Integrating Predictive Modeling with Control System Design for Managed Aquifer Recharge and Recovery Applications

Thursday, 18 December 2014
Zachary William Drumheller1,2, Julia Regnery2,3, Jonghyun Harry Lee2,4, Tissa H Illangasekare2,3, Peter K Kitanidis2,4 and Kathleen M Smits2,3, (1)Colorado School of Mines, Mechanical Engineering, Golden, CO, United States, (2)Engineering Research Center for Re-Inventing the Nation's Urban Water Infrastructure (ReNUWIt), National Science Foundation, Stanford, CA, United States, (3)Colorado School of Mines, Civil and Environmental Engineering, Golden, CO, United States, (4)Stanford, Civil and Environmental Engineering, Stanford, CA, United States
Aquifers around the world show troubling signs of irreversible depletion and seawater intrusion as climate change, population growth, and urbanization led to reduced natural recharge rates and overuse. Scientists and engineers have begun to re-investigate the technology of managed aquifer recharge and recovery (MAR) as a means to increase the reliability of the diminishing and increasingly variable groundwater supply. MAR systems offer the possibility of naturally increasing groundwater storage while improving the quality of impaired water used for recharge. Unfortunately, MAR systems remain wrought with operational challenges related to the quality and quantity of recharged and recovered water stemming from a lack of data-driven, real-time control.

Our project seeks to ease the operational challenges of MAR facilities through the implementation of active sensor networks, adaptively calibrated flow and transport models, and simulation-based meta-heuristic control optimization methods. The developed system works by continually collecting hydraulic and water quality data from a sensor network embedded within the aquifer. The data is fed into an inversion algorithm, which calibrates the parameters and initial conditions of a predictive flow and transport model. The calibrated model is passed to a meta-heuristic control optimization algorithm (e.g. genetic algorithm) to execute the simulations and determine the best course of action, i.e., the optimal pumping policy for current aquifer conditions. The optimal pumping policy is manually or autonomously applied. During operation, sensor data are used to assess the accuracy of the optimal prediction and augment the pumping strategy as needed. At laboratory-scale, a small (18”H x 46”L) and an intermediate (6’H x 16’L) two-dimensional synthetic aquifer were constructed and outfitted with sensor networks. Data collection and model inversion components were developed and sensor data were validated by analytical measurements.