Chapter 9 Common Malfunctions

7/30/2019 Chapter 9 Common Malfunctions

1/64

395

C H A P T E R 9

Common Malfunctions 9

Process machines are subject to manymalfunctions that range from internal forcing functions, to self-excited mecha-

nisms, to externa l forces, plus a myria d of physical phenomena tha t ma y impose

dynamic loads upon the machinery. Some of the malfunctions (e.g., unbalance)are common to all rotating machines. Other excitations (e.g., gear mesh forces)

are unique characteristics for a particular type of mechanical device. These vari-

ous excita t ions are norma l for a ll moving elements, and they form a significan t

portion of the beha vioral pa ra meters for t he specifi c piece of ma chinery.

When the forces responsible for the excitations increase beyond normal or

expected limits, this is often detrimental to the integrity of the equipment. This

physical cha nge in a pplied forces is often detecta ble as a change in th e machin-

ery vibrat ion r esponse cha ra cterist ics. In order to provide the dia gnostician with

some addit ional insight into these mechanical relat ionships, these excitat ions

and the resultant malfunctions will be examined. Sequentially, the common

types of machinery malfunctions are reviewed, and specific case histories are

presented in this chapter. These common malfunctions are applicable to most

rotat ing ma chines, an d th ey include forced a s w ell as free vibration mecha nisms.In a ddit ion, a series of unique excita t ions associat ed with specifi c machine types

will be presented and discussed in the following chapter 10.

SYNCHRONOUS RESPONSE

The synchronous, or running speed, or fundamental, or 1X motion of a

rotat ing element is an inherent chara cterist ic of every ma chine. It should be rec-

ognized that all machines function with some level of residual unbalance. All

ma chines must operate wit h some fi nite clearan ce between stat iona ry an d rota t-

ing elements. Since it is physically impossible to produce a perfectly st ra ight a nd

concentric rotor, another source of synchronous motion is apparent. In addition,

all ma chines a re supported by va rious compliant structures a nd foundat ions.Vibration response measurements on any machine with virtually any

transducer will reveal a component at rotational frequency. Not surprisingly, this

universally common excita t ion accounts for t he ma jority of th e ma chinery m al-

function mechanisms. Un fortuna tely, the a na lysis of 1X vibrat ion is signifi cantly


2/64

396 Chapter-9

complicated by the fact that many different mechanical malfunctions appear as

changes in the rotational motion. Hence, the machinery diagnostician is faced

with the real dilemma of observing a single frequency, and attempting to diag-

nose the origin of an increased vibration amplitude.The fi rst line of at t ack resides in reviewing the t ra dit iona l relat ionship dis-

cussed earlier in this t ext as:

(9-1)

It is clear tha t the vibra tion response is directly proportional t o the a pplied

force when the restraint or stiffness is held constant. As force increases, the

resulta nt vibr a tion response will a lso increase. This ty pe of relat ionship is intrin-

sic with the fundamental concepts associated with activit ies such as rotor bal-

ancing. It is also clear that vibration response is inversely proportional to the

restra int or st iffness w hen the a pplied force is held consta nt . In this condition, as

the restraint decreases, the resultant vibration response will increase. Forinsta nce, a bea ring w ith increasing clear an ces w ill typically exhibit a reduction

in support stiffness (i.e., restraint). Assuming a constant unbalance force, the

rotor w ill vibrate a t a higher response level.

The third possibility for changes in vibration response amplitudes must

consider variat ions in both the force and the restraint . As stated by Donald E.

B ently in an issue of Orb i t1 magazine:Vibr ati on is usuall y eit her th e r esult of a bowed r otor or is th e resul t of a

force or m oment acting on t he sti ffn ess of that machi ne to that force or m oment . Assuch, vibrat ion i s actu all y a r ati o and fr equent ly n ot an end objective measure-ment in i tsel f. Remember th at forces and moment s flow t hr ough, moment s ar emeasur ed across, and :

Th us, to best r ead th e behav i or of a m achin e, it i s oft en n ecessar y to knowBOT H the numerat or an d th e denomin ator of th is sim ple relati onship Obvi-ously, you m ust m ake some sor t of assum pti on of sti ffn ess or for ce i n ord er t o havea knowl edgeable vibr ati on measurement . We do thi s r egula r ly and wi ll cont in ueto do so. However, to im pr ove our capabil it i es of oper ati ng machin er y, the mea-sur ement of observ ed oper ati ng d ynam ic stiff ness wi l l become more importan t i nth e fut ur e, as eit her th e numera tor (Dynam ic Forces) may be in corr ect, or t hedenomi nat or (Dynam ic Sti ffn ess) may be incor r ect.

In ma ny insta nces it is extraordinarily diffi cult t o quant ify the active forces,

or the associated restraints (st iffness). This inability to define actual machine

para meters often yields to a n investigat ion of changes in response constituents.For instance, the effective rotor support characteristics were discussed in chap-

ter 4. This discussion concluded that for most types of process machinery that

1 Dona ld E. B ently, Vibration levels of machinery, Orb i t, Vol. 13, No. 3 (September 1992), p. 4.

ResponseFo rce

Res t r a i n t ----------------------------=

D y n am i c Mo t i o n V i b r a t i o n ( ) D y n am i c ForcesorMomen t s D y n am i c St i f f ness--------------------------------------------------------------------------------------=


3/64

Synchronous Response 397

the effective st iffness is related to the oil film stiffness and the overall bearing

housing s tiffness in equ a tion (4-16). This expression is r esta ted a s follows :

(9-2)

The bearing housing stiffness includes the support pilings and foundation,

the grout a nd ba seplat e, the bearings or ma chinery pedesta ls, plus the st iffness

of the bearing housing itself. Often a visual inspection of the machinery will

identify t he condit ion of these mecha nical elements. For exam ple, it is quite clear

when a bearing housing is loose on a pedestal, or when grout degradation has

occurred. If these support elements remain in good condition, then the oil film

stiffness chara cterist ics should be examined.

One of the most powerful and commonly available tools for evaluating bear-

ing condit ion is an exam ina tion of th e journa l position wit hin t he bearin g. This is

performed with radially mounted X-Y proximity probes as discussed in chapter 6

of this text. Specifically, the change in shaft centerline position was determinedw ith th e vector example previously display ed in F ig. 6-20. For purposes of com-

pleteness, this sa me dia gra m is r eproduced in th e follow ing Fig. 9-1. Within this

diagram, the change in probe DC gap voltages may be vectorially summed to

determine the overall shift in journal position from an initial stop

condit ion to an

operating

posit ion of the shaft w ithin the bear ing.

Substa ntia l cha nges in radial sha ft posit ion ar e often associat ed with bear-

ing damage. This is particularly true for a horizontal machine that has experi-

enced dam age t o the bott om half of the bearing. In t hese instances, the probe gap

volta ges will reveal a vertical d rop of the sha ft int o the babbitt . This type of dam-

age often results in a change of the synchronous vibration combined with the

posit ion shift . In some cases t he 1X shaft vibra tion will increase, as shown in t heinduction motor case history 44. In other situations, the running speed vibration

will decrease, as illustrated by the refrigeration compressor case history 50. In

both cases, the change in rotational speed vibration was associated with a dis-

tinct va riat ion in bearing support st iffness. It should also be mentioned that the

Fig. 91 Shaft CenterlineShift Vector As MeasuredWith X-Y Proximity Probes

1Kef f------------ 1

Koi l----------- 1

Kh sg-------------+=

13545

90

180 0

Shaft CenterlineShift Vector

4.4Mils

@45

5.4

4Mil

s@

81

3.2Mils@

135


4/64

398 Chapter-9

analysis of both of these case histories depended heavily on the measured

changes in sha ft centerline posit ion da ta .

Furthermore, variations in nominal bearing clearance (i.e., too big or too

sma ll) represents one of the most common ty pes of mechan ical problems on pro-cess machinery. These abnormal clearances may be due to poor installation,

attrition of the bearings during operation, or the presence of some forcing func-

t ion tha t is hammeri ng out

the bearing cleara nce. Clearly, there are visua l a nd

measurement m ethods for evalua ting cha nges in rotor support st iffness. If t hese

techniques do not identify any change in restra int (st iffness), then it is reason-

able to conclude that the vibrat ory change is primarily associat ed with a change

in th e applied force(s). The rema inder of this cha pter w ill be devoted t o an exa m-

ination of common forcing functions, and traditional excitation mechanisms.

These malfunctions will also be illustrated with descriptive case histories

M

ASS

U

NBALANCE

Mass unbalance represents the most common type of synchronous excita-

tion on rotating machinery. Every rotor consists of a shaft plus a series of inte-

gral disks used for turbine wheels, or thrust collars. Turbomachinery rotors may

a lso include a series of slip on element s such a s compressor w heels, pump impel-

lers, thrust collars, spacers, coupling hubs, etc. Although each item is typically

ma nufactured t o high dimensiona l tolera nces, a residual unba lance is present in

each element. It is self-evident tha t t he residual unba lance for a single machine

disk may be satisfactory, but the combined effect for a stacked rotor may be com-

pletely una ccepta ble as described by J ohn Ea st

2

. To addr ess th is issue, a va riety

of tools and techniques have evolved to correct mass unbalance problems. Since

this is a fundamental problem with all rotating machinery, chapter 11 of this

text ha s been devoted to a detailed explana tion of ma ss unba lance response, andthe va riety of methods used to determine a nd correct rotor unbala nce.

