Cable insulation test voltages: how high is too high?
Stephen Drennan - Electrical Engineer
There are many ways of testing the insulation of electrical equipment using a variety of voltages, frequencies and test methods. The Megger Group supplies a comprehensive range of testers for such applications, from 50 V to 15 kV insulation resistance testers, through VLF and AC Tan Delta test sets to diagnostic Dielectric Frequency Response instruments and HiPot or proof testers using AC or DC up to 80/800 kV.
This article sets out to resolve the confusion that sometimes exists regarding acceptable voltage levels for cable testing, and what is meant by “DC testing” of cables in various contexts.
Defining the terms
When used in a specific context, many engineering terms have a clear, well defined meaning. When they are taken out of context or used casually, however, those same terms can become ambiguous and confusing. A good example is “high voltage”.
Many national and international standards spell out unambiguously the voltages that can properly be designated as ELV (extra low voltage), EHV (extra high voltage) and everything in between. In common usage, however, the phrase “high voltage” means very different things to a commercial HVAC engineer used to working at 110 or 230 V, a distribution engineer working with 11 kV systems and a transmission engineer whose work involves 132 kV or 765 kV transmission lines. Test equipment is often used across these disciplinary boundaries and, partly at least because of the loose use of terminology, there can be confusion about which test voltages and methods are appropriate – and which are potentially harmful – in particular applications.
The problem – XLPE solid dielectric cables
Concerns about high voltage testing arose as a result of the behaviour of XLPE cables when they were subjected to the same maintenance regime as had previously been applied to laminated cable types. In the early 1990s, some invaluable research into the factors affecting the aging of XLPE cables was carried by Dr N N Srinivas at EPRI (Electric Power Research Institute), and others such as Dr M Mashikian at the University of Connecticut and Prof.dr.ir. F.H. Kreuger at Delft University.
DC testing versus overvoltage testing
The resulting papers all refer to what is variously known as ‘proof testing’, ‘withstand’ testing or ‘hi-pot’ testing, by which they mean that ‘high’ voltages (there’s that word again), relative to the working voltage of the system, are applied to cables to see if breakdown occurs during the test. For example, a 40 kV test voltage might be used to test a 15 kV system cable. In the context of these cable-testing regimes, the papers also refer to ‘DC testing’ to differentiate it from AC testing at similar voltages. The researchers are not, however, talking about all DC tests irrespective of the voltage used – after all, a multimeter uses a DC voltage from 0.5 to 2.5V to test for continuity but this certainly wouldn’t be included! The researchers are concerned only with ‘high’ DC voltages, but what does ‘high’ mean in this context?
The EPRI investigation
The EPRI report begins by stating “DC high-voltage testing of cables is used in an effort to detect gross imperfections or deterioration…”
In this context, the voltages in question are made absolutely clear in statements such as:
“DC testing at 40kV will cause a reduction in life of an accelerated-aged XLPE insulated cable” and
“DC testing at either 70kV or 55kV prior to aging does not appear to influence the cable life.”
The investigations looked at three classes of XLPE cables: those that were new, those that had been naturally aged, and those that had been subjected to accelerating aging in the laboratory by running them at about twice normal working voltage at high temperature.
The samples were then split into two groups, and one group was subjected to a DC overvoltage test, while the other group was not. The DC overvoltage test voltages applied to the cables were from 3.8 to 5.2 times the AC design voltage of the cables, typically 40 kV to 68 kV. Both sets of samples were then further run at ‘accelerated aging’ AC voltages and the final failure time of the samples compared.
With some tests, the samples that had been subjected to DC overvoltage testing failed sooner than the untested samples. For example, two of the tested cables failed after 346 and 887 days while their untested counterparts survived for more than 928 days. The results were by no means clear-cut, however, as 32 other cables in the study showed no statistically significant difference between the tested and untested samples.
EPRI Research Conclusions
Nevertheless, in the light of earlier laboratory work on the microscopic analysis of water-tree generation, and taking into account the limited ability of DC overvoltage to induce failure during test, the researchers concluded that while DC overvoltage testing on new cable involved no risk of cable degradation, there was a potential risk of accelerated aging on already-aged XLPE cable.
The research suggested that the problem with overvoltage testing related to the induction of an electric field in the insulation of the order of 230 V per thousandth of an inch (a ‘thou’ in British parlance or a ‘mil’ in American). In the metric world, this is equivalent to 9050 V/mm. This electric field is a problem for significantly aged XLPE insulation as the dielectric strength of such a cable can fall below 300 V per thousandth of an inch (12,000 V/mm). From this point on, overvoltage stress can measurably damage the insulation.
The research also established that when the
insulation is new, its dielectric strength is of the order of 1,100 V per thousandth of an inch (44,000 V/mm). This is about four times the field strength produced during DC testing which will, therefore, have no effect on new insulation.
With the above in mind, it’s important to realise that the voltage stresses produced in the cable are not simply a result of applying a DC voltage – an AC overvoltage will also accelerate aging – but are primarily due the high voltage used in overvoltage tests.
Under-voltage insulation tests
Not all insulation tests on cables are carried out at high voltages. In fact, many DC tests are typically carried out 2.5 kV or 5 kV. These electrical stresses resulting from such tests are only one-eighth to one-sixteenth of the dielectric strength of even a badly aged XLPE cable. There is no evidence at all that this causes any problem for the insulation. These values are in fact significantly less than the voltage/dielectric strength ratios which have been proven to give no problem to new XLPE cables.
