A scenario-based MCDA framework for wastewater infrastructure planning under uncertainty

Wastewater infrastructure management is increasingly important because of urbanization, environmental pollutants, aging infrastructures, and climate change. We propose a scenario-based multi-criteria decision analysis (MCDA) framework to compare different infrastructure alternatives in terms of their sustainability. These range from the current centralized system to semi- and fully decentralized options. Various sources of uncertainty are considered, including external socio-economic uncertainty captured by future scenarios, uncertainty in predicting outcomes of alternatives, and incomplete preferences of stakeholders. Stochastic Multi-criteria Acceptability Analysis (SMAA) with Monte Carlo simulation is performed, and rank acceptability indices help identify robust alternatives. We propose step-wise local sensitivity analysis, which is useful for practitioners to effectively elicit preferences and identify major sources of uncertainty. The approach is demonstrated in a Swiss case study where ten stakeholders are involved throughout. Their preferences are quantitatively elicited by combining an online questionnaire with face-to-face interviews. The trade-off questions reveal a high concern about environmental and an unexpectedly low importance of economic criteria. This results in a surprisingly good ranking of high-tech decentralized wastewater alternatives using urine source separation for most stakeholders in all scenarios. Combining scenario planning and MCDA proves useful, as the performance of wastewater infrastructure systems is indeed sensitive to socio-economic boundary conditions and the other sources of uncertainty. The proposed sensitivity analysis suggests that a simplified elicitation procedure is sufficient in many cases. Elicitation of more information such as detailed marginal value functions should only follow if the sensitivity analysis finds this necessary. Moreover, the uncertainty of rankings can be considerably reduced by better predictions of the outcomes of alternatives. Although the results are case based, the proposed decision framework is generalizable to other decision contexts. (C) 2016 Elsevier Ltd. All rights reserved.