Quantitative Forecasting for Energy Applications

Speaker: Professor Daniel M. Tartakovsky (CV)

Abstract:

Design and operations of energy systems rely, to a large degree, on mathematical models whose veracity is often questionable. Selection of a proper model is driven by a quantity of interest (i.e., by specific questions one is asked) and by tolerance to risk (i.e., by the level of predictive uncertainty one is willing to tolerate). We elucidate this problem by considering several energy-related applications that range from batteries to pipelines, and span the scales from nanometers to kilometers. The first application deals with design of hierarchical nanoporous carbon-based materials that exhibit high specific electric double layer capacitances and high rate capability in an organic electrolyte, which makes them attractive for energy storage. The second application is related to optimal siting of ground heat exchangers, which exploit differences between air and soil temperatures to heat and/or cool buildings. The third application stems from the need of insurance companies to estimate the risks of a pipeline rapture and associated environmental remediation costs. Put together, these problems illustrate an application-centered modeling philosophy that is the opposite of a proverbial hammer in search of a nail.