ORIGINAL PAPER Multi-objective optimal design for flood risk management with resilience objectives Hsin-Ting Su 1 • Sai Hung Cheung 1 • Edmond Yat-Man Lo 1,2 Ó Springer-Verlag GmbH Germany, part of Springer Nature 2018 Abstract In flood risk management, the divergent concept of resilience of a flood defense system cannot be fully defined quanti- tatively by one indicator and multiple indicators need to be considered simultaneously. In this paper, a multi-objective optimization (MOO) design framework is developed to determine the optimal protection level of a levee system based on different resilience indicators that depend on the probabilistic features of the flood damage cost arising under the uncertain nature of rainfalls. An evolutionary-based MOO algorithm is used to find a set of non-dominated solutions, known as Pareto optimal solutions for the optimal protection level. The objective functions, specifically resilience indicators of severity, variability and graduality, that account for the uncertainty of rainfall can be evaluated by stochastic sampling of rainfall amount together with the model simulations of incurred flood damage estimation for the levee system. However, these model simulations which usually require detailed flood inundation simulation are computationally demanding. This hinders the wide application of MOO in flood risk management and is circumvented here via a surrogate flood damage modeling technique that is integrated into the MOO algorithm. The proposed optimal design framework is applied to a levee system in a central basin of flood-prone Jakarta, Indonesia. The results suggest that the proposed framework enables the application of MOO with resilience objectives for flood defense system design under uncertainty and solves the decision making problems efficiently by drastically reducing the required computational time. Keywords Flood risk management  Multi-objective optimization  Non-dominated sorting genetic algorithm (NSGA-II)  Resilience  Monte Carlo  Surrogate model 1 Introduction Following widespread flood risk exposures at a global scale, resilience has become topical in flood risk manage- ment, as following earlier resilience concepts of Holling (1973) for ecological systems. The resilience of a flood defense system can be broadly defined as the capacity to resist, respond to, recover from, and adapt to the flood impact and reach sustainability over time in an uncertain environment (Kaufmann et al. 2016). This is in line with the general framework of prevention (mitigation), pre- paredness, response and recovery (PPRR) as widely adopted for resilience management (Godschalk 2003). For flood risk assessment, quantifiable performance metrics must be defined to evaluate the mitigation ability of flood defense systems while accounting for the uncertain nature of floods. Despite many different indicators proposed, no single indicator exists to quantify the resilience in all aspects directly (de Bruijn 2004; Fekete and Hufschmidt 2014). Therefore, using multiple resilience indicators is recommended. The thrust of this paper is on modelling the resilience of a flood defense system (specifically a levee system) in multiple aspects, which depend on the proba- bilistic features of the flood incurred damage cost under the uncertain nature of rainfalls. A multi-objective optimal design framework which enables explicit and simultaneous consideration of multiple quantifiable resilience indicators is developed to determine the optimal protection level of a & Sai Hung Cheung SHCHEUNG@ntu.edu.sg 1 School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639798, Singapore 2 Deputy Director of Institute of Catastrophe Risk Management, Nanyang Technological University, Singapore 639798, Singapore 123 Stochastic Environmental Research and Risk Assessment https://doi.org/10.1007/s00477-017-1508-7