Thermal Runaway: how numerical simulation can help you in developing safer Li-Ion batteries

Table of Contents

1. Why Thermal Runaway is the main safety concern for BEVs
2. What is and what are the main causes of Thermal Runaway
3. Best practices in Thermal Runaway modelling
4. POWERTECH Engineering Case studies
5. Final Remarks

1.??????Why Thermal Runaway is the main safety concern for BEVs

Safety concerns are currently recognized as one of the main obstacles that hinders large-scale applications of lithium batteries in Electric Vehicles (EVs), according to Xuning?et al. [1]. In the quest for sustainable transportation alternatives, during the last few years BEVs have gained momentum in terms of customer appeal and market share. Lithium, widely adopted as conductor material for batteries, has contributed to this trend for its high specific energy and performance, but when it comes to thermal stability it shows its shadow side. Although the incidence of cases is low, some BEVs fire have been reported by the media. These reports may significantly hinder the marketability of BEVs and the cost of BEVs recalls for OEMs and battery suppliers. These fires usually occur due to the so-called Thermal Runaway (TR).

The Thermal Runaway is the most critical phenomenon concerning the reliability and safety of Li-Ion batteries; it is a rapid and uncontrolled self-heating of a battery cell due to exothermic chemical reactions, that end up with catastrophic release of cell energy. TR is the Mr. Hyde of any battery chemistry, but it becomes particularly important for Li-Ion batteries safety because of their intrinsic features. Li-Ion batteries have high energy density, the electrolytes are flammable, and they contain metal oxides with oxygen available which can lead to combustion: the energy release may be higher than the pure electric energy stored. Even a depleted battery can bring about harmful incidents!

A thermal runaway event can lead to the expulsion of so called “vent gases” which are reactive gases emitted by the cell. Then these gases can ignite, giving way to flaming combustion. The gases emitted may even end up in an explosion if they are not burnt right away and they accumulate in the battery cell or battery pack casing.

A video of TR is available HERE and other examples of TR of Li-Ion batteries are reported at this LINK.

2.??????What is and what are the main causes of Thermal Runaway

Thermal runaway occurs if the cell exceeds a critical temperature, above which further heating is irreversible. This critical temperature depends on battery cell chemistry and is in the order of about 70-80°C. Once started, the TR follows a process of chain reaction, during which the decomposition reactions of the battery component materials usually occur one after another.

We can subdivide the root causes of TR in three main categories: mechanical, electrical and thermal abuses.

The cause mostly related to vehicle crashes is the mechanical abuse: after a collision the shell or the battery pack is damaged and no more able to contain and guarantee that the battery cell remains intact. Also, a component may penetrate the cell casing (e.g., a nail).

Then, we have the electrical abuse. It is commonly caused by an internal short circuit, caused in turn by consequences of mechanical abuses but also by manufacturing defects and overcharge or overdischarge caused by a faulty Battery Management System (BMS).

Finally, the third category is the thermal abuse. Malfunctioning cooling systems or defective thermal management strategies not able to guarantee cell and battery optimal operating temperature, are part of this group. Indeed, the overheating of battery cells above a certain threshold can be a cause of Thermal Runaway. If you are interested in how simulation can help you in developing and validating the Thermal Management of a BEV, have a look at one of our latest case study: Virtual BEV thermal management control development by means of an integrated GT-SUITE model of HVAC and battery cooling circuits

3.??????Best practices in Thermal Runaway modelling

Historically, battery safety studies have been mainly performed through experimental approaches supported by post-mortem battery analyses from field failures as well as by safety and abuse tolerance tests. These tests, however, are affected by several limitations:

• complexity: the phenomenon is intrinsically dangerous and special care has to be taken to enclose the battery cell or pack in a robust and fireproof test bench
• destructive tests: the TR is an irreversible process that ends up in the complete failure of the battery (and the surroundings) and this translates in a high costs of the experimental test
• they are technically infeasible during the early development of large batteries.

That’s why at PWT, we are convinced that the development of TR modelling methodologies is a powerful tool to understand and predict TR, leveraging available experimental tests for model validation and reducing the need of specific experimental tests.

Although the phenomenon is complex, a virtual model aimed at reproducing TR is only bound by computational power or simulation time. Different approaches to model TR exist. For instance, the TR can be described at different scales (such as, for example, at material scale, at cell or at pack scale) or aiming for different levels of prediction capabilities.

In this last perspective, Empirical models are a first step: they are based on experimental data which are interpolated to approximate relations between measured variables. In this case the physics behind the phenomenon is hardly represented and the “prediction” relies on the quality of the data. By using empirical models, the assessment quality of the TR for Li-Ion batteries with a chemistry different from the one used during the experimental tests, is actually poor. Still, these models can be used in a BMS fault-diagnosis algorithm to forecast a possible TR event and raise a warning.

Analytical models are the next level of complexity in TR modelling. These models are composed by a set of physically based equations describing Li-Ion cells behaviour in specific domains (e.g., the electrochemical domain using the Ohm laws, the thermal domain with heat transfer relations, …). Depending on the quality of the data and the model complexity, they can provide a high level of prediction accuracy. Since they reproduce the phenomena at a higher level, the analytical models lack the ability to properly assess local effects.

