Influence of Model Parameter Uncertainty on the Evaluation of Function State of Drainage Network

The uncertainty of hydraulic model parameters of drainage network comes from the randomness in the spatial and temporal distribution of rainfall, the complexity of underlying surface and the uncertainty in the process of design, construction and use of pipeline network. To explore the influence of parameter uncertainty on functional state evaluation of nodes and pipelines, an uncertainty analysis model was established by considering the uncertain parameters in the process of precipitation, runoff, flow, and hydrodynamic calculation of pipeline network. According to the parameters in the modeling process and the reviews from literatures, fourteen parameters in the modeling process of SWMM model were selected as the uncertainty parameters in this study. The parameter sensitivity analysis was performed by analyzing the influence to the simulation results of the hydraulic model by the increment analysis of model parameters. Five sensitive parameters were selected for the uncertainty analysis of the hydraulic model of the drainage system. Based on the SWMM simulation engine, Monte Carlo simulation method was used to calculate the failure probability of nodes and pipelines, as well as the corresponding estimators and variation coefficients of the occur time and volume of node overflow, and the fullness degree of pipes. The difference between the certainty model and the uncertainty model was analyzed by comparing the hydraulic simulation results of nodes and pipes. Case study results show that the process of rainfall input and model calculation is greatly influenced by the uncertainty of parameters. It is necessary to consider the uncertainty of parameters in the whole process of hydraulic models of the drainage system. Compared to the certainty condition, the uncertainty analysis model is capable of identifying the nodes and pipes at high risk of failure. For the nodes and pipes with high failure probability identified in the uncertainty model, the relative difference between the uncertainty results and the average value of the certainty results of the occur time and volume for node overflow comes up to 29.13% and 10.41%, respectively, and the max relative difference for the fullness degree of pipes is 22.31%.