(172c) CFD Analysis of Fireball and Explosion Hazards in Battery Energy Storage Systems

Hazards from a thermal runaway event in Battery Energy Storage Systems (BESS) have the potential to cause substantial damage, threatening not only system functionality, but also the safety of firefighters, first responders, and other persons in the vicinity. While fire is the most common hazardous outcome from thermal runaway, delayed ignition and buildup of the off-gas vapor cloud have the potential to result in a deflagration or vapor cloud explosion (VCE). As the number and pace of BESS deployments continue to increase, the increased scale and energy density of the systems deployed are tending to lead also to more complex and dense structures while compounding the associated hazards. Simplified methods of analysis, such as the multi-energy method for VCE loadings or hand calculation methods summarized in NFPA 68 for vent sizing, are able to provide estimates of blast effects for simple conditions that vent directly to external conditions, but they do not provide a detailed estimate of the conditions throughout a BESS structure. Furthermore, the impact of deflagration arising from one BESS module, typically a shipping container, on other modules is not easily quantified with such simplified methods.

Arup has been working in the last five years with a new computational fluid dynamics (CFD) software, Viper::Blast, that is optimized to run on a graphics processing unit (GPU), increasing the speed and accuracy of our blast analyses by multiple orders of magnitude in terms of runtime and/or model size. This increased speed allows for larger, more detailed models as well as the development of batch analyses based on Monte Carlo and other statistical methods. By moving towards statistical, risk-based methods rather than deterministic analyses, Arup is able to understand the importance of ignition location, cloud size, gas composition, structural layout, and other details in how they influence blast overpressures and propagation.

Arup has applied this VCE methodology to a generic BESS module based on a shipping container with comparisons to publicly available experimental data from partially vented VCE’s. The flammable vapor cloud is modeled as a homogeneous mixture, predominantly hydrogen, based upon cell off-gas compositions reported in literature obtained through testing such as the UL 9540 test method. Multiple flame speeds are modeled, which can be explicitly identified in Viper::Blast, based upon a distribution between the laminar flame speed of the off-gas products as a lower-bound and the velocity of detonation as an absolute upper-bound. Results of the numerical analysis include pressure-time histories at key locations within the container as well as visualization of the blast pressures venting from the container. Comment is made on how this type of analysis can be used to provide an alternate method of vent sizing that identifies the external impact of vented explosion products. It is anticipated that this method may provide a more detailed approach for modeling VCEs and the associated hazards from BESS failure.