ANALYSIS OF CROSS-LAMINATED TIMBER
CHARRING RATES UPON EXPOSURE TO NON-
STANDARD HEATING CONDITIONS
Alastair I. Bartlett1, Rory M. Hadden1, Luke A. Bisby1, & Angus Law2
1School of Engineering, University of Edinburgh, UK, 2Arup, UK
The use of engineered timber products such as cross-laminated timber (CLT) is of increasing
interest to architects and designers due to their desirable aesthetic, environmental, and structural
properties. A key factor preventing widespread uptake of these materials is the uncertainty regarding
their performance in fire. Currently, the predominant approach to quantifying the structural fire
resistance of timber elements is the charring rate, which allows estimation of residual cross-section
and hence strength. The charring rate is usually determined by testing timber specimens in a furnace
by exposure to a ‘standard fire’. However, it is recognized that the resulting charring rates are not
necessarily appropriate for non-standard fire exposures or for characterizing the structural response in
a real timber building.
The effect of heating rate on the charring rate of CLT samples is investigated. The charring rate
resulting from three heating scenarios (constant, simulated ‘standard fire’ and quadratically
increasing) was calculated using interpolation of in-depth temperature measurements during exposure
to heating from a mobile array of radiant panels, or in a Fire Propagation Apparatus (FPA). Charring
rate is shown to vary both spatially and temporally, and as a function of heating rate within the range
0.36–0.79 mm/min. The charring rate for tests carried out under simulated ‘standard fire’ exposures
were shown to agree with the available literature, thus partially verifying the new testing approach;
however under other heating scenarios the Eurocode charring rate guidance was found to be
unconservative for some of the heat flux exposures in this study. A novel charring rate model is
presented based on the experimental results. The potential implications of this study for structural fire
resistance analysis and design of timber structures are discussed. The analysis demonstrates that
heating rate, sample size and orientation, and test setup have significant effects on charring rate and
the overall pyrolysis, and thus need to be further evaluated to further facilitate the use of structural
timber in design.
Cross-laminated timber (CLT) is an engineered mass timber product which has recently been
gaining popularity in the construction industry due to its aesthetic, environmental, and structural
appeal. It is engineered from multiple timber panels, which allows the impact of imperfections such as
voids and knots to be reduced compared to solid timber1. Adjacent lamellae are glued together with
their grain directions perpendicular, thus providing significant structural capacity in both directions.
Primarily used as structural wall and floor slabs, CLT offers bending strengths competitive with
concrete. This creates the opportunity for multi-story, architecturally unique buildings composed
predominantly of exposed structural timber. Due to the precedence of steel and concrete construction
for multi-story buildings during the last century, available fire design guidance for mass timber lags
behind current architectural vision, and places potentially arbitrary restrictions on application of CLT,
thus stifling innovation.
To enable safe, sustainable, resilient and efficient performance-based design of mass timber
structures, a detailed understanding of the structural behaviour of CLT at elevated temperatures, and
additional knowledge of the pyrolysis, ignition, and combustion processes is essential. Current design
guidance2 defines the fire resistance of timber as the duration it can withstand furnace exposure to a
standard temperature-time curve without its loadbearing, insulation, or integrity abilities being
compromised. These depend on three factors: (1) the contribution of any fire protection applied to the
exterior of the structural timber, such as plasterboard, (2) heating and charring of the timber, and (3)
the residual loadbearing capacity of the timber section, its integrity, and insulation performance3.
Whilst the fire resistance of timber can be relatively easily increased by adding gypsum plasterboard4,
this does not allow the desired architectural vision of exposed mass timber to be realised. Sacrificial
protection through charring is thus the preferred (aspirational) method of providing fire resistance.
TIMBER PYROLYSIS AND CHARRING
When timber is heated, it undergoes thermal degradation, producing pyrolysis gases, liquid
tar, and a rigid carbonaceous char5. The process of char formation is typically assumed to occur at
temperatures close to 300°C, with pyrolysis of wood typically commencing around 200°C6-8.
Pyrolysis and the subsequent combustion of wood has been observed to consist of four distinct
(1) Up to 200°C a mostly inert heating stage is observed, during which free water present in the
voids evaporates. Pyrolysis in this temperature range, if it occurs, is very slow, increasing
slightly after the sample has dehydrated13. Production of flammable gases is low during this stage
– the main products, in addition to water vapour, being small amounts of carbon dioxide and
carbon monoxide, as well as formic and acetic acids11. Structural changes in the complex lignin
molecules have been observed at temperatures as low as 65°C12 to 100°C14; these may result in a
reduction of structural strength due to the glass transition temperature of lignin being in the range
of 60-170°C15-17. Below 200°C char production is dominated by hemicellulose decomposition.
Prolonged heating to these temperatures can produce char from hemicellulose, leaving cellulose
largely unreacted11, 12, 18.
