iT la Cee ae idely used n areas of high standards Soe meat Peace cs Rr ee eee eee cn) Pan) ag ts) Tes en the risk of un- ee eeu) Methods for performing the induced eee aca aT ey eee ae eaten Ruse etd Beet cen oc cn Sy KEYWORDS Cee Seay voltage testing, partial di presto harge co Diagnostic testin of cast-resin of transformers ‘The development of cast-resin transformers was initiated many decades ago because of the flammable nature of ‘mineral oil and the risk of oil spillage with, the related contamination of the ground. Since then transformer technologies have been well developed and are now widely used in applications where safety is of major importance, for example in buildings, in industrial applications, or for ecologically sensitive areas, such as ground water protection areas, The powerand the nominal voltage of cast-esin transformers were increased. cast resin im the last few years. Nowadays, cast resin transformers are built up to 72.5kV and more than 60 MVA. They can also be equipped with on-load tap changers (LTO), Although cast-resin transformers are called “maintenance free, periodic diagnostic measurements are requested often — in particular, when outages would cause high follow-up costs. Some insurance providers offer cheaper tariffs when the risk of outages is minimized by periodic diagnostic measurements. For cast-resin transformers the diagnostic tools are limited. During manufacturing TRANSFORMERS MAGAZINE | Volume 5, Issue 2ng routine tests are performed according to TEC 60076-11 (1) + Tums ratio measurement + Winding DC resistance measurement + Measurement of voltage ratio and phase displacement + Measurement of short-circuit impedance Load and no-load measurement Applied voltage test Induced AC withstand voltage test Partial discharge measurement (Un >3.6 KV, acceptance level 10 pC) Figure 1. CPC 100 with matching transformers www.transformers-magazine.com In order to determine the transformer condition directly on site, only a limited amount of test procedures can be used Especially when it comes to assessing the insulation condition, common test methods which are typically performed ‘on oil-filled transformers are not suitable For instance, dissolved gasin oil (DGA) analysis on cast-resin transformers is not applicable. Vice versa, the power/ dissipation factor measurement on cast resin transformers is highly affected by the ambient conditions. Therefore, partial discharge (PD) measurements yield the most valuable information on theinsulation condition. On-siteinduced voltage test on cast-resin transformers, combined with PD measurements, is described in the following section 2. Partial discharge Measurement For partial discharge (PD) measurements the frequency of the test voltage should be higher than the nominal frequency to avoid excessive excitation current during the test, ‘The most common reason for failures of cast-resin transformers is the electric breakdown of the cast-resin insulation between turns or parts of the windings. In many cases, partial discharges occur before the breakdown happens. By testing the windings with induced voltage in combination with a sensitive PD measurement, the risk of unforeseen breakdowns can be minimized, TSN sean For on-site PD tests on cast-resin transformers, a voltage source with adequate power is needed. ‘The frequency ofthe outputvoltage should be higher than. the nominal frequency of the transformer to avoid excessive magnetizing currents. In order to minimize the required power, the source should deliver an output voltage with variable frequency. The main inductance of the magnetic core, the capacitance of the high voltage winding and the coupling capacitor build a parallel resonance circuit. The frequency ofthe test ‘voltage should be tuned to the resonance frequency of this parallel resonance circuit to reduce the needed power to the lowest level If the resonance frequency is too high, an additional capacitor can be connected to the low voltage winding to reduce the frequency down to the upper limit of 500 Hz, according to TEC 60060-3 [2]. Figure 1 shows the setup with the CPC 100, a compact voltage source with a weight of 30 kg/ 66 Ibs ‘he built-in amplifier can deliver up to SKVA at frequencies from 15 to 400 Hy. Tt has a “Soft switching” design to avoid disturbances by the switching semi- conductors. With an additional matching transformer, the output vollage of the amplifier can be matched to the required excitation vollage of the transformers low voltage side. This universal matching transformer can deliver voltages. from 50 V up to 400 V in steps of 50 V (at frequencies of 100 Hz or higher) and up (o 800 V (at frequencies of 200 Hz or higher). During the factory acceptance The most common cause of cast-resin turns or parts of the windings See iz 87DIAGNOSTICS Field experience has shown that voltages up to 130 % of the rated voltage are a good compromise to find partial discharge faults but not stress the coil too much er tegiee) eo atsi2} *» ros ° ° Figure 2. Impedance ofthe test circuit vs. frequency test (FAT), three phase generators re used as voltage source. To check the integrity of the coils onsite, tis sulicient to test each phase individually using single phase ‘excitation, In the worst-case scenario a voltage rise above the rated voltage could lead to PD activity which