By Steven Popper
The annual meeting of the Society for Decision Making under Deep Uncertainty (DMDU) will feature two panels on applications to energy and infrastructure planning. This is not surprising. These are in many ways the classic sectors in which DMDU concepts and methods first arose. Therefore, these presentations will be of interest not only to practitioners in these fields but also to those interested in DMDU technique more generally and who seek ideas and guidance for applications to other venues.
What makes energy and infrastructure so paradigmatic for DMDU in general? Certainly, the long service lifetimes planned for these systems plays a leading role. The time has long since passed when it was credible to rely upon past economic, climatological, technological or even societal rules of thumb and presume they would reliably extend across decades into the future. No public agency with plan review authority would accept such premises today. But what then is to be done? Generate scenarios? That is the usual course and represents the current state of the art – or does it?
Analysts applying DMDU concepts and methods can – and do – routinely advance the frontiers of this presumed state of the art. To be sure, many of these techniques can rightly be seen as being grounded in scenario thinking. But whereas most scenario approaches require analysts to select ex ante what are likely to be the main drivers of future outcomes as are then represented in several canonical scenarios (high/low economic growth; 1o or 2.5o ambient temperature rise; etc.,) the application of DMDU concepts often switches the process – and makes the analyses more practically useful thereby. The scenarios arise analytically ex post through iterative examination of myriad cases or simulation runs with varying assumptions. The methods also provide a system-specific definition for what constitutes an important scenario: it is one that conveys important information about the relationship between alternative courses of action and resulting outcomes. This permits a characterization of uncertainties that is both operational and situational without recourse to probabilistic assumptions regarding values of important variables in the future. This also means that the analytical output is framed precisely in the same terms as the policy deliberation it seeks to inform. Pretty cool stuff!
I will chair the annual meeting’s second session on this topic, to be held on 11 November. As is typical for this Society and of the missions it has set for itself, the talks themselves will represent a wide sphere of application. In the first of three presentations, Victor Espinoza and Edmundo Molina will take us on a tour of the pipeline infrastructure of the home country for this year’s meeting: Mexico. How may planners anticipate future requirements and do so by reducing costs, all the while meeting the challenge of future economic, political or physical risks? They develop an analytical framework that yields a robust medium and long-term strategy for the management of the Mexican gas pipeline network.
Next, Katie Popp, Prerna Singh, and Adjo Amekudzi-Kennedy review DMDU methods to understand and characterize their strengths and weaknesses for transportation decision-making within the scope of evolving black swan events during the COVID-19 pandemic. They explore the usefulness of combining multiple approaches to leverage various strengths and harness complementary attributes. Transportation involves quite literally the connection of everything to everything else – a notoriously difficult planning problem. Thus, their review will provide insights of potential value to other possible areas of application.
Finally, the team of Vivienne Liu, Amandeep Gupta, David Gold, Patrick Reed, and Lindsay Anderson provide us with a foray into the deep issues regarding future electric grid design and management, especially the transition from reliance on large, centralized distribution infrastructure to micro-grids. Again, the methodology will be as interesting for participants as the direct object of their inquiry. Using the Evolutionary Multi-Objective Direct Policy Search (EMODPS) method, they build a simulation framework for the daily energy management problem. Their work embraces quantification of tradeoffs, exploratory modeling of candidate microgrid control strategies, and scenario discovery to understand the dominant uncertainties – basically in itself a short-course practicum on key DMDU methods! These could each be the subject of an hour-long seminar. We will do it all in 60 minutes. This session will be oriented toward conveying the Big Ideas without straining the eyes or patience of attendees from across the globe who will by joining virtually. We will work to set the table in three, short, set-piece presentations while providing time for questions from and engagement with the audience. Clearly, this will be just the starting point for multiple conversations – and that is precisely what the DMDU Society is designed to be! Join us, if you can.
About this Blog Post:
This blog post is part of a series of posts contributed by the chairs of the 2020 DMDU Annual Meeting. For more information about the Annual Meeting, including registration, visit our website at 2020.deepuncertainty.org. We hope to (virtually) see you soon!