THE INITIAL PROMISE OF COMPOSITE INSULATORS WAS INTOXICATING: a product that was lightweight, easy to install, resistant to gunshots and less likely to track in corrosive environments. And we have found that composite insulators for the most part do in fact live up to that promise. But, our industry was caught off-guard when the first cases of brittle fracture were reported. Utilities found that under certain conditions, a transverse crack could propagate across the internal fiberglass rod and bring power lines down. This article provides an insight into the process of insulator degradation and outlines steps utilities can take to address this uncommon yet critical problem.

INVESTING IN RESEARCH

Several critical issues regarding the mechanical and electrical integrity of composite — also called nonceramic and polymer — insulators have been investigated in our laboratory since 1993, initially at the Oregon Graduate Institute (Portland, Oregon, U.S.) and then, since 1996, at the University of Denver (Denver, Colorado, U.S.). Our insulator research was jointly supported between 1993 and 2004 by the Electric Power Research Institute (EPRI; Palo Alto, California, U.S.), Bonneville Power Administration (Portland), Western Area Power Administration (Lakewood, Colorado), Pacific Gas and Electric (San Francisco, California), Alabama Power Co. (Birmingham, Alabama, U.S.), and the National Rural Electric Cooperative Association (Arlington, Virginia, U.S.). Glasforms Inc. (San Jose, California) and NGK-Locke (Baltimore, Maryland, U.S.) also supported some of the work. Over the years, we have identified various mechanical and electrical failure modes of the insulators, and have provided recommendations on how to avoid some of these failures in service.

BRITTLE-FRACTURE INVESTIGATION

One of the mechanical failure modes of the insulators is a failure process called brittle fracture, which is caused by stress-corrosion cracking (SCC) of glass-reinforced polymer (GRP) rods. This particular failure mechanism has generated significant angst among electric utilities because the phenomenon has not been well understood, nor is it easy to predict. Typically, a brittle fracture is only found when an insulator fails and drops the line. Brittle-fracture failures, though not frequent, have caused line outages at various utilities over the years. Therefore, we undertook major research efforts to better understand the causes of brittle fracture and to prevent these failures in service by the proper design of GRP composite rods used as load-carrying components of the insulators.

Various groups of researchers around the world have made several attempts to understand the process and to provide potential means of avoiding it in service. In our insulator research, particular attention has been given to understanding the brittle-fracture process, simulate the process under in-service conditions and recommend feasible remedies. The brittle-fracture research in our laboratory was initiated as a result of 14 insulator failures experienced on a 345-kV line in 1990/1991 by a major U.S. utility. It was further expanded in 1996 after another large U.S. utility on the West Coast experienced insulator failures on a 500-kV line in 1995.

In brittle fracture, large cracks are formed inside the GRP rods and run perpendicularly to the long axis of the insulators (Figs. 1 and 2). The failure can occur deep inside the fitting or just above the hardware. We also noticed that in the presence of grading rings, the failure above the hardware always occurred above the rings. Due to the complexity of the problem, several different analyses are required as part of the failure investigation. Brittle fracture can only be determined if the following three analyses of composite-fracture surfaces are performed: macroscopic, microscopic and chemical. All these fundamental analyses must be performed if we are to be absolutely sure that an insulator has failed because of brittle fracture.

Our research efforts regarding brittle fracture were predominantly experimental in nature and involved comprehensive stress-corrosion testing of various insulator composites based on E-glass and ECR (boron free) glass fibers embedded in polyester, epoxy and vinyl-ester resins with and without fillers. An example of a stress-corrosion-fracture surface developed under laboratory conditions in a GRP rod is shown in Fig. 3. We investigated two different types of ECR fibers with low- and high-seed counts. The seeds are gaseous inclusions (voids) inside the fibers left from the glass fiber manufacturing process. We tested the composites under different conditions in different acids, including nitric acid, and different types of water. We also conducted several high-voltage stress-corrosion-fracture experiments involving corona discharges.

We have also developed a comprehensive model of brittle fracture that can explain the different types of brittle-fracture failures of in-service composite insulators (Fig. 4). In addition, we have also performed both mechanical and electric field numerical simulations to explain several brittle-fracture failures and understand the effect of the electric field on the acid-formation process in service.

Our research indicates that brittle fracture is predominantly caused by the formation of nitric acid either on the external surface of the insulators or inside the insulators above their fittings if either the acid or water is allowed to ingress into end-fittings (Fig. 4). Laboratory evidence supported these findings, as traces of nitric acid were found on the brittle-fracture surfaces of several composite insulators We have always maintained that the failure process is complex and caused by three different stresses: mechanical, electrical and environmental. During the failure process, these three stresses are never constant and continuously change as the process progresses. According to our brittle-fracture model, the electric field is absolutely critical to generate nitric acid strong enough to cause brittle fracture.

As shown in our study, the resistance to brittle fracture of composite insulators is strongly affected by such factors as fiber and resin type, surface-fiber exposure, resin-fracture toughness, moisture absorption and interfacial strength. Out of three E-glass/polymer composites investigated in our research and supplied by a single supplier, the resistance to the initiation of SCC in nitric acid of E-glass/modified polyester was found to be 10 times lower than for E-glass/epoxy and approximately 200 times lower than for E-glass/vinyl ester. The ECR-glass/polymer rods exhibited vast improvement over the E-glass/polymer rods. The highest resistance to the initiation of SCC damage in the ECR glass-based composites was found for the low-seed ECR-glass fibers embedded in either epoxy or vinyl-ester resins. The ECR glass composites were shown to be equally resistant to the propagation of SCC even under highly accelerated testing conditions, which was not the case for the E-glass fiber-based composites.