(2) At about 200°C to 300°C pyrolysis reactions remain slow, and most evolved gases remain non-
combustible11. The primary pyrolysis reactions occur in this range10, with hemicellulose typically
decomposing at lower temperatures, around 200-260°C, followed by cellulose around 240-
350°C, and finally lignin around 280-500°C8, 11, 19, with the onset of combustion following in the
presence of oxygen6, 10, 11. Cellulose may break down in two modes, one favouring production of
char and the other the production of volatiles8. Due to the slower reaction rates in this
temperature range, break down of cellulose will favour the former, exothermic reaction, yielding
more char than at higher temperatures11, 20.
(3) At about 300°C to 500°C pyrolysis reaction rates significantly increase6, 10-12, 21 as chain scissions
occur in the cellulose producing levoglucosan molecules8, 12, 19, 22, 23; these then rapidly
decompose further to produce volatiles11, 19, 21-23 such as methane, formaldehyde, hydrogen, and
methanol, providing the main pyrolysis gases supporting flaming combustion24. Char is formed
rapidly in this temperature range6, 10, 12, creating an insulating layer as shown in Fig. 1. Once a
char layer is formed the rate of additional char formation decreases to a lower, quasi-constant,
value6, 8, 9, 13, 20, 25-28 which serves to slow the progress of in-depth heating and delays further
pyrolysis reactions11. It should be noted that depths up to 40mm beneath the char layer
experience temperatures considerably above ambient7, 25, 29-34. The char yield is largely dependent
on the chemical makeup of the wood, with different species with different ratios of polymers
giving different char yields. Woods typically comprise about 50% cellulose, 25% hemicellulose,
and 25% lignin8, 10, 18, with considerable variations between species. As discussed, the char yield
of cellulose depends on its rate of pyrolysis. Lignin yields significantly more char6, 8, 11, 35, 5,
typically 40-50% of its original mass, and is the main contributor to char yield6, 8.
(4) At temperatures above 500°C, rapid secondary oxidation of char occurs6, 10-12, although the
temperature for this to occur has been quoted as low as 300°C25, 36 to 400°C36, 37 and as high as
700°C with increased external heat flux36, 38; it also depends on oxygen concentration36.
Secondary oxidation reduces the thickness of the char, thus reducing the insulation capacity and
consequently allowing additional in-depth heating of the sample and further pyrolysis.
It can thus be seen that the rate of char formation is governed by the rate of pyrolysis reactions,
therefore the factors affecting char formation shown in Table 1 are the factors affecting pyrolysis.
This is clearly an extremely complex thermo-physical-chemical process which is often not accounted
for by structural and fire engineers when making decisions regarding the use of exposed timber in
Table 1: Factors governing pyrolysis rates of timber
Density3, 6, 8, 12, 24, 28-30, 33, 39-42
More material packed into the same space requires more energy per
unit volume to decompose
Moisture content3, 6, 12, 14, 24,
26, 28, 41-43
Energy is required to evaporate absorbed moisture and drive it from
Species3, 6, 12, 24, 29-31, 41, 42, 44
This governs the ratios of lignin, cellulose, and hemicellulose, which
give different char yields
Permeability6, 12, 24, 29, 41, 42, 45
Volatiles can escape much more readily along the grain than across it
Oxygen concentration6, 10-12,
26, 37, 38, 43
Increased oxygen concentration allows for greater oxidation of char
and combustion of pyrolysis gases
Grain direction6, 8, 42
Influences permeability and thermal conductivity46
Sample orientation33, 39, 47, 48
Affects convective conditions, gas flow, flame behaviour, and
Sample size6, 27, 28
Affects the flame size, and thus thermal feedback and heat exchange
Heat flux3, 6, 8, 12, 30, 31, 40, 49
Higher fluxes provide more energy for pyrolysis reactions
A typical sample of CLT (and resulting in-depth temperature profile) exposed to an imposed heat flux
is illustrated in Fig. 1; the char layer has a lower density than the original timber and has negligible
structural strength. However, it has a comparatively low effective thermal conductivity as compared
to the virgin timber, enabling it to act as insulation for the remaining timber. Additional discolouration
exists below the char layer due to the pyrolysis reactions described above. It can also be seen in Fig. 1
that the first lamella of the CLT sample has started to delaminate due to failure of the polyurethane
CURRENT DESIGN APPROACHES AND TEST METHODS
Whilst it is clear that the pyrolysis and subsequent combustion of timber are complex
phenomena dependent on numerous factors, and that significant chemical and physical changes
resulting in strength loss occur before the onset of charring, this is not explicitly taken into
consideration in current design guidance2 or test methods. For instance, The Eurocode2 prescribes a
reduced cross-section method for calculating the residual structural strength of fire exposed structural
timber. An effective cross-section is assumed, equal to the initial cross-section minus the thickness of
the char layer plus an additional empirically-based “zero-strength” layer. The residual section is then
assumed to have full strength. The thickness of the char layer is calculated by multiplying the time of
exposure (to standard50 heating conditions) by a constant one-dimensional charring rate of
0.65mm/min; this fails to address the complexities noted above, such as the initially higher charring
rate or the dependency of charring on factors listed in Table 1 (with the exception of a single
additional value of 0.50mm/min for hardwoods with density over 450kg/m3). The so-called “zero-
strength” layer attempts to account for the additional heating and loss of mechanical properties that
occurs below the char line, and is assumed to be equal to 7mm although this has been demonstrated to
lead to unconservative results51. This value was determined experimentally for one specific setup, and
has been found to be heavily dependent on setup, failure mode, and assumed ambient properties34.
