Abstract
Monitoring networks are designed based on a developed methodology linked contaminant transport simulations and optimization models. Advection and dispersion simulation generates plume realizations while the use of Latin Hypercube Sampling, one of the Monte Carlo techniques, accounts for uncertainty in transport parameters and in contaminant source characteristics. Optimization model using Genetic Algorithm provides a given number of wells in order to maximize the detection probability. Application of the method to a two-dimensional hypothetical domain with steady state flow illustrates the utilization of the developed method and provides monitoring networks with sufficiently high accuracy under the condition with or without preexisting wells as well as various flow magnitudes. Results reveal that designed well locations adequately cover the vacant area where plumes pass through, leading to increasing the detection probability. Results also show that wells are located relatively close to the source and the detectable limit and probability distribution affect on monitoring network design. In addition, the effect of the number of generated plume realizations on the network design is investigated.
