Thermal decomposition models for wood: Comparing parallel and sequential reaction schemes for fire modeling applications

Understanding the thermal decomposition of lignocellulosic materials is critical for modeling wildland and wildland-urban interface (WUI) fires, which have become increasingly destructive worldwide. The thermal decomposition of Douglas fir and red oak wood, representing softwood and hardwood species, respectively, is investigated to assess the impact of reaction-scheme choice on semi-global kinetic models used in computational fire solvers. Thermogravimetric analysis experiments are conducted at multiple heating rates, and the effects of sample preparation and crucible configuration on decomposition behavior are evaluated to establish appropriate experimental protocols; coarse particle samples with pierced-lid crucibles are recommended as a practical compromise between minimizing thermal lag and avoiding altered decomposition behavior from aggressive processing. Three parallel and three sequential reaction models of similar complexity are formulated within the Fire Dynamics Simulator framework, calibrated using multi-rate optimization, and evaluated against an independent temperature program designed to test predictive performance outside the optimized conditions. The five-reaction sequential model provides the best training accuracy for both species, with errors approaching the experimental noise floor, while requiring no compositional analysis inputs. When evaluated against the independent prediction dataset, the sequential models consistently outperformed the parallel models, with the five-reaction sequential model achieving the lowest prediction error for both species. None of the parallel-scheme components yield transferable kinetics across wood species; however, the closest cross-species agreement is found when single first-order reactions represent decomposition to char, which also extrapolate comparably to more complex formulations under unexplored heating scenarios. Thermogravimetric analysis of isolated hemicellulose and lignin fractions confirmed fundamentally different thermal decomposition behavior between species, corroborating the non-transferability of component kinetics.