Fire Performance of Laminated Veneer Lumber (LVL) Warren LANE Research Student Warren.Lane@xtra.co.nz Andy BUCHANAN Professor Andy.Buchanan@canterbury.ac.nz Peter MOSS Associate Professor p.moss@civil.canterbury.ac.nz Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand Warren Lane is undertaking post-graduate research in engineering (Fire) after 25 years experience in civil & structural consulting, and government project work. He returned to Canterbury University to learn from those at the forefront of developments in the field, to apply current research in practice. Summary This paper describes the results of an investigation into the fire performance of laminated veneer lumber (LVL), considering both the early fire hazard and the structural fire performance. To determine ignition properties and charring rates, cone calorimeter tests were carried out on blocks of LVL with a number of grain orientations, each for a range of heat flux exposures. Some samples included embedded thermocouples. Charring experiments were also carried out on three LVL beams, approximately 100 x 300mm in cross section and 2 m long, tested in a pilot scale furnace following the ISO834 standard fire curve. Embedded thermocouples were used to record layer temperatures, which were then used to calculate the residual cross section for prediction of structural behaviour in fire conditions. A load-bearing fire resistance test was carried out on a larger LVL beam spanning 4 m in a full size furnace, following the ISO834 standard fire curve. Keywords: Charring, fire performance, ignition, laminated veneer lumber, LVL 1. Introduction Fire is one of the most unpredictable and dangerous accidents that may occur in our environment, especially in residential buildings. Fire development and behaviour and the effects of fire on structural members are very complex because of the large number of variables involved. Ignition is one of the essential fire properties of a material and must always be considered in any assessment of fire hazard. The ignition of wood is more complex than for many other materials, especially because a layer of char is formed. The time to ignition depends on the species, moisture content, inherent variability of wood as a natural material, and the grain orientation of the specimen when exposed to the incident heat flux. In the case of laminated veneer lumber (LVL), the inherent variability of the timber may be a lesser factor because the raw material is carefully selected. Once ignition has occurred, a layer of char is formed as the wood burns. A structural wood member will lose load capacity as the wood is converted to charcoal which has no strength. The thickening char layer protects the remaining wood, resulting in a predictable rate of charring below the surface. The rate of development of this charred layer determines how long the member can continue to carry load before the strength of the remaining unburned wood material is exceeded. A thin layer of heat-affected wood below the char layer will have reduced strength and stiffness. LVL is made from thin veneers of timber glued together into large planks, which can be sawn into structural members. LVL is similar to thick plywood except that all the veneers have the same grain orientation and the continuous process can produce very large lengths. This paper describes an investigation into the fire performance of laminated veneer lumber (LVL) made from New Zealand grown radiata pine.
2. Test Material The LVL used in these tests was manufactured in New Zealand by Carter Holt Harvey Ltd using veneers 3 mm thick, glued into panels of up to 105 mm thickness. The average measured density and moisture content for the material used in each series of tests is shown in Table 1. Table 1: Density and moisture contents of samples tested. Density kg/m 3 Moisture content % Dry density kg/m 3 Ignition & instrumented 607 13.7 556 Pilot tests 604 12.0 540 Loaded test 608 12.4 541 3. Ignition Tests Ignition tests were carried out at five different levels of heat flux and two main grain orientations, with three replications of each test. The grain orientations were on the exterior veneer face (face grain) and on the edges of the cut veneers parallel to the grain of the veneers (edge grain). The results of these tests are shown in Figures 2 for the face grain and 3 for the edge grain where the Figure 1 LVL Grain Orientation inverse of the square root of the temperature has been plotted against the incident heat flux, together with the line of best fit. Two methods were used to estimate the ignition properties of the LVL. The first method is that of Mikkola & Wichman [1] and the second one is by Delichatsios, Panagiotou & Kiley [2]. These methods were recommended by Ngu [3] as having the best correlations based on his testing of the ignition properties of New Zealand timbers. The resulting values for the surface ignition temperature (T ig ) and the thermal inertia (k c) of the LVL samples, along with the estimate of the critical heat flux (q cr ) are shown in Table 2. 0.350000 0.300000 0.300000 0.250000 0.250000 Tig -1/2 (s-1/2) 0.200000 0.150000 tig -1/2 = 0.0047 q" e - 0.036 R2 = 0.9437 Tig -1/2 (s-1/2) 0.200000 0.150000 tig-1/2 = 0.005 q"e - 0.0495 R2 = 0.9753 0.100000 0.100000 0.050000 0.050000 0.000000 0 10 20 30 40 50 60 70 Heat Flux q"e (kw/m2) Figure 2 Ignition data LVL Face Grain. 0.000000 0 10 20 30 40 50 60 70 Heat Flux q"e (kw/m2) Figure 3 Ignition data LVL Edge Grain. Table 2: Estimated ignition properties, MW Mikkola & Wichman [1], DPK Delichatsios, Panagiotou & Kiley [2] Face grain Edge grain MW [1] DPK [2] MW [1] DPK [2] Critical heat flux q cr (kw/m 2 ) 7.66 7.66 9.90 9.90 Surface ignition T ig ( o C) 251 341 294 380 temperature Thermal inertia k c (kw/m2) 2 s 1.080 0.558 0.678 0.393 2
