Open Access
Open access
volume 22 pages 100326

Development of compact mechanism for lithium-ion battery venting gas fires using Cantera ordinary differential equation neural network algorithm

Publication typeJournal Article
Publication date2025-06-01
scimago Q1
wos Q1
SJR1.460
CiteScore7.9
Impact factor6.0
ISSN2666352X
Abstract
Lithium-ion battery fires pose significant challenges to the development of electric vehicles and energy storage systems due to their potential hazards and complex combustion behavior. To address these issues, this work develops the Cantera Ordinary Differential Equation Neural Network (CODENN) algorithm, which combines the computational power of neural ordinary differential equations with Cantera's advanced chemical kinetics modeling. This integration allows for the optimization of a wide range of chemical reactions, improving both the precision and versatility of reaction mechanism development. Using CODENN, a compact mechanism (COM) was developed by optimizing the Arrhenius parameters of the RED mechanism, which had been overly reduced from the detailed CRECK2003 mechanism (114 species, 1999 reactions). CRECK2003 was chosen for its proven accuracy in predicting the combustion properties of LIB venting gases. The resulting COM mechanism, with 30 species and 213 reactions, achieves a high fidelity of 94.8 % in predicting ignition delay times across equivalence ratios from 0.3 to 2.5, demonstrating the reliability and robustness of CODENN algorithm. Further analysis of speciation data and CO net production rates shows that the COM mechanism closely aligns with the species evolution of the ORI mechanism during autoignition, while exhibiting notably more intense CO production and consumption than the ORI mechanism. Path flux analysis indicates that, despite having shorter reaction chains than the ORI mechanism, the COM mechanism preserves the fundamental physical logic of fuel consumption (CH₄) leading to H₂O formation while introducing additional pathways for the generation and consumption of H and OH radicals. Sensitivity analysis across diverse equivalence ratios and temperatures consistently identifies reaction R5 (H + O₂ ≤> O + OH) as the most temperature-sensitive reaction, underscoring its critical role in reaction kinetics of LIB venting gases.
Found 
Found 

Top-30

Journals

1
Applications in Energy and Combustion Science
1 publication, 100%
1

Publishers

1
Elsevier
1 publication, 100%
1
  • We do not take into account publications without a DOI.
  • Statistics recalculated weekly.

Are you a researcher?

Create a profile to get free access to personal recommendations for colleagues and new articles.
Metrics
1
Share
Cite this
GOST |
Cite this
GOST Copy
Li M. et al. Development of compact mechanism for lithium-ion battery venting gas fires using Cantera ordinary differential equation neural network algorithm // Applications in Energy and Combustion Science. 2025. Vol. 22. p. 100326.
GOST all authors (up to 50) Copy
Li M., Hu H., Lu L., Zhang H. Development of compact mechanism for lithium-ion battery venting gas fires using Cantera ordinary differential equation neural network algorithm // Applications in Energy and Combustion Science. 2025. Vol. 22. p. 100326.
RIS |
Cite this
RIS Copy
TY - JOUR
DO - 10.1016/j.jaecs.2025.100326
UR - https://linkinghub.elsevier.com/retrieve/pii/S2666352X25000081
TI - Development of compact mechanism for lithium-ion battery venting gas fires using Cantera ordinary differential equation neural network algorithm
T2 - Applications in Energy and Combustion Science
AU - Li, Mengjie
AU - Hu, Hao
AU - Lu, Li
AU - Zhang, Huangwei
PY - 2025
DA - 2025/06/01
PB - Elsevier
SP - 100326
VL - 22
SN - 2666-352X
ER -
BibTex
Cite this
BibTex (up to 50 authors) Copy
@article{2025_Li,
author = {Mengjie Li and Hao Hu and Li Lu and Huangwei Zhang},
title = {Development of compact mechanism for lithium-ion battery venting gas fires using Cantera ordinary differential equation neural network algorithm},
journal = {Applications in Energy and Combustion Science},
year = {2025},
volume = {22},
publisher = {Elsevier},
month = {jun},
url = {https://linkinghub.elsevier.com/retrieve/pii/S2666352X25000081},
pages = {100326},
doi = {10.1016/j.jaecs.2025.100326}
}
Profiles