mechanical behavior of an under fault conditions, case...
TRANSCRIPT
International Journal of Rotating Machinery1999, Vol. 5, No. 3, pp. 219-230Reprints available directly from the publisherPhotocopying permitted by license only
(C) 1999 OPA (Overseas Publishers Association) N.V.Published by license under
the Gordon and Breach SciencePublishers imprint.
Printed in Malaysia.
Mechanical Behavior of an Industrial Gas Turbine underFault Conditions, a Case History*
STEFAN S. FLORJANCIC a’t, NOEL LIVELY b and RICHARD THOMAS
aABB Power Generation, CH-5401 Baden, Switzerland," bKentucky Utilities Company, E.W. Brown Generation Construction,Burgin, KY 40301, USA," CBently Nevada Machinery Diagnostic Services, Baton Rouge, LA 70817, USA
(Received 24 April 1998," Revised 5 May 1998," In finalform 7 July 1998)
The resolve of journal bearing field problems and the full vibrational assessment of anindustrial gas turbine, simple cycle rotor train, ABB’s GTllN2 is presented. Three unitsexperienced several times damage of the journal bearings on the compressor end. Theanalysis of the damage, including tests, indicated that insufficient jacking oil flow was thecause. The jacking concept was corrected and starting history has proven reliable operation,resulting in not one additional failure.
During the same period, two last stage blades failed in one of the units damagingneighboring blades additionally. The machine tripped under the unbalance. Some auxiliarypiping and the babbitt of the journal bearing was damaged. However, the rotor came to a safestop with only some rubbing, and the turbine casing fulfilled its containment functionperfectly.
In order to better understand the dynamic behavior of the train, an extensive vibrationmeasurement program was decided between Kentucky Utilities and ABB, and executedby Bently Nevada. The results proved excellent rotor balance, verified original designparameters, and no fluid induced instabilities were found. This joint effort allowed to fullyassess and prove the rotor dynamic integrity of the gas turbine.
Keywords. Case history, Journal bearings, Rotordynamics, Unbalance, Vibration measurement,Instability
1. INTRODUCTION
Three ABB GTllN2s, an upgrade of the provenGT11N1, went into commissioning at KentuckyUtilities (KU) during 1994. In the beginning of1995, after a significant number of starts had beenaccumulated on the fleet, it was realized that the
This paper was originally presented at ISROMAC-7.Corresponding author. Tel.: 056 205 7516. Fax: 056 2054690.
219
most loaded journal bearing on the compressor endwas damaged several times on various units. Theanalysis of the journal damage, including tests inthe spin pit, indicated that insufficient jacking oilflow during start up and shut down was the cause.The jacking concept was corrected by a higher flowoil pump and the subsequent starting history has
220 S.S. FLORJANCIC et al.
conclusively proven that these strongly loadedjournal bearings can operate very reliably, as notone additional failure has occurred.At the end of the same period, two last stage
blades of one of the machines failed due tosimultaneous high stochastic excitation at twonatural .frequencies. Neighboring blades were
damaged additionally, resulting in a big unbalance.The excessive vibrations caused the machine to tripimmediately, and the turbine casing fulfilled itscontainment function perfectly. Due to high vibra-tions some auxiliary piping was damaged, leadingto the loss of lube oil and the destruction of thebabbitt in the journal bearing liner. However, therotor came to a safe stop, and only some rubbingand burrs had to be rectified on this part.At this stage, it was decided between KU and
ABB, to launch an extensive vibration measure-ment program including unbalance testing in orderto better understand the dynamic behavior of themachine and the damage and rubbing pattern ofthe train. This program was contracted to BentlyNevada Corporation (BNC), with extensive infra-structure support of KU, and definitions andsupervision of the measurement by ABB. Theresults of the measurement proved excellent rotor
balance, and verified original assumptions on
rotordynamic design parameters. Additionally, itwas found that there are absolutely no fluid inducedinstabilities. The joint effort of the owner, the OEMand the contracted vibration measurement consul-tants allowed to fully assess and prove the rotor
dynamic integrity of the gas turbine.
2. PLANT DESCRIPTION,COMMISSIONING PHASE, ANDBEARING FAILURES
The E.W. Brown Combustion Turbine GeneratingFacility is a peaking plant located about 56 km(35 miles) southwest of Lexington, KY adjacent toan existing coal fired steam generating plant. Thepeaking station is designed to be unmanned withminimal physical connections to the steam plant,and is built around four simple cycle ABB GT11N2machines (Fig. 1) the first two planned for com-
mercial operation in 1994, followed by one each in1995 and 1996. The primary fuel is natural gasbacked up by #2 liquid fuel, with water injectioncontrolled NOx requirements of 42 and 65 ppm,respectively. The site design conditions are 32C(90 F) and 50% relative humidity at an elevation of270m (884 feet) above sea level.
