Decision-making under uncertainty for LT-ATES systems in complex subsurface settings: application to a low-transmissivity aquifer

Abstract

Low-temperature aquifer thermal energy storage systems (LT-ATES) store excess thermal energy in the subsurface and recover it when needed to heat and cool buildings sustainably. Complex subsurface settings are increasingly targeted for storage. Existing design guidelines cannot be readily applied because they were developed for homogeneous sandy aquifers. In complex settings, subsurface uncertainty translates into uncertainty about system feasibility or efficiency. This paper introduces a method to establish a decision tree for ATES based on uncertainty analysis. It combines Monte Carlo simulations of groundwater flow and heat transport with sensitivity analysis, joint probability distribution estimation and kernel density estimation. For illustration the method is applied to a low-transmissivity aquifer where hydraulic and thermal feasibility are the main prediction targets. Sensitivity analyses based on clustering reveal the impact of model parameters and engineering actions on both prediction targets. In the specific case study, we find for example that storage conditions with transmissivity below 20 m²/d lead to inefficient systems. Desirable storage conditions have transmissivity above 40 m²/d, while intermediate values require thorough risk analysis. Thermal breakthrough risk is higher when longitudinal dispersivity is above 3 m. Further, our approach results in some minimum requirements in terms of subsurface properties that have to be reached for which an investment is justified. On top, the decision tree proposes target engineering actions to decrease investment risk while optimizing return. As such, the proposed method effectively guides early investment decisions for LT-ATES in complex subsurface settings without requiring field data.