You are on page 1of 9

Tensile and compression properties through

the thickness of oriented strandboard


Caryn M. Steidl
Siqun Wang
Richard M. Bennett

Paul M. Winistorfer

Abstract
It is well known that the density varies through the thickness of oriented strandboard, with the faces being much denser than the
core. Density varies through the thickness because of consolidation characteristics of the wood elements during pressing in a
hot-press. Hence, the mechanical properties should vary through the thickness of the panel. To determine the variation in strength and
stiffness through the thickness of the panel, a commercial oriented strandboard was sawn into 15 layers to obtain thin-layer specimens
for tension and compression testing. Specimens were obtained both parallel and perpendicular to the length of the panel. The specimens were tested in tension using straight-sided specimens and unbonded tabs. For specimens parallel to the length of the panel, the
face layers had a tensile strength approximately an order of magnitude greater than the core. Greater face tensile strength was due to a
combination of strand orientation and density. An apparatus was designed to test the thin specimens in compression. The average
compression strength was significantly higher than average tension strength. However, the average compression modulus of elasticity
was significantly lower than average tension modulus of elasticity. These layer tension and compression properties were related to the
vertical density profile with high r2 values (> 0.75), thus indicating that a strong linear relationship exists. The layer properties were
used to predict the panel properties.

riented strandboard (OSB) is one


of the many engineered wood products
that is gaining increased use in both residential and commercial construction.
Although OSB has been used commercially for over 20 years, there are still
many aspects of the behavior and properties that are not fully understood. For example, it is well known that density varies
through the thickness of the panel, with
the faces having higher densities than the
core. This enhances the flexural behavior
because the denser faces have a greater
stiffness, creating a product that is analogous to an I-beam. However, it is not
known how the stiffness varies through
the thickness of the panel, or how the
stiffness is related to the density profile.
72

Increased understanding of the behavior


of OSB will enhance the further development and efficient use of this engineered
product. The objective of this work is to
build a relationship that would be predic-

tive of engineering flexural properties


from the vertical density profile. This is
accomplished by determining the engineering properties of individual layers of
the OSB.

The authors are, respectively, Former Graduate Student, Dept. of Civil and Environmental
Engineering, 221 Perkins Hall, The Univ. of Tennessee, Knoxville, TN 37996-2010 (currently Structural Designer, BKV Group, 222 North Second St, Minneapolis, MN 55401); Assistant Professor, Tennessee Forest Products Center, The Univ. of Tennessee, P.O. Box 1071,
Knoxville, TN 37901-1071; Professor, Dept. of Civil and Environmental Engineering, The
Univ. of Tennessee; and Former Professor and Director, Tennessee Forest Products Center
(currently Professor and Department Head, Dept. of Wood Science and Forest Products, 210
Cheatham Hall, Virginia Tech, Blacksburg, VA 24061-0323). The authors would like to thank
colleagues Chris Helton, William W. Moschler, and Ken Thomas for their helpful assistance.
The authors would also like to thank J.M. Huber for providing experimental materials. This
paper was received for publication in July 2001. Article No. 9350.
Forest Products Society Member.
Forest Products Society 2003.
Forest Prod. J. 53(6):72-80.
JUNE 2003

Previous researchers have examined


various aspects of through-the-thickness
properties of wood composite panels.
Perhaps the most commonly obtained
through-the-thickness panel property is
the density, or the vertical density profile (VDP) (Strickler 1959). The vertical
density profile of wood composites is
formed from a combination of actions
that occur both during consolidation and
also after the press has reached its final
position (Wang and Winistorfer 2000).
In order to examine the variation of internal bond (IB) through the thickness of
a panel, Xu and Winistorfer (1995)
sawed an OSB specimen into 9 layers,
and measured the IB of each layer. Although there was a positive correlation
between IB and density, the degree of
correlation was small (r2 between 0.20
and 0.25). The lowest IB did not always
occur in the low density core layer, and
the highest IB did not necessarily occur
in the high density face layers. The layer
sawing technique was also employed to
obtain specimens for water absorption
(WA) tests (Xu et al. 1996). These tests
revealed that WA was positively correlated to layer density and layer thickness
swell.

loading tests. Using the density gradient


in successive 1/32-inch increments
along with the developed nonlinear
modulus-density relationships resulted
in improved predictions, 92 percent for
the two-point loading and 102 percent
for the single-point loading test.