Initial hot commissioning began with liquid fuelonly on the first two units in April 1994 sinceconstruction of the natural gas pipeline was not
complete. Operational and hardware related diffi-culties delayed commercial operation and resultedin a major inspection of one ofthe units. During this
inspection the compressor bearing was found to bemoderately wiped, where upon the OEM made thedecision to have the bearing repoured. The failuremode assessment was not conclusive, two possibil-ities were considered to be potentially the root
cause, either high loading, or contaminated lube oil.The machine was reassembled with the repairedbearing which failed during the first idle speed
FIGURE GTllN2.
INDUSTRIAL GAS TURBINE UNDER FAULT 221
attempt during recommissioning. Inspection of thebearing revealed heavy wiping. Again, the causewas unclear, either poor quality pour ofthe repairedbearing, or again contaminated lube oil was judgedto be the cause.By this time hot commissioning of the second
unit on gas fuel was well under way with base loadoptimization runs being attempted. During theseruns high bearing metal temperature alarms andassociated protective load shedding were observed,as well as unexpected temperature differences be-tween the two axially spaced thermocouples ofindividual bearing shells. The same behavior even-
tually was seen in the first unit as well. Additionally,the third unit experienced two compressor bearingfailures during its initial hot commissioning beforeachieving base load. In both cases it was determinedthat high loading was a contributing factor but notlikely to be the root cause.As not only the E.W. Brown GT11N2 units were
in the commissioning phase, a considerable amountof starts were amassed during a short period oftime, and six compressor bearing failures mani-fested in babbitt wiping in three individual unitswithin a total of 224 start stop cycles. An overviewof starts and failures before and after rectificationof the problem is given in Table I. Apparently,looking at the overall picture contaminated lube
TABLE Starts and operating hours before and after jack-ing oil system modification
Unit Jacking oil system (new--additional)
Starts Operating hours
Old New Old New
71 66 381 2652 69 57 310 2283 7 82 10 3634 0 53 0 2215 11 59 39 3566 16 61 56 3037 50 64 192 3028 0 87 0 15069 0 78 0 101810 0 79 0 1578
Total 224 686 988 6140
oil statistically could not be the root cause for thefailures on all units involved.
3. BEARING INVESTIGATIONS ANDSOLUTION
The bearing is an ABB design, low power loss,single horizontal feed bearing. It is similarly shapedto a lemon shape bearing and has one pocket at thebottom to lift the rotor with jacking oil during lowrotational speeds.The specific loading of the bearings, around
3 MPa (435 psi)is considerable, however, the OEM’slong term experience on gas as well as on steamturbines clearly indicated this load to be bearable.This is partially possible because of the high gradebabbitt used as a standard which corresponds toat least an ASTM Grade 3 type ofmaterial. In orderto re-check the journal bearing characteristics andthe material loading, the pressure distributionwithin the bearing clearance was calculated forthe dimensions of the bearing with an in-housecode based on Glieniecke (1970). The specific steadystate load case, and transient speed conditionswere considered. Typical pressure distributions are
shown in Fig. 2 for two speeds. At nominal speedthe maximum pressure is clearly below allowablecompression stresses for the babbitt.
Scrutinizing available temperature trend datalogs indicated temperature spikes as shown inFig. 3 during start up or coast down for severalfailed bearings. Therefore, the oil film for variousrotor speeds has been calculated, and the maximumpressure as well as the minimum clearance can beplotted against speed. To prove these theoreticalvalues, tests in the spin pit of the OEM have beendone. The journal bearing of a full size GT11N2rotor was instrumented with eddy current andpressure probes.As very small displacement deviations were
attempted to be measured, and the probes weresubject to thermal drifts (no temperature compen-sation) no absolute values could be measuredreliably over the time of the runs. However, film
222 S.S. FLORJANCIC et al.
FIGURE 2 Journal bearing pressure distribution at 270 rpm (left) and 2600rpm (right).
3’600
3’200
2’800
2’400
2’000
1’600
1’200
800
180--Rotor Speed1 160
-..---Bearing Tern pereature
I’- "2-Ii "" T 120140i, ,,o
: T o
0 008:00 08:05 01:10 08:15 08:20 08:25 08"30 08:35
Tim
FIGURE 3 Temperature spike during coast down.
