Foul sewer networks face many challenges related to new pressures and ageing infrastructure. There is a need to be able to evaluate how networks will be able to adapt to varying population densities, urban development and ecologic changes. Some suggest the use of exploratory models to test large numbers of network configurations, intended as alternative responses to the driving pressures. However, in order to carry out exploratory modelling, a trade-off between computational time and accuracy must be achieved. An approach to generate and size foul sewer networks which allow computational savings was developed in a collaboration between Eawag, ETH Zurich and TU Delft.This approach was used in this study to evaluate the computational time savings and accuracy of a generated sewer network in hydraulic performance assessment. Two case studies of 7km2 (Port Phillip) and 57km2 (Melbourne), and different land uses were used to evaluate the ability of the generated network to represent a real network. It was found that computational time was reduced for both case studies, by a maximum factor of 10. Hydraulic performance was compared under high, typical and low flow conditions. It was found that high flow parameters are better represented in small case studies, where network capacity reduction and topology differences are less evident between real and generated networks. For low flow conditions, percentages of network length at risk of sedimentation were well represented for both case studies, with the larger case study showing slightly better performance. Under typical flow conditions, it was shown that topology simplification in the generated network leads to significant changes in times of concentration between networks, and hydrograph discrepancies, especially for the larger case study. However, general trends, including distribution of pipe cumulative flow percentage for pipes in sedimentation, or pipe diameter for pipe surcharge, are well represented by the generated network. Therefore, it is found that the generated network should be used to evaluate hydraulic performance trends in generated networks, as well as global values (eg. flood volume) if the network differences (eg. capacity, path length) are taken into account. Finally, future work needs are highlighted which could strengthen the findings of this study, including use of real network flow and hydraulic performance data. Moreover, the computational time savings for a full assessment (rather than just hydraulic evaluations) should be quantified and compared.