利用电导率实时预测地层水滤液污染度  

Filtrate Contamination Predictions in Formation Water Using Conductivity in Real Time

作　　者：冯永仁[1] 刘海波[1] 左有祥 支宏旭 孔笋[1] 周明高[1] FENG Yongren;LIU Haibo;ZUO Youxiang;ZHI Hongxu;KONG Sun;ZHOU Minggao(Oilfield Technology Branch,China Oilfield Services Limited,Longfang,Hebei 065201,China)

机构地区：[1]中海油田服务股份有限公司油田技术事业部,河北廓坊065201

出　　处：《测井技术》2022年第5期511-516,共6页Well Logging Technology

基　　金：“十三五”国家科技重大专项“大型油气田及煤层气开发超低渗地层测试技术与装备”(2017ZX05019-004)。

摘　　要：传统的水基钻井液滤液污染度的估算方法是在不同的泵抽时间点,井下取水样,将其带回钻井平台,进行现场水分析;假设地层水中不含某种离子,而水基钻井液滤液中含有大量这种离子,就可以通过计算确定滤液污染度。但这种方法不能实时监测滤液污染度,增加了作业成本和风险,需要开发新的滤液污染度估算方法。泵抽取水样时测量的电导率为实时预测地层水滤液污染度提供了有效途径。为此,在实验室测量了钻井液滤液、纯水、纯水+不同离子浓度盐的电导率值。发现电导率随离子浓度呈线性变化关系,得到电导率的线性混合规则。基于此线性混合规则,只要知道纯滤液和纯地层水的2个端点值,就可以根据实时测量的电导率值,预测任何时间的滤液污染度。纯滤液端点值由开始泵抽的值确定,纯地层水的端点值则利用泵抽过程的数据训练机器学习模型,然后外推到时间、体积无限大或预定阈值的方法确定。将该方法嵌入到地层测试取样的实时解释软件中,并用于海上实际取样,得出的结果与水分析结果相比,一致性很好。新方法可为实时取样决策提供有效工具,节省作业成本,降低作业风险。The traditional method for estimating water-based mud filtrate contamination is to take water samples at different pumping times and bring them back to the drilling platform for on-site water analysis.It is assumed that there is no certain ion in the formation water and a large amount of this ion in the filtrate of water-based drilling mud,thus we should determining the mud filtrate contamination through calculation.However,this method cannot monitor filtrate contamination in real time,thereby increasing the operation cost and risk.Therefore,it is necessary to develop a new method for estimating filtrate contamination.The conductivity measured while pumping provides an effective way to predict mud filtrate contamination in real time.Hence,we measured the conductivity values of water-based mud(WBM)filtrate,pure water,pure water+salt ions with different concentrations.It is found that conductivity changes linearly with ionic concentrations,thus obtaining the linear mixing rule of conductivity.As long as the endpoint values of WBM filtrate and pure formation water are known,contamination can be predicted in terms of conductivity measured in real time.The end point of WBM filtrate can be determined by an initial value at the beginning of pumping.The endpoint value of pure formation water can be determined by training a machine learning model based on real-time sampling data,and then extrapolated to an infinite time,volume or a predetermined large value.This method is embedded in the real-time interpretation software of formation test and applied to field cases offshore.The obtained results are compared with those of on-site water chemistry analysis,and the consistency is very good.The new method provides an effective tool for real-time sampling,thus saving operation cost and reducing operation risk.

关 键 词：地层测试 钻井液滤液污染 机器学习 电导率 泵抽时间 地层取样 

分 类 号：P631.84[天文地球—地质矿产勘探]