Application Note
PD fault location on cable lengths
Author Dr.-Ing. Kay Rethmeier |
[email protected] Date Aug, 2010 Related OMICRON Product MPD 600, MPD 540, CAL 542, CAL 543, 3-Band Software Option Application Area Cable, Cable length Version v2.0 Document ID ANP_10035_ENU
© OMICRON
Page 1 of 17
Content 1
Introduction: Partial discharge pulses propagate .............................................................................3
2
Locating PD faults: Using reflection....................................................................................................4
3
Determining the pulse charge: Using filters .......................................................................................6
4
Reflections: Influence on PD filters .....................................................................................................7
5
PD fault location with the MPD 600......................................................................................................8 5.1 Switching over to EXPERT MODE ("Mode" menu) ........................................................................9 5.2 Making the 3 MHz bandwidth available ("Advanced" menu) ..........................................................9 5.3 Select filter settings fCenter=1.6 MHz and delta f=3 MHz ("Q" tab) ............................................ 10 5.4 Finding the calibration pulse and echo (TDR) ............................................................................. 10 5.5 Measuring PD faults ("Measure" tab) .......................................................................................... 12 5.6 Statistical localization: PD fault location on long cables .............................................................. 13
6
Special aspects to be considered when using the CAL 543 online
7
Influence of non-sinusoidal test voltages ........................................................................................ 15
calibrator .................... 14
Please use this note only in combination with the related product manual which contains several important safety instructions. The user is responsible for every application that makes use of an OMICRON product.
OMICRON electronics GmbH including all international branch offices is henceforth referred to as OMICRON. © OMICRON 2010. All rights reserved. This application note is a publication of OMICRON. All rights including translation reserved. Reproduction of any kind, for example, photocopying, microfilming, optical character recognition and/or storage in electronic data processing systems, requires the explicit consent of OMICRON. Reprinting, wholly or in part, is not permitted. The product information, specifications, and technical data embodied in this application note represent the technical status at the time of writing and are subject to change without prior notice. We have done our best to ensure that the information given in this application note is useful, accurate and entirely reliable. However, OMICRON does not assume responsibility for any inaccuracies which may be present. OMICRON translates this application note from the source language English into a number of other languages. Any translation of this document is done for local requirements, and in the event of a dispute between the English and a nonEnglish version, the English version of this note shall govern.
© OMICRON 2013
Page 2 of 17
1
Introduction: Partial discharge pulses propagate To check production quality, partial discharge (PD) measurements are performed on cable lengths or cable drums. Unlike concentric elements, cables are subject to propagation processes which cannot be ignored. And while this allows PD fault localization to be performed, it also requires several special features when performing PD measurement. Due to reflections, partial discharges generally occur as pulse groups, which can lead to superimposed oscillations of the PD filter required to determine the charge value as per IEC 60270 [1].Depending on the PD fault location, and thereby also the pulse spacing, this can cause too high values(positive superposition), too low values (negative superposition) or even the PD signals to be erased completely [2]. The latter two cases are most serious here, as an existing PD fault is either not recorded in full or not detected at all. When setting the parameters of the PD measuring system, it is therefore vital to ensure that the filter bandwidth and band-pass position (center frequency) are chosen in such a way that no significant underestimate of the PD fault can occur. To offer a simplified summary of the best approach, we can say that a broad filter bandwidth should be chosen so as not to fall into individual spectral resonance dips, such as can be generated by double pulses. A PD filter set as wide as possible covers a large number of both resonance dips and excess peaks (see also Fig. 8),which allows a correct PD measurement to be made due to this averaging effect. This highbandwidth filter should operate with as low a frequency as possible (e.g. starting from 50 kHz), so that many spectral components of the PD pulses which are not subjected to any or only minor propagation loss can be recorded. This thereby allows long cables to be measured. As an additional feature, from MPD software version1.5.1 onward the "Negative Superposition Elimination Mode" is available. This uses three PD filters, all working simultaneously, to eliminate the risk of underestimating PD pulse groups.