Finally, Multi-Scale Multi Domain (MSMD) models can be adopted. These are a set of analytical models covering different physical aspects which are coupled to obtain multi-physics capabilities (chemical, electrical and thermal models).

An example of a state-of-the-art TR MSMD model is presented by Abada et al., [2] where 1D- 3D CFD electrochemical submodels and 3D thermal submodels are coupled together. With this approach, temperature non-uniformities, which are fundamental for TR, can be tracked and taken into account by the 3D model. Thermal abuse of Li-Ion cells is indeed greatly affected by the local distributions of heat and materials. Moreover, various thermal boundary conditions can be simulated considering local cooling or heating effects: the 3D model of the battery can also be coupled with a 1D model of the cooling system.

A MSMD model aimed at describing Thermal Runaway includes, at least, two submodels:

• a Chemical model that describes the chain reaction that takes a battery cell to thermal runaway. The two most used reactions mechanisms adopted for the 3D chemical submodel are: the Kim reaction mechanism (from [3]) which describes the chain reactions during TR events accounting for 4 different exothermic reactions and the Ren reaction mechanisms [4], which account for 6 different exothermic reactions. While these models were developed for a given battery chemistry, they can be extended to other Li-Ion batteries with limited effort.
• a 3D-CFD thermal model of the battery cell (and/or pack depending on the aim of the simulation) and the surrounding environment. The 3D thermal model of the battery requires the cell geometry, materials, and boundary conditions. The heat source from the chemical exothermic reactions is considered, but also the heat source from an electrical model is useful as well. Specific care should be taken in the modelling of the cell geometry to physically represent the heat transfer in all directions. Contact resistance plays a role, too!

A MSMD model may include additional sub-models: electrical models (to account for electric energy balance before and during the TR), ageing models or venting models.

5.??????POWERTECH Engineering Case studies

In PWT, we’ve reproduced in a virtual 3D-CFD environment the nail penetration experiment from Zhang et al [5]. A MSMD model, which includes cell reaction kinetics and heat transfer across the cell, is employed and the nail penetration in the single cell is modelled as a local heat source. As a result, the simulation can predict the onset of combustion within the cell, prompted by the mechanical damage, with a relatively low computational effort (25 minutes).

We’ve also modelled the propagation of the thermal runaway in a battery pack to understand how cooling conditions and pack geometry can speed up (or slow down) this phenomenon. Starting from the detailed geometry of an air-cooled battery pack for e-bike applications, we’ve prepared a MSMD simulation setting (chain reaction kinetics and 3D-CFD Conjugated Heat Transfer simulation). If a TR event occurs in a single cell, we’ve seen that slight differences in inflow and environmental BCs play a significant role in TR propagation, as shown in the figure below.

Gas vent and auto-ignition simulations are another topic we’ve been investigating to assess the spread of vented gas across battery packs and the flame propagation after auto-ignition occurs.

5.??????Final Remarks

The thermal runaway is a very complex and uncertain phenomenon, and although the methodology to reproduce TR is easier said than done, numerical simulation can help in diagnosing issues and prevent possible TR events. For a given and proved TR event, numerical models can be used as virtual test rig for hardware modifications (e.g., to improve the thermal isolation among battery cells) and the design of a safe and reliable Li-Ion battery for BEV application.

That’s why the development of reliable simulation tools and methodologies for hybrid, electric or hydrogen-powered vehicles is our focus and objective. Since 2007.

If you want to know more about how numerical simulation can help you in disclosing the secrets of Thermal Runaway, feel free to contact us at [email protected]. We will be glad to chat with you!

References

[1] Xuning F., Minggao O., Xiang L., Languang L., Yong X., Xiangming H., “Thermal runaway mechanism of lithium ion battery for electric vehicles: A review”, Energy Storage Materials, https://dx.doi.org/10.1016/j.ensm.2017.05.013

[2] Abada, S., Marlair, G., Lecocq, A., Petit, M., Sauvant-Moynot, V., et al.. “Safety focused modeling of lithium-ion batteries: A review”, Journal of Power Sources, Elsevier, 2016, 306, pp.178-192.10.1016/j.jpowsour.2015.11.100

[3] Kim, G-H., Pesaran, A., Spotniz, R., “A three-dimensional thermal abuse model for lithium-ion cells”, Journal of Power Sources 170, pp. 476–489, 2007

[4] Ren, D., Liu, X., Feng, X., Lu, L., Ouyang, M., Li, J., He, X., “Model-based thermal runaway prediction of lithium-ion batteries from kinetics analysis of cell components”, Applied Energy, Volume 228, 2018, Pages 633-644, ISSN 0306-2619, https://doi.org/10.1016/j.apenergy.2018.06.126

[5] Zhang, L., Zhao, P., Xu, M., Wang, X., “Computational identification of the safety regime of Li-ion battery thermal runaway”, Applied Energy 261, 114440, 2020