**Urban wastewateR system Layout model**

## Research Discription

The urban wastewater system is one of the most important infrastructures for developing a sustainable city. Traditional methodology for water system design are various, such as enumeration, Monte Carlo simulation, heuristic search and genetic algorithm .But most of them are single-objective : least-cost. The Urban wateR system Layout model(URL) is a multi-objective optimization model with objects of cost, pollutant and source-recycle. It consists of Monte Carlo sampling, non-dominated sort genetic algorithm(NSGA-II) and graph algorithms. Firstly, URL will confirm the number and the location of waste water treatment plants (WWTPs). Then an optimal solution sets will be found by NSGA-II algorithm. After a sufficient number of Monte Carlo samplings, a set of Pareto optimal solutions (POSet) was obtained.

## Research Contents

NSGA-II is a faster and efficiency version of NSGA, which has been successfully applied to many Multi-Objective Problems (MOPs) in various study area. Examples of study with NSGA-II includes flood mitigation, pollution management, and imaging observation. The efficiency of NSGA-II has also draw attention. Improving efficiency of this research would be focus on algorithm optimization and parallelization

We implanted a two layer parallel URL model. Figure below shows the framework.

This model was applied to Fengtai, south of Beijing which has 40 spatial units and 5 spatial WWTPs. With optimization and parallelization, it would be applied to Hefei which has 8000 spatial units and 11 potential WWTPs. And it would be parallel with more than 600 cores.

Figure below shows one potential layer of Hefei presented by this model. The left figure presents the service area distribution and different colors indicates different WWTPs' service area. The right figure presents the utilization of reclaimed water while the purple area means use relaimed water and the green area means not.

The whole calculations take 104 times monte-carlo samplings with population size : 500 and iteration steps : 500. Finally, We get a 182 Pareto optimal layout set (Figure below). Each point in the figure represent a potential layout includes the number and the location of WWTPs, the service area distribution, reclaimed water ulitization and treatment technologies selections. And each layouts in the Pareto optimal set are dominated by none of the others which means these layouts can not improve any object function without setback another one. Desicion-maker of the master plan would select layouts with object values in different level accroding to the actual demand of economic, environmental and resouce recycle performance.