Andrews (1998) determined that there


was a negative correlation (r = -0.65) between the location of maximum density
in the tension face layer and the bending
modulus of elasticity (MOE) of the
panel. As the maximum density location
moved closer to the panel surface, the
stiffness of the panel increased. The location of the maximum density influenced the MOE more than the density
value itself. The same relationship was
true for modulus of rupture (MOR), but
the correlation was lower (r = -0.33).

Carll and Link (1988) studied the


layer behavior of 0.5-inch-thick aspen
and Douglas-fir flakeboard panels. The
1/8-inch-thick f ace layers and
1/4-inch-thick core layers were tested in
tension and compression. A logarithmic
relationship was developed between the
tensile or compressive MOE and the
specific gravity and wave speed. These
relationships were used with the density
measured in 6 layers through the thickness (10%, 15%, 25%, 25%, 15%, and
10% of the thickness) to predict the
bending MOE. The predicted MOE was
consistently 10 to15 percent higher than
the measured MOE.

Geimer et al. (1975) examined the effects of layer characteristics on 3-layer


particleboards. In one series of tests,
face and core layers were separated and
tested for stiffness in tension parallel to
the board surface. They suggested that
there was a nonlinear relationship between the MOE of the face layer and the
density of the face layer, while there was
a linear relationship between MOE of
the core and density of the core layer.
Laminate theory was used to predict
board properties from layer properties.
The predicted stiffness averaged 78 percent of the measured stiffness in
two-point loading tests and 87 percent of
the measured stiffness in single-point
FOREST PRODUCTS JOURNAL

One of the more extensive studies on


engineering properties was conducted
by Geimer (1979). He measured tension, compression, and bending MOE
and failure stress of full-thickness
flakeboards made with uniform densities throughout their thickness and different degrees of flake alignment. Logarithmic relationships between stiffness
(or strength) and specific gravity and
wave speed were developed. Several important behavioral aspects were determined from this work. The failure stress
(or MOR) was highly correlated with
the stiffness. The stiffness of boards
with a density gradient could be predicted to within 20 percent using the
stiffness-density relationship from uniform density boards. The bending MOR
was almost double that of the tension
failure stress for the same level of stiffness.

Grant (1997) examined the effects of


strand alignment on the mechanical
properties of OSB. A mathematical
model was constructed to describe the
relationship between the orientations of
the individual surface layer strata and
the unidirectional MOR and MOE of
OSB panels. The models confirmed the
positive influence of strand alignment
on MOE, but it was found that there was
a marginal return in improvement of the
mechanical properties above a certain
threshold. As expected, the contribution
to strength and stiffness was found to di-

Vol. 53, No. 6

minish from the outer surface to the


core.
Xu conducted a series of theoretical
studies on the effects of various layer
properties on the MOE of the panel. Xu
and Suchsland (1998) concluded in a
study that assumed a uniform vertical
density profile: 1) the panel MOE was
not influenced by particle size; 2) the
MOE decreases as the average
out-of-plane orientation angle of particles increases; 3) the MOE increases linearly with an increase of either board
density or compaction ratio (CR); and 4)
in-plane orientation improves MOE in
the orientation direction but reduces
MOE across the orientation direction,
with MOE in the across orientation direction leveling off after the percent
alignment exceeded approximately 60
percent. Xu (1999) used laminate theory
and simulated linear layer MOE-layer
density relationships to examine the effect of different VDPs on the panel
MOE. The analysis showed that the
MOE benefits from the high density surface layer and increases linearly with an
increase in peak density, but the maximum MOE does not occur when the
peak density is located at the extreme
board surface.
These previous studies have examined many different aspects of the
through-the-thickness behavior of wood
composite panels. There has been an attempt to relate many of these properties
to the vertical density profile. The purpose of this study was to determine the
through-the-thickness tension and compression strength and stiffness of a commercial OSB panel and to relate the
layer mechanical properties with the
layer density. A much thinner layer than
previously used was chosen for this
study to better define the variation of the
engineering properties through the
thickness. The layer properties were
used to predict panel properties.

