Analyzing system safety in lithium-ion grid energy storage

Abstract

As grid energy storage systems become more complex, it grows more difficult to design them for safe operation. This paper first reviews the properties of lithium-ion batteries that can produce hazards in grid scale systems. Then the conventional safety engineering technique Probabilistic Risk Assessment (PRA) is reviewed to identify its limitations in complex systems. To address this gap, new research is presented on the application of Systems-Theoretic Process Analysis (STPA) to a lithium-ion battery based grid energy storage system. STPA is anticipated to fill the gaps recognized in PRA for designing complex systems and hence be more effective or less costly to use during safety engineering. It was observed that STPA is able to capture causal scenarios for accidents not identified using PRA. Additionally, STPA enabled a more rational assessment of uncertainty (all that is not known) thereby promoting a healthy skepticism of design assumptions. We conclude that STPA may indeed be more cost effective than PRA for safety engineering in lithium-ion battery systems. However, further research is needed to determine if this approach actually reduces safety engineering costs in development, or improves industry safety standards.

Upholding Safety in Future Energy Systems: The Need for Systemic Risk Assessment

“…These results confirm findings from comparable studies. The omission of parameters, such as nitrogen content of gas or its odorization, as illustrated above, is in line with several critiques that observe the difficulty of providing a full inventory of hazards with analytic approaches, such as HAZOP and FMEA (i.e., References [4,[18][19][20]). For HAZOP, this is partly due to its inherent structure that determines hazards by focusing on the P&ID and relies on the team brainstorming to generate a design intent and possible deviations from it [42].…”

“…Hence, it is a clear example of how the system becomes more complex. In line with previous studies [18][19][20], we are particularly interested in how the two approaches differ in dealing with digital instrumentation, the number and nature of identified hazard scenarios, and ways in which uncertainty is accounted for.…”

“…Some recent exceptions underline the untapped potential for systemic risk assessment in the energy sector. Rejzek and Hilbes stress their added value in the design of digital instrumentation for nuclear power plants [18]; Rosewater and Williams analyzed safety in new battery designs using a systematic risk assessment and conclude that it creates more causal scenarios for accidents compared to their analytic counterpart, in addition to providing a better answer to hazards related to uncertainties that come with new technologies [19]; and Karatzas and Chassiakos find that a systemic risk assessment is instrumental in identifying gaps in current risk assessments in the context of the increased complexity of domestic energy systems [20]. Their results suggest that systemic methods are indeed more appropriate for identifying and potentially mitigating novel challenges associated with a transition to renewable energy sources.…”

Battery Management System Hardware Concepts: An Overview

“…Afterwards, design aspects related to the safety of the BMS itself are discussed. According to [59], the main hazards inside lithium-ion battery packs are exposure to high voltage, arcing, fire, vented gas combustibility and vented gas toxicity. In combination with suitable battery pack design, the safety functions of the BMS should minimize the likelihood of occurrence and severity of these events.…”

Human factors and ergonomics systems-based tools for understanding and addressing global problems of the twenty-first century

“…The study that most closely aligns to the types of global problems envisaged by this paper is Rosewater and Williams' (2015) investigation of a lithium-ion energy storage system. The system investigated by Rosewater and Williams (2015) was relatively small, although it is fairly easy to see how this type of analysis might be scaled-up to larger energy grid systems. In general though STAMP has been characterised as difficult to use and the graphical outputs have been difficult to interpret (Underwood, Waterson, & Braithwaite, 2016).…”