From a recognition standpoint , mass unbalance will normally produce a

transient Bode plot as shown in Fig. 9-2. This calculated plot for a forced unbal-

ance spring-mass-damper system was extracted from Fig. 2-19 of this text. At

speeds well below the resonance, the vibration response will vary as the speed

squa red. The a pplied centrifuga l force ma y be estima ted by eq ua tion (9-3):

(9-3)

where:

F

cent

= Centrifugal Force Due To Residual Unbalance (Pounds)

Mass

= Effective Mass of Residual Unbalance (Grams)

Radius

= Effective Radius of Residual Unbalance (Inches)

RPM

= Shaft Rotational Speed (Revolutions / Minute)

2 J ohn R. Ea st , Turbomachinery Ba lancing Considerat ions, Pr oceedi ngs of the Twentieth Tu r-bomachin ery Sym posium,

Turboma chinery L abora tory, Texas A&M U niversity, College Sta tion,Texa s (Sept emb er 1991), pp. 209-214.

Fcent M ass Rad i us RPM4 000,---------------

2

=


5/64

Mass Unbalance 399

Thus, if the speed is doubled, the centrifugal force will be increased by a

factor of four. If the mechanical system is totally linear, the observed rotor vibra-

tion will a lso exhibit a fourfold increase in ma gnitude. For units such a s t urbines

or compressors that operate above a translational crit ical speed, the shaft

rota tes a bout the ma ss centerline (principal inertia a xis). This beha vior produces

a plateau region of constant 1X amplitude and phase as shown in Fig. 9-2. Thus,

moderat e changes in speed (within the plat eau r egion) will have m inimal effect

upon the synchronous vibra tion vectors.

When a machine is operating at full speed and load, a step change in the

runn ing speed vectors (amplitud es an d/or phase) may be indica tive of a ma ss

unbala nce shift . Par t of a blade shroud or other minor a tta chment w ill manifest

as a 1X vector chan ge in a ccordance with the defl ected mode shape. Major unba l-

ance changes such as a blade loss will also produce a running speed vector

change. In th is case, the bala nce cha nge will be significa nt, a nd concern should

be placed on how ar e we goin g to shu td own th is machin e, and pass th r ough thecri t i cal w ith out causing fur ther d amage. In this type of situation, a rapid coast-down is desirable to minimize t ime within the bandwidth of the crit ical speed

domain. It might be a dvisable to bring the ma chinery tra in down under full loadto slow it down a s fast as possible.

Typically, a pure ma ss unba lance problem w ill appear as a forwa rd a nd cir-

cular shaft orbit. The orbit could also appear elliptical if the machine contains a

significant difference in vertical versus horizontal stiffness, or if the rotor is sub-

Fig. 92 Typical MassUnbalance Response AsDescribed By A Calcu-lated Bode Plot

180

150

120

90

60

30

0

PhaseLag(Degrees)

=0.2

Low Damping, =0.1

=0.5

=2.0

=1.0

High Damping,=2.0

=0.1

0

1

2

3

4

5

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Amp

litude

Ra

tio

Critical Speed Frequency Ratio

Low Damping, =0.1

=0.2

=0.5

=2.0=1.0

()


6/64

400 Chapter-9

jected to a sha ft preload in a ddit ion t o the unba lance. Fina lly, the phase r elation-

ship across the rotor will be in accordance with the deflected mode shape, and

the location of the probes along the axis. It is easy to be confused by traditional

rules tha t say t hings like identi cal ph ase angl es acr oss a machi ne are repr esen-tat iv e of mass un bal ance. This statement is only true for a specific set of condi-tions, for a particular group of machines. Many cases of mass unbalance (e.g.,

turbine generator case history 39) will exhibit a phase relat ionship other tha n a

pure inphase motion across the machine.

It is highly recommended tha t t he diagnostician become intima tely familiar

with the specific t opics of shaft mode shapes in chapter 3, dynamic signal cha ra c-

terist ics cha pter 5, and rotor bala ncing cha pter 11 before at tempting to dia gnose

a mass unbalance problem. Furthermore, the study of 1X synchronous behavior

of a rotat ing system w ill reveal importa nt informat ion on the specific chara cter-

istics of th e ma chine. This ty pe of informat ion will provide signifi cant benefi ts in

ma lfunction diagnosis of the rota ting equipment.

B

ENT

OR

B

OWED

S

HAFT

Bent rotors and shaft bows represent another major class of synchronous

1X motion. It was previously mentioned that all machine parts contain some

fi nite am ount of residual unba lance. In a similar ma nner, all assembled horizon-

tal rotors (and some vertical rotors), will exhibit varying degrees of rotor bows.

In some cases, such a s a light weight pinion w ith a short bearing span, the mid-

span deflection will be minimal. In other cases, the rotor will deform due to grav-

ity. That is, the rotor will bend or bow under the influence of its own weight. An

example of this type of rotor would be a large gas turbine or steam turbine rotor.

These types of rotors will display a change in the gravitat ional bow by simply

lifting the rotor off a set of rollers, and setting it back down.

The shaft bow m ay be purely a gra vitat iona l bow, or it ma y be a t hermally

induced bow. In either case, the force associat ed with the bent sh aft is equal t o

the shaft st iffness t imes the init ial bow radius. This is considerably different

from t he ma ss unba lance case, where th e init ial force is equal t o the product of

residual unba lance mass, ra dius, and speed squared. The bent sha ft w ill exhibit

a va riable speed cha ra cterist ic tha t is similar t o the diagram in Fig. 9-3 that wa s

extracted from Fig. 2-18. This calculated plot for a forced spring-mass-damper

syst em wa s ba sed upon the application of a const a nt force. This is identica l to the

bent rotor condition where the applied force is equal to the deflection times the

shaft spring constant as shown in equation (9-4).

(9-4)

where:

F

bow

= Applied Force Due to Shaft Bow (Pounds)

Deflection

= Maximum Midspan Deflection Due to Shaft Bow (Mils)

Shaft Stiffness

= Lateral Shaft Stiffness (Pounds / Inch)

FbowD ef l ect i on S ha ft St i f f ness

1 000,

----------------------------------------------------------------------------------=


7/64

Bent or Bowed Shaft 401

Since shaft st iffness is usually a very large number, a m odera te deflection

(i.e., bow) will result in a substantial radial force. From Fig. 9-3, it is noted that

at speeds well below the resonance, the vibration response is equal to shaft bow.

Hence, a symmetrical rotor supported between bearings would display a maxi-

mum response at the middle of the rotor. A proximity probe or dial indicator

mounted at this location would reveal a runout vector equal to the bow magni-tud e an d loca tion. At speeds w ell above the resonance, the ma gnit ude of th e bow

would approach zero as t he ma chine tends t o rota te a bout t he ma ss center, which

w ould be equivalent , or m ore precisely, coincident w ith th e rotor bow center.

In the vast majority of cases, a thermal or gravitat ional shaft bow may be

r olled out

by extended operation at slow roll speeds. In most instances the 1X

vectors are monitored, and when minimum and constant runout values are

achieved, the machinery may be safely started with a minimal shaft bow. In

other situations, the severity of the bow is of such a magnitude that the rotor

cannot be stra ightened by slow rolling. Localized heat applicat ion ma y be used to

relieve the bow, or the rotor may be suspended vertically and heated in an oven.

In other cas es th e rotor must be completely scra pped (e.g., ca se hist ory 17).

In a ll cases, the diagnostician must recognize tha t rotor bows ar e inherent

with rotating machinery. Furthermore, the shaft bow may consist of complexcurves instead of a simple catenary. Also, the synchronous force from any shaft

bow will vectorially int eract w ith t he synchronous forces due t o unbala nce or a ny

other 1X forcing function. Hence, the final measured rotational speed vectors

probably include contributions from more than one synchronous excitation.

Fig. 93 Typical LateralResponse Due To A ShaftBow As Described By ACalculated Bode Plot

180

150

120

90

60

30

0

PhaseLag(Degrees)

=0.2

Low Damping, =0.1

=0.5

=2.0

=1.0

High Damping,=2.0

=0.1

0

1

2

3

4

5

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Amp

litud

eRa

tio

Critical Speed Frequency Ratio

Low Damping, =0.1

=0.2

=0.5

=2.0=1.0

()


8/64

402 Chapter-9

Case History 20: Repetitive Steam Turbine Rotor Bow

A 35,000 horsepower stea m t urbine wa s subjected to a n extensive overha ul

that included replacement of the 11 stage rotor. The maintenance work was per-formed with minimal difficulties, but major problems were encountered when a

routine overspeed check could not get past slow roll conditions. Although initial

uncoupled turbine runout vibration levels were quite acceptable, the 1X vectors

significant ly increased a t slow roll. At 1,000 RP M, vibration a mplitudes reached

2.5 and 3.1 Mils,

p-p

a t t he governor a nd exha ust end s respectively. This ma nifes-

tat ion of high vibration was combined with in-phase deflection of the turbine

rotor. Specifically, the governor end horizontal probe had a 261 phase angle,

with 265 displayed by the exhaust end horizontal pickup.

