Professional Documents
Culture Documents
Designing A Strand Orientation Pattern For Improved Shear Properties of OSB
Designing A Strand Orientation Pattern For Improved Shear Properties of OSB
Abstract: As oriented strand board (OSB) increases in use, improving the mechanical properties is of importance. Improved in-plane shear
properties will allow for more efficient use as well as open up other opportunities in engineered wood products with high shear stresses. Based
on classical laminated plate theory, composite laminates with 45° laminate alignment patterns produce higher in-plane shear modulus and
strength compared with typical 0°=90°=0° alignment. Oriented strand board was manufactured 13.3-mm thick with 0°= þ 45°= − 45°= −
45°= þ 45°=0° and 0°=90°=0° alignment patterns, and in-plane shear, bending, nail connection, and small-scale shear wall properties com-
pared to commercial OSB. Results showed a 24% increase in shear modulus for the 0°= þ 45°= − 45°= − 45°= þ 45°=0° alignment when
compared with the 0°=90°=0° alignment using a method similar to a standard in-plane shear test. Results also show 10% reduction in bending
modulus of elasticity in the parallel direction. Small-scale shear wall tests were insensitive to changes of in-plane shear properties. Lateral nail
connection tests showed no reduction in connection yield load, implying that these panels can be used in similar applications without affecting
connection properties. DOI: 10.1061/(ASCE)MT.1943-5533.0001033. © 2014 American Society of Civil Engineers.
Author keywords: Wood; Digital imaging techniques; Nails; Shear modulus; Shear walls.
Materials and Methods vanes spaced at 50 mm, as shown in Fig. 1. The vanes were
130 mm deep. The forming box was 910 by 910 mm with the align-
Manufacturing of OSB ment screen having an adjustable height above the surface of the
strand mat. Free-fall distance of the wood strands from the bottom
A six-layer alignment pattern of 0°= þ 45°= − 45°= − 45°= þ of the alignment vanes to the surface of the mat was not greater than
45°=0° and a typical 3-layer pattern of 0°=90°=0° were manufac-
the typical length of a wood strand (approximately 100 mm).
tured in the laboratory. Both panel types were manufactured with
Vibration was achieved from a variable speed electric motor with
the similar target density and similar furnish weight. Therefore, the
an off-center counterweight attached to the side of the forming box
0°= þ 45°= − 45°= − 45°= þ 45°=0° panel had 1=6 of the total
(Fig. 1). The speed of the motor was adjusted to achieve the greatest
panel weight for each layer, whereas the 0°=90°=0° panel had
amount of vibration of the forming screen. Two forming screens
1=3 of the total panel weight per layer. For this research, the 0°= þ
were used consisting of vanes set at 0° and 45°. The forming
45°= − 45°= − 45°= þ 45°=0° alignment pattern will be referred to
screens were removed, rotated 90°, and reinstalled to create the
as ½0= þ 45= − 45s, where “s” represents “symmetrical.” The pat-
different alignment patterns. Photos of the alignment patterns of
tern is symmetrical about the central axis. The 0°=90°=0° layup pat-
the OSB strands before pressing are shown in Fig. 2. A thin thermo-
tern will be referred to as 0=90=0. To preserve the bending
couple wire was placed in the center of the panel to monitor core
properties of the OSB, both alignment designs had the face layers
temperature during pressing. The weight of strands used in each
with 0° alignment. OSB also tends to warp after cooling from hot
panel was held constant. Typically, OSB has lower quality strands
pressing if the alignment pattern is not symmetrical about the
in the core than in the faces, resulting in a lower density core and a
central axis, so the ½0= þ 45= − 45s alignment prevents warping.
higher density face material. This, along with the pressing schedule
Wood Strand Preparation of the panel, leads to a vertical density gradient in the panels.
Aspen strands from the Great Lakes area were procured. Strands Similar quality strands were used throughout the panels manufac-
were dried to a moisture content of 6% to 8%. They were approx- tured in the current study; as a result, the variation in density
imately 100 mm in length, and the length-to-width aspect ratio of through the thickness is minimal. A consistent pressing schedule
the strands was about 4∶1. The average thickness of strands for all panels manufactured also minimized the variation in density
was 1.2 mm. through the thickness.