4. Charring Tests Charring tests were carried out using small blocks of LVL in the cone calorimeter and tests of larger unloaded LVL beams in a pilot scale furnace. 4.1 Cone calorimeter tests Instrumented tests were carried out using the cone calorimeter to determine the char depths with time. Eighteen samples of LVL measuring 100 mm square and 85 mm deep were instrumented using thermocouples placed in the central region of the blocks and at depths of 13, 25, 38, 50 and 63 mm from the face as shown in Figure 4. These samples were exposed Figure 4 Layout and depth of thermocouples to a constant heat flux of 35, 50 or 65 kw/m 2. The char depths were determined by using the thermocouples to locate the 300 o C isotherm which characterises the char front within the samples. From the results of these tests, the time for each thermocouple to reach 300 o C was interpolated for each thermocouple depth and the results averaged. Figure 5 shows the plot of char depth against time for each grain orientation and each heat flux. Char Depth (mm) 70 60 50 40 30 20 10 0 35kW Face 35kW Edge 50kW Face 50kW Edge 65kW Face 65kW Edge Uninst. 35kW Face Uninst. 35kW Edge Uninst. 50kW Face Uninst. 50kW Edge Uninst. 65kW Face Uninst. 65kW Edge 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 Time (min) Figure 5 Instrumented & un-instrumented char depth results Eighteen additional un-instrumented LVL blocks of the same dimensions were exposed to the same heat fluxes for specific periods of time; 20, 40 & 60 minutes. At the end of the test time, the blocks were immediately removed from the heat source, the char removed, and remaining solid LVL measured and recorded. The uninstrumented results for each of the heat fluxes and the different grain orientations have been included in Figure 5. It can be seen that very good correlations were obtained between the instrumented and the uninstrumented test results. For each of the 18 un-instrumented tests, the cumulative charring rate was calculated, and the average for each heat flux, including both grain orientations, was plotted against time as shown in Figure 6. 4.2 Pilot Furnace Tests Three tests on instrumented LVL beams exposed to the standard ISO 834 design fire were carried out using the pilot furnace at the BRANZ Fire Research facility. In the first two tests, the beams were 105 mm x 300mm deep and 2200 mm long. In the third test, two 66 mm thick by 360 mm beams were glued together using resorcinol adhesive prior to being instrumented, to give a beam size 132 mm x 360 mm deep. The beams were instrumented at two sections along the beam with 12 thermocouples installed at depths of 18 mm and 36 mm as shown in Figure 7. The instrumented sections were at the 1/3rd points in tests 1 and 2, and at one 1/3rd point and the centre for test 3. The thermocouples at the corners were affected by fire exposure on two faces whereas all other thermocouples essentially had single face exposure. 3
1.20 1.00 Char Rate (mm/min) 0.80 0.60 0.40 0.20 35 kw /m2 50kW/m2 65kW/m2 Average 35 kw/m2 Average 50 kw/m2 0.00 Average 65 kw/m2 10 20 30 40 50 60 70 Figure 6 LVL Char Rate from un-instrumented cone calorimeter tests Figure 7 Instrumented Section LVL beam Pilot Tests The results from the tests in the pilot furnace are shown in Figure 8 where they are compared with the results from the cone calorimeter tests. The current design charring rate of = 0.65 mm/min from NZS 3603 [4] is also shown in the Figure. A typical beam cross-section after the fire test is compared with the original section in Figure 9. 1.20 1.00 0.80 Char Rate (mm/min) 0.60 0.40 35 kw /m2 50kW /m2 65kW /m2 Average 35 kw/m2 0.20 Average 50 kw/m2 Average 65 kw/m2 Pilot Tests Average 0.65mm/min NZS 3603: 1993 0.00 10 20 30 40 50 60 70 Time (min) Figure 8 LVL char rate from the pilot furnace tests and the un-instrumented cone calorimeter tests compared with the NZS 3603 [4] charring rate. Figure 9 Beam cross-sections before and after testing in the pilot furnace The average char rate obtained from the 3 pilot tests varied from = 0.72 mm/min at 18mm depth to = 0.70 mm/min at 36 mm depth. The char rate of 0.72 mm/min was used to predict the behaviour of the load test in the full scale furnace. From the pilot test results the time lag at each thermocouple between recording 100 o C and 300 o C was evaluated. This was used with the cumulative charring rate for that particular thermocouple to calculate the thickness of LVL at a temperature over 100 o C. This layer of material is likely to have reduced mechanical properties because of the elevated temperatures. The estimated thickness of the temperature-affected material varied from 6 to 13 mm, with an average thickness of 9 mm. This information is useful because some design methods require an estimate of the thickness of zerostrength heated wood below the char layer [5]. The Australian timber code [6] states that a zerostrength layer thickness of 7.5 mm should be used, which is consistent with the findings of this study, recognising that not all of the wood between 100 o C and 300 o C will have zero strength. 4
