|NDT.net March 2004 Vol. 9 No.03|
| 2nd MENDT Proceedings
DETECTION OF AGEING HIGH-VOLTAGE ELECTRICAL CABLES USING FLUORESCNECE SIGNALSE. Hegazi
Center for Applied Physical Sciences, Research Institute,
King Fahd University of Petroleum and Minerals, Dhahran 31261,
Corresponding Author Contact:
ABSTRACTSamples of XLPE insulation material from electrically damaged and non-damaged underground high voltage cables were examined using laser-induced fluorescence techniques. Four laser sources of wavelengths at 355nm, 457nm, 488nm and 514.5nm, in the CW and pulsed modes, were used as excitation source for the XLPE material and the resulting fluorescence signal were measured and processed in the spectral region between 380nm and 700 nm using a high-resolution SPEX monochromator. It is found that the intensity of the fluorescence signal and its spectral shape vary systematically with the ageing status of the cables. This suggests that it is possible to use such emitted fluorescence signal as a tool to monitor the deterioration of high voltage underground cables and also to monitor, non-intrusively, the condition of the newly-manufactured insulation clad on-line. The paper studies also the secondary effect of the laser radiation itself on the intensity and shape of the fluorescence spectra.
INTRODUCTIONElectric power is distributed from main power generators to consumers through underground distribution cables operating at voltages of about 8kV. The structure of these cables consists of four standard layers. In the core there is a standard conducting aluminum (or brass) cable, followed by a semiconducting polymer layer to smooth out electric field inhomogeneities due to possible imperfections in the conducting core, followed by a thick insulation layer made of crosslinked polyethylene (XLPE) to prevent breakdown, and finally followed by an outer rubber-type shield for physical protection against the environment and for other electrical purposes. A new (i.e., neat) distribution cable usually has an electric breakdown strength of about 1200 V/mil for 175 mil thick XLPE insulation, which in theory should be well within the operational safety margin. However, this is not always the case. Some of these cables breakdown in an unexpected manner long before reaching their presumed lifetime, typically once their local breakdown strength drops down to about 200V/mil .
The interest here is not to identify the circumstances that lead to these breakdowns but rather to investigate possible tools that can be used to predict them before happening. One of the promising tools that can be utilized in this subject is laser-induced fluorescence spectroscopy. When a molecule absorbs light energy it becomes excited to a higher energy level for a short time before it loses back this energy. If the loss of this energy is in the form of re-emitted radiation it is called luminescence. Fluorescence is a form of luminescence that occurs in a very short time, i.e., within picoseconds to nanoseconds, (immediate loss of energy), and fluorescence spectroscopy is the technique that analyzes the emitted fluorescence signal.
Fluorescence spectroscopy is a very sensitive technique when studying the chemical reactions of any fluorescing species in the material. When dispersed, the observed fluorescence signal from these species can provide a very useful spectrum that carries within it information about the condition of these fluorescing species and consequently about the deterioration condition of the whole material that houses them. Therefore, if any of the three layers that wrap the conducting core of the distribution.4 cable does include such fluorescing species then these species can be used as probing subjects to monitoring the deterioration condition of the cable . An interesting layer in the distribution cable is the XLPE insulation material. Although, in itself, the XLPE material is not a fluorescing material, the additives that are incorporated in it to resist all kinds of thermal, physical, and electrical stresses are mainly fluorescing species [3,4]. With use, these additives become affected and start losing their protective characteristics that the cable relies on, which in turn contributes strongly to the deterioration of the cable itself. The way these additives “age” with use can be easily monitored by investigating their fluorescing capabilities, and then can be used to estimate the ageing status of the cable.
In the present work we investigate the fluorescence spectrum that is emitted from the XLPE insulation material upon excitation with four different wavelengths 355nm, 457nm, 488nm and 514.5nm. Three different distribution cable conditions have been used: neat (unused), used (but not electrically damaged), and electrically damaged. The work reports also on the secondary photo-degradation effect that the cables could suffer using laser radiation, which if not taken into consideration, could lead to false information. We also compare our results with those obtained by Ref. 5 in which, we believe, some peculiar artifacts due to irregularities in the spectrometer appeared in their data and consequently in their conclusions.
EXPERIMENTAL SETUPThe experimental setup was typical for a straightforward fluorescence spectroscopy and is shown in Fig. 1. Samples, cut in wedged shapes, were irradiated with a laser beam and the resulting fluorescence was focused onto the entrance slit of a scanning monochromator. The exit slit of the monochromator was fitted with a photomultiplier tube to detect the dispersed fluorescence signals and to send them, in turn, to a digital multi-meter and a computer for further analysis and plotting. Four laser beams were used in the study; namely, blue ( 457nm), greenish blue (488nm), and green (514nm) which were all produced using a CW Argon laser (Spectra Physics), and ultraviolet laser beam (355 nm) form the third harmonic of a.5 pulsed YAG laser (Spectra Physics GCR). Different laser powers were tried in the case of Argon laser which ranged between 20mw and 1w, while only one value of energy was used for the YAG laser (2 mJ/ 10 pulses). Two monochromators were used in an alternative manner in the setup to detect any possible undesirable background due to irregularities in the dispersive elements themselves. The first was a was a low resolution 0.3 m monochromator (photophysics) equipped with a 500nm blazed grating with 1100 lines/nm similar to the one used in Ref 5, and the second is a high resolution 1-m SPEX. The entrance and exit slits in both monochromators were kept fixed at 2mm. The photomultiplier used was a side-on Hamamatsu R818 covered with a suitable filter to block the laser beam wavelength. The digital multi-meter was an HP 3478A and it was interfaced with a personal computer.
