Quantifying the effects of forecast uncertainty on the role of different battery technologies in grid-connected solar photovoltaic/wind/micro-hydro micro-grids: An optimal planning study

Designing a reliable and robust micro-grid (MG) aided by energy storage devices requires quantifying the parametric uncertainty associated with input data time-series, among other types of uncertainties – and in particular, the uncertainty in forecasted meteorological, load demand, and wholesale electricity price time-series. Given the computational complexities involved, the recent battery-storage-supported MG capacity optimisation literature fails to simultaneously characterise multiple sources of data uncertainty. This paper, therefore, seeks to address this knowledge gap by hedging against a wide array of parametric uncertainties at the same time, whilst exploring the potentially salient implications of high-dimensional uncertainty characterisation for the costing and configuration of battery-supported, grid-tied MGs. To this end, this paper introduces a novel computationally efficient, stochastic MG design optimisation model that enables the simultaneous handling of multiple uncertain inputs. Notably, the model yields in-depth, accurate, and robust infrastructure decision-making support, particularly with regard to the optimal size of battery storage, by specifically analysing the worst-case, most likely case, and best-case uncertainty characterisation scenarios. To demonstrate the effectiveness of the model within a community renewable energy project scheme, a case study of a grid-connected, battery-storage-supported MG in the town of Ohakune, New Zealand is presented. Importantly, the numerical simulation results indicate that the life-cycle cost of the MG would have been underestimated by as much as ~8% (equating to NZ$0.3 m) in the most likely case – with an associated ~15% (equating to 143 kWh) battery size underestimation – if the variability inherent in forecast inputs was not factored into the analysis. Also, the proportion of the energy storage capacity to the total renewable power generation capacity is ~71%, ~76%, and ~68% in the best-case, most likely case, and worst-case scenarios, respectively, suggesting that the risk-seeking and risk-averse MG planning strategies have a slightly contradictory effect on the storage-to-generation ratio. Comprehensive sensitivity analyses of different battery technologies with different specific power, specific energy, and depth-of-discharge specifications have also demonstrated the robustness and validity of the model, whilst additionally identifying the sodium‑sulfur (NaS) battery technology as the most profitable choice in grid-connected MG development applications – which can be primarily attributed to the fact that it provides the best trade-off between the specific power and specific energy.