This unusual bowed rotor behavior was repeated on multiple runs covering

a time period of two days. At this point, historical rotor records were examined,

and the following general conclusions and observations were reached:

r

All of the attempted turbine solo runs were aborted due to high rotor vibra-tion. This wa s indicat ed by the contr ol room monitoring inst rument a tion, as

well as the physical sensation of excessive deck vibration.

r

During the occurrence of the high turbine vibration, a pure translational

(inphase) shaft bow was clearly evident across the rotor.

r

For th is particular t urbine, sha ft vibra tion amplitudes should rema in essen-

tia lly consta nt between 300 and 1,400 RP M.

r

Init ia l appeara nce of the sha ft bow wa s independent of speed.

r

Init ia l appeara nce of the bow wa s a function of t ime and t empera ture. Spe-

cifically, the shaft bow appeared following approximately two hours of oper-

at ion in a wa rm (>200F) casing.

r

Cold sha ft run out vectors w ere repeata ble, and consideration of any cracked

sha ft ma lfunction was discontinued.r

Rotor inspection records revealed acceptable runout along the length of the

turbine. Furthermore, the shop balance was performed to low levels of

residual unba lance. Hence, the observed behavior w as not a ssociat ed with a

cold rotor bow, or a ma ss un ba lan ce problem.

r

Records revealed tha t t he spare turbine rotor insta lled during t his overha ul

had not operated in the turbine casing for five years. At that point in his-

tory, the plan t suffered a cat ast rophic fire, and t his spare rotor was essen-

t ia l ly baked

in the casing for several day s as the fi re burned itself out.

A turbine rotor is ma chined from a solid forging, and all w heels a nd t hrust

collars are integral with the shaft . This type of rotor assembly is heat treated

an d tempered a s part of the manufa cturing process. These heat treat ments of the

a lloy forging are performed to obtain specific mechan ical properties, a nd th ey are

implemented by controlled heating and cooling of the rotor. Hence, the turbine

rotor is constructed of a steel alloy that was subjected to various heat cycles dur-

ing fabrication. It is reasona ble to believe tha t such a n a ssembly might be sensi-


9/64

Bent or Bowed Shaft 403

t ive to the heating an d cooling an omalies associat ed with the previous plant fi re.

B ased upon these measurements and observa tions, it wa s reasoned tha t the

turbine rotor contained a residual stress that was probably inflicted during the

fire. This residual stress manifested as a shaft bow whenever the rotor washeat ed above ambient temperature. Conversely, the shaft bow w as not apparent

when the rotor was cold. It is logical to assume that the emergency shutdown

produced a therma l bow a s th e hot rotor rested betw een bearings, on the inter-

stage labyrinths. As the fire subsided, the turbine cooled over a period of several

days. It was postulated that as the rotor cooled, it returned to a straight condi-

tion at ambient temperature, and locked i n

the residual stress from the therma l

bow. Since all rotor repa irs, inspections, runout checks, a nd sh op ba lan cing were

performed with a cold rotor, this type of internal stress would be undetectable.

If this hypothesis was accurate, then correction of the residual thermal

shaft bow would require addit ional heat treatment, combined with continuous

rotation of the turbine rotor. Obviously, this type of repair would be difficult to

perform in most shop repair facilities. However, the turbine casing provided a

means t o heat the rotor wit h inlet steam, plus the ability to turn th e rotor at con-

tr olled speed. Thus, th e opportunit y existed to perform a n on-line st ress r elief of

the rotor within the a ctual t urbine casing. The ASM I nterna tional defi nes stress

relieving as: H eati ng to a suit able temper atu r e, hold in g long enough t o redu cer esidu al str esses, and t hen cool in g sl owl y enough t o mi ni m ize th e devel opment ofnew r esid ual str esses

.

From th is common defi nition, both h eat ing a nd cooling must be combined to

stress relieve the rotor. During original manufacturing of this rotor, the heat

treatment temperatures are quite high. It must be recognized that the exhaust

casing has much lower temperature limits (circa 300F). Thus, any field stress

relieving of the rotor in the t urbine casing must be limited in t he heat soak tem-

perature. Based upon the mechanical parameters of the installed turbine sys-

tem, t he follow ing on-line rotor pseudo-str ess r elieving procedure w a s d eveloped:

1.

Operate the turbine at a slow speed of nominally 300 RPM for approxi-

ma tely 30 minutes w ith t he sealing steam off, an d a cool turbine casing.

This is the cooling portion of the cycle.

2.

Apply sha ft sea ling steam , and allow t urbine speed to increase t o approxi-

ma tely 500 RP M with the improved vacuum.

3.

Increase speed, an d monitor the ra dial sha ft vibrat ion a t both journa l bear-

ings. Continue to increase speed until t he unfi ltered radia l vibration am pli-

tud es approach a ma ximum of 4.0 to 5.0 Mils,

p-p

(ba sed on conservat ive use

of the 12 to 14 Mil diam etrical t urbine bear ing clea ra nce).

4.

Operat e the tur bine at this h eat soak condition for a pproximately 60 min-

utes. During this t ime period, the rotational speed and st eam fl ow shouldbe adjusted to mainta in a maximum unfi ltered ra dia l vibra t ion a mpli tude

betw een 4.0 a nd 5.0 Mils,

p-p

, combined with a ma ximum exha ust casing

tempera tur e of 250F, plus a ma ximum speed of 1,400 RP M.

5.

Following 60 minutes of high speed, high temperature, and high vibration


10/64

404 Chapter-9

opera tion; the tur bine speed should be reduced back to 500 RP M. Seal

steam should t hen be removed, and rota tional speed returned ba ck to 300

RP M for a r epea t of the cooling cycle, Step 1.

6.

The previous St eps 1 thr ough 5 should be repeat ed until th e shaft vibra tion

am plitudes rema in essentially constant between 300 an d 1,400 RPM. When

this consistency of slow roll vibrat ion da ta is a chieved, the turbine ma y be

sta rted, and operated norma lly.

This procedure wa s implemented, and t he fi rst 60 minute hot run w as lim-

ited to 654 RPM. A total of six additional cold to hot runs were completed, and

the results of these consecutive runs are summarized in Fig. 9-4. This diagram

consists of 1X radia l vibrat ion a mplitudes measured a t t he end of each cold 30

minute run, plus each hot 60 minute run. Note tha t t hese 1X filt ered amplitudes

are slightly less than the unfiltered, overall vibration levels mentioned in the

procedure. For simplificat ion, only the horizonta l probes at g overnor a nd exha ust

bearings are shown. The vertical probes exhibited identical characteristics.

Within Fig. 9-4, the 1X vibration amplitudes at the governor journal are

depicted by t he solid lines and circular plott ing sym bols. The exhaus t end a mpli-

tudes are defined by dotted lines and square plott ing symbols. Data points

acquired at the end of a cold run a t 300 RP M a re identifi ed by the open plott ing

symbols. The 1X vibrat ion a mplitudes measured a t t he end of a hot r un a re rep-

resented by the solid symbols. For each of the hot runs, the rotational speed at

the end of the run is listed for each pair of hot dat a points. From this summa ry

diagram, it is apparent that the maximum attainable speed during each hot runsuccessively increased from run t o run. In add it ion, the hot vibration a mplitudes

across the turbine tra cked up and down in unison.

A significant portion of this plot is noted in the lower right hand corner.

Within this region, the measured amplitudes at the conclusion of the seventh

Fig. 94 Variation Of 1XVibration Amplitudes Dur-ing Multiple Heating andCooling Cycles

J

J

J

J

J

J

J

J

B

B

B

B

B

B

B

B

1 2 3 4 5 6 7 80

1

2

3

4

5

1XShaftVibration(Mils,p-p)

Consecutive Run Number

Gov Cool

Exh Cool

J

Gov Hot

B

Exh Hot

702Rpm

740Rpm

785Rpm

946Rpm

1,2

03Rpm

1,508Rpm

980Rpm

654Rpm

Slow Speed at 300 Rpm


11/64


12/64

406 Chapter-9

E

CCENTRICITY

Eccentricity of one ma chine part with respect to a nother represents a less

common cat egory of rota tiona l speed excita tions. Norma lly, shaft s a nd most rotorelements are ground on-centers. The material center is thereby concentric with

the initial center of rotation, and eccentricity is generally not a problem. How-

ever, there are occasions when a machine part is bored off-center. Although the

ma jority of the rota ting a ssembly may be stra ight a nd concentric, the presence of

an eccentric element can impose a significant rotational speed force.

If the eccentric element is a minor part of the rotor assembly, the resultant

1X forces may be insignificant compared to the other active synchronous forces.

However, if the eccentric element represents a substantial portion of the rotating

a ssembly, or if it is loca ted a t a moda lly sensit ive locat ion (e.g., th e coupling hub),

then the eccentricity may be a problem. The actual forces associated with an

eccentric element m ay be determin ed from t he followin g equa tion (9-5):

(9-5)

where:

F

ecc

= Radial Force Due to Eccentricity of a Mechanical Element (Pounds)

Weight

element

= Weight of the Eccentric Mechanical Element (Pounds)

Eccentricity

= Radial Eccentricity of Machine Element (Mils)

RPM

= Shaft Rotational Speed (Revolutions / Minute)

La rge ma chine elements or high rotat iona l speeds a re the m ost susceptible

to high forces due to an eccentric element. In many respects, an eccentric ele-

ment appears similar to a shaft bow at low rotational speeds. Both mechanisms

provide large shaft displacement amplitudes at slow speeds. However, the forces

from a bowed rotor may remain constant at all speeds in accordance with equa-

tion (9-4). The radial forces from an eccentric element will vary with the speedsquared as described by expression (9-5). Naturally this all becomes much more

complicated when machines with flexible rotors and multiple mode shapes are

discussed. In a ll cases, eccentric ma chine elements on a r otor should be avoided,

a nd one source of potent ial s ynchronous excita tion removed from considera tion.

