Multi-Objective Management of Saltwater Intrusion in Groundwater. Optimization under Uncertainty

Coastal aquifers are very vulnerable to seawater intrusion through, for example, the overdraft of groundwater exploitation or insufficient recharge from upstream. Problems of salt-intrusion into groundwater have become a considerable concern in many countries with coastal areas. There have been a number of studies that have tried to simulate groundwater flow system in regions under threat of saltwater intrusion into coastal aquifers. These aquifer systems are characterized by either a single layer (unconfined) or multiple layers with varying hydraulic properties. These are necessary parameter inputs of such simulation models. In order to control saltwater intrusion, planning and management models have been reported in the literature under varying forms of saltwater intrusion management models. These have been based on combinations of the simulation models and optimization programs. They address optimal groundwater pumping and recharge schedules with or without surface water supply for conjunctive use. Saltwater intrusion management problems are necessary to be multi-objective. In this work, the multi-objective management schemes that are based on the minimization of the operational costs and the saltwater intrusion length as their objectives are proposed for the first time. In the literature, these saltwater intrusion management models are mostly based on deterministic aquifer parameters, e.g. the transmissivity is assumed to be precisely known. Moreover, in reality the aquifer parameters cannot be described as constants because they vary spatially. Therefore, the output of deterministic simulation models gives only a first impression of the saltwater intrusion problem. In this thesis also the uncertainty of the input parameters is taken into account in the computations. Hence, the stochastic optimization approach will be introduced to the multi-objective management of saltwater intrusion in groundwater. Moreover, the relationships of the saltwater intrusion lengths (tip and toe) with respect to stresses have a non-linear nature. Thus, their response matrices, which are the coefficients of the objective functions and constraints, are not fixed. This introduces even more complexity to the linear programming in the sense of determining the exact coefficients for the optimization problems. In this thesis, multi-objective management models are developed for a single-layered confined aquifer system. The mean value of the transmissivities is used in the deterministic management problem and realizations of random values of the hydraulic conductivity within a given range are used for the stochastic management problem. In both problems, the Second-Order Cone Optimization programming (SOCO) is applied for solving the single and multiobjective problems.