Increasing demand for overhead transmission line components such as insulators has brought into the market new, previously often unknown suppliers. This has created a large spectrum of performance and quality that are not necessarily easy to identify relying only on existing standards or specifications. This is true for all insulator technologies. End user procurement offices are not yet prepared for a differentiated approach based on technical merits and, in most cases, still rely on a relatively simple technical description in line with some national standard. Today, however, more and more standards engineers are asking for more stringent technical criteria to be introduced into their specifications in order to reinforce their own selection tools and avoid them from inadvertently qualifying low quality suppliers. It might, therefore, be time to review the standards and redefine quality and performance. End users themselves bear some responsibility in the quality of the products they purchase since not all take the time to carry out detailed audits of potential suppliers nor do all send qualified representatives to factories to randomly select samples for the acceptance tests. This edited contribution to INMR by J.M. George, E. Brocard, S. Prat, F. Virlogeux, D. Lepley of the Sediver Research Center in France reviewed key points to take into account, most covering glass and porcelain and some more specific to glass. It also contributes to ongoing work intended to revise IEC 60383, which is the result of concern based on research published in the recent past.
Status of Standards
The most commonly used standards for overhead line insulators are ANSI, IEC and CSA. While many insulators are being installed in China, this remains a separate market environment with its own standards, although Chinese GB standards are often based on IEC. These standards describe test protocols but little if any material description. Problems with poor quality of materials can sometimes show up after only several years of service and should therefore ideally also be part of the review. Similarly, experience has been gained over the years from weaknesses of some designs not yet integrated in the battery of tests that should help weed out ‘low performers’. The following Tables compare IEC 60383, ANSI C29-2B and CSA 411-1-16. It appears very rapidly that ANSI and IEC are less demanding than CSA, that in many respects goes a few steps further by introducing new tests. The separation between type tests and sample tests, which differs between standards, (e.g. quantities as well as test itself) should also be noted. A key question is whether or not all manufacturers are capable to demonstrate consistency and, as a result, if some tests should be repeated at the level of acceptance tests.
Metal End Fittings
Description of materials to be used for components remains relatively vague in ANSI, calling only for commercially available, malleable ductile iron or steel. IEC 60120 does not give any detail either. This is in contrast to CSA that describes in detail the type of casting and steel (e.g. minimum elongation of 10% for malleable and 12% for ductile iron, with a requirement of cold impact strength as per ASTM A370). Also, Cotter keys are described only in CSA with clear reference to type of stainless steel (i.e. ASTM A580 or 580M S30400, S32100 type 304, 314, 321 or EN 10088-1 type 1.4301 with Wicker hardness above HV150). IEC 60372 does not specify any material reference.
Cement is another major component of an insulator. All the standards today ask for the expansion test (per ASTM C151/151M) when the insulator is made using Portland cement. For aluminous cement, there is little description. Besides, aluminous cement can be cured in cold or hot water – the cheaper being to produce in cold water. Is there any benefit to ask for hot cured cement and, if so, what are the reasons and expected benefits?
Table 4 describes the main oxides present in various cements. Portland cement has a much higher content of CaO than aluminous cements (approximately 40% in a final mortar mix for a content above 60% when considering a pure cement paste prior to mixing with other additives). The possible transformation into Gypsum (larger crystal) could lead to insulator failure through cement expansion (called cement growth) if the batch of cement is prone to expansion. This is the reason it needs to be tested as per ASTM C151. Aluminous cements have much less CaO and much more Al2O3. The figures in this table are typical of pure cement but, in fact, manufacturers use mortars where the cement is the major but not only component. The rate of Alumina should therefore be considered around 30% in an industrial mortar used in the assembly process of insulators.
Checking for Portland versus aluminous is easy. Indeed, some utilities have initiated a chemical check of a sample of insulators received after they discovered that they had been sent Portland cement assembled units even though the drawing called for aluminous cement. A simple chemical analysis will show the respective contents of CaO and Al2 O3. Aluminous cements should be comprised of at least 30% alumina oxide.
Curing aluminous cement in water can take place through two different crystallographic patterns. If cured in cold water (i.e. ambient temperature), crystals will take on a hexagonal shape that is unstable and will progressively convert into a stable cubical stage. This conversion process can take years unless the cement is cured in hot water (around 70°C) in which case conversion is immediate. Obviously, hot curing will be more costly yet it leads to more stable mechanical performance. It has also been demonstrated that when cold-cured aluminous cement is progressively converted from hexagonal to cubical shape there is a drop in strength (which eventually will be restored when conversion reaches completion). Fig. 1 describes this phenomenon.