From a detection standpoint , shaft bows and eccentric elements can be

determined in the shop with accurate runout checks as described by J ohn E ast

3

.

Once the ma chine is a ssembled, runouts can be detected at low speeds w ith rela-

tive sha ft sensin g proximity probes. Ca sing velocity coils and a ccelerometers w ill

probably not detect either mechanism at slow roll speeds. However, the casing

vibration transducers will pick-up the influence of a bow or an eccentricity at

higher speeds when the radial forces are significant. For instance, case history

21 considers a situation where a pinion coupling hub was bored off-center, and

the resulta nt eccentricity ha d a considerable influence upon t he ma chinery.

3 J ohn R. Ea st , Turbomachinery Ba lancing Considerat ions, Pr oceedi ngs of the Twentieth Tu r-bomachin ery Sym posium, Turboma chinery L abora tory, Texas A&M U niversity, College Sta tion,Texa s (Sept emb er 1991), pp. 209-214.

Fecc Weigh te lement Eccen t r i c i t y RPM5 930,---------------

2=


13/64

Eccentricity 407

Case History 21: Seven Element Gear Box Coupling Bore

The machinery train in question consists of a lime drying kiln driven by a

variable speed synchronous motor through a seven element speed reducing gearbox. The drive portion of this train is depicted in Fig. 9-6. The kiln dryer itself

consisted of a long cylindrical tube that is mounted on rollers, and it is inclined

with respect to grade. Wet lime is loaded into the top end of the kiln, and dry

lime is extracted from the lowest elevation. This massive cylindrical kiln rotates

slowly, with a ma ximum speed of 1.8 revolutions per minu te.

The kiln drive system is located on an elevated platform. Initial operationrevealed una cceptable vibration am plitudes on the plat form, an d va rious st ruc-

tural modifications were implemented. Following these support improvements,

the drive train continued to exhibit uncomfortable vibration amplitudes. The

machinery was not equipped with any vibration monitoring instrumentation,

an d init ia l readings were limited to casing m easurements. The large seven ele-

ment gea r box exhibited low vibrat ion a mplitudes, and the problem appeared to

be confined to the variable speed motor. The outboard motor bearing displayed

casing vibra tion levels of 3.0 to 4.0 Mils,p-p. At the in board, or coupling end motor

bearing, the unfiltered horizontal motion varied between 4.0 to 5.0 Mils, p-p.

I t w as clear t ha t the kiln and the gear box ar e massive structures, and an y

casing vibration would be substantially suppressed by their respective steel

structures. Conversely, the motor driver consisted of a fairly light frame, and itwas susceptible to excitation from a variety of sources. A complete survey was

performed with tr iaxial casing velocity measurements at each main bearing,

plus various locations on t he support st ructure. It wa s quite evident th at the pri-

mary excitation frequency occurred at motor running speed. Axial vibration of

Fig. 96 MachineryArrangement For LargeKiln Drive

SevenElementSpeed

ReducingGear Box

First Pinion

Last Bull Gear

SynchronousMotor

3 - 50 to 500 HP120 to 1,200 Rpm

Output to Ring Gearon Kiln OD

1.8 Rpm Maximum

High Casing Vibrationon Motor IB


14/64

408 Chapter-9

the motor, gear box, an d st ructure wa s minima l a t a ll locations. The ma jor vibra -

tion appeared to be running speed motion of the motor in a radial direction. The

casing orbits presented in Fig. 9-7 depict the unfiltered and 1X filtered behavior

at the motor inboard bearing housing. Operating at an average speed of 1,190RP M, the motor coupling end bear ing housing exhibited a forwa rd a nd elliptical

orbit at motor running frequency. As noted in the unfiltered plots, the maximum

horizontal motor casing vibration approached 4.0 Mils,p-p. Simultaneously, the

compa nion casing orbit on t he pinion input bear ing housing (not shown) revealed

casing displacement amplitudes of less than 0.5 Mils,p-p.

Further investigation, and visual inspection revealed excessive vibration of

the coupling assembly. Since the normal threshold for visual observation of

vibrating surfaces is in the vicinity of 10 Mils, p-p, i t was considered significant

tha t t he coupling could be vibrating tw ice as much as the motor bearing housing.

This observat ion prompted furth er investiga tion into the abs olute motion of

the coupling. To implement t his mea surement, a separat e sta nd w as constructed

from steel plate and angle iron to support a pair of X-Y proximity probes on

either side of the coupling. One pair of probes observed the 8.25 inch diameter

coupling hub on the motor shaft. The other pair of proximity probes were posi-

tioned over t he 9.75 inch dia meter pinion coupling hu b. All four proximity probes

were su pplement ed by velocity coils to a llow for potent ial correction of th e sha ft

vibration signals for motion of the probe supporting structure. As it turned out,this correction was not required, and the direct probe signals from the pinion

coupling hu b a re presented in F ig. 9-8.

This data reveals a dist inctive forward circular pattern to the pinion cou-

pling hub displacement. Since the hub outer diam eter wa s not m achined or pre-

Fig. 97 Motor Bearing Housing Casing

Vibration At Full Operating Speed

Fig. 98 Pinion Relative Shaft Vibration At

Full Operating Speed


15/64

Eccentricity 409

pared for proximity probes the hub surface imperfections were visible, and

overall a mplitudes in th e vicinity of 12.0 Mils,p-p were document ed. The hu b sig-

na l fi ltered a t t he rotat ional speed (1X) of 1,190 RP M exhibited an a vera ge circu-

lar a mplitud e of 10.0 Mils,p-p. I t sh ould a lso be mentioned tha t t he coupling hub

on the motor side displayed circular amplitudes of nominally 5.0 Mils, p-p. This

steady state data at full speed suggested that the high vibration might be origi-

nating at the pinion instead of the motor. In support of this preliminary conclu-

sion, it wa s understood th at uncoupled, and un loaded, motor vibrat ion w as quite

low. That does not necessarily give the motor a clean bill of health, since many

motor problems only appear under load. Nevertheless, it did suggest that per-

ha ps the pinion might be the culprit , an d th e light motor might be just respond-

ing to a forced vibra tion condit ion.

Additional perspective on t his problem wa s ga ined by th e a cquisit ion an d

an alysis of va riable speed informat ion. Specifica lly, vibration da ta wa s recorded

durin g a sh utd own of the kiln, and t he B ode plot show n in Fig. 9-9 obta ined from

th e X-Y proximit y probes positioned over th e pinion coupling h ub. Note tha t t hesynchronous 1X amplitudes and phase angles remained essentially constant

from th e top speed of 1,190 RP M to t he minim um s a mple point of 186 RPM . This

type of behavior is certainly representative of an eccentric mechanical element.

In this case, the pinion coupling hub wa s th e primary suspect .

Another perspective of the kiln shutdown was gained from Fig. 9-10, that

documents a t ime history plot of the coast down. In th is diagra m, the 1X ampli-

tude and phase are plotted against t ime from 0 to 60 seconds. In addit ion, the

Fig. 99 Bode Plot Of Pinion Vertical HubVibration During A Typical Coastdown

Fig. 910 Time History Coastdown FromVertical Prox Probe On Pinion Coupling


16/64

410 Chapter-9

rotative speed is included at the top of the same plot. The various humps in this

summary plot are due to the fact that the main kiln cylinder carries a tremen-

dous amount of inertia from the rotating kiln, plus the internal lime. During a

routine coa std own of this ma chine, th e kiln slows down fr om full opera ting speedof 1,190 RPM, and comes to a stop in approximately 6 seconds. The kiln tube

(with the moving lime) then begins a reverse rotation, and drives the gear box

and motor in a reverse rotation. The input pinion and motor reach a peak speed

of 800 RPM before the train starts to slow down. As shown in Fig. 9-10, the unit

experiences two more forward, and one more reverse rotation sequence before

the tra in comes to a fi na l stop.

Note that the rotational speed amplitude peaks at about 8.0 Mils, p-p irre-

spective of a forward or a reverse cycle. When the unit is rotating in a forward

direction, the vertical probe phase angle is approximately 250. During reverse

rota tion, the 1X pha se an gle is in the vicinity of 80. Thus, a n ominal 170 rever-

sal in the high spot occurs as the pinion hub rotates in a forward or a reverse

direction. This is close enough to 180 to conclude that there was a complete

reversal of the phase relationship between forward and reverse rotation. This

documented behavior also helps to substantiate the hypothesis of an eccentric

coupling hub.

Additiona l dat a at var ious loads provided no other useful informa tion, and

it was finally concluded that the coupling hub was bored off center. The physical

configuration of the pinion extension did not allow simultaneous dial indicator

mea surement s of th e pinion hub versus th e pinion sha ft. However, wh en the cou-

pling hub w as removed, it wa s determined tha t t he shaft bore wa s indeed off cen-

ter by approximately 8 to 10 Mils. Naturally the coupling supplier was slightly

embarrassed, and they provided a concentrically bored coupling assembly in a

short period of time. The inst a llat ion of th is correctly bored coupling ha lf on the

pinion sha ft solved the problem.

SHAFT PRELOADS

Another category of potentia l ma lfunctions tha t a re genera lly applicable to

all rotating machinery is the topic of shaft preloads. The presence of various

types of unidirectional forces acting upon the rotating mechanical system is a

normal a nd expected cha racterist ic of ma chinery. J ust a s residual unba lance,

rotor bows, and component eccentricity are inherent with the a ssembly of rotat -

ing elements, the presence of shaft preloads are an unavoidable part of assem-

bled mechanical equipment.

