The Role of Chemistry and Physics in the Charring of Timber in Realistic Fires
By Franz Richter, Panagiotis Kotsovinos, and Guillermo Rein
Mass timber is an emerging construction material for modern tall buildings since it is sustainable and leads to a fast and low-cost construction process in comparison with traditional materials.1 Around the world, skyscrapers are being proposed of timber up to 70 stories high (350 m). The construction of buildings made of timber — tall ones in particular — naturally poses questions regarding fire safety due to timber being perceived as having a low resistance to fire. However, every construction material has its risks in a fire and knowledge-based design can reduce such risks to a reasonable level.
When a structural element made of timber is heated, it develops a black char layer that slowly propagates through the element. This char layer increases thermal resistance, due to lower thermal diffusivity between the fire and timber. In other words, it slows down the degradation of timber by slowing down the heating process. When assessing the structural fire resistance of timber elements, it is typically assumed that the char layer does not have any residual strength, while the layer of intact timber underneath maintains all of its strength. Therefore, the strength of a timber section can be calculated when the rate of charring has been determined. This method is called the reduced cross-section method.2
Current design standards, such as the Eurocode, assume a universal constant charring rate2 based on the duration of the exposure to a standard fire. However, the standard fire does not represent a real fire.3 It only considers flashover conditions where the room burns uniformly and at the same temperature throughout (see Figure 1, left). A cooling phase is not included.
These assumptions only hold for small compartments, while in large open-plan compartments, such as modern office floors, fire tends to travel. Parametric, localised and travelling fires4 capture this behaviour, or part of it. Figure 1 illustrates these differences between the standard fire and a real fire, using travelling fires as an example of real fires.
A travelling fire grows to a certain size and then moves through the area. Ahead of the flames, the structure experiences pre-heating at relatively low temperatures (far-field) and is only exposed to high temperatures when the flames arrive (near-field). After the flame front passes, the structure receives far-field heating again. Figure 1 (center) illustrates a typical temperature-time curve for a travelling fire that results in a non-uniform thermal environment and can lead to longer burning durations compared with post-flashover fires. As a result, travelling fires can have a more-detrimental thermal impact on a structure.4
There is limited guidance about timber charring under realistic fires and results from standard fire tests are deemed inappropriate.5 This article outlines some of the physics and chemistry of charring to build a scientific model that can predict the charring of timber under realistic fire conditions.
Figure 1. Illustration of the difference between the standard and a travelling fire in terms of fire dynamics and temperatures. Note: The timescale for the standard fire and the travelling fire is indicative since only the latter uses real time. This figure is under CC BY, 2018.
Modelling Across Scales
Charring is part of the process of pyrolysis, which is the simultaneous chemical and physical change of a solid that provides the gaseous fuel for the flame above its surface.6 It is a complex interplay between heat and mass transfer as well as chemistry, as illustrated in Figure 2. In chemistry, this refers to the breakdown of the large polymer structure of timber into smaller molecules of gas and char.
Two other processes — drying and smouldering — also affect the dynamics of pyrolysis. Ahead of the pyrolysis front (see Figure 3), heat is absorbed to evaporate moisture (drying front). Moisture content, therefore, hinders charring. Behind the pyrolysis front, oxygen may directly oxidise (burn) the carbon-rich char. This smouldering process releases heat and consumes the char. Smouldering thus enhances charring, but could be a self-sustained process (no self-extinguishment). Despite the importance of smouldering in slowly reducing the fire resistance (char layer thickness) of timber, it is not considered in current models.
It is conventionally assumed that physical processes dominate charring, and that chemistry is of minor importance. However, a simple dimensional analysis reveals that both are important. Before a char layer forms, heat transfer controls the heating of the sample. Once sufficiently high temperatures are reached, a pyrolysis zone (thin char layer) develops. This development is controlled by chemistry. Heat transfer (across the char layer) would dominate once a thick char layer forms. An accurate model of charring has to incorporate both heat transfer (physics) and chemistry. In fact, smouldering makes the char layer thinner, which favours the chemistry-controlled process of charring. Unfortunately, the chemistry of charring (pyrolysis and oxidation) is not commonly considered when assessing the structural fire resistance of timber.
Figure 2. Overview of the different physical and chemical processes which take place or affect the charring of timber. This figure is after N Bal 2009 and under CC BY, 2018.
Figure 3. Sketch of the different stages of burning timber showing the three fronts (drying, pyrolysis and smouldering) (image of burning timber from Reszka, 2008 ). This figure is under CC BY, 2018.
To gain a deeper understanding of the charring of timber, we studied each controlling process in isolation. As a result, we adopted the approach of going across scales. At the microscale (mg samples), we studied and deduced the chemistry of charring using a TGA. At the mescoscale (kg samples), we studied the heat transfer of charring using a cone calorimeter and equivalent setups. At the macroscale (structural components), we can, through upscaling of the model and the reduced cross-section method, predict the loss of strength of timber in several heating scenarios.