12 panels were used for small-scale shear wall tests. Sixteen specimens. Digital image correlation has been successfully used
commercial panels (10 for shear/bending and six for small shear for wood products by Sinha and Gupta (2009). It uses a pair of
walls) were cut to similar specifications as for the laboratory manu- high-definition digital cameras that image a surface coated with
factured panels. All panels, both manufactured and commercial, a contrasting black-and-white speckle pattern. A series of images
were conditioned in a standard room maintained at 65% relative are then captured during testing. The software measures the move-
humidity and 20°C until they were tested. At testing, the measured ment of a specified set of pixels in the image during the test and
average moisture content of the panels was 6%. The commercial correlates back to a reference image. The size of the specified set of
panels were on an average 11.5 mm thick with a specific gravity pixels used in this study was 21 × 21 pixels. For every iteration,
of 0.59. The commercial panels were manufactured by LP Building this square block of pixels moves over by 5 pixels. The cameras
Products and met the exposure 1 sheathing grade requirement. were calibrated using a surface with a known speckle pattern on
a special calibration plate. The cameras were set to capture an im-
In-Plane Shear Test age every second during the loading of the specimen. Each image
was tagged to specific load data received from the MTS 407
One specimen per panel was tested for in-plane shear based on the hydraulic controller. The DIC software then maps the surface by
ASTM D2719 Method C (ASTM 2007). The specimen was correlating the movement of the contrasted pixels to calculate
610 mm in height by 390 mm in width. The ASTM D2719 Method strain. Shear strain can be extracted from the DIC output data.
C procedure requires bonding heavy lumber rails to the long edges Thus, full-field shear strain contour plots were developed for the
of the shear specimen with adhesive. Steel brackets were bolted to in-plane shear and small-scale shear wall tests. Resolution of the
the specimen and were used in place of the adhesive-attached shear strain measured was 0.0002 strain. An assumption made
lumber rails, as shown in Fig. 3 [similar to the testing bracket used by using optical surface strain measurement was that the strain
in Shrestha (1999)]. The brackets were made from 19-mm-thick on the surface of the OSB material represented the strain through
steel plate with 7 holes in each for 13-mm-diameter bolts to clamp the thickness of the material due to strain compatibility. Factors that
the brackets to the specimen. Holes were drilled through the speci- would cause local variations in the surface strains include the
men for the bolts to clamp the steel brackets together. The space presence of voids in the material, as well as resin and density
between the brackets was 200 mm. The brackets were then pulled distribution.
in tension to create a shearing force on the specimen. The specimen
was loaded at a rate of 1.3 mm per minute using an MTS 160-kN
hydraulic actuator (model # 244.23) on the universal testing In-Plane Shear Data Analysis
machine. Shear strain (εxy ) from the in-plane shear test was measured using
the DIC measurement technique. Shear modulus, Gxy , was calcu-
Digital Image Correlation lated using the following equation:
An optical, noncontract strain measurement system based on the
P 1
digital image correlation (DIC) technique was used to measure Gxy ¼ × ð1Þ
shear strain (εxy ) on the surface of the shear and wall test εxy L×t×2
where (P=εxy ) = slope of the plotted load versus shear strain curve Load and deflection at mid-span were recorded and bending
in the linear region, which was at a load between 18 and 27 kN; L = modulus of elasticity (MOE) and modulus of rupture (MOR) were
length of the specimen; and t = thickness of the specimen. calculated using Eqs. (3) and (4), respectively as recommended in
Shear strength was found using the following equation: ASTM D3043–Method B:
Pmax 7 × L3 P
τ xy ¼ ð2Þ MOE ¼ × ð3Þ
L×t 324 × I Δ
where Pmax = peak load measured on the specimen during the shear Pmax × L × t
test. The sample size for the in-plane shear tests was 10 for each MOR ¼ q ð4Þ
12 × I
panel type (n ¼ 10).
where L = span length; t = thickness of the specimen; I = moment
of inertia of the specimen; Pmax = maximum load on the specimen;
Bending Tests and P=Δ = slope of the load deflection curve. The span between the
The parallel (strong axis) and perpendicular (weak axis) bending load points was L=3.
tests followed the ASTM D3043–Method B (ASTM 2011) test The linear portion of the curve was used for MOE calculations
procedure. The tests were essentially three-point bending tests and was found to be at loads between 89 and 178 N for the parallel
conducted on an INSTRON Series 5582 universal testing machine. direction and between 89 and 133 N for the perpendicular direction.
