RSS - Julien Harou - Identifying future-proof infrastructure investment portfolios that meet multiple societal goals
Julien J. Harou, Professor of Water Management, Chair of Water Engineering, The University of Manchester
At present infrastructure investment planning in England’s water companies seeks the least cost schedule of scheme implementation that bridges a conservative future estimate of the supply-demand gap. We describe how this method has been developed and used over the last 15 years and provide example applications including a regulator-funded investment model covering the whole of South-East England. We identify the main benefits and limitations of the approach; limitations include that many different portfolios are nearly least cost; the framework (constrained economic optimisation) cannot easily incorporate goals such as robustness, resilience, ecological performance, etc.; the framework only provides one solution rather than helping decision-makers understand their options and the trade-offs they imply, etc.
Based on these limitations we identify several recent ‘decision-making under uncertainty’ frameworks which show promise for planning future resource systems. The talk focuses on one such framework which uses multi-criteria multi-scenario search methods. Its objective is to find combinations of resource system investments that are robust to future uncertainties and meet a range of different societal goals. Results are presented as scatter plots showing Pareto-optimal trade-off curves (or surfaces) made of points, each representing a non-dominated bundle of new system assets (and potentially the policies that regulate their joint use). Results presented as trade-offs help decision-makers identify which balance of performance they seek for their system and the investment choices they imply. We apply the approach both to London’s water supply system, and to the investment in new African dams. Future research directions are discussed in the conclusion.
More information about the speaker can be found here.