A Model Based Approach To Optimize Controls Of A Large LCI .
A model based approach to optimize controls of a large LCI VSDfor minimal grid-side sub-synchronous torsional interactionCopyright Material PCIC EuropePaper No. PCIC Europe (IS-32)Pieder JörgABB MV DrivesAustrasseSwitzerlandAdrian TreschABB MV DrivesAustrasseSwitzerlandAbstract - Local power generation units in an industrialgrid have a certain risk for interacting with powerelectronics loads (drives, VAR compensators, gridinterties), especially if these are of comparable power.Interaction is most prone to happen at the lowest torsionalnatural frequencies of the power generation equipment.The paper reports, how the level of such sub-synchronoustorsional interaction (SSTI) is affected by the dynamics ofthe controls of a large VSD.Unlike other factors, namely the mechanical properties ofthe generator and the grid impedances, the dampingeffect of the VSD's control system can be adjusted tosome extend via parameterization. A linear model basedapproach is reported and demonstrated with a 20MWdrive installation, which allows optimizing subsynchronous torsional damping while maintaining thenecessary dynamic response to load- and referencechanges in the VSD system. In the proposed approach,only the critical frequencies of the generator and thedriven equipment have to be known, while grid data andinfluence of other loads needs is segregated from theoptimization.Index Terms — Variable Speed Drives, LCI, SubSynchronous Torsional Interaction, Interharmonics,Electro-Mechanical Interaction, Closed-Loop Stability.I.INTRODUCTIONTwo very similar phenomena have been discussed inliterature separately, mainly because they have beenobserved in different fields of application. On one hand,when planning for variable speed drive (VSD) poweredturbo machinery, the chemical, oil and gas industry knowsit has to deal with the excitation of rotating mechanicalsystems (loads) by harmonic and inter-harmonic torquepulsations, which are created as “by-product” of theelectric frequency conversion.On the other hand, when high voltage DC (HVDC)power transmission is installed close to power generation,similar electro-mechanical interactions are observedbetween the rotating mechanical system of the generatorand the adjacent HVDC inverter substation. Thephenomenon is called sub-synchronous torsionalinteraction (SSTI) in power transmission and distribution,as instability normally shows at the lowest (and thus subsynchronous) torsional natural frequency of the generator.The power electronics circuit of an LCI-type (loadcommutated inverter) VSD and the power conversionfrom AC to DC and back to AC of a classic HVDC link arebased on the same topology, the simplest representativeof which is shown in Fig. 1.Electronic ISBN: 978-3-9524059-1-8Martin BruhaABB MV DrivesAustrasseSwitzerlandFig. 1: Simplest circuits representing an LCI based VSD(above) and a classic HVDC link with local powergeneration (below), showing their similarity.The similarity is obvious and so it is for the describedelectro-mechanical interaction: torsional interaction withdriven rotating equipment in case of VSD and torsionalinteraction with the generating rotating equipment.However the approach in engineering is historically verydifferent. For VSDs, pulsating torques are calculatedanalytically for the feed-forward controlled converter/motorsystem. This torque is applied for torsional analysispurposes to the respective motor mass in a mechanicaltrain model [1].For HVDC installations, a damping contribution pertorsional natural frequency of the feed-forward controlledinverter is analytically derived and then compared to thenatural mechanical and electrical damping of thegenerators and the local loads to grid [2]The importance of closing the control loop, with thespeed resp. motor position feedback for quantification ofharmonic and inter-harmonic impact, has only recentlybeen highlighted for electric motor drives [3-6] , while forHVDC, some control contribution was factored into thedamping coefficients, however without making themdependent on input, e.g. frequency dependent gridimpedance behavior.This paper deals with a field case where subsynchronous torsional interaction occurred betweenpower generation equipment and a large VSD, partly incombination (and supported) by torsional interaction withthe driven equipment, in this case a complex turbocompressor train.II.A.FIELD CASE DESCRIPTIONSystem DescriptionThe system under study is a large LNG productionfacility with local power generation and a predominantpower electronic load, given by 3 large 20MW LCI drives
