模拟退火组合优化算法在油气储层随机建模中的应用

石油天然气的勘探与开发是一个涉及多学科的领域。各学科提供的资料观察方法不同,观察尺度不同(即承载大小不同),分辨率不同,可靠性不同。资料和知识的表达方式也不相同,既有数字化的资料,也有定性的描述。将各种资料综合在同一个储层定量化数值模型之中,是目前储层建模的趋势。文中讨论了在储层随机精细建模中使用模拟退火组合优化算法的方法和步骤,包括目标函数的构造、选择初始状态、抽样和扰动机制、退火策略参数的选择以及怎样综合多种类型的资料等问题。指出在储层建模中使用模拟退火优化算法有两方面的优点,即易于综合多种类型的资料;可以客观地再现储层变量的空间相关结构。其主要缺点是计算量大,有必要进行并行处理。

Applying Simulated Annealing Combinatorial Optimization Algorithm to ReservoirStochastic Modeling

Exploration and exploitation of hydrocarbons involve multiple disciplines. Each of these disciplines supplies data in different way, in different scale ( support ), in different resolution and in different reliability. To integrate mul-tiple disciplines data in a numerical reservoir model is a tendency of reservoir modeling at present day. This paper discusses the methods of and problems encountered in applying simulated annealing combinatorial optimization al-gorithm to reservoir stochastic conditional simulation, which include the methods of constructing and updating ob-ject function, choosing initial state, perturbing mechanism, annealing schedule and integrating data from differ-ent sources. Over other reservoir stochastic modeling techniques, two principal advantages of simulated annealing method are that it can integrate different kinds of data more feasibly and can keep reservoir spatial structure more accurately. Its main disadvantage is the vast amount of CPU time needed.