Developing a Flexible Simulation-Optimization Framework to Facilitate Sustainable Urban Drainage Systems Designs Through Software Reuse

Abstract

In stormwater management, increasing attention is paid to sustainable urban drainage systems (SuDS), which include porous pavements and rain gardens. Various numerical modeling software systems have been developed to simulate the hydrological performances of SuDS, e.g., GIFMOD and SWMM can be respectively applied to study the hydrological processes in small-scale SuDS and their effectiveness in large-scale drainage networks. However, it is sometimes desirable to combine the features of different software, such that the hydrological processes of various spatial-temporal scales can be simulated more accurately. Reusing the existing code to create new software, however, can be challenging, as various combinations of the software of very different structures might be interested. Therefore, this study develops a method and a toolbox in R programming language to couple and reuse existing software without modifying their source code. This toolbox automates the software application processes, including input file creation, simulation, and post-processing. A modeling scheme is adopted that the outflow time series from SuDS are first modeled using the software of choice, which are then treated as external inflows to the drainage networks modeling software. This toolbox also integrates evolutionary optimization algorithms and performance-assessment frameworks to form a flexible simulation-optimization framework for solving design optimization problems. For demonstration, SWMM, a data-driven model of SuDS, and a relative performance evaluation framework are coupled to solve a SuDS allocation problem in an urban catchment. This research demonstrates that file-based software coupling methods can be useful for reusing existing modeling software in developing integrated modeling systems.