剩余油分布的智能预测方法  

Intelligent prediction methods of remaining oil distribution

作　　者：黄开桦 黄旭日[1] 杨剑 HUANG Kaihua;HUANG Xuri;YANG Jian(School of Geosciences and Technology,Southwest Petroleum University,Chengdu Sichuan 610500,China)

机构地区：[1]西南石油大学地球科学与技术学院,四川成都610500

出　　处：《石油化工应用》2023年第12期66-69,共4页Petrochemical Industry Application

摘　　要：剩余油分布预测的常见方法是以井资料、测井资料和储层结构信息为基础,通过油藏数值模拟的方法实现含水率、产油量空间分布与变化规律的预测。对于井网复杂、生产周期长的高含水率老油田,以传统方法进行剩余油分布预测常受限于计算量过于庞大。本文通过简化数值模拟实现步骤,引入神经网络模型来改进剩余油分布预测的传统方法。新方法在某砂岩储层地区进行实际应用,与含水饱和度插值方法相比,含水率分布的连续性更强;与全工区油藏数值模拟方法相比,运算时间大大缩短。The common method of remaining oil distribution prediction is based on well data,well logging data and reservoir structure information,and through reservoir numerical simulation,the spatial distribution and change law of water cut and oil production can be predicted.For the old high water cut oilfield with complex well pattern and long production cycle,the remaining oil distribution prediction by the traditional method is often limited by the large amount of calculation.In this paper,by simplifying the steps of numerical simulation,neural network model is introduced to improve the traditional method of remaining oil distribution prediction.The new method has been applied in a sandstone reservoir area.Compared with the water saturation interpolation method,the water content distribution is more continuous.Compared with the reservoir numerical simulation method in the whole working area,the calculation time is greatly shortened.

关 键 词：剩余油预测 神经网络 数值模拟 含水率 

分 类 号：TE327[石油与天然气工程—油气田开发工程]