(Fig. 2) and power generation consisting of up to 4 gasturbine generators (GTGs) of about 40MVA each, withroughly 180MVA short-circuit capacity (SCC) per GTGunit. Particular is the presence of local power generationequipment close to the VSD, as the generator and itsprime-mover (here the gas-turbines) are equally sensitiveequipment from a torsional standpoint. There is noconnection to another supply grid.B.SSTI Risk AssessmentThe interaction between an HVDC converter and amachine is assessed by means of the unit interactionfactor (UIF), a method presently not used outside thepower generation and transmission community. Thefactor is evaluated for one VSD at the time. For this VSD,one-by-one, each power system configuration isconsidered, while always one unit (“the last unit to comeonline”) is checked for interaction risk with that VSD. If theUIFi of one VSD against the i:th generation unit, with i-1other generation units running already, is higher than 0.1,then there will be a risk of SSTI [7]. These assessmentcalculations are always made prior to an HVDC project ina so called screening study . If this study shows that thereis no risk of adverse interaction, no further SSTI detailedstudy will be made.Due to the similarity of power electronics topology, therisk for sub-synchronous torsional interaction betweenVSD and power generation, or even between compressortrain and power generation (through the VSD) is proposedto be evaluated by using the same factor. The factor forone VSD and the i:th unit coming online is defined asfollows:el. loadsUIFi where:UIFiMWVSDMVAiSCiSCtotFig. 2: The system under study, with the 3 largecompressor drives as the dominant load and the 4 gasturbine driven generators.The system has various critical frequencies, as all thethree compressor trains and all the four generators dohave relatively weakly damped low frequency torsionalmodes. The dominant modal frequencies are given inTable ITABLE IDominant torsional natural frequencies of the plantEquipment1st TNF2nd TNFDesignationGTG (4 identical)9.5Hz29.7Hz1st compressor9.5Hz18.6Hznd9.83Hz23.46Hz2 compressorrd3 compressor14.9Hz21.8HzIt can be seen, that the frequencies are very close toeach other. A particularity of the plant is the coincidenceof the first mode of compressor train #1 with the first modeof the GTGs.MWVSDMVAiæSC i öçç1 è SC tot ø2Unit interaction factor of i:th generation unitMW rating of the VSDMVA rating of the i:th generator machineShort circuit capacity at the VSD supply busexcluding the i:th generation unitShort circuit capacity at the VSD supply busincluding the i:th generation unitAfter the experience with the here presented system,we recommend evaluating the UIF per train for eachproject, as if the train would run solo i.e. MWVSD equalsthe power of one unit. This is typically only the caseduring the commissioning, but also this operation needsto be ensured.The unit interaction factors (UIF) for one VSD running isevaluated and given in Table II for different number ofGTGs in operation.TABLE IIUIF per individual GTG and 20MVA drive load#GTG in operationUIF of one GTG unit20.125 (à SSTI risk)30.05640.031For the case where 2 GTGs are running, the qualitativeevaluation shows a risk for sub-synchronous torsionalinteraction, a risk which materialized in field during thecommissioning phase of the plant, when the respectiveoperating conditions were met.The evaluation of the UIF is clearly only an indicator forSSTI-risk. In practice we have observed SSTI also for theconfiguration of 3 GTG running with 1 VSD as load, andwith very little additional loads (typical for commissioningphase).The numbers that go into the UIF also clearly neglectthe fact, that the VSD is an active load, i.e. that the control
