<p>This study investigates the thermal and hydraulic performance of a geothermal reservoir through integrated numerical simulation, geological uncertainty quantification, and surrogate-based well placement optimization. A three-dimensional single-phase geothermal reservoir model is developed using the MATLAB Reservoir Simulation Toolbox (MRST), incorporating coupled mass and energy conservation equations for fluid flow and heat transport in porous media. Reservoir performance is evaluated using key indicators, including produced-fluid temperature, water production rates, instantaneous heat extraction, and cumulative thermal energy recovery. To assess the impact of subsurface heterogeneity, an ensemble of one hundred permeability realizations is generated using sequential Gaussian simulation, representing realistic geological uncertainty in sandstone reservoirs. The resulting ensemble simulations demonstrate significant variability in thermal breakthrough timing and cumulative heat extraction, highlighting the strong influence of permeability heterogeneity on geothermal system performance. Furthermore, a surrogate-based global optimization algorithm is employed to optimize producer’s well locations with the objective of maximizing cumulative extracted thermal energy over a ten-year production period. The optimization results show that producer–injector spacing plays a dominant role in delaying thermal breakthrough and improving heat recovery. Compared to a random well configuration, the optimized layouts can increase cumulative thermal energy extraction by more than 30% in the studied case, while marginal improvements are achieved over a conventional corner-well configuration. Overall, the results demonstrate that combining numerical simulation, uncertainty analysis, and surrogate-based optimization provides a promising and computationally efficient framework for evaluating geothermal reservoir performance and supporting well placement decisions under geological uncertainty.</p>