However, we strongly emphasize that just because flat fracture surfaces form in a GRP rod and run perpendicular to the rod axis, Figs. 1 and 2 should not be immediately used as definite evidence of brittle fracture. If crimped composite insulators (Fig. 5) are incorrectly designed and manufactured — excessive crimping deformations, incorrect design of the metal end-fittings, excessive in-service mechanical loads — failure of the rod can also occur very close to the end-fitting, and the macroscopic fracture surface will be flat and almost perpendicular to the long axis of the rod (Fig. 6). In this case, the macro-fracture features will be almost indistinguishable from the fracture features caused by brittle fracture. More about the effect of crimping on the insulator strength can be found in the many technical papers published on insulator-failure mechanisms.

MOISTURE ABSORPTION AND LEAKAGE CURRENT STUDIES

Since brittle fracture is always caused by water/acid ingress into the insulators, our laboratory initiated major efforts to understand the effect of different fibers, resins and manufacturing conditions on the moisture-absorption properties of the GRP composites. We ranked different insulator composites for their resistance to water absorption. However, in order to do so, we had to develop new procedures that could be used to evaluate water absorption and water movement in the insulator composites. Finally, since water also has a detrimental effect on the development of leakage currents, we developed new experimental and numerical techniques for the determination of moisture absorption versus leakage current relationships for different composite systems.

Through our research, we have shown that, on one hand, modified polyester-based composites absorb moisture much faster than other composites based on epoxy and vinyl-ester resins. In addition, this type of composite exhibited single-phase Fickian diffusion (Fig. 7). On the other hand, epoxy-based composites absorb water at a much slower rate (Fig. 7). However, these materials have a tendency not to equilibrate (non-Fickian, multiphase diffusion, might take many years to reach full saturation). We also showed that vinyl-ester-based composites took on the least amount of moisture more slowly and their diffusion was single phase.

In our laboratory, we performed water-diffusion electrical testing on GRP-composite rods following the ANSI standard C29.11 Section 7.4.2 that can be used to evaluate electrical properties of insulator composites. We investigated the rod materials based on either E-glass or ECR-glass fibers with modified polyester, epoxy and vinyl-ester resins, and we studied the effects of composite-surface sandblasting, mechanical preloading and nitric-acid exposure on the electrical properties of the composites. In the end, we found that the amount of absorbed moisture did not correlate with the leakage current change (Fig. 8). Furthermore, we were able to show that the composites with high-seed ECR glass fibers had the highest leakage current.

Subsequently, we have redesigned the ANSI test by replacing the composite-rod specimens with composite hollow core cylinders. Using the new test, two very different composites — ECR (low-seed) glass/epoxy and ECR (high-seed) glass/modified polyester — were tested for their leakage currents under controlled-moisture diffusion conditions at 50°C (122°F) and 80% RH. The data shown in Fig. 9 clearly demonstrate an enormous effect of seeds in the fibers on leakage currents. For the same amounts of absorbed moisture, the modified polyester composites with ECR high-seed glass fibers exhibited more than 150 times higher leakage currents than the composite with the ECR low-seed glass fibers.

BRITTLE FRACTURES ONLY ONE OF POSSIBLE FAILURE MECHANISMS

The fact is that the quality of the design and manufacturing of composite insulators have dramatically improved over the years. The insulators, which failed by brittle fracture in the 1990s, were susceptible to moisture ingress into energized end-fittings. Those insulators were also based on the composites, which, as we have shown in our research, had very low resistance to brittle fracture. We strongly suspect that such insulators are most likely no longer in service. The manufacturers have made major efforts to protect the insulators against moisture ingress into their fittings by applying new designs. In addition, as our research has shown, there are composite systems with almost-perfect resistance to brittle fracture. Therefore, considering the recent developments, we can safely state that brittle fracture can be fully prevented in the future based on knowledge now available on the causes of brittle fractures, as well as methods available to mitigate this phenomenon.

We've learned a great deal about the failure mechanisms and operational characteristics of composite suspension insulators in the more than 30 years since they were first introduced. As with any new product, we went through a learning curve. Today, as vendors work closely with research institutions and their utility customers to apply the knowledge gained to the next generation of insulators, we should be able to reap the promise of the insulators that were put forth decades ago.

---

Maciej S. Kumosa (http://myprofile.cos.com/mkumosa) received his master's and Ph.D. degrees in applied mechanics and materials science in 1978 and 1982, respectively, from the Technical University of Wroclaw in Poland. He is currently a professor of mechanical engineering and the director of the Center for Advanced Materials and Structures at the University of Denver. In the past, Kumosa worked for six years at the University of Cambridge in England. His research interests include the experimental and numerical fracture analyses of advanced-composite systems for electrical and aerospace applications. A comprehensive series of Kumosa's papers on this topic are available at www.engr.du.edu/CFAMS/CFAMS_Downloads.htm. Kumosa is on the editorial board of the Journal of Composites Science and Technology. mkumosa@du.edu