Difference of Oil Film Thickness between 0and 12 I/rain Jacking Oil Flow
Rotor Speed [lhnin]
Oil Pressure and Max. Oil Film Pressureversus Speed
18
o.,: 8 Me.U’eo (.r...r; ’261a’(*r---- ’-"- -’
0
Rotor Speed [rpm]
FIGURE 4 Typical relative clearance and pressure measurement.
thickness changes with controlled changes ofjacking oil flow in a short time span were easilyarrived at. Additionally, the oil film pressure as afunction of speed was recorded.
Figure 4 indicates how well the change inpredicted clearance matches the measured differ-ence in clearance as a function of rotor speed and
jacking oil flow. With 121/rain (3.2 gpm) jackingoil flow the minimum oil film thickness can beincreased by 30-401.tm (1.2-1.6mils) at 400-500 rpm, and more below those speeds.For reason of balance, the maximum oil film
pressure cannot exceed the jacking oil pressurewithin the bearing pocket. The second half of
INDUSTRIAL GAS TURBINE UNDER FAULT 223
Fig. 4 shows nicely how the jacking oil limits themaximum oil film pressure at speeds below 500 rpmas it prevents the minimum bearing clearance to befurther reduced. On the other hand, measurementmatches the calculated film pressures well at higherspeeds, where the jacking oil supply flow is notsufficient to maintain the required high pressure inthe pocket anymore (there is an orifice between theline of supply and the bearing pocket).
Hence, sufficient jacking oil flow at an adequatepressure level not only maintains a minimum oilfilm thickness at low rotational speeds but it alsolimits peak pressure loading of the babbitt.As discussed with other bearing manufacturers
and the author of Glienicke (1970), a minimum filmthickness of about 20-30m (0.8-1.2 mils) has tobe maintained to ensure safe operation of a jour-nal bearing without scratching or even wiping thebabbitt, even if fine lube oil filters are in use.Additionally, even a high grade babbitt will notsustain the excessive oil film peak pressure loadingwhich occurs in the absence of sufficient jacking oilflow at low rotational speeds for a highly loadedbearing.Thejacking oil supply ofthe new auxiliary system
for the GT11N2 units hence has been checked. Witheffectively less than 0.71/min (0.2 gpm) the flow,and therefore the minimum clearance was found tobe marginal, leading to the experienced 2.7% start/stop failures. Evaluation of the investigationsperformed led the OEM to reset the jacking oilflow to 61/min (1.6 gpm), and thereby increasingthe minimum film thickness by more than 20 tm(0.8 mills) at low speeds. Modifications were doneon all existing units.
After the rectification, and by the time this articlewas written, the GTllN2 fleet had accumulatedmore than three times the start stop cycles (andmore than six times the fired hours) than with themarginal jacking oil flow without one new failure ofthe bearing babbitt; see Table I.From those statistics it hence can be concluded
that energy efficient, highly loaded bearings aresafe during steady state operation at nominalspeed. No further failures have to be expected if
sufficient care is taken during start up and coastdown conditions, i.e., if sufficient jacking oilflow is available during low transient rotationalspeeds.
4. BLADE INCIDENT WITH FAULTCONDITION UNBALANCE
During a test run intended to observe the abovementioned bearing behavior on one ofthe KU unitsthe machine was approaching base load when ittripped on very high vibration. The fire whichensued in the machine’s enclosure and exhaustductwork was extinguished quickly and safely bythe CO2-system. A visual inspection revealed thatno parts were ejected from within the machine, butconsiderable damage to the turbine/generator setsuch as broken lube oil line connections (leading tothe fire), slight movement, broken journal bearings,and failure of the generator exciter support hadoccurred. A cursory examination from the exhaustend of the turbine revealed that several last rowblades had failed. The OEM issued a stop run orderwhile an investigation and failure analysis wasperformed.The root cause of the excessive unbalance was
identified to be two failed last stage blades. Detailed
investigation of the telemetric measurement takenon these blades in conjunction with metallograph-ical investigations of the fracture surface andFinite Element Method (FEM) stress and modalanalysis revealed a rather surprising effect. Whilethis blade row was not in resonance with anyengine rotating frequency harmonic, both, its firstand second natural frequencies were subject tounusually high, though not excessive stochasticexcitation. Individual resulting stress amplitudeshad been analyzed before and were clearly withinacceptable limits, i.e., by far not sufficient for a highcycle fatigue (HCF) failure. However, the secondfrequency proved to be almost exactly an integermultiple of the first one. Additionally, the modalFEM analysis indicated that the two mode shapeswere resulting in a maximum stress amplitude at
224 S.S. FLORJANCIC et al.
almost the identical location in the airfoil trailingedge to platform transition. This coincidencesporadically led to the stress amplitudes to addarithmetically rather than statistically at approxi-mately the lower natural frequency period in anarea of considerably high mean stresses where thecentrifugal load of a last stage blade acts in thevicinity of a notch. Hence, the failure only tookplace after a number of fired hours untypically highfor HCF.
