By Dr. Olaf Goll
This is the second part of a 2-part series. The first part appeared in the June 1997 issue of Electricity Today. Correction: In last month's article the sentence beginning "Today, a voltage level of 3 volts and a test duration..." should have read "Today, a voltage level of 3 Vo and a test duration..." Vo is the cable's phase-ground voltage, derived from phase-phase voltage (usually printed on the cable jacket) divided by 3 (or Ø - Ø x 0.5774).
A combination of very low frequency and recovery voltage testing has been successfully demonstrated for on-site evaluation of the state of low-voltage cables that have been in service for some time. The technique is of particular importance for testing ageing polyethylene and cross-linked polyethylene insulated cables.
Hagenuk looks at the customer's experience whenever possible. The aim of our study was to check the suitability of the vlf method for testing medium-voltage cables using as much statistical material as possible. The data collected covered around 5500 tests, corresponding to a total cable length of 10 000 km. Both new and aging cable networks were tested.
The results are as follows:
a) Testing of new installations: On new installations 3500 tests were performed. Only six breakdowns occurred, with faults detected in both sleeves and cables.
b) Testing of serviced cables. The results of 2000 tests are available in this instance. About half of them concerned installations with unreliable sleeves. Some 150 defective sleeves and 50 cable faults were discovered. In the case of cable tests, without particular sleeve problems, 300 breakdowns occurred. Approximately 20 per cent of the cable was replaced. The criterion for replacement was more than 3-5 breakdowns during testing of each cable run.
c) All massive weak points will break down if the test duration is long enough
d) Field experience and research results show that the vlf test method is suitable for both lead paper and pe/xlpe cables. The test parameters for a vlf cosine square wave voltage 3Vo and a test duration of one hour arrived at in practice are confirmed by the scientific results.
Supplement to breakdown testing
The results of breakdown testing in respect to the continued serviceability of a cable installation are limited, as only locally restricted weak points such as mechanical damage, cavities and individual advanced water trees can be detected. Integral analysis of cable insulation quality is not possible. Assessment of the cable insulation with regard to quality and aging is possible in principle using dielectric diagnosis methods.
In addition to tan d (dissipation), which has been known for some time, the return voltage measurement method is also available. Experiments and on-site cable diagnosis have shown that this method, which has already proved itself in the case of oiled-paper insulation, is also suitable for testing plastic insulated medium-voltage cables.
Test Circuit
Figure 3 shows that the test circuit can be used for return voltage measurement.
The system consists of a dc voltage source for charging the cable, an hv switch S1 and a discharge device consisting of a switch S2 and a discharge resistor RD. The value of the resistor RD is dimensioned so that the sudden discharge does not cause any traveling wave characteristics in the cable. Following the discharge, the electrical measuring sensor is connected to the cable via switch S1. The digitized measuring signals are transferred to the PC for evaluation and filing. The voltage characteristics for such a measurement are shown in Figure 4.
The charge voltage, which is generally less than 2 Vo, is applied to the cable being tested for the time to (up to a few minutes). Following the discharge period td, the voltage at the cable rises again. The characteristic of the return voltage rises exponentially up to a maximum and then falls again with a larger time constant.
The voltage shape of this transient reaction is determined partly by the insulating material itself and partly by the aging of the insulation (such as water trees) and can thus be evaluated as a way of assessing the aging condition of the cable.
Figure 5 shows results typical of on-site measurement on three 20kV cable runs with different degrees of aging. The insulation of all three cables consisted of homopolymer xlpe. Cable 1 (length 600 m laid in 1989) shows a small return voltage of about 4V at the peak. Return voltage values of this magnitude represent a minor degree of aging in this cable type. The measurements performed on cable 2 produced a return voltage of around 9V. This result indicates more advanced aging than in the case of cable 1. The third curve shows the result measured on a badly damaged cable. The return voltage increases to almost 20V. Based on the present level of knowledge, it can be said that the greater the absolute value of the return voltage, the greater the damage to the cable. This result is supported by comparative tan d measurements at 0.1Hz on the cable runs. The measuring results for both procedures and in the cable data are summarized in Table 1.
Both methods were used to perform tests on cables with different insulating materials. As the cables had not been operating long, it can be assumed that aging is negligible. The results of these measurements are summarized in Table 2.
The measuring results for cables 5 and 6 show both high return voltage values and tan d values, and indicate a high degree of ageing. These high values are due to the dielectric properties of the copolymer xlpe insulating material used for the cables.
