Friedrich Enkert, Europe/CIS Sales Director, Megger explains how to diagnose faulty cables even if you've tried everything!
While power cables sometimes fail soon after being put into service into service, for example, due to bad workmanship, failure is usually associated with ageing of the insulation in a cable that has been in service for a long period and with the associated cable joints. The fact that many of these cables are underground present real challenges to identify, locate and repair.
Diagnostic testing
It's clear there is a need for diagnostic testing that will provide reliable information about the insulation condition of a cable and its susceptibility to failure. In theory, such testing would best be carried out at power frequency in order to replicate, as accurately as possible, the normal working conditions of the cable.
In practice, however, this approach is problematic. Cables present a highly capacitive load to the test set, which means if testing is carried out at power frequency, a large amount of reactive power is required to charge the cable.
To meet this requirement, a test set necessarily has to be physically large, heavy and expensive. For this reason, power frequency testing of cables is not commonly used.
An alternative is DC testing, using an insulation test set that typically operates at 1 or 5 kV. Such test sets are small, convenient and inexpensive, and when they are used to carry out tests of suitable duration, typically a minute, this is a particularly useful form of testing as it is non-destructive and provides a first-line insight into the condition of the cable insulation.
It has to be said though that while DC testing at these modest voltages provides a guide to the presence or absence of serious faults, it's less useful in providing information about general degradation and ageing of insulation. In addition, this form of testing does not model the cable's normal working conditions and so does not replicate the stress distribution in the insulation when the cable is operating at power frequency.
Detection of insulation failures at higher voltages may require high voltage tests, often at two or three times the working voltage of the cable. There is a serious concern that high voltage dc testing can unnecessarily damage the cable under test. In particular, polarisation of the insulation may course flashovers later when the cable is going back to service.
The water tree problem
Water trees start as minute inclusions of water vapour within the insulation at the production or installation phase and grow into tree-like structures. Water trees in themselves do not cause immediate cable failure, but the danger is a water tree may continue to develop until the field stress in the remaining insulation is too high and an electrical tree is started. This can lead to electrical leakage and failure of the cable insulation.
These problems and concerns led to the development of yet another approach to cable testing - the use of AC at very low frequency (VLF). Commonly, the frequency used is 0.1 Hz. At such a low frequency, the capacitive reactance of even a long cable is relatively low, so the charging current is manageable even for a conveniently small test set.
Because the test is carried out with alternating current, the cable is not subject to a unipolar applied voltage for long periods, so no polarisation will take place and the cable will not be damaged after putting back into operation.
Types of very low frequency (VLF) test sets
Commercially available VLF cable test sets can be divided into two categories:
• Those that use a test waveform that is sinusoidal
• Those that use a cosine-rectangular waveform
The advantage usually claimed for testing with sinusoidal technology is it lends itself to being combined with other insulation evaluation techniques, such as tan delta measurement and partial discharge (PD) diagnostics.
While this is true, it's important to remember testing is being carried out at a frequency that typically differs by a factor of 500 or 600 from the normal operating frequency of the cable. The PD results may not, therefore, reflect those that would be obtained if the tests were carried out at power frequency.
Although water treeing can be better identified at these lower frequencies, the tan delta diagnosis on cables is usually performed at 0.1Hz.
Cosine-rectangular test sets are much less power consuming than their sinusoidal counterparts.
At first glance, the cosine-rectangular waveform looks almost like a square wave.
Closer examination shows the rising and falling edges of the waveform very closely match the rising and falling edges of a sine wave with a frequency of around 50 Hz.
This means the stresses generated in the cable insulation by VLF cosine-rectangular testing are close to those it will experience in normal power frequency operation.
This wave shape is recommended by DIN VDE standards, the HD620 and 621 harmonisation documents and IEEE400 standard. It allows tests to be performed on loads with higher capacitance - that is, longer cables - than a sinusoidal test set would allow.
This article has largely concentrated on insulation testing as an aid to determining how likely a cable is to fail, for the very good reason that preventing or at least predicting faults is always preferable to reacting after they have happened.
However, cable faults do occur, and when they do the emphasis changes to locating them as quickly as possible, which can be difficult if the cable is buried underground.
Logical approach to cable fault location
The key to speedy fault location is to adopt a logical approach. After the cable has been isolated and made safe, the first step is prelocation to find the approximate location of the fault.
First, it is advisable to check the insulation resistance between all conductors and the cable sheath, typically using a kΩ range on an insulation tester or a multimeter. This determines whether it is a high resistance or low resistance fault, information useful in deciding the pre-location technique to be used. Using a kΩ range rather than an insulation test avoids conditioning the fault and altering its nature, which could make prelocation more difficult.
The next stage is to use a time-domain reflectometer (TDR). Again, this will not condition the fault but, for open- or shortcircuit faults, it will be able to show the distance to the fault. This is sometimes called pulse echo low-voltage prelocation.
If the fault is of a higher resistance, alternative methods need to be used. These are often called arc reflection methods and involve using a surge generator to apply a high energy, high voltage charge to the cable in addition to the pulse echo signal.
This generates an arc at the site of the fault effectively making a temporary short circuit that can be seen by a TDR. Available techniques include arc reflection, impulse current and voltage decay.
Finally, pinpointing can be used to identify the exact fault location. The most common method relies on detecting the acoustic and electromagnetic signals emitted at the fault location when the cable is impulse tested by a surge generator (thumper).
Cable testing to both predict and then deal with faults is a vital concern for all those involved with the distribution of electrical energy. A wide range of test techniques and test equipment are available to allow this concern to be effectively addressed, but cable testing can, nevertheless, be a challenging task.
For this reason, a resource that's as important as the test equipment itself is access to expertise that will help with selecting the best equipment for the job, and using it in such a way it delivers the best results. There are very few suppliers that can offer equipment and in-depth expertise that covers the whole gamut of cable test requirements but, following its recent acquisition of SebaKMT, Megger can do exactly that.© Power & Water Middle East 2013
