|
Strain (mm/mm)
St
re
ss (
M
Pa
)
Specimen1
Specimen2
Specimen3
Specimen5
Specimen6
Specimen7
Specimen8
Figure 3.3. OSB stress versus strain for different specimen.
0
5
10
15
20
25
30
35
40
45
50
0
0.05
0.1
0.15
0.2
0.25
0.3
Strain (mm/mm)
St
re
ss (
M
Pa
)
Specimen1
Specimen2
Specimen3
Specimen5
Specimen6
Specimen7
Specimen8
Figure 3.4. Zoom-in of Figure 3.3.
50
4.
Finally, the particle locations of the compacted mats were input into a new MPM
simulation for tensile loading (in the horizontal direction of Figure 3.5). The
tensile simulation used fiber angles from the undulating strands and implemented
imperfect interfaces between strands using imperfect interface methods for MPM
(Nairn 2007A). The numerical calculations gave tensile properties as a function of
mat compaction and interface parameter
D
t
. The various values of D
t
spanned the
range from perfect interface (1/
D
t
= 0) to beyond experimental results for D
t
of PF
resin (Nairn and Le 2009).
5.
Because the mats were randomly created, all simulations were repeated for five
randomly selected initial mat structures. Error bars of some curves show standard
deviation or 95 percent confidence interval range of modulus results from the
various structures.
For numerical temporal convergence, the MPM time step was set to 0.4 d/c,
where d is the dimension of the elements in the background grid and c is the maximum
wave speed in the strand. For wood, the maximum wave speed is in the longitudinal
direction and for unmodified and VTC strands it is:
s
m
m
kg
GPa
E
V
/
5328
/
350
936
.
9
3
=
=
=
ρ
and
s
m
m
kg
GPa
E
V
VTC
/
5136
/
910
24
3
=
=
=
ρ
.
Four material points were used for each background element or cell. For spatial
convergence, there has to be a sufficient number of particles in the thickness direction of
each strand. The simulations that were first tried had one cell or two particles in the
thickness direction, but the results did not work well. They failed to resolve the glue-line.
We then moved to 4, 6 and 8 particles. The simulations with 6 and 8 particles gave
reasonable and similar stress, strain and strand undulation at higher levels of compaction.
However, the simulations with 8 particles took too long to finish. Therefore, all
simulations used 6 particles (3 cells) across the thickness of each strand.
51
The OSB structures were generated based on the strand length, and width, strand
gaps, and their standard deviations. In this study (this chapter), the strand length was 150
mm with standard deviation of 20 mm. The strand end to end spacing or gap was 15 mm
with standard deviation of 4.95 mm. In the core plies, the strand width was 25 mm with
standard deviation of 3 mm. The strand side-to-side spacing or width gap was 10 mm
with standard deviation of 1 mm. Figure 3.5 shows a sample MPM simulation of
commercial OSB at different levels of compaction, from 0 percent compaction to 40
percent compaction. There was more undulating of strands as the level of compaction
increased.
52
a)
0% compaction
b)
10% compaction
c)
20% compaction
d)
40% compaction
Figure 3.5. Sample simulation of OSB at different levels of compaction.
3.5.2 Tension Method
The OSB structures were compacted to different levels of densification with the
rate of 2m/sec. Archives of MPM results were saved at increments of 2% compaction.
The output of mat structures for specific levels of compaction were then simulated by
loading them in tension. The loading in tension was done for compactions of 0%, 10%,
20%, 30% and 40%. The tension results were used to find OSB tensile modulus. To keep
53
the tension test in the elastic region, the tensile load was applied only for very small
strain and the plastic energy was tracked during the analysis. During elastic loading, the
total elastic energy for a composite with imperfect interfaces is (Hashin 1990):
dS
u
D
dV
C
U
t
t
S
V
tot
2
]
[
2
1
2
1
int
∫
∫
+
=
σ
σ
(3.4)
where
U
tot
is the total energy that includes elastic strain energy (first term) and interface
energy (second term),
V
is volume,
C
is
stiffness matrix,
σ
is stress tensor,
S
int
is
interfacial surface area, D
t
interfacial stiffness parameter, and [u
t
] is slippage
(displacement) along the interface
.
The MPM output was set to record both the elastic
strain energy and the interface energy. From the output, total potential energy was used to
calculate the modulus of elasticity. By definition, the total energy during tensile loading
is related to effective modulus in that direction,
eff
E
, by
Lth
E
U
eff
tot
2
2
1
ε
=
(3.5)
where
t
is thickness =1mm,
h
is compressed height, and
L
is length of specimen in mm.
Rearranging gives
Lth
U
E
tot
eff
2
2
ε
=
(3.6)
The output results for interfacial energy and elastic strain energy were summed and used
to calculate modulus. Simulations were run for various D
t
and for different levels of
compaction. The results of elastic modulus as a function D
t
for different levels of
compaction are disused below.
3.5.3 Verify the Model
To verify the simulation method for MOE, a test simulation was run on a
homogenous specimen with known MOE. Figure 3.6 shows an MPM model for a block
of material loaded in tension in the x direction. The input MOE was 11000 MPa, Poison’s
ratio was 0.33 and density was 0.5 gram/cm
3
. To calculate MOE, the strain energy was
recorded (here there was no interfacial energy). Figure 3.7 plots effective elastic modulus
54
(MOE) by Eq 3.6 versus time. At early times there are dynamic effects. But after t =
0.021 sec, the effective modulus levels off and is close to the input MOE (11000 MPa).
To understand the time to level off, the transit time for stress waves was calculated. The
axial wave speed for a tensile stress wave is:
Do'stlaringiz bilan baham: |