Since the chemistry has not been studied in detail, we had to develop our own model of the kinetics. We modelled timber as a linear superposition of its main components, which are present in different amounts in all species of wood, to capture the behaviour of different species. We used literature about mass loss rate data from mg samples in a small furnace (TGA) for validation. In contrast to previous research, our kinetic model predicts accurately the mass loss rates of different timber species under heating conditions and oxygen concentrations spanning multiple experiments (over 80).
We also validated the kinetic model extensively with blind predictions. We then implemented this kinetic model into a mesoscale heat and mass transfer model to validate the complete charring model against the state-of-the-art experiments of Kashiwagi and Ohlemiller.8
Figure 4 shows that our preliminary predictions about the mesocsale of both the char depth and the charring rate are in good agreement with the experiments. The model captures the increased speed of charring with oxygen concentration well, although it under-predicts the charring rates before 200 seconds. The research is ongoing and we are still analysing the reasons for these under-predictions. We also have not included drying in the model yet, since that model is still being developed.
Mindeguia, et al.,9 recently developed a charring model for timber — pyrolysis only — that was able to predict charring rates under standard and parametric fires (different heat conditions). Like our model, theirs included both chemistry and heat transfer, but only mesoscale experiments were adopted to build the chemistry and heat transfer model. Nevertheless, our work and the work of Mindeguia, et al., demonstrates that generalized charring models could improve the accuracy of charring predictions over conventionally used tools — such as the constant charring rate — since they capture the effect of heating conditions, oxygen concentrations, and timber species. In fact, the Eurocode charring rate is non-conservative in our heating conditions, as shown in Figure 4.
Our works also illustrates that going across scales is a promising path for developing such charring models. It leads to improving predictions and new insights on the underlying physics and chemistry.
Figure 4. Comparison between the measurements (symbols)  and predictions (lines) at the mesoscale at different oxygen concentration at 40 kW/m2. The predictions of the Eurocode  are illustrated by the dotted line. The char depths were calculated using the conventional 300 °C isotherm. Moisture content of the samples is estimated to be around 9 %.
Charring of timber is influenced by a complex interplay between heat and mass transfer, as well as chemistry. The charring rate, therefore, could vary depending on the heating condition, oxygen concentration and timber species. Conventional charring rates and furnace tests may therefore not be representative of real fire conditions. By going across scales, we were able to study each main physical and chemical process in isolation and use this knowledge to build a combined generalized model of charring.
Such generalized models show great potential to predict charring under different heating conditions, oxygen concentrations and of different timber species. Since these models are based on first principles, they can predict charring rates under any condition, once validated. It is believed that with these tools at hand, engineers will be able to estimate the fire resistance of timber structures more accurately. Current and future work involves working with fire safety engineers to model the response of structural timber elements at the macroscale.
Franz Richter and Guillermo Rein are with the Department of Mechanical Engineering, Imperial College London, UK. Panagiotis Kotsovinos is with ARUP, UK.
1D. Barber, “Tall Timber Buildings: What’s Next in Fire Safety ?” Fire Technol., pp. 1–6, 2015, doi: 10.1007/s10694-015-0497-7.
2EC5, “Eurocode 5: Design of Timber Structures. ENV 1995-1-2: General Rules-Structural Fire Design,” European Committee for Standardization, Brussels, Belgium, 1994.
3D. Drysdale, An Introduction to Fire Dynamics. Chichester, UK: John Wiley & Sons, Ltd., 2011.
4E. Rackauskaite, P. Kotsovinos, A. Jeffers and G. Rein, “Structural analysis of multi-storey steel frames exposed to travelling fires and traditional design fires,” Eng. Struct., pp. 271–287, 2017, doi: 10.1016/j.engstruct.2017.06.055.
5A. I. Bartlett, R. M. Hadden, L. A. Bisby and A. Law, “Analysis of cross-laminated timber upon exposure to non-standard heating conditions,” in 14th International Conference and Exhibition on Fire and Materials, 2015.
6G. Rein, “Smoldering Combustion,” in SFPE Handbook of Fire Protection Engineering, New York, NY: Springer New York, 2016, pp. 581–603.
7P. Reszka, “In-Depth Temperature Profiles in Pyrolyzing Wood,” University of Edinburgh, 2008.
8T. Kashiwagi, T. J. Ohlemiller and K. Werner, “Effects of external radiant flux and ambient oxygen concentration on nonflaming gasification rates and evolved products of white pine,” Combust. Flame, pp. 331–345, 1987, doi: 10.1016/0010-2180(87)90125-8.
9J. Mindeguia, G. Cueff, V. Dréan, G. Auguin and J. Mindeguia, “Simulation of charring depth of timber structures when exposed to non-standard fire curves,” J. Struct. Fire Eng., 2017, doi: 10.1108/JSFE-01-2017-0011.