See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267564830

Coupled heat and moisture transfer in timber beams exposed to fire

Article · January 2006

CITATIONS READS

7 103

2 authors:

S. Schnabl Goran Turk

University of Ljubljana University of Ljubljana
44 PUBLICATIONS   441 CITATIONS    109 PUBLICATIONS   1,207 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Teaching spatial statistics for second cycle students at the department of geodesy. View project

Research on buckling behavior of CFST columns View project

All content following this page was uploaded by Goran Turk on 03 March 2015.

The user has requested enhancement of the downloaded file.

 Coupled heat and moisture transfer in timber beams exposed to fire

Simon Schnabl
young researcher
University of Ljubljana, Faculty of Civil and Geodetic Engineering
SI-1115 Ljubljana, Jamova 2, Slovenia

Goran Turk
associate professor
University of Ljubljana, Faculty of Civil and Geodetic Engineering
SI-1115 Ljubljana, Jamova 2, Slovenia

Summary
This paper describes the modelling of timber beams behaviour when one or more faces are exposed
to fire. A computer program for the prediction of two dimensional temperature and water content
distribution in timber beams exposed to fire has been developed. The model consists of differential
equations for coupled heat and moisture transfer with the corresponding boundary conditions.
Furthermore, the model also predicts the char formation in the wood beam as a function of its
temperature, moisture, density, etc. Owing to non-linear system of governing equations, solution is
obtained by numerical procedures such as finite difference method. Results obtained by numerical
model are be compared to the analytical solutions and experimental results published in the
literature.

1. Introduction

Ever since prehistoric times wood has been recognized as a very popular construction material.
Nowadays it has commonly been used in various fields of engineering. Due to its remarkable
properties it has been used by many architects and designers of modern buildings. The capability to
predict the behavior of wood, particularly when exposed to fire, has become increasingly important
in the field of fire safety engineering.

The purpose of this paper is to study and understand a charring behavior of wood and compare those
results to the one-dimensional charring rate models for wood presented in the literature and to the
experimental results published by Fredlund (1993).

When numerically analysing mechanical behavior of load bearing timber beams in natural fires, the
contributions of shrinkage-swelling, temperature, viscous creep and mechano-sorptive strains are of
major importance. The development of all aforementioned strains is strongly affected by the actual
temperature and moisture content distribution in the beams. Therefore, the determination of the
spatial and temporal distribution of temperature and moisture content over the element according to
ambient conditions is the first key phase of the analysis. When wood as a charring material is
exposed to fire, it undergoes thermal degradation, i.e. pyrolysis. The pyrolysis of wood is a complex
interplay of chemistry, heat, and mass transfer. It changes the wood to char and gases (such as
carbon dioxide and hydrocarbons) and consequently changes the density and other material
characteristics. The process of thermal degradation starts when the temperature of wood reaches a
certain threshold value which depends on the kind of wood, but is generally around 300 ◦ C.

Because of the importance and complexity of the pyrolysis of wood, there is a substantial volume of
work on the pyrolysis and charring of wood as a porous media, but the extensive literature review is
not the aim of this work. Experimental observations of charring behavior prove the mutual effect of
the temperature and moisture content gradients in wood, but it is rarely taken into consideration in
the computational analysis of charring in fire situations. In this paper the transient heat and moisture
transfer over a timber beam exposed to standard fire conditions is considered. The governing
equations of simultaneous heat and moisture transfer in porous media like wood were provided by
Luikov (1966).

Assuming a homogeneity of the temperature and moisture content field along the beam, the 2-D
Luikov equations are solved for the cross-section of the timber beam. Due to rectangular
cross-section the finite difference method using an equidistant mesh of finite difference points is
chosen for the solution. For the spatial integration the symmetric formulae based on quadratic shape
functions are introduced, whereas for the time-integration linear shape functions are employed.

2. Model theory and governing equations

Heat and mass transfer is governed by the two second order non-linear partial differential equations
(Luikov, 1966). The first equation describes heat conduction governed predominantly by
temperature gradient but affected also by the effect of phase-change and heat on sorption and
desorption, which depend by the speed of moisture changes. The second equation describes moisture
diffusion governed predominantly by moisture potential but is also considerably affected by
temperature gradients. Furthermore, anisotropy and temperature and moisture content dependent
material properties are assumed. The equations can be written as:
∂T ∂w ∂  ∂T  ∂  ∂T 
ρ Cp − ρ Cm (ε hLV + γ) = kx + ky , (1)
∂t ∂t ∂x ∂x ∂y ∂y
∂w ∂   ∂w ∂T  ∂   ∂w ∂T 
ρ Cm = DM x + δx + DM y + δy . (2)
∂t ∂x ∂x ∂x ∂y ∂y ∂y