Alternative values of 19mm34 or time increasing values starting at 10mm for tension, and 18mm for
compression42 have been proposed. The variation in possible parameters for different scenarios
demonstrates that to provide robust and efficient design guidance, particularly for the case of non-
standard heating scenarios such as those being applied in the performance based design of buildings
across Europe, it is necessary to focus not only on charring rates, but also on the pyrolysis behaviour
as a whole, and to examine the temperature profile in a section, relating this to reduced structural
Figure 1: A CLT sample after 60 minutes of exposure to a quadratically increasing incident heat flux
showing char layer, additional discolouration, final temperature profile*, and onset of delamination
*it should be noted that towards the end of the test thermocouples near the surface intersected cracks, thus measuring gas
temperature rather than solid temperature, so actual temperatures in the char layer will have been significantly higher than
Fire resistances of structural elements of any material exposed to fire are currently established using
furnace tests, wherein test elements are placed in a furnace and subjected it to a specific, essentially
arbitrary52 time-temperature curve until failure50. Temperature is the primary control variable in the
test. However this does not allow active control of the thermal energy absorbed by the sample53, as
this depends on the convective conditions and radiative feedback within the test furnace, which will
differ from one furnace to another, as well as on the thermal properties of the structural element54.
On the contrary, most small scale tests performed on timber to determine combustion characteristics
have been undertaken in cone (or similar) calorimeters55, wherein thermal exposure is defined by
incident heat flux. The cone calorimeter provides better control over the thermal exposure conditions,
however structurally loaded tests cannot be performed. Summaries of the benefits and shortcomings
of each testing method are given in Table 2.
Tests undertaken to the standard fire50 in furnaces to date have shown average charring rates over the
time of fire exposure (as determined from the 300°C isotherm, despite the fact that char may begin to
form significantly before this) to vary from 0.36-0.82mm/min32, 43-45, 48, 56, 57. These tests, together with
a range of cone calorimeter tests, have provided the background knowledge presented in Table 1.
Table 2: Relative benefits and shortcomings of furnace and cone calorimeter testing methods
Allows element-scale testing
including structural loading
Relatively poor control and
repeatability in thermal exposure
Enables determination of average
charring rates and fire resistances
under standard heating conditions
Costly and time-consuming
Thermal exposure varies from one
material (and furnace) to another54
Well defined thermal exposure with
high degree of repeatability
Lack of ability to model real fire
behaviour and reflect the structural
and architectural situations in which
the material will be used in
Allows analysis of combustion gases
Allows analysis of burning rate
through mass loss measurements
Relatively inexpensive and rapid
Table 2 shows a serious disconnect between the goals and outcomes of large- and small-scale
testing of timber. To address this issue, the majority of the tests presented herein were performed at an
intermediate scale using a novel testing methodology called the Heat Transfer Rate Inducing System
(H-TRIS)53, 54. The intermediate scale enables limited structural loading, repeatable thermal exposure
and boundary conditions, calorimetric analyses, and is much more cost-efficient than traditional
furnace testing. Tests were also run using an FM Global Fire Propagation Apparatus (FPA)59 to study
effect of scale on the observed charring rates.
Tests in the Heat Transfer Rate Inducing System (H-TRIS)
H-TRIS comprises an array of radiant panels mounted on a linear motion system, allowing a pre-
determined time-history of heat flux to be imposed on a sample set opposite the panels. A detailed
description of the apparatus is presented by Maluk53, 54. Sitka spruce (average density 426kg/m3) and
Scots pine (average density 501kg/m3) samples of CLT of dimensions 300mm×200mm×120mm,
formed from three 40mm lamellae, were tested in a vertical orientation under a range of heating
scenarios (see below). Type K thermocouples (TCs) were inserted through the sides of the samples at
5mm depth increments from the exposed surface to 40mm depth, with an additional TC at 60mm.
Curve fitting was used to locate the position of the 300°C isotherm on the resulting thermal profile at
each time-step; this was used to give char depth, and thus charring rate.
Tests in the FM Global Fire Propagation Apparatus (FPA)
The FPA comprises four tungsten filament lamps intended to give uniform irradiation over the surface
of a sample59. It is similar to the cone calorimeter in size and test method, however the rapid thermal
response of the lamps means that, unlike the cone calorimeter, tests can be run under controlled
varying time-histories of heat flux. Douglas fir CLT samples (average density 490kg/m3) of
dimensions 90mm×90mm×80mm, formed from two 40mm lamellae, were tested in a horizontal
configuration. Mass loss was recorded during testing, which allowed charring rate to be approximated
from mass loss rate according to Eq. 10, 26:
where β is charring rate (mm/min), mf
'' is the mass loss rate per unit area (kg/m2.s), and ρw is the
density of wood (kg/m3).