From an init ial categorization standpoint , shaft preloads may be divided

into two fundamental groups. The first group would address the preloads that

originat e with in th e ma chinery. These internal preloads ma y be due to a ny or a llof the following common mechanis ms:


17/64

Shaft Preloads 411

r G ravita t ional P reloads

r B ear ing Preloads

r Internal Misalignment Preloads

r Gear Mesh Forces

r Fluid Preloads

Gravitational preloads on horizontal rotors are responsible for the rotorbow or sag that was discussed earlier in this chapter. Again, this is part of the

normal and unavoidable characteristics of process machinery. This is generally

not a desi gn cor r ectabl eproblem. It is a physical phenomena t ha t must a lway s beconsidered, and d ealt w ith on a ma chine by ma chine basis.

Bearingpreloa ds represent one of th e ma chinery design considera tions. Asdiscussed in chapter 4, bearing preloads are typically expressed as a non-dimen-

sional nu mber betw een 0 a nd 1. A bear ing preloa d of 0 indicat es no bea ring load

upon the shaft . Conversely, a bearing preload of 1 indicates a shaft to bearing

line conta ct (i.e., ma ximum preloa d of 1). The computa tion of bear ing preload is

based upon the difference in curvature between the shaft and the individual

bear ing pa d. As expressed by equa tion (4-6), bearing preloa d is d etermined by :

(9-6)

Where Cb is the bearing clearance which is equal to the bearing radius

minus t he journa l ra dius. The pad cleara nce Cpis equal to the pad r adius minus

the journal radius. On segmented bearings it is common to find that the pad

ra dius is greater t ha n th e bear ing ra dius. Thus, the pad cleara nce Cpis greater

than bearing clearance Cb, and their ratio is less than one. From equation (9-6)

the bearing preload must therefore be less than 1. Typically, the bearing design-

ers will employ preloads tha t va ry betw een 0.1 and 0.4.

Clearance reduction at the center of the pad forces the oil to converge into

an oil wedge. The operational characteristics of bearing stiffness, damping, and

eccentricity posit ion w ill vary w ith t he preload, a nd th e actua l bearing confi gura-

tion. Obviously, the characteristics of a fixed pad bearing, such as an elliptical

bearing, will be different from a five shoe tilting pad bearing. However, in either

case an oil wedge will be developed, and tha t oil wedge will have a n a ssociat ed

pressu re profi le (Fig. 9-20). The direct a ction of this pr essur e profile upon t he

journal is a preload force. Again, this is part of the normal behavior of the rotat-

ing machine, but the amount of preload, and the associated journal force is

ad justa ble (with in limits) by va rying the pa d ra dius an d bearing geometry.

The third type of interna l ma chine preload is a t t r ibuted to internal mis-

alignment of elements. This ca n va ry fr om offset or cocked sea ls or bear ings, todistorted diaphragms or stators, to a variety of rub situations. It is virtuallyimpossible to quantify all of the potential combinations of misaligned internal

ma chine elements. However, the common chara cterist ic tha t t hey all sha re is the

generation of a load or force against the shaft . Some of these preloads may

Pre l oad 1CbCp-------

=


18/64

412 Chapter-9

relieve themselves during normal operation. For instance, a laby seal rub might

occur during init ia l sta rtup on a ma chine, an d th e expanded sha ft to seal clear-

an ces ma y never rub aga in. Other preload s such as a distorted stat or will gener-

a lly rema in const a nt , and w ill continue to provide a force upon th e rotor.The fourth type of internal preload is associated with gear mesh forces.

These are signifi cant load s th at must alw ays be considered. To demonstra te the

magnitude of gear contact forces, the values calculated in case history 24 are

repeated. These loads a re for a simple pinion - bull gear a rra ngement . The forces

were computed for a transmitted load of 4,000 horsepower, a pinion speed of

5,900 RP M, and a bull gear s peed of 1,920 RP M. The signifi cant element w eights

an d forces for t his gear box ar e summa rized as follows:

P inion Weight ......................... 220 Pound s

B ull Gea r Weight ................. 1,630 Pounds

Sepa ra tion Force.................. 4,550 Pounds

Ta ngent ial Force................ 10,880 Pounds

From this summ ar y, it is clear th at the ma jor forces wit hin a gear box are

the gear contact forces. The magnitudes of the separation and the tangential

forces place the gear weights into the role of a secondary influence. These gear

forces are used in the development of gear box bearing coefficients, and in the

init ial estima tion of the journa l running posit ion. It is importa nt to consider the

journal operating position during alignment of the gear box, and recognize that

the bull gear and pinion bearings are subjected to significant radial preloads

from the gea r forces.

The fifth type of fluid preloads is applicable to many types of rotatingmachines. For instance, the unbalanced radial force in a volute pump is an obvi-

ous case of fluid forces acting directly upon a rotor. A less obvious example of

fl uid forces w ould be the behavior of a multista ge and multilevel turbine duringstar t up. I t ha s been documented tha t par t ia l steam a dmission to the fi rst sta ge

nozzles may cause a lift ing force on the rotor when the first nozzle segment is

located in the bottom half of the turbine casing. This radial force may be suffi-

cient to lift the rotor, and allow the governor end bearing to go unstable. Hence,

the vertical shaft preload would work against the stabilizing gravitat ional force

to drive the ma chine into an other type of ma lfunction. In a ll cases, the ma chin-

ery diagnostician must be awa re of these types of physical interactions, an d must

strive to understand and address the fundamental forces behind the observed

vibratory motion.

The second ma jor cat egory of shaft preloa ds considers t he ar ra y of potent ial

external preloads or forces. For example, the following short list identifies some

of the common external shaft or machinery preloads:

r Coupling Misalignment

r Locked Coupling

r Therma l or Externa l Forces


19/64

Shaft Preloads 413

The problem of coupling misalignment is common to most types of rotat-ing machinery. Fortunately, the machinery community has devoted considerable

tim e an d effort to develop solutions an d t echniq ues for execution of correct s ha ft

alignment. In fact , many operating facilit ies have applied these tools and tech-niques to develop very successful machinery alignment programs. In these oper-

ating plants, misalignment has ceased to be a problem. Nevertheless, when

misalignment between ma chines is present, the sha ft preload forces ma y be sub-

sta ntia l, an d may result in prema ture mecha nical failure.

A related mechanism to shaft misalignment is the problem of a lockedcoupling. Primarily this occurs on oil lubricated gear type couplings. Sincethese couplings are designed as flexible members with a tolerance to misalign-

ment, the abnormal condition of locked coupling teeth will violate the intended

behav ior of th e flexible design. A locked coupling ma y beha ve simila rly t o a mis-

aligned coupling. In may cases the coupling is locked due to excessive misalign-

ment. In other cases, the locked coupling may develop during operation due to

the a ccumulat ion of sludge betw een t he teeth. In either case, the resulta nt forces

on both machinery shafts are unwelcome preloads that may damage machinery

components on either side of the locked coupling.

The th ird ca tegory of external preloads is associated with the presence ofa ny n umber of potent ial extern a l forces or moments on th e ma chinery. This kind

of preloa d could be due to ba seplat e stra in imposed by a d egra ding grout or foun-

da tion. External preloads could a lso arise from piping stra in upon the ma chine.

For instance, the case history 49 describes the effect of a piping moment upon a

compressor, and the coupled turbine driver. External preloads may influence the

coupling alignment, or they may distort bearing housings, casings, or other

mechanical at tachments.

In essence, shaft preloads are a normal part of rotating equipment that

must be addressed. It is useful to recognize that preloads have different levels of

severity. For instance, some preloads such as gravity or fluid based forces may beclassified as softpreloads that are generally non-destructive. Other preloads,such as misalignment or gear contact forces, may be considered a s hardpreloadsthat can be damaging to the machinery. A third severity classification for shaft

preloads would be the destabi l iz ingva riety. This t ype of preloa d m a y oppose t henorma l rotor or bea ring forces, and it ma y a ct to desta bilize the rotor. The sever-

ity of these destabi l iz ingpreloads ma y a lso vary from softt o hard, depending onthe final influence upon the process machinery.

Preload detection is predicated upon the recognition of abnormalities in

radial vibration. The following three characteristics are used to identify the pres-

ence of ra dial sha ft preloads:

r Normal Orbita l Motion

r Abnorma l Sha ft C enterline Posit ionr Abnorma l Sha ft versus Ca sing Motion

Abnormal orbital m otion is d emonstra ted in F ig. 9-11, tha t d epicts a n a rra y

of four shaft orbits with different levels of radial preloads. The orbit in Sketch A


20/64

414 Chapter-9

displays a normal forwa rd circular patt ern w ith no sha ft preload . Applicat ion of

a slight preload to the shaft in Sketch B forces an elliptical shape to the shaft

motion. The moderate preload of Sketch C drives the orbit into a bananashape.Finally, the heavy preload case shown in Sketch D results in a Figur e 8orbit.This type of analysis is quite appropriate to machines with fluid film bearings,

an d sha ft sensing proximity probes. In th e vast ma jority of the cases, the direc-

tion of th e preload is perpendicula r to the ma jor a xis, a nd t he severity of the pre-

load is dependent on the ellipticity of the orbit.

