Natural Full-Scale Fire Test on a 3 Storey XLam Timber Building

Andrea FRANGI

Institute of Structural Engineering, ETH Zurich, Switzerland Giovanna BOCHICCHIO and Ario CECCOTTI

CNR-IVALSA, San Michele all’Adige, Italy Marco Pio LAURIOLA

Timber Engineering, Florence, Italy Summary An extensive research project, called SOFIE and sponsored by the Italian Province of Trento, on the structural behaviour of timber buildings made of prefabricated cross-laminated solid timber panels is currently carried out at CNR-IVALSA in Italy. The research project aims at supplying documentation and information on the use of XLam timber panels as structural elements, in order to increase its use in particular for residential multi-storey buildings. Following the shaking table test of a full-scale 3 storey XLam timber building carried out in Tsukuba, Japan in June and July 2006, a natural fire test on the same building has been recently carried out at the Building Research Institute in Tsukuba. The paper presents the main results of the natural full-scale fire test.

1. Introduction Construction industry has drawn attention to new construction methods based on building systems. Wood particularly fits the requirements of the idea “built with system” because of its easy manufacturing possibility. An example of a successful construction system are large prefabricated cross-laminated (XLam) solid timber panels for load bearing wall and floor assemblies. XLam timber panels have become increasingly popular not only for residential but also for office, retail and industrial buildings in particular in Austria and Italy. XLam timber panels are produced in the factory according to the structural drawings taking into account door/window/stair openings, transported from the factory directly to the site and then joined on site with simple and rapid connections. Insulation and facade elements can be connected easily to the timber panels. Unlike light timber frame constructions, where single timber studs are responsible for the transfer of the vertical loads, the use of large solid timber panels allow the transfer of high vertical loads and guarantee a high building stiffness and robustness. Other main advantages of this new building system are an excellent thermal insulation and air tightness. The use of large solid timber panels is also favourable in case of fire, as the risk of fire spread through void cavities is reduced in comparison to light timber frame constructions. However, large solid timber panels increase the fire load in the room.

Fig. 1 Example of cross-laminated solid timber panel [1] XLam timber panels are produced from industrially dried spruce boards which are stacked crosswise and glued together over their entire surface. Depending on the purpose and static requirement, XLam timber panels are available with 3, 5, 7 or more board layers. The width of the single boards usually varies between 80 and 240mm, the thickness between 10 and 35mm. The cross-section is symmetrical. Each single board is visually or machine graded and can be jointed

using finger joints. The size and form of cross-laminated solid timber panels is limited by production, transportation and erection possibilities. Figure 1 shows an example of XLam timber panel [1]. An extensive research project, called SOFIE, on the structural behaviour of XLam timber buildings is currently carried out at CNR-IVALSA in Italy. The research project aims at supplying documentation and information on the use of XLam timber panels as structural elements, in order to increase its use in particular for residential multi-storey buildings. Prescriptive fire regulations often still restrict worldwide the use of timber. However, in the last couple of years, many countries have started to introduce performance based fire regulations or liberalized the use of timber for buildings. These regulations open the way for new applications. In taking advantage of the new possibilities it is essential to verify that fire safety of timber buildings is not lower than of buildings made of other materials. One area of research of the SOFIE project focuses on the experimental and numerical analysis of the fire performance of XLam timber buildings. A large number of fire tests have been performed on unloaded XLam timber panels at CNR-IVALSA using ISO-fire exposure. The testing program permits to study the influence of different parameters on the fire performance of XLam timber. Further in order to analyze the global behaviour of XLam timber structures a natural full-scale fire test on a 3 storey XLam timber building has been carried out at the Building Research Institute in Tsukuba, Japan. The paper presents the main results of this natural full-scale fire test.

2. Description of the test specimen The natural full-scale fire test was performed in a 3 story timber building with an area of about 7x7m and a height of about 10 m. The building main structure consisted of 4 outer 85 mm thick XLam timber walls and an inner 85 mm thick XLam timber wall placed parallel to the E-W direction (see figure 2).

