Automated modelling and design of urban stormwater control measures : doctoral dissertation
Avtomatizirano modeliranje in načrtovanje ukrepov za obvladovanje padavinskih voda v urbanih območjih : doktorska disertacija

english

doktorska dela

2022

automated modelling   civil engineering   doctoral thesis   domain knowledge   equation discovery   processbased modelling   rainfall-runoff model   stormwater control measures   urban drainage

Univerza v Ljubljani, Fakulteta za gradbeništvo in geodezijo
(Obvezni izvod spletne publikacije)

Stormwater control measures (SCMs) are technical elements designed to prevent and mitigate the negative effects of uncontrolled stormwater flow. To investigate the impact of SCMs at different spatial scales, SCM models need to be integrated into or coupled with catchments rainfall-runoff (RR) models. Moreover, approaches that include auto-calibration of the RR model and objective-driven automated SCM design can be of great assistance. This research focused on applying automated modelling by using the Process-Based Modelling Tool (ProBMoT) to (a) find the most suitable RR model combining the choices among multiple alternatives for describing hydrological processes, (b) to calibrate RR model parameters against measured data, and (c) to design SCMs based on a target catchment outflow. To this end, a new domain knowledge library that includes RR and SCMs models was encoded. Furthermore, a multi-criteria decision analysis was performed to rank SCMs according to additional criteria. The mark is derived from SCM design parameters and can range between 0 and 5.1, where 0 is the best score. Given the conceptual model of the case study area and the knowledge library, ProBMoT first explored 9 alternative structures of RR models. On average, the models achieved a value of the Nash-Sutcliffe coefficient of 0.82, confirming the efficiency of the proposed approach for calibration of model parameters. Given the target catchment outflow, six SCM scenarios (i.e., bio-retention cell, rain garden, green roof, infiltration trench, detention pond, and storage tank) were designed by using ProBMoT to reduce the peak and total volume of the catchment outflow. Finaly, the developed approach for multicriteria decision analysis was applied to evaluate the designed SCMs. The results showed that landscapeintegrated SCMs (i.e., detention pond, infiltration trench, rain garden, bio-retention cell) are better scored (1.44 - 2.24) than building-integrated SCMs (i.e., green roof, rainwater-harvesting storage tank) (2.57 – 2.73) and grey infrastructure (i.e., underground storage tank) (3.48). The results confirmed the effectiveness of applying automated modelling to the field of RR modelling and optimal SCM design. Furthermore, they underline the importance of multi-criteria SCM evaluation.

12.05.2023