21
were made, outer strands had the same MOE and thickness (E
1
and t
1
). The central strand
may be different (E
2
and t
2
). The stiffness k for each specimen was determined from
slope of the load versus displacement plot for the specimen.
Interfacial
stiffness per surface area
D
t
(MPa/mm) was extracted using shear-lag
theory with an imperfect interface. Let k be measured specimen stiffness, then
2
1
2
2
2
)
2
1
(
1
L
R
L
R
C
C
W
E
t
k
+
+
+
=
∞
λ
λ
(Eq. 2.2)
where
(
)
(
)
+
+
+
+
−
+
=
∞
2
4
2
1
4
4
2
1
2
)
2
1
(
1
2
2
2
2
2
β
β
λ
β
βλ
β
λ
β
λ
l
Csch
l
R
t
dkE
lR
l
Tanh
R
l
Tanh
R
C
C
C
/
∞
C
is the ratio of the compliance of the bonded section of the specimen relative to the
compliance where the three strands equally deform as a unit.
Here
dk
G
t
G
t
E
t
RE
t
+
+
+
=
2
2
1
2
2
2
2
2
6
3
2
1
λ
λ
β
is the
shear-lag parameter, which in modern shear-lag theory
can account for interfacial stiffness properties (Nairn 2004, 2007A).
t
D
dk
1
=
,
2
1
t
t
=
λ
,
2
1
E
E
R
=
,
)
2
1
(
2
2
R
WE
t
l
C
λ
+
=
∞
The only unknown in k is
D
t
or the interfacial property. Thus given experimental results
for k and strand E and t,
D
t
can be calculated.
Figure 2.3 shows a sample shear-lag calculation of DLS specimen k as a function
of interfacial stiffness. The stiffness deceases as 1/
D
t
increases.
Given experimental
results for specimen k,
D
t
can be found by numerical solution. The process is illustrated
graphically in Figure 2.3 where “Expt k” is mapped to the curve, to find 1/
D
t
.
The dotted
lines illustrate propagation of error in k (
∆
k) to error in 1/
D
t
(
∆
(1/
D
t
)). To minimize
error in 1/
D
t
, the curve should be as steep as possible. One way to achieve this goal is to
keep
bond length, l, as short as possible. All bond lengths here had l = 25 mm, which was
22
thought to be as short as possible while still having samples with reliable bonds and with
ease of use in the mechanical tests.
Figure 2.3. Analysis of interfacial stiffness.
2.2.4 Statistical Analysis of Experimental Data
Statistical analysis for these data was performed with standard one sample t-test
using S-PLUS statistical software Version 8.0 and Microsoft Excel 2003 statistical
analysis tools. All comparisons of interfacial properties and MOE were based on a 95%
confidence level.
An analysis of variance (ANOVA) comparing different adhesive surface coverage
was performed within the treatments and found that the
interfacial property had
statistically significant different results.
500
900
1300
1700
2100
2500
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
Do'stlaringiz bilan baham: