Parasitic effects caused by static frequency converters may endanger the reliable operation of drives or turbine shaft arrangements. Converter drive applications for large plants up to 500MVA can cause shaft voltages up to SOOV,, if a non optimised converter design is chosen. The influence of the most suitable solution and their influence upon electromagnetic interference (EMI) and the shaft voltages is shown and conformed by measurements in real power plants.
Parasitic effects caused by static frequency converters may endanger the reliable operation of drives or turbine shaft arrangements. Converter drive applications for large plants up to 500MVA can cause shaft voltages up to SOOV,, if a non optimised converter design is chosen. The influence of the most suitable solution and their influence upon electromagnetic interference (EMI) and the shaft voltages is shown and conformed by measurements in real power plants.
Parasitic effects caused by static frequency converters may endanger the reliable operation of drives or turbine shaft arrangements. Converter drive applications for large plants up to 500MVA can cause shaft voltages up to SOOV,, if a non optimised converter design is chosen. The influence of the most suitable solution and their influence upon electromagnetic interference (EMI) and the shaft voltages is shown and conformed by measurements in real power plants.
2004 35t h Annual IEEE Power Elecrronirs Specialists Conference Aar hm, Germany, 2004
Static frequency converters with reduced parasitic effects
Oliver Drubel Max Hobelsberger Drubel@,ieee.org ALSTOM (Switzerland) Ltd., Turbogenerator Development, Zentralstr. 40 CH-5242 Birr, 0liver.drubel~alstom.nower.com ALSTOM (Switzerland) Ltd. Customer Service Absl r acl : Parasitic effects caused by static frequency converters, like EMI, may endanger the reliable operation of drives or turbine shaft arrangements and diminish its lifetime. Converter drive applications for large plants up to 500MVA can cause shaft voltages up to SOOV,, if a non optimised converter design is chosen. This phenomenon is investigated. Design options for the converter and the plant arrangement are discussed. The influence of the most suitable solution and their influence upon electromagnetic interference (EMI) and the shaft voltages is shown and conformed by measurements in real power plants. Keywor ds: Static frequency converter, EMI, shaft voltage, power application I INTRODUCTION Recent developments in power generation equipment show an increasing growth in the installed unit power of gas turbines. Gas turbines with up to 260 MW are proven design with high reliability. Whereas steam turbines are capable to run up independently, gas turbines have to be started and driven until they turn fast enough to allow ingnition and self-sustaining operation. Ignition may not be possible until up to 80% of nominal speed. For running up the turbine it is state of the art to use the turbo-generator as motor. Electrical power is supplied to the generator-motor by a variable-speed starting device like a static frequency converter (SFC). A typical start arrangement is givenin fig. 1. This SFC has to deliver considerable amounts of power, e.g. 5-15 MW for a typical large gas turbine. The SFC is disconnected from the generator after running up the gas- turbine to ignition and self-sustained operation. The generator-motor is consequently used as generator. Usually this start-up operation with active SFC takes only a couple of minutes, e.g. 7 minutes which is only a short period of time. Because of this short period some effects and phenomena during start-up were not discovered until recently. However due to modern use of gas-turbine plants to generate peak-load turbines must be switched on and off several times per day; and in between these times of full operation there are also prolonged periods of active SFC-operation for maintenance purposes. Therefore shaft voltages during start static frequency conversion have to be cared for as well. A key component to reduce shaft voltages is the design of the SFC [3]. Therefore the SFC is within the focus of this investigation. 11OBSERVED EFFECTS AT A TYPICAL POWER PLANT Strong electromagnetic interferences occurred at sensors, video screens and distributed control equipment (DCS) at a typical 300 MVA gas-turbine plant during SFC-operation. The powerful SFC and the close proximity of control equipment to the SFC has been the main reason for these disturbances. The grounding and shielding of systems was then improved to reduce the interference to acceptable levels which appeared to be quite costly. Furthermore it was observed during maintenance works that the fuse at the NDE shaft-grounding module ("RC-module" in Fig. 1: Electrical power plant equipment for starting the gas Fig. 2) had blown. turbine 0-7803-8399-0/04/$20.00 82004 IEEE. 4365 2004 35th Annual I EEE Power Electronics Specialists Conference The grounding arrangement is shown in Fig. 2 below. The shaft has its main grounding at the drive-end (DE) of the generator. At the non-driven end (NDE) the shaft is additionally grounded by a mainly capacitive element. Therefore no low-frequency current can flow. Howwer high- frequency voltage peaks (typically caused by the SCKs I thyristors ofthe excitation system) are strongly damped [ I ] . Tvbb. - Aacken. Germany, 2004 turbine. Every 6.66 ms a huge peak appeares, followed by a much lower peak after 3.33 ms. It is evident that the rectifier at the input-side of the SFC has to be considered as the primary cause. 111 INVESTIGATION OF THE SFC GENERATED SHAFT VOLTAGE Within electrical machines shaft voltages occur on account of different effects [1-6,10]. With the application of power electronic devices, new types of shaft voltage sources occur [3-61. In general shaft voltages due to power electronics are either capacitive or inductive. After the introduction of power electronics for static excitation the voltage peaks of the rectifier bridge caused severe voltage peaks on the shaft [I ]. These kind of peaks occurred during normal operation. The voltages were coupled by the capacitance between the rotor-field winding and the shaft. Therefore the shaft current is different at the drive end in comparison with the non-drive end. Reichert, Amman, Posedel and J oho designed a method to suppress the shaft voltages in 1988. In order to avoid a direct coupling of the shaft at the non-drive end (NDE) they coupled the shaft capacitively to ground, see [I ]. Some other publications related to experience with converter fed asynchronous machines [4,6] an inductively coupled the shaft voltage. Here the currents at drive end and non-drive end must be the same. Synchronized measurements at the DE-side and the NDE-side showed that a positive pulse at the NDE-side coincides with a negative pulse at the DE-side and vice versa. So the picture of a closed current loop arises with the current being induced by changing magnetic fields in the stator core. This hypothesis is supported by the observation that the pulses at NDE have low- impedant sources. Current strengths of up to 70A were measured which could not be explained by capacitive coupling. These kinds of shaft voltages are clearly inductively caused with the SFC as main root cause. The speed independent time period of 3.33ms with strong pulses every 6.66 ms reveals the strong relation to the SFC's rectifier. A typical electrical design of a converter fed plant drive is given in Fig.4. Fig. 4 reveals a strong asymmetry concerning the arrangement of the main choke between the rectifier and converter. According to this model current measurement coils (Rogovsky coils) and voltage sensors have been installed at a multitude of significant locations in the real equipment and corresponding measurements were done. ". SMIGxmdm WGnumm WGxmdm nom DEM& R C M C" B m CVBRUd Fig 2: Grounding scheme of shaft train with KC-module Measurements determined voltage peaks with abnormally high amplitudes, fig.3, during active SFC-operation to be the cause ofthe blown fuse. 200 speed n i l 89 rpm 150 100 50 "I - 3 0 0 > 5 -50 -100 -150 -200 -c v) 0.01 0.02 0.03 -25OF; " ' I ' I ' " I ' ' I ' I ' Time (s) Fig. 3: Shaft voltage peaks at NDE during turbine run up The shown voltage pattern is in its timely distribution of peaks independent of the turning speed of the running-up m 2004 35th Annul IEEE Power Electronics Specialists Conference Anchen. Germany. 2W4 The static frequency converter consists of two thyristor bridges. It is a current-type converter: Between both bridges a D.C. circuit exists with one choke only in the positive branch. The negative branch connects both bridges directly. The bridge at the grid side is used to adjust the voltage level of the D.C. circuit (and hence the DC current) and works asl a rectifier. During one period of 20ms each of the 6 thyristors is switched on and off. Three of them are connected to the positive branch (link) with the choke, the other three are directly connected to the output -rectifier bridge. The bridge at the machine side works as a converter. At lower speeds it operates in block mode, afterwards it works as LCI converter. 1. It was determined that the start-up transformer could not be the source of the pulses. A clear relationship between the shaft voltages and the transformer currents should exist in case of the transformer as voltage source. Figure 5 shows the summation of all three transformer phase currents. Shafi vol aee at non d& end I50 IO0 SO 0 -50 -100 -150 Fig. 5: Relationship between the summation current of all transformer phases and the shaft voltage. No relationship between the shaft voltage and the summation current can be found in Fig. 5 . 2. Whereas the transformer as source of the new shaft voltage phenomenon could be excluded, the SFC seems to be the real source. It is obvious from Fig. 6 that a strong common-mode current (summation of all phase currents) flows into the SFC and out of it again without alteration. The conclusion is that the current loop must be closed via the generator, ground, the shielding of the SFC supply-cables and their shielding capacitances. There are no other connections left. The similarity between curve shapes of common mode current and shaft voltage is also striking. 100 0 -100 Sh&vol ege IMP 4456) -2001 fB -50 t v I I I I J I 0.0194 0.01% 0.0198 Time t (s) Fig. 6 : Current into the SFC and out of the SFC 3. In [ 6] it was shown how the shaft voltage is induced by a circulating flux flow through the generator core. The circulating flux flow can only occur in case of a difference between the current in the positive and negative axial direction. Therefore the current of the stator terminals and the neutrals is measured, Fig. 7. -200 arrent at Etdmr neutral wp 4440) -240 b -280 -300 (MP4436) -3201 - 340b, I , , , , I , , , , ~, , , I , , , , I , 0.01976 0.01978 0.0198 0.01982 0.01984 Time t (s) Fig. 7: Current into one generator terminal and out of the generator neutral connection Indeed figure 7 reveals a phase difference of about I0-15p between the current at the phase-terminal and at the neutral connection. The time constants of the shaft voltage-pulses and the difference current between the stator terminal-current and neutral-current are fairly similar. It will cause a circular magnetic flux in the stator core and hence induce the shaft- voltage pulses. In the following more emphasis is given to voltages measured in or at the SFC during these switching processes. All these voltage-surges and oscillations contribute to overall EM1 and also to increased stress of components. It can be seen in Fig. 8 that during valve-change-over of the corresponding conducting thyristors of the rectifier-bridge very strong current-oscillations appear in the two connected cables and their shields. 4367 2004 35th Annual IEEE Power Elecrrunics Specialisrs Conference Aachen, Germany, 2004 0.021 I 0.0212 o' 021 Time 1 (s) Fig. 8: Current in the grounding-connections of the transformer - SFC cable a) Phase U, b) Phase V, c) Phase W These currents flow in differential mode and do not influence the shaft voltage. They produce however strong EM1 due to the relatively high frequencies involved. Fig. 9 and Fig. 10 show the considerable common-mode voltage pulses and steep voltage slopes to which the whole SFC-system and also the generator winding is subjected. The common mode voltage is given by the summation of the three terminal voltages, F 0 5 10 I S 20 25 Tirnr(rnr) Fig. 10 Voltages to ground at the SFC input terminals r j - 2 - F . . ' ' ' ' ' ' ' ' ' ' ' ' . . 1 1 . ' ' ' ' ' 0 5 10 I5 20 25 ~~ Timc(rnr) .~ Fig. 1 I Voltages at the links inside the SFC The negative-link potential voltage (curve 3) is pushed down very steeply to low values. At the same time the potentials at the SFC-inputs are pushed up. However, when the positive- link thyristors are switching almost no effect on the shaft- voltage can be observed. This is due to the "protecting" effect of the inductor in the positive link. In Fig. 11 it is clearly visible that the shaft-voltage pulses coincide with every switching of the rectifier-thyristors connected to the negative link. Fig. 12 gives a magnified view of the processes. 2.5 sh#' *a.mYwh-t m .nm*ndsma-' w , W . b 8 r n W d r n ,A.--- -.-. -. --._ _ _ _ . _ _ wYp.*n..-. -d.r .V* , >-0.5 0 0.W5 0.01 0.015 0.02 0.025 0.1 0.035 0.04 0045 0.05 Time (I) - 1. 5 --I_ Fig. 9: Common mode voltage at SFC-input terminals -2.5 6.4 6.5 6.6 6.1 6.8 6.9 I 7.1 7 ~ Time ,"lo Fig. 12 Voltages of links inside the SFC The potential of the negative link starts to slope steeply downwards. The potential of the voltage between the choke and generator follows and also the shaft voltage follows. This is again a clear indication that switching of the rectifier 4368 2004 35th Annual IEEE Power Electronics Specialists Conference thyristors connected to the negative link cause the pulses on the shaft voltage. IV IMPROVED SFC DESIGN The influence of the SFC design is one of the dominant factors for the parasitic effects, which can he seen at the shaft. The direct connection of rectifier thyristors to the negative link and with it to the generator connections has been identified above as main cause for high frequency pulses. This configuration can be used as long as the shaft voltages stay within appropriate limits [1,7-91. If these limits are exceeded the SFC needs to be modified. One very effective modification is to use a split choke within the d.c. circuit of the SFC. A split or preferably symmetric inductor will prevent steep gradients and HF-pulses from travelling into the generator. The choke will significantly reduce the voltage gradients to which the generator winding is subjected and keep i t at acceptable values. This design ensures a very smooth operation. It is also an economically acceptable solution because no large additional components are needed in general: The main inductor is split into two halfs, which is often quite feasible. No big changes are caused electrically to the SFC by this change: The internal DC- current path and its control is still dominated by the same inductances. The beneficial influence of the choke on the high-frequency generation ofthe SFC can he seenin Fig. 13. The shaft voltage during a commutation at the positive link of the SFC is only a third or even less than during a commutation at the negative link. The main reason for this is the strong HF-damping by the choke, see fig. 13 below. In Fig. 13 it is clearly visible that the HF-content of the voltage after the choke is almost negligible. Aachen, Germany, 2004 ShaRvoltage(MP4456) 6000 MOO 4000 d.c. link voltage befors &e choke, tranrformrside (MP 4455) d.c. linkvoltage af er &e choke, (MP4454) -,,,E -2oooE Y U 0.0176 0.0178 0.018 -4WE -30;j174 Time t (s) Fig. 13 Voltages before and after the choke, positive link Beside the influence of the SFC (steepness of voltagelcurrent changes), other parameters like the cable length between the start up transformer and the SFC as well as between the SFC end the generator have to be taken into account. Generally, modifications to converters within plants of some IOOMVA output have to be based on a deep understanding of the phenomenon involved. Therefore design modifications to the static frequency converter (SFC) must take into account all main components which are involved. Modifications need firstly to he simulated by using simplified mathematical models of the whole plant with all the main par& involved. The effectiveness has to he shown in simulations. The proposed design improvement is a result of these simulations. VI SUCCESSFULLY IMPLEMENTED MODIFICATION The influence of the modifications could be proven in the meantime at several power plants which showed high shaft- voltage pulses. All the plants had long cables between start- up transformer and SFC resulting in high coupling capacitance to ground at the inputs of the SFC. The main modification has been the implementation of a symmetrical choke in both d.c. links of the SFC. The influence of this change within the frequency converter design is shown in fig. 15. 4369 2004 35111Annual I EEE Power Elecrronics Specialists Conference Aachen, German),, 2004 Timet @s) Fig. 15: Measured shaft voltages with and without symmetrical choke The modification has been a complete success. The high- frequency components (-5OkHz) of the shaft-voltage pulse have almost completely disappeared. They are efkctively blocked by the inductors in both links. There is no low- impedance ground-loop anymore for HF-bursts caused by the SFC. Only a signal of considerable lower frequency (-10 kHz) and much lower amplitude is left. The voltage peaks have been reduced to a third of the original value. High frequency signals causing EM1 have almost been eliminated by this relatively simple modification. Furthermore it became clear that the source-impedance of the pulses could be considerably increased by the additional choke. That means that the capacitive grounding module at the NDE-side of the generator can reduce the voltage pulses even further without being subjected to increased electricaVthermal stress due to high HF-currents. It was possible to reduce the level of shaft voltages even below the expected level. V CONCLUSIONS Measurements in a 300MVA power plant allowed to identify the mechanism which causes the observed HF shaft-voltage pulses at converter-fed large electrical machines. The root causes for these pulses were a SFC which produces common- mode HF-voltage pulses together with a mainly capacitively conducting electrical loop which couples thc pulses inductively onto the shaft. Based on the improved understanding of the phenomenon, modifications have been implemented in a power plant which were able to reduce the shaft voltages by a factor of 3.6. The key factor for the successful voltage reduction has been a design-change within the converter. Instead of using a single choke in the internal DC-loop in an asymmetric configuration, two chokes are arranged in a symmetrical way. The two chokes effectively block HF-pulses and reduce EM1 considerably. VI REFERENCE [ I ] Amann C., Reichert K., J oho R., Possedel Z.: SHAFT VOLTAGES IN GENERATORS WITH STATIC EXCITATION SYSTEMS - PROBLEMS AND SOLUTI ON, IEEE Trans. Energy Conversion, vol. 3, no. 2, june 1988. [2] Torlay I. E., Corenwinder C., Audoli A., Herigault J., Foggia A.: Analysis of Shaft Voltages in Large Synchronous Generators, International Electric Machines and Drives Conference, 9-12 may 1999, Seattle, Washington, US. [3] Link P. J .: Minimizing Electric Bearing Currents in ASD Systems, IEEE Industry Application Magazine, J ulyiAugust 1999, pp. 55 - 66. [4] Miitze A.: Bearing Currents in Inverter-Fed AC-Motors, Dissertation University of Darmstadt, 23.1.2004. [5] Cheng S. , Lip0 T., Fitzgerald D.: Modeling of Motor Bearing Currents in PWM Inverter Drives, IEEE Trans. on Industry Applications, vol. 32, no. 2, march- April 1993 [6]Hausberg V., Seinsch H. 0.: Kapazitive Lagerspannungen und Strome bei umrichtergespeisten Induktionsmaschinen, Electrical Engineering (82). pp, 153-162, 2000. [7] Amman C. U.:Wellenspannungen in grossen, statisch erregten Turbogeneratoren, Dissertation ETH-Zurich, 1988. [8] VDE 0141/5.76: ,,Bestimmungen f~ Erdungen in Wechselstromanlagen f& Nennspannung, VDE Verlag [9] AIEE Committee Report: ,,Voltage gradients through the ground under fault condition, Trans. AIEE Part 111, (1958) pp. 669-692. [IO] Hausberg V., Seinsch H. 0.: Kapazitive Lagerspannungen und Strljme bei umrichtergespeisten Induktionsmaschinen, Electrical Engineering (82). pp. 153-162, 2000. 4370