Probabilistic modelling of sewer system overflow emissions

The main advantages of the probabilistic modelling of urban drainage system s are the possibility of quantification and efficient reduction of the risk of 'wrong' management decisions or engineering designs, as well as the les s biased mathematical modelling. Such probabilistic modelling achieves quan tification of the uncertainty in the field of model-output. In the modellin g of sewer system overflow emissions, a lot of uncertainties are involved: uncertainties about the spatial rainfall-input, model simplifications of th e sewer system and runoff catchment, and biases and uncertainties about the model parameters. A methodology has been presented to quantify the different uncertainty-sour ces in a decomposed way by separating uncertainties about the model structu re, the model input and the model parameters. It has been applied to an exi sting sewer system and WWTP, for which the emission modelling has been perf ormed both by a full hydrodynamic and a simplified conceptual model. In thi s application, the importance of having accurate (spatial) rainfall data an d a thoroughly calibrated runoff model has been noticed. A well-considered simplification of a full hydrodynamic model, for instance by a conceptual r eservoir model, has a negligible influence on modelled values of sewer syst em emissions. Such simplification achieves a reduction of calculation times , which is a necessity in probabilistic modelling. (C) 1999 Published by El sevier Science Ltd on behalf of the IAWQ. All rights reserved.