Metallographical analysis of the failed surface aswell as available material property data provedthe failure scenario of HCF with high mean stress.With this insight, the design change in the criticalbut very limited blade area was straight forward,the two frequencies were detuned and the meanstress was lowered. Again, Table I clearly proves thesuccess of the design change with the number ofoperating hours, i.e., "old, Unit 1" until failure ascompared to "new, Unit 8 through 10", the latterunits being originally supplied with the modifiedlast stage blade.The two failed blades were reflected by the
casing, and, in a domino-like effect damaged otherblades in their vicinity to various extents. Theresulting total equivalent blade out was around 10neighboring blades, see Fig. 5, an unbalance thatmost industrial gas turbine rotors would not becapable of carrying without major destructionFlorjancic et al. (1997).
557585
52 10112 Initial Failure
1 at #15a749 13
1447 15
42
FIGURE 5 Unbalance by blade loss in the last turbine stage.
As the machine was opened for analysis andrepair, it became apparent that the gas turbine rotorhad survived the extremely large imbalance incidentvery well. Even though considerable rubbing hadoccurred in the compressor section it resulted inrelatively minor damage to the rotor. The surfacerubbing and some burrs were rectified easily. How-ever, it also became apparent that improvementto the existing structure, mainly auxiliary piping,was possible and could be implemented with rea-
sonable effort.Increasing tle rigidity of the broken generator
exciter support again was straight forward withadditional and stronger bolting to sustain loadsmuch higher than originally anticipated. Theauxiliary piping was further investigated. Appar-ently, a balance between distance of supports,piping flexibility, and attached masses (flanges)had to be found. The design of such "low tech"but very obviously still essential items has thereforebeen investigated in very detail. Design principleswere improved to an outlay as insensitive as pos-sible to very high transient excitations, althoughthey only occur during uncommon but extremefault conditions. It was found that designs can beestablished which will withstand the extremeloading of a last stage failure but that such pip-ing will need thorough inspection in the aftermath,as the endurance limit almost certainly will beexceeded during such an event, i.e., such highloading can be sustain for a limited number ofcycles, only.
Rectifying design principles after such an event
certainly does lead to an overall improved productsafety. However, the need for a comprehensiveunderstanding and prove of the theoretical back-ground of rotordynamics coupled to structuraldynamics was strongly felt after the incident. TheOEM and the owner of the gas turbines thereforedecided to undertake a joint effort to establishbaseline rotor- and structural dynamic base linedata of the units as installed on site. This investiga-tion was seen as a unique possibility to check andprove design parameters, a task impossible to fulfillwith high accuracy on a test stand. With the owners
INDUSTRIAL GAS TURBINE UNDER FAULT 225
best interest to fully understand his equipment, andhence, with his full support the following investiga-tion could be undertaken.
5. MEASUREMENT OF GAS TURBINEROTOR AND STRUCTURAL DYNAMICS
Figure 6 shows all measurement locations along therotor train as defined by the OEM. Bearings arenumbered through 6, starting at the turbine end.For a better assessment ofmode shapes, the locationat the intermediate shaft (spool) was chosen inaddition to standard locations at bearings. Specialattention was also given to the two bearing loca-tions of the gas turbine itself. Here, shaft mea-surements were taken inboard and outboard ofthe bearing, to avoid inaccurate measurement inpotential mode shape nodes around the bearing,and to assess shaft to stator alignment during opera-tion. The movement of the bearing shell relative toits support structure was measured in bearing 1,planes 4 and 5.The displacement measurements of bearing 1,
planes 2, 4 and 5, of bearing 2, plane 2, and of theintermediate shaft and generator in general are non-standard. An additional non-standard measure-ment is the absolute vibration at the intermediateshaft location which was introduced to gain fullrelative and absolute vibration characteristics alongthe entire rotor train. Hence, extensive instrumenta-tion and wiring was needed. This part of the projectwas executed by the owner, with his own staffon site. He also installed a measurement trailer
adjacent to the gas turbine enclosure, to which heled all, standard and non-standard vibration mea-surement channels.Measurement and data reduction itself was
contracted to Bently Nevada Corporation (BNC).The OEM specified the locations and type ofmeasurement needed, and BNC was responsibleto establish the measurement hardware set-up, datastorage and reduction, as well as to perform theentire measurement. During the measurement, theowner was operating the unit jointly with OEMcommissioning staff. Special care had to be taken,as deliberately large trial unbalance weights were
installed alternatively on both sides of the gasturbine to assess its characteristic behavior.