circuits with their settings for controlling e.g. thyristorfiring, intermediate DC current or electric motor speed doaffect the torsional stability for equipment connected toeither side of the converter system.Measured Sub-Synchronous Torsional InteractionDuring one of the first start-up attempts of the train #2,we found the DC link current of the VSD and, asconsequence, the air-gap torques of both load motor andgenerators, to behave as shown in the waterfall diagramin Fig. 3. The diagram shows the frequency content of theDC-link current over time. The DC-link current is a verygood representative of air-gap torque. From left to right,the compressor train accelerates from 1200rpm (20Hzelectrical) to 3600rpm (60Hz). The acceleration ramp lastsabout 500s, which is shown on the x-axis.The spectra are plotted in logarithmic scale, in order toreveal all the signal content The diagonal lines areinterharmonic torque pulsations of the LCI. They arechanging as a function of speed. The horizontal lines arespeed independent pulsations, such as the torsionalnatural frequencies or integer harmonics of the gridfrequency. The interharmonics of the LCI go to zero atseveral speeds, e.g. for 12* line frequency - 12*motorfrequency this can be seen around 420s, where a motorfrequency of 50Hz (3000rpm) is eling of the LCI current control loopIn search for a simplified model, which is good enoughto capture the effects encountered during commissioningof the plant, it was decided to focus on the current controlloop and to think of the voltage fluctuations due totorsional oscillations on generator and motor asdisturbances to that circuit.Fig. 4 shows a block diagram from which we will derivethe transfer functions of the current control loop. Bothinverters are producing a DC-voltage out of theirrespective AC voltages. The motor side AC voltage, andthus also the DC voltage, are by principle controlled sothat the voltage is proportional to the motor statorfrequency. Theoretically, the rectifier would always berunning at the same operating point, which is only slowlyadapted in practice for system performance optimization.The line side AC voltage is turned into a variable DCvoltage which has to drive the desired current through theDC-link inductor against the rectified motor voltage.AC voltage deviations from pure fundamental voltage,e.g. due to harmonic distortion by other equipment (linesie) or induced by torsional oscillations of the electricmotor (load side), are entering into the DC link throughthe respective inverter: Frequencies other than theinverter operating frequency are passing unchanged intothe DC voltages (Udc1 resp. Udc2 in Fig. 4) and aredriving corresponding AC currents of the same frequencyin the DC-link inductor. This can for example be seen inthe spectra in Fig. 3, where the excitation of the torsionalnatural frequencies is becoming visible at the identicalfrequencies in the DC-link.-50402. TNF-60-7030LoadC.III. MODEL BASED APPROACH-801. 450500Fig. 3: Waterfall diagram of VSD DC-link current of a startup of compressor train #2 (y: frequency [Hz], x: time [s],color-map: logarithmic intensity of frequency component).DC current is representative for air-gap torque for bothmotor and generator.Around 400s into the measurement, the same interharmonic is crossing a torsional natural frequency around9.5Hz. The TNF is excited and then latches into aninstability. Further analysis revealed it was the first TNF ofthe GTGs and in consequence, lateral vibrations in theGTG’s gearbox were already approaching trip limits.The second TNF of the train #2 can also bedistinguished. It is visible in this logarithmic plot as soonas some load is transferred through the VSD. Theexcitation level is however within acceptable limits.At this point, a model based approach was chosen tounderstand the role of the converter and its controller toreject sub-synchronous voltage due to GTG torsionaloscillations and to decouple the load mechanical systemand the GTG mechanical system from each other.Fig. 4: Schematic diagram of current control loop of anLCI, where the mechanical oscillations of both grid andmotor-load are modeled as voltage disturbance.Fig. 5 shows a block diagram of the current controlloop, which directly can be transformed into a transferfunction. For small signal analysis, this linear modelneeds not to include the motor side, as it represents aconstant DC voltage inside the loop.
