Controlling the Genetic Algorithm Parameters by Binding It to Simulated Annealing (Case study: Petroleum Engineering)

One of the most common optimization algorithms is genetic algorithm which is used is different problems. There are some internal parameters for the genetic algorithm that changing them alters the application of the algorithm. To find the best optimizer’s parameters, it is usual to change one parameter and set other ones to a constant value, and again change another parameter and set others to a fixed value. This method needs the different runs of the optimizer (with different optimizer parameters) and it is clear that is very time consuming. Here for this purpose the genetic algorithm is coupled with simulated annealing. Thus, genetic algorithm optimizes the problem and simultaneously simulated annealing optimizes the parameters of the genetic algorithm. Afterward its application tested in a petroleum engineering problem. In some oil wells, gas is injected at the bottom of oil wells to bring the oil to the surface. This operation is called gas lift. Usually in gas lift operation there is a limited amount of gas that should be allocated between some wells in a way that the total produced oil be maximized. Here the genetic algorithm coupled with simulated annealing was used to find the best gas allocation which maximizes the oil production. Results show that this new mean is much faster than changing variable method (as previously mentioned) in addition to it, the quality of its optimum point is much better than other methods (changing variable method).

Controlling the Genetic Algorithm Parameters by Binding It to Simulated Annealing (Case study: Petroleum Engineering) Keywords:

changing variable method , parameter control , parameter tuning , petroleum engineering