Multi-criteria decision analysis for water supply infrastructure planning under uncertainty
Our centralized water supply systems are aging. Despite their success in reliably providing high quality drinking water, nowadays especially small utilities (e.g. less than 10’000 inhabitants supplied) are illprepared to face possible future challenges. The fragmented structure of the water supply sector leads to a lack of institutional, financial and personnel resources for professional management and planning of water supply systems. Current planning is furthermore challenged by insufficient knowledge and data about the prevailing water infrastructure condition and future rehabilitation demand. It usually ignores future dynamics and planning uncertainties, as well as alternatives to the perpetuation of the status quo. Infrastructure decision making is usually not transparent and only few stakeholders are included into the decision process.
This thesis presents an approach to overcome these shortcomings and support long-term water supply infrastructure planning under uncertainty in a multi-stakeholder context. Thereto, methods from multicriteria decision analysis (MCDA), strategic asset management (SAM), pipe failure modeling, and scenario planning were combined, adapted, and further developed. The suitability of the approach was validated in a case study in Switzerland. To improve the prediction of pipe service life in view of data scarcity, it was shown how the knowledge of experts can be quantitatively assessed and integrated into the calibration of pipe survival models by means of Bayesian inference. Similarly, knowledge gained from three mid-size to large Swiss water networks was used to improve calibration of a novel pipe failure model. It is demonstrated that this failure model is able to deal with the common data situation, and mitigate overestimation of the time to failure caused by the absence of data from already replaced pipes. The failure model was combined with a rehabilitation model to assess the performance of 18 rehabilitation strategies under four future scenarios for a small water utility. MCDA was used to compare these alternatives under different preferences concerning three objectives. The analysis revealed that the common strategy, purely reactive rehabilitation, is not recommendable in most cases and that annual replacement of 1–2 % of the network by condition might be a good strategy for the utility in question.
These findings were considered during the definition of alternatives for a second MCDA study aiming at identifying ‘good water supply infrastructure’ alternatives. Eleven alternatives and an objectives hierarchy consisting of 44 fundamental objectives and 30 attributes were developed together with stakeholders. The alternatives differ not only with regard to rehabilitation management, but also technical, managerial, and organizational aspects. The outcomes of all alternatives regarding these attributes were predicted under four future scenarios to account for uncertainties about the future development. The approach for the elicitation and modeling of preferences includes the imprecision of the stated preferences as well as uncertainties of preference parameters which were not elicited (the aggregation model, marginal value functions, risk attitude, and scaling factors). Preferences of ten selected stakeholders were then elicited and probability distributions of the ranking of alternatives based on these preferences were obtained. Despite differences in the individual rankings, a potential compromise solution could be proposed and ways for potential adaptation and improvement of other alternatives be indicated. In general, alternatives with good outcomes regarding groundwater protection, water quality, supply reliability, and realization of the rehabilitation demand received the highest ranks, as these were also among the most important objectives for the majority of the stakeholders. As operation and management do considerably contribute to the performance regarding the latter three objectives, the importance of a thorough infrastructure and rehabilitation management cannot be neglected. In view of the possible ranges of the outcomes, the objectives of ‘high social acceptance’ (e.g. disturbance by unnecessary road works, resources autonomy), and to some extent also ‘low costs’ were judged less important.
Multi-criteria decision analysis proved useful to support the long-term planning of water infrastructures under uncertainty and in a multi-stakeholder framework. The usual extrapolation of the status quo was overcome. Future dynamics and uncertainties could be incorporated by combining decision making and modeling with scenario planning, besides the quantitative consideration of uncertainties in making predictions, and evaluating the results. With the presented approaches for the modeling of pipe failures and rehabilitation, the methods and tools for the assessment of the current condition and future rehabilitation demand of small water networks despite a difficult data situation are now available.