Current Control Open Loop: MagnitudeMagnitude [dB]100Fig. 5: Block diagram of the current control loop fortorsional stability study purposes.C (s ) 1I nomæ T s 1ö k P çç iè Ti s øThe rectifier is modeled as first order delay and scaledwith the nominal rectifier DC voltageU DC1nom . The delayof a 12-pulse rectifier is assumed to be maximum 1/12 ofthe period and minimum zero. The mean-delay is TU thusthe line frequency period T divided by 24.R(s) where1 U DC 1nom1 sTUTU T / 24average delay time based online frequency period TThe DC-link is modeled as an ideal inductor.L( s) sLDCThe disturbances in the AC line voltage and motorvoltage are fed with correct sign as AC disturbances tothe DC-voltage that drives the current in the DC-linkinductor. The two noise sources are applying voltage tothe DC-link inductor with opposite sign. The same voltagedisturbance on both side would consequently becancelled and not impact the current regulation, which istypical for common mode voltages. However – whenreferring back to the full schematic in Fig. 4 – it willexchange energy through the DC-link without affecting theDC-current.Depending on practical implementation (e.g. fordifferent VSD vendors), additional blocks such as currentfeedback filters etc. will have to be added.If we want to study stability of the current controller withrespect to reference, but also with respect tonoise/disturbance, we can now plot the bode diagram ofthe open loop. The diagram in Fig. 6 is taken for one ofthe 20MVA VSD in the system of study. These are dualchannel 6p LCI, with – per channel – nominal DC-linkcurrent of 2062A. nominal rectifier DC-voltage of 5103Vand a DC-link inductance of 7.5mH. The current controlleris set according the symmetric optimum. This means thekp and the Ti is chosen such that the 0dB crossing is nearthe maximum phase and in the middle of that section ofthe gain, which falls at -20dB. This tuning principle resultsin a kp of 0.1 and a Ti of 200ms. Setting the controlleraccording symmetric optimum would be the typicalparameterization for such a VSD and established thebase-line for the following experiments.0-50-110011010102310Current Control Open Loop: Phase-100-150Phase [ ]The current controller is of classic PI type. The controlpart of the model is formulated in p.u., with scaling bynominal current I nom at the input:50-200-250-300-110010110102310Fig. 6: Open loop bode diagram of the current controlloop in Fig. 5 with controller settings according symmetricoptimum (kp 0.1, Ti 0.2), x-axis labeled in rad.One can nicely distinguish the following areas in themagnitude plot. At the low frequency end, the curve dropsat 40dB/decade due to the DC-link and the I-part of thePI-current-controller acting together. Then, starting fromaround 1Hz, the curve is falling with 20dB per decadeonly. This is the effect of the DC-link inductor. At higherfrequencies, the first order delay of the rectifier starts tocome into the picture. In addition, there is a second orderlow pass filter in the current measurement in thisparticular VSD, which will reduce gain and add phaseabove 500Hz. However with this setting, the filter is notentering the topic of interest of this paper.The closed loop transfer functions of the control systemfor the reference and for the disturbances are very similarand do have the same denominator. This means, that forstability, the diagram in Fig. 6 is valid to study both cases.B.C (s ) R (s ) L(s )1 C (s ) R ( s ) L ( s )reference to outputL (s )1 C (s ) R(s ) L(s )disturbance to outputExperimental dataThere were two things out of the first experimental startup in Fig. 3 that needed an explanation:1)2)Why is the current controller not countering theoscillation at 9.5Hz in the DC-link current?ststThe 1 TNF of GTG and the 1 TNF of train #2 arevery close: Is there a combined electro-mechanicalinteraction with two mechanical systems involved?In order to answer the first, we turn to the bode diagramin Fig. 6. With 9.83Hz, the first torsional natural frequencyof the compressor train #2 is above the 0dB point of theopen loop control characteristic. Furthermore, the transferfunction from disturbance to DC-link current can beplotted for various controller gains now. When looking atFig. 7, it can be seen, that for a kp of 0.1, a voltagedisturbance applied to the DC-link inductor will be
