Professional Documents
Culture Documents
rod
= U
r
rod
E
22
rod
/ r
rod
( 1
23
rod
) (1)
3
where
rr
rod
and
rod
are the radial and tangential stress components in the rod, r
rod
is
the
radius of the rod, U
r
rod
is the radial displacement on the surface of the crimped rod, E
22
rod
is the Youngs modulus of the rod transverse to the fibers, and
23
rod
is the Poissons ratio
in the cross-section of the rod.
The mechanical strength of a composite insulator with crimped end-fittings can be
determined by pulling the GRP rod out of the end-fittings [3-6, 14-16]. The insulator
strength will depend on the amount of compression applied during manufacturing, the
frictional properties of the rod/fitting interface and the geometry of the fitting. There are
three important factors related to the fitting geometry which strongly influence the
insulator strength. The first factor is the length of the fitting. Obviously, long fittings
provide higher insulator strength than short ones. The second important factor is the size
of a gap between the internal surface of the fitting and the rod surface before crimping is
applied. The third factor is the internal surface of the fitting. Smooth surfaces will require
larger compression deformations applied to the fitting and the rod surface in comparison
with rough surfaces (for example threaded surfaces) for the same strength properties of
the insulators.
The other factor which also can have a significant affect on the insulator strength is the
material of the fitting. Since the mechanical bonding between the GRP rod and fitting in
crimped composite insulators is achieved due to the plastic deformation of the fitting
material during crimping, the effect of plasticity of the fitting on the insulator strength
could be significant. Different elasto-plastic properties of the fitting materials can
noticeably influence the internal stress distributions in the GRP rod and thus the insulator
strength for the same amount of compression applied to the fitting.
2. Analysis of Field-Failed Suspension Insulators due to Improper Crimping
Improper crimping procedures can have serious effects on the short- and long-term
strengths of composite suspension insulators. Two examples of damage caused by
improper crimping procedures will be shown and discussed in this section. If crimping
deformations in the GRP rods generated during manufacturing are excessive, the
composite rod can be significantly damaged inside the hardware. Multiple cracks can be
generated inside the rod (inside the fitting) near the rod/hardware interface. An example
of this type of damage is illustrated in Figure 2.
Another example of insulator damage due to improper manufacturing is illustrated in
Figures 3a and 3b. A 115 kV suspension composite insulator that failed in service by rod
fracture was examined. The insulator failed under relatively low tensile stresses, much
lower than the tensile strength of the GRP rod in longitudinal tension. The damage zones
presented in Figure 3a and 3b consists of numerous large transverse and longitudinal
cracks in the GRP rod. The mechanical failure of the rod occurred just above the top
surface of the end-fitting. Most importantly, the transverse cracks in the rod are planar in
nature (see Figure 4) and closely resemble the brittle fracture cracks previously reported
[1,18,19,21], if examined macroscopically. However, under higher magnifications
significant differences were observed between the surface fracture morphologies of the
4
insulators failed by brittle fracture and the damage in the insulator shown in Figure 3a
and 3b. Under higher magnification (see Figure 4) the fracture surfaces are still planar
without however flat fracture surfaces of the individual fibers (see Figures 5a and 5b). It
should be pointed out here that when suspension insulators fail in service by brittle
fracture, the fracture surfaces of glass fibers are also planar, however, consisting of
clearly defined mirror and hackle zones. It can be clearly seen in Figures 5a and 5b that
the fracture surfaces of the glass fibers are certainly not planar, and are highly irregular.
This effect can be even better observed by examining the SEM micrographs presented in
Figures 6 and 7. In Figures 7a and 7b the most typical surface fracture morphologies of
individual fibers from the insulator in Figures 3a and 3b are illustrated. Not only the
fracture surfaces are highly irregular but very often they consist of longitudinal cracks in
the glass fibers (see Figure 7a). If compare with a typical fiber fracture caused by stress
corrosion (brittle fracture, see Figure 7c [21]) the differences between these two fracture
modes are apparent.
In order to determine a potential cause of failure of the insulator shown in Figures 3-7 a
unidirectional E-glass/epoxy specimen was mechanically damaged by applying large
tensile and bending forces. Subsequently, the fracture surface of the specimens was
examined by SEM. An SEM micrograph of this fracture surface is shown in Figure 8.
Clearly, the microscopic features of the surfaces presented in Figures 5-8 are very
similar. In particular, the micrographs shown in Figure 5b, 6b and 8 appear to be almost
identical. Therefore, it was concluded that the damage to the rod in the insulator shown in
Figure 3a and 3b was caused by a combination of manufacturing crimping stresses and
mechanical tensile stresses concentrated in the rod near the fitting in service.
The mechanically damaged insulators illustrated in Figures 2 and 3 were provided to this
research by US electrical utilities as examples of mechanical failures of suspension
composite insulators. These types of failures have not been detected in insulators
manufactured by NGK.
3. Finite Element Computations
In order to explain mechanical failures of crimped composite insulators a detailed
analysis of internal stresses in their crimped ends is required. Therefore, a crimped
composite suspension insulator was modeled in this study using ANSYS 5.2 [24]. Two-
and three-dimensional axisymmetric models of the insulator end were generated, as
illustrated in Figures 9 and 10. The 3-D model was composed of 8-node brick elements.
