LID Optimization for Flood Mitigation in Urban Drainage Systems Considering Phasing of Construction, Climate Change Scenarios and Using Machine Learning

Abstract Submission: This abstract presents a novel methodology for phasing the implementation of Low Impact Development (LID) systems in urban drainage systems (UDSs). The methodology consists of the combination of various elements including physically based event simulation of UDSs, event hyetograph generation at various return periods considering climate change, and machine learning for the identification of suitable locations and extensions of LIDs in the various UDS sub-catchments and in the various phases of construction. The machine learning algorithm used is the biobjective genetic optimization with the total cost of the interventions and the flood volume to be simultaneously minimized, to yield a Pareto front of optimal solutions in the trade-off between the two objective functions. The applications to the calibrated UDS serving the industrial area of Brescia (Northern Italy) prove the capability of the methodology to offer cost-effective solutions to significantly attenuate flooding by reducing the rainwater volumes entering the underground network, with no need for structural interventions underground. The methodology also proves flexible to accommodate variations in intense rain events because of climate change.

Learning Objectives/Expected Outcome (Optional) :