Span length for the parallel bending tests was 558 mm and for the The peak load was used for MOR calculations. The thickness of
perpendicular bending tests was 330 mm. Length between the each specimen was used for MOE and MOR calculations.
load points for the parallel bending test was 190 mm and for
the perpendicular bending test was 110 mm. Span-to-depth
Nail Connection Test
ratios for the parallel and perpendicular tests were 44 and 26, re-
spectively. All bending test specimens were cut to approximately The lateral nail tests for connections between panels and dimension
50 mm in width per ASTM D3043. Load rate for the parallel lumber members were conducted in accordance to ASTM D1761
bending test was 6.8 mm= min and for the perpendicular span (ASTM 2006). Lateral nail resistance was determined for
was 0.094 mm= min. The load rate was calculated to keep the the 0=90=0, ½0= þ 45= − 45s, and commercial panels. Two types
extreme fiber strain rate limited to 0.0015 mm=mm= min in of tests were conducted, using 38 × 89-mm lumber loaded
accordance with ASTM D3043 (ASTM 2011). Seven parallel perpendicular and parallel to grain, representing two typical con-
and six perpendicular bending samples were cut from each panel. nection configurations found in a shear wall, i.e., edge connections
Hence, the sample size for the bending tests was 70 for each panel to studs and top and bottom plate connections. The nails used were
type in the parallel direction (n ¼ 70) and 60 for each panel type in manufactured by Senco, Inc. (Cincinnati, Ohio) and were 75 mm in
the perpendicular direction (n ¼ 60). length and 3.8 mm in diameter. These 10d nails were hand driven
with edge distances for both tests of 19 mm. The loading rate was plateau observed in the curve. Eighteen walls were tested with six
5 mm= min. Specimen width was 50 mm, and one nail connection of each type of panel (0=90=0, ½0= þ 45= − 45s, and commercial).
test was performed for each panel.
Small-Scale Shear Wall Data Analysis
Nail Connection Test Data Analysis
The load versus deflection curve was recorded using the hydraulic
The nail tests were conducted using an INSTRON Series 5582 uni- cylinder deflection. Load versus deflection curves were compared
versal testing machine. Load and deflection data were recorded and for each wall type. The DIC image analysis was also performed on
analyzed to determine the yield point using the linear offset method the small-scale shear wall tests. Sample size for the small-scale
(American Forest and Paper Association 2012). The linear portion shear wall tests was 6 for each panel type (n ¼ 6).
of the load and deflection curve is offset by 5% of the diameter of
the nail shank. The line is then extended to find the yield load where
it intersects the original load versus deflection curve. The sample Results and Discussion
size for the nail connection tests was 10 for each panel type and
each direction (n ¼ 10). Shear Test
Average shear moduli for the 0=90=0, ½0= þ 45= − 45s, and com-
Small-Scale Shear Wall Test
mercial panels are presented in Table 1. Shear strain (εxy ) was mea-
The wall test was based on ASTM E564 (ASTM 2012), modified to sured using the DIC technique. The DIC software can average the
accommodate the 610 × 610-mm walls tested. The wall panels for shear strain over a specified area, called the “area of interest.” Area
0=90=0, ½0= þ 45= − 45s, and commercial specimens were nailed of interest for the average shear strain measurement for the shear
to 2 × 4 nominal (38 × 90 mm) select structural lumber studs with test was between the two bolts on the extreme ends of the testing
a single sill plate and single top plate. The nails used were 75 mm in brackets and between the steel rails. Average shear strain was plot-
length and 3.8 mm in diameter and were pneumatically driven. The ted against the corresponding force to produce a load versus shear
10d nails were driven with an edge distance of approximately strain (εxy ) curve. The slope of the curve in the linear region was
19 mm. Nail spacing was 100 mm around the entire wall. The sill used to calculate the shear modulus (Gxy ). A two-sample t-test as-
