。Due to factors such as physical properties and fluid properties of the long 3 reservoir in the M oilfield resulting in complex logging response characteristics information,it is difficult to accurately identify reservoir fluid properties using single logging information.In order to improve the accuracy of identifying the fluid properties of the long 3 reservoir in the study area,the general methods of fluid identification analysis were sorted out based on a large amount of literature,and a comparison of the application of multiple methods of fluid identification was carried out in combination with the existing actual logging information,commonly used discriminatory methods to achieve the purpose of rapid and accurate identification of fluid properties that are difficult to achieve.Therefore,this paper applies the random forest principle,which can combine multiple information analysis and less human interference,to the fluid property identification of reservoirs in the study area,and the method has been tested to achieve the best identification results with an accuracy rate of 90.47%,which provides an idea for intelligent algorithms to solve geological problems.

关 键 词：测井解释 流体识别 随机森林 

分 类 号：TE122[石油与天然气工程—油气勘探]