An Object-Oriented Finite Element Processing Model For Oriented Strand Board Wood Composites
An Object-Oriented Finite Element Processing Model For Oriented Strand Board Wood Composites
An Object-Oriented Finite Element Processing Model For Oriented Strand Board Wood Composites
INTRODUCTION
Since its introduction in the early 80’s, oriented strand board (OSB) has rapidly been
recognized in the marketplace as an alternative to plywood in the residential building
construction industry. OSB is an engineered wood product, which is made from small
diameter logs of fast growing species like aspen, poplar or pine. To produce this wood
composite, logs are first flaked into strands of dimensions varying from 70-150 mm in length,
5-30 mm in width and 0.5-1.0 mm in thickness. These strands are dried to a desired moisture
content, blended with resin (normally less then 5% by weight) and formed into a mat. The mat
formation process induces an in-plane or horizontal mat density variation due to random
overlapping of the strands. Finally, pressure and temperature are applied to the mat to
consolidate the wood elements and cure the resin.
During the hot pressing operation, heat and pressure are simultaneously applied from the press
platens to the mat. With the platen temperature being as high as 220 °C, the mat surface layers
are heated rapidly upon press closing. The elevated temperature then turns the moisture in the
wood into steam, which flows into the inner layers causing convective heat flow. The change
in local environmental conditions affects the wood consolidation characteristics and causes a
non-uniform deformation distribution in the mat thickness direction. After the resin is cured, a
small amount of springback is observed during the press opening, which leaves a noticeable
vertical density profile across the panel thickness in the final product. Obviously, the density
profile can have a substantial impact on bending and other properties of OSB panels.
To gain a better understanding of the complex pressing processes, models with different levels
of complexity have been developed [1-6]. In most cases, the heat transfer and the moisture
movement have been addressed [1-3], while the mat consolidation or deformation has been the
focus of models presented in other studies [4-6]. However, no attempt has been made to
integrate the solutions of all mentioned physical phenomena into a comprehensive model.
In this paper, an object-oriented finite element model is presented to simulate OSB mat
behaviour during pressing. In the first part, an overview of the model assumptions and
solution strategies is presented. Then, case studies are presented to demonstrate the merit of
the model as a practical tool for the panel manufacturers.
MODEL DEVELOPMENT
The structure of the model is composed of three major components: a database, a user
interface and a finite element method (FEM) solver (Fig. 1). The database contains all the
inputs to run the model (i.e. solution options, press definition, press cycle, boundary conditions
and material properties). The user interface allows the user to define a run and to monitor the
evolution of key parameters during the run. The FEM solver, which is the heart of the model,
contains all the calculation algorithms. To solve the complex problem, a modular approach [7]
is used for the FEM solver as shown in Fig. 1. The mat and the press are described at any
given time by a collection of state variables (e.g. temperature, moisture content, steam
pressure, displacement and press opening). The pressing process is simulated by consecutive
execution of modules describing various physical phenomena (i.e. heat and mass transfer, resin
cure, stress and deformation). The modular nature of the proposed model allows for flexible
incorporation of any changes that need to be implemented in the future.
The object-oriented method was applied in the design and implementation of the FEM solver.
Application of object-oriented methods for the development of engineering software is
relatively recent [8-10]. There are many advantages of using these methods. The most
important one is that the programs are very modular which make them easier to maintain and
to understand. More detailed description of the object-oriented approach can be found in [8].
For our application, the objects created and their relationship are presented in Fig. 2.
Model Description
In this model, the mat of wood strands is discretized into 1-D linear isoparametric elements in
the mat thickness direction (out-of-plane) as depicted in Fig. 3. Performing the analysis on
only a 1-D section is believed to be adequate, since most in-plane gradients are small compared
to out-of-plane gradients and thus can safely be ignored. However, the in-plane steam flow has
been included in the solution by using a ‘leak term’ at each node. We believe that the mat edge
pressure condition has to be considered as a significant quantity of steam can escape from the
mat edges. The mat is subdivided into a series of layers where different initial conditions,
material properties and strand dimensions can be assigned (Fig. 3).
FEM SOLVER
INPUT
MODULE
PRESS CONTROL
MODULE
INPUT
FILE
STRESS-DEFORMATION HEAT TRANSFER
MODULE MODULE
USER STATE
DATABASE VARIABLES
INTERFACE
RESIN CURE MASS TRANSFER
MODULE MODULE
OUTPUT
FILE
OUTPUT
MODULE
Fig. 1 Flow chart of the model components with the FEM solver architecture.