© OMICRON 2013
Page 3 of 17
2
Locating PD faults: Using reflection As already mentioned in section 1, a single PD pulse also generates multiple measurable pulses that can be detected by the PD measuring system. Starting from the original location of the partial discharge, two charge pulses continue along the cable in both directions. Here, one pulse can make its way to the measuring device directly, while the second pulse, running in the opposite direction, must first be reflected by the far end of the cable before it then runs toward the measuring device and can be detected by the PD measuring system (TDR, Time Domain Reflectography). Figures 1 and 2 highlight the propagation and reflection process.
Figure 1: Propagation from the f ault location in both directions
Figure 2: Reflection of the (blue) portion running to the right at the far end of the cable
Due to multiple reflections, a single PD pulse can therefore be detected multiple times by the PD measuring system (see Fig. 3).
Figure 3: Approximately exponential attenuation of the calibrator pulse through multiple reflections
The time intervals that occur here between the pulses are constant and represent twice the cable length. Since the PD measuring system generally only measures at one end of the cable ("near" end, , working end, etc.), the pulses have to run through the cable twice to then be detected. The dropping amplitude of the measured pulses is caused by the signal attenuation in the cable. The cable acts as a distributed low-pass filter here, which can also have a more pronounced effect with long propagation distances. Via the pulse propagation speed v, the time differences t or _t and cable lengths L or PD fault locations s can be summarized as shown in equation 1.
© OMICRON 2013
Page 4 of 17
(1)
The propagation speed can therefore be determined when the cable length is known and the time difference has been measured. This is classed as sufficiently frequency-independent here and is determined by the materials used in the cable (εr) or its structure (e.g. thickness of conductive layer and insulation, resulting εr,res). Using equation 2, the propagation speed can be calculated for known geometries and materials.
,
,
(2)
Typical values of XLPE-insulated cables are 150 m/ µs to 190 m/ µs. Half the propagation speed, i.e. v/2, is often also used as the numerical value without explicitly stating this. In this case, the values are then in a range from 75 m/ µs to 95 m/ µs. Since the outgoing and returning signal, and thereby twice the run length 2 L, always has to be taken into account when performing the PD location calculations, these two factors of ½ and 2 cancel each other out in the calculation, which is generally viewed as a simplification. However, further procedures are always based on the actual propagation speed v. Unlike calibration pulses, which are fed in from one side of the cable, PD signals can also originate from the cable. This leads to a significantly different picture than the one shown in Figure 3. This is shown in the time domain chart (see Fig. 4).
Figure 4: PD fault with reflections
Here we can generally observe three different, yet repetitive pulse intervals. The time interval between pulse 1 and pulse 3 (double red arrow in Fig. 4) corresponds to twice the cable length and should already be known from the calibration. The time that elapses from pulse 1 to pulse 2 (double black arrow) represents the PD fault location, seen from the far end. The time that elapses from pulse 2 to pulse 3 (double green arrow) also represents the PD fault location, although seen from the near end of the cable. Due to the more intuitive reference point (0 meter starting at the measuring system or the high voltage feed), this seems like the most suitable approach, although the third pulse is not always available for measurement due to signal attenuation. The time that elapses from pulse 1 to pulse 2 (taking into account the plausibility of the interval from pulse 1 to pulse 3 where available) should therefore be used as standard to determine the PD fault location. Equation 3 shows the conversion to fault location s, measured from the far end of the cable.
∆
(3)
Remark: In the event that the PD is not decoupled at the end of the cable, but instead for example at a cross-bonding link box, the TDR procedure can also be used. However, due to the unusual measurement location, slightly different formulae are used here [3].
© OMICRON 2013
Page 5 of 17
3
Determining the pulse charge: Using filters To determine the IEC-compliant pulse charge of a partial discharge, the PD current signal can be integrated in the time domain. Equation 4 shows the relationship between the charge q and the timedependent current signal i(t) or voltage signal u(t).
The MPD PD measuring system provides this charge calculation option. In the "Q" tab in Expert mode, a time domain can be selected as the type of integration (see Fig. 5).
Figure 5: Time domain integration with the MPD
Figure 6: Integration in the time domain
In the time domain view of the oscilloscope, the integration boundaries t1 and t2 can be adjusted to the PD pulse plot using two independent cursors (see Fig. 6). However, integration in a frequency domain is far more common. The PD signal, which initially still has a high bandwidth, is filtered using a band-pass filter here. The filtered signal then represents the pulse charge and is displayed in the measuring system. The MPD supports frequency domain integration and allows stepless selection of the filter's center frequency. There is a predefined selection for the filter bandwidth, although this can also be set manually by the user (see Fig. 7).