Experimental method
Specimen preparation
All specimens were cut from one 4by 8-foot, 23/32-inch-thick commercial
southern pine OSB panel. The OSB
panel was production sanded and
bonded with a diphenyle-methylene
diisocyanate (MDI) resin. Fourteen
sample sets were cut parallel to face
strand orientation and 14 sample sets
were cut perpendicular to face strand
73

orientation. One bending specimen,


three tension, and three compression
specimens were cut from each sample
set, as shown in Figure 1. Bending specimens were 3 by 19-1/4 inches as specified in ASTM D 1037 (1996). The 2- by
2-inch specimens were cut from the
bending specimens after testing to measure the vertical density profile. Compression specimens were 1.5 by 4 inches
and tension specimens were 1.5 by 8
inches before making thin slices.

Figure 1. Detail of offset cuts for tension and compression specimens.

To better define the engineering properties through the thickness of the panel,
a thin layer was chosen. However, as the
layer gets thinner, it behaves less like a
homogeneous material. We chose to use
15 layers through the thickness of the
panel for tension and compression testing, resulting in a specimen thickness of
0.047 inch. The final tension and compression specimens were 1 inch wide by
8 inches long and 1 inch wide by 4
inches long, respectively. The OSB material was a multi-layered alignment
panel. OSB face layer strands were
aligned opposite to the core layer; those
specimens cut parallel to panel length
produced approximately three specimens parallel to face strand alignment,
nine specimens perpendicular to face
strand alignment, and three specimens
parallel to face alignment, respectively,
through the panel thickness. The opposite was true for the specimens cut perpendicular to face strand alignment.
These layer changes were confirmed by
visual inspection after the thin specimens were obtained.
Three full-thickness tension and three
full-thickness compression pieces were
required for each sample set because
only five thin specimens could be obtained from each full thickness piece.
When the 15 thin specimens are arranged according to their position in the
thickness of the board, they reflect the
23/32-inch full-thickness board.

Figure 2. Cutting diagram for one sample set.


74

A cutterhead consisting of six


7-1/4-inch diameter, 0.094-inch-thick,
18-tooth carbide-tipped blades with appropriately sized spacers was mounted
on an arbor in a milling machine to
achieve the desired sample thickness of
0.047 inch. The full-thickness pieces
were initially 1.5 inch wide, so they
could be held in a vice attached to the
milling machine table. The machine was
set to cut to a depth of 1 inch and to feed
JUNE 2003

Table 1. The effective panel MOE and MOR.


Property

No. of specimens tested

Mean

COV

Parallel

14

934,000

0.08

Perpendicular

14

505,000

0.07

Parallel

14

5,230

0.18

Perpendicular

14

3,520

0.10

183

41.2

0.03

MOE (psi)

MOR (psi)

Average density (pcf)

Table 2. Layer properties for perpendicular to panel length tension specimens.

Layer

Strand
orientation

Density

No. of
specimens tested

Mean

SDa

- - - - (pcf) - - - -

Strength
Mean COV

MOE
Mean

COV

- - - - - - - - - (psi) - - - - - - - - - -

Perpendicular

14

52.5

2.50

616

0.37

3.79E + 05

0.36

Perpendicular

13

49.1

1.71

522

0.47

2.85E + 05

0.34

Perpendicular

13

44.7

1.63

384

0.41

2.38E + 05

0.33

Parallel

14

40.3

1.28

616

0.61

5.09E + 05

0.62

Parallel

13

37.4

0.69

699

0.41

5.32E + 05

0.36

Parallel

14

36.1

0.37

576

0.33

4.50E + 05

0.31

Parallel

14

35.2

0.32

495

0.34

3.82E + 05

0.32

Parallel

14

34.7

0.33

446

0.27

3.77E + 05

0.36

Parallel

14

34.8

0.29

594

0.42

4.23E + 05

0.29

10

Parallel

14

35.5

0.46

546

0.22

5.33E + 05

0.34

11

Parallel

14

36.9

0.64

672

0.33

4.65E + 05

0.37

12

Parallel

14

39.1

0.85

615

0.59

4.64E + 05

0.43

13

Perpendicular

13

42.5

1.45

484

0.39

2.50E + 05

0.31

14

Perpendicular

14

46.7

1.58

466

0.26

2.49E + 05

0.29

15

Perpendicular

14

50.2

1.16

634

0.41

3.33E + 05

0.25

13.7

41.0

1.02

558

0.39

3.91E + 05

0.35

Average
a SD

= standard deviation.