As a cautionary note, the machinery diagnostician should not confuse the

pr eloaded orbitwith the normal ly el l ipt i cal orbi t. Machines with significant dif-ferences between vertical a nd horizonta l bearing st iffness will nat ura lly display

a n elliptical orbit. This t ype of orbit is st rictly a function of the bear ing geometry,

an d must be considered a s norma l a nd proper behavior.For example, Fig. 9-12 displays a diagram of an elliptical bearing with a

counterclockwise rotating shaft. The normal elliptical orbit is noted in a proper

running posit ion in the lower right hand quadrant. This same diagram also

Fig. 911 Changes InShaft Orbits With Increas-ing Levels of Radial ShaftPreloads

Fig. 912 Normal VersusAbnormal Shaft CenterlinePosition In An EllipticalBearing Assembly

Sketch D - Heavy Preload

Sketch C - Moderate PreloadSketch A - No Preload

Sketch B - Slight Preload

45 45

Vertical Probe Horizontal Probe

Vertical

Clearance

Horizontal

Clearance

NormalOrbit

PositionCCW

Rotation

AbnormalOrbitPosition


21/64

Shaft Preloads 415

depicts another elliptical orbit residing in the lower left hand quadrant. The sec-

ond orbit is obviously in the wrong place for a CCW rotating machine. This

improper radial position represents the second method of detection of shaft pre-

loads. That is, the calculated journal centerline position should be in the properlocation within the clearance of the bearing. If the shaft centerline position

resides at an abnormal location, the possibility of a damaged bearing, or the

presence of a ra dia l preloa d should be suspected.

Three ca utiona ry notes should be added to this t ype of evalua tion. First, the

diagnostician must know the proper running posit ion of the journal within the

specific bearing before attempting to pass judgment on any field data. For

insta nce, a fi ve shoe t ilt ing pad bear ing will display a vertical a t t itude angle, an d

normal position for this type of bearing is considerably different from the previ-

ously discussed elliptical bearing. If the eccentricity position and the attitude

an gle are not known, then th e diagnostician should consider an FE A ana lysis of

the specific bearing confi gurat ion a s discussed in chapter 4.

Second, the ma chinery diagn ostician must be working w ith a properly cali-

brated proximity probe system to measure the true running position of the jour-

nal within the bearing. This includes accurate probe scale factors plus a correct

probe orientat ion dia gram . In some instan ces, it may be necessary t o install four

ra dial proximity probes at 90 increments. In t his a pplicat ion, the dia metrically

opposed probes are summed to determine an average shaft position in each

orthogonal direction. This is more work, but it does enhance accuracy of the

ra dial posit ion measurement.

Third, the init ial proximity probe DC gap voltages must be accurately

known to a llow a confi dent calculation of the shaft centerline posit ion. It is diffi-

cult to genera lize on th e precise condit ion to obta in th e at stopga p volta ges. Nor-ma lly, this da ta is obtained prior t o sta rtup w ith w ar m oil circulat ing. To be safe,

it is recommended that DC voltages be tracked with a computer-based system

that will identify t ime, speed, and other useful information such as oil supplytemperature, am bient temperature, etc.

Finally, on machines with accessible bearing housings, it is desirable to

acquire X-Y casing vibration response measurements. The casing probes should

be placed in the sa me an gular orientat ion a s the sha ft sensing proximity probes.

In a ddit ion, th e ca sing probes should be loca ted a s close as possible to the mount -

ing point of the proximity probes. The casing da ta must be integra ted t o displa ce-

ment, and the 1X synchronous vectors compared directly against the runout

compensated shaft displacement 1X vectors. Under normal conditions, the casing

motion should be smaller than the shaft vibration, and the casing phase angles

should lag behind the sha ft vibrat ion a ngles.

For a machine with a radial preload, and a compliant support , the shaft

vibration ma y be suppressed. In this condition, the norma l sha ft vibra tion within

the bearing is tran smitted to the casing a nd surrounding structure. From a mea-surement standpoint , the casing 1X vibration amplitudes may exceed the shaft

motion, and the shaft to casing phase relationship may appear abnormal. This

fi na l criteria is not a totally conclusive test , but it does provide addit ional insight

into th e mecha nics of the ma chinery.


22/64

416 Chapter-9

RESONANT RESPONSE

Machines and structures all contain natural frequencies that are essen-

tia lly a fun ction of stiffness a nd ma ss. As described in previous chapt ers, th e fun-damental relationship may be described by the following expression:

Recall t ha t this expression wa s developed for a simple spring-ma ss syst em,

and it basically identified the lowest order resonant frequency. For more complex

mechanical systems an entire family of resonant responses must be addressed.

For example, consider a turbine compressor set mounted on a mezzanine struc-

ture, and connected with a flexible coupling. The potential array of anticipated

na tura l, or resonant frequencies are summa rized as follows:

r Lateral Crit ical Speeds

Turbine Tra nsla tiona l (1st)

Turbine P ivota l (2nd)

Turbine B ending (3rd)

Compressor Translational (1st)

Compressor Pivotal (2nd)

Compressor Bending (3rd)

r Torsiona l C ritica l S peeds

Turb ine (1st)

Turb ine (2nd)

Compressor (1st)

Compressor (2nd)

r Rotor Element Resonances

Coupling Na tura l Axial

Coupling Lateral

Turbine Blades

Compressor Im pellers

r Acoustic Resonances

Externa l Piping Systems (including stubs an d branches)

Internal P assa ges Within Ca sings

r St ructural Resona nces

P iping Systems

Structura l S teel Systems

Machinery Pedestals

B aseplate, Founda tion, or Ground Support Syst em

This list of potent ia l resonant frequencies ca n be intimida ting. The ra nge of

natural frequencies may vary from 60 CPM (1 Hz) for the foundation and sup-

port systems, to 1,800,000 CPM (30,000 Hz) for turbine blade natural frequen-

N a t u r a l FrequencyS t i f f nessMa s s

--------------------------


23/64

Resonant Response 417

cies. With this extended range of natural frequencies it is somewhat amazing

that process machinery can be designed to operate without violating many of

these system natural frequencies. One of the tools used by machinery designers

is th e Ca mpbell diagr am for plott ing na tura l frequencies versus excitat ions. Thiswas initially introduced in chapter 4. For reference purposes, the Campbell plot

from Fig. 5-4 is reproduced in Fig. 9-13. P lease note t ha t t he na tur a l frequencies

(rotor la tera l modes) are shown a s horizonta l lines off the vertical a xis, and the

va rious machine excita tions a re depicted as t he slan ted lines. These slant ed lines

are the 1X, 2X, 3X, etc. multiples of rotational speed that the machine will pro-

duce as speed increases from slow roll to the norma l operat ing point.

An intersection between curves identifies points of potential resonantresponse. That is, a natural frequency exists, and an excitation source is present

at the same frequency. This coincidence results in a stimulation or activation of

the resonance. The Campbell plot is an interference diagram that is commonly

developed during t he design sta ges of a new m achinery tra in. The a ctual num ber

of system natural frequencies may vary from 20 to 50 or more. Furthermore, the

actua l number of machinery excita t ions may also be significant , and 10 to 30 sig-

nifi cant excitat ions are not unusual.

The general characteristics of each type of resonance must always be con-

sidered. For example, a structural resonance has minimal damping, and it will

exhibit a sharp response with a high Qamplification factor. Resonances of thistype have a narrow bandwidth, and the excitat ion frequency must be precisely

equal to the resonance for any action to occur. On the other hand, rotor lateral

resonances include damping from the bearings and seals, and they are active

over an a ppreciable frequency r a nge (ba ndw idth ). These type of rotor resona nces

may be activated by an excitat ion force that falls anywhere within the band-

width of the damped rotor resonance. The resonant frequency may also change

Fig. 913 Campbell Diagram Describing Interference Between Several Forced MachineryExcitations And A Partial Group of Turbine Natural Lateral Resonances

0

2,000

4,000

6,000

8,000

10,000

12,000

10,000 20,000 30,000 40,000 50,000

Na

tura

lFrequency

(Cyc

les

/Minu

te)

Excitation Frequency (Cycles/Minute)

1st

2nd

3rd

4th

5th

10X

5X

2X

1X

Na

tura

lFrequenc

ies

Exc

ita

tions

Translational - 1st

Pivotal - 2nd

Rotor Bending - 3rd

4th

5th

1X

10X

5X2X


24/64

418 Chapter-9

with ma chine speed. For insta nce, bear ing st iffness an d da mping chara cterist ics

vary with rotational speed. In most cases, the bearing parameters that apply as

the ma chine passes t hrough a crit ical speed ar e different from th e normal oper-

at ing condition. Hence, the sa me resona nce may appear a t different frequenciesdue to the infl uence of other a ssocia ted ma chine elements (e.g., bear ings).

Additionally, the exciting mechanism does not have to occur at rotating

speed, or any even order harmonic. The excitation may occur at a subsynchro-

nous (below running speed), or a supersynchronous (above running speed) fre-

quency. For instance, a ma chine may sta rtup w ith a locked 43% oil whirl

instability. Presumably, the frequency of this instability would increase in direct

proportion to the rotational speed. If the machine operated above twice the criti-

cal speed, the 43%whirl would eventually coincide with the rotor balance reso-

nance during startup. This coincidence between the rotor resonance and the oil

wh irl w ould proba bly result in a re-excita tion of the critical. The oil whirl w ould

evolve into an oil wh ip, with potentia lly da ma ging implicat ions t o the ma chinery.