Differences between Testing Methods
A key difference between the two testing methods used, in addition to their length-scales and
orientations, is their radiative emission spectra. The FPA has its spectral energy emission peak at 1.15
microns,59 whereas radiant panels (such as H-TRIS) typically have spectral energy emission peaks
above 2 microns60. However, it has been found for timber char that:
where α is the absorptivity and λ is the radiation wavelength in microns61. Thus for the FPA, α = 0.95,
and for H-TRIS, α = 0.91, and the difference in absorbed radiation after the char has formed is
therefore likely to be minimal.
To examine the effect of heating rate upon charring rate, three different heating scenarios were
considered as shown in Fig. 2: (1) simulated standard fire exposures; (2) constant heat fluxes; and (3)
quadratically increasing heat fluxes. The two simulated standard fire curves selected were the
standard cellulosic fire curve50, and the standard slow heating fire curve62 to allow comparison against
furnace test data. Constant initial incident heat fluxes of 30kW/m2 and 50kW/m2 were also used.
Additionally, arbitrary “slow”, “medium” and “fast” growing quadratically increasing heat flux curves
were selected to study the effect of continually increasing heat fluxes on charring, with rates of
8.33W/m2min2, 12.5W/m2min2, and 16.7W/m2min2, respectively. All of these were tested with H-
TRIS under unloaded and unpiloted conditions for a total duration of 60 min. Unpiloted FPA tests
were run at a constant 30kW/m2, and to the simulated standard fire curve, to allow comparison against
H-TRIS and furnace tests.
Figure 2: Heating scenarios tested in this work
To convert standard time-temperature curves into imposed heat-flux curves for H-TRIS, an inverse
heat-transfer model was developed to determine the required imposed incident heat flux to obtain
internal temperature gradients as would be experienced during furnace testing of timber elements. Full
details of the inverse model are given by Bartlett63. This approach produced temperature gradients
0 10 20 30 40 50 60
Imposed Heat Flux (kW/m2)
Simulated Standard Fire Curve Simulated Slow Heating Fire Curve
"Slow" Quadratic Curve "Medium" Quadratic Curve
"Fast" Quadratic Curve Constant 30kW/m^2
comparable to those observed in furnace testing and was deemed to be sufficiently robust for an initial
comparison of the influences of heating rate on charring rate.
RESULTS AND DISCUSSION
Effects of Thermal Exposure on Charring Rate
Table 3 gives test outcomes for the simulated fire curve and quadratically increasing heat flux
tests. One of the H-TRIS tests at 30kW/m2 was stopped after around 25 minutes, so only the peaks are
compared for these tests.
Table 3: Test results for simulated fire curves and quadratically increasing heat fluxes
Time to char
Time to flaming
Figure 3 shows the evolution of charring rate with time for each of the six simulated standard fire
tests. It shows that, far from a constant rate of 0.65mm/min, the charring rate is subject to an initial
peak before dropping off to a lower, quasi-steady value. Whilst the time-average values over the full
test duration are close to the Eurocode value, it should be noted that the initial peak has a significant
impact on this. For example, for the standard fire curve tests, the post-peak behaviour of the FPA tests
and the H-TRIS tests are similar, but the differences in peak behaviour yield significantly different
time-average charring rates. Taking an average charring rate over the whole test then is not an
accurate way of representing charring, and the effect of charring peaks must be considered due to
typical durations of compartment fires.
Figure 3: Evolution of charring rate with time for samples exposed to simulated standard fires
Whilst tests were undertaken on different species with different densities, the range of densities for
each species overlapped, thus there was no significant different in density between the species tested.
The lignin contents of Sitka spruce, Scots pine, and Douglas fir are similar (28%64, 27%65, and 25%66
respectively), so this was also not considered to be an important factor.
Figure 4 shows the evolution of charring rate with time for each of the nine quadratically increasing
heat flux tests. Again, the charring rate is subject to an initial peak before dropping off to a lower,
quasi-steady value. Due to the similarity of the magnitudes of each of the heating curves in the first
thirty minutes, no obvious trend was seen in the locations or magnitudes of the peaks. After the peak,
as with the simulated fire curve tests, behaviour was independent of the applied incident heat flux,
thus there was no clear variation in behaviour between the three quadratically increasing heating
curves was observed. It was noted that the charring rates observed for these tests were considerably
lower than those for the more severe simulated standard fire curves, thus to enable rational, efficient
design, different rates of charring should be assumed for different heating scenarios.
Figure 4: Quadratically increasing heat flux charring rate with time
0 10 20 30 40 50 60
Charring rate (mm/min)
Simulated Standard Fire Curve (H-TRIS)
Simulated Slow-heating Fire Curve
Simulated Standard Fire Curve (FPA)
0 10 20 30 40 50 60
Charring rate (mm/min)
"Slow" Quadratic Curve
"Medium" Quadratic Curve
"Fast" Quadratic Curve
Due to the variance in the test durations for the constant heat flux tests, only the peaks are compared.