The 2-D axisymmetric model used 8-node isoparametric elements. The point-to-surface
contact elements were placed between the rod surface and the internal surface of the
fitting. The finite element computations were performed assuming geometric non-
linearity as well as elastic and elasto-plastic properties of the fitting material. The
composite rod was assumed to be linear-elastic.
The geometry of the finite element model follows the dimensions shown in Figure 9. The
following elastic properties of the GRP rod were assumed:
5
E
11
= 49 GPa, E
22
= 13 GPa, G
12
= 4.8 GPa, G
23
= 4.3 GPa,
12
= 0.34,
23
= 0.5.
where E
11
is the longitudinal Youngs modulus, E
22
is the transverse Youngs modulus,
G
12
and G
23
are the shear moduli and
12
and
23
are the Poisson ratios of the GRP rod
material.
The end fitting was assumed to be made out of steel with the following properties:
E = 213600 MPa, = 0.3,
Y
= 300 MPa,
U
= 400 MPa and
U
= 0.2.
where E is the Youngs modulus and v is the Poisson ratio of the steel as provided by
NGK whereas
Y
,
U
and
U
are the yield strength, tensile strength and elongation at
failure, respectively. The plastic and strength properties of the fitting were estimated from
the open literature for this particular steel.
The analysis was divided into two steps, that is, crimping and pulling. In the first step,
crimping was simulated on the surface of the end-fitting. In the second step, the rod was
gradually pulled out of the end-fitting. The stresses in the GRP rod caused only by
crimping as well as crimping and pull-out can be derived from the two-step finite element
simulations. In addition, the load/displacement curves from the numerical pull-out can be
determined.
The actual crimping deformations applied during manufacturing were measured on the
surface of the fitting of an insulator with the standard stress conditions from NGK [14,16]
by comparing the diameters of the fitting in the compressed (plastically deformed) and
uncompressed areas (third method described in Section 1.3). The measurements were
performed in 24 locations on the surface of the fitting and then the average compression
was determined. The crimping appeared to be uniform around the circumference of the
end-fitting, and was almost uniform along the longitudinal direction with a small
reduction close to the top surface of the fitting. At first, a perfectly uniform crimping
profile was assumed (see Figure 9(a)) and the crimping was simulated via prescribed
displacements applied onto the external surface of the fitting. The value of the friction
coefficient at the rod/fitting interface was taken as 0.3 (measured and provided by NGK).
The plastic deformation of the fitting during crimping was modeled assuming an elasto-
plastic model with work-hardening presented in Figure 11. Secondly, the crimping
conditions in the finite element analysis were modified to simulate linearly distributed
crimping deformations along the length of the fitting.
4. Results and Discussion
4.1. Load-Displacement Diagrams
The numerical and experimental load-displacement diagrams are shown in Figure 12. The
experimental curve was obtained by subjecting the insulator manufactured by NGK with
the standard crimping conditions to axial tension on an MTS testing machine [14,16].
The insulator failed by rod pull-out. After the mechanical test, the GRP rod was carefully
6
examined. No damage of any kind was detected on the surface of the rod. It can be seen
in Figure 12 that the numerically determined load-displacement diagram from the fully
non-linear finite element computations is very close to the experimental curve for this
particular insulator. The difference between the maximum loads, obtained experimentally
and numerically, is only 2.9%.
The frictional forces between the rod and the end-fitting increase as the pulling force is
increased. The highest load takes place right before the entire rod slides inside the end-
fitting. It can be seen in Figure 12 that there is a small drop on the experimental curve at
the highest point. This might be caused by the fact that the friction coefficient during the
pull-out experiment is not constant. Most likely, the drop in the load on the experimental
curve coincides with the transition from static to dynamic friction. In the finite element
model, a constant friction coefficient was assumed therefore the numerical curve does not
exhibit a similar load drop. When the rod starts sliding out of the fitting, the contact area
between the fitting and the rod decreases. Thus, lower pulling forces are required to pull
the rod out of the fitting after the maximum load is achieved.
The numerical and experimental load-displacement curves shown in Figure 12 are very
close to each other. It can therefore be concluded that the numerical computations of the
rod-pull out experiment closely simulate the actual experimental conditions. Thus, the
internal stress distributions in the insulator end can be determined from the finite element
analysis with a high degree of accuracy.