Figure 7: Frequency domain integration
© OMICRON 2013
Page 6 of 17
4
Reflections: Influence on PD filters Multiple pulses in the time domain lead to spectrums tainted with resonances in the frequency domain. Figure 8 shows a typical frequency spectrum, such as recorded during a PD measurement.
Figure 8: Resonance-tainted PD spectrum
As already touched upon in Chapter 1, there is a risk here that measurements may be taken in the range of a resonance dip when using a narrow-band PD measuring filter. This leads to the PD signal being severely underestimated or even completely erased. And while moving the PD filter slightly to the left or right may seem to resolve this issue at first, the position of these resonance dips is dependent on the PD fault location (actually dependent on the double pulse time difference). Since this is not known, it is not possible to determine an optimum fixed filter position prior to performing the PD measurement. Using a wider filter (e.g. 3 MHz), as symbolized in Figure 8 by the area shaded blue, can help here. This filter now covers multiple resonance dips and peaks. The FFT signal is thereby averaged over a broader range, which means that the extremes can be smoothed out. When using a narrow bandwidth filter (e.g. 100 kHz bandwidth), no averaging process can affect the measured value. Figure 9 shows a double pulse diagram, as was recorded within the scope of the MPD type test [4] as per IEC 60270.
Figure 9: MPD600 double pulse diagram
Here, the measured charge value is applied with reference to the charges actually fed in and based on the timed double pulse interval for three different filter settings. When we look at the filter settings of the 300 kHz center frequency and 100 kHz bandwidth (pink curve), we can see that the measurement results are not affected when the interval between the two pulses is sufficiently large (e.g. 15 µs). 100 % (corresponds to 1.0 p.u.) of the pulse fed in is measured and the charge information is not distorted in any way.
© OMICRON 2013
Page 7 of 17
However, if we then reduce the impulse interval to approximately 4.75 µs, we see that with 0.75 p.u. the value displayed is significantly lower. Depending on which standard is being used, this is not permissible and must be avoided. For these filter settings, PD measurements in line with the standards can therefore be performed up to a double pulse interval of approximately 5.5 µs (recordings that overestimate the charge are generally not viewed as critical). The black curve in Figure 9 shows better double pulse behavior. At 300 kHz, a wider band-pass has been selected here. This means that PD measurements in line with the standards can be performed up to a double pulse interval of 2.5 µs, which represents an improvement over the narrower 100kHz filter. Consequently, the best double pulse resolution can be achieved when using the widest filter available, i.e. 3 MHz. In simplifying the whole procedure, we can say that the double pulse resolution time that can be achieved is determined by the reciprocal value of the filter bandwidth selected (see Equation 5).
≅
(5)
Remark: If no PD measuring filter is used, as is the case with the time domain integration of the MPD, the problem of double pulse resolution would seem to be solved. However, it is important to note that the bandwidth of this time domain integration is determined by the ultimate bandwidth of the A-D converter of 32 MHz and/or by the limits of the anti-aliasing filter necessary for correct digitalization of approximately 16 MHz. To determine the maximum double pulse resolution, a bandwidth of 16 MHz is therefore to be used here. It may also be necessary to take into account any other external components that limit the bandwidth, such as the CPL 542 impedance measuring device at approximately 6 MHz.
5
PD fault location with the MPD 600 After several theoretical introductions, the following will now focus on how to set up the MPD PD measuring system correctly. The information and instructions given are based on the Expert mode of software version 1.5.1 (or higher). Users can determine which version they are using by clicking on the "Help" menu and then clicking "About..." (see Fig. 10).
Figure 10: Determining which software version is installed
The latest released software version can be downloaded free of charge at any time by registered customers via OMICRON's homepage (www.omicron.at). We strongly recommend always using the latest software version!
© OMICRON 2013
Page 8 of 17
5.1
Switching over to EXPERT MODE ("Mode" menu)
Figure 11: Selecting the Expert mode
The EXPERT MODE allows access to all software settings, some of which are automated and not available to users in the BASIC MODE or CABLE MODE. The software can be switched over from one mode to another via the "Mode" menu (see Fig. 11).
