You are on page 1of 12

1> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1

Brittle Fracture Failure of Composite (Non-


Ceramic) Insulators
Maciej S. Kumosa

is a failure process called brittle fracture [1-26] which is


Abstract— In this work, additional input is provided into the caused by the stress corrosion cracking (SCC) of GRP rods
ongoing discussion on the mechanisms, causes and remedies of [27-41,52]. Several attempts have been made over the years
brittle fracture failures of composite (non-ceramic) insulators. [1-26] to understand the process and provide potential means
This has been done partially to complement the recently
published article on this topic entitled “ IEEE Task Force
of avoiding it in-service.
Report: Brittle Fracture in Non-Ceramic Insulators” [1]. At The reason why the brittle fracture problem has drawn so
first, the most important characteristics of the brittle process are much attention is not because of the number of brittle fracture
described in this study. Then, two important examples of failures, which has certainly not been significant [1], but
insulator failures by brittle fracture are shown and their causes because of its characteristic features, which are:
explained. Finally, several recommendations on how to avoid - the failure is catastrophic,
brittle fracture in service are presented to supplement those
already given in [1]. This work should improve the current state
- the failure is unpredictable; can occur after a few
of knowledge on brittle fracture of composite insulators as well months or after a few years of being in service,
as provide a potential means of avoiding it in-service. - there are no reliable monitoring techniques which
could be used to monitor composite insulators in
Index Terms—Brittle Fracture, Composite, Non-Ceramic, service for potential brittle fracture failures,
Polymer Insulators, Failure Analysis, Prevention. - the causes of failures, if one reads the available
literature on this topic, are uncertain due to a number of
confusing journal articles and conference presentations.
I. INTRODUCTION
B. Brittle fracture research at OGI and DU
A. Composite (non-ceramic) insulators Mechanical failure mechanisms of composite high voltage

C
OMPOSITE suspension insulators are used in insulators have been extensively studied at the Oregon
overhead transmission lines with line voltages in the Graduate Institute (OGI) [9-15,30, 44-46] and presently at the
range of 69 kV to 735 kV. The insulators, also called University of Denver (DU) [16,17,19, 21, 24-26,31-43,47,48].
as either non-ceramic, polymer or polymeric insulators This has been done in addition to the other research activities
rely on unidirectional glass reinforced polymer (GRP) on this topic [1-8,18,20,22,23]. In the insulator research at
composite rods as the principle load-bearing component. The OGI and DU particular attention has been given to the
rods are manufactured by pultrusion and the constituents are understanding of the brittle fracture process that can occur in
either polyester, vinyl ester, or epoxy resins reinforced with composite insulators in service. SCC, which causes brittle
either E-glass or ECR-glass (also called boron-free E-glass) fracture of unidirectional E-glass/polymer composites, is
fibers. In spite of many benefits, which the insulators can offer caused by the combined action of mechanical tensile stresses
in comparison with their porcelain counterparts (high along the fibers and a corrosive acidic environment [27-41].
mechanical strength-to-weight ratio, improved damage The brittle fracture research in our laboratory was initiated in
tolerance, flexibility, good impact resistance, and ease of 1992 as a result of numerous insulator failures on a 345 kV
installation), they can fail mechanically in service by rod line [9]. It was further expanded in 1996 after additional
fracture. One of the mechanical failure modes of the insulators insulator failures on a 500 kV line [14]. Since we had already
some past experience with SCC [27-29], we were selected to
Manuscript received March 1, 2005. This research has been supported perform this research.
since 1992 by the Electric Power Research Institute (under five consecutive The major accomplishments in this insulator research
contracts) and a consortium of electric utilities consisting of the Bonneville
Power Administration (under four contracts), Western Area Power regarding brittle fracture have been:
Administration (four contracts), Alabama Power Company (two contracts), - explanation of the 345 kV and 500 kV brittle fracture
Pacific Gas and Electric (three contracts), National Rural Electric Cooperative failures [9-16],
Association (one contract). It was also supported by Glasforms, Inc (one
contract) and NGK-Locke (one contract). - identification of the type of acid responsible for brittle
M. S. Kumosa is with the Center for Advanced Materials and Structures, fracture [17, 26],
Department of Engineering, University of Denver, 2390 S. York St., Denver, - simulation of brittle fracture with and without high
CO 80208 (phone: 303-871-3807; fax: 303-871-4450; e-mail:
mkumosa@du.edu).
voltage [10-12,15,19,24],
2> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2

