Modelling SuDS at the Street-Scale.

Sustainable drainage systems (SuDS) are increasingly implemented to manage urban runoff, yet their hydrological representation across spatial scales remains challenging. Key knowledge gaps arise in three areas: the representation of unsaturated flow within engineered media, the reliability of simplified device-scale models, and the validity of aggregation strategies for heterogeneous SuDS systems at the street-scale.
The aim of this thesis was to examine how model detail, process representation, and spatial aggregation interact across scales to shape the accuracy, and applicability of SuDS simulations. To achieve this, a structured investigation was undertaken. First, the representation of unsaturated percolation within engineered media was evaluated and refined through analysis of hydraulic conductivity formulations. Second, commonly used device-scale simplifications were systematically assessed against physically explicit SWMM-LID representations to quantify their impact on runoff predictions. Third, aggregation strategies for heterogeneous SuDS deployments at the street-scale
were examined using Monte Carlo simulation to evaluate the trade-offs between model efficiency and hydrological accuracy. Finally, the proposed modelling framework was demonstrated through application to a real-world urban catchment.
At the device-scale, unsaturated drainage was shown to exert strong control over detention behaviour, and a revised hydraulic conductivity formulation (NewHCF) was proposed. This provided more stable and structurally consistent predictions than conventional exponential or power-based approaches. Also, at the device-scale, simplified representations frequently overestimated SuDS performance and were unreliable when internal drainage dynamics governed system response. At the street-scale, typology-based aggregation (LumpN) preserved overall runoff dynamics (NSE generally > 0.9 and exceeding 0.95 in application), whereas full aggregation (LumpOne) produced severe errors in heterogeneous systems, including runoff underestimation of up to 80–90% in
extreme cases.
Overall, the study demonstrates that reliable SuDS modelling requires a balance between model detail, predictive accuracy and computational complexity, with modelling strategies needing to adapt as objectives and levels of aggregation change.