The average charring rates over the peaks are shown in Table 4. Auto-ignition was observed in the
50kW/m2 tests within the first minute; no flaming ignition occurred during the unpiloted 30kW/m2
tests. Over the course of the flaming 30kW/m2 tests, an initial large flame was observed at ignition,
before rapidly reducing to a much weaker flame, around 1 to 2cm in height, for the remainder of the
Table 4: Test results for constant heat fluxes
Peak duration [mm:ss]
Time-average Charring Rate
over Peak [mm/min]
It is clear from Table 4 that there is a significant increase in charring rate over the peak with
increasing heat flux, again showing the need for heating rate dependent charring rates.
Effects of Testing Method on Charring Rate
As already discussed, the differences in radiation spectra between H-TRIS and the FPA are minimal
(~4%) for char, and it has been shown above that minor differences in imposed heat flux have no
discernible effect on the charring. It should also be noted that different methods were used for
measuring char depth in each set-up, for practical reasons. It has previously been found that the mass
loss rate method compares well with other methods10, 67, and hence this difference should not be
critical, however this should be examined further in future work. Thus, any differences between H-
TRIS and FPA results are likely due to the effects of sample size and orientation.
Tables 3 and 4 show that the time-average charring rates found in the FPA are noticeably lower than
those found in H-TRIS, as were the ignition times. This is in contrast to previous research which has
found that ignition times tend to be higher for vertical samples than horizontal samples8, 68. Larger
sample sizes (0.6m diameter cf. 0.1m diameter) have been shown to have larger initial heat release
rate per unit area peaks, by around 55%27, and for a different material, ignition times are
approximately 50% lower as sample size increases from 50 to 140mm square, with this effect greatest
for smaller sample sizes69. The ignition time was only 10% higher for a vertical sample at 30kW/m2
imposed heat flux68, therefore it appears that the differences in sample size dominate, thus yielding
higher charring rates for H-TRIS. Thus, combining these two parameters, a difference of
approximately 35% is to be expected. It is noted however, that the difference in the observed charring
rate peaks is much greater than this, around 2 to 3 times greater in H-TRIS than the FPA, thus the
difference may not be attributed solely to the sample size and orientation. As aforementioned, the
differences in emission spectra for char are negligible, but are significant for uncharred timber70. This
effect should be investigated through comparisons with cone calorimeter experiments at the same
scale in order to quantify the differences due to this parameter.
Whilst significant differences are observed between the test methods, the cause of these differences is
largely understood. It is hoped that with further testing across different apparatus, an accurate method
for scaling and interpolating across test methods can be developed.
Spatial Variation of Charring Rate
In addition to the temporal variation discussed herein, and the dependency of charring rate on heating
rate and sample size and orientation, spatial variation of charring rate was observed over single
samples. This was most notable in samples in which the top or front layer was formed of two different
sections of wood, as in Fig. 5. In the larger samples, it was often observed that one side of the sample
would visibly char sooner than the other, and this was attributed to local variations in grain orientation
and structure, density, and the presence of local imperfections, such as voids, splits and knots,. The
small-scale samples, however, revealed a much more obvious cause, as shown in Fig. 5. In the sample
shown in Fig. 5, the segment of wood on the left has a grain direction that is close to parallel with the
surface; the segment of wood on the right is closer to perpendicular. Whilst the grain is not exactly
parallel or perpendicular to the surface, they are of orientations which restrict and allow the flow of
volatiles respectively, as illustrated. As discussed in Table 1, permeability is greater along the grain
than across it, thus charring was faster on the right segment of the sample than the left.
Figure 5: Spatial variation of charring rate in a typical CLT sample with variable grain directions
In structural design using CLT, multiple large panels will be used, which will have significant
variations in density, grain direction, and void/knot content, so it is important to understand the effects
these variations have on the overall pyrolysis behaviour and structural response. Through small-scale
testing, these effects can be more easily quantified, allowing more efficient single-variable testing,
and the determination of the overall effects of material variability. However, the couple thermo-
mechanical performance of CLT panels can only be determined from tests at larger scales. This
clearly shows the need for testing at multiple scales, as performed herein, to properly understand and
account for all the necessary factors.
CHARRING MODEL AND DESIGN GUIDANCE
Due to the importance of the initial charring peak identified herein, a new, empirical charring
model is proposed to account for this, as shown in Fig. 6.
Whereas the Eurocode has a single-parameter model, with a constant value β = 0.65mm/min, the
model proposed herein has a trilinear charring rate, with the initial peak charring rate, β1 starting at a
time tchar, and then the second charring rate, β2 = 0.65mm/min, starting at a time tpeak. Thus:
for 0 ≤ t < tchar
for tchar ≤ t < tpeak
!for tpeak ≤ t
Whilst the minutiae are not presented herein, the parameters used in Fig. 6 are presented in Table 4. It
is suggested that β2 remain constant at 0.65mm/min, as the post-peak behaviour for all tests tended to
fluctuate around this value, and tchar, tpeak, and β1 should vary with heating rate, orientation, and sample
properties, as it has been shown (e.g. in Table 3) that the current Eurocode guidance is over-
conservative for some heating scenarios, and non-conservative for others.