- identification of several critical conditions leading to should not be immediately used as definite evidence of brittle
brittle fracture [9-17,19], fracture. According to [47,48] a crimped composite insulator
- providing a ranking of the commonly used GRP rod can fail mechanically in-service by rod fracture caused not by
materials for their resistance to stress corrosion brittle fracture but overcrimping. If crimped composite
cracking in nitric acid and thus brittle fracture insulators are incorrectly designed and manufactured
[21,19,36-40], (excessive crimping deformations, incorrect design of the
- recommendations of several stress corrosion tests for metal end fittings, excessive in-service mechanical loads, etc.)
the initiation [21,30,36-40] and propagation [10-12,19, failure of the rod can also occur very close to the end fitting
24,31-35] of brittle fracture cracks in unidirectional and the macroscopic fracture surface will be flat and almost
GRP composites. perpendicular to the long axis of the rod (Fig. 2)[47]. In this
case, the macro-fracture features will be almost
II. CHARACTERISTIC FEATURES OF BRITTLE FRACTURE indistinguishable from the fracture features caused by brittle
fracture. This type of failure can occur at either end fitting. If
A. Brittle fracture description crimping deformations are large, the loads at rod fracture
In brittle fracture large cracks are formed inside the GRP could be quite low [47,48]. Most likely some of the reports in
rods and run perpendicular to the long axis of the insulators the past of “brittle fracture failures” at the cold ends must have
(Fig. 1). Due to the complexity of the problem several been related to the rod failure by a combination of
different analyses are required as part of the failure overcrimping and in-service mechanical tensile loads. The
investigation. Brittle fracture can only be determined if the only way to distinguish this type of failure from brittle
following three analyses of composite fracture surfaces are fracture is by performing a detailed microscopic analysis of
performed: macroscopic, microscopic and chemical. These the rod’s fracture surface [47].
fundamental analyses must be performed, all of them, to be
absolutely sure that an insulator has failed by brittle fracture.
1) Macroscopic analysis: It has been shown in [9-16]
that brittle fracture can occur either inside or just outside of
the energized end fitting (see Fig. 1). There is a relationship
between the location of failure and the position of the grading
ring. When grading rings were present, a vast majority of
failures occurred in the rod just above the grading ring
[9,10,12,16,26]. Also the location of failure was related to the
size of the grading rings. For higher line voltages with larger
grading rings the location of failure was found to occur at
larger distances from the end fitting [9,10,14,16]. In the
absence of the grading rings the failure location was found to
occur usually inside the fitting [10,12]. In some sporadic cases
the failure was a combination of transverse cracks in the
composite rods inside and just outside the fittings linked by
long axial splits along the rod [10,12].
A model has been proposed [9,10,14-16,19,25,26], which
explains the location of failure in the insulators, e.g. failure
inside the fitting or above the grading ring. The model is
based on the formation of nitric acid either on the external
surface of the insulators due to water droplet coronas (failure
inside the fitting), or inside the insulators in the presence of
water and partial discharges (PD) (failure above the
hardware). To support the model, electric field calculations
were performed [10,13,19] in order to demonstrate the
possibility of PD that are necessary to create nitric acid in
service inside macroscopic delaminations and other large
cracks that are always associated with the brittle fracture
process. It has been shown in [10,13,19] that if the cracks are
partially filled with water, the field concentration is high
enough to initiate PD inside the transmission composite
insulators near their energized ends, and to produce
nitrogenous species (nitrates).
The fact that in brittle fracture transverse fracture surfaces Fig. 1. Examples of brittle fracture failures of 500kV suspension insulators;
form in the GRP rod and run perpendicular to the rod axis (a) failure inside the fitting and (b) failure above the hardware [14,26].
3> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 3

Fig. 3. Mirror (*), mist (**) and hackle (***) zones on the fracture surface of
E-glass fiber exposed to nitric acid [47].

2) Microscopic analysis: The microscopic observations


of numerous field-failed insulators have shown that the brittle
fracture surfaces in the GRP rods formed either inside or
above the hardware exhibit noticeably different micro-features
[9,10,12,14,15,19]. If failures occurred inside the fittings, the
surfaces were usually heavily contaminated by thick surface
deposits (see for example Fig. 5a). For the failures above the
hardware the amount of surface contamination was found to
be significantly lower. In addition, more resin decomposition
by PD was observed on the composite fracture surfaces above
the hardware than inside the fitting. These differences were
explained in [10-16,19] and the internal and external brittle
failure processes were simulated in small composite GRP
specimens under laboratory conditions [10-12,14,19].
The most characteristic feature of SCC of unidirectional E-
glass/polymer composites and thus brittle fracture of non-
ceramic insulators is the formation of the mirror, mist and
hackle zones on the fracture surfaces of broken fibers. These
three zones can be easily observed for example in Fig. 3 of
this paper. The size of the mirror zones does not depend on
the chemical environment (usual misconception) but is related
to the magnitude of the mechanical stresses [27]. The fiber
fracture under stress corrosion is caused by the ion exchange
mechanism in which hydrogen ions from an acidic solution
replace metal ions (Al, Ca, Fe, Mg) in the fibers. This
weakens the fibers causing their fracture under low tensile
stresses [27].
According to the experimental observation presented in
[49], cracks in brittle materials such as glasses suffer a
dynamic instability, which makes them unable to accelerate up
to high velocities predicted by classical theories of dynamic
fracture. Therefore, the transition from the mirror to hackle
zone is related to a critical crack tip velocity above which the
fracture process is highly unstable [34,49]. Therefore, it can
be shown that the size of the mirror zones depends primarily
Fig. 2. Failure characteristics of overcrimped 115kV insulator; (a) macro-
on the applied stress. The entire fracture process, after the
damage zone, (b) composite fracture surface and (c) typical example of formation of a ”C-type” flaw, is very fast but not directly
fractured fiber from the surface in Fig. 2b [47]. related to an acid attack on glass fibers [27,34]; an acid is only
4> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 4

necessary to initiate a “C-type” flaw. If we consider the fittings can be traced along the insulators, which failed in-
average crack tip velocity across a glass fiber to be about 200 service, by brittle fracture. By tracing the metal ions, the
m/s [49] then the time for fracture of an average fiber with extent of water/acid ingress inside the insulators can be
14µm diameter is approximately 70ns. From the moment of established. Also, the presence of other contaminates such as
the flaw initiation until the final failure of the fiber, the sulfur, chlorine, sodium, potassium, etc., which could
process is dynamic not static (another usual misconception) contribute to the failure process, can be established by
[34]. chemically analyzing the deposits [9,10,12,14,15,19].
The zones shown in Fig. 3 are usually damaged during a 3) Chemical analysis: According to [1] “the definitive
corrosion process. This process is called post failure damage method of detection of brittle fracture consists of an elemental
[10,12,27], occurring after the fracture of individual fibers, analysis of the relative depletion of calcium and aluminum
and is caused by the chemical attack of a corrosive ions in the fiberglass core rod. Scanning electron microscope
environment on the newly formed fiber fracture surface. In (SEM) and energy dispersive X-ray (EDX) analysis have been
some cases, numerous surface cracks are formed in the mirror used in studying the corrosive morphological damage and
zones due to the leaching of aluminum, calcium and other elemental depletion analysis of fiberglass. Using these
metal components out of the glass fibers (see Fig. 4). techniques, the decreases in calcium and aluminum relative to
silicon can be determined. Other methods that are useful in
this detection include Auger spectroscopy to define the
constituents of the glass fiber”.

Fig. 4. Post failure damage on the fracture surface of E-glass fiber exposed to
oxalic acid [10].