In Eqs. (1–2) kx , ky represent thermal conductivity (W/mK) in two space directions x and y of a
rectangular cross-section of the beam. Similarly, DM x and DM y represent conductivity coefficient of
moisture content (kg/m s ◦ M), ρ represents density (kg/m3 ), cp heat capacity (J/kgK), T
temperature (◦ C), cm moisture capacity (kg/kg ◦ M), ε ratio of vapor diffusion coefficient to
coefficient of total moisture diffusion, hLV heat of phase change (J/kg), γ heat of sorption or
desorption (J/kg), w moisture potential (◦ M), t is time (s) and δx , δy thermogradient coefficient
(◦ M/K) in two independent orthogonal directions.

Moisture potential w is related to moisture content V by linear relation

V = cm w. (3)

The special solution of the system of governing differential equations (1–2) has to satisfy the initial
and boundary conditions. The initial conditions prescribe the temperature and moisture potential of
the cross-section of the beam at the initial time t = 0

T (x, y, 0) = T0 (x, y), (4)

and

w(x, y, 0) = w0 (x, y). (5)

The boundary conditions prescribe the heat and moisture flow on the exposed boundaries of
cross-section. It is assumed that the flow magnitudes depend on the differences between temperature
and moisture potentials on the boundary and in the surrounding air. In the heat boundary condition
the effect of evaporation on the heat flow is added while in the moisture boundary condition the
effect of temperature gradient is taken into account. Thus, the boundary conditions at the exposed
surface of the exterior siding are given by balancing heat conduction and moisture transfer at the
surface with the radiative and convective heat and convective moisture input. The boundary
conditions can be written as
∂T ∂T
−kx enx − ky eny = hc (T − TA ) + εR σ(T 4 − TA4 ) + (1 − ε) hLV hm (w − wA ), (6)
∂x ∂y
and
∂w ∂w ∂T ∂T
DM x enx + DM y eny + DM x δx enx + DM y δy eny = −hm (w − wA ), (7)
∂x ∂y ∂x ∂y
where enx and eny are components of the normal to the boundary surface, hc and hm are convective
heat transfer coefficient (W/m2 K) and convective moisture transfer coefficient (kg/s m2 ◦ M),
respectively. TA is the temperature and wA is the moisture potential of the ambient. The second term
in the right-hand side of Eq. (7) represents the radiative heat input where εR is the effective surface
emissivity of the exterior siding and σ is the Stephan-Boltzmann constant for radiation,
(σ = 5.671 · 10−8 W/m2 K 4 ).

The ambiental moisture potential is obtained through sorption isotherms. The effects of stress state
and history of sorption are neglected. The above system of the equations Eqs. (1–7) are solved using
finite difference method and a computer program written in M atlab enviroment.

3. Numerical examples

3.1 One-dimensional charring

The effect of charring of wood in timber construction has been a subject of intensive research for
many years. Charring of wood during fire has been extensively studied and considerable efforts have
been given to the development of theoretical models for wood charring that are intended to take into
account more complex geometries and variation of heat exposure. Unfortunately, no completely
satisfactorily charring model has yet been developed. On the other hand, an extensive char rate data
is available for simple one-dimensional wood slabs. Data is also available for two-dimensional
timbers, but most of this data is limited to larger cross-sections. Hence, a one-dimensional case of a
wood slab made of spruce, with a thickness d, exposed to the standard fire (standard fire curve ISO
834, 1999) is considered in order to compare the charring rate of the wood slab with the
one-dimensional empirical models presented in the literature. Generally, charring behavior of wood
can either be described by the mass loss rate (g/s) or by the rate of advance of the formed char front
from the original surface (mm/s). The latter definition has been more widely used because it enables
the determination of an effective residual cross-section area commonly employed in timber design
calculations. The rate of charring is a complex process which depends upon the interaction between
the pyrolysis of wood and the generation of heat, both of which are a function of a number of factors
such as the species, density moisture content, permeability and thermophysical properties. Since the
material properties at elevated temperatures are difficult to obtain, constant material properties of
wood and char are used. The data for this case is as follows
T0 = 20◦ C, w0 = 13◦ M, wA = 4◦ M, ρ = 370 kg/m3 , kwood = 0.12 kchar = 0.15 W/(m K),
DM = 2.2 × 10−8 kg/(m s ◦ M), hLV = 2500 kJ/kg, hc = 22.5 W/(m2 K), ε = 0, 3,
(8)
cp wood = 1530J/(kg K), cp char = 1050 J/(kg K), cm = 0.01 kg/(kg ◦ M), δ = 2.0 ◦ M/K,
hm = 2.5 × 10−6 , d = 0.3 m, γ = 0.