Several cable samples were provided to us by a local electric company (SCECO) with mainly three different aging conditions; namely, new (neat), used but not damaged and damaged. Two other neat samples from other suppliers were also used. However, all of the neat samples from the different suppliers showed exactly the same spectra; hence we concentrated our study to only the samples given to us by SCECO. Care was taken to ensure that the position and orientation of each of the wedged XLPE samples was the same for all of them. This was done by using a holder that was kept fixed relative to the collective lens system in front of the monochromator and to the angle of the impinging laser beam.
RESULTS AND DISCUSIONThe resulting fluorescence emission spectra of the neat XLPE insulating material in the visible region, when irradiated by CW laser radiations at 488nm, 457nm (50mW each), and by pulsed laser radiation at 355nm (2 mJ) are shown in Figure 2. The monochromator used was the low resolution Photophysics brand. The intensities at the emission wavelength of 508.8nm for all of these three spectra have been normalized to unity for comparison purposes only since the intensities of the spectra are not necessarily relative to each other. The three peaks indicated by asterisks in the diagram do not actually rise from the fluorescence emission itself but rather from weak Raman bands of some carbon-hydrogen stretching vibration modes.
It is clear from Figure 2 that the insulating material of the cable does give rise to a fluorescence spectrum in the visible region extending up to 700nm. This is interesting because it means that the material could be excited by visible laser wavelengths which are cheaper to obtain, more flexible to deal with in fiber optics, and less damaging to the photo-characteristics of the cable than UV lasers. Fluorescence spectra excited by the green wavelength at 514.5nm will be also shown later. It should be mentioned that the weak peaks at about 580, 620, 680nm appearing on the fluorescence spectra of the 488 and 457nm excitation wavelengths in Figure 2 are mere artifacts as will be explained shortly.
The fluorescence from the insulating material originates from the protective additives that are included and not from the XLPE material itself. When these additives deteriorate with use the fluorescence response from them will be different from its original case. This situation can be seen in Figure 3, which shows a comparison between the fluorescence spectra of neat (unused), neat (used but not damaged), and damaged cables for the same excitation wavelength, i.e., pulsed 355- nm laser radiation. The resulting fluorescence spectra here are plotted without pre-normalizing the intensities. It is clear in this Figure that the fluorescence spectra vary systematically with the deterioration case of the insulating material in two ways. The first is in the overall intensity of the fluorescence signal, i.e., the more the cable is deteriorated the weaker the fluorescence band from 470nm to 700nm will be. And the second is in the change of the shapes of the spectra themselves. For example, the ratio between the intensity at 500nm to the intensity at 600nm (I500/I600) decreases systematically with the deterioration case. This is an interesting result because, in principle, one can use a ratio of this sort to monitor the status of the insulating material from which the ageing of the electric cable can be predicted. There will be no need to use a scanning monochromator to do that. It suffices one to use only two photomultipliers fixed at two different monochromator settings, for example, one at 500nm and the other at 600nm, and then measure only the intensity ratio at these two wavelengths. However, care must be taken to avoid monitoring the intensities at the weak narrow peaks that appear on top of the broad spectrum as shown in the damaged cable case in Figure 3, since they are not related to the fluorescence spectra themselves.
There is another interesting piece of information that can be derived from the fluorescence intensity even without scanning the monochromator. This has to do with how the fluorescence intensity at any particular emission wavelength decays with time. In this case it is the photo-degradation properties of the additives that will be used as a tool to monitor the status of the additives. In Figure 4, we show the decay of the fluorescence intensity at 540nm as a function of time exposure to the CW laser radiation for the XLPE insulating material of the neat (not used) and damaged cable samples. The excitation wavelength here was at 488nm and the laser power was at 50mW. It can be seen that, initially, the fluorescence from the damaged cable decays much faster than that from the neat cable. This decay then stabilizes after a while, indicating that the spots that are being irradiated with the laser beam have been totally photo-degraded. This stabilization in the fluorescence intensity decay occurs at an earlier time in the case of the damaged cable than that of the neat cable, which means that the fluorescence lifetime becomes shortened as the cable deteriorates. This is also an interesting result, which can be used to determine the ageing of the cable if a careful calibration of the fluorescence intensity decay as a function of laser exposure time is made. It is important to note here that even though using the laser to photo-degrade may seem as a destructive test its actual effect is in fact very negligible because the size of the laser beam is very small (<0.25-cm diameter) compared to the dimension of the cable.
