Methods for Evaluating the Contribution of Exposed Timber to Post-Flashover Enclosure Fire


Full Title - Engineering Methods for Evaluating the Contribution of Exposed Timber to Post-Flashover Enclosure Fires

 By Colleen Wade, Michael Spearpoint and Charles Fleischmann

Introduction

Around the world, there is increasing demand to use more timber in buildings, both for architectural reasons and to promote sustainability (Barber 2015). Examples include the use of cross-laminated timber (CLT) and similar engineered timber products.

CLT is a multi-layered wood panel system that typically has between three and seven layers, with each layer comprising wood boards assembled and adhered perpendicular to each other.

Since timber is a combustible material, if directly exposed to the fire, it can provide additional fuel that changes fire dynamics and the overall severity of an enclosure fire. These changes can be due to large areas of exposed timber or a failure of protective linings such as gypsum plasterboard. In the case of CLT, new layers of virgin wood may be exposed due to debonding of lamella if the adhesive does not maintain strength when its temperature increases (e.g., some PU adhesives), sometimes leading to a fire that does not burn out (Brandon and Östman, 2016).

The charring behaviour of timber in real fire conditions is of great interest in being able to predict the structural performance of timber elements. This requires methods for predicting char rates under a wide range of realistic conditions and not just for standard fire resistance tests.

The performance of enclosures with exposed CLT is an area of active research, with a range of engineering approaches that vary in degree of complexity:

  1. Very simple methods, such as including all the exposed timber surfaces as part of the fire load calculation when using the EC1 parametric equations (CEN 2002). This can be quite conservative.
  2. Reasonably simple engineering methods, such as that proposed by Barber, et al. (2016) or Brandon (2018), also based on the EC1 parametric fire equations. These methods use an iterative procedure to estimate the char depth beneath the timber surface while adjusting the fire load energy density (FLED) at each iteration to include a contribution from the charred material. The methods do not consider the effects of debonding of the base layer, and the limitations of the parametric fire equations still apply (e.g., valid ranges for fire energy density, floor area and opening factor). The calculations can be carried out using spreadsheets.
  3. Simple engineering method incorporated within a zone-type computer model, a somewhat similar approach to method 2, developed by Wade, et al. (2018). This provides additional flexibility for the user, beyond a simple spreadsheet method, since it is possible to simulate fire development within the enclosure based on the underlying mass and energy conservation equations than by using parametric time-temperature correlations.
  4. Advanced methods, including pyrolysis models, more-advanced approaches that are often quite complex and inevitably involve numerical methods and computer models. Improved methods for treating thermal properties of timber have been proposed (Hopkin, 2012) and there are a number of different pyrolysis models, ranging from correlations between the imposed heat flux and resultant charring rate, to much more-detailed kinetic multi-component models that address the various reactions that describe the decomposition of the wood constituents. These approaches tend to be under continuing development and require extra caution, since some of the models may not have the necessary accuracy or conservatism to be considered a practical method for engineering design.

Prediction of Charring in a Zone-Type Model

The zone-type model B-RISK (Wade, et al., 2016) has been adapted to simulate enclosure fire dynamics with exposed timber surfaces under post-flashover conditions. Any initial design fire can be used with a specified FLED value. At flashover (defined in the model as when the average upper layer gas temperature reaches 500ºC), the total heat release rate (HRR) is assumed to be some multiple of the ventilation-controlled rate for the enclosure. The multiplying factor can be thought of as a global equivalence (GE) ratio that is applied in the model.

Experiments on timber enclosures have shown that a large amount of external burning can occur. For example, Hakkarainen (2002) reported that a larger part of the pyrolysis gases burned outside the enclosure opening where exposed timber surfaces were involved, compared to when the timber surfaces were protected by a non-combustible material. A GE factor of 1 results in no external burning, but a longer fire duration (i.e., stoichiometric burning), while a GE factor of 2 results in an equal amount of burning inside and outside the opening with a corresponding shorter fire duration.

For an enclosure with cellulosic contents and non-contributing surfaces, a GE factor of 1.3 (with some external burning) represents ventilation-controlled burning of wood cribs (Babrauskas, 2016). With or without exposed timber surfaces, the peak temperatures reached inside the enclosure are expected to be somewhat similar. A one-dimensional finite difference heat transfer calculation for the walls and ceiling is used to calculate the depth of char (assumed to be the 300ºC isotherm) beneath the timber surfaces, with the total quantity of charred material contributed by the surfaces added to the overall fuel load at each time step as the fire progresses.

A CLT enclosure experiment conducted at Carleton University by McGregor (2013) illustrates application of this model. The enclosure was 4.5 × 3.5 × 2.5 m high with a single door-type opening 1.07 × 2.0 m high. The fuel load was bedroom furniture that gave a FLED of 550 MJ/m2. Figure 1 shows the predicted total HRR and upper layer gas temperature for the enclosure in which the CLT was fully protected; Figure 2 is for the fully exposed CLT enclosure.

