There are similar problems about geometry optimization of boxes, positioning of cutting slices on assigned areas which are both solved by means of Genetic Algorithms (GA).
GA are not efficient in terms of CPU time but at the same time are capable of moving the degrees of freedom in a quite randomized but hierarchical way without the user having to teach/write the constraints and usual models that are required by standard methods.
I expect your model featuring a reduced amount of degrees of freedom (i.e. the number of gas wells in the reservoir). Therefore, it might be viable and worth considering the forced approach of grid search. This is an exhaustive method where you can put your gas wells onto a spatially-discretized grid of the reservoir and distribute the N gas wells on the intersection points of that grid and evaluated the objective function. The best objective function would identify the optimal position of the gas wells. For any configuration of gas well, even before evaluating the objective function, you could check for its consistency and any possible geometrical, i.e. minimum-distance, violation of the assigned constraints.
Hope this may help you.
Knowing the variables, and if measurable, central composite design or box=behnken design of response surface methodology will do.
- Badges
- Science topic
- Similar topics
- Engineering
- Chemical Engineering