5.1. Measurement Equipment
The transducer suite for this investigation,consisted of proximity probes, velocity seismictransducers, and thermocouples. The additionalproximity probes in planes 2 of Bearing #1 and #2allowed to evaluate the actual rotor position anglewith respect to the vertical direction and theeccentricity ratio at midspan of both turbinebearings. Planes 4 and 5 of Bearing #1 were
introduced to measure the "tilting" of the bearingliner against the housing during operation. Thevelocity seismic transducers measure the absolutemotion of the bearing housing at both Bearings #1and #2, while thermocouples were used to mea-sure bearing metal temperatures and cooling airtemperature in four of the support struts forBearing #1.
B1 B2 IS
3 3
1452 2
B3 B4 B5
FIGURE 6 Measurement set-up.
226 S.S. FLORJANCIC et al.
Shaft relative vibration measurements at Bear-ings #1 and #2, as well as bearing liner to bearinghousing relative displacement measurements at
Bearing #1, were obtained with a permanentlyinstalled BNC proximity probe transducer system.Transducer calibration was consistent with BNCspecifications of 7.gmV/gm (200mV/mil), +5%.Frequency response characteristics for proximityprobes are often stated as DC (zero frequency)to 10,000Hz (600,000cpm). However, at higherfrequencies, displacement amplitudes are quitesmall, and typically fall below the signal to noiseratio of the measurement system. For this reason,the most significant data from proximity probesoccurs in the frequency domain of DC to 1,500 Hz(90,000 cpm).
Bearing housing absolute velocity measurementswere made at Bearings # and #2 with permanentlyinstalled seismic transducer systems. Transducercalibration was within the nominal manufacturer’sspecifications of 19.7 mV/mm/s (500 mV/in/s), witha useable frequency response offrom 10 to 1,000 Hz(600-60,000cpm). The housing absolute velocitydata was integrated to displacement and then postprocessed with the shaft relative data to obtainshaft absolute motion.One permanently installed proximity transducer
was used as the once per revolution trigger probe.It is installed at 0 at the turbine to intermediateshaft coupling.
The vibration transducers are named to providethe location of the transducer along the machine
train, i.e. Bearings #1 and #2 etc., and the angularorientation of the transducer, i.e. Y (45 L) or X(45 R).The transducer convention adopted by the owner
is to view the machine train from the drivenmachine (generator) looking back to the driver(gas turbine). Using this convention, rotation is inclockwise (CW) direction. At all five bearings, the Ytransducer is mounted at 45 left from 0 (top deadcenter), and the X transducer is mounted 45 rightfrom 0. One must maintain this convention, i.e. atrue Cartesian coordinate system, where the / Yaxis (vertical transducer) is always 90 counter-clockwise of the +X axis (horizontal transducer),in order to properly evaluate the vibration responsedata. A transducer overview for Bearing #1 andBearing #2 is exhibited in Fig. 7.
5.2. Data Processing Instrumentation
All steady state and transient startup and shutdownvibration response data was obtained via ten BNC208 Data Acquisition Instrument Units (DAIU).Two units were paired to form a 16 channel digital,real time, data acquisition package. Five 16 channelinstrumentation grade digital tape recorders were
installed to provide taped backup data. All vibra-tion and process variable data was acquired byBNC’s Machinery Diagnostic Services personnel.
FIGURE 7 Bearing #1 (turbine side); Bearing #2 (compressor side).
INDUSTRIAL GAS TURBINE UNDER FAULT
TABLE II Comparison of measured and calculated bearing and structural support characteristics
227
Brg. Bearing stiffness 10
BNC
Vert. Horiz.(N/m) (N/m)
Bearing damping x 10 Structural stiffness 109
ABB BNC ABB BNC
Vert. Horiz. Nom. Vert. Horiz.(N/m) (N/m) (Ns/m) (Ns/m) (Ns/m)
ABB
Vert. Horiz. Vert. Horiz.