Figure 6: Standard fire H-TRIS tests compared with Eurocode and new empirical charring model
Table 4: Tentative parameters for a new trilinear charring model
As mentioned previously, charring is just one part of the overall pyrolysis process. Thus, it is
suggested that this model be further developed to allow the analytical determination of temperature
profiles for any given heating curve, to allow a more accurate and robust determination of residual
strength than currently provided by adoption of the Eurocode’s zero-strength layer procedure.
The charring rate of CLT samples of varying sizes under different heating conditions has been
investigated. It is shown that charring rates are dependent on heating rate, test setup, and sample size
and orientation. Average charring rates under exposure to a simulated standard fire curve were found
to be 0.70mm/min using an intermediate scale radiant panel apparatus (H-TRIS), giving strong
agreement with existing literature, but were significantly higher and lower for more and less severe
fires, respectively. This confirms that current design guidance is likely unsuitable for use with heating
scenarios other than the standard fire curve in a fire testing furnace.
0 10 20 30 40 50 60
Char Depth (mm)
Simulated Standard Fire Curve FPA Tests
Simulated Standard Fire Curve H-TRIS Tests
Consistent with the available literature, charring rate was found to not be constant with time, as is
currently assumed in design guidance, but rather subject to an initial increased peak value before
dropping off to a lower, quasi-steady value. The duration and magnitude of the charring peak was
found to be dependent on heating rate, test setup, sample size, and orientation. Whilst the tail was
found to be independent of heating rate over the heating scenarios considered. The peak was found to
occur within the first 30 minutes testing. This is comparable to the duration of a typical compartment
fire (REF), and thus the effect of the peak will be significant in real buildings as opposed to furnace
tests where it is much less relevant.
Due to different behaviours observed due to different test setups, there is a need to examine the effects
of parameters such as radiation emission source, size, and orientation separately to allow robust
comparisons between methods. Due to the competing benefits of small- and large-scale testing, tests
must be carried out at multiple scales and robust methods of inter-scale comparison developed.
The pyrolysis process governing charring and the residual strength of timber has been shown to be
highly complex, and suitable methods of refining understanding are necessary to deliver testing and
engineering tools of appropriate complexity and robustness. To fully understand the problem of
timber in fire it is necessary to look beyond charring and better understand the evolution of through-
thickness temperature profiles within timber, and subsequently relate this to changes in mechanical
properties. Mechanical losses have been observed at temperatures well below those to cause charring,
yet this is not adequately accounted for in design guidance. To enable robust design of engineered
timber structures, further work will move towards developing comprehensive tools to predict through-
thickness temperature profiles with time as a function of fire exposure, and will link this to reductions
in mechanical properties. Only by adopting a holistic approach can resilient fire safety design be
This project was carried out at the University of Edinburgh, and the authors would like to
acknowledge funding and support from Ove Arup and Partners Ltd, the Engineering and Physical
Sciences Research Council, and Edinburgh’s School of Engineering.
1Guan, Z. & Rodd, P. 2001. Hollow steel dowels—a new application in semi-rigid timber
connections. Engineering Structures, 23, 110-119.
2CEN 2004. Eurocode 5. Design of timber structures. . General. Structural fire design. Brussels:
European Committee for Standardisation.
3White, R. H. 1995. Analytical methods for determining fire resistance of timber members. The SFPE
handbook of fire protection engineering. 2nd ed. Quincy, MA: National Fire Protection Association.
4Schmid, J., König, J. & Köhler, J. Design model for fire exposed cross-laminated timber. Proc of the
sixth International conference Structures in Fire, Lancaster, US, 2010.
5Hosoya, T., Kawamoto, H. & Saka, S. 2007. Pyrolysis behaviors of wood and its constituent
polymers at gasification temperature. Journal of Analytical and Applied Pyrolysis, 78, 328-336.
6Friquin, K. L. 2011. Material properties and external factors influencing the charring rate of solid
wood and glue!laminated timber. Fire and Materials, 35, 303-327.
7Buchanan, A. H. 2001. Structural design for fire safety, Wiley New York.
8Drysdale, D. 2011. An introduction to fire dynamics, John Wiley & Sons.
9Wichman, I. S. & Atreya, A. 1987. A simplified model for the pyrolysis of charring materials.
Combustion and Flame, 68, 231-247.
10Inghelbrecht, A. 2014. Evaluation of the burning behaviour of wood products in the context of
structural fire design. International Master of Science in Fire Safety Engineering MSc, The
University of Queensland, Ghent University.
11Browne, F. L. 1958. Theories of the combustion of wood and its control.
12White, R. H. & Dietenberger, M. 2001. Wood products: thermal degradation and fire.
13Yang, L., Chen, X., Zhou, X. & Fan, W. 2003. The pyrolysis and ignition of charring materials
under an external heat flux. Combustion and Flame, 133, 407-413.
14Reszka, P. & Torero, J. 2008. In-depth temperature measurements in wood exposed to intense
radiant energy. Experimental Thermal and Fluid Science, 32, 1405-1411.
15Li, W., Sun, N., Stoner, B., Jiang, X., Lu, X. & Rogers, R. D. 2011. Rapid dissolution of
lignocellulosic biomass in ionic liquids using temperatures above the glass transition of lignin.
