First International Conference on Saltwater Intrusion and Coastal Aquifers Monitoring, Modeling, and Management. Essaouira, Morocco, April 2325, 2001 Chance Constrained Optimal Management in Saltwater- Intruded Coastal Aquifers Using Genetic Algorithms M.K. Benhachmi 1 , D. Ouazar 1 , A. Naji 2 , A.H-D. Cheng 3 , K. EL Harrouni 4 1 Ecole Mohammadia d’IngØnieurs, Morocco 2 FacultØ des Sciences et Techniques de Tanger, Morocco 3 University of Delaware,USA 4 Ecole Nationale dArchitecture, LabHAUT, Morocco Abstract A chance-constrained management model with an economic objective of coastal aquifers threatened by saltwater intrusion is developed. The model explicitly considers that the toe location of saltwater intrusion is a function of random variables such as physical parameters and boundary conditions. This model incorporates the quantity, quality and economic aspects of groundwater for sustainable use of the coastal aquifers and extends the existing deterministic model to stochastic one. The management objective allows for a plausible scenario for economic planned withdrawal and salinity control in coastal aquifers. The steady-state model simulates groundwater flow on a horizontal plane using sharp-interface analytical approach. This model is incorporated into a simple genetic algorithm (SGA) and chance constrained optimization. The chance constraint is divided into an expected value component and a stochastic one. Both the expected value and stochastic component were evaluated using the perturbation technique based on the second order Taylor series approximation. When the input physical parameters and boundary conditions are given as mean and standard deviation, the solutions, either the toe location of the saltwater wedge or the entire interface location, are predicted as a mean and standard deviation. The chance constrained programming method transforms the probabilistic model to a deterministic one and therefore facilitate the solution and reduces memory requirements and computational time. The management model is illustrated through a hypothetical unconfined coastal aquifer system. The results show that incorporating uncertainty into coastal aquifer optimization model using chance constraint programming coupled to SGA could be a practical method for making decisions on optimal pumping rates and scenario exploitation schemes. 1 INTRODUCTION: In coastal areas, several wells have been drilled to satisfy the increasing by water demand, which has caused unacceptable drawdowns and deterioration of water pumped quality. A set of well-established withdrawal and management policies is necessary to achieve more efficient management and operation of these aquifers. Initial efforts to support and improve the development and operation of groundwater systems by simulation and optimization techniques were made in the early 1970s. Since then, various types of groundwater management models have been proposed and successfully applied to realworld aquifer systems. Many reviews on the types of groundwater management models and their applications are made by [Gorelick, 1983], and [Yeh, 1986]. The management models applications in saltwater intrusion, is relatively recent, [Cummings and McFarland, 1974], [Shamir, et al., 1984], [Willis and Finney, 1988], [Barlow, Wagner and Belitz, 1996], [Emch and Yeh, 1998], [El Harrouni et al., 1998],[Das and Datta, 1999 ], [Cheng et al , 1999], [Loaiciga and Leipnik, 2000]. Most of these problems have been investigated in a more complex setting involving various management objectives. Concerning saltwater intrusion into wells, it is often addressed in an indirect manner such as constraining drawdown at control points, or minimizing the intruded saltwater volume. These studies were