Decomposition of uncertainty propagation through networks of heterogeneous energy systems
Abstract:
Model-based design and analysis of any complex dynamical system is never exact and must consider uncertainties at different stages of the analysis process. In model-based analysis, uncertainty is introduced from many different sources including inadequate assumptions (poor parameter choice), or inappropriate structure or equations within the model. In the experimental or operative environment, uncertainty also comes from many sources including external (e.g. weather) or internal (e.g. social behavior) disturbance.
In this work we present an uncertainty analysis paradigm that not only calculates uncertainty and sensitivities between these specific sources and key outputs of the system, we also decompose the pathway that uncertainty proceeds through the entire set of dynamics. In this way we can illuminate where uncertainty is amplified or attenuated throughout the system. Our specific work is associated with building energy modeling (which is a heterogeneous interconnected environment), while the future direction is to expand the spatial scale to study models of interconnected communities including smart grid architecture that contains uncertain sources of energy production (e.g. wind generation).