5.2
Making the 3 MHz bandwidth available ("Advanced" menu) Due to the software upgrade from version 1.4 to 1.5, a different set of filter bandwidths is preset by default. However, the 3 MHz bandwidth (which is no longer available in this set) is the most suitable for locating PD faults. The option "Use high-bandwidth filters" makes the familiar set of filter bandwidths from version 1.4 available for use, including the important 3 MHz bandwidth.
Figure 12: Making high-bandwidth filters available
Using the highest possible bandwidth ensures that when resonances occur in the frequency spectrum, as can be the case with PD measurements on cables, no partial discharges are erased by the "negative superposition" effect, which would mean that they could no longer be measured. The high-bandwidth filter covers multiple resonance locations here, both negative and positive (see the "Introduction" and "Reflections" chapters, see also Fig. 8).
© OMICRON 2013
Page 9 of 17
5.3
Select filter settings f Center=1.6 MHz and delta f=3 MHz ("Q" tab)
Figure 13: Setting up the PD filter
To determine the charge in the frequency domain, the filter data should be set in such a way that measurements are performed with the greatest possible filter bandwidth ("delta f"). The center frequency ("fCenter ") should be selected in such a way that the integration takes place from virtually 0Hz. A value of 1.6 MHz is recommended, making the pass-band of the filter between 100 kHz and 3.1 MHz. This ensures minimum influence of the PD pulse attenuation, also for PD faults occurring further away.
5.4
Finding the calibration pulse and echo (TDR) Note: Due to the maximum detectable time range of 32 µs in the "Small Scope View", only those cable lengths that do not exceed this runtime of 32 µs can be measured in this way. In practice this means that with an assumed pulse propagation speed of 180 m/ µs the length of the cable to be tested is limited to approximately 2800 meters (twice the runtime must always be applied due to recording of the pulse echo). To optimally exploit the 32 µs time window, the "pretrigger delay" value in the "PD" tab should be reduced to 1 µs. For longer cables, statistical localization is available (see the "Statistical localization" chapter).
Figure 14a: Displaying and measuring the original and echo in an oscilloscope image
© OMICRON 2013
Page 10 of 17
Figure 14b: Displaying and measuring the original and echo in an oscilloscope image st
Set the desired calibration level (e.g. 5 pC) and show both the calibration pulse and its 1 echo from the "far" end in the "Time" display (it may be possible to see many reflections which are not relevant). If the cable length is known, the propagation speed (v, NOT v/2) must continue to be adjusted until the meter display shows TWICE the cable length (the calibration pulse must run there AND back, meaning it passes through the cable twice). Finally, the charge must be calibrated as usual in the "Q" tab by entering the calibrator value in the "QIEC (target)" field and then pressing the "Compute" button
Figure 15: Performing a charge calibration
© OMICRON 2013
Page 11 of 17
5.5
Measuring PD faults ("Measure" tab)
. Figure 16a: Measuring the PD fault
Figure 16b: Measuring the PD fault
If there is a PD fault, both the original pulse and the echoes can be measured. The relevant information for st nd locating the PD fault is the time that elapses from the 1 to the 2 pulse. This ALWAYS describes the fault location seen from the FAR END (away from the measurement), i.e. the "non-connected" far end of the cable. It is also important to take into account the fact that the measured value displayed is too high by a factor of 2 (outward and return path of the echo). The value displayed must therefore be halved. st rd The interval between the 1 pulse and the 3 pulse ALWAYS corresponds to the cable length and should already be familiar from the calibration process.
Note: The interval from pulse 2 to pulse 3 also describes the fault location, seen from the measuring position, i.e. from the connected end of the cable. In terms of a more intuitive evaluation, this is the more suitable method, as it essentially measures "from where you are sitting". However, depending on the attenuation of the PD signal, the 3rd pulse is not always available. The best approach is therefore ALWAYS to perform evaluation using pulses 1 and 2 and thereby measure the fault location "from behind".
