一种新型混合优化算法及其在优化油田开发中的应用

【作者】 李晓明；

【导师】 张烈辉；

【作者基本信息】 西南石油学院 ， 油气田开发工程， 2004， 博士

【摘要】 开发油气田时油藏工程师的主要任务之一是确定油气田的布井方案以及评估不同方案投入实施时的赢利情况，并从中选出最佳方案。合理的布井方案要求在保证油气田经济效益的前提下，采用井数最少的井网，并最大限度地控制地下储量。在经济和自然条件允许的情况下，应尽可能在油气藏有限寿命期内采出更多的油、气，以获取最大经济效益，因此在布井方面要确定合理的生产井数以及布井方式、位置。这是涉及油气田基本建设的最中心问题，也是涉及油气田今后生产效果的根本问题。 对油、气、水井位置的优化非常复杂，依赖于油藏和流体特征，驱动方式，井和地面设备的规格，以及各类经济参数，因此这是一个多决策变量问题，所提出的每一种方案都跟一整套决策变量有关。显然，对这些变量值的选择与组合就是一个最优化问题。常规优化技术由于不能很好地处理这种多变量优化问题及目标函数可能不连续、不可微的情况而不被采用，而通常所采用的规则井网布井方式又缺乏灵活性、风险性较大，常常使油田不能获得最大经济效益。 针对这一问题，本文提出了一种新型优化技术，该技术是基于标准遗传算法、基于适应值共享机制的小生境运算，以及单纯形局部搜索法的一种混合优化算法。该算法最大的优势在于对函数性态无要求，能够处理的变量数目多，收敛速度快、精度高，尤其适用于大规模搜索的整体优化问题。本文将该方法应用于优化油田开发当中，以使油田获得最大累积产量为目标对油井位置进行了优化：同时以使油田获得最高经济效益(净现值)为目标优化了油井数量及其对应位置，取得了一定成果，并为今后油藏工程师在布井方案的优化以及如何提高经济效益的问题上提供了一种新的思路及方法。更多还原

【Abstract】 One of the main tasks of a reservoir engineer is to develop a well-arrangement scheme and evaluate the profitability of different schemes when put into execution, and the best one will be selected for adoption. A rational scheme requires making use of the least well number to control the reserves by the greatest degree. The most hydrocarbon should be produced to obtain maximal profit within economic and natural limits. Therefore, we must design a rational well number, placement and mode. This is the most central problem involved in the oilfield infrastructure and an essential problem related to the future production effect.Optimal placement of oil, gas or water wells is a complex problem that depends on reservoir and fluid properties, type of drive, well and surface equipment specifications, as well as various economic parameters. Thus it is a multiple decision variables problem and each proposed scheme is relative to a wide set of decision variables. It is obvious that the choice of their values is an optimization problem. Since they can not deal well with this kind of problem of multiple variables, discontinuous and non-differentiable functions, traditional optimization technologies are usually not adopted. Furthermore, regular well pattern lacks flexibility and takes a risk in losing money in business.A new-style hybrid optimization algorithm that is based on standard genetic algorithm(GA), niching technology based on fitness-sharing model, and the simplex method is presented in this paper. The best advantages of this algorithm are that it has no function property requirements and can deal with large numbers of variables with a fast convergence speed and high precision. Thus it is especially fit for global-optimization problem on a large scale. This technology is applied in optimizing reservoir development in this paper. Firstly it is used for well placement optimization with the object of maximizing the cumulative production and then for well number optimization with the object of maximizing the net present value(NPV). This work provides reservoir engineers with a new idea and a new type of method in how to optimize well-arrangement scheme and increase economic profits.更多还原