分层抽样方法在岩心柱取样中的应用

Application of stratified sampling method in core plug sampling

吕洲 王玉普 李莉 吉伟平 侯秀林 刘达望 葛政廷 李扬

LÜ Zhou WANG Yupu LI Li JI Weiping HOU Xiulin LIU Dawang GE Zhengting LI Yang

中国石油勘探开发研究院, 北京 100083 2. 中国工程院, 北京 100088 3. 中国石油化工集团公司, 北京 100728 4. 中国石油 长庆油田公司, 西安 710018

Research Institute of Petroleum Exploration & Development, PetroChina, Beijing 100083, China 2. Chinese Academy of Engineering, Beijing 100088, China 3. China Petrochemical Corporation, Beijing 100728, China 4. Changqing Oilfield Company, PetroChina, Xi'an, Shaanxi 710018, China

通过储层岩心实验得到的数据是地质研究、测井解释和油气藏工程计算的基础。岩心柱样品的选择是储层岩心实验的第一步，其取样精度对实验结果具有重要的影响。当岩心柱样品取样数量受限的情况下，传统的简单随机抽样方法难以满足测量精度的要求。针对传统抽样方法的局限性，阐述了分层抽样方法的原理，并探讨了分层抽样参数的选取标准。在实测数据的基础上，使用Visual Basic语言编程实现蒙特卡洛算法，对随机抽样和分层抽样的测量精度进行了比较。结果表明：在相同抽样数量的前提下，分层抽样的数据精度明显优于简单随机抽样。因此，根据不同岩心实验的要求，提出了基于分层抽样的岩心柱样品的选择方法。该方法可以预估取样精度，并据此设计合理的样品数量，当实验样品数量受限时，可以有效地提升取样精度。

The data obtained from reservoir core experiments are the basis of geological research, logging interpretation and oil and gas reservoir engineering calculations. The selection of a sample plug is the first step in a reservoir core experiment, and its sampling accuracy has an important effect on experiment results. When the sample number is limited, the conventional random sampling cannot meet the requirements of measurement accuracy. In view of the limitations of the traditional sampling methods, this paper described the principle of a stratified sampling method, and discussed its selection criteria. Based on measured data, the Monte Carlo algorithm was implemented using Visual Basic programming, and the measurement accuracy of random sampling and stratified sampling was compared. The results showed that, the data accuracy of stratified sampling was better than that of simple random sampling under the same sampling quantity. Therefore, according to the requirements of different experiments, a selection method of core plug samples based on stratified sampling was proposed. The method can estimate sampling accuracy and design a reasonable number of samples accordingly. When the number of experiment samples is limited, the sampling accuracy can be improved effectively.

国家科技重大专项（2017ZX05013-006）和中国石油天然气股份有限公司科学研究与技术开发项目（2016B-1306）联合资助。

https://doi.org/10.11781/sysydz201802274