SEMINAR: Rishi Adiga Event as iCalendar

20 September 2017


Venue: Room 201, 70 Symonds Street

Integer Programming Optimization of Geothermal Production

PhD candidate, Department of Engineering Science

Geothermal power generation has a high capital cost, a large portion of which is accounted for by well costs. Hence, it is imperative to maximize value from wells by determining an optimal drilling and production policy. An important technology used when making well sighting decisions is computer simulation, which is commonly used as a tool to aid manual decision making in a time and labour intensive process.

This project uses Mixed Integer Programming (MIP) to create a framework for automating this process. Surrogate models were created to represent candidate well locations and their interactions by Net Present Values, calculated from a small number of simulation runs. These NPVs were then used with binary decision variables in MIP models to select optimal well combinations for maximizing NPV. Optimal solutions for the surrogate models were found to be near optimal with respect to the numerical reservoir model for small numbers of wells selected.

Plant capacity is another variable that factors into the decision making process. The best way to split plant capacity between wells drilled was also investigated by simulation for different scenarios. The current formulation assumes that all wells are drilled at the start and are run to depletion, that there are no limits on production, and only makes drilling decisions. The next steps in this research include incorporating decisions in time and on production limits in the optimization to create comprehensive production policies, testing the framework or different simulation models, and then introducing uncertainty to the model.