【作者】 宋杰鲲；

【导师】 张在旭；

【作者基本信息】 中国石油大学 ， 石油工程管理， 2007， 博士

【摘要】 油藏经营管理是20世纪90年代新兴起来的一种新的油藏管理模式,目前已成为国内外石油公司高效开发油气田的基本模式,也成为当前油气田开发领域的一个研究热点。油藏经营管理是一项系统工程,系统决策是其核心,不确定性是系统决策的一个重要特征,以不确定优化理论为基础进行油藏经营管理系统决策研究具有重要的理论和现实指导意义。本文运用系统观点,对油藏经营管理进行分析,提出油藏经营管理系统的概念与特征,通过国内外油藏经营管理系统运行机制对比,建立了上下层为经营承包关系的系统运行机制,明确系统决策的主要内容为产量成本分配、措施规划和单井措施优选,并给出系统决策的不确定性、多目标性和群体性等特征;对随机规划、模糊规划、区间规划等不确定优化理论以及遗传算法、人工神经网络等智能优化算法进行系统总结,论证了各种区间数序关系的一致性和各种区间线性规划转化形式之间的包含与被包含关系,给出了完整、统一的参数线性规划转化形式,为油藏经营管理系统不确定性决策模型和方法的建立提供理论基础;以产量成本分配基本程序为主线,分别应用回归预测、灰色预测和神经网络预测对产量指标进行预测,应用回归预测和灰色预测对单位操作成本指标进行预测,给出产量、成本指标的区间描述形式,建立产量成本分配区间规划模型,包括产量最大化模型、成本最小化模型、利润最大化模型和净现值最大化模型,并给出各种模型的参数规划转化形式、模型参数含义及具体求解步骤;以措施规划的基本步骤为基础,应用随机性统计和模糊性统计对稳增产措施效果进行描述,建立稳增产措施的随机规划模型和模糊规划模型,包括期望值模型、机会约束规划模型、机会约束目标规划模型和相关机会规划模型,并给出各种模型的融不确定模拟、神经网络、遗传算法于一体的混合智能算法;以单井措施优选基本步骤为基础,按照指标体系构建原则,构建了包括技术指标和经济指标共八项指标的单井措施优选评价指标体系,提出了群组层次分析法赋权、区间数多属性决策方法排序(包括基于可能度的区间数排序方法、投影法和TOPSIS法)和Borda集结方法终排序的单井措施优选方法体系;通过具体实例表明,应用不确定优化理论进行油藏经营管理系统决策是切实可行、科学有效的。更多还原

【Abstract】 Advanced reservoir management is a newly developed reservoir management pattern in 1990’s. At present it has become the basic pattern of effectively developing oil & gas fields in domestic and foreign petroleum company, and a hot branch of oil & gas field development domain. Reservoir management is system engineering, and system decision is core. Uncertainty is an important characteristic of system decision, so applying uncertain optimization theory to research system decision has important theory and reality-guiding sense. This paper analyzes reservoir management from the system viewpoint, proposes the concept and characteristic of reservoir management system, establishes system operation mechanism, specifies the content of system decision which includes the allocation of production and cost, measure programming, and measure selecting of single well, and gives the characteristic of system decision such as uncertainty, multi-objective, and group electing. It summarizes the uncertain programming theory including stochastic programming, fuzzy programming and interval programming and intelligent optimization algorithms including genetic algorithm and artificial neural network, especially proves the uniformity of all kinds of interval number sequential relation and the containing & contained relation of all kinds of interval linear programming transformation, and presents the complete and unified parameter linear programming transformation form. Based on the allocation procedure of output and cost, it uses the production using regression, gray theory and neural network to predict production and uses regression and gray theory to predict cost, gives their interval number description form, constructs the interval programming models of allocating production and cost which include output maximization, cost minimization, profit maximization and net present value (NPV) maximization, at the same time gives their parameter programming transformation form, parameter meaning of model and the concrete solution steps. Based on the procedure of measure programming, it applies the random statistics and fuzzy statistic to descript the effect of measure, constructs the stochastic programming models and fuzzy programming models, which include the expect value model, chance-constrained programming, chance-constrained goal programming, and dependent-chance programming, and presents the hybrid intelligent algorithm including uncertain simulation, neural network, and genetic algorithm. Based on the measure selecting of single well, it establishes the evaluation index system according to the construction principle, which includes eight indexes that reflect the technical effect and economical effect of measure, put forwards a selection method, which entrusts weights through group AHP, uses interval number multiple attribute decision making to sort the measures, and applies Borda method to rank the measures finally. Examples testify the feasibility and scientific of applying uncertain optimization theory to reservoir management system decision.更多还原