A method for adjusting the peakedness of design storms to prevent bias in hydraulic modelling simulation

Abstract:

In the United Kingdom (UK), decision-makers rely on outputs from hydraulic semi-distributed and fully distributed models to inform funding decisions and other decisions related to planning applications, connection consent and adoption of new drainage networks. The current modelling practice in the UK requires applicants submitting funding, connection consent or planning applicants to apply design (synthetic) rainfall events in certain durations and magnitudes to hydraulic models in order to provide the required modelling outputs. The modelling outputs, such as flood volume, discharge rate and flood count are then used to satisfy various requirements defined by the respective authorities to secure the required funding or consent. With more frequent flooding incidents taking place every year around the UK, the hydraulic modelling outputs are becoming questionable as most drainage networks would have been designed for no flooding up to 1 in 30-year return period. In this paper, we address the practical challenge of sewer system design and planning at a catchment scale by providing an alternative approach to traditional design storm application. As the peak intensity and the peak timing significantly affect the runoff peaks simulated in urban catchments of significant imperviousness, we focus on the peakedness factor, defined as a ratio of maximum intensity to average rainfall intensity as a key factor that contributes to the bias in hydraulic modelling outputs. The Cranbrook and Norwich catchment models have been used as case studies to evaluate the hypothesises by driving the catchment models using historical storms, design storms and modified design storms to test the proposed storm modification method. We also use both catchment models to implement SuDS on a catchment scale and test the betterment achieved. In this paper, we conclude that design storms contain conceptual errors and should not be used to run catchment models, for optioneering assessment or funding decisions. The proposed design storm modification reduces the bias that occur by adopting the practice of using design storms in comparison to continuous data. We suggest that, should use of design storms be required, then a modification process be applied. We also conclude that the role of SuDS is underestimated when using design rainfall.