This article is contributed by Brennan Campbell – Ph.D., Battery Expert at AVL, Thomas Glatz – Business Development Manager Advanced Simulation Technologies at AVL, and Ben Playfoot – Segment Lead Electrification at AVL
Introduction: The Evolving Landscape of Battery Safety
As lithium-ion batteries become the backbone of modern vehicles and energy systems, safety, alongside cost, has emerged as a defining challenge for EVs, stationary storage, next-generation aviation, power tools, and beyond. High-profile incidents including thermal runaway events, system failures, and the battery plant fire in Moss Landing, CA, have accelerated the push for stricter safety regulations (see Fig. 1). From a global perspective, China’s upcoming battery safety regulations are slated to be significantly more stringent mandating no fire and no explosion (see Fig. 2). European and U.S. markets are expected to follow China’s regulatory lead.
Figure 1: Growing flames at Moss Landing battery fire (source: insideclimatenews.org)
Figure 2: Global Thermal Propagation Regulations Overview [1]
Battery safety engineers need more than robust materials to meet increasingly strict regulations to balance safety and performance. The following core concepts help minimize the risk of thermal propagation:
- Choosing the right cell chemistry and cell packaging
- Avoiding gas ignition
- Stopping cell-to-cell propagation
- Avoiding arcing and preventing short-circuiting
- Providing sufficient heat sinks
To meet these evolving regulations and to help ensure battery thermal safety across all applications, AVL has developed a “No Flame-Out” pack design that embodies all the above concepts. This design was made possible through a “three-legged stool” approach, which combines simulation, empirical testing, and analytics. By employing an integrated approach combining multi-scale simulation, rigorous safety testing, and data-driven analytics, engineering teams can develop safer, more compliant battery designs on a shorter time scale. In the following sections, we will take a deeper look into this approach, including examples of model validation and advanced methods leveraging AI.
Developing Safe Designs through Virtualization
While physical testing is a cornerstone of a Design Verification Plan (DVP), it cannot cost-effectively capture all real-world abuse scenarios. Engineers need to predict hazards and validate safety strategies long before hardware even exists. Simulation enables running thousands of virtual “what-if” scenarios, which reduce destructive tests and accelerate development timelines [2].
How can engineers utilize modeling to reduce Thermal Propagation (TP)?
Accurate thermal propagation modeling begins with a solid understanding of normal operating conditions including:
- Ohmic Heating: for designing cooling circuits, thermal design
- Equivalent Circuit Models: fast running, BMS development
- Electrochemical Models: deep chemistry understanding, normal and destructive
- Batemo Cell Model Library: ready to go, highly accurate benchmarked models
Beyond normal operation, a range of hazard scenarios are available to help ensure safety:
- Basic effects including cell heat release, venting, melting, burst discs, flammability
- Particle behavior
- Arc probability
- Solid body chemistry
- Pack deformation
Figure 3: Simulating hazardous scenarios
Engineers can leverage system-level simulation, 3D computational fluid dynamics (CFD) and hybrid solutions in combination with one another.
- Real-time capable system simulation focuses on electrochemical and thermal modeling, fast-charging analysis, and system-level safety from cell to pack.
- 3D-CFD simulation is used to perform detailed simulations of thermal design, thermal runaway, vent gas flow, and solid particle ejecta.
Figure 4: Comparison of the two major physics-based simulation methodologies for a 15min battery module thermal propagation event.
Advanced Effects: From Cell Heat Release to Arcing and Particle Ejection
At the core of each advanced thermal propagation simulation is accurate modeling of cell heat release. Figure 5 shows three common approaches used in the development process. Achieving an exceptionally reliable prediction of heat generation and venting gas enables virtual validation before committing to costly physical pack level tests.
Figure 5: Approaches for modeling cell heat release, including validation for different cell types and chemistries. [3]
Regardless of the heat release level, achieving a no flame-out design depends on eliminating potential ignition sources. Evaporated electrolytes can ignite due to excessive heat or electrical arc. To minimize risk, arcing probability is simulated using Paschen’s law taking into account geometry, distance, and pressure. In addition to arcing risk, damage caused by hot particle ejection can be simulated by predicting melting point and material removal. Figure 6 illustrates how simulating these effects provides engineers with critical insights for designing safer battery packs.
Figure 6: Simulating arcing and particle ejection. [4]
Hybrid and AI-Driven Workflows
Integrating 1D system models with 3D simulation provides a balance between speed and detail, however this is only the starting point for more sophisticated workflows. Engineers can use AI trained on these simulations to explore entire design spaces, while applying the same models for instant trade-off analysis among safety, cost, and performance. The training process utilizes real and simulated data or a combination of both.
Figure 7: Engineers can combine 1D and 3D models with Machine Learning to create superfast models that scan the entire design space for optima.
Battery Testing: Capturing the Full Picture of Battery Behavior
Simulation models require test data to validate the accuracy of models in the final design. For battery safety this does not mean just standard cycling or validation. To achieve a no flame-out pack design, engineers must combine advanced simulation with rigorous empirical testing to characterize how cells and sub-assemblies behave under extreme abuse conditions. These tests are conducted in a controlled, instrumented safety test chamber: a controlled environment designed to contain high-energy events and capture critical data without external flame release (see Fig. 8).
This safety test protocol replicates worst-case scenarios such as overcharge, nail penetration, and thermal ramp conditions. High-speed thermocouples, pressure transducers, and gas analyzers record transient phenomena including venting onset, gas composition, and heat release rates, while multi-angle video analysis documented flame evolution and particle ejection. These experiments characterize the two-phase nature of thermal runaway, an initial slow temperature rise followed by a rapid, severe escalation which is critical for sizing heat sinks and designing venting paths.
