Physics-constrained machine learning for reduced composition space chemical kinetics
Modeling detailed chemical kinetics is a primary challenge in combustion simulations. We present a novel framework to enforce physical constraints, specifically total mass and elemental conservation, during the reaction of ML models’ training for the reduced composition space chemical kinetics of la...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Cambridge University Press
2025-01-01
|
| Series: | Data-Centric Engineering |
| Subjects: | |
| Online Access: | https://www.cambridge.org/core/product/identifier/S2632673625100129/type/journal_article |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849319422363172864 |
|---|---|
| author | Anuj Kumar Tarek Echekki |
| author_facet | Anuj Kumar Tarek Echekki |
| author_sort | Anuj Kumar |
| collection | DOAJ |
| description | Modeling detailed chemical kinetics is a primary challenge in combustion simulations. We present a novel framework to enforce physical constraints, specifically total mass and elemental conservation, during the reaction of ML models’ training for the reduced composition space chemical kinetics of large chemical mechanisms in combustion. In these models, the transport equations for a subset of representative species are solved with the ML approaches, while the remaining nonrepresentative species are “recovered” with a separate artificial neural network trained on data. Given the strong correlation between full and reduced solution vectors, our method utilizes a small neural network to establish an accurate and physically consistent mapping. By leveraging this mapping, we enforce physical constraints in the training process of the ML model for reduced composition space chemical kinetics. The framework is demonstrated here for methane, CH4, and oxidation. The resulting solution vectors from our deep operator networks (DeepONet)-based approach are accurate and align more consistently with physical laws. |
| format | Article |
| id | doaj-art-be6fcea4e8b2495081fe283090c9b96d |
| institution | Kabale University |
| issn | 2632-6736 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Cambridge University Press |
| record_format | Article |
| series | Data-Centric Engineering |
| spelling | doaj-art-be6fcea4e8b2495081fe283090c9b96d2025-08-20T03:50:27ZengCambridge University PressData-Centric Engineering2632-67362025-01-01610.1017/dce.2025.10012Physics-constrained machine learning for reduced composition space chemical kineticsAnuj Kumar0Tarek Echekki1https://orcid.org/0000-0002-0146-7994Department of Mechanical and Aerospace Engineering, https://ror.org/04tj63d06 North Carolina State University , Raleigh, NC, USADepartment of Mechanical and Aerospace Engineering, https://ror.org/04tj63d06 North Carolina State University , Raleigh, NC, USAModeling detailed chemical kinetics is a primary challenge in combustion simulations. We present a novel framework to enforce physical constraints, specifically total mass and elemental conservation, during the reaction of ML models’ training for the reduced composition space chemical kinetics of large chemical mechanisms in combustion. In these models, the transport equations for a subset of representative species are solved with the ML approaches, while the remaining nonrepresentative species are “recovered” with a separate artificial neural network trained on data. Given the strong correlation between full and reduced solution vectors, our method utilizes a small neural network to establish an accurate and physically consistent mapping. By leveraging this mapping, we enforce physical constraints in the training process of the ML model for reduced composition space chemical kinetics. The framework is demonstrated here for methane, CH4, and oxidation. The resulting solution vectors from our deep operator networks (DeepONet)-based approach are accurate and align more consistently with physical laws.https://www.cambridge.org/core/product/identifier/S2632673625100129/type/journal_articlechemistry accelerationdeep operator networksphysics-informed machine learningreduced order modelingsurrogate modeling |
| spellingShingle | Anuj Kumar Tarek Echekki Physics-constrained machine learning for reduced composition space chemical kinetics Data-Centric Engineering chemistry acceleration deep operator networks physics-informed machine learning reduced order modeling surrogate modeling |
| title | Physics-constrained machine learning for reduced composition space chemical kinetics |
| title_full | Physics-constrained machine learning for reduced composition space chemical kinetics |
| title_fullStr | Physics-constrained machine learning for reduced composition space chemical kinetics |
| title_full_unstemmed | Physics-constrained machine learning for reduced composition space chemical kinetics |
| title_short | Physics-constrained machine learning for reduced composition space chemical kinetics |
| title_sort | physics constrained machine learning for reduced composition space chemical kinetics |
| topic | chemistry acceleration deep operator networks physics-informed machine learning reduced order modeling surrogate modeling |
| url | https://www.cambridge.org/core/product/identifier/S2632673625100129/type/journal_article |
| work_keys_str_mv | AT anujkumar physicsconstrainedmachinelearningforreducedcompositionspacechemicalkinetics AT tarekechekki physicsconstrainedmachinelearningforreducedcompositionspacechemicalkinetics |