The results clearly show that the measuring result is greatly influenced by the insulating material. Confusing cable aging from the absolute values of the measuring results is therefore possible only if the basic characteristics of the cable are known. In practice, this requirement can be met only with great difficulty, as mixed cable runs of different makeups are often laid and the basic dielectric characteristics of laid cables are not generally known during later investigations.
In the past, attempts have been made to improve the diagnostic method based on return voltage measurement with a view to obtaining results that are largely independent of the dielectric characteristics of the insulating materials. A procedure based on the depolarization current method is reported. Unlike the return voltage method, this procedure measures not the voltage but the current. It revealed that the size of the depolarization current is proportional to the size of the charge voltage Ñ that is, the greater the charge voltage the greater the value of the depolarization current. This relationship was established on non-service-aged medium voltage cables. If on the other hand the cables have severe water tree damage, the relationship is no longer proportional. In such cases, the value of the depolarization current rises disproportionately.
This behavior can also be observed in the case of return voltage measurement. Figure 6 shows the result of such a measurement on cable 1. The diagram shows the maximum values of consecutive single return voltage measurements in which the charge voltage was increased from 2kV to 2Vo.
The curve progression shows that the absolute value of the return voltage increases in proportion to the charge voltage. The diagram in Figure 7 on the other hand, shows the measuring result for a cable with water tree damage. (cable 3).
In the case of the damaged cable (3) it can be observed that the absolute value of the return voltage increases disproportionately in relation to the charge voltage. This result was also obtained when the depolarization currents in service-aged cable were measured. To assess the degree of damage, it is therefore possible to specify a factor that can be calculated from the ratio of the return voltage at a charge voltage of Vo and 2Vo. For the new cable (1), the factor is 2, while a factor of 3 is calculated for the damaged cable (3). Cables with a depolarization current factor of > 3 were reported to have had very severe water tree damage.
Independent Result
Using this evaluation and the procedure progression of return voltage measurement, it is possible to obtain a result virtually independently of the characteristics of the insulating material, as it is not the absolute value of the return voltage that is evaluated but the ratio of two measuring results at different charge voltages.
A cable diagnostic system is now available based on this technique, and is already in use at some utilities. For the evaluation the cable is charged with a low dc voltage ranging in steps from 0.5Vo to 2Vo and the returning voltage (also called recovery voltage) is evaluated with respect to its linear relationship to the charge voltage.
For the evaluation process this means that a new cable has a linear response: charge voltage and return voltage arc proportional to each other. The damage brought about by aging in an older cable produces an increasingly non-linear result. This non-linearity can vary from the factor of 2 for a new cable up to more than 5 for a completely damaged cable.
Extensive field tests with the first CD units have proved the qualification and capability of the new cable diagnostic system to evaluate the damage by water trees and the general aging condition of cables. The results of return voltage measurements of water tree size and their amount and following breakdown tests will be available soon.
The three-phase connection and measurement results in significant time-saving. Potential compensation of leakage current measurement is not required, which permits easy and rapid connection of the test leads.
External effects such as temperature, weather conditions and so on are of little significance and have been observed under only the most extreme conditions.
The CD System software operates with normal Windows software on a 386 or higher laptop or PC. The software itself requires very little handling from the operator, and all measurements can be prepared and set up in the office prior to the measurement. On site the test then requires only the start of the test routine and the manual activation of the high voltage by a push-button.
The complete test runs automatically, and for safety reasons requires only the manual permission of the "Charge voltage on" push-button for each sequence. After 40-60 minutes the test is completed and the software will display all measured values and curves and, of course, as a final result the "Diagnostic Factor". All these values and diagrams are directly prepared for printed report and storage. Special operator training is not required.
Conclusion
VLF tests are suitable for both pe/xlpe and mixed cable networks. This test method is good for revealing local singular fault joints and well-advanced water trees. With dielectric diagnostics such as return voltage measurement, only overall assessment of the cable insulation is possible. Detection of singular fault points is not possible. The combination of the two methods produces a powerful procedure that can satisfy the requirements of on-site testing of medium-voltage cables. Local weak points can be traced with the VLF method, and the degree of aging can be assessed with the return voltage method to achieve target-oriented reinvestment in cable systems.
Dr. Olaf Goll is head of research and development at Hagenuk GmbH.