the samples into the cutterhead at a rate


of 1.75 feet per minute.
Only five specimens could be obtained from each full-thickness tension
and compression piece. Layers 1, 4, 7,
10, and 13 were obtained from the first
piece. Layers 2, 5, 8, 11, and 14 were obtained from the second piece. Layers 3,
6, 9, 12, and 15 were obtained from the
third piece. This was accomplished by
cutting all of the first pieces, offsetting
the blades 0.047 inch into the sample,
and cutting all of the second pieces. The
third pieces were cut in a similar fashion, thus obtaining samples representative of a full-thickness board. Clarification of the offset specimens is shown in
Figure 2.
After all tension and compression
pieces were cut through the thickness, a
bandsaw was used to trim off most of the
1/2-inch gripping edge. An end mill was
used for the final trimming to separate
FOREST PRODUCTS JOURNAL

the thin specimens. In order to prevent


damage to the specimens during this
process, plastic spacers were inserted in
the sawkerfs. This process successfully
produced thin specimens of the desired
quality and thickness.
Tension testing
The necked-down, or dog-boned-shaped,
specimens typically used in tension testing were not used in this research due to
the difficulty of making these specimens
from the thin OSB slices. Rather,
straight-sided specimens with a uniform
rectangular cross section were used,
which is similar to the way thin polymer
matrix composites are tested (ASTM
1995). Self-aligning g rips with
non-bonded tabs were used to pull the
specimen. The tabs were small pieces of
wood with sandpaper adhered to them.
This set-up proved to be successful.
Most samples failed between the grips

Vol. 53, No. 6

with only 13 percent of the specimens


failing within the grips.
The specimens were tested under deflection control at a rate of 0.020
in./min. Strain was measured using an
extensometer consisting of two linear
variable displacement transducers
(LVDTs), one on either side of the specimen to correct for bending. The gage
length was 4 inches and an average deflection was used in data analysis.
The MOE of the tension and compression specimens was obtained from the
slope of the stress-strain curve. Some
stress-strain curves had a small initial
vertical portion before the extensometer
began recording deformation. A minimum stress was chosen just above the
vertical portion of the stress-strain
curve. The beginning slope value was
calculated for the portion of the
stress-strain curve between the minimum stress and the minimum stress plus
30 percent of the failure stress. The minimum stress was incrementally increased by 2 percent of the failure stress
and a regression calculated the slope of
the next 30 percent of the curve. Once
the upper stress reached the maximum
stress, the incremental process stopped.
The maximum slope calculated during
the step-wise regression was taken as the
tension MOE. Thus, the reported tension MOE is the least squares fit over the
steepest 30 percent of the stress-strain
curve. Values of r2 for the linear regression were typically greater than 0.9.
There were 420 tension specimens
cut. While handling and loading samples in the testing machine, some samples broke, thereby reducing the number
of tests. Except for one, all of the samples that failed in handling were for
strands perpendicular to the applied tension. Thus, there may be some small
bias in the tension perpendicular to
strands results. Tension tests were conducted on a total of 403 specimens: 206
cut perpendicular to face strand orientation and 197 cut parallel.
Compression testing
Published test methods for wood panels in compression (ASTM 1994) suggest an apparatus that utilizes spring
steel to provide the lateral support. Due
to the thin specimens of this study and
make-up of OSB, this apparatus was not
feasible because the steel tines could
pierce into the thin specimens or
75

Compression tests were conducted on


a total of 403 specimens: 209 cut perpendicular to face strand orientation and
194 cut parallel. One complete parallel
sample set was lost due to an error in the
cutting process. Only two other specimens were lost in handling. Some of the
compression samples became wedged
in between the UHMW blocks after failure. However, the maximum load taken
by the specimen before wedging was
noted, and the compression MOE was
calculated using the appropriate data.
Other testing
A commercial densitometer (QMS
Density Profile System QDP-01X) was
used to measure the vertical density profile, with the density being measured at
0.02-inch increments through the thickness of the specimen. There were 183
density profiles determined. Density
points ranging between the start and end
points for each layer were averaged to
find the average density for each layer.

Figure 3. Compression test apparatus.

Table 3. Layer properties for parallel to panel length tension specimens.

