A data-driven integrated framework for sustainable evaluation of grey-green urban drainage strategies under rainfall...

Urban areas face increasing pressure from population growth and climate variability, which intensify flood risk and strain water-related infrastructure. While grey and green drainage strategies are widely promoted to enhance urban resilience, their comparative hydraulic performance and economic feasibility remain challenging to evaluate due to the high computational demands of hydrodynamic modelling. This study proposes a novel data-driven integrated framework that combines hydrodynamic simulation, machine learning-based surrogate modelling, and cost-benefit analysis to support environmental management decisions. The methodology is applied to an exploratory case study in Quito, where alternative grey and green infrastructure scenarios are assessed under multiple rainfall return periods. Machine learning models were developed to emulate key hydraulic responses in the urban drainage system, including flooded volumes at manholes and peak flows in conduits. The surrogate models accurately reproduced hydrodynamic outputs, with coefficients of determination frequently exceeding 0.97 for higher return periods, while substantially reducing computational requirements. Beyond hydraulic performance, the integrated economic evaluation indicates that benefits exceed investment costs in all scenarios. Benefit-cost ratios are greater than one, with green strategies outperforming grey alternatives, although both options are economically viable over a 30-year lifespan. By coupling process-based modelling, data-driven emulation, and economic analysis within a unified framework, this study advances scalable and computationally efficient decision-support tools for sustainable urban flood management under climate variability.