© OMICRON 2013
Page 12 of 17
5.6
Statistical localization: PD fault location on long cables If the 32 µs time window is not sufficient for TDR location, users can switch to the MPD's statistical localization procedure. This is only limited by the "correlation depth" that can be adjusted by the user. The default value of 32 µs should always be changed here and set to the specific cable length. In the statistical method, the filtered charge signal is used instead of the high-bandwidth PD signal. The wide 3 MHz filter allows a minimum "blind range" of approximately 300 ns (1/3 MHz) here, which corresponds to the system's reaction time (double pulse resolution). PD faults that occur at a distance of less than 30 meters (at v=180m/ µs) from one or both sealing ends may therefore not be detected as two pulses.
Figure 17: Statistically measuring PD f aults
Note: It is also possible to switch to time domain integration for determining the charge to further "increase" the filter's bandwidth. Only the anti-aliasing filter is then active at 16 MHz, which causes the "blind range" to be further reduced.
The fault distances shown in the "LOC" tab must NOT then be halved: Unlike TDR measurements, this statistical algorithm has been specially set up for cable PD measurements and therefore already takes the outward and return run of the pulses into account.
© OMICRON 2013
Page 13 of 17
Note: The statistical evaluation process excels through the fact that it determines and displays ALL detected st nd rd time differences. However, it is no longer possible to assign the 1 , 2 or 3 pulse from the classic TDR image, as there are only ever 2 pulses shown in relation to one another to determine a time difference. This leads to a situation in which clusters occur in the "LOC" view, which correspond to the pulse intervals 1 after 2, 1 after 3, yet also 2 after 3 (and also all other imaginable combinations). As such, it is not possible to determine which cluster originates from which pair of pulses using this approach and therefore also from which side the fault distance determined has to be measured. With an identical PD fault location, the first cluster displayed could therefore represent the short time interval between 1 and 2 (all seen from the FAR end of the cable) OR the short time interval between 2 and 3 (thereby starting from the NEAR end of the cable). In certain circumstances, the attenuation percentage (y-axis) could potentially be of use here, as it also describes the propagation behavior of the pulses or echoes. The FIRST cluster, i.e. the shortest time difference, is the one relevant for further assessments. This describes EITHER a PD fault AT THE NEAR END, although with pulses 2 and 3. As such, the displayed PD fault location is to be viewed from the NEAR END (measurement end). OR the 1st cluster describes a PD fault AT THE FAR END, but with pulses 1 and 2, which means that it is to be viewed FROM THE FAR END (as is the case with classic TDR). If the st percentage value of the 1 cluster (=SMALLEST time interval) is lower than that of the 2nd cluster, it is fair to assume that the fault is AT THE FAR END and thereby that we are dealing with pulses 1 and 2 (with assumed overexponential attenuation, the attenuation has a stronger relative effect on small time intervals than on large intervals). The fault distance specified is therefore seen FROM THE FAR END of the cable. Cluster 2 with the higher percentage value is therefore THE SAME PD fault location, seen from the NEAR END.
6
Special aspects to be considered when using the CAL 543 online calibrator If OMICRON's CAL 543 online calibrator [5] (see Fig. 18) is used for calibration prior to performing a PD measurement or for live calibration during a PD measurement, attention must be paid to a special aspect of statistical localization.
Figure 18: CAL543 online calibrator for li ve calibration (system energized)
© OMICRON 2013
Page 14 of 17
Figure 19: Influence of the CAL543 on the LOC graphs
In addition to the clusters that occur due to reflections of the calibrator and the PD filter, exponentially sloping clusters are also to be detected starting from the time index of approximately 104 µs. These are originally generated by the double pulse inherent to the CAL 543 online calibrator system. A first positive pulse is followed at a time interval of approximately 104 µs by a second, negative charge pulse of the same magnitude. In the "LOC" view, this pair of pulses corresponds to the cluster highlighted by the red arrow in Figure 19. The time difference of approximately 104 µs and an amplitude relationship (second pulse to first pulse) of approximately -100% is detected by the measuring system here. All other equidistant and exponentially sloping clusters result from multiple reflections throughout the length of the cable being tested. In summary, we can say that while the CAL 543 online calibrator does indeed generate additional clusters (not caused by PD) in the statistical localization view, these can be reliably detected due to the initial amplitude behavior of +/- 100% and as such can be classed as irrelevant for the PD fault measurements. The only thing that cannot be ruled out is a potential overlapping of actual PD clusters that display a pulse difference of precisely this value of 104 µs.
