Evolutionary Testing of Hydraulic Simulator Functionality

A method for automatic functional testing of hydraulic simulators is proposed. The method is based on using genetic algorithms to search for network parameter values at which the simulator under test computes solutions that do not satisfy the governing network equations. The search is made by maximizing the residual of the governing equations. The application of the method to the latest version of the EPANET hydraulic simulator demonstrates its efficiency in detecting incorrect results. The results of quantitative assessment of the functional adequacy of the EPANET solver by random testing are presented. The paper provides examples of hydraulic networks and of parameter value combinations for which incorrect results occur. An example of the use of automatic functional testing together with automatic convergence testing in a comprehensive study of the flow control valve model of the EPANET solver is given.