Compaction
Mat
Curve
Steam
Solution BC
Vector
Fig. 2 Objects description and relationship, (italic objects are utility objects used by the
main objects in bold).
Layer 3
Layer 2
In-plane
‘leak term’
Layer 1
z
y
Mat Mesh
The heat transfer module calculates the distribution of temperature (T) during pressing by
solving the heat transfer equation for conduction and convection in a porous medium:
∂T ∂ 2T ∂T
ρ mcm = k m 2 − ρ s cs v z +Q (1)
∂t ∂z ∂z
where ρm, cm and km are the mat density, specific heat and conductivity, ρs, cs and vs are the
steam density, specific heat and velocity, and Q is a heat flux term which contains water phase
change heat, resin cure heat generation and heat loss due to in-plane steam flow. A backward-
Euler time stepping scheme is used to solve Eqn 1. The material properties and various heat
fluxes are updated at each time step. The mat thermophysical properties (cm and km) are
calculated using the local mat density, temperature and moisture content, the steam properties
(ρs and cs) are computed from the steam tables based on local steam state, temperature and
pressure. The heat fluxes caused by steam in-plane flow and water phases changes are
obtained using the in-plane steam flow rate and the steam generation rate computed in the mass
transfer module to be discussed in the next section. Convective conditions are applied at the
boundaries to account for the contact thermal resistance between the press platens and the mat.
The mass transfer module calculates the distribution of vapor pressure (P) and moisture
content (MC). Assuming that steam behaves like a perfect gas and flows in the mat according
to Darcy’s law, the steam pressure governing equation can be written as follows:
Rs T ∂P ρ s K z ∂ 2 P ∂m L ∂mSG
= + + (2)
ε ∂t µ s ∂z 2 ∂t ∂t
where Rs, ρs, µs are the steam gas constant, density and viscosity, Kz and ε are the mat
permeability and porosity, ∂mL ∂t and ∂mSG ∂t are the steam in-plane flow rate and steam
generation rate (vaporization or condensation). A backward-Euler time stepping is used to
solve Eqn 2. The material properties and steam fluxes are updated at each time step. The mat
permeability (Kz) is calculated based on the mat local density. The steam properties (Rs, ρs and
µs) are computed from the steam tables using local steam state, temperature and pressure. The
moisture content is updated at each time step assuming that it corresponds to the steam
generation rate:
∂MC ∂mSG
≅ (3)
∂t ∂t
An impermeable (no flow) condition is assumed at the platen-mat boundaries. The steam
generation rate is calculated based on the steam state obtained from the local temperature and
pressure. The state of steam can be liquid, boiling and condensing. During vaporization or
condensation, the steam generation rate is obtained from the local heat flux and the steam
latent heat of phase change. The steam in-plane flow rate is computed assuming that a
pressure gradient is present from the mat center to the edges and is a function of the mat in-
plane permeability and the mat in-plane dimensions.
Stress-deformation module
The stress-deformation module calculates the mat deformation and stresses during pressing by
solving the stress equilibrium equation:
(
∂ σz − P )+F =0 (4`
∂z
z
where σ z , P and Fz are respectively the mat stress, steam pressure and body force. For the
current model, the contribution of the steam pressure and the body force are neglected. The
mat is treated as an elasto-plastic material with moisture and temperature dependencies of
wood mechanical properties. To solve Eqn 4 with material non-linearities, a Newton-Raphson
solution is used. Zero displacement condition is assumed at the bottom boundary, while
prescribed displacement or pressure is assumed at the top boundary depending whether the
press is in position or pressure control. At each time step, the mat compaction properties are
updated based on the local deformation, temperature and moisture content. For this model, the
viscoelastic behaviour of wood is neglected, this will mainly affect the accuracy of the
prediction of the stress relaxation observed during the hold period [11]. The mat density is
updated at each time step from the mat local strain. During the press opening phase, the mat
springback behaviour is modeled assuming that the mat behaves as a linear elastic material
where the elastic modulus or springback modulus is a function of the local resin degree of cure
and mat density.
The resin cure module calculates the resin degree of cure (α) assuming that the curing rate is a
function of the degree of cure and the temperature:
∂α
= f (α , T ) (5)
∂t
where f(α,T) can be an empirical or semi-empirical function proper to the type of resin used. A
large number of functions have been developed for epoxy or polyester resin systems [12] and
for a phenol-formaldehyde system [13]. In general, the function contains constants that have
to be determined from resin curing tests. At each time step, the curing rate is calculated based
on local temperature and degree of cure according to the curing function specified by the user.
