A High Performance Framework for the Optimization of Geothermal Systems, Comparing Energy Production and Economic Output

Abstract

Geothermal heat production for direct use often faces economic challenges that might hinder its more widespread deployment. Design and operation choices in geothermal field development require the evaluation of a multitude of factors with different levels of uncertainty. Energy production and economic output of geothermal systems can therefore be very sensitive to well production rates and reservoir properties such as permeability, while regulatory constraints could further increase complexity. In this work, a high performance framework for optimization of geothermal systems is developed using the Delft Advanced Research Terra Simulator (DARTS) and the Sequential Least Squares Programming (SLSQP) optimization algorithm. An optimization is performed in a system with two doublets by changing the rates of each doublet separately every year, over a thirty year period. Two different objective functions are optimized for: energy generation and NPV. With the maximum injection constraint active both objective functions arrive to almost identical solutions, while removing the constraint improved both the energy generation and NPV only for the NPV objective function. The framework achieved an average convergence time below 5 hours for the full optimization cycle and an average simulation runtime just below 10 seconds. This analysis suggest that the NPV is a better suited optimization function than the energy generation as it encompasses more system aspects. The high performance of the optimization framework will prove increasingly important as more uncertainties are considered in the optimization process.