基于最优控制与SPSA算法的水驱油藏优化方法研究

发布时间：2019-01-06 08:26

【摘要】：水驱是石油二次开采的一种重要方式,在驱替过程中,由于油藏地质的非均质性,使注入水优先进入高渗透层,高渗路径一旦见水,这些路径以外区域内的石油就被高渗路径短路而无法被驱替采出。在水驱中如何进行高效合理的开发油田,提高油田控水稳油效果,减缓油田产量递减,成为了急需解决的问题。而现有的水驱油藏优化控制算法均存在优化效果不明显等问题。本文正是在这种背景下,制定最佳的注入策略,进行水驱油藏优化控制算法的研究。在现实的油藏开发中是具有一定的指导和应用意义的。在研究油藏动态系统时,本文以经济净现值为目标函数,选择饱和度为状态变量,将产油量和注水量作为控制变量,并选择连续性方程、边界限定等作为约束条件,从而建立水驱油藏优化控制的数学模型。根据油藏的特性,选择两类优化算法来研究水驱油藏优化问题。一类是有梯度算法即最优控制算法,关于最优控制算法的选择,本文选择最速下降法,在最速下降法中,梯度的求解是一大难点。本文在伴随法求解梯度的基础上,针对伴随法计算速度慢且复杂的缺点对伴随法做了改进,以提高计算的效率和减少计算的复杂度,由此形成了改进最速下降法。另一类是无梯度类算法即SPSA (Simultaneous Perturbation Stochastic Approximation)算法。该算法主要通过估计目标函数的近似梯度来逐渐逼近最优解,这样就避免了对于复变函数的梯度进行复杂的求解。根据两种算法的研究过程,分别设计水驱油藏优化仿真程序,用Eclipse建立油藏模型,结合MATLAB进行编程仿真。本文采用的优化方法在实际的油藏优化中还是能取得较好的效果,且成本低、耗费人力少。
[Abstract]:Water drive is an important way of secondary exploitation of petroleum. Due to the heterogeneity of reservoir geology, the injected water enters into the high permeability zone first, and the high permeability path once the water appears, Oil in areas outside these paths is short-circuited by high permeability paths and cannot be driven out. How to develop oil fields efficiently and reasonably in water drive, improve the effect of controlling water and stabilizing oil production, and slow down the decline of oil production has become an urgent problem to be solved. However, the existing optimal control algorithms for water drive reservoirs have some problems, such as poor optimization effect and so on. It is under this background that the optimal injection strategy is established and the optimal control algorithm for water drive reservoir is studied in this paper. It has certain guiding and application significance in the realistic reservoir development. When studying the reservoir dynamic system, this paper takes the economic net present value as the objective function, selects the saturation as the state variable, takes the oil production and the water injection rate as the control variable, and selects the continuity equation, the boundary limitation and so on as the constraint conditions. Thus the mathematical model of optimal control of water drive reservoir is established. According to the reservoir characteristics, two kinds of optimization algorithms are selected to study the water drive reservoir optimization problem. One is the optimal control algorithm with gradient algorithm. For the selection of optimal control algorithm, this paper chooses the steepest descent method, in which the solution of gradient is a big difficulty. In this paper, the adjoint method is improved to improve the computational efficiency and reduce the computational complexity, aiming at the slow and complex computing speed of the adjoint method, and the improved steepest descent method is formed. The other is the non-gradient class algorithm, the SPSA (Simultaneous Perturbation Stochastic Approximation) algorithm. The algorithm approximates the optimal solution gradually by estimating the approximate gradient of the objective function, thus avoiding the complex solution of the gradient of the complex function. According to the research process of the two algorithms, the water drive reservoir optimization simulation program is designed, the reservoir model is established by Eclipse, and the simulation is carried out with MATLAB. The optimization method adopted in this paper can still achieve good results in the actual reservoir optimization, and the cost is low and the manpower is less.
【学位授予单位】：西南石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TE357.6

【参考文献】