Due t o the severity of this problem, addit ional discussion a nd explana tion will be

presented lat er in t his chapter.

As discussed th roughout t his text , ma chinery systems exhibit a wide va ri-

ety of nat ural resona nces. In the va st m ajority of cases, these resona nces rema in

dormant, and their presence goes undetected. However, when an excitation does

appear, or w hen the mechanical char acterist ics of the syst em undergo a chan ge

(due to failure or attrition); the idle resonance may become adversely excited.

The solution to these occurrences typically resides in identification of the

changes in physical machine parameters. With respect to resonance problems,

the machinery diagnostician should always examine the mechanical system for

evidence of variation of mass, stiffness, or the application of a new force. When

the va riant is discovered, the solution is close at ha nd.

Finally, there are groups of natural frequencies that may be discounted

from a design standpoint , but they may have to be examined during a detailedma chinery a na lysis. For insta nce, torsiona l resona nces on a turbine driven com-

pressor would probably not be a ma jor caus e for concern. However, if mechanica l

failures indicated the existence of twisting forces, torsional vibration should be

considered. Torsiona l modes ha ve low da mping (high Qs) a nd t hey ha ve caused

ma ny fa ilures on large turbines.

Conversely, torsional resonances on a reciprocating engine would be of sig-

nificant interest during init ial design, and acceptance testing. However, if the

main bearings displayed babbitt failures, the lateral vibration should be evalu-

a ted. Thus, a group of potent ial resonan ces a nd/or excita tions sh ould not be elim-

inated from possibility just because they may not apply. The other extreme of

performing deta iled examina tions on mecha nisms t ha t have n o possible relat ion-

ship to the immediate problem should also be avoided. As always, the machinery

diagnostician must exercise good engineering judgment when selecting or elimi-na ting potential r esona nces for a ma chinery problem.


25/64


Case History 22: Re-Excitation of Compressor Resonance

Man y refrigeration compressors in operating facilit ies such a s a mmonia or

ethylene plan ts are configured w ith m ultiple side stream s plus one or more lev-els of extraction. These refrigeration compressor systems are designed to main-

ta in specifi c flow ra tes an d refrigera nt t empera tures. In ma ny cases, the process

requirements are quite stringent, and the compressor is optimized to operate

w ithin a very limit ed performa nce envelope. These requirement s pose some chal-

lenging system and machinery design considerations that often results in

ma chines that ar e diffi cult to operat e at non-designconditions.For instance, consider a large six stage propylene refrigeration compressor

that draws in excess of 30,000 horsepower at design operating conditions. The

rotor weighs 7,100 pounds, and the normal operating speed for 100%plant load

is 3,860 RPM. Obviously this speed will be subjected to control system adjust-

ment dependent on the actual number of operating cracking furnaces, and the

associat ed refrigerant load. A typical sta rtup of this ma chine is described by the

Bode plot shown in Fig. 9-14, plus the polar plot shown in Fig. 9-15. This tran-sient da ta wa s a cquired from the horizonta l proximity probe at the coupling end

of the compressor. On this machine, the horizontal vibration is normally larger

than the vertical motion, and the discharge bearing vibration is slightly higher

than the outboard suction end bearing. Hence, the data presented in Figs. 9-14

and 9-15 represents the highest vibration amplitudes encountered during a nor-

mal ma chinery t ra in star tup.

This sta rtup wa s quite smooth w ith a clearly defi ned compressor fi rst crit i-

cal speed at 2,020 RPM. The observed critical response range extended from

Fig. 914 Bode Plot Of RefrigerationCompressor Startup To Minimum Governor

Fig. 915 Polar Plot Of Refrigeration Com-pressor Startup To Minimum Governor


26/64

420 Chapter-9

a pproxima tely 1,600 to 2,500 RP M. The peak response wa s somewh a t sh a rp, but

that was attr ibuted to the fast ramp programmed into the electronic governor.

Hence, this tra nsient response wa s not indicative of any mechanical a bnormality.

A complete set of steady state data was acquired twenty-four hours aftersta rtup. The orbit an d t ime doma in plots for the discharge end bea ring a re pre-

sented in Fig. 9-16 at a n a vera ge speed of 3,680 RP M. The compressor orbits ar e

forwa rd, with a n elliptical pat tern a t th e discharge, an d a circular motion at t he

suction (not sh own ). In a ddit ion, a subsynchronous insta bility w as visible at the

dischar ge bearing. The domina nt d irection of this low frequency motion wa s hor-

izontal, and the frequency oscillated between 1,980 and 2,100 CPM. Extended

observation of this subsyn chronous component r evealed t ha t 2,060 CP M w as the

ma jor frequency, and pea k am plitudes rea ched 1.5 Mils,p-p at the discharge hori-

zontal. Time averaged behavior is displayed on the FFT plot shown in Fig. 9-17

with a defined peak at 2,060 CPM, and an average horizontal amplitude of 1.1

Mils,p-p. I t sh ould a lso be mentioned t ha t subsynchronous vibrat ion a mplitudes

at the suction bearing generally remained below 0.25 Mils,p-p.This 2,060 CPM frequency is recognized as the first critical speed of the

compressor rotor. As previously noted, the st art up da ta displayed a tra nslat iona l

balance resonance (first critical) at 2,020 RPM. However, as stiffness characteris-

tics change with speed, journal eccentricity, and temperature, the startup first

critical is generally different from the critical response observed at full speed.

Hence, the 2,060 CP M component is considered t o be a r e-excita tion of th e com-

pressor first critical speed. This phenomena has occurred for over twenty years,

an d previous stu dies have correlated t he compressor inst ability w ith t he extrac-

Fig. 916 Refrigeration CompressorInboard Bearing Orbit And Time Base Plots

Following Normal Train Startup

Fig. 917 Refrigeration CompressorInboard Bearing Spectrum Plots Following

Normal Train Startup


27/64


tion sidestream. Historically, startup of the propylene refrigeration system pro-

duces flow and pressure fluctuations in the extraction line. These fluid based

excita t ions are t ra nsmitt ed to the discharge end of the compressor rotor via t he

horizontal extraction nozzle. The observed shaft motion is primarily horizontal,and the largest excitation occurs at the coupling (discharge) end of the rotor.

Fluid varia t ions in the extra ction strea m provides a broadban d excita t ion to the

compressor rotor, and the first critical speed is generally excited. In addition, the

turbine first critical is also driven due to the close proximity between the com-

pressor and turbine balance resonant speeds. Although the dominant subsyn-

chronous vibration occurs at the compressor discharge bearing, the other

machinery train bearings display low level excitations. However, as the refriger-

ation systems become l i ned out, the extraction stream flow instability dimin-ishes, and excitation of the critical speeds of both rotors normally disappears.

In support of this explana tion, the vibrat ion response dat a acquired with an

average speed of 3,860 RPM, and a full process is shown in Figs. 9-18 and 9-19.

Although 1X runn ing speed vectors ha ve experienced minor cha nges, the subsyn -

chronous motion no longer exists. Examina tion of the t urbine dat a reveals a sim-

ilar a bsence of motion at th e fir st critica l. Hence, the document ed re-excita tion of

the compressor first crit ical speed under unstable extraction flows was elimi-

nated by establishing normal loading of the propylene refrigeration system.

Again, this behavior is totally consistent with the historical behavior of this

machinery t ra in .

Fig. 918 Refrigeration CompressorInboard Bearing Orbit And Time Base Plots

Under Normal Process Load

Fig. 919 Refrigeration CompressorInboard Bearing Spectrum Plots Under

Normal Process Load


28/64

422 Chapter-9

MACHINERY STABILITY

Throughout t his text , the a ttr ibutes of well balanced an d properly aligned

machines operating with concentric rotor elements have been repeatedlyendorsed. Reductions in shaft preloads are generally associated with reduced

forces and extended machinery life. In fact , many machinery problems that

appear to be extraordinarily complex are often beat into submission merely by

corrections to the basic mechanical parameters of balance, alignment, and ele-

ment concentricity. There is an added dividend provided by smooth running

machines in the area of incrementally improved efficiency. In essence, more of

the input energy goes into productive work instead of being wasted on mechani-

cal a bnormalit ies.

However, the uninit iated may be surprised to find that in some situations,

these fundamental corrections may result in an inoperable machine. There are

many documented instances of properly executed balance or alignment correc-

tions tha t ha ve resulted in signifi cantly h igher vibra tion response amplitudes. In

many of these cases, an examination of the vibratory characterist ics hasunveiled the presence of a new frequency component. Often this new vibration

component occurs at frequencies below rotating speed, and this subsynchronous

motion is often a ssociat ed wit h ma chinery inst ability. Although this genera l defi -

nition of instability is not rigorously correct, it is still used throughout most

industria l locations.

The very na ture of centrifuga l ma chinery provides the funda menta l mecha-

nism for this type of behavior. In all cases, it must be recognized that centrifugal

ma chines consist of rotat ing cylinders or disks confi ned wit hin sta t ionary cylin-

ders. If clearances between cylinders are large, there is no possibility for interac-

tion betw een sta t ionary a nd rotat ing parts. For example, a 6 inch diam eter shaft

rotating w ithin a 20 inch diameter annulus w ill function in the same ma nner as

it w ould in free space. However, as clea ra nces decrease, there is increa sed oppor-

tunity for interaction between elements. For instance, if the 6 inch diameter

shaft now rotates inside a 6.008 inch diameter bearing; interaction between

cylindrical elements now exists across the contained fluid. The fluid might be

stea m, a process ga s, a process liquid, oil in a sea l, or oil conta ined wit hin a bear-

ing. The general t ype of behavior for a cylinder rota ting in side of a st a tiona ry cyl-

inder is depicted in Fig. 9-20.

