Computational Fluid Dynamics analysis of Cross-Laminated Timber compartment fire behaviour based on chemical reactions

Abstract

In recent years, the building industry is experiencing a shift towards timber buildings driven by the need for a sustainable building industry. A popular building typology consists of Cross-Laminated Timber (CLT) panels. There is an ongoing trend for leaving CLT panels exposed, meaning without protective fire-resistant cladding. This enhances the aesthetical appeal of the building but raises questions regarding fire safety.

This thesis attempts to model the fire dynamics within a CLT compartment with Fire Dynamics Simulator (FDS), which is a CFD software. Hereby, special attention is given to modelling the pyrolysis and combustion reaction for the fire exposed CLT panels. The ability to model the burning behaviour of wood in FDS is a big step forward to be able to move from large-scale experiments to computational models.

This thesis validates the FDS model by comparing the simulated results with the experimental series by Olivier (2019). This series assessed small-scale cubical compartments with 0,5 m sides and a varying amount and orientation of CLT.

Extensive analyses of the simulated results show that the energy release due to the burning CLT is overestimated by approximately 50%. The simulated burning rates range from 1,15 – 1,33 mm/min, while experimentally measured burning rates are approximately 0,60 mm/min which is more commonly assumed in practice. The fire growth phase is approximated well, however validation of the decay phase is poor.

The simulated gas temperatures vary significantly more locally compared to the experimental data. Simulated gas temperatures in the middle of the compartment are underestimated by approximately 30%. On the other hand, gas temperatures near the burning CLT surfaces are approximated well. It is therefore concluded that the radiation from the burning CLT is poorly approximated and has a further reach in real-life experiments.

Extensive sensitivity analyses have shown that the simulated results in the FDS model are highly sensitive to the mesh cell size and the heat of combustion of wood. It is concluded that the three-dimensional mesh should consist of 20 mm cubical cells positioned within a proximity of 10 – 20 cm from the CLT surfaces. In the remainder of the computational domain 100 mm cubical cells should be used to minimise the number of cells in the simulation while maintaining simulation accuracy.

Conclusively, this thesis provides significant added value by enhancing the knowledge within the CFD modelling research field for FSE. The presented FDS model has potential for serving as a tool to roughly predict the fire dynamics within a CLT compartment. However, further optimisation of the model is required before practical applications in daily engineering can be justified.