The simulation has assumed that 77% of burning (1/1.3) occurred inside the enclosure in the protected case, but equal burning inside and outside the opening for the fully exposed CLT compartment. At around 39 minutes in the latter case, debonding of some of the CLT lamella was observed; this is apparent in Figure 2, where the HRR can be seen to increase after this time. The model simulation did not account for debonding, so a divergence between the measured and predicted quantities occurs after 40 minutes.

McGregor estimated charring rates using thermocouples embedded into the structure at different depths below the exposed surface. For thermocouples at a depth of 24 mm, the recorded time to reach 300ºC ranged from 30 to 39 min. The predicted char depth in the wall surface shown in Figure 3 falls within this range. Considering the position of the 200ºC isotherm in Figure 3, it can be seen that it is within about 7 minutes of the observed time of debonding where 200ºC has been proposed as a debonding criterion for PU adhesives (Aguanno 2013) and 35 mm was the thickness of the lamella. While debonding could be accounted for in the model by adjusting boundary conditions in the finite difference calculations, it is prudent from a design perspective to prevent the phenomenon by using temperature-resistant adhesives in the manufacture of the CLT.

Conclusion and Further Research

As the demand for greater use of timber in the construction of buildings increases, there is a corresponding need for design methods and models that can accurately account for the effect that timber may have on the behaviour of an enclosure fire. It is necessary to consider charring behaviour in realistic fire scenarios, not only standard fire-resistance tests, to create confidence that exposed timber structures can resist burnout and not collapse. In the case of CLT structures, additional considerations such as potential debonding effects may also be important.

The work discussed here is the subject of ongoing research at BRANZ and the University of Canterbury, New Zealand, where different wood pyrolysis sub-models are being investigated as part of the development of B-RISK. This includes an integral sub-model for the wood burning rate responding to the incident heat flux, as well as a kinetic sub-model for wood pyrolysis.

Colleen Wade is with BRANZ Ltd, New Zealand. Charles Fleischmann is with the Department of Civil and Natural Resources Engineering, University of Canterbury, New Zealand. Michael Spearpoint is with Olsson Fire & Risk, UK.

References

Aguanno, M. 2013. “Fire Resistance Tests on Cross-Laminated Timber Floor Panels: An Experimental and Numerical Analysis.” Master of Applied Science Thesis, Ottawa, Canada: Carleton University.

Babrauskas, V. 2016. “Heat Release Rates.” In SFPE Handbook of Fire Protection Engineering, edited by M.J. Hurley, 5th ed.New York: Springer.

Barber, D. 2015. “Tall Timber Buildings — What’s next in Fire Safety?” Fire Technology 51 (6):1,279–1,284. https://doi.org/10.1007/s10694-015-0497-7.

Barber, D., R. Crielaard, and X. Li. 2016. “Towards Fire Safety Design of Exposed Timber in Tall Timber Buildings.” In Proceedings of WCTE 2016 World Conference on Timber Engineering. Vienna, Austria.

Brandon, D. 2018. “Fire Safety Challenges of Tall Wood Buildings — Phase 2: Task 4 Engineering Methods.” FRPF-2018-04. Quincy, MA: Fire Protection Research Foundation.

Brandon, D., and B. Östman. 2016. “Fire Safety Challenges of Tall Wood Buildings — Phase 2: Task 1 — Literature Review.” FRPF-2016-22. Quincy, MA: Fire Protection Research Foundation.

CEN. 2002. “Eurocode 1: Actions on Structures — Part 1-2: General Actions — Actions on Structures Exposed to Fire, Annex B.” European Committee for Standardization.

Hakkarainen, T. 2002. “Post-Flashover Fires in Light and Heavy Timber Construction Compartments.” Journal of Fire Sciences 20(2):133–75. https://doi.org/10.1177/0734904102020002074.

Hopkins  , D. 2012. “Predicting the Thermal Response of Timber Structures in Natural Fires Using Computational ‘Heat of Hydration’ Principles.” Fire and Materials 37 (4): 311–27. https://doi.org/10.1002/fam.2133.

McGregor, C.J. 2013. “Contribution of Cross Laminated Timber Panels to Room Fires.” Master of Applied Science Thesis, Ottawa, Canada: Carleton University.

Wade, C.A., G.B. Baker, K. Frank, R Harrison, and M.J. Spearpoint. 2016. “B-RISK 2016 User Guide and Technical Manual.” Study Report SR364. Porirua, New Zealand: BRANZ.

Wade, C.A., M.J. Spearpoint, C.M. Fleischmann, G.B. Baker, and A.K. Abu. 2018. “Predicting the Fire Dynamics of Exposed Timber Surfaces in Compartments Using a Two-Zone Model.” Fire Technology 54(4):893–920. https://doi.org/10.1007/s10694-018-0714-2.