and top plates were bolted to the test machine. The sill plate was suming equal variances was performed on the results comparing the
rigidly fixed to the test frame base [Fig. 4(b)], and the top plate was 0=90=0 alignment with the ½0= þ 45= − 45s alignment. The ½0= þ
deflected vertically, causing a racking force on the specimen. 45= − 45s showed a 24% increase (p-value ¼ 0.001) in measured
Displacements were measured from the MTS hydraulic cylinder average shear modulus when compared with the 0=90=0 alignment
attached to the top plate. The specimen was also sprayed with a pattern. This result agrees with predictions from the CLPT calcu-
black-and-white speckle pattern [Fig. 4(a)] for the DIC strain mea- lations, which show an increase in shear modulus when the align-
surements. No stud hold-downs were used in the test. The test was ment is changed from 0=90=0 to ½0= þ 45= − 45s. Average
halted when the load-deflection curve reached a maximum with a measured shear modulus for the 0=90=0 layup pattern was 23%
Table 1. Summary of Test Results for In-plane Shear, Bending, and Small-Scale Wall Tests and Coefficients of Variation (COV %) and Sample Sizes for Each
Panel Type
In-plane shear Strong axis Weak axis Small-scale wall
test (n ¼ 6 bending (n ¼ 70 bending (n ¼ 60 test (n ¼ 6 per
Property per panel type) per panel type) per panel type) panel type)
Gxy COV T xy COV MOE COV MOR COV MOE COV MOR COV Ppeak COV K COV
Panel type (GPa) (%) (MPa) (%) (GPa) (%) (MPa) (%) (GPa) (%) (MPa) (%) (N) (%) (N=mm) (%)
0=90=0 1.22 16.8 9.16 20.1 9.53 20.5 28.43 25.6 2.48 21.9 13.23 27.0 4079 23.7 649 25.7
½0= þ 45= − 45s 1.52 10.3 9.50 11.7 8.50 14.6 26.35 25.8 2.69 24.4 13.99 24.8 4311 20.0 518 31.1
Commercial 0.99 6.50 6.83 5.20 9.82 14.1 27.49 21.8 2.09 14.5 9.14 19.6 3144 21.9 470 24.1
Bending Test
The ½0= þ 45= − 45s alignment pattern resulted in a lower average
bending modulus of elasticity (MOE) in the parallel direction
(strong axis bending) and an increase in average MOE in the
perpendicular direction (weak axis bending), as summarized in
Fig. 5. Typical in-plane shear failure
Table 1. The parallel direction is in line with the surface strands.
A two-sample t-test assuming equal variances was performed on perpendicular axis when compared with 0=90=0 alignment. The
the bending results comparing the 0=90=0 alignment with the ½0= þ MOR of the ½0= þ 45= − 45s panels showed a 7% reduction in
45= − 45s alignment. The ½0= þ 45= − 45s parallel direction the parallel direction (p-value ¼ 0.027) when compared with the
(strong axis bending) MOE reduced by 10% (p-value < 0.001) 0=90=0 alignment. There was a 5% increase in MOR in the
when compared with the 0=90=0 alignment. With a 0=90=0 align- perpendicular direction for the ½0= þ 45= − 45s alignment when
ment, 1=3 of the strands are oriented at 0° on the tension surface of compared with the 0=90=0 alignment; however, this result was
the bending specimen, whereas in the ½0= þ 45= − 45s alignment statistically inconclusive because of high variation. Statistically
pattern, only 1=6 of the strands are oriented at 0°. Strands oriented
inconclusive results in MOR were attributable to high horizontal
at 0° on the tension surface have a very significant impact on
density variability in the specimens. Horizontal density gradient
bending properties, and this reduction in strong axis bending
MOE is expected. The ½0= þ 45= − 45s alignment pattern could differences increase when alignment becomes more uniform, as
produce a product that needs high shear properties as well as in the laboratory panels as compared with commercial panels
good bending properties. The ½0= þ 45= − 45s perpendicular (Kruse et al. 2000). Horizontal density control in the panel manu-
direction (weak axis bending) MOE showed an 8% increase facturing process was achieved with visual inspection during the
(p-value ¼ 0.027) when compared with the 0=90=0 alignment strand alignment process. This process seemed to occasionally
pattern. This result is expected because of the ½0= þ 45= − 45s produce outliers in the data, thus causing difficulty in statistical
alignment pattern having more strands oriented toward the conclusions.
Fig. 8. Typical failure of bottom plate in small-scale shear wall test with: (a) side views; (b) end view