7
Influence of non-sinusoidal test voltages Partial discharge measurements can be performed using various test voltage types (e.g. sinusoidal, cosine square wave, oscillating voltage, DC) and frequencies (operating frequency, VLF, increased frequency - e.g. with associated iron core). While for phase-resolved PD display, and also for PD application/extinction voltages, one single PD fault can generate significant differences in some cases [5] [6], this does not in principle affect PD fault localization. The only thing to be aware of here is that the test voltage source should not generate any pulse-shaped interference, which is then propagated along the cable in the same way as potential PD pulses and therefore picked up by the PD measuring system. Major disturbing pulses generate oscillations here, which may be superimposed over or overlap smaller PD signals. The best approach here is to prevent any disturbing factors from occurring in the test setup or at least reduce them on the HV source side through suitable high voltage filters (this also applies equally to resonance systems, which generate a sinusoidal test voltage pattern by covering the power loss by the block-by-block supply of energy).
© OMICRON 2013
Page 15 of 17
List of Literature [1] IEC 60270, "High-voltage test techniques - Partial discharge measurement", Version 2000, 3rd Edition [2] IEC 60885-3, "Electrical test methods for electric cables", Part 3: "Test methods for partial discharge measurements on lengths of extruded power cable" [3] Rethmeier, Balkon, Obralic, Kalkner, Plath: "Advantages and Limits of PD Localization by Time Domain Reflectography with Signal Decoupling at Joints of High Voltage Cable Systems", ISH 2007 - 15th International Symposium on High Voltage Engineering, Ljubljana, Slovenia, 27 - 31 Aug 2007 [4] Holle, Plath, Schon, Lucas: "Typprüfung eines digitalen TE-Messsystems nach IEC 60270", ETG Konferenz "Diagnostik elektrischer Betriebsmittel", Kassel, 2006 [5] Seifert, Kessler, Plath, Emanuel, Boschet: "Influence of parasitic effects on online calibration of partial discharge measurements", ISH 2009, 16 International Symposium on High Voltage Engineering, Cape Town, SA, Aug. 2009 th
[6] Pepper: "Grundlagenuntersuchung zum Teilentladungsverhalten in kunststoffisolierten Mittelspannungskabeln bei Prüfspannungen mit variabler Frequenz und Kurvenform", Dissertation TU-Berlin, 2003 [7] Rethmeier, Mohaupt, Seifert, Bergmann, Kalkner: "Sinusoidal 0.1Hz VLF test voltage for diagnostic PD measurements of high voltage cable systems", ISH 2007 - 15th International Symposium on High Voltage Engineering, Ljubljana, Slovenia, 27 - 31 Aug. 2007
© OMICRON 2013
Page 16 of 17
OMICRON is an international company serving the electrical power industry with innovative testing and diagnostic solutions. The application of OMICRON products provides users with the highest level of confidence in the condition assessment of primary and secondary equipment on their systems. Services offered in the area of consulting, commissioning, testing, diagnosis, and training make the product range complete. Customers in more than 130 countries rely on the company's ability to supply leading edge technology of excellent quality. Broad application knowledge and extraordinary customer support provided by offices in North America, Europe, South and East Asia, and the Middle East, together with a worldwide network of distributors and representatives, make the company a market leader in its sector.
Americas OMICRON electronics Corp. USA 12 Greenway Plaza, Suite 1510 Houston, TX 77046, USA Phone: +1 713 830-4660 +1 800-OMICRON Fax: +1 713 830-4661
[email protected]
Asia-Pacific OMICRON electronics Asia Limited Suite 2006, 20/F, Tower 2 The Gateway, Harbour City Kowloon, Hong Kong S.A.R. Phone: +852 2634 0377 Fax: +852 2634 0390
[email protected]
For addresses of OMICRON offices with customer service centers, regional sales offices or offices for training, consulting and commissioning please visit our web site.
Europe, Middle East, Africa OMICRON electronics GmbH Oberes Ried 1 6833 Klaus, Austria Phone: +43 5523 507-0 Fax: +43 5523 507-999
[email protected]
www.omicron.at www.omicronusa.com