Layer

Strand
orientation

No. of
specimens tested

Density
Mean

SDa

- - - - (pcf) - - - -

MOE
Mean

COV

- - - - - - - - - (psi) - - - - - - - - - -

Parallel

14

53.6

3.14

2185

0.36

1.08E + 06

0.26

Parallel

14

49.5

2.04

1640

0.33

7.87E + 05

0.19

Parallel

14

44.3

1.73

730

0.61

4.29E + 05

0.44

Perpendicular

14

39.9

1.17

288

0.37

1.42E + 05

0.53

Perpendicular

12

37.4

0.58

227

0.22 1.10E + 05

0.45

Perpendicular

12

36.3

0.38

166

0.25

9.56E + 04

0.25

Perpendicular

14

35.6

0.30

163

0.32

1.27E + 05

0.46

Perpendicular

11

35.1

0.28

187

0.27

1.04E + 05

0.38

Perpendicular

14

35.1

0.31

147

0.24

1.16E + 05

0.64

10

Perpendicular

11

35.6

0.43

164

0.35

1.56E + 05

1.22

11

Perpendicular

11

36.7

0.54

259

0.49

1.76E + 05

0.37

12

Perpendicular

14

39.1

0.98

347

0.49

2.86E + 05

0.45

13

Parallel

14

42.7

1.42

807

0.45

5.77E + 05

0.46

14

Parallel

14

47.0

2.01

1568

0.42

8.43E + 05

0.18

15

Parallel

14

51.6

1.69

1992

0.54

9.17E + 05

0.30

13.1

41.3

1.13

725

0.38

3.96E + 05

0.44

Average
a SD

= standard deviation.

through any voids, which are not uncommon in the core specimens. Hence,
a compression testing apparatus was
designed to provide lateral support, yet
preserve the integrity of the specimen
(Fig. 3). Full drawings of the device can
be found in Steidls thesis (Steidl 2000).
Lateral support is provided by two ultra
high molecular weight (UHMW) plastic
blocks. UHMW was chosen for its low
coefficient of friction with OSB, which
76

Strength
Mean COV

Results

was measured as 0.25. Preliminary tests


confirmed that there was minimal load
transfer between the specimen and lateral supports.
Like tension testing, the compression
specimens were tested under deflection
control at a rate of 0.020 in./min. Since
an extensometer could not be attached to
the specimen, strain was determined
from the testing machine crosshead
movement.

The effective panel MOE and MOR


are given in Table 1. An average vertical
density profile of the panel is shown in
Figure 4. The average panel density was
41.2 pcf. Average density for each layer
is given in Tables 2 to 5.
Tables 2 and 3 list the number of samples tested, average density, average tension MOE, and average tension strength
by layer for perpendicular and parallel
cut specimens. Figure 5 shows typical
failures for perpendicular (L1 and L15)
and parallel (L5 and L9) to panel length
specimens. Tables 4 and 5 list the number of samples tested, average density,
average compression MOE, and average
compression strength by layer for perpendicular and parallel cut specimens.

Discussion
Tension properties
For specimens parallel to the length of
the panel, the face layers had a tensile
strength and MOE approximately an order of magnitude greater than the core
(Table 3). This was due to a combination of a denser face and the face strands
being oriented parallel to the applied
tension. For specimens perpendicular to
the length of the panel, the tensile
strength and MOE were relatively uniform through the thickness (Table 2).
The denser faces, with the strands oriented perpendicular to the applied tenJUNE 2003

Table 4. Layer properties for perpendicular to panel length compression specimens.

Layer

Strand
orientation

Density

No. of
specimens tested

Mean

SDa

- - - - (pcf) - - - -

Strength
Mean COV

MOE
Mean

COV

- - - - - - - - - (psi) - - - - - - - - - -

Perpendicular

14

52.4

2.25

860

0.46 1.98E + 05

0.37

Perpendicular

14

49.0

1.61

1013

0.32 1.53E + 05

0.22

Perpendicular

14

44.7

1.57

893

0.36 1.44E + 05

0.38

Parallel

14

40.2

1.17

1129

0.43 2.29E + 05

0.43

Parallel

14

37.4

0.66

1106

0.33 2.27E + 05

0.24

Parallel

14

36.0

0.36

895

0.38 2.08E + 05

0.34

Parallel

14

35.2

0.32

849

0.36 1.83E + 05

0.30

Parallel

14

34.7

0.33

764

0.36 1.93E + 05

0.30

Parallel

14

34.8

0.29

906

0.40 2.08E + 05

0.26

10

Parallel

14

35.4

0.47

1127

0.41 2.48E + 05

0.41

11

Parallel

14

36.9

0.62

983

0.32 2.20E + 05

0.27

12

Parallel

14

39.1

0.84

837

0.40 1.65E + 05

0.38

13

Perpendicular

13

42.5

1.49

923

0.50 1.40E + 05

0.38

14

Perpendicular

14

46.7

1.52

922

0.29 1.53E + 05

0.24

15

Perpendicular

14

50.3

1.09

1024

0.50 1.64E + 05

0.39

13.9

41.0

0.97

949

0.39 1.89E + 05

0.33

Average
a SD

= standard deviation.