(N/m) (N/m) (N/m) (N/m)
1.12.61.61.20.02
0.7 3.8-4.0 1.4-1.5 5.3 14-15 2.4-2.5 5.3 3.51.4 5.2-6.1 1.8-2.0 7 19-22 3.2-3.5 10.5 7.91.4 1.4-1.9 0.6-0.8 3.5 5.8-7.6 1.0-1.3 8.8 7.00.53 1.1-1.5 0.4-0.6 3.5 4.5-6.0 0.8-1.0 8.8 7.00.01 0.2-1.3 0.03-0.2 0.9 0.6-1.2 0.1-0.2 0.2 0.04
TABLE III Comparison of measured and calculated critical speeds
Modalrange (cpm)
BNC measurement data Stability analysis Response analysis
DAMP Active component DAMP Active component Active component
GT IS GE EX GT IS GE EX GT IS GE EX
450-500 3-7 x800-950 9-12 x1050-1200 6-10 x1300-1500 9-12 x1600-1850 8-21 x2000-2200 302400-2700 3-27 x2800-3100 3-7 x
Structural modex x Structural mode (coupled?) x x x x
x 15-18 x x x x x xStructural mode x x x x
x 16-17 x x x xx 37-58 x x x x x xx x 24-25 x x x x x
x 17-18 x x x x x x x
The DAIU is an eight channel instrument thatcaptures and processes simultaneous vector andwaveform data for each channel. A vector datarecord consists of five types of information:
(1) Direct (overall) vibration amplitude.(2) IX, 2X, nX amplitude.(3) Gap voltage.(4) Machine speed (rpm).(5) Time.
The sampling rate was 128 samples/revolutionwith a discrete sample occurring over 8 revolutions
(1024 samples) for the machine speed at hand. Eachacquisition unit (pair) was connected to a 486notebook computer, with a total of five systems,80 measurement channels were available. Thevibration/process variable data acquisition wasaccommodated via the five data acquisition sys-
terns. Each data acquisition computer was oper-ating with ADRE(R) for Windows(R) softwaredeveloped by BNC.Measurements are compared to calculated results
in the next paragraph, an overview is given inTables II and III. More detailed information aboutmeasurement results, i.e., type of data reduction,mode shapes, extraction of structural response dueto unbalance, etc., is beyond the scope of this
publication and can be found in Florjancic et al.(1998).
6. LINEAR ROTORDYNAMICS
Measured results were compared to originally cal-culated natural frequencies and forced responseruns, based on the model shown in Fig. 8. To cover
uncertainties, the support stiffness (structure and
228 S.S. FLORJANCIC et al.
G enerator
G as Turbine
FIGURE 8 Standard rotordynamic model.
bearing in series) of the rotordynamic model isvaried in a sensitivity study over a considerablerange, i.e., typically from 108 to 109 N/m (6.105-6. 1061bs/in). Individual properties derived fromthe unbalance measurements compare well withthose combined values. More detailed stiffnessestimates based on a Finite Element (FE) modelof the gas turbine support structure had beenprovided to allow BNC to assess measurementmore easily. The calculated values correspondwell with the measured ones, as can be seen inTable II. NB.: Measured properties of Bearings #3through 5 are less reliable as unbalance excitationhas only been applied to the gas turbine part of therotor train.
Discrepancies in bearing stiffness and dampingare in the order of a factor 2-3. They are caused tosome extent by tolerances in bearing diameter andlength. However, the strongest influence on param-eter discrepancies result from variations in effec-tive static load on each bearing due to non-perfectrotor train alignment (as discussed in Florjancicet al. (1998) (Fig. 8), from effectively non-parallelaxes of rotor and bearing liner which is not takeninto account in the theoretical prediction, andthe limited accuracy of measurement data. Thecomparison of the parameters given clearly indi-cates the realistic accuracy of theoretical prediction(of a "perfect system") compared to effectivemachine behavior.
Differences in structural stiffnesses are somewhatlarger than in bearing parameters. The extent and
complexity of the support structure is considerable,including many bolted and fitted interfaces. TheFE model shown in Fig. 9 is not sophisticatedenough to represent exactly the effective multi-layermachine support structure. However, even a more
detailed model would show discrepancies due to
tolerances, and, more importantly, due to vary-ing conditions in interfaces. Non-linearities ininterfaces (friction and looseness) also change thestiffnesses according to effective pre-load, i.e., withthe state of alignment. It was therefore chosen tolimit the scope of the FE model, and all interfacesand their variations were not included for theanalysis.
Critical speeds of the original design study werealso compared to measured values. Again, as testunbalances had been put into the gas turbine rotoronly, critical speeds of the generator part of therotor train were only weakly excited and are lessaccurate to determine. An overview of predictedand measured critical speeds is given in Table III.
