A Stochastic Dynamic Programming Model for Planning ASR Operations Under Uncertainty

Aquifer Storage and Recovery (ASR) systems provide environmentally safer storage locations for surplus water that help augment water resources during periods of high demand and low supply.  ASR operations are advantageous over other artificial recharge methods in that they can be implemented in both unconfined and confined formations.  However to be successful in meeting demands, it is critical that ASR facilities be operated in an optimal manner.  This issue is particularly relevant as water supplies available for storage and subsequent recovery tend to exhibit considerable inter-annual variability due to climatic fluctuations and other stress factors.  The primary focus of this work is to develop a decision support system (DSS) to evaluate optimal ASR operations under supply-side uncertainty.  The DSS is based on a stochastic dynamic optimization modeling approach based on the Markov Decision Process with known probability distribution for the supplies.  The DSS provides an estimate for the long-run average monthly operation policies.  The implementation of the DSS and its utility will be presented using an illustrative case-study.