Modelling transient flame spread over thick PMMA slabs in high-temperature concurrent airflow using a Bayesian-calibrated simplified scale model
(2026) In Combustion and Flame 287.- Abstract
Transient concurrent flame spread over solid surfaces involves multiple interacting physical processes, where simplified modelling with effective parameter identification is still challenging. This study presents a modelling framework that combines scaling-based formulation with Bayesian parameter inference to quantitatively capture the transient behaviors of concurrent flame spread. Experimental data under varying airflow speeds and temperatures are used to train the model. Results show that the scale model, using the inferred parameters, can accurately predict the time evolution of key quantities, including the positions of flame tip, flame drag front, pyrolysis front, pyrolysis end, and mass loss rate, confirming the validity of the... (More)
Transient concurrent flame spread over solid surfaces involves multiple interacting physical processes, where simplified modelling with effective parameter identification is still challenging. This study presents a modelling framework that combines scaling-based formulation with Bayesian parameter inference to quantitatively capture the transient behaviors of concurrent flame spread. Experimental data under varying airflow speeds and temperatures are used to train the model. Results show that the scale model, using the inferred parameters, can accurately predict the time evolution of key quantities, including the positions of flame tip, flame drag front, pyrolysis front, pyrolysis end, and mass loss rate, confirming the validity of the approach. Most inferred parameters are consistent with values reported in the literature, while the Markov Chain Monte Carlo (MCMC) framework can also identify the influence of the experimental boundary conditions and reflects it in the inferred flow-related parameters. The mechanisms behind the increase in flame spread rate with airflow speed and temperature are also analysed. Flow speed accelerates flame spread by increasing flame temperature, reducing flame stand-off distance, enhancing surface heat flux and mass loss, and extending the preheating zone. In contrast, the flow temperature has a milder effect, primarily by increasing flame temperature and reducing re-radiative heat loss from the solid surface. This study demonstrates the potential of combining scaling models with Bayesian inference to model complex transient flame spread phenomena.
(Less)
- author
- Ma, Yuxuan
; Huang, Yajun
LU
; Chen, Yuhang
; Madsen, Dan
LU
; Wahlqvist, Jonathan
LU
; Ren, Fangsi
; Nakaya, Shinji
; Tsue, Mitsuhiro
; van Hees, Patrick
LU
and Hu, Longhua
- organization
- publishing date
- 2026
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Concurrent flame spread, Data-driven approach, Flow temperature, Scale model, Solid combustion
- in
- Combustion and Flame
- volume
- 287
- article number
- 114929
- publisher
- Elsevier
- external identifiers
-
- scopus:105034215918
- ISSN
- 0010-2180
- DOI
- 10.1016/j.combustflame.2026.114929
- language
- English
- LU publication?
- yes
- id
- d2c8b61c-5de9-4699-938c-7dd0b618749f
- date added to LUP
- 2026-04-24 15:12:47
- date last changed
- 2026-04-24 15:13:32
@article{d2c8b61c-5de9-4699-938c-7dd0b618749f,
abstract = {{<p>Transient concurrent flame spread over solid surfaces involves multiple interacting physical processes, where simplified modelling with effective parameter identification is still challenging. This study presents a modelling framework that combines scaling-based formulation with Bayesian parameter inference to quantitatively capture the transient behaviors of concurrent flame spread. Experimental data under varying airflow speeds and temperatures are used to train the model. Results show that the scale model, using the inferred parameters, can accurately predict the time evolution of key quantities, including the positions of flame tip, flame drag front, pyrolysis front, pyrolysis end, and mass loss rate, confirming the validity of the approach. Most inferred parameters are consistent with values reported in the literature, while the Markov Chain Monte Carlo (MCMC) framework can also identify the influence of the experimental boundary conditions and reflects it in the inferred flow-related parameters. The mechanisms behind the increase in flame spread rate with airflow speed and temperature are also analysed. Flow speed accelerates flame spread by increasing flame temperature, reducing flame stand-off distance, enhancing surface heat flux and mass loss, and extending the preheating zone. In contrast, the flow temperature has a milder effect, primarily by increasing flame temperature and reducing re-radiative heat loss from the solid surface. This study demonstrates the potential of combining scaling models with Bayesian inference to model complex transient flame spread phenomena.</p>}},
author = {{Ma, Yuxuan and Huang, Yajun and Chen, Yuhang and Madsen, Dan and Wahlqvist, Jonathan and Ren, Fangsi and Nakaya, Shinji and Tsue, Mitsuhiro and van Hees, Patrick and Hu, Longhua}},
issn = {{0010-2180}},
keywords = {{Concurrent flame spread; Data-driven approach; Flow temperature; Scale model; Solid combustion}},
language = {{eng}},
publisher = {{Elsevier}},
series = {{Combustion and Flame}},
title = {{Modelling transient flame spread over thick PMMA slabs in high-temperature concurrent airflow using a Bayesian-calibrated simplified scale model}},
url = {{http://dx.doi.org/10.1016/j.combustflame.2026.114929}},
doi = {{10.1016/j.combustflame.2026.114929}},
volume = {{287}},
year = {{2026}},
}