Structural modes obviously cannot be predictedby the model of Fig. 8, as the support structure ismodelled only by a spring (no mass). Other naturalfrequencies and modal dampings compare well withthe measured values. The portions of the rotorparticipating in the calculated and measured vibra-tion mode shape also are closely related, consid-ering that unbalance was placed on the gas turbineportion, only. Identifying critical speeds fromforced response calculations is indicated in additionto the natural frequencies. As structural modes
INDUSTRIAL GAS TURBINE UNDER FAULT 229
Center Support
Pendulum
Exhaust Casing Support
FIGURE 9 Detailed rotordynamic model.
are at low frequencies, and the rotor modes athigher frequencies, closer to the running speed, are
predicted well, the simple model is sufficientlyaccurate to allow a safe prediction of vibrationalbehavior of the rotor during steady state operationand higher transient rotor speeds. At lower speedtransients, accuracy is of minor importance, asforced responses are smaller.A more detailed rotordynamic FE model of the
unit was established, including the mass distribu-tion of the stator part, i.e., the gas turbine casingand the silo combustor, as well as the FE model ofthe support, rather than just support stiffnesses; see
Fig. 9. As expected, results compare well with thesimpler model, and the measured structural modearound 1,300-1,500cpm is now predicted. Addi-tional effort would have to be undertaken to fullymodel the multi-layer rotor-stator structure inorder to predict all modes with higher accuracy,see also Florjancic et al. (1998) for discussions.However, as the measured results indicate, the
simpler assessment is of sufficient accuracy andinformation to assess the safe rotordynamic behav-ior. The overall results fully justify the judgmentof predicted bearing and support parameters to begood and adequate compared to measured ones.The detailed model was used to simulate big
unbalances in the area of the last stage blade. Not
surprisingly, it was found that linear rotordynamicscannot account for an unbalance as it had occurredduring the blade loss incident experienced. Even thepresent design of an unbalance insensitive rotor,Florjancic et al. (1997), has to rely on iterative timestep calculation of the non-linear bearing coeffi-cients for unbalances exceeding the equivalent ofseveral single blades.A study executed in this manner indicated that
information on the behavior of journal bearingswith eccentricities close to one, i.e., at the beginningof oil film break down and during metal to metalcontact is needed for huge unbalances but notavailable in literature. Additionally, it was realizedthat a time step coupling of rotordynamics with a
non-linear structural code, one influencing theresults of the other due to impacting, is needed to
fully assess the consequences of a huge unbalance.A possibility to test bearings to destruction andto couple rotordynamics with non-linear structuralcodes has been outlined in Childs (1996), but thecoupled tools are not established.
7. CONCLUSIONS
The joint effort of the owner and the OEM of an
industrial gas turbine allowed for the successful
230 S.S. FLORJANCIC et al.
solution of a field problem, and the full assessmentof the rotordynamic behavior or the gas turbine asbuilt and installed.
Detailed theoretical and experimental investiga-tions on the journal bearings used proved thatstrongly loaded journal bearings work very reliablyif sufficient care is taken with the according auxi-liary equipment. Speed transient corditions are ofprimary importance, and sufficient jacking oil hasto be provided in order to ensure a minimum oilfilm thickness. This ascertains that the local filmpressure does not exceed limits acceptable for thebabbitt material, that no rubbing will occur, and itadditionally makes the bearing insensitive to con-taminated lubrication oil. Operational history afterrectification of the marginal jacking oil flow statis-tically proves the theoretical and experimentalfindings.A blade loss incident lead to a unusually large
unbalance of the gas turbine. After the safe shutdown it was found that the rotor also in practicebehaves as unbalance insensitive as described inFlorjancic et al. (1997). Damage to the rotor wasminor rubs and burrs. Structural damage to auxil-iaries was of more concern, and design principleswere changed to ensure that in the future suchpiping will safely sustain the fault conditions.The extensive vibration measurement leading to
enhanced and proven understanding of thedynamics of the rotor train was only possible as acooperation of the owner, responsible for theinfrastructure, the OEM defining the program anddata needed, and the contracted vibration specialistsexecuting the campaign and establishing appro-priate data reduction methods. Comparison of themeasured and reduced data with design valuesproved sound design tools and accuracy had beenachieved. Parameters used for the modeling of the
support structure were measured to be correct, andhence, the dynamic behavior of the rotor train hadalso been predicted with reliable accuracy.
Non-linear rotordynamics was found to beneeded to assess a huge unbalance equivalent toseveral blades. However, tools to couple rotor-dynamics and the stationary components of the gasturbine under impact would be needed but are notreadily available.