The evolution in strength of cold-cured cement cannot be predicted in time and value. To better determine such effects, a residual strength test can be a good indicator. Another approach to differentiate cold from hot-cured is based on thermo-gravimetric analyses (TGA) with identification of hydrates present in the mortar. CAH10 (CaO, Al2O3, 10H2O) and C2AH8 (2CaO, Al2O3, 8H2O) are the hexagonal unstable hydrates described previously. C3AH6 is the cubic stable hydrate. The cold-cure process introduces temporary hydrates such as CAH10 and C2AH8 that progressively react to become C3AH6 (3CaO, Al2O3, 6H2O).
Among the reasons for strength to be lower during this unstable phase, is difference in density of these hydrates, meaning higher porosity and therefore lower strength (as per Table 5).
The TGA spectrum shown in Fig. 2 compares cold-cured and hot-cured aluminous cement. It is easy to discriminate which is being used in insulators and a number of utilities leaning towards aluminous cement now specify only hot-cured. So does the CSA. Evaluation of the performance and type of curing can also be verified through a residual strength test, as described further below.
Purity of Glass
Unlike porcelain that by nature has a heterogeneous structure made of a variety of crystals, glass is amorphous and has no structure. This makes glass a perfect dielectric material that does not age. However, during the glass melting process it is possible to have impurities in the melt. Among other reasons, this has a direct impact on quality in regard to ‘self-shattering’. These impurities usually, but not only, come from wear and tear of refractory walls.
While a broken glass disc (also called ‘stub’) is not a problem for the performance and safety of a line, excessive numbers of stubs could become a concern requiring maintenance. Although there is no standard dealing with this issue, utilities should systematically request performance certificates from other utilities where self-shattering rate has been reported. Ideally, such certificates should refer to large quantities supplied outside the manufacturer home country to ensure complete objectivity. At least 3 such certificates that deal with deliveries in quantities of more than 100,000 units in service for 10 years should be requested. A good benchmark for acceptable self-shattering rate is 1/10,000/year.
Table 3 above describes several thermal shocks intended to help weed out glass shells that contain impurities. Major manufacturers have implemented specific additional tests to further improve the quality of glass targeting some specific impurities. Technical specialized literature on glass purity refers to procedures used in the flat glass industry such as “soak test” or others, and some manufacturers advertise a full compliance to such treatments to explain their quality strategy. Reality is more complex since glass discs are not flat and do not have regular thicknesses across their volume. Based on the above-mentioned research in the early 1980s, Sediver has customized such processes to take into consideration these particularities demonstrating a consistent benchmark level with a shattering rate at or below 1/10000. The quality level reached today is a combination of the acquired knowledge from the factory quality indicators and line surveys where actual shattered discs are being counted. While there is no standard describing the optimum processes involved (this knowhow cannot be disclosed by manufacturers), there are still many ways to make the evaluation of the effectiveness of the glass manufacturing processes through actual performance certificates from the field as explained earlier.
Molding & Toughening
Molding and toughening are important steps in the process and need to be considered carefully. The quality of the insulator could be compromised with glass shells containing defects from any of these operations. Section 3.2 will describe test procedures which can be helpful in the screening of these aspects, and it will appear very obviously that steep front test really makes sense in a sample test plan intended to check for consistency. Toughening is what makes glass strong enough to be used for overhead transmission lines. Some will try to convince the market that “the more the best”. In fact, this process requires a careful approach and both product shape and tools have to be designed as a combination. Fig. 3 shows a typical example of a digital simulation of the cooling process during toughening showing how intimately product and tools are connected to precisely match thermodynamic requirements.
Some utilities that acquired insulators from various suppliers have reported surprises when dealing with breakage patterns and therefore questioned the quality of toughening. This is much different from what happens when small chips or flakes appear after strong impact on the glass surface. This is in fact always possible and acceptable when the flake remains in the volume of glass under compression.
Some engineers and experts recommend making the assessment of the quality of the toughening by breaking samples of glass based on size and distribution of the glass pieces. The impact strength test could be reviewed with a description of the typical breakage pattern expected.
Type Tests & Sample Tests
Residual Strength Test
A residual strength test is performed to demonstrate the ability of an insulator to keep a minimum mechanical strength once it is damaged. This mechanical test is therefore performed on broken discs with the skirt removed if it is porcelain and a stub if it is glass. By design the mechanical load is transferred between cap and pin mostly through compression. Therefore, the design of the head of the insulator as well as the cement are critical components which performance can be assessed through this test.
