Mechanical and Civil Engineering Seminar

Short Bio:
Sarah is a PhD Candidate at MIT in the Institute for Data, Systems, and Society where her work focuses on resilient infrastructure systems. Her current research on water supply infrastructure evaluates the potential for flexible infrastructure design and planning combined with information collection to mitigate risk from a variety of uncertainties including future demand growth, groundwater availability, and climate change. Before starting her PhD, Sarah worked as a researcher and consultant on issues at the intersection of water, energy, and environment at IHS CERA. She was formerly the Director of Product Development at the startup Sourcewater, where she led the development of the first online marketplace for water and wastewater for energy companies. Sarah has an S.M. in Technology and Policy from MIT and a B.A. in Physics and Economics from the University of Pennsylvania.

Abstract:
Planners in water and other civil infrastructure domains make decisions that will impact the sustainability, reliability, and cost of critical services for decades in the future. These decisions rely on forecasts about supply, demand, and external risks to the system that are inherently uncertain. To address uncertainties with potential to be reduced over time with more information, I develop a method that combines Bayesian inference with multi-stage stochastic programming. I use this approach to assess both the value of information and the value of flexibility — in the form of physical infrastructure design, operations, or staged development. I demonstrate the utility of this framework using real-world cases with contrasting uncertainties. In Melbourne, Australia a flexible, modular approach to desalination capacity can reduce the risk of overbuilding, saving an expected $1 billion over a 30-year planning period. In Riyadh, Saudi Arabia, poor groundwater characterization creates uncertainty in aquifer depletion. Flexible timing of new capacity can reduce cost risk without impacting reliability. Finally, in Mombasa, Kenya, model uncertainty — or disagreement across climate models in projections of change — can be reduced as new observations are collected. This highlights the value of a flexible design in which a dam be easily be modified to increase reservoir storage in the future. Across all cases, deciding whether investments in flexibility are worthwhile rests on value judgments about the impact water scarcity, planners' appetite for risk, and society's willingness to invest in the well-being of future generations.