The groundwater system is often polluted by different sources of contamination where the sources are difficult to detect. The presence of contamination in groundwater poses significant challenges to its delineation and quantification. The remediation of a contaminated site requires an optimal decision making system to identify the pollutant source characteristics accurately and efficiently. The source characteristics are generally identified using contaminant concentration measurements from arbitrary or planned monitoring locations. To effectively characterize the sources of pollution, the monitoring locations should be selected appropriately. An efficient monitoring network will result in satisfactory characterization of contaminant sources. On the other hand, an appropriate design of monitoring network requires reliable source characteristics. A coupled iterative sequential source identification and dynamic monitoring network design, improves substantially the accuracy of source identification model. This paper reviews different source identification and monitoring network design methods in groundwater contaminant sites. Further, the models for sequential integration of these two models are presented. The effective integration of source identification and dedicated monitoring network design models, distributed sources, parameter uncertainty, and pollutant geo-chemistry are some of the issues which need to be addressed in efficient, accurate and widely applicable methodologies for identification of unknown pollutant sources in contaminated aquifers.

Effective environmental management and remediation strategies are required to remediate contaminated water resources. Accurate characterizing of unknown contaminant sources is vital for selection of appropriate environmental management plan and reduction of long term remedial costs. In order to characterize the sources of contamination, the aquifer boundary conditions and hydrogeologic parameter values need to be estimated or specified. In real life contaminated aquifers, often there are sparse and inaccurate information available. On the other hand, extensive collection of data is very costly. The uncertain and highly variable natures of water resources systems affect the accuracy of contaminant source identification models. In this study, an optimal source identification model incorporating Adaptive Simulated Annealing optimization algorithm linked with the numerical flow and transport simulation models, is designed to identify contaminant source characteristics. The fuzzy logic concept is used to identify the effect of hydrogeological parameter uncertainty on groundwater flow and transport simulation. The fuzzy membership values incorporate the reliability of specified parameter values in to the optimization model. An illustrative study area is used to show the potential applicability of the proposed methodology. The incorporation of fuzzy logic in source identification model increases the applicability of contaminant source detection models in real-life contaminated water resources systems.

A fuzzy genetic algorithm (FGA) is used to obtain the least-cost design of looped water distribution networks (WDNs). The FGA is a GA-based search method in which fitness evaluation is carried out using a fuzzy decision system (FDS). Unlike previous fuzzy water network optimisation techniques in which different properties were introduced by fuzzy membership functions and the decisions were made by the means of crisp logic, a fully FDS is used to simulate the tolerance in constraint deviation in network elements. The method is applied to two wellknown networks currently appearing in the literature as case studies. For each case, based on network layout and optimisation constraints, a FDS is introduced by determining fuzzy parameters, membership functions and the rule set. The results prove the capability of the FGA method of finding optimum solutions in addition to providing a dynamic algorithm that gives engineers more practical insight into easily applying system uncertainty in the optimisation algorithm. NotationC cost objective function CON constant number D vector defining diameter of pipes DP number of new pipes difHead head deviation constraints difVelocity velocity deviation constraints E vector defining rehabilitation process H head in nodes H min , H max minimum and maximum allowable head HP vector defining pumps' power L K length of the pipe Kth segment NJ number of network nodes NP vector defining number of parallel pumps working in each pump station PEN fuzzy representative of constraints PM number of pump stations RMSE constraint deviation measurement value RP number of existing pipes T vector defining tank capacity TK number of new tanks V velocity in pipes

Abstract-Groundwater contamination is one of the serious environmental problems. Effective remediation strategies require accurate characteristics of contamination sources. Contamination source identification approaches need accurate flow and contaminant transport simulation models. In order to obtain reliable solutions, the simulation models need to be provided with reliable hydrogeologic information. In real life scenarios usually sparse and limited hydrogeologic information is available. In this study two hydraulic conductivity sampling networks are ranked based on their effectiveness in identifying reliable contamination source characteristics. Using multiple realizations of hydraulic conductivity fields, and the location and size of the contaminant plume at different monitoring stages, an index of reliability is estimated for each hydraulic conductivity sampling network. It is demonstrated that the source characteristics identified by utilizing the sampling network with higher index of reliability results in more accurate characterization of contamination sources. Therefore the developed methodology provides a tool to select an appropriate hydrogeologic sampling network for more efficient characterizing of contamination sources.

No abstract