Figure 4. Average vertical density profile of OSB panel.

sion in testing, had approximately the


same strength and MOE as the less
dense core, where the flakes were
aligned with the applied tension. In
specimens with perpendicular strand
orientation (L1 and L15), the failure is
typical of perpendicular to grain failures
(Fig. 5). Specimens with parallel strand
orientation (L5 and L9) experience failure within the strands.
All of the tensile properties had relatively high scatter, as reflected by an average coefficient of variation (COV) of
39 percent. This was expected due to the
FOREST PRODUCTS JOURNAL

small specimen size used in this work.


Locally, there is a high variation in the
properties of OSB; however, there is
much less variation in panel properties
as reflected by much lower COVs for the
bending specimens.
Compression properties
For specimens parallel to the length of
the panel, the face layers had a compressive strength and MOE approximately
an order of magnitude greater than the
core (Table 5), which was similar to the
tension strength and MOE. For perpendicular to panel length samples, the

Vol. 53, No. 6

compression strength and MOE were


relatively uniform through the thickness
(Table 4). Typically, failure occurred in
the compression specimens as a result of
strands sliding over one another. The average COV in compression properties
was 36 percent.
The compressive MOE was about half
the tensile MOE, averaging 0.46 of the
tensile MOE for specimens with the
load parallel to the strands and 0.49 of
the tensile MOE for specimens with the
load perpendicular to the strands. Carll
and Link (1988) also found the compressive MOE to be less than the tensile
MOE for OSB, although the difference
was smaller. They reported the full
thickness compressive MOE of aspen
and Douglas-fir OSB panels as about 89
percent of the tensile MOE. Geimer
(1979) reported compressive MOE of
approximately 92 percent of the tensile
MOE for Douglas-fir OSB panels.
The compression strength was
considerably greater than the tensile
strength. The compressive strength averaged 1.52 times the tensile strength for
specimens with the load parallel to the
strands, although the outer two face layers had essentially the same tensile and
compressive strengths. The compression strength averaged 1.99 times the
tensile strength for specimens with the
load perpendicular to the strands. This is
opposite of the findings of Geimer
(1979) who found the compressive
strength of Douglas-fir OSB panels to
be approximately 80 percent of the tensile strength. We do not have a good explanation for this, although we note that
our compression testing method did provide full lateral support of the specimen.
Relating layer properties
to vertical density profile
In order to relate the vertical density
profile to the layer tension/compression
properties, layers were grouped by compression and tension, and for perpendicular and parallel strand orientation. The
tension/compression MOE and strength
values were plotted versus density (Figs.
6 and 7) and a linear regression was obtained with a corresponding r2 value
(Table 6). The r2 values are high values
(reasonably close to 1), thus the data
have a positive linear relationship with
the high density values producing the
higher property values. A log transform
model was also investigated since it has
been used by others (Geimer 1979, Carll
77

3-inch-long strands. Geimers results


were obtained based on an estimated
strand alignment of 35 percent, which
results in an estimated sonic velocity ratio of 1.59. Geimers strength values for
1.5- and 3-inch-long strands are essentially the same, so only the 3-inch results
are plotted.
Several observations can be made
from Figure 8. Although Geimer (1979)
used a log transform model, the relationship between the mechanical properties
and density is essentially linear over the
range of interest. The slope of the
MOE-density relationship and the
strength-density relationship from this
work is similar to that from Geimer
(1979). Geimers data shows a much
greater stiffness and strength than that
measured in this work.

Figure 5. Failure photo of specimens cut perpendicular to panel length (bottom),


specimens L1, L5, L9, and L15 (L1 and L15 perpendicular to strand orientation, L5
and L9 parallel to strand orientation).
Table 5. Layer properties for parallel to panel length compression specimens.

