Using genetic algorithms for reservoir characterisation

Reservoir characterisation is the process of describing a hydrocarbon reservoir, in terms of the parameters of a numerical model, so that its performance can be predicted. We describe the use of a specially designed genetic algorithm to search for the reservoir description that is most likely to match the measurements made on the reservoir. The genetic algorithm uses six separate chromosomes for different types of reservoir parameters. Three of the chromosomes have multi-dimensional real number structures, while the other three chromosomes are one-dimensional binary bit arrays. Specially designed crossover and mutation operators have been created to work with the non-standard genome structure. The method has been tested on a realistic, complex synthetic reservoir model, and compared with a simulated annealing (SA) algorithm. We have shown that our genetic algorithm produces better results than the simulated annealing algorithm and results which are comparable to what might be achieved by hand. Also, we have shown that the performance of the genetic algorithm is robust to the details of how it was set up. Given the ease with which the method can be cheaply parallelised, its robustness to lost or corrupted solutions, and that it returns a suite of good solutions, it is an ideal method to implement as an automatic reservoir characterisation algorithm.