| dc.description.abstract | The building industry has, in recent years, more frequently used different engineered wood products, like cross-laminated timber (CLT), glue-laminated timber and I-joists. The increased popularity is due to the many advantages of building with wood, like the possibility of prefabrication, the low carbon dioxide footprint, the easy handling and mounting of wood, and the aesthetic look.
Conversely, when building with wood, additional combustible mass is introduced into the buildings. Given that these wooden products are exposed (i.e., visually present), they would produce pyrolysis gases when subjected to a fire and thus have an impact on the fire dynamics.
For over a decade, different research groups have conducted compartment fire experiments with exposed CLT. Most of those experiments have been conducted in relatively small compartments with small ventilation openings. Thus, the role of the exposed CLT is better understood in small compartments than in large. Small compartments with exposed CLT could be relevant for certain buildings. However, CLT is used in a variety of different rooms, including open-plan offices, lobbies, canteens, dwellings, kindergartens, schools, etc. In other words, the great variation in the use of CLT, including room size, geometry, and orientation, challenges the current understanding of how exposed CLT affects a fire.
The main focus of this thesis has been to increase the knowledge of the fire behaviour in large compartments with exposed CLT. The methodology consisted of conducting two large-scale (95 m2) compartment fire experiments. In the first experiment, named #FRIC-01, the ceiling was exposed, while in the second experiment, #FRIC-02, both the wall and the ceiling were exposed. The compartment had four open window openings along one wall, which caused a well-ventilated compartment. The fuel load density was representative of an office building and was represented by a continuous wood crib on the floor. These experiments aimed to better understand how two different configurations of CLT affect the fire dynamics, including fire spread inside the compartment, external flames, charring rate of CLT, decay phase, self-extinguishment of flames and delamination.
In both experiments, the ignition of the CLT ceiling triggered a clear change in the fire dynamics, in which flames spread under the ceiling and caused a strong radiative heat flux to the wall and the wood crib. The increased radiative heat flux effectively preheated the wood crib (and the wall in #FRIC-02) and led to a significantly faster spread across the wood crib than before the CLT was ignited.
In #FRIC-01, a new behaviour was observed, in which the flames in both the ceiling and of the wood crib travelled back and forth three times. As such cycles have not been reported earlier, we have named them flashing waves. Three such waves were observed before the fire was fully developed in the fourth wave. Despite retraction of the flames, the wood crib fire grew larger after each wave, contributing to a significantly faster fire spread rate than before the CLT ceiling ignited. In total, it took 13 minutes from ignition of the ceiling until the fire was fully developed. After a few minutes of intense burning, the flames in the ceiling started to extinguish, and over a period of 11 minutes, all flames in the ceiling were extinguished. This occurred while the wood crib was still burning. No reignition was observed within a total duration of four hours.
In #FRIC-02, the contribution of having CLT exposed in both a wall and the ceiling was clearly seen. From ignition of the CLT ceiling, it took only 91 seconds before the entire compartment was burning. The fire spread was dominated by the rapid flame spread under the ceiling and upper part of the wall first. This caused a strong radiative heat flux to the wood crib and the other parts of the wall. The fire spread rate after ignition of the ceiling was 15 m/min across the ceiling and 11.7 m/min across the wood crib, corresponding to a fire growth rate faster than the ultrafast fire growth rate defined by Eurocode 1 (EN 1991-1-2).
During the most intense burning phase, large external flames emerged mainly out of one window. For some period, the flames covered almost the entire window opening, reached above the facade wall (5.2 m), and extended about 3 m horizontally from the window. The flames effectively reduced the inflow of air through that window, resulting in more air being supplied through the other windows. This imbalance in air supply contributed to large temperature variations throughout the compartment. These non-symmetrical external flames are believed to be mainly due to the wind coming diagonally from behind the corner of the compartment, but also due to the very rapid fire spread. The characteristic behaviour of the external flames was successfully reproduced in a CFD simulation when similar wind conditions as in the experiment were considered. This strengthens the hypothesis that the non-symmetrical external flames were influenced by the wind conditions.
After 10 minutes of intense burning, the gas temperatures started decaying, and also with this CLT configuration (wall and ceiling exposed), the flames of the CLT self-extinguished. About 50 minutes after the start of the decay phase, multiple small flames appeared at the surface of both the CLT wall and ceiling. Within 10 minutes, all combustible surfaces were burning again, corresponding to a second flashover. The temperatures were, for a short period, almost as high as after the first flashover. After that, the fire intensity varied strongly over the next 100 minutes but continued to burn until it was manually extinguished after almost 3 hours. This ongoing fire can be explained by the build-up of the CLT with thick (40 mm) outer layers and thin (20 mm) intermediate layers and the use of a regular PUR adhesive known to cause delamination.
A side topic of this thesis has been to study the charring of I-joists in light timber framed assemblies with combustible insulation. In recent years, there has been increased interest in combining I-joists with new combustible insulation products, like wood fibre and cellulose insulation. However, as there is no available design model for calculating the load-bearing capacity of this combination when exposed to a fire, the outspread of this combination has been limited.
This part of the research was aimed at producing experimental data to better understand the charring of I-joists and the recession rate of combustible insulation when these products are combined. This data could later be used to develop or validate design parameters for combustible insulation.
Through five experiments, a combination of I-joists with different flange sizes and different combustible insulation types (wood fibre, cellulose and phenolic foam) were exposed to the standard fire curve (ISO 834) in a medium-scale furnace. Thermocouples were embedded into and outside of the flanges and used to determine the charring rate of the I-joist and the recession rate for the insulation. After exposure, the final char depth and the remaining cross-section were measured.
The charring rates were compared against calculated values based on the design model for rectangular cross-sections in the current Eurocode 5 (EN 1995-1-2) and the new model for I-joists in the draft of the new Eurocode 5 (prEN 1995-1-2).
Compared to those models, the charring rates were mainly on the conservative side. The charring rates decreased with increasing flange size and were comparable for the flanges of solid wood and laminated veneer lumber (LVL).
Overall, the combustible insulation protected the I-joists well, and the recession rates were lower than values reported for glass wool insulation. The lowest values were obtained by cellulose- and wood fibre insulation. Due to few repetitions, the results must be considered as indicative. Still, the results strongly indicate that biobased, and thus more sustainable, insulation types deserve a great market share in the future.
Altogether, the experimental work in this thesis has contributed to improved knowledge of both CLT and I-joists and could be considered a small but important step towards a more sustainable and fire-safe building sector. | en_US |