One of the most significant outcomes is the validation of safe venting strategies. Pack-level concepts demonstrates that gases initially exceeding 1300 °C can be cooled to below 100 °C before exiting the pack, eliminating external ignition risk, and meeting “no fire, no explosion” requirements. Propagation studies further reveal that failure accelerates when adjacent cells experience simultaneous mechanical deformation and electrolyte exposure, driving the introduction of thermal barriers and mechanical separators to interrupt heat transfer and prevent cascading failures.
Figure 8: AVL Safety Test Chamber
Abuse Testing Activities for Safety and Compliance
To ensure safety and regulatory compliance, engineers perform several evaluations:
- Quantify heat release and pressure rise during runaway for accurate thermal modeling.
- Characterize vent gas composition and ignition thresholds to inform venting design.
- Validate cooling strategies that reduce vent gas temperature from ~1300 °C to <100 °C.
- Identify propagation accelerants, leading to thermal and mechanical isolation measures.
Compliance with global standards (UN38.3, IEC, ISO, SAE, GB) is embedded in the workflows. These evaluations commonly involve dedicated bunkers for each test; AVL has streamlined this testing by developing a modular Safety Test Chamber for cost-effective execution and high throughput. The empirical dataset from these programs feeds directly into 1D and 3D simulations, enabling high-fidelity prediction of propagation behavior and iterative optimization of module partitioning, venting architecture, and material selection for thermal buffering.
While full-scale pack testing remains essential for final validation, the combination of detailed small-scale abuse testing and validated simulation models reduces destructive pack tests to a fraction of traditional programs, saving time and cost without compromising safety assurance. This closed-loop approach is instrumental in achieving a pack architecture that demonstrates zero external flame release under worst-case conditions.
Figure 9: Examples of custom battery data dashboards for post-processing for lab/field.
Engineering teams can harness data analytics to process thousands of channels from abuse tests and simulations, detecting critical events such as abnormal resistance growth or runaway precursors. The image above shows how discharge capacity and dynamic resistance trends are visualized for anomaly detection and degradation tracking. These insights validate models, refine test strategies, and accelerate design decisions.
Once products are deployed, analytics shift to field data. Real-world performance testing through digital twins and fleet-level monitoring help identify patterns such as cell imbalance or early-cycle degradation; these signals cannot be reproduced in a lab. These findings feed back into simulation and testing, improving predictive fidelity and guiding new design iterations.
Analytics completes the Testing–Simulation–Analytics framework by:
- Validating models against empirical and field data.
- Highlighting gaps in test coverage and informing new abuse scenarios.
- Driving continuous design optimization for thermal propagation mitigation and lifecycle safety.
This feedback loop ensures that every pack design evolves from data-driven foresight making safety not just a validation step, but an ongoing process.
Figure 10: Data analytics completes the state-of-the-art “three-legged stool” battery development framework
Conclusion: A No Flame-Out Battery Design
By applying the “three-legged stool” approach: Testing, Simulation, and Analytics, engineering teams can deliver a No Flame-Out Design for applications covering mobility, grid storage, aviation, and marine:
- Testing: capturing real behavior under normal and extreme conditions.
- Simulation: modeling complex interactions to guide safer design.
- Analytics: transforming data into continuous insight and prediction.
This method allows engineering teams to achieve the following KPIs ensuring safety and compliance:
- Gas Cooling: Vent gases reduced from ~1300 °C to <100 °C before exiting the pack.
- Propagation Control: Thermal runaway contained within the initiating module; no cell-to-cell propagation beyond design limits.
- External Flame: Zero external flame observed under worst-case abuse conditions.
- Regulatory Compliance: Meets “no fire, no explosion” requirements per GB 38031:2025 and UN ECE R100 proposals.
- Testing Efficiency: Full-scale pack tests minimized through validated small-scale testing and simulation correlation.
Figure 11: A closer look at the NFO battery design [5]
References
- Juergen Schneider, “Virtual Models, Real Safety: New Approaches to Safe Battery Design”, AVL Webinar, (2025), https://www.avl.com/en-at/webinars/virtual-models-real-safety-new-approaches-safe-battery-design
- Zaman Sadeghi and Thomas Glatz, “Modeling and Simulation to Accelerate Automotive Battery Design and Development Time with Henkel and AVL, Webinar, Battery Forum, 2025, https://www.youtube.com/watch?v=1Idtvxksl2M
- Thirumalesha Chittipotula, Lucas Eder and Thomas Uhl, “Application of a Chemical Kinetic Modeling Approach within a Fully Coupled Computational Fluid Dynamics Simulation of Battery Cells during Thermal Runaway,” SAE Int. J. Elec. Veh. 14(2):261-279, 2025-, https://doi.org/10.4271/14-14-02-0014.
- Thirumalesha Chittipotula, Lucas Eder and David Schellander, “Hazard Predictions Models for Battery Module and Packs: Flammability, Particle Ignited Vent Gas, Arcing without and with Particles” The 19th International Conference on Fluid Flow Technologies At: Budapest, Hungary, 2025
- Paul Schiffbaenker and Tobias Stahl, “Acting upon regulation GB38031-2025 “No Fire – No Explosion”” Whitepaper, AVL and H&Z Group 2026, https://www.avl.com/en-at/form/custom-downloads?download=50521 at https://www.avl.com/en-at/engineering/e-mobility-engineering/battery-development-electric-vehicles