From this diagra m, it is ant icipated tha t the rotat ing element establishes a

minimum running clearance to the stat ionary cylinder. For an oil film bearing,

this clear an ce would normally be identifi ed as the minimum oil fi lm. The active

forces a cross th e minimum oil film include the fl uid ra dial force, plus a t an gen-

tial component. In this simple example, these two forces should be vectorially

equal t o the sha ft load. Thus, the oil fi lm forces are in equilibrium with the sha ft

load. If Fig. 9-20 was representative of a journal and bearing in a horizontalmachine, the shaft load would primarily consist of the shaft weight. Further-

more, the described system w ould exhibit a minimum oil fi lm in t he lower r ight

hand quadrant of the bearing. Due to the counterclockwise rotation, it is intui-

t ive tha t the shaf t would cl imbthe lower right ha nd side of the bea ring. Addi-


29/64


30/64

424 Chapter-9

the ma ximum velocity, or 50% of th e rotat ional frequency. However, in a rea l

world mechanical system, the difference between journal and bearing surface

conditions, the bearing geometry, bearing clearances, and loads will result in an

avera ge oil film velocity th at is somewha t less than a simple 50% arit hmeticavera ge. The velocity profile will not be a st ra ight line rela tionship, and t he aver-

a ge oil fi lm velocity w ill be less th a n 50%. This ma nifests a s a frequency th a t

appears a t s omething less th an 50%of running speed. In pra ctice, ra t ios betw een

35%and 49%are commonly observed and documented.

This is th e essence of a self-excitedmechan ism such as oil whir l. Tha t is, thephysical geometry and loading of the mechanical system allows the establish-

ment of a subsynchronous vibration component in oil lubricated bearing. This

motion is forwa rd, circular, and typically a ppear s betw een 35%and 49%of shaft

rotative frequency. The simplest case is known as oi l wh i r l, and it is generallydetected in ma chines w ith pla in sleeve bearings.

A visual dichotomy is always encountered when oil whirl orbits are

observed on an oscilloscope. Specifically, this is a forward circular mechanism,

but th e Keyphasor dots seem to spin backwar ds a gainst rotat ion. This gives the

appeara nce of a ba ckwa rds motion, and a misinterpreta tion of the dat a is possi-

ble. In a ctuality, the ba ckwa rds spinning K eypha sor dots a re perfectly correct

for a forwa rd subsynchronous vibra tory component. In order t o demonstra te th is

behavior, consider Fig. 9-22 which describes severa l revolution of a sha ft experi-

encing a forwa rd subsynchronous w hirl.

In this diagram of a counterclockwise rotating system, consider the first

Keyphasor dot at posit ion A . Assume a negative going Keyphasor pulse, and

the expected blank-br ightsequence indicates a CCW rotation. As the shaftmakes one complete revolution, the predominant subsynchronous component

ha s only completed a bout 45%of a full cycle (160). At this inst an t in t ime, thesecond Key dot appears a t posit ion B. Note tha t the blank-br ightsequence is

still consistent, and representative of forward CCW motion. As additional cyclesare completed, the K ey dot continues t o lag furt her a nd furt her behind. Thus,

Fig. 922 Keyphasor Dot

Precession For Subsynchro-nous Excitation Occurring AtLess Than 50% Of MachineRotative Speed

A

D

E

F

C

B

CCWShaft

Rotation

K Dots

K Dots Move CW


31/64

Machinery Stability 425

a visua l observa tion on an oscilloscope will reveal th e dots moving from A t o C t o

E. In the second group, the dots move from B t o D t o F. Therefore, the dots

appear t o roll backwards in t ime, but th e blank-br ightsequence specifi cally iden-

tifies a forward orbit .This chara cterist ic can be put t o good adva nta ge when viewing live subsyn-

chronous data in an orbital display on an oscilloscope. That is, when the dots

appear to move in a counter-rotation direction, the subsynchronous component

occurs at less than 50%of running speed. Conversely, when the dots appear to

move in the same direction as shaft rotation, the subsynchronous component

occurs at a frequency that is greater than 50%of rotative speed. The middle con-

dit ion of a pair of fi xed Keypha sor dots indicat es a frequency tha t is locked onto

50%of runn ing speed.

This is a n extra ordina rily importa nt concept to remember an d a pply. It ma y

be extended to three fi xed Keyphasor dots tha t identify a frequency component

at exactly one-third of running speed. Four fi xed Key dots identify a fre-

quency component at exactly one-fourth of running speed, etc. In many

instances, the orbital observation of fixed dots is faster, easier, and more accurate

tha n performing a n FF T an alysis of the data . Finally, the rate of Keypha sor dot

rotation is directly related to the frequency difference between the subsynchro-

nous component a nd 50%of running speed. If t he Key dots a re moving very

slowly, the su bsynchronous component is qu ite close to 50% of running speed.

Conversely, if the dots ar e rolling around t he orbit a t a ra pid pace, the subsyn-

chronous component is considera bly removed from one ha lf of running speed.

The vibration characteristics of a machine experiencing oil whirl are dem-

onstrat ed in Fig. 9-23. Operating at 2,832 RP M, the orbit a nd t ime base da ta in

the top plot represents t he unfi ltered signals from a rotor kit w ith a diametr ical

Fig. 923 Orbit & Time Base Of Oil Whirl Fig. 924 Spectrum Analysis Of Oil Whirl


32/64

426 Chapter-9

bearing clearance of 18 Mils. The orbit-time-base data shown in the bottom of

Fig. 9-23 describes t he low a mplitude runn ing s peed (1X) vibrat ory beha vior. For

precise frequency identification of the subsynchronous component, the same

information is viewed in the frequency domain in Fig. 9-24. From the FFT plotsthe w hirl frequency is 1,365 CP M w hich is equivalent t o 48%of sha ft rota tive

speed. Again, the run ning speed motion is dwa rfed by th e violent oil whirl excita-

tion at 48%of running speed.

Under the proper circumstances, oil whirl may turn into oil whip. If the

machine operates at a speed of twice the critical, the potential for whip exists. In

the previous example of a 48%whirl, the operating speed was about 2,832 RPM.

This speed is below the fi rst crit ical. I t should be noted tha t t he bandw idth of the

critical speed range for this machine extends from 3,000 to 3,800 RPM. This

ma chine went into whirl at 1,600 RPM, and ma inta ined a steady 48%of running

speed whirl during the initial speed ramp. This behavior is shown in Fig. 9-25

describing a cascade plot of vertical vibration spectra versus speed.

In this plot, the 48%Whirl tracked running speed until 6,200 RPM when

the Whirl frequ ency (6,200 RPM x 48%= 2,980 CP M) bega n t o move into th e crit-

ical speed range. As shown in the cascade plot, the whip frequency remained

locked into th e critical speed ra nge of 3,000 to 3,800 RPM a s speed contin ued to

increase. Under this condition, the increased rotor speed had minimal influence

upon the wh irl. Tha t is, even a t t he top speed of 9,800 RP M, the oil wh ip wa s st ill

tra pped w ithin t he na tura l rotor resonance range. Thus, a resonant response can

be obta ined from a n on-synchronous excita tion.

Since this type of oil whip behavior involves the re-excitation of a major

rotor resonance, it is rea sonable to expect tha t oil whip ma y be a potentia lly dan -

Fig. 925 Cascade PlotRevealing The TransitionFrom Oil Whirl To Oil Whip


33/64

Machinery Stability 427

gerous and destructive mechanism. In numerous field cases, machines have

operated successfully for many years with low levels of oil whirl. The whirl

existed, it wa s nondestructive, and it w as tolera ted. However, few ma chines ha ve

successfully survived any type of extended operation with appreciable levels ofoil whip instability.

A closer examination of the orbital and time domain vibration of this oil

w hip case is presented in th e orbit a nd t ime bas e plots of Fig. 9-26. Note th a t t he

unfiltered amplitudes are in the vicinity of 9.0 Mils,p-p, and the rotative speed

wa s 8,420 RP M. From th e orbit plots it is clearly demonstra ted tha t t his is a for-

ward and circular mechanism. The companion spectrum plot of the vertical and

horizonta l probe signals is shown in Fig. 9-27. It r eveals t he precise oil whip fre-

quency of 3,360 CPM. Under this combination of frequencies, the whip occurs at

40% of rota tive speed. Also note tha t t he whip frequency of 3,360 CP M falls

directly int o the critical speed ra nge of 3,000 to 3,800 RP M previously ident ifi ed.

Extended operation under this oil whip condition would probably be haz-

ar dous to the equipment. Sta ted in another wa y, it is genera lly agreed that s ta r-

tup a nd shut down ra mps should specifically minimize the t ime required to pass

through rotor critical speeds. Under no conditions shall a machine be allowed to

dwell within the bandwidth of the rotor resonance. However, in an oil whip con-

dit ion, the machine is continually running at operating speed, and the rotor is

violently shaking at its natural resonant frequency. It is no wonder thatma chines w ith oil w hip often experience signifi cant mecha nical fa ilures.

Oil whirl a nd w hip serve as an introduction into the broad t

Date post:	14-Apr-2018
Category:	Documents
Upload:	ewfsd
View:	247 times
Download:	0 times

Chapter 9 Common Malfunctions

Documents