An investigation was carried out to identify the origin of the small narrow peaks that sometimes appear superimposed on top of the broad fluorescence emission bands. This issue is important to clarify in order to choose the right wavelengths at which the intensities of a fluorescence emission spectrum should be measured. In Ref 5, Menzel gave a rather elaborate description of the way the geometrical positioning of the cable sample in front of the laser beam depended on the accuracy of the resulting fluorescence intensity. This prompted us to repeat our measurements under several different conditions. We noticed that there was a different system of spectral lines that always appeared underneath the true fluorescence spectrum of the cable, which was not paid attention to in Ref 5. This system becomes more prominent as the fluorescence intensity from the cable becomes weaker. Figures 5a and 5b demonstrate this behavior.
In Figure 5a the same sample was irradiated with the green light of the Ar Ion laser of 514.5-nm and at three different laser powers, 1W, 250mW, and 50mW, while keeping the exact same geometry fixed in all three cases. It can be seen that the resulting fluorescence spectrum from the XLPE material underwent two different types of changes. One is due to the photo-degradation effect as described above, and the other is due to the appearance of the narrow lines. The weaker the laser power, and hence the weaker the resulting fluorescence signal, the more prominent these lines appear in the background as it is clear in Figure 5a.
In Figure 5b a comparison between the fluorescence spectra of the neat and the photo-degraded XLPE samples, at the same geometrical positioning, is shown. In this particular case the geometrical setting was chosen such that the intensity of the narrow lines at the background and was minimized. It can be seen that the shapes of the two spectra in Figure 5b are slightly different from each other due to the photo-degradation effect as expected and that, again, the weaker one at the bottom shows more background than the more intense one at the top. Therefore, using as little laser intensity as possible to avoid the photo-degradation effect produces, at the same time, an increasing background effect that could change the measured intensity ratios at many wavelengths.
Several measurements with and without optical filters and with different excitation wavelengths were made to identify these narrow lines. At some point, the low resolution monochromator (Photophysics) was even replaced by a second one of the same brand but still this did not get rid of the background. Finally, a completely different material was used in place of the XLPE. The material was almost 100% transparent to the visible light and was used as a mere light scattering object to record only the background. Figure 6 shows the resulting background along with the fluorescence spectra of the XLPE sample under two conditions: a) with minimal background and b) considerable background. One can clearly see that the same exact background still appeared even when the XLPE sample had been removed. This simply means that the background has nothing to do at all with the XLPE material. It was only after replacing the monochromator with a completely different brand that the background was found to disappear completely.
This can be seen in Figure 7 in which the fluorescence spectra from the neat XLPE sample and the photo-degraded sample were excited by a CW laser of 488nm and were measured using a 1-m SPEX monochromator instead of the low resolution one. Therefore, it became immediately conclusive that the low resolution monochromator of the Photophysics brand, of which we happen to possess two of them and of which Menzel also happened to use in Ref 5, was the reason for produce the unfortunate deformation in the dispersed spectrum (i.e., ghost lines). Therefore it is important, when dealing with fluorescence signal of such small intensity that a good quality dispersive grating should be used otherwise the intensity ratio measurements will not be accurate.
Finally, we should note here that it would be better to replace the whole detection system with a multi-channel CCD camera fitted with a reasonably good quality dispersive grating to minimize also the photo-degradation of the XLPE sample.
CONCLUSIONSamples from the XLPE insulation layer of high-voltage electric cables were separately irradiated by one of the visible CW laser radiations at 457, 488, and 415.5nm, and by a pulsed laser radiation at 355nm. The resulting fluorescence emission spectra from these samples were then measured in the visible range using a scanning monochromator. It was found that the shapes of the spectra changed systematically with the electrical condition of the cable, such that the ratio between the intensity of the fluorescence signal at 500nm, say, to that at 600nm was found to decrease systematically as the electrical condition of the cable deteriorates. Such information could be effectively used to monitor the ageing of the electric cable and to predict its breakdown. It was also found that the photo-degradation of the XLPE insulation using a laser beam could also be used to determine the status of the cable. Damaged cables were found to produce fluorescence signals of shorter decay lifetimes than those of neat cables. This is also an interesting result which makes it possible to monitor the ageing of the electric cable by testing the temporal characteristics of the fluorescence signal while a small spot of the cable is being photo-degraded. The present work also found that, because the excitation laser radiation is preferred to be weak, the resulting fluorescence signal will also be weak, which will necessitate the use of good quality dispersive gratings to avoid any significant overlap from other spectral backgrounds.
ACKNOWLEDGEMENTThe author acknowledges the support of the Research Institute of King Fahd University of Petroleum and Minerals. He also wishes to thank Mr. A. Hamdan and Mr. J. Mastromarino for their help in the initial stages of this work, and Mr. K. Soufan, Dr. B. A. Chokri, and Dr. M. H. Shwehdi for donating the samples.