remains active ‘even afier returning to nominal operation voltage. Exceeding the nominal voltage during testing helps to identify such PD activity and to classify its severity According to [1], during the FAT a pre- siress voltage is applied for 30 seconds Afterwards, the test voltage is applied for another three minutes. On site, either these test cycle durations can be used oran Matching ‘Transformer Matching ‘Transformer Matching “Transformer Figure 3, Test system for 15 KVA with three CPC 100 88 adapted cyclecan be agreed upon. In order to avoid unwanted breakdowns of older ransformers, high-voltage ests. should only be performed with lower voltages than those used on new transformers in the factory: Field experience has shown that voltages up to 130 % of the rated ‘voltage are a good compromise to find PD faults but not stress the coil too much. Power transformers up to 3 MVA can be tested using a test power of 5 KVA, With the CPC 100, the impedance of the whole test circuit can be measured in amplitude and phase angle. Fig. 2. If the frequency is set to the maximum impedance (red curve), the required test power is minimized If the required power to energize the transformer exceeds the power output of one CPC 100, up to three devices can operate in parallel, Fig. 3. One CPC 100 works as a master; the other two work in a synchronous mode as slaves. To- gether they can deliver 15 kVA on three 230 V 1/16 A plugs. With this portable test system, cast-tesin transformers up to 25 MVA can be tested directly on site 3, Electromagnetic interference from the surroundings PD measurements outside of Faraday cages are troublesome due to electro- Transformer under Test TRANSFORMERS MAGAZINE | Volume 5, Issue 2Frequency 2 Pulse in 3CFRD after —"® addition of phasors Frequency 1 Frequency 3 Figute 4. Single PD impuise entered inthe 3CFRO diagram When higher power is required, up to three CPC 100 devices can operate in parallel, allowing on-site PD testing of transformers up to 25 MVA ‘magnetic interference from the surround ings. Modern digital PD instruments provide special filtering methods which can reduce or even eliminate such noise problems, One possibility to separate PD signals from ambient noise sources is to use the BPARD or 3CFRD method (3). Both Dar De Se pri’ beth baa ji | beixdiidis | Peake open te methods make use of a so-called star diagram which represents the relations between three measured signals. ‘The 3PARD method uses three units to si ‘ultaneously pick up PD signals, while for the 3CFRD method, the digital PD instrument can measure PD impulses si multancousy at three different frequen. cies. For each frequency, a PD impulse is evaluated. Afterwards, the discharge lev dls obtained at three frequencies, using a single instrument, are used to mark a corresponding point in a star diagram, Fig 4 Assuming that different PD and noise sources differ in their frequency spectrum, each source will show in a different cluster in the star diagram. By tuning the three frequency’ settings, an optimal separation of the individual clusters can be achieved. As a last step. each luster is evaluated separately without interference from the remaining sources. This way a reliable evaluation of all PD sources can be achieved on site, even while strong disturbances are present, Fig 5. 4. Case studies 4.1 Case 1 ‘A PD measurement was performed on a 33 MVA cast-resin transformer with a rated voltage of 30 kV. Figure 6.1 shows the “Phase resolved PD” (PRPD) diagram at 40 KV. The noise is about 8 pC. The analysis with the 3CFRD is shown in Figure 6.2. The filtered signal, in Figure 6.3, shows a clear pattern of internal void discharges [4] with 6 pC, although the PDs are below noise level. ‘The extinction voltage was above 36 KV, which is 20 % higher than the rated volt- age. As a conclusion, during normal service no PD sources are active inside the transformer. Fitered interference signals ‘Separation through 3PARD or 3CFRD representation ‘Separated PO activity Figure 5, Separation through 3PARD or 3CFRD representation www.transtormers-magazine.com 9DIAGNOSTICS Figure 6. (1) PRPD without 3CFRD filtering with interference; (2) star diagram with noise (op cluster) and PD cluster (marked); (3) PD pattem of voids with 3CFRD filtering Figure 7. Breakdown fault on cast-resin transformer 42 Case2 After a single-phase failure of 2.5 MVA. cast-resin transformer, Fig. 7, it was in- vestigated whether the remaining two windings had also been damaged, A partial discharge measurement on both, outer windings was undertaken with 130 % of the rated voltage to prove that these windings were free of PD. No PD. could be detected above the noise level of 15 pC. It was then decided to replace the faulty winding and to put the trans- former back into service Figure 8 shows the PD_ measurement results of phase B after installing the new winding on the transformer. During the on-site PD measurement interference signals of 17 pC were measured. The only visible cluster was related to the ambient noise. In this case, the so- called inverse 3CFRD was used, The only visible cluster was selected and filtered out, Fig. 82 Instead of showing the impulses related to this cluster all Figure 8 (1) Unitered PO measurement; (2) 3CFRD cluster selection; (3) PRD with inverse SCFRD fering 90 TRANSFORMERS MAGAZINE | Volume 5, Issue 2remaining pulses outside this cluster ae shown. Figure 8.3 shows the inverse principle of 3CFRD filtering while the noise cluster was selected. Itcan be seen that no other impulses are visible, which leads to the conclusion that the new installed coil has no measurable PD. ‘The inverse 3CFRD filtering should be performed carefully in order to avoid filtering out any PD signals 43 Case 3 In an industrial 20 KV grid, all cables are equipped with HF PD couplers at the terminations to make biannual routine monitoring measurements of the whole grid. including the cables and the transformers, During 2 routine ‘monitoring test. partial discharges were detected on one ofthe cable terminations. ‘To find out ifthe PDs are in the cable or in the connected transformer, an off-line PD measurement was catried through. ‘The cable did not show any PD, but the transformer did. Figure 9 shows the setup of a PD measurement on the transformer under test. On all. phases PDs were detected and showed typical patterns of internal voids with discharge levels up to 101 pC, Fig 10. "The inception voltage was found tobe 1.0xUs, while the extinction voltage was below nominal voltage (0.6* Us) on all three phases. ‘This means that the transformer has continuous partial discharges during ‘operation and should be replaced. As a consequence, the operator decided to replace the transformer completely 44 Case 4 PD measurements were done ona 30 KV / 6 KV 1.17 MVA cast-resin transformer, Figure 11 shows the applied test setup In order to energize the 6 kV side of the transformer, an additional step-up trans- former was needed, The required 10 kW power to create excitation of one phase of both transformers was delivered by three synchronized CPC 100s, The test volt- age of 1.3xU, was applied for 60 seconds as agreed upon on site. Figure 12 shows the PD measurement result of phase B, which shows no PD impulses above the noise level of 8 pC Conclusions On-site induced voltage testing in com- bination with partial discharge measure www.transformers-magazine.com ane o Figute 10, PO measurement results Modern digital PD nals from ambient n struments separate @ sources using 3PARD or 3CFRD methods ments is a powerful tool to assess the insulation condition of cast-resin trans formers. In order to conduct the measurement, a voltage source with variable frequen: cy is needed, Depending on the power consumption, up to three OMICRON CPC 100 units can be synchronized. By doing so, the output power can be in- creased up to 15 kVA. Depending on the transformer design and the required 1DIAGNOSTICS The 3CFRD method evaluates discharge levels obtained with three different fre- quency settings and uses star diagram to identify clusters which separate PD and noise sources test voltage, power transformers up to 25 MVA can be measured using this syn- chronization functionality. Nevertheless, itis stlla flexible and portable solution for ‘on-site testing, asall required components ‘weigh less than 30 kg/ 66 Ibs each, Usually, different types of noise sources arepresent during on-site measurements ‘Thus, the digital PD instrument offers several ways to cope with high disturbances. One proven. tool for filtering noise and separating different PD sources is the so-called 3CFRD ‘method. Evaluating the discharge levels, obtained with three different frequency settings, the influence of noise sources ‘ean be minimized and each PD source «can be evaluated individually. Bibliography 1] IEC 60076-11, Power transformer Part 11: Dry-type transformers, national Electrotechnical Commission, Geneva, 2004 2] IEC 60060-3, High-voltage test tech- niques, Part 3 Definitions and require ments for on-site testing, International Electrotechnical Commission, Geneva. 2006 3] K.Rethmeier, A. Obralic, A. Kraetge, M. Krdger, W. Kalkner, R. Plath, Improved Noise Suppression by real time pulse-waveform analysis of PD pulses and pulse-shaped disturbances, International symposium on high volt- age engineering (ISI), Cape Town, 2009 iA] Cigré 676, Partial Discharges in Transformers, Working Group D1.29, Cigte 2017 92 C foes cat Eri i e N step-up TR Peis Figure 12. PD measurement results on phase B Authors ‘Michael Kriiger is principal engineer with OMICRON clectronics GmbH in Klaus, Austria. He studied electrical engineering at the Technical University of Aachen (RWTH) and the University of Kaiserslautern (Germany) and graduated in 1976 (Dipl.-Ing }.In 1990 he received the Dr. (PHD) degree from the Technical University of Vienna ) | Michael Krager has more than 35 years of experience in high voltage engineering and insulation diagnosis on GIS, instrament transformers, cables, power transformers and rotating machines He has published many papers about electrical measurements on different assets and holds 15 patents. He is a member of VDE, Cigre and IEEE and participates in several working groups for OEVE, IEC and Christoph Engelen holds an MSc degree in electrical engineering from the RWTH Aachen University in Germany, After he graduated in 2013, he started his professional career as an application engineer at OMICRON electronics where he focused on transformer diagnosis Since 2017 hes part of the product management and specializes in the business development of cast-resin transformer testing TRANSFORMERS MAGAZINE | Volume 5, Issue 2