The comparison of different charring models is presented in Fig. 1.

Eurocode 5 (2004)
6 AS 1720.4-1990
White and Nordheim (1992)
5 Lawson et al (1952)
Char depth (cm)

Schaffer (1967)
4 Present model

0 10 20 30 40 50 60
Exposure time (min)

4 1: Comparison of different charring models to present one

Figure
Char depth (cm)

3
Most of the aforementioned models (Eurocode 5, 2004, AS 1720.4, 1990, Schaffer, 1967) suggest
constant charring rates. The use of these charring rates is convenient but does not accurately reflect
the actual charring behavior of wood. To account for the char non-linearity, White (1988) developed
2
a non-linear, one-dimensional char rate model based on the results of forty one-dimensional wood
slab charring tests. Later on, White and Nordheim (1992) developed a non-linear empirical model
for charring rate of eight different wood species. The comparison to the present model in the case of
1
spruce is shown in Fig. 1. The non-linear charringFredlund
rate model was also developed
(computational model) by Lawson et al.
Fredlund
(1952) who studied the charring rates of spruce timber (experiment)
beams of different thickness at 12% moisture
present model
content when exposed to standard ASTME E 119 (2000) heating regime. The comparison in Fig. 1
10
shows, that the models proposed in AS20 1720.4-1990
30 and by Schaffer
40 50 differ considerably. The
(1967)
Lawson’s and Schaffer’s models differ slightlyExposure time (min)but are virtually equal at 60
at the beginning,

Eurocode 5 (2004)
6 AS 1720.4-1990
White and Nordheim (1992)
5 Lawson et al (1952)
)
 Eurocode 5 (2004)
minutes. Similarly, the6model proposed
AS 1720.4-1990
by Eurocode 5 (2004), White’s and Nordheim’s model and
White
the model proposed in this paper differ and Nordheim
considerably in the(1992)
first 15 minutes, but later show similar
5 Lawson et al (1952)

Char depth (cm)

results. All models described aboveSchaffer
are relatively
(1967)simple to use. However, all empirical models are
limited to one-dimensional
4 cases. Also in allmodel
Present empirical models it is assumed that the charring of
wood starts instantaneously after exposure to fire. In reality, this is not the case. In our model, the
charring starts when the
3 temperature of wood reaches the temperature of pyrolysis, which is around
300 ◦ C. This happens nearly 4 minutes after the fire starts.
2
To validate our numerical model for a coupled heat and moisture transfer of wooden beams exposed
to fire we compare our1numerical results to the experimental results obtained by Fredlund (1988),
see Figure 2.
0 10 20 30 40 50 60
Exposure time (min)

4
Char depth (cm)

1
Fredlund (computational model)
Fredlund (experiment)
present model
10 20 30 40 50
Exposure time (min)

Figure 2: Comparison of experimental and numerical

Eurocode 5 (2004)results for penetration of the char layer as a
6 with initialAS
function of time for spruce moisture content of 14.5%
1720.4-1990
White and Nordheim (1992)
5 Lawson et al (1952)
Char depth (cm)

Schaffer (1967)
The comparison between the calculations andmodel
Present the experiment is based on measured and calculated
4
temperature-dependent thermo-mechanical properties of wood and char given by Fredlund (1988).
The example given below 3 deals with a specimen of spruce of initial moisture content of 14.5%. The
results are also compared to the numerical model proposed by Fredlund (1988). Figure 2 shows a
comparison of the penetration/formation
2 of the char layer as a function of fire exposure time. In the
Fredlund’s calculations, the char front has been defined as the point at which density has dropped to
300 kg/m3 , while in our1model the formation of the char is defined as the point at which temperature
reaches 300 ◦ C. As it can be seen from the Figure 2, there is a very good agreement between
measured and calculated results
0 of 10
the penetration
20 depth
30 of the 40
char. This50allows us
60to conclude that
the proposed mathematical model is generallyExposure
appropriate
timefor(min)
the prediction of charring behavior of
wooden beams exposed to fire.
3.2 A two-dimensional charring

In this case the formation of a char layer in a wooden beam exposed on three sides to standard fire
conditions defined in ISO 834 (1999) is considered. The upper edge is thermally isolated. The
original beam cross-section is rectangular with dimensions 10×15 cm. The beam cross-section is
discretized by the finite difference mesh of 30×45 points. Material properties are assumed to be the
same as they are in the first one dimensional case. The results of the simulation at times 1.2, 10, 20
and 30 minutes after exposure to fire are set out in Figures 3–4.