The post failure damage is an important phenomenon very


useful in the failure analysis. Its characteristics can be used to
determine the cause of SCC for a composite structure. The
amount of post failure damage is related to the acid type, its
concentration and the time of exposure [10,12, 27,30]. For
example, nitric acid of pH 1 will not develop post failure
damage within a few weeks of exposure whereas other acids
of the same concentration (sulfuric, hydrochloric, oxalic) will
after a few hours or days [10,12,27,30]. In the case of oxalic
acid the post failure damage can be so extensive (Fig. 4) that
the entire fracture surface including the mirror, mist and
hackle zones is essentially shattered.
Another very important issue related to the micro-fracture
features of composite insulators is the nature of the deposits
on their fracture surfaces, in particular, when they were
formed inside the fitting (see Fig. 5) [9,10,12,14,15]. Usually,
the deposits are so thick that the fracture surfaces of individual Fig. 5. Examples of surface deposits on brittle fractures of composite
broken fibers cannot be seen. Critical information about the insulators [10,14,15].
causes of failure by brittle fracture can be gained by Certainly, aluminum and calcium are depleted in the glass
chemically analyzing the deposits [9,10,12,14,15,19]. For fibers on the composite fracture surfaces in the field failed (by
example, metal ions such as iron (Fe) and zinc (Zn) from the brittle fracture) composite insulators [10]. Also, the depletion
5> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 5

of Al and Ca indicates SCC and thus brittle fracture as a cause suspension insulators (Fig. 6) indicating that the failure was
of failure as stated in [1]. However, EDX and SEM could not caused by nitric acid [17,26]. It was also shown in [26] that
be successfully used to determine the depletion in the field the highest concentration of nitric acid existed on the fracture
failed E-glass fibers on the brittle fracture surfaces surface on both sides of a separation caused by water ingress
[9,10,14,15]. Since the chemical composition of the deposit along the GRP rod/housing interface (Fig. 6, sites 2, 3 and 4).
shown in Fig. 5 consisted predominantly of Ca (with traces of This has finally proven that the model proposed in [9-16,19]
Fe, Zn, S, Cl, etc.), the application of the two techniques could regarding the nitric acid formation process inside the
not determine the actual composition of the glass fibers on the insulators is correct.
composite surfaces. This could only be accomplished using The most important information about the causes of brittle
Auger spectroscopy with depth profiling [10]. EDX showed
fracture failures is not in the E-glass fibers but in the surface
the chemistry of the deposit, not of the fibers.
deposits (contamination). Since the surface chemistry is so
important, recommendations must be provided describing the
procedures for handling field-failed insulators. After failure,
fracture surfaces in composite insulators should be
immediately protected from any further contamination, for
example by placing both sides of a field failed insulator in
plastic bags. The composite fracture surfaces should not be
touched with fingers, sprayed with chemicals etc., which is
usually the case.
B. Results and frequency of occurrence
It is most unlikely that all the cases of brittle fracture
failures reported in past were fully analyzed macroscopically,
microscopically and chemically as recommended here and in
[1]. We also suspects that many failures reported in the past
were incorrectly identified as brittle fracture instead of as
overcrimped failures. Also, many brittle fracture failures have
not been reported at all. In our research we have identified
approximately five 115 kV, twenty 345 kV and four 500 kV
insulators [9,10,12,14,19], which failed by brittle fracture.
We must agree that the probability of insulator failure by
brittle fracture substantially increases with an increase in line
voltage [1]. Certainly, the higher field concentration will
create an environment more suitable for the initiation of brittle
fracture [10,13,19]. Therefore, the only model of brittle
fracture, which could explain these observations, is a model,
which takes into consideration not only the presence of
moisture and mechanical stresses but also the effect of the
electric field [25]. However, even if our high voltage brittle
fracture model is accepted, it cannot be used to predict these
failures since all critical electrical, mechanical and chemical
conditions for brittle fracture are unknown at present, and
most likely, will never be known.
Fig. 6. FTIR analysis of failed by brittle fracture suspension composite
insulator; (a) failure morphology with site locations for FTIR and (b) FTIR C. Examples and explanations of brittle fracture failures
spectra for the sites in Fig. 6a. Band at 1386.8cm-1 indicates presence of
The failures described in [1] as cases D and E were quite
nitrates [26].
widely publicized in the insulator community at different
Even if the depletion process in glass fibers in a field-failed IEEE meetings. These two cases, Cases D [9-13,15,19] and E
insulator is determined, this will still not establish the actual [14], were also extensively studied at OGI and DU. It is our
cause of failure. In order to determine the actual composition opinion that the two cases are of particular importance due to
of a chemical environment responsible for brittle fracture a the sheer number of failures in Case D and high voltage in
detailed surface analysis must be performed, not by using Case E.
Auger, SEM and EDX, but for example by employing Fourier 345 kV failures
Transform Infrared (FTIR) spectroscopy which can detect The failures occurred in 1990/1991 on a 345 kV line.
nitrates [17, 26]. Using FTIR, nitrates were detected on the Fourteen brittle fracture failures occurred resulting in the drop
composite surface in the brittle fracture zones of several of the conductor (see Fig. 7) within two years of service. In
6> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 6

addition almost 200 insulators were found to be severely the fittings near the triple point shown in Fig. 8a,b. It was also
damaged. We had full access to all the failed and damaged established that all the units on the 345 kV line were not
units removed from the line [9]. The failure analysis of the protected against moisture ingress into their end fittings. No
damaged units (not the failed ones) was found to be silicone gel was applied at the fitting/rod interface to achieve
particularly useful. The damage process was observed at the moisture seal.
different stages of its development in the damaged insulators.
In order to explain the 345 kV failures numerous
experimental techniques were employed [10,12]. The field-
failed units were examined using optical and scanning
microscopes on the macro and micro-levels [9,10,12,15,19].
Detailed chemical analyses of the fracture surfaces taken from
the failed units were performed. Unique new experimental
techniques were developed to simulate the brittle fracture
process under laboratory conditions with and without high
voltage fields. All of the above proved that the failures were
caused by brittle fracture. However, the exact causes of the
failures were much more difficult to determine. Finally, after
simulating numerically the mechanical performance of the
insulators with the epoxy cone end fittings using advanced
finite element techniques [15,19] (see Fig. 8), it was
established that a manufacturer must have applied excessive
mechanical loads during proof testing [15,19]. The large loads Fig. 7. 345 kV suspension insulator failed in-service by brittle fracture
applied to the insulators during proof testing crushed the [9,10,15,26].
epoxy cones inside the fittings, as stated in [1] and in [15,
19], allowing easy access of water to the GRP rod of the
insulators. In addition, water was also allowed to stay inside

Fig. 8. Finite element model of suspension insulator with the epoxy cone end fittings; (a) end fitting design (schematic) (b) deformed finite element mesh and (c)
7> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

mechanical stress distributions in the cones [15,19].

and between 2.5 and 3 for an epoxy based composite [10,12].


500 kV failures
The failures described as Case E of [1] were also studied in
this laboratory [14]. The failures were caused by spilt epoxy
on the hardware during manufacturing (see Fig. 9) allowing
easy access of water into the end fittings [1,14].

Fig. 10. Damaged rubber housing in 500 kV composite insulator near its hot
end [14].

Fig. 9. Spilt epoxy on the upper surface of the hardware of 500kV suspension
composite insulator [14].

The research performed on the 500kV units [14] did not


provide any significant new information regarding the macro
and micro-features of brittle fracture. It showed however three
very important new things [14]:
1. The amount of water responsible for brittle fracture
must have been extremely small. All units from the
line, including the failed ones, passed the dye
penetration test.
2. Large amounts of water present on the surface of the
GRP rods in the insulators did not lead immediately to
brittle fracture [14]. A couple of insulators from the
line were found with the rubber housing near the hot
end significantly damaged and the GRP rod completely
exposed to moisture, without brittle fracture (Fig. 10).
3. If water and PD are present at the rod/housing interface
above the hardware, cracks can be found in the rubber
housing initiating at the interface and, then, propagating
outwards across the rubber towards the external surface
of a composite insulator (see Fig. 11).
The formation of the cracks in the housing shown in Fig. 11
could explain the origin of the damage to the housing
presented in Fig. 10. It could also suggest that when such
cracks are present, large amounts of water will penetrate the
housing through those cracks, further accelerating the damage
process in the housing. However, if a large quantity of water
is present inside an insulator, nitric acid forming at the rubber
housing interface will not be concentrated high enough. As a
result, the critical chemical conditions for brittle fracture (acid
concentration) will not be satisfied [16]. It has been shown
Fig. 11. Cracks in rubber housing at different stages of their formation; (a)
that the critical pH for SCC in nitric acid for a polyester type partial crack and (b) through crack [14].
composite with E-glass fibers is somewhere between 3 and 3.5
8> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 8

D.Causes of brittle fracture in the GRP rods, can be designed for their very high resistance
Obviously, if one knows the causes of brittle fracture one to SCC in nitric acid [19,21,24,25,36-40], high resistance to
can also provide remedies. Therefore, a few different models moisture absorption [21,42], high resistance to mechanical
of brittle fracture have been proposed over the years. The surface damage [21], high resistance to the development of
most important ones ( Models I-III) are listed below. leakage currents [21,43], and to ozone exposure [21]. The
final recommendations on how to prevent brittle fracture by
Model I. According to [22] “the failure of in-service NCI
the proper chemical composition of the GRP rods were given
in the brittle fracture mode can occur under the influence of
in [25,52]. Below, the most important conclusions are listed
water and mechanical stresses, and the failure is more likely to from our rod design research performed on several GRP rod
happen with water than with acids”. composites from a single supplier based on E-glass and ECR-
Model II. According to [18,23] “the brittle fracture of the glass (with low and high seed counts) fibers embedded in the
FRP [Fiber Reinforced Polymer] rod of composite insulators modified polyester, epoxy and vinyl ester resins.
is associated with an acid, derived from the hardener, lodged Resistance to SCC in Nitric Acid
on the surface of the rod combined with the ingress of water at The resistance to brittle fracture of insulator composites is
the same location and a mechanical tension load applied to the strongly affected by such factors as fiber and resin type,
insulator.” surface fiber exposure, resin fracture toughness, moisture
Model III. Brittle fracture is caused by nitric acid being absorption, and interfacial strength [19,21,24,25,27-41,52].
formed in service due to electrical discharge, ozone and Out of three E-glass/polymer composites (based on modified
moisture [6,8-12,14-17,19-21,24-26]. The acid can either be polyester, epoxy and vinyl ester resins), the resistance to the
formed by water droplet coronas on the insulator surface near initiation of SCC in nitric acid of E-glass/modified polyester
its energized end or by PD inside the insulator, close to its was found to be 10 times lower than for E-glass/epoxy, and
end. approximately 200 times lower than for E-glass/vinyl ester.
The only agreement between Models I - III is that brittle The ECR-glass/polymer composites exhibited vast
fracture is initiated by moisture ingress into a composite improvement over the E-glass/polymer materials. The highest
resistance to the initiation of SCC damage in the ECR glass-
insulator. The issue of the chemical causes of brittle fracture
based composites was found for the low seed ECR-glass
has been recently discussed in [25]. In particular, the
fibers embedded in either epoxy or vinyl ester resins. The
credibility of the models I-III has been evaluated. It was
ECR glass composites were shown to be equally resistant to
shown that the only model, which can explain all the aspects the propagation of SCC even under highly accelerated testing
of brittle fracture is Model III. This model also considers a conditions, which was not the case for the E-glass fiber based
strong effect of the electric field on the brittle fracture process. composites.
E. Manufacturing prevention Resistance to Moisture Absorption
The resistance to moisture absorption of insulator
Design: For the prevention of brittle fracture of non-
composites affects their insulation properties, resistance to
ceramic insulators an excellent moisture seal and the use of
SCC and brittle fracture, and their overall mechanical
the proper sheath thickness are very important [1]. It has properties [19,21,42,43,52]. The E-glass-modified polyester
already been recommended in [15] that a high quality sealant system exhibited by far the worst resistance to moisture
should be used to protect the interface between the fitting and absorption with a very high rate of moisture diffusion and
the rod/housing. It has also been stated in [1] that the grading high maximum moisture absorption in comparison with the
ring is important since it will minimize corona cutting in the other two E-glass based materials. They had very similar rates
rubber housing, which is certainly correct. However, none of of moisture absorption, however the epoxy-based system did
the field failed insulators described as Cases D and E of [1], take more moisture than the vinyl ester-based material. A
and other units from the same lines [9,10,14], showed any slight fiber effect was found to exist on the moisture
evidence of any damage to the rubber housing from the absorption properties of the composites based on the three
outside. Only in two insulators, which did not fail by brittle different resins. In general, composites based on the ECR
fracture for Case E, numerous cracks were found in the (high seed)-glass fibers took smaller amounts of moisture than
housing propagating from inside out (Fig. 11) [14]. Therefore, their E-glass and ECR-glass (low seed)-glass fiber-based
it is doubtful that grading rings will prevent brittle fracture if composites.
water ingress is allowed through the fitting. Since the failure Resistance to Interfacial Failure
above the hardware, which is the most common type, is Interfacial splitting along the glass/fiber interfaces
associated with the movement of moisture along the rod near accelerates brittle fracture and therefore should be minimized
[19,52]. The resistance of the interfaces to failure caused by
the rod/housing interface, the grading will only postpone the
shear in dry E-glass/polymer composites was found to be the
process, but do not prevent it.
lowest in E-glass/modified polyester followed by E-glass
1) Materials: If the GRP rods can fail by SCC when
epoxy (approximately 50% higher) and E-glass/vinyl ester
exposed to nitric acid and mechanical tensile loads, the easiest
(approximately 4% higher) than for the epoxy based
way to prevent this process from occurring in-service is by the
composite). Immediately after full saturation with distilled
chemical optimization of the rods making them immune to
water at 50ºC the resistance to interfacial splitting was only
brittle fracture. Unidirectional glass/polymer composites, used
slightly reduced. The quality of the interfaces of the ECR-
9> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 9

glass fiber composites has not been investigated. It can be to brittle fracture with the ECR (low seed)- glass/vinyl ester
assumed however that their resistance to interfacial splitting being the best. The ECR (low seed)/epoxy and E-glass/vinyl
would be similar to the E-glass/polymer systems. ester must also be ranked very high [52].
Resistance to Overcrimping Damage 2) Process Controls: The GRP rods should be free from
If overcrimping can cause brittle fracture [1], it should be large voids, cracks or any other type of macroscopic damage,
minimized. It has been shown in our work that the resistance as stated in [1] to minimize the probability of brittle fracture.
to surface compression caused by crimping of the E-glass and Certainly, such material imperfections could create problems
ECR-glass composites is scale dependent [21]. On a micro- caused by the development of PD especially in the presence of
scale the vinyl ester based composites (with E-glass and ECR- moisture. However, such imperfections can easily develop in-
glass fiber) exhibited the highest resistance to compressive service if water is allowed into the insulators. [22]. Therefore,
surface damage, which was attributed to a high fracture the fact that voids and cracks are not present in the rods
toughness of their resin. The resistance of the vinyl ester during manufacturing does not mean that these imperfections
based systems was followed by the epoxy and modified will not develop in service causing problems.
polyester based composites. On a meso-scale, the resistance to Some insulator manufacturers occasionally performed
surface compression strictly depends on the amount of glass sandblasting of their composite rods to increase adhesion
fibers left exposed on the composite surfaces during between the rods and the rubber housing [21,38,39,52].
pultrusion. The macro-resistance, on the other hand, depends During sandblasting the surface of the rods can be severely
on the type of fibers, polymers, and the strength of their damaged. This damage is usually created in the glass fibers
interfaces. left on the composite surface. The initial assumption was that
Resistance to Leakage Currents sandblasting could have an extremely negative effect on the
The resistance to leakage currents of insulator composites resistance of the GRP rods to brittle fracture; therefore, this
[21,43] could be more important for their insulation properties effect was investigated in [21,38,39,52]. It was found however
than for their resistance to brittle fracture. The fact is however that low/medium sandblasting could actually improve slightly
that the electric field plays a critical role in brittle fracture the resistance of the composites to SCC in nitric acid. This
[25]. Therefore, high leakage currents could accelerate the was attributed to the relaxation of mechanical residual stresses
nitric acid formation process inside the insulators. After on the surface of the composites due to sandblasting. This
boiling in deioniozed water with 0.1% (by weight) of NaCl for clearly indicates that not all types of damage to the composites
four days, E-glass/modified polyester developed significantly can negatively affect their resistance to brittle fracture.
higher leakage current that the E-glass/epoxy and E- It has been suggested in [1] that over-stressing the GRP rod
glass/vinyl ester systems, caused by its high water absorption. due to crimping can lead to brittle fracture. This is certainly an
For the same amounts of absorbed moisture by their resins, the important issue related to the insulator design and its
ECR (high seed)-glass fiber composites developed several manufacturing. The fact is that extensive damage to the GRP
hundred times higher leakage currents than their E-glass rods can be created by insulator manufacturers by applying
counterparts. The leakage currents in the composites with excessive crimping [44,47]. Therefore, this problem has been
recently developed ECR (low seed)-glass fibers were similar investigated in great details by us in [19,21,44-48]. It has been
to the E-glass-based systems. The high leakage current for shown that overcrimping will certainly lead to the type of
high seed ECR-glass fiber composites subjected to moisture is fracture, which was described in Section II.B.1 of this work.
one of the major reasons why most manufacturers have not
At the same time, not a single insulator failure by brittle
used these materials until now.
fracture initiated by overcrimping has been documented [1].
Resistance to Ozone Damage
However, excessive and badly distributed crimping
The three E-glass/polymer composites were exposed to
ozone for a period of nine days [21]. It was found that the deformations on the rod surface inside the fittings in
bonding environment in the modified polyester based combination with high mechanical loads will significantly
composite was severely compromised by ozone. The epoxy increase the tensile axial stresses in the rods just outside the
based system showed similar evidence of damage, but to a fittings [44-48]. Then, if nitric acid is present on the surface of
much smaller extent. The E-glass/vinyl ester material was the rod in this location, the probability of failure by brittle
found to be the most resistant to ozone. fracture could increase due to the higher mechanical stresses.
Ranking of Composites for Resistance to Brittle Fracture The cracks caused by overcrimping [44,47], depending on
We have not yet identified all critical material related their location, size and orientation could reduce the resistance
factors affecting brittle fracture. In addition, such known of the insulators to brittle fracture. However, the critical
factors as the resistance of insulator polymers and their mechanical flaw size and type, and the loading conditions for
interfaces with glass fibers to the decomposition by discharge, the initiation of SCC are unknown at present. Therefore, we
acids, electric wind, etc., has not been systematically can also speculate which composite system (see the Materials
investigated. Also, no systematic studies of the resistance to part of this section) will have the highest resistance to brittle
moisture movement along E-glass and ECR-glass composites fracture if initiated by overcrimping. It can be safely
have been performed. Considering however the thoroughly assumed, however, that the vinyl ester based composites will
examined material properties of several different composite have the highest resistance to brittle fracture due to its high
systems for high voltage insulation applications, E-
fracture toughness. Certainly more research is required to
glass/modified polyester exhibits by far the lowest resistance
10> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 10

understand the effect of overcrimping on the initiation of financial support the insulator research at OGI and DU would
brittle fracture have not been possible. I am also very grateful to my
colleagues, Dr. P. K. Predecki, Dr. D. Armentrout and, in
F. User Prevention
particular, my son L. Kumosa for their help in the preparation
Composite insulators should not be mechanically damaged of this manuscript.
at any stage; manufacturing, transportation, installation.
Certainly, damage caused by mishandling “may result in a REFERENCES
brittle fracture or some other type of failure” as stated in [1].
[1] J. T. Burnham, T. Baker, A. Bernstorf, C. de Tourreil, J.-M. George, R.
However, the evaluation of the effect of mechanical damage Gorur, R. Hartings, B. Hill, A. Jagtiani, T. S. McQuarrie, D. Mitchell, D.
(of various types) on the electrical and mechanical in-service Ruff, H. Schneider, D. Shaffner, J. Yu and J. Varner, “IEEE Task Force
performance of composite insulators is not a straightforward Report: Brittle Fracture in Non-Ceramic Insulators,” IEEE Transactions
on Power Delivery, Vol. 17, No. 3, pp. 848-856, July 2002.
problem. Numerous different types of failure caused by [2] H. D. Chandler, R. L. Jones and J. P. Reynders, “Stress Corrosion of
mishandling can exist in non-ceramic insulators. At the same Composite Long Rod Insulators,” in Proceedings of the 4th
time, we still do not know what is acceptable and what is not International Symposium on High Voltage Engineering, 1983, paper
23.09, Athens, Greece, 5-9 September, 1983.
regarding the mechanical damage and its effect on the short- [3] B. Noble, S. J. Harris and M. J. Owen, “Stress Corrosion Cracking of
and long- term properties of the insulators and, in particular, GRP Pultruded Rods in Acid Environments,” J. Material Science, Vol.
on brittle fracture. 18, pp. 1244-1254, 1983.
[4] H. Chandler and J. Reynders, “Electro-Chemical Damage to Composite
The critical type of damage for the initiation of brittle Insulator,” CIGRE 1984 Session, Paper 33-08, 1984.
fracture in composite insulators subjected to the combined [5] S. J. Harris, B. Noble and M. J. Owen, “Metallographic Investigation of
effect of mechanical, electrical, and environmental stresses is the Damage Caused to GRP by the Combined Action of Electrical,
Mechanical, and Chemical Environments,” J. Material Science, Vol. 19,
an issue which most likely is never going to be correctly
pp. 1596-1604, 1984.
addressed. The best example of the complexity of this [6] A. Akhtar, J. S. Nadeau, J. Y. Wang, D. P. Romily and C. Taggart,
problem is gun shot damage. In certain areas, the insulators “Brittle Fracture of Non-Ceramic Insulators,” Report for the Canadian
can be severely damaged by gunshots [51]. When damaged by Electrical Association (186 T 350), Prepared by the British Columbia
Hydro and Power Authority, September 1986.
gunshots the insulators are entirely open to moisture ingress. [7] M. J. Owen, S. J. Harris and B. Noble, “Failure of High Voltage
In addition, huge amounts of mechanical damage are created Electrical Insulators with Pultruded Glass Fiber-Reinforced Plastic
to the GRP rods. It would be difficult to imagine any worse Cores,” Composites, Vol. 17, pp. 217-226, 1986.
[8] A. Akhtar and J. Y. Wong, “Failure Analysis of Brittle Fracture in
type of mishandling of the insulators in-service than by Nonceramic Insulators,” J. Composite Technology and Research, Vol. 9,
gunshots. Yet, not a single composite insulator has been pp. 95 –100, 1987.
reported to fail by brittle fracture initiated by gunshots. [9] M. Kumosa, “Failure Analysis of Composite Insulators from the Craig
Bonanza Line,” Final Report to the Western Area Power Administration,
Therefore, the only credible recommendation, which can be Oregon Graduate Institute, Portland Oregon, 1994.
provided on how to avoid brittle fracture due to mishandling [10] M. Kumosa, Q. Qiu, M. Ziomek-Moroz, A. Moroz and J. M. Braun,
is related to the moisture sealing conditions, which should “Micro-Fracture Mechanisms in Glass/Polymer Insulator Materials
never be compromised during manufacturing, transportation under Combined Effects of Electrical, Mechanical and Environmental
Stresses”, Final Report to the Bonneville Power Administration, Electric
and installation. Power Research Institute and the Western Area Power Administration,
Oregon Graduate Institute, Portland, Oregon, July 1994 (under contract
III. CONCLUSIONS DE-AC79-92BP61873).
[11] M. Kumosa, Q. Qiu, E. Bennett, C. Ek, T. S. McQuarrie and J. M.
The brittle fracture process in composite non-ceramic high Braun, “Brittle Fracture of Non-Ceramic Insulators,” in the Proceed.
voltage insulators has been discussed in this study. This was Fracture Mechanics for Hydroelectric Power Systems Symposium'94,
Canadian Committee for Research on the Strength and Fracture of
done partially to supplement the information recently provided Materials (CSFM), BC Hydro, Sept. 1, pp. 235-254, 1994.
in [1]. The most important failure characteristics, examples [12] Q. Qiu, “Brittle Fracture Mechanisms of Glass Fiber Reinforced
and causes of failure, and possible prevention methods have Polymer Insulators,” Ph.D. Thesis, Oregon Graduate Institute of Science
& Technology, Portland, Oregon, October 1995 (with Dr. M. Kumosa as
been presented in this work in an attempt to improve the the thesis advisor)
current state of knowledge on brittle fracture of the insulators. [13] N. Fujimoto, J. M. Braun, M. Kumosa and C. Ek, “Critical Fields in
It has been shown that despite the recent major advances in Composite Insulators: Effects of Voids and Contaminations,” Ninth
the failure analysis of the insulators there are still several International Symposium on High Voltage Engineering, Graz, Austria,
August 28-September 1, 1995.
critical issues which need to be correctly addressed in the [14] M. Kumosa and Q. Qiu, “Failure Investigation of 500 kV Non-ceramic
future in order to fully understand this failure process and, in Insulators” Final Report to the Pacific Gas and Electric Company,
particular, to be able to prevent it from occurring in-service. Department of Engineering, University of Denver, May 1996.
[15] M. Kumosa, H. Shankara Narayan, Q. Qiu and A. Bansal, Brittle
Fracture of Non-Ceramic Suspension Insulators with Epoxy Cone End-
ACKNOWLEDGMENTS Fittings, Composites Science and Technology, Vol. 57, pp. 739-751,
1997.
The author is especially grateful to Dr. J. Stringer and Dr. [16] Interview with Maciej Kumosa, Research of Brittle Fractures in
A. Phillips of EPRI, Mr. C. Ek, D. Nicols and Mr. R. Stearns Composite Insulators, Insulator News & Market Report, pp. 46-51,
of BPA, Mr. O. Perkins and Mr. F. Cook of WAPA, Mr. D. July/August, 1997.
Mitchell of APC, Mr. D. Shaffner of PG&E and Dr. T.S.
McQuarrie of Glasform, Inc. without whose technical and
11> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 11

[17] A. R. Chughtai, D. M. Smith and M. Kumosa, “Chemical Analysis of a [39] L. Kumosa, M. Kumosa and D. Armentrout, “Resistance to Stress
Field-Failed Composite Suspension Insulator,” Composite Science and Corrosion Cracking of Unidirectional Glass/Polymer Composites Based
Technology, Vol. 58, pp. 1641-1647, 1998. on Low and High Seed ECR-glass Fibers for High Voltage Composite
[18] C. de Tourreil, L. Pargamin, G. Thevenet and S. Prat, “Brittle Fractures Insulator Applications,” Composites Part A, Vol. 34, No. 1, pp. 1-15,
of Composite Insulators: Why and How they Occur”, Power 2003.
Engineering Society Summer meeting 2000, Vol. 4, pp. 2569-2574, [40] D. Armentrout, M. Kumosa and T. S. McQuarrie, “Boron Free Fibers for
2000. Prevention of Acid Induced Brittle Fracture of Composite Insulator GRP
[19] M. Kumosa, “Fracture Analysis of Composite Insulators,” EPRI, Palo Rods,” IEEE Transactions on Power Delivery, Vol. 18, No. 3, pp. 684-
Alto, CA: 2001. 1006293. 693, 2003.
[20] M. Kuhl, “FRP Rods for Brittle Fracture Resistant Composite [41] D. Armentrout, M. Gentz, L. Kumosa, B. Benedikt and M. Kumosa,
Insulators”, IEEE Transactions on Dielectrics and Electrical Insulation, “Stress Corrosion Cracking in a Unidirectional E-glass/Polyester
Vol. 8, No. 2, pp. 182-190, 2001. Composite Subjected to Static and Cyclic Loading Conditions,”
[21] M. Kumosa, “Failure Analysis of Composite High Voltage Insulators”, Composites Technology & Research, Vol. 25, No. 4, pp. 202-218, 2003.
EPRI, Palo Alto, CA: 2002.1007464. [42] L. Kumosa, B. Benedikt, D. Armentrout and M. Kumosa, “Moistures
[22] J. Montesinos, R. S. Gorur, B. Mobasher and D. Kingsbury, Mechanism Absorption Properties of Unidirectional Glass/Polymer Composites
of Brittle Fracture in Non-Ceramic Insulators, IEEE Transactions on Used in Non-Ceramic Insulators”, Composites Part A, Vol. 35, pp.
Dielectrics And Electrical Insulation, Vol. 9, No. 2, April 2002. 1049-1063, 2004.
[23] F. Schmuck and C. de Tourreil, “Brittle Fractures of Composite [43] D. Armentrout, M. Kumosa and L. Kumosa, “Water Diffusion into and
Insulators. An Investigation of their Occurrence and Failure Mechanisms Electrical Testing of Composite Insulator GRP Rods”, IEEE
and Risk Assessment”, for CIGRE WG 22-03 Transactions on Dielectrics and Electrical Insulation, in press, Vol. 11,
(http://www.corocam.com/papers.htm). No. 3, pp. 506-522, 2004.
[24] M. Kumosa, L. Kumosa and D. Armentrout, “Can Water Cause Brittle [44] M. Kumosa, “Analytical and Experimental Studies of Substation NCIs,”
Fracture Failures of Composite Non-Ceramic Insulators in the Absence Final Report to the Bonneville Power Administration, Oregon Graduate
of Electric Fields?,” IEEE Transactions on Dielectrics and Electrical Institute of Science & Technology, Portland, Oregon, 1994.
Insulation, Vol.. 11, No. 3, pp. 523-533, 2004. [45] A. Bansal, A. Schubert, M. V. Balakrishnan and M. Kumosa, “Finite
[25] M. Kumosa, L. Kumosa and D. Armentrout, “Causes and Potential Element Analysis of Substation Composite Insulators,” Composites
Remedies of Brittle Fracture Failures of Composite (Non-Ceramic) Science and Technology, Vol. 55, pp. 375-389, 1995.
Insulators,” IEEE Transactions on Dielectrics and Electrical Insulation, [46] A. Bansal and M. Kumosa, “Mechanical Evaluation of Axially Loaded
Vol. 11, No. 6, pp.1037-1048. Composite Insulators with Crimped End-Fittings,” Journal of Composite
[26] A. R. Chughtai, D. M. Smith, L. S. Kumosa and M. Kumosa, FTIR Materials, Vol. 31, No 20, pp. 2074-2104, 1997.
Analysis of Non-Ceramic Composite Insulators, IEEE Transactions on [47] M. Kumosa, Y. Han and L. Kumosa, “Analyses of Composite Insulators
Dielectrics and Electrical Insulation, Vol. 11, No. 4, pp. 585-596, 2004 with Crimped End-Fittings: Part I – Non-linear Finite Element
[27] D. Hull, M. Kumosa, J. N. Price, “Stress Corrosion of Aligned Glass Computations,” Composites Science and Technology, Vol. 62, pp. 1191-
Fiber-Polyester Composite Material,” Materials Science and 1207, 2002.
Technology, Vol. 1, pp. 177 –182, 1985. [48] M. Kumosa, D. Armentrout, L. Kumosa, Y. Han and S.H. Carpenter,
[28] M. Kumosa, D. Hull, J. N. Price, Acoustic Emission from Stress “Analyses of Composite Insulators with Crimped End-Fittings: Part II –
Corrosion Cracks in Aligned GRP, J. of Materials Science, Vol. 22, pp. Suitable Crimping Conditions,” Composites Science and Technology,
331-336, 1987. Vol. 62, pp. 1209-1221, 2002.
[29] M. Kumosa, “Acoustic Emission Monitoring of Stress Corrosion Cracks [49] M. Marder and J. Fineberg, “How Things Break,” Physics Today,
in Aligned GRP,” J. Phys. D: Appl. Phys. Vol. 20, pp. 69-74, 1987. September 1996, pp. 24-29.
[30] Q. Qiu and M. Kumosa, “Corrosion of E-Glass Fibers in Acidic [50] T. S. McQuarrie, “Improved Dielectric & Brittle Fracture Resistant Core
Environments,” Composites Science and Technology, Vol. 57, pp. 497- Rods for Non-Ceramic Insulators,” Insulator 2000 World Congress on
507, 1997. Insulator Technologies for the Year 2000 & Beyond, Barcelona, Spain
[31] D. Armentrout, T. Ely, S. Carpenter and M. Kumosa, “An Investigation November 14-17, 1999.
of the Brittle Fracture in Composite Materials used for High Voltage [51] J. T. Burnham and R. J. Waidelich, “Gunshot Damage to Ceramic and
Insulators,” J. Acoustic Emission, Vol. 16, No. 1-4, pp. S10-S18, 1998. Nonceramic Insulators,” IEEE Transactions on Power Delivery, Vol. 12,
[32] M. Kumosa et al., “Micro-Fracture Mechanisms in Glass/Polymer No. 4, pp. 1651-1656, 1997.
Insulator Materials under the Combined Effect of Mechanical, Electrical [52] L. Kumosa, M. Kumosa and D. Armentrout, “Resistance to Brittle
and Environmental Stresses,” Final report to BPA, APA, PG&E, WAPA Fracture of Glass Reinforced Polymer Composites Used in Composite
and NRECA, University of Denver, Denver, Colorado, December 1998. (Non-Ceramic) Insulators”, IEEE Transactions on Power Delivery, in
[33] S. H. Carpenter and M. Kumosa, “An Investigation of Brittle Fracture of press, 2005.
Composite Insulator Rods in an Acidic Environment with Static or Maciej S. Kumosa received his Master’s and
Cyclic Loading,” Journal of Materials Science, Vol. 35, Issue 17, pp. Ph.D. degrees in Applied Mechanics and
4465-4476, 2000. Materials Science in 1978 and 1982 from the
[34] T. Ely and M. Kumosa, “The Stress Corrosion Experiments on an E- Technical University of Wroclaw in Poland.
glass/Epoxy Unidirectional Composite,” J. Composite Materials, Vol. He is currently a Professor of Mechanical
34, pp. 841-878, 2000. Engineering and the Director of the Center for
[35] T. Ely, D. Armentrout and M. Kumosa, “Evaluation of Stress Corrosion Advanced Materials and Structures at the
Properties of Pultruded Glass Fiber/Polymer Composite Materials,” J. University of Denver. In the past, Dr. Kumosa
Composite Materials, 35, pp. 751-773, 2001. worked six years at the University of
[36] M. Megel, L. Kumosa, T. Ely, D. Armentrout and M. Kumosa, Cambridge in England. Between 1990 and
“Initiation of Stress Corrosion Cracking in Unidirectional Glass/Polymer 1997 he was an Associate Professor of
Composite Materials,” Composites Science and Technology, Vol. 61, pp. Materials Science and Electrical Engineering
231-246, 2001. at the Oregon Graduate Institute in Portland,
[37] L. Kumosa, D. Armentrout, M. Kumosa, “An Evaluation of the Critical Oregon. Dr. Kumosa’s research interests include the experimental and
Conditions for the Initiation of Stress Corrosion Cracking in numerical fracture analyses of advanced composite systems for electrical and
Unidirectional E-glass/Polymer Composites,” Composites Science and aerospace applications. He has performed research for a variety of funding
Technology, 61, pp. 615-623, 2001. agencies in the US including the National Science Foundation, Air Force
[38] L. Kumosa, D. Armentrout and M. Kumosa, “The Effect of Sandblasting Office of Scientific Research, NASA Glenn, Electric Power Research Institute
on the Initiation of Stress Corrosion Cracking in Unidirectional E- and a consortium of US electric utilities and insulator manufacturers
glass/Polymer Composites Used in High Voltage Composite (Non- consisting of the Bonneville Power Administration, Western Area Power
Ceramic) Insulators,” Composites Science and Technology, Vol. 62, No. Administration, Alabama Power Company, Pacific Gas and Electric Company
15, 2002, pp. 1999-2015, 2002. and National Rural Electric Cooperative Association, Glasforms, Inc. and
12> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 12

NGK-Locke. Dr. Kumosa has published over 85 publications in numerous


composites, materials science, applied physics, applied mechanics, general
science and IEEE international journals, and 45 publications in conference
proceedings. Dr. Kumosa is on the Editorial Board of the Journal of
Composites Science and Technology (http://myprofile.cos.com/mkumosa).