Fig. 2 Geometry of the 3-storey timber building used for the natural full-scale fire test

The floors (incl. roof) consisted of 142 mm thick XLam timber panels connected to the walls by means of steel brackets and screws. South and West facades of the test building were covered with insulating and finishing materials while on the North and East sides of the building the XLam timber walls remained uncovered. The roof was entirely covered with silica calcium boards.

Wall 3 Plaster 10 mm

Fire proof gypsum wallboard 12 mm

Mineral wool 27 mm

Standard gypsum wallboard 12 mm

Wood Fibre 120 mm

XLam Panel 85 mm

(1)

(2)

(6)

(3)

(4)

(5)

Outside

Inside

Wall 4 Fire proof gypsum wallboard 12 mm

Fire proof gypsum wallboard 12 mm

Mineral wool 27 mm

Mineral wool 27 mm

Standard gypsum wallboard 12 mm

Standard gypsum wallboard 12 mm

XLam Panel 85 mm

(1)

(2)

(6)

(7)

(3)

(4)

(5)

Fire Room

Room B

Wall 5 Standard gypsum wallboard 12 mm

Mineral wool 27 mm

XLam Panel 142 mm

Standard gypsum wallboard 12 mm

Mineral wool 27 mm

(1)

(2)

(3)

(4)

(5)

Fire Room

Room A

Floor

and ceiling

of room

fire

Wood flooring 20 mm

Sand 60 mm

XLam Panel 142 mm

Fireproof gypsum wallboard 12 mm

Mineral wool 27 mm

Concrete topping 50 mm

Polyethylene sheet

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Fig. 3 Cross-section of the walls and floors of the fire room (see figure 2) The fire room with dimensions of 3.34x3.34x2.95m was located on the first floor and presented two window openings with the dimensions of about 1.0x1.0m and a door with the net dimensions of 0.9x2.1m. During the fire test the door certified with fire resistance of 60 minutes remained closed. At the beginning of the fire test, each window made of standard double glass was opened at one quarter of the width (i.e. the surface opened was about 0.26x0.94m, see fig. 4 left). The window openings above the fire room consisted of double reinforced glass and were closed at the beginning of the fire test. All other window openings were sealed with gypsum plasterboards. Figure 3 shows the cross-section of the walls and floors of the fire room (see also figure 2). The following materials

used in the construction were supplied according to the following Japanese standards: • Standard gypsum plasterboard : JIS A 6901 GB-R 12.5mm • Fire proof gypsum plasterboard: JIS A 6901 GB-F 12.5mm • Mineral wool (rock wool) 27 mm: JIS A 9521 25mm • Concrete topping: JIS A 5308 Ready-mixed concrete 18N/mm2

3. Test measurements During the tests, the temperature at more than 100 locations was measured and recorded with thermocouples as described in EN 1363-1 [2]. The thermocouples were located on the room surface as well as within the wall, ceiling and floor elements. At the windows of the room above the fire room the glass temperature was measured on the internal surface of the window. The temperature in the fire room was measured in the middle of the room with five thermocouples placed at distances of 0.1, 0.74, 1.48, 2.22 and 2.85m from the floor. Further eight flat thermocouples (two for each wall) were placed close to the floor and the ceiling at a distance of 0.1m from the surface of the gypsum plasterboards. Two air pressure transducers were placed in front of each window of the fire room, one in the top and one in the bottom of the opened side. The air pressure transducers permitted the measurement of the pressure of the incoming and outgoing gas; in the same position additional thermocouples were placed. In this way for each window the temperature and the pressure of the gas flux passing through was obtained in two different positions. The gas concentrations of oxygen, CO and CO2 were recorded with a gas measuring instrument placed in the fire room at 2.20m height over the floor. Two heat flux measuring instruments were placed outside each window of the fire room at a distance of 3m from the surface of the glass. Two infrared thermo cameras were placed outside the building in front of the South and West side wall of the fire room. The fire test was documented by 9 video cameras and photos, as well as test protocols.

