Risk-based factorial probabilistic inference for optimization of flood control systems with correlated uncertainties

In this paper, a risk-based factorial probabilistic inference method is proposed to address the stochastic objective function and constraints as well as their interactions in a systematic manner. To tackle random uncertainties, decision makers' risk preferences are taken into account in the decision process. Statistical significance for each of the linear, nonlinear, and interaction effects of risk parameters is uncovered through conducting a multi-factorial analysis. The proposed methodology is applied to a case study of flood control to demonstrate its validity and applicability. A number of decision alternatives are obtained under various combinations of risk levels associated with the objective function and chance constraints, facilitating an in-depth analysis of trade-offs between economic outcomes and associated risks. Dynamic complexities are addressed through a two-stage decision process as well as through capacity expansion planning for flood diversion within a multi-region, multi-flood-level, and multi-option context. Findings from the factorial experiment reveal the multi-level interactions between risk parameters and quantify their contributions to the variability of the total system cost. The proposed method is compared against the fractile criterion optimization model and the chance-constrained programming technique, respectively.