In the case of cold cured aluminous cement new insulators would show very high values, but progressively, if tested while the conversion (section 2.2) is in progress the strength could drop close to half of their initial values. The phenomenon can be artificially accelerated if the samples are immersed for some time in hot water. Figure 7 shows test results on cold cured aluminous cement insulators which were immersed for various times in hot water. The dip is obvious and can reach 50% of the rating of the insulator when applying a 2σ standard deviation.
Today, ANSI does not specify any thermal pre-conditioning prior to the residual strength test. IEC and CSA call for a pre-conditioning at around 75°C maximum but this may not be high enough to make this risk visible. All standards should ideally include pre-conditioning at temperatures of around 85°C or 90°C. For hot-cured cement, since conversion is completed during manufacturing, such a test should show no variation in strength. Among the reasons for being careful with residual strength of insulators, consider the U.S. where new NESC Guide Rule 277 allows utilities to load lines up to 65% of the rating of insulators. Current standards, as shown in Table 1, do not offer a buffer. To this effect, a value of residual strength after thermal pre-conditioning of 80% seems an adequate move to assure reliability of insulators. This test needs to be a sample test, possibly in addition to chemical verification of the cement itself.
Steep Front Test
IEC has developed a steep front test standard (IEC 61211) to test dielectric under severe overvoltage conditions (2.5 or 2.8pu) and without need for oil (as per the traditional oil puncture test where the dielectric is tested under power frequency conditions). Today, there is sufficient data available to support this steep front test as replacement for the oil test. This test identifies insulators that contain defects that can be generated during manufacture, whether micro-cracks or structural defects in porcelain bodies or molding and other defects in toughened glass insulators.
It has also now been established that when doing a puncture test in oil the main parameter governing the result is not the insulator but the oil. Yet little is being said about oil characteristics in ANSI. CSA and IEC at least give a resistivity value to be set between 106Ωm and 109Ωm but nothing about dielectric strength. Poor quality insulators can go through the oil test while good insulators could fail because of thermal gradient in the oil if it is too resistive. Standards should ideally all converge toward steep front testing, provided it follows IEC 61211. Since this test offers a good indication of insulator quality, it should be recommended not only as a type test but also as a sample test, through a random selection from a batch of insulators.
CSA calls for the steep front test not only as a normal type test but also after the thermo-mechanical test. By pre-stressing the insulator, this test simulates the behavior of an insulator not only when it is new but also after several years of service (where it may have been subject to a variety of environmental and service condition stresses).
Looking at various standards applicable to insulators, there could be real value in harmonizing all standards around CSA parameters. These are more stringent, especially with regard to thermo-mechanical pre-conditioning. Indeed, insulator failures during benchmark sessions have been recorded under CSA whereas no issues had been visible with other testing protocols. In each case, an assembly or glass quality problem was traced back as being the root cause.
There are no RIV values specified in the standards, except for ANSI. At the same time, there is a debate around the possible introduction of an RIV test in IEC. RIV requirements can be a good indicator of quality of some aspects linked to shape or assembly of an insulator. Specified values should, however, take account of such parameters as size of insulator and possibly maximum voltage seen by the most stressed unit in the string (i.e. at the line end). This can be done either by carrying out a voltage distribution evaluation in a laboratory or by making an assumption, such as 10% or in some cases 15% of the phase-to-ground voltage of the line under consideration. Having a specification with low values will show little if anything. Taking values that are too severe could lead either to over-design of the insulator (in cap size) or to using artificial means to meet such values. An example of the latter is where one manufacturer used a coating to cover the cement but that rapidly disappeared after several hours in a mild salt fog environment and subsequently resulted in higher RIV values.
RIV generated by bad cap-to-glass seal (or connection) is usually much more difficult to generate than from the pin side, unless the gap under the cap is extremely poor. Photos below show an example where the gap is on the order of 2 mm, leading to corona but at a high voltage level. Similarly, manufacturers, in some cases, use a plastic ring under the cap. Often made with poor plastic, these will degrade over time and temperature. RIV values would then progressively increase. Such a solution should therefore be avoided to the benefit of a classical ‘flock’ deposit at base of the cap.
Other ideas, such as a RIV test under wet conditions, have been described based on different behavior during test observations. However, caution is needed here since it is important to understand whether differences can be correlated to weaknesses, defects, flaws or low quality designs. This is not yet clearly established.
The following Table can be a basis for future upgrades or changes in standards or specifications. Beyond the tests themselves, it is key to understand that random sampling to perform tests, so far only performed at the level of type tests, can prove highly instrumental in demonstrating consistency and quality.