This entire project showed how the cooperationbetween owner, OEM, and subcontractors can leadto success and gain for all parties involved.
Acknowledgments
The authors wish to thank their companies and allthe staff involved for the support given to writethis publication, proving how well the coopera-tion went. Special thanks are given to R. Kellererfor his outstanding work on the journal bearings,E. Holoehr for his many rotordynamic studies, andJ. McElhaney for his compilation of the compar-ison of calculated versus measured results.
References
Childs, D.W. (1996) An R&D program to develop validatedcomputer codes for predicting survival and secondary damageof combustion-gas-turbine/generator systems associated withblade-loss incidents, Texas A&M University, College Station,Texas.
Florjancic, S.S., Franklin, W. and Lively, N. (1998) Vibrationmeasurement techniques on an industrial gas turbine rotor,43rd ASME Turbo Expo, Stockholm.
Florjancic, S.S., Pross, J. and Eschbach, U. (1997) Rotordesign in industrial gas turbines, 42nd ASME Turbo Expo,1997, 97-GT- 75.
Glienicke, J. (1970) Experimentelle Ermittlung der statischenund dynamischen Eigenschaflen von Gleitlagern ftir schnel-laufende Wellen Einfluss der Schmierspaltgeometrie undder Lagerbreite, Fortschritt Berichte der VDI Zeitschriften,Reihe 1, Nr. 22, VDI-Verlag, Duesseldorf.
EENNEERRGGYY MMAATTEERRIIAALLSSMaterials Science & Engineering for Energy Systems
Economic and environmental factors are creating ever greater pressures for theefficient generation, transmission and use of energy. Materials developments arecrucial to progress in all these areas: to innovation in design; to extending lifetimeand maintenance intervals; and to successful operation in more demandingenvironments. Drawing together the broad community with interests in theseareas, Energy Materials addresses materials needs in future energy generation,transmission, utilisation, conservation and storage. The journal covers thermalgeneration and gas turbines; renewable power (wind, wave, tidal, hydro, solar andgeothermal); fuel cells (low and high temperature); materials issues relevant tobiomass and biotechnology; nuclear power generation (fission and fusion);hydrogen generation and storage in the context of the ‘hydrogen economy’; andthe transmission and storage of the energy produced.
As well as publishing high-quality peer-reviewed research, Energy Materialspromotes discussion of issues common to all sectors, through commissionedreviews and commentaries. The journal includes coverage of energy economicsand policy, and broader social issues, since the political and legislative contextinfluence research and investment decisions.
SSUUBBSSCCRRIIPPTTIIOONN IINNFFOORRMMAATTIIOONNVolume 1 (2006), 4 issues per year Print ISSN: 1748-9237 Online ISSN: 1748-9245Individual rate: £76.00/US$141.00Institutional rate: £235.00/US$435.00Online-only institutional rate: £199.00/US$367.00For special IOM3 member rates please emailssuubbssccrriippttiioonnss@@mmaanneeyy..ccoo..uukk
EEDDIITTOORRSSDDrr FFuujjiioo AAbbeeNIMS, Japan
DDrr JJoohhnn HHaalldd, IPL-MPT,Technical University ofDenmark, Denmark
DDrr RR VViisswwaannaatthhaann, EPRI, USA
FFoorr ffuurrtthheerr iinnffoorrmmaattiioonn pplleeaassee ccoonnttaacctt::Maney Publishing UKTel: +44 (0)113 249 7481 Fax: +44 (0)113 248 6983 Email: [email protected] Publishing North AmericaTel (toll free): 866 297 5154 Fax: 617 354 6875 Email: [email protected]
For further information or to subscribe online please visitwwwwww..mmaanneeyy..ccoo..uukk
CCAALLLL FFOORR PPAAPPEERRSSContributions to the journal should be submitted online athttp://ema.edmgr.com
To view the Notes for Contributors please visit:www.maney.co.uk/journals/notes/ema
Upon publication in 2006, this journal will be available via theIngenta Connect journals service. To view free sample contentonline visit: wwwwww..iinnggeennttaaccoonnnneecctt..ccoomm//ccoonntteenntt//mmaanneeyy
NNEEWW
FFOORR 22000066
Maney Publishing on behalf of the Institute of Materials, Minerals and Mining
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttp://www.hindawi.com Volume 2010
RoboticsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Journal ofEngineeringVolume 2014
Submit your manuscripts athttp://www.hindawi.com
VLSI Design
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Modelling & Simulation in EngineeringHindawi Publishing Corporation http://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
DistributedSensor Networks
International Journal of