An Improved and Efficient Surrogate Simulator for Groundwater S/O Models

Continual developments and advances in Simulation/Optimization (S/O) models have led to the development of cost-effective and improved groundwater management implementations. S/O models eliminate the excessive time consumed by trial-anderror use of simulation models to reach a “good” strategy. S/O models of groundwater studies have historically used one of two Response Matrix-based surrogate simulators (here termed ICGO1 and ICGO2). This research will enrich these S/O models with an innovative and efficient methodology (ICDUAH). The main difference among the different surrogate simulators is in the handling of the stress periods of stimulation and observation. ICDUAH combines the capabilities of ICGO1 and ICGO2 and enhances the composite. ICDUAH can employ stress periods of different durations while requiring fewer unitstimulus simulations than ICGO2. Also, ICDUAH uses a time varying unit stimulus for each stimulus location to increase efficiency. ICDUAH groups consecutive stress periods of equal durations into a management era. Moreover, it allows users, if they wish, to set up their own management eras irrespective of the relative duration of stress periods. All stress periods of a management era must be of equal duration. Each steady-state stress period must have its own management era. A unit stimulus will be applied at the first stress period of each management era. This research compares ICGO1, ICGO2 and ICDUAH performance for an optimization problem that maximizes groundwater extraction from a hypothetical study area subject to constraints on state variables. The simulation model of the hypothetical study area uses 12 transient stress periods. The S/O model optimizes for two formulations that differ only in the stress period setup. Posed bounds resemble those for protecting senior surface water or groundwater rights, from proposed management actions. Cycling used in conjunction with the Response Matrix approach allows linearly solving non-linear systems with acceptable accuracy and stability. The result is a better and more efficient surrogate simulator for solving non-linear groundwater management problems. The proposed tool is an improvement over the traditional response matrix algorithm (ICGO1) because it is capable of handling a broader spectrum of situations. The created surrogate simulator is more efficient than the ICGO2 approach because it does not require a unit-stimulus simulation at each stress period