Parsons Environmental Science Seminar
Friday, November 16, 2018 at 12:00pm to 1:00pm
Parsons Laboratory, 316 Building 48
Learning and flexibility for water supply infrastructure planning under uncertainty
Sarah Fletcher, Postdoctoral Associate , Civil and Environmental Engineering
Water supply infrastructure planning must account for uncertainty in future supply and demand in order to ensure long-term reliability. Recent planning methods have used scenario-based approaches to develop robust planning strategies that perform well across many possible future climates. Robust approaches, however, face high risk of overbuilding expensive infrastructure if the worst outcomes are not realized. In this seminar, I assess the potential to learn about uncertainty over time and enable flexible, adaptive infrastructure planning approaches to maintain reliability at reduced cost. To do this, I develop a novel planning framework that uses Bayesian statistical modeling to update climate change uncertainty. These dynamic uncertainty estimates are used to characterize climate change uncertainty in a non-stationary stochastic dynamic program. I apply this framework to a reservoir sizing problem in Mombasa, Kenya. Results show that high potential to reduce uncertainty over time enables large value in a flexible dam design in which the dam can be raised in the future to increase storage capacity. The value of flexibility depends on the social discount rate, technology choice, and the value society places on water reliability. This approach shows promise in targeting large-scale investments for climate change adaptation where necessary while identifying opportunities to rely on smaller, incremental approaches.