Groundwater contamination is a major problem in many parts of the world. Remediation of contaminated groundwater is a tedious, time consuming and expensive process. Pump and treat (PAT) is one of the commonly used techniques for groundwater remediation. Simulation-optimization (S/O) models are very useful in appropriate design of an effective PAT remediation system. Simulation models can be employed to predict the spatial and temporal variation of contaminant plumes. Optimization models, on the other hand, can be used to minimize the cost of pumping or recharge. Generally, grid or mesh based models using Finite Difference Methods (FDM) or Finite Element Methods (FEM) are used for groundwater flow and transport simulation. Recently, Meshfree (MFree) based numerical models have been developed due to the difficulty of meshing and remeshing in these methods. The MFree Point Collocation Method (PCM) is a simple MFree method to simulate coupled groundwater flow and contaminant transport. It saves time for pre-processing such as meshing or remeshing. Evolutionary algorithm based techniques such as for particle swarm optimization (PSO) and genetic algorithms (GA) have been found to be very effective for groundwater optimization problems. In this paper, a simulation model using MFree PCM for unconfined groundwater flow and transport and a PSO based optimization model are developed. These models are coupled to get an effective S/O model for the groundwater remediation design using PAT. The S/O model is applied to the remediation design of an unconfined field aquifer polluted by Total Dissolved Solids (TDS) by using pump and treat and flushing. The model provides an effective remediation design of pumping rate for the selected wells and costs of remediation.