Adding a Computationally-Tractable Probabilistic Dimension to Meta-Heuristic-Based Microgrid Sizing

A robust solution to the optimal micro-grid (MG) sizing problem requires comprehensive quantification of the underlying parametric uncertainties - particularly, the uncertainty in forecasts of meteorological, load demand, and wholesale electricity price time-series data. However, the associated data-driven processes for probabilistic uncertainty quantification are computationally expensive. Accordingly, the mainstream meta-heuristic-based MG sizing literature has failed to concurrently quantify more than four sources of forecast uncertainty. To address this knowledge gap, this paper introduces a novel computationally efficient, probabilistic MG sizing model that enables the simultaneous treatment of any (reasonable) number of data uncertainty. This provides a platform to characterize the uncertainty in ambient temperature and river streamflow for the first time in the MG optimal sizing literature. Importantly, the model supports the associated long-term strategic MG energy planning optimization processes through in-depth analyses of the worst-case, most likely case, and best-case planning scenarios. To demonstrate the utility of the proposed model for community MG projects, a case study is presented for the town of Ohakune, New Zealand. Notably, the numeric simulation results have shown that the whole-life cost of the conceptualized MG would have been underestimated and overestimated by as much as 17% and 30% respectively in the best-case and worst-case scenarios if the problem-inherent uncertainties were not explicitly factored into the associated techno-economic analyses.

Automated summary

This paper introduces a novel probabilistic MG sizing model that can simultaneously handle multiple sources of data uncertainty and support long-term strategic MG energy planning optimization processes. A case study is presented to demonstrate the utility of the model in community MG projects.