Green Chemistry, 13, 2038-2047.
16Sakata, I. & Senju, R. 1975. Thermoplastic behavior of lignin with various synthetic plasticizers.
Journal of Applied Polymer Science, 19, 2799-2810.
17Salmén, L. 1984. Viscoelastic properties ofin situ lignin under water-saturated conditions. Journal of
Materials Science, 19, 3090-3096.
18Lautenberger, C., Sexton, S., & Rich, D. 2014. Understanding Long Term Low Temperature
Ignition of Wood. International Symposium on Fire Investigation Science and Technology. College
19Hirschler, M. M. & Morgan, A. B. 2008. Thermal decomposition of polymers. SFPE handbook of
fire protection engineering, 3, 1-112-1-143.
20Milosavljevic, I., Oja, V. & Suuberg, E. M. 1996. Thermal effects in cellulose pyrolysis:
relationship to char formation processes. Industrial & Engineering Chemistry Research, 35, 653-
21Shen, D., Fang, M., Luo, Z. & Cen, K. 2007. Modeling pyrolysis of wet wood under external heat
flux. Fire Safety Journal, 42, 210-217.
22Kawamoto, H., Morisaki, H. & Saka, S. 2009. Secondary decomposition of levoglucosan in
pyrolytic production from cellulosic biomass. Journal of Analytical and Applied Pyrolysis, 85, 247-
23Shen, D. & Gu, S. 2009. The mechanism for thermal decomposition of cellulose and its main
products. Bioresource Technology, 100, 6496-6504.
24White, R. H. & Nordheim, E. V. 1992. Charring rate of wood for ASTM E 119 exposure. Fire
Technology, 28, 5-30.
25Spearpoint, M. & Quintiere, J. 2000. Predicting the burning of wood using an integral model.
Combustion and Flame, 123, 308-325.
26Mikkola, E. Charring of wood based materials. Fire Safety Science–Proceedings of the Third
International Symposium. London: Elsevier Applied Science, 1991. 547-556.
27Ritchie, S. J., Steckler, K. D., Hamins, A., Cleary, T. G., Yang, J. C. & Kashiwagi, T. The effect of
sample size on the heat release rate of charring materials. Fire Safety Science: Proceedings of the
Fifth International Symposium, 1997. 177-188.
28Schaffer, E. L. 1967. Charring Rate of Selected Woods - Transverse to Grain. DTIC Document.
29Tran, H. C. & White, R. H. 1992. Burning rate of solid wood measured in a heat release rate
calorimeter. Fire and materials, 16, 197-206.
30White, R. H. & Tran, H. C. 1996. Charring rate of wood exposed to a constant heat flux.
31Frangi, A. & Fontana, M. 2003. Charring rates and temperature profiles of wood sections. Fire and
Materials, 27, 91-102.
32Friquin, K. L., Grimsbu, M. & Hovde, P. J. Charring rates for cross-laminated timber panels
exposed to standard and parametric fires. World Conference on Timber Engineering, 2010. 20-24.
33Yang, T.-H., Wang, S.-Y., Tsai, M.-J. & Lin, C.-Y. 2009. Temperature distribution within glued
laminated timber during a standard fire exposure test. Materials & Design, 30, 518-525.
34Schmid, J., Just, A., Klippel, M. & Fragiacomo, M. 2014. The Reduced Cross-Section Method for
Evaluation of the Fire Resistance of Timber Members: Discussion and Determination of the Zero-
Strength Layer. Fire Technology, 1-25.
35Yang, H., Yan, R., Chen, H., Lee, D. H. & Zheng, C. 2007. Characteristics of hemicellulose,
cellulose and lignin pyrolysis. Fuel, 86, 1781-1788.
36Ohlemiller, T., Kashiwagi, T. & Werner, K. 1987. Wood gasification at fire level heat fluxes.
Combustion and Flame, 69, 155-170.
37König, J. 2006. Effective thermal actions and thermal properties of timber members in natural fires.
Fire and materials, 30, 51-63.
38Kashiwagi, T., Ohlemiller, T. & Werner, K. 1987. Effects of external radiant flux and ambient
oxygen concentration on nonflaming gasification rates and evolved products of white pine.
Combustion and flame, 69, 331-345.
39Yang, T.-H., Wang, S.-Y., Tsai, M.-J. & Lin, C.-Y. 2009. The charring depth and charring rate of
glued laminated timber after a standard fire exposure test. Building and Environment, 44, 231-236.
40Lizhong, Y., Yupeng, Z., Yafei, W. & Zaifu, G. 2008. Predicting charring rate of woods exposed to
time-increasing and constant heat fluxes. Journal of Analytical and Applied Pyrolysis, 81, 1-6.
41Lau, P. W., White, R. & Van Zeeland, I. 1999. Modelling the charring behaviour of structural
lumber. Fire and materials, 23, 209-216.
42Cachim, P. B. & Franssen, J.-M. 2010. Assessment of Eurocode 5 charring rate calculation methods.