Layer

Strand
orientation

No. of
specimens tested

Density
Mean

SDa

- - - - (pcf) - - - -

MOE
Mean

COV

- - - - - - - - - (psi) - - - - - - - - - -

Parallel

13

53.8

1.89

2606 0.33 5.12E + 05

0.25

Parallel

13

49.3

1.92

1835 0.30 3.79E + 05

0.31

Parallel

13

44.2

1.61

1128 0.38 2.05E + 05

0.36

Perpendicular

13

39.7

1.09

486 0.35 7.29E + 04

0.38

Perpendicular

13

37.4

0.56

429 0.39 5.58E + 04

0.28

Perpendicular

13

36.3

0.36

345 0.35 4.90E + 04

0.19

Perpendicular

12

35.6

0.30

365 0.38 5.13E + 04

0.27

Perpendicular

13

35.0

0.28

334 0.66 4.84E + 04

0.38

Perpendicular

13

35.0

0.30

367 0.37 5.13E + 04

0.31

10

Perpendicular

13

35.6

0.42

331 0.26 4.34E + 04

0.19

11

Perpendicular

13

36.7

0.54

589 0.58 8.37E + 04

0.54

12

Perpendicular

13

39.0

0.96

753 0.48 1.32E + 05

0.55

13

Parallel

13

42.6

1.37

1609 0.26 2.99E + 05

0.30

14

Parallel

13

46.9

1.93

1877 0.31 3.69E + 05

0.22

15

Parallel

13

51.5

1.42

1814 0.48 3.63E + 05

0.40

12.9

41.2

1.00

991

0.33

Average
a SD

0.39 1.81E + 05

= standard deviation.

and Link 1988). Although both the linear and log transform fit the data about
equally well, the r2 values for the linear
model averaged 3 percent higher than
the r2 values for the log transform
model. Thus, the linear model was used.
78

Strength
Mean COV

MOE and strength data for tension


parallel to the strand orientation is compared to other models in Figure 8. The
model from Xu (1999) is for aspen OSB.
The results from Geimer (1979) are for
Douglas-fir OSB with both 1.5- and

Prediction of panel properties


Fundamental engineering mechanics
relationships can be used to predict both
the stiffness and strength of the panel
from the layer properties (e.g., Geimer
et al. 1975). The predicted panel MOE
from the layer properties was 61 percent
of the measured panel MOE for parallel
to panel length, and 47 percent of the
measured panel MOE for perpendicular
to panel length. Geimer et als (1975)
predicted panel MOE was also less than
measured, averaging 87 percent of the
measured panel MOE. Carll and Link
(1988) overpredicted panel MOE by 10
to 15 percent.
In an attempt to characterize the
through-the-thickness mechanical behavior of OSB, much thinner layers were
used in this study than in previous studies. This probably affected the results.
As thin layers are removed from the
full-thickness panel, several of the
strands become severed, with a portion
going to each adjacent layer, leaving
only partial strands to contribute to
strength and stiffness. If these partial
strands were neglected, only the complete strands would serve as the effective
thickness of the layer. The average
strand thickness was approximately half
the specimen thickness. Thus, an effective thickness could be considered as
half the specimen thickness, which
would explain the low predictions.
The predicted panel strength from the
layer properties was 20 percent of the
measured strength for parallel to panel
length and 25 percent of the measured
JUNE 2003

Table 6. Regression equations for tension/compression properties vs. layer density.a


MOE vs. density

Perpendicular

MOE = 13500 (DEN) - 351600

Parallel

MOE = 30220 (DEN) - 662200

Perpendicular
Parallel

Strength vs. density

0.804

STR = 27.14 (DEN) - 778

0.929

0.824

STR = 84.84 (DEN) - 2554

0.890

MOE = 8220 (DEN) - 234900

0.880

STR = 40.39 (DEN) - 1016

0.797

MOE = 13230 (DEN) - 278200

0.777

STR = 75.12 (DEN) - 1798

0.845

Tension

Compression

aUnits

are psi for modulus of elasticity (MOE) and strength (STR) and pcf for density (DEN).

strength for perpendicular to panel


length. The strength prediction was
about half as accurate as the stiffness
prediction. This is consistent with the
findings of Geimer (1979), who found
that the bending MOR was almost twice
the tension and compression strength.
Although we do not have a complete explanation for this, we suggest that part of
the reason is nonlinear behavior near ultimate. A shift in the neutral axis towards the compression face near ultimate would cause an underestimation of
the strength using linear elastic theory.