amplified and have a relatively larger current asconsequence. At a torsional natural frequency, thiscurrent will again excite the generator (or the motor)further and lead to more voltage disturbance. Or in otherwords: The control system of the VSD has a negativecontribution to the damping of the torsional naturalfrequencies and risk for SSTI/closed loop torsionalinteraction is high.With a phase shift close to 90 degrees, the torqueproduced by this current has also the worst possiblephase-relationship, as air-gap torque of the generator (orthe motor) will be following motor mass speed by 90degrees as well, thus optimally “pumping” the torsionalresonance like this.30Magnitude [dB]2010Impedance plot of VSD [gain]kp 0.1kp 0.2kp 0.3kp 0.4kp 0.5kp 0.60-10-20-30-110100110102310Impedance plot of VSD [phase]100Phase [ ]500-50-100-150-110100110102310Fig. 7: Transfer function from a voltage disturbance tothe actual DC-link current, plotted for different gains of thecurrent controller and scaled to p.u.With the previously described model, a series ofexperimental load tests with the different kp settings as inFig. 7 were planned and – due to availability – executedon train #1. In these tests, load is varied in steps, whilespeed is always close to 100%. With the island grid of thesystem under study, both train #1 and train #2 showed thesame behavior, which was to be expected as all VSDswere identical. With base-line settings for the currentcontrollers, the first TNF of the GTGs was undamped toan extent that was not tolerable. Both trains and theGTGs have their first TNFs very close to each other, fortrain #1 and GTGs they are even identical to the first digitafter the comma. Consequently, the train #1 took also partmuch stronger in the SSTI of the VSD with thegenerators.Fig. 8 shows the waterfall plots of the DC-link current,which largely corresponds also to air-gap torque of theelectric motor. For values smaller than kp 0.2 themeasurement sequence could not be completed, sincetorsional oscillation turned out to be too large forincreased load currents. The series of full measurementswas only possible starting from kp 0.2. With increasingkp, the frequency, at which the undamping is thestrongest and gives most rise for torsional instability, isshifted from the 1st TNF (kp 0.1 or smaller) to the 2nd TNF(kp 0.3) and finally above both TNFs (kp 0.6). The sameeffect is observed on both sides, at the GTGs as well asat the compressor trains.Fig. 8: Waterfall plots of DC-link current under variousloads at steady speed for different current controller gains.DC-link current largely corresponds to electric motor airgap torque.
Any kp 0.2 is already sufficient not to excite the GTGs1st TNF, however due to the presence of the 2n
A linear model based approach is reported and demonstrated with a 20MW drive installation, which allows optimizing sub-synchronous torsional damping while maintaining the necessary dynamic response to load- and reference-changes in t
work/products (Beading, Candles, Carving, Food Products, Soap, Weaving, etc.) ⃝I understand that if my work contains Indigenous visual representation that it is a reflection of the Indigenous culture of my native region. ⃝To the best of my knowledge, my work/products fall within Craft Council standards and expectations with respect to
The modern approach is fact based and lays emphasis on the factual study of political phenomenon to arrive at scientific and definite conclusions. The modern approaches include sociological approach, economic approach, psychological approach, quantitative approach, simulation approach, system approach, behavioural approach, Marxian approach etc. 2 Wasby, L Stephen (1972), “Political Science .
based or region-based approach. Though the region-based approach and edge-based approaches are complementary to each other the edge-based approach has been used widely. Using the edge-based approach, a number of methods have been proposed for low-level analysis viz. image compressi
akuntansi musyarakah (sak no 106) Ayat tentang Musyarakah (Q.S. 39; 29) لًََّز ãَ åِاَ óِ îَخظَْ ó Þَْ ë Þٍجُزَِ ß ا äًَّ àَط لًَّجُرَ íَ åَ îظُِ Ûاَش
Collectively make tawbah to Allāh S so that you may acquire falāḥ [of this world and the Hereafter]. (24:31) The one who repents also becomes the beloved of Allāh S, Âَْ Èِﺑاﻮَّﺘﻟاَّﺐُّ ßُِ çﻪَّٰﻠﻟانَّاِ Verily, Allāh S loves those who are most repenting. (2:22
Athens Approach Control 132.975 Athens Approach Control 131.175 Athens Approach Control 130.025 Athens Approach Control 128.95 Athens Approach Control 126.575 Athens Approach Control 125.525 Athens Approach Control 124.025 Athens Approach Control 299.50 Military Athinai Depature Radar 128.95 Departure ServiceFile Size: 2MB
Mendelr Model-1988, 1992, The Jacob Kounin Model -1971, Neo-Skinnerian Model-1960, Haim Ginott Model (considered non-interventionist model approach) -1971, William Glasser Model-1969, 1985, 1992 (Quality school), Rudolf Dreikurs Model (Model of democracy)-1972, Lee and Marlene Canter Model (Assertive Discipline Model is one of the most spread
GMS TUTORIALS MODFLOW - Conceptual Model Approach Two approaches can be used to construct a MODFLOW simulation in GMS: the grid approach or the conceptual model approach. The grid approach involves working directly with the 3D grid and applying sources/sinks and other model parameters on a cell-by-cell basis.