Fire technology, 46, 169-181.
43Cedering, M. Effect on the charring rate of wood in fire due to oxygen content, moisture content and
wood density. Proceedings of the Fourth International Conference Structures in Fire (SiF’06), 2006.
44Njankouo, J. M., Dotreppe, J. C. & Franssen, J. M. 2004. Experimental study of the charring rate of
tropical hardwoods. Fire and materials, 28, 15-24.
45Hugi, E., Wuersch, M., Risi, W. & Wakili, K. G. 2007. Correlation between charring rate and
oxygen permeability for 12 different wood species. Journal of wood science, 53, 71-75.
46Roberts, A. Problems associated with the theoretical analysis of the burning of wood. Symposium
(International) on Combustion, 1971. Elsevier, 893-903.
47Torero, J. L. 2008. Flaming ignition of solid fuels. SFPE Handbook of Fire Protection Engineering,
48Frangi, A., Fontana, M., Knobloch, M. & Bochicchio, G. 2008. Fire behaviour of cross-laminated
solid timber panels. Fire Safety Science, 1279-1290.
49Silcock, G. & Shields, T. 2001. Relating char depth to fire severity conditions. Fire and materials,
50ISO 1999. ISO 834-1: Fire resistance tests. Elements of building construction. Part 1: General
Requirements. Geneva, Switzerland: Internatinal Organisation for Standardization.
51Schmid, J., König, J. & Köhler, J. Fire-exposed cross-laminated timber-modelling and tests. World
Conference on Timber Engineering, 2010.
52Gales, J., Maluk, C. & Bisby, L. 2012. Structural fire testing - where are we, how did we get here,
and where are we going? 15th International Conference on Experimental Mechanics. Porto,
53Maluk, C. & Bisby, L. 2012. 120 years of structural fire testing: Moving away from the status quo.
54Maluk, C., Bisby, L., Terrasi, G., Krajcovic, M. & Torero, J. L. 2012. Novel Fire Testing
Methodology: Why, how and what now?
55ISO 2002. ISO 5660-1: Reaction-to-fire tests -- Heat release, smoke production and mass loss rate.
Part 1: Heat release rate (cone calorimeter method). Geneva, Switzerland: Inernational
Organisation for Standardization.
56Frangi, A., Fontana, M., Hugi, E. & Jübstl, R. 2009. Experimental analysis of cross-laminated
timber panels in fire. Fire Safety Journal, 44, 1078-1087.
57Wilinder, P. 2010. Fire resistance in cross-laminated timber. Bachelor thesis. Jönköping University,
58Law, A., Bartlett, A., Hadden, R. & Butterworth, N. 2014. The Challenegs and Opportunities for
Fire Safety in Tall Timber Construction. In: Galea, E. R. (ed.) Second International Tall Building
Fire Safety Conference. Greenwich, London: University of Greenwich.
59ASTM 2013. Standard Test Methods for Measurement of Material Flammability Using a Fire
Propagation Apparatus (FPA). West Conshohocken, PA: ASTM International.
60Comeford, J. 1972. The spectral distribution of radiant energy of a gas-fired radiant panel and some
diffusion flames. Combustion and flame, 18, 125-132.
61Grosshandler, W. & Monteiro, S. 1982. Attenuation of thermal radiation by pulverized coal and
char. Journal of Heat Transfer, 104, 587-593.
62CEN 1999. EN 1363-2 Fire resistance tests. Part 2: Alternative and additional procedures. Brussels,
Belgium: European Committee for Standardization.
63Bartlett, A. 2014. Charring Rates of Cross-Laminated Timber Under Standard and Non-Standard
Heating Scenarios. MEng, University of Edinburgh.
64Moore, J. 2011. Wood properties and uses of Sitka spruce in Britain, Forestry Commission.
65Kilpeläinen, A., Peltola, H., Ryyppö, A., Sauvala, K., Laitinen, K. & Kellomäki, S. 2003. Wood
properties of Scots pines (Pinus sylvestris) grown at elevated temperature and carbon dioxide
concentration. Tree Physiology, 23, 889-897.
66Erickson, H. D. & Arima, T. 1974. Douglas-fir wood quality studies part II: Effects of age and
stimulated growth on fibril angle and chemical constituents. Wood science and technology, 8, 255-
67Butler, C. 1971. Notes on charring rates in wood, Department of the Environment and Fire Offices'
Committee, Joint Fire Research Organization.
68Shields, T., Silcock, G. & Murray, J. 1993. The effects of geometry and ignition mode on ignition
times obtained using a cone calorimeter and ISO ignitability apparatus. Fire and materials, 17, 25-
69Hadden, R., Alkatib, A., Rein, G. & Torero, J. 2014. Radiant Ignition of Polyurethane Foam: The
Effect of Sample Size. Fire Technology, 50, 673-691.
70Girods, P., Bal, N., Biteau, H., Rein, G. & Torero, J. L. Comparison of pyrolysis behaviour results
between the cone calorimeter and the fire propagation apparatus heat sources. 10th International
Symposium on Fire Safety Science, 2011. International Association for Fire Safety Science, 889-