Conclusion

Figure 6. Regression relationship between compression and tension MOE vs.


layer density.

Figure 7. Regression relationship between compression and tension strength vs.


layer density.
FOREST PRODUCTS JOURNAL

Vol. 53, No. 6

The sawing technique, as described in


this study, was successfully used to prepare individual 0.047-inch-thick layers
through the thickness of a commercial
OSB panel. These layers provided tension/compression strength and stiffness
values. For specimens parallel to the
length of the panel, the face layers had
tensile and compressive properties approximately an order of magnitude
greater than the core. For perpendicular
samples, the tensile and compression
properties were relatively uniform
through the thickness. This behavior
was due to a combination of strand orientation and density. The average compression strength was significantly
higher than average tension strength.
However, the average compression
MOE was significantly lower than average tension MOE. These layer tension
and compression properties were related
to the vertical density profile with high
r2 values (>0.75), thus indicating a
strong linear relationship exists. The
layer properties were used to predict
panel bending properties.
This study enhances the understanding of the mechanical behavior of OSB
panels. Coupling this understanding of
how the vertical density profile affects
the through-the-thickness mechanical
properties with research on the effects of
the VDP on other parameters (such as
79

Geimer, R.L. 1979. Data basic to the engineering design of reconstituted flakeboard. Proc.
13th Inter. Particleboard/Composite Materials Symposium. Washington State Univ.,
pp. 105-125.
___________, H.M. Montrey, and W.F.
Lehmann. 1975. Effects of layer characteristics on the properties of three-layer particleboards. Forest Prod. J. 25(3):19-29.
Grant, D. 1997. Effects of the through-thickness
strand alignment distribution on the unidirectional bending properties of oriented strand
board. Masters thesis. Laval Univ., Quebec,
Canada.
Steidl, C.M. 2000. Layer properties of oriented
strandboard. Masters thesis. The Univ. of
Tennessee, Knoxville, TN.
Strickler, M. 1959. Effect of press cycles and
moisture content on Douglas-fir flakeboard.
Forest Prod. J. 9(7):203-215.

Figure 8. Comparison of tension parallel to strand MOE and strength to other


data.

IB and thickness swell), the effects of


changes in the VDP can be ascertained.
The information can be used to enhance
manufacturing and the development of
optimal vertical density profiles.

Literature cited
American Society for Testing and Materials
(ASTM). 1994. Standard test methods for
wood-based structural panels in compression.
ASTM D 3501-94. ASTM, West Conshohocken, Pa.
__________1995. Standard test method for tensile properties of polymer matrix composite

80

materials. ASTM D 3039M-95a. ASTM,


West Conshohocken, Pa.
__________1996. Standard test methods for
evaluating properties of wood-base fiber and
particle panel materials. ASTM D 1037-96a.
ASTM, West Conshohocken, Pa.
Andrews, C.K. 1998. The influence of furnish
moisture content and press closure rate on the
formation of the vertical density profile in oriented strandboard: Relating the vertical density profile to bending properties, dimensional
stability and bond performance. Masters thesis. The Univ. of Tennessee, Knoxville, TN.
Carll, C.G. and C.L. Link. 1988. Tensile and
compressive MOE of flakeboard. Forest Prod.
J. 38(1):8-14.

Wang, S. and P.M. Winistorfer. 2000. Fundamentals of vertical density profile formation
in wood composites. Part 2. Methodology of
vertical density formation under dynamic
conditions. Wood and Fiber Sci. 32(2):
220-238.
Xu, W. 1999. Influence of vertical density distribution on bending modulus of elasticity of
wood composite panels: A theoretical consideration. Wood and Fiber Sci. 31(3):277-282.
__________ and O. Suchsland. 1998. Modulus
of elasticity of wood composite panels with a
uniform vertical density profile: A model.
Wood and Fiber Sci. 30(3):293-300.
__________and P.M. Winistorfer. 1995. Layer
thickness swell and layer internal bond of medium density fiberboard and oriented strandboard. Forest Prod. J. 45(10):67-71.
_________, _________, and W.W. Moschler.
1996. A procedure to determine water absorption distribution in wood composite panels.
Wood and Fiber Sci. 28(3):286-294

JUNE 2003

You might also like