Since the corners are subjected to heat and moisture transfer from two directions, charring is faster at
these corners. Consequently, the formation of a char starts always at these corners. As a result, a
rounding effect occurs and shortly after the ignition the remaining load bearing cross-section is no
longer rectangular.

t = 1.2 min t = 10 min Char

15 15 300

250

Temperature (°C)
10 10 200
Height (cm)
Height (cm)

150

5 5
100

50

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Width (cm) Width (cm)

Figure 3: Temperature distribution

t = 20 min in cross-section of spruce beam
t = 30with Char of
min initial moisture content
13% and the transformation of wood into a char at 1.2 and 10 minutes 300
15 15

250
Temperature (°C)

10 10 200
Height (cm)
Height (cm)

150

5 5
100
 50

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Width (cm) Width (cm)

t = 20 min t = 30 min Char

15 15 300

250

Temperature (°C)
10 10 200

Height (cm)
Height (cm)

150

5 5
100

50

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Width (cm) Width (cm)

Figure 4: Temperature distribution in cross-section of spruce beam with initial moisture content of
13% and the transformation of wood into a char at 20 and 30 minutes

4. Conclusions

A mathematical model was developed to solve a coupled heat and moisture transfer and charring
behavior of timber beam when exposed to fire. Since the analytical solution is seldom obtainable, the
problem was solved numerically by finite difference method. The results were tested on the one
dimensional case for which comparison was made with the results obtained numerically and
experimentally. A comparison of different charring models showed that the charring rates obtained
by the proposed mathematical model were proven to be in excellent agreement with the results
obtained experimentally by Fredlund (1988). The model showed also to be in good agreement with
the Eurocode 5 (2004) model and model proposed by White and Nordheim (1992), while compared
to other charring models differ considerably. The same model was used to analyze a
two-dimensional behavior of wood beam exposed to fire from three sides. Faster charring at the
corners and typical rounding effect were observed. Thus, we can conclude that a relatively simple
mathematical model is generally appropriate for the accurate prediction of the thermomechanical
behavior of timber beams exposed to fire.

Acknowledgment
The work of S. Schnabl was financially supported by the Ministry of Education, Science and Sport of
the Republic of Slovenia under contract 3311-02-831625. The support is gratefully acknowledged.
5. References
View publication stats

[1] Luikov A. V. 1966. Heat and Mass Transfer in Capillary-porous Bodies. Pergamon Press,
Oxford.

[2] Fredlund B. 1988. A model for heat and mass transfer in timber structures during fire. A
theoretical, numerical and experimental study. Report LUTVDG/(TVBB–1033). Department of
fire safety engineering, Lund institute of science and Technology, Sweden.

[3] Fredlund B. 1993. Modelling of heat and mass transfer in wood structures during fire. Fire
Safety Journal Vol. 20, pp. 39–69.

[4] ISO 834. 1999. Fire-resistance test – Elements of building construction-Part 1. General
requirements. ISO 834-1, International organization for standardization. Geneva, Switzerland.

[5] EN 1995-1-2:2004. Eurocode 5: Design of timber structures Part 1-2: General-Structural fire
design.

[6] AS 1720.4. 1990. Timber structures – Fire resistance of structural timber members.

[7] Schaffer E.L. 1967. Charring rate of selected woods-transverse to grain. Research paper
FPL-69. USDA Forest Product Laboratory, Madison, Wisconsin, USA

[8] White R. H. 1988. Charring rates of different wood species. PhD Dissertation. University of
Wisconsin, Madison, WI, USA.

[9] White R. H. and Nordheim E. V. 1992. Charring rate of wood for ASTM E 119 exposure. Fire
Technology Vol. 28, No. 1, pp. 5–30.

[10] Lawson D. I. Webster C. T. and Ashton L. A. 1952. Fire endurance of timber beams and floors.
Journal of Structural Engineering Vol. 30, No. 2, pp. 27–34.

[11] ASTM. 2000. Standard test methods for fire tests of building construction and materials.
Standard designation E 119-00. West Conshohocken, PA:ASTM.