The press controller module updates the press pressure or opening according to the press cycle
specified. The temperature of the platens is assumed constant. The press cycle is defined by a
series of segments similar to the procedure used for commercial presses. The press can be
controlled in position or pressure and feedback from the measured pressure and position is
obtained by the mat response calculated by the stress-deformation module.
CASE STUDY
To demonstrate the uses of the present model, a pressing simulation of OSB mat composed of
commercial strands is presented. A 35x35 cm2 mat with a target thickness of 11 mm and a
target density of 650 kg/m3 is meshed with 20 elements are shown in Fig. 4. A convection
thermal boundary condition (h=400 W/°Cm2) is used on top and bottom boundaries to account
for the thermal resistance between the mat and the press platens maintained at a constant
temperature (200 °C). Impermeable steam flow conditions are assumed on all boundaries.
Fixed displacement (uz=0) is assumed at the bottom platen and prescribed pressure (σz=Ppress)
or displacement (uz=uplaten) is assumed at the top platen interface depending whether the press
is in pressure or position control. Time steps of 0.05 seconds are used for the solution. The
material properties are obtained from our characterization tests or from the literature. The
press cycle used is generally composed of four segments as shown in Fig. 4. The simulation
results are compared with experimental results obtained for similar mat specifications and
pressing conditions.
Convection h
σz=Ppress
Top platen
or
uz=uplaten
Fig. 4 Mesh and press cycle definition for the case study.
Results
One of the most important process variables controlled by OSB producers is the mat initial
moisture content. The effect of the initial moisture content on mat consolidation behaviour and
core temperature profile is investigated. The predicted mat consolidation is consistent in trend
with experimental observations: an increase in initial moisture content leads to a reduction in
time to reach the target thickness and time of pressure relaxation (Fig. 5). However, the
predicted stress relaxation is significantly lower, particularly for the 9% moisture content case,
which may be caused by viscoelastic effects ignored in the present model. The predicted core
temperature profile follows the trend observed experimentally (Fig. 6): an increase in initial
moisture content increases the heat transfer to the core and reduces the maximum core
temperature. The extra amount of water generates steam which contributes to increase the
convective flow, but consumes more energy. One important effect captured by the model is
that the boiling temperature corresponding to the plateau observed in Fig. 6(a) is about 140 °C.
This is attributed to the fact that the core internal pressure is significantly higher than the
atmospheric condition. Such a coupling effect between the heat and mass transfer process
seems well simulated in the present model.
One of the major issues during pressing is the development of internal steam pressure because
it can potentially cause delaminations or ‘blows’ upon press opening. The mat density
significantly affects the magnitude of the internal pressure. At high density, the mat is not as
permeable and the flow of steam out of mat is thus reduced, which causes the internal pressure
to build up. To demonstrate this effect, the model was used to simulate the pressing of a low
density mat (600 kg/m3) and a high density mat (700 kg/m3). The predicted internal steam
pressure in the core is compared to the experiment results in Fig. 7. Again, the model captures
well the observed effect.
4 20 4 20
1% 1%
Mat pressure (MPa)
(a) (b)
Fig. 5 Effect of initial mat moisture content on mat pressure and press opening profiles, (a)
simulated, (b) measured.
220 220
1% 1%
Temperature (°C)
Temperature (°C)
180 9% 180
9%
140 140
100 100
60 60
20 20
0 100 200 300 0 100 200 300
Pressing time (sec) Pressing time (sec)
(a) (b)
Fig. 6 Effect of initial mat moisture content on core temperature profiles, (a) predicted, (b)
measured.
320 320
280 Low Low
Steam pressure (kPa)
280
240 High 240 High
200 200
160 160
120 120
80 80
40 40
0 0
0 100 200 0 100 200
Pressing time (sec) Pressing time (sec)
(a) (b)
Fig. 7 Effect of mat density on core steam pressure profiles, low and high correspond to a
mat density of 600 kg/m3 and 700 kg/m3 respectively, (a) predicted, (b) measured.
Finally, the vertical density profile is an important characteristic of OSB panels. Among the
factors affecting the shape and magnitude of the profile, the press closing rate is one of the
most important. The density profiles obtained with a fast and slow press closing times (20 and
90 seconds) are shown in Fig. 8. The density profile shape is affected, particularly close to the
panel surfaces. Similar result was found in refence [14] (Fig. 8(b)) and it is believed that pre-
mature curing of resin in the surface layers is the main cause to this phenomenon. It occurs
when the adhesive cures before the mat is plastically compressed. This condition which leads
to higher local springback during press opening is more susceptible to occur in a slow closing
rate pressing cycle.