Under-voltage DC insulation testing can therefore be used as part of commissioning and maintenance procedures without concern about damaging XLPE cables. Indeed it is frequently used by utilities with, for example, a 10-minute test time between each phase and screen, with a pass level of 10GΩ. Other utilities use this form of testing, in conjunction with other tests, to check consistency between phases.
Why DC Overvoltage Testing is an issue for XLPE
Although both DC and AC overvoltages can accelerate aging, at typical test durations of say 30 minutes the issue for XLPE is much worse with DC than AC. This is because the electric field, being maintained in the same direction for the test duration, can create undesirable space charges within XLPE insulation; when the cable is subsequently re-energised these charges remain, causing very high localised stresses. The normal AC stress plus the space-charge can start an electrical tree in the insulation which may develop into a fault and decrease service life. It could take as long as 24 hours for space charges to dissipate after a DC overvoltage test and, in most cases, leaving a cable out of service this long is impractical.
Solutions for testing XLPE cables
It is reasonable to argue that any test intended to determine the status of insulation should evaluate the system under test in a manner as close as possible to its normal operating conditions. So, for a cable system intended for service at AC power frequency, an AC overvoltage test at 50/60Hz might be considered the most representative test, particularly as the field-reversals will avoid the generation of persistent space-charges.
However, at power frequency a cable appears as a large capacitive load, with values of 300 pF/m being typical. A 500 m length of 66 kV cable being tested at 100 kV AC will, therefore, present a capacitive load of 470 kVA. Clearly, supplying this load would demand a large, heavy and very costly test system. And, if such a test system were powered from a single-phase 400 V source, the input current required would exceed 1 kA! Even if a series resonant test set which reduces the input power requirement were to be employed, it would still be large and expensive. However, sometimes there is no alternative and access to this kind of specialist equipment is needed from time to time by some utilities and by many equipment and cable manufacturers.
In the general field environment, however, VLF (very low frequency) testers often provide an acceptable and much more convenient option, but these still require careful consideration of voltage levels and test techniques.
VLF AC Techniques
Clearly a reduction in the capacitive loading effect of the cable will help to facilitate practical testing so VLF testing is conducted at frequencies below 1 Hz. Reducing the test frequency to 0.1Hz, the frequency most often used for VLF testing, means that the output power needed from the tester is reduced by a factor of 500, which makes it a much more practical field-testing proposition.
The IEEE Guide for Field Testing cables using VLF (IEEE 400.2, Table 1 ) summarises the VLF test voltages applicable to various cable types, dividing each into categories for installation, acceptance, and maintenance testing.
Test durations for VLF testing are significantly longer than those for 5/10 kV DC testing. Recommended durations are usually 30 or 60 minutes per test, which can make for a lengthy process when it is necessary to test each phase separately.
Field research on test-related failures has been carried out to back up the original laboratory research that led to the development of VLF testing. This showed that “VLF tests at IEEE Std. 400.2 levels do not significantly damage cable systems”.
Note however, that this research also warns against raising the recommended VLF voltage values for field-aged cables in an attempt to shorten the test time (to say 15 minutes per phase) as this can cause multiple failure problems .
New cables, in contrast, can take higher voltages as defined in, for example, IEC605202-2, which includes a test voltage for new cables of 3Uo at 0.1Hz for 15 minutes . So this is another instance where it is necessary to be clear about how high is high enough and how high is too high!
One might consider that the very low frequencies employed might not adequately represent the stresses in the cable when it is operating at power frequency. For this reason, an adapted waveform known as “cosine-rectangular” is often used on VLF test sets. This waveform is substantially a square wave with rising and falling edges that closely match the slope of a power frequency sine wave. This means that the stresses produced in a cable when testing with a cosine-rectangular waveform are more representative of those the cable will experience in normal operation.
Cosine-rectangular VLF for testing longer cables
This cosine-rectangular waveform is recommended by IEC, DIN VDE standards, the HD620
harmonisation documents and IEEE400 . The CIGRE paper on experience of testing cables in the USA did not find any significant difference in the diagnostic capability of sine versus cosine-rectangular wave shapes, but the cosine-rectangular equipment allows tests to be performed on loads with higher capacitance and so makes it possible to test longer cables than those which can be tested using a comparable sinusoidal test set.
Horses for courses…
Practical advice for field testing of cables can be summarised thus:
1. Insulation testing at 2.5 kV or 5 kV (under-voltage testing)
- Can be carried out on HV and MV equipment, including XLPE cables, without fear of inducing faults, either as a low-cost, Go-NoGo test, or, on some equipment, as a diagnostic insulation test, utilising techniques such as step voltage, polarisation index or dielectric discharge. With the voltages and durations customarily used, there is no evidence for degradation of XLPE cable insulation.
2. DC ‘Hi-Pot’ proof testing at 40kV/70kV or above (over-voltage testing)
- Can be carried out when commissioning any new cable, although certain types of defect may be missed.
- Should not be carried out for testing of aged XLPE or other solid dielectric cables during the maintenance cycle but can be carried out on laminated cable types.
3. VLF testing
- Can be applied to both laminated-dielectric and solid-dielectric cable types.
- VLF testing uses a frequency of 0.1Hz, which resolves the issues with DC Hi-Pot (over 40kV) testing on XLPE or mixed cables because the direction of the electric field alternates.
- The reduced power requirement compared with power frequency testing means that the test equipment can be made transportable and it costs less.
- VLF can test long cables due to the low levels of current required and this capability is maximised using the cosine-rectangular test option.