Techno-economic assessment of photovoltaic plants considering high temporal resolution and non-linear dynamics of battery storage

Abstract

The addition of energy storage systems for photovoltaic plants in order to provide reliability and flexibility in their dispatch is a very active line of research, as it studies plant design configurations and different dispatch strategies to improve the coupling of variable renewable solar generation with conventional hydro/thermal electric grids to enable large-scale participation of renewables in these systems. Nevertheless, battery degradation is a critical parameter to assess the plant’s lifetime performance as it depends upon the battery’s operational principles, design and usage pattern, which coupled with their inherent non-linearities makes them a potential source of uncertainty in assessments, should the involved phenomena be overly simplified. In this work, we explore different photovoltaic plant configurations with batteries operating under a dispatch optimization algorithm that considers operational power forecasts, market pricing information and a series of constraints to improve their coupling with electric networks. The battery system is modelled considering the electrical, thermal and degradation phenomena occurring during its life cycle, as well as considering a variable charge/discharge efficiency and it’s simulated at high temporal resolution to account for any sub-hourly non-linear dynamics. The results show that considering time steps over 30 min for battery simulation during dispatch can lead to noticeable differences in key performance indicators when compared to one-minute cases and it was found that 10-minutes offered the best compromise in accuracy and computational cost. Additionally, it was found that, depending on their design, effective yearly degradation rates for batteries are highly variable and can reach values that deviate substantially from commonly-used constant assumptions, demonstrating the value of degradation modeling for lifecycle assessment. Finally, it was observed that plant configurations suited for specific operational modes such as smoothing and peak shaving could be obtained by analyzing their dispatch characteristics under the dispatch optimization algorithm, instead of following the usual approach of setting the operational mode and tuning the plant for it. This finding suggests that a design philosophy that adapts the plant’s design to the dynamic the market’s dynamic could be highly suitable for electric systems in which no central agent determines the expansion of the system.