5. Load-bearing tests The 105 x 300 mm LVL beam that was used for this test is shown in Figure 10 prior to being loaded and exposed to the standard ISO 834 design fire, in the full-scale furnace at the BRANZ Fire Research facility. The beam was subjected to an ambient temperature bending test to obtain a value for the Modulus of Elasticity (E). From the E value and by interpolation from the manufacturer s data [7] a value for the mean bending strength (f b )was estimated. The following properties were used for design of the 105 x 300 mm LVL beam: Modulus of elasticity E = 11.15 GPa Mean bending strength f b = 55.58 MPa Ultimate section capacity M n = 87.5 kn.m Bending moment in test M = 34.1 kn.m The beam was simply supported over a span of 4.4 metres and subjected to a central point load of 30kN applied by means of a hand-operated hydraulic jack. The load was kept constant during the test and the mid-span deflection of the beam was recorded. Figure 11 shows the beam after the fire test, while Figure 12 shows the plot of applied load and central deflection. Figure 10 LVL beam before the fire test Figure 11 LVL beam after the fire test It was observed through the observation ports of the furnace that charring of the surface did not begin until 3 or 4 minutes into the test. This delay in the onset of charring is not considered in most design methods which assume a constant rate of charring from the time of flashover [5]. 200 35 180 30 160 140 25 Deflection (m m) 120 100 80 Deflection, mm 20 15 60 Load, kn 10 40 20 5 0 0 0 5 10 15 20 25 30 35 40 Figure 12 Load and deflection plots 5
6. Discussion Based on a charring rate of 0.72 mm/min. from the cone calorimeter and pilot furnace tests, the loaded beam was expected to fail after approximately 38 minutes. It can be seen from Figure 12, that the beam actually failed at about 37 minutes after the start of the fire. Measurements of the remaining cross-section approximately 7 minutes after the test finished confirmed that the rate of charring for the section was similar to the pilot tests. Considering the above findings and observations the following two options for fire design of LVL can be considered: 1. Design using the char rate = 0.72 mm/min (found experimentally in this study) to calculate a reduced cross section which can be used with normal temperature properties (characteristic bending stress and modulus of elasticity) with no zero-strength layer of LVL below the char line. 2. Design using the char rate = 0.65 mm/min (complying with NZS 3603 [4]) to calculate a reduced cross section which can be used with normal temperature properties, with an allowance for a 7 mm to 9 mm zero-strength layer of LVL below the char line. 7. Conclusions A series of ignition and charring tests was carried out on laminated veneer lumber (LVL) manufactured from radiata pine. The results were similar to those previously found for solid timber of the same species. More investigations should be carried out to determine the effects of the delay in the onset of charring under exposure to post-flashover fire conditions, and also on the thickness and properties of the zero-strength layer below the char line. A simple method for predicting the fire performance of radiata pine LVL exposed to post-flashover fires is to use the experimentally found char rate = 0.72 mm/min to determine a reduced crosssection, and design using normal temperature properties without considering a heat-affected layer of wood below the char line. 8. References [1] Delichatsios, M.A., Panagiotou, T.H. and Kiley, F.,.The Use of Time to Ignition Data for Characterizing the Thermal Inertia and the Minimum (Critical) Heat Flux for Ignition or Pyrolysis., Combustion and Flame, Vol. 84, 323-332, 1991. [2] Mikkola, E. and Wichman I.S.,.On the Thermal Ignition of Combustible Materials., Fire and Materials, Vol. 14, 87-96, 1989. [3] Ngu, Chu Kan, Ignition Properties of New Zealand, a research project report presented as partial fulfilment of the requirements for the degree of Master of Engineering in Fire Engineering, University of Canterbury, Christchurch, New Zealand, 1997. [4] NZS 3603:1993 Code of practice for Timber Design, Standards Association of New Zealand [5] Buchanan, A.H. Structural Design for Fire Safety. John Wiley & Sons, Chichester, U.K. [6] Standards Association of Australia, 1990. Timber Structures, Part 4: Fire Resistance of Structural Timber Members. AS 1720.4-1990. [7] Carter Holt Harvey Ltd, LVL Properties Evaluation Documents 6