800 900
Fast Fast
Density (kg/m )
Density (kg/m )
Slow Slow
3
800
3
700 700
600
Core (layer 8)
600 500
0 1 2 3 4 5 6 7 8 9 10 11 12 1 3 5 7 9 11 13 15
Vertical position (mm) Vertical position (layer)
(a) (b)
Fig. 8 Effect of press closing time on final panel density profile, fast and slow correspond to
a press closing time of 20 and 90 seconds respectively, (a) predicted, (b) measured from [14].
A 1-D object-oriented finite element pressing model for oriented strand board wood
composites has been developed. The model consists of several modules which simulate:
• heat and mass transfer, cure kinetics, wood compaction
• realistic boundary conditions and
• commercial pressing cycle.
A case study has been presented to demonstrate the usefulness of the current model. In
general, the model captures well the complex interactions between all the main pressing
variables, although some magnitude discrepancies exist between the model predictions and
experimental observations. Variations in local material properties and pressing conditions can
affect the results. Future model development will be possible by only adjusting specific
modules without affecting the global structure.
ACKNOWLEDGMENTS
Financial support from the Structural Board Association (SBA), the Science Council of British
Columbia (SCBC) and the Natural Sciences and Engineering Research Council (NSERC) of
Canada is gratefully acknowledged. We would like to acknowledge the contribution of Mrs.
Caroline Simon, Mr. Julien Ferriani, Mr. Maik Hirschberg and Mr. Craig Wilkinson for their
work on the measurement of the mat pressing behaviour.
REFERENCES
1. Humphrey, P.E. and Bolton, A.J., “The Hot Pressing of Dry-formed Wood-based
Composites Part II. A Simulation Model for Heat and Moisture Transfer, and Typical Results”,
Holzforschung, Vol. 43, No, 3, 1989, pp. 199-206.
2. Kayihan, F. and Johnson, J.A., “Heat and Moisture Movement in Wood Composite
Materials During the Pressing Operation - A Simplified Model”, Numerical Methods in Heat
Transfer, Volume II, John Wiley & Sons Ltd, 1983, pp. 511-531.
4. Lang, E.M. and Wolcott, M.P., “A Model for Viscoelastic Consolidation of Wood-
strand Mats, Part I. Structural Characterization of the Mat Via Monte Carlo Simulation”,
Wood and Fiber Science, Vol. 28, No. 1, 1996, pp. 100-109.
5. Lang, E.M. and Wolcott, M.P., “A Model for Viscoelastic Consolidation of Wood-
strand Mats, Part II. Static Stress-Strain Behaviour of the Mat”, Wood and Fiber Science, Vol.
28, No. 3,: 1996, pp. 369-379.
6. Dai, C. and Steiner, P.R., “Compression Behaviour of Randomly Formed Wood Flake
Mats”, Wood and Fiber Science, Vol. 25, No. 4, 1993, pp. 349-358.
7. Hubert, P., Johnston, A., Poursartip A. and Vaziri, R., “A Two-Dimensional Finite
Element Processing Model for FRP Composite Components”, Proceedings of the Tenth
International Conference on Composite Materials, Whistler, British Columbia, Canada,
August 14-18, 1995, Vol. III: Processing and Manufacturing, pp.149-156.
9. Besson, J. and Foerch, R., “Large scale object-oriented finite element code design”,
Computer Methods in Applied Mechanics and Engineering, Vol. 142, 1997, pp. 165-187.
10. Mackie, R. I., “Using Objects to Handle Complexity in Finite Element Software”,
Engineering with Computer, Vol. 13, 1997, pp. 99-111.
11. Wolcott, M.P., Kamke, F.A. and Dillard, D.A., “Fundamentals of flakeboard
manufacture:Viscoelastic behavior of the wood composite”, Wood Fiber Science, Vol. 22, No.
4, 1990, pp. 345-361.
12. Yousefi, A. and Lafleur, P.G., “Kinetic Studies of Thermoset Cure Reactions: A
Review”, Polymer Composites, Vol. 18, No. 2, 1997, pp. 157-168.
13. Barry, A. O., Peng, W. and Riedl, B., “The Effect of Lignin Content on the Cure
Properties of Phenol-Formaldehyde Resin as Determined by Differential Scanning
Calorimetry”, Holzforschung, Vol. 47, 1993, pp. 247-252.
14. Winistorfer, P.M., Young, T.M. and Walker, E., “Modeling and Comparing Vertical
Density Profiles", Wood and Fiber Science, Vol. 28, No. 1, 1996, pp. 133-141.