4.2. Internal Stresses and Strains in the Insulator
The strain and stress distributions in the insulator end after crimping, calculated using the
3-D model, are presented in Figure 13(a-c). The plasticity of the metal end-fitting was
included. The equivalent stress criterion was used as the criterion of plasticity. If the
equivalent stress in an element exceeds the yield strength of the metal (which was
assumed to be 300 MPa) the element is considered to be plastically deformed. The
accumulated effective plastic strains, as defined in reference [24], are shown in Figure
13(a) whereas the stress components
rr
and
at the
highest pull-out load on the numerical curve shown in Figure 12. As illustrated in Figure
15, the stress components
rr
,
and
z
are no longer uniformly distributed along the z
direction. The tensile stress
z
along the rod inside the fitting increases along z, and is
proportional to the integration of
rz
, which represents the friction along the interface
between the rod and the end-fitting:
z
=
2
0
r
rod
z
rz
dz (2)
The stresses in the fitting are not a major concern, whereas the stresses in the GRP rod
are critical for this research. One apparent phenomenon is that the stresses in the rod are
almost constant in the radial direction. The stresses on the rod surface are plotted in
Figure 16. As shown in Figure 16(a),
rr
and
Y
Strain
Stress
Figure 11. Elasto-plastic finite element model for the end-fitting material.
33
0
50000
100000
150000
200000
0 2 4 6 8 10 12
Displacement (mm)
T
e
n
s
i
l
e
L
o
a
d
(
N
)
Experimental
FEM
Figure 12. Experimental and numerical load-displacement diagrams from rod pull-out with uniform crimping.
34
( a ) ( b ) ( c )
Figure 13. Strains and stresses in the end fitting after crimping for the 3-D model; (a) accumulative effective plastic strain, (b)
rr
and
(c)
.
35
Figure 14: Strains and stresses in the insulator after crimping for the 2-D model; (a) accumulated
effective plastic strain, (b)
rr
and (c)
.
A = -450 MPa
B = -350 MPa
C = -250 MPa
D = -150 MPa
E = -100 MPa
F = -50 MPa
G = 0
H = 100 MPa
( c )
A = -450 MPa
B = -350 MPa
C = -250 MPa
D = -150 MPa
E = -100 MPa
F = -50 MPa
G = 0
H = 100 MPa
( b )
A = 0.001
B = 0.002
C = 0.004
D = 0.006
E = 0.008
F = 0.012
G = 0.016
H = 0.020
( a )
36
Figure 15: Axial
z
, radial
rr
and tangential
A = -450 MPa
B = -350 MPa
C = -250 MPa
D = -150 MPa
E = -100 MPa
F = -50 MPa
G = 0
H = 100 MPa
( c )
A = -450 MPa
B = -350 MPa
C = -250 MPa
D = -150 MPa
E = -100 MPa
F = -50 MPa
G = 0
H = 100 MPa
( b )
A = 50 MPa
B = 100 MPa
C = 200 MPa
D = 300 MPa
E = 500 MPa
F = 700 MPa
G = 1000 MPa
H = 1400 MPa
( a )
37
-300
-200
-100
0
0 20 40 60 80 100
z (mm)
(MPa)
( a )
-400
-200
0
200
400
600
800
1000
0 20 40 60 80 100
z (mm)
(MPa)
( b )
Figure 16. Stresses on the rod surface (a) after uniform crimping and (b) at the
highest load during pull-out with plasticity
z
rr =
rr =
z 0 rz 0
z = r = 0
rz
z = r = 0
Fitting GRP Rod
38
-300
-200
-100
0
0 20 40 60 80 100
z (mm)
(MPa)
( a )
-400
-200
0
200
400
600
800
1000
0 20 40 60 80 100
z (mm)
(MPa)
( b )
Figure 17. Stresses on the rod surface (a) after uniform crimping and (b) at the
highest load during pull-out without plasticity
z
rr =
rr =
rz
z = r = 0
Fitting GRP Rod
z 0 rz 0
z = r = 0
39
0
50000
100000
150000
200000
250000
300000
0 2 4 6 8 10 12
Displacement (mm)
T
e
n
s
i
l
e
L
o
a
d
(
N
)
Experimental: standard stress condition
FEM: uniform crimping with plasticity
FEM: uniform crimping, no plasticity
FEM: type 1 linear crimping, no plasticity
FEM: type 2 linear crimping, no plasticity
Figure 18. Experimental and numerical load-displacement diagrams during rod pull-out for the end fittings with different crimping
profiles.
40
Figure 19a. Deformation of the rod surface after uniform crimping.
41
Figure 19c. Deformation of the rod surface after type 2 linear crimping.
42
Figure 19b. Deformation of the rod surface after type 1 linear crimping.
43
-1000
-800
-600
-400
-200
0
0 20 40 60 80 100
z (mm)
(MPa)
( a )
-800
-400
0
400
800
1200
1600
0 20 40 60 80 100
z (mm)
(MPa)
( b )
Figure 20: Stresses on the rod surface (a) after type 1 liner crimping and (b) at the
highest load during pull-out without plasticity.
z
rr =
rr =
z 0
rz 0
z = r = 0
rz
z = r = 0
Fitting GRP Rod
44
-800
-600
-400
-200
0
0 20 40 60 80 100
z (mm)
(MPa)
( a )
-800
-400
0
400
800
1200
0 20 40 60 80 100
z (mm)
(MPa)
( b )
Figure 21: Stresses on the rod surface (a) after type 2 liner crimping and (b) at the highest load
during pull-out without plasticity.
z
rr =
rr =
z 0 rz 0
z = r = 0
rz
z = r = 0
Fitting GRP Rod