4. Fire load and fire ignition In building fires, the contents (movable fire load) as well as combustible construction materials contribute to the total fire load. In a typical residential room, the contents consist mostly of beds, tables, cabinets, electronic apparatus, etc. The fire room was equipped with two typical mattresses made of polyurethane. As additional movable fire load, several wooden cribs were located in the module (see fig. 4 right). The wooden cribs were ignited with common fire starts.

Fig. 4 Detail of the window opening of the fire room (left) and fire load made of two mattresses and several wooden cribs (right) Table 1 gives the movable fire load as well as the additional fire load due to the combustible construction materials for the fire test. For the calculation of the fire load, a net heat of combustion Hu of 17.5 MJ/kg for wood as suggested in EN 1991-1-2 [3] was assumed. For the calculation of the additional fire load due to combustible construction materials, only the combustible wooden flooring has been considered with a participation factor of 0.5, i.e. it was assumed that only 50% of

the wooden flooring contributed to the fire load. However, after the gypsum plasterboards and the rock wool insulation began falling off, the X-lam timber panels also contributed to the fire load. The total fire load density (calculated over the floor area) for the fire room was approximately 790 MJ/m2. Table E.4 in EN 1991-1-2 gives for rooms of residential buildings an average fire load density of 780 MJ/m2. Material Participation

factor Total weight

[kg] Hu

[MJ/kg] Qfi

[MJ] Wooden flooring 0.5 94.5 17.5 827 Polyurethane mattresses 1.0 33.0 23.0 759 Additional wooden cribs 1.0 378 17.5 6613 Fire starts for fire ignition 1.0 2.5 40.0 100

Total fire load [MJ] 8299 Floor area [m2] 10.5

Fire load density calculated over 10.5 m2 floor area [MJ/m2] 790

Table 1 Fire load for the natural full-scale fire test

5. Test results

5.1 Visual observations After fire ignition, fire grew slowly due to low ventilation in the fire room (at the beginning of the fire test, each window was opened only at one quarter of the width). The glass of the window on the South side started falling down after about 20 minutes, on the West side after about 30 min. (see fig. 5 left). The complete collapse of both windows occurred after about 36 minutes. As the failure of the glasses of the windows progressed, the intensity of the fire inside the room and the external burning out the windows became more severe. Flashover, i.e. where all unprotected combustible material burns, occurred after about 40 minutes (see fig. 5 right). After 53 minutes the door of the fire room fell off, leading to smoke penetration into the adjacent room B. After about 55 minutes the fire intensity started declining and the fire was controlled and manually extinguished by fire fighting actions as planed after 60 minutes. The window openings above the fire room did not fail and thus no fire spread into the upper level was observed.

Fig. 5 Fire development after about 32 minutes (left) and after 40 minutes (right) after fire ignition In two different series of full scale fire tests recently performed by ETH [4] and VTT [5] on timber compartments, flashover occurred between 4 and 7 minutes after the fire ignition. In these cases, at the beginning of the fire test, the windows were completely opened in order to guarantee adequate ventilation in the fire room.

5.2 Temperatures Figure 6 shows the room temperatures measured in the middle of the fire room at distances of 0.1, 0.74, 1.48, 2.22 and 2.85m from the floor (see fig. 6 right). A non uniform temperature distribution over the height of the fire room was measured during the first 35 minutes, with the highest temperature measured close to the ceiling and the lowest temperature close to the floor. After about 35 minutes all temperatures rose within a few minutes to flashover confirming the increased fire intensity observed during the test.

2.85m

2.22m

1.48m

0.74m

0.1m

Fire room

Fig. 6 Temperatures measured in the middle of the fire room at distances of 0.1, 0.74, 1.48, 2.22 and 2.85m from the floor

Fig. 7 Temperatures measured at the interface of the different layers of the North side wall (see figure 2 for the position of the wall) Figure 7 and 8 show the temperatures measured at the interface of the different layers of the North side wall and the South side wall (see fig. 7 right and 8 right for the position of the thermocouples). It can be seen that the temperature behind the two layers of gypsum plasterboard grew very rapidly after about 47 minutes for the North side wall and after about 50 minutes for the South side wall.

Assuming a critical temperature of about 600°C [6], it can be assumed that the two layers of gypsum plasterboard failed after about 57 and 53 minutes, respectively. During the last 10 minutes of the fire test it was observed that the gypsum plasterboards partially fell off confirming the temperature measurements. Based on the rapid increase of the temperature measured behind the rock wool insulation it can be assumed that the rock wool insulation fell off quasi immediately after being exposed directly to fire, i.e. after falling off of the gypsum plasterboards. This assumption is confirmed by visual observations after the fire test.

Fig. 8 Temperatures measured at the interface of the different layers of the South side wall (see figure 2 for the position of the wall) Figure 9 shows the temperatures measured at the interface of the different layers of the ceiling of the fire room (see fig. 9 right for the position of the thermocouples). On the ceiling only a layer of gypsum plasterboard was used. It can be seen that the temperature behind the gypsum plasterboard grew very rapidly after about 35 minutes. Based on the temperature measured, failure (i.e. fall off) of the gypsum plasterboard probably occurred after about 40 minutes, while the rock wool insulation failed after about 45 minutes.

Fig. 9 Temperatures measured at the interface of the different layers of the ceiling of the fire room

5.3 Damages After the fire test it was observed that the gypsum plasterboards in the fire room had completely fallen off confirming the visual observations during the fire test and the temperatures measured in the wall and floor assemblies. On the other side, the rock wool insulation was partially still in place, thus the timber surface was partially protected from charring. The charring depth of the wall and floor assemblies was measured with a mesh of 300mm. Figure 10 shows, for example, the measured charring depths for the East side wall. It can be seen that the charring depth varied between 5 and 10 mm. Similar results have been also measured for the other walls and the ceiling.

Fig. 10 Charring depth measured on the East side wall after the fire test

6. Conclusions A full-scale test on a 3-storey building made of Xlam solid timber panels was performed under natural fire conditions to check the global performance and find possible weaknesses of the timber structure. The Xlam solid timber panels were protected by one or two layers of non combustible gypsum plasterboards. The test has confirmed that with pure structural measures it is possible to limit the fire spread to one room even for timber structures. In the room above the fire compartment no elevated temperatures were measured and no smoke was observed. Further by protecting the timber structure with gypsum plasterboards the damage of the Xlam solid timber panels was relative small.

7. References [1] EN 1995-2:2004, Eurocode 5: Design of timber structures - Part 2: Bridges, CEN, Brussels,

2004. [2] EN 1363-1:1999, Fire resistance tests - Part 1: General requirements, CEN, Brussels, 1999. [3] EN 1991-1-2:2002, Eurocode 1: Action on structures Part 1-2: General action – Actions on

structures exposed to fire, CEN, Brussels, 2002. [4] Frangi A., Fontana M, “Fire Performance of Timber Structures under Natural Fire

Conditions”, 8th International Symposium on Fire Safety Science (IAFSS), Beijing, China, September 18-23, 2005, Proceedings edited by Hughes Associates, Baltimore, USA, ISBN 0-9545348-0-8, pages 1111-1122.

[5] Hakkarainen T., “Post-Flashover Fires in Light and Heavy Timber Construction Compartments”, Journal of Fire Sciences, Volume 20, March 2002.

[6] Sultan M.A., “A model for predicting heat transfer through non-insulated unloaded steel stud gypsum board wall assemblies exposed to fire”, Fire Technology, Volume 32, No.3, pp 239-259, 1996.