Sie sind auf Seite 1von 82

67

C H A P T E R

3

Rotor Mode Shapes 3

T

he anal ysi s of machi nery vi brati on char-
acteri sti cs must be based upon a sol i d worki ng knowl edge of the dynami cs asso-
ci ated wi th a rotati ng mechani cal system. I tems that i nuence rotor moti on
i ncl ude the shaft constructi on, support l ocati ons, and the di stri buti on of masses
such as i mpel l ers, spacers, and coupl i ngs. Certai nl y the sti ffness and dampi ng
characteri sti cs of the beari ngs and machi ne support structure wi l l pl ay an
i mportant rol e i n i nuenci ng rotor behavi or. I n addi ti on, the rel ati onshi p
between the operati ng speed range, and the system cri ti cal s are very i mportant.
These factors are parameters for establ i shi ng the response characteri sti cs
of the rotor system, and the associ ated dynami c rotor mode shapes. I t i s mean-
i ngful to recogni ze that many machi ne characteri sti cs can be adequatel y
descri bed by appl yi ng the basi c concepts of mass and support di stri buti on, sys-
tem vi scous dampi ng, pl us sti ffness of the major el ements and support structure.
I t must al so be recogni zed that mode shapes may be forced deecti ons based
upon the acti ve system forces. They may al so be natural modes of vi brati on (res-
onances), pl us a combi nati on of the two to yi el d a forced resonant response.

M

ASS



AND

S

UPPORT

D

ISTRIBUTION

The di stri buti on of wei ght and supports al ong the l ength of a rotor estab-
l i shes the stati c deecti ons, pl us the stati c beari ng l oads. For exampl e, consi der a
constant di ameter shaft that i s si mpl y supported between two poi nts as shown i n
Fi g. 3-1. Thi s shaft wi l l have a maxi mum deecti on at the mi dspan, and each
support l ocati on wi l l carry one hal f of the shaft wei ght.
I f one support i s moved to the rotor mi dspan, the condi ti on descri bed i n Fi g.
3-2 wi l l occur. I n thi s case, the maxi mum deecti on occurs at the unsupported
end of the shaft. The l oad appl i ed to each support wi l l now be dependent upon
the support characteri sti cs. Speci cal l y, i f the l eft support i s a free support, i t
wi l l not have a stati c l oad. Under thi s condi ti on, the center support wi l l carry the
enti re wei ght, and the shaft wi l l be bal anced on thi s center pi vot. On the other
hand, i f the l eft support i s connected to the rotor (e.g., a beari ng), i t wi l l produce
a verti cal reacti on i f the mi ddl e support i s not perfectl y centered. For thi s reason,
68 Chapter-3

overhung machi nes such as power turbi ne rotors for dual shaft gas turbi nes, or
overhung bl owers must be careful l y exami ned to determi ne the stati c and
dynami c beari ng l oads, and di recti ons.
Next consi der the addi ti on of a concentrated l oad (e.g., an i mpel l er) at the
mi ddl e of the rotor. The di agram presented i n Fi g. 3-3 represents the deecti on
associ ated wi th the addi ti onal force appl i ed at the mi dspan of the si mpl y sup-
ported shaft. Cl earl y the center deecti on must i ncrease when the addi ti onal
l oad i s appl i ed. I n addi ti on, wi th the supports l ocated at the shaft ends, i t i s rea-
sonabl e to concl ude that the total wei ght (shaft pl us mi dspan l oad) wi l l be
equal l y shared between the two supports.
Fi g. 3-4 i l l ustrates the condi ti on of an overhung rotor wi th the addi ti on of a
center wei ght. I n thi s congurati on, one support i s l ocated di rectl y bel ow the
concentrated mi dspan l oad. I n thi s case, the mode shape, and the maxi mum
deecti on are i denti cal to Fi g. 3-2 wi th zero external l oad. However, the force bal -
ance has been al tered, and the center support must now carry the shaft wei ght
pl us the center l oad.

Fig. 31

Simply Supported Shaft With
Static Deection Due To Beam Weight

Fig. 32

Overhung Shaft With Static
Deection Due To Beam Weight

Fig. 33

Simply Supported Shaft Deected By
Beam Weight And Center Load

Fig. 34

Overhung Shaft Deected By
Beam Weight And Center Load
1/2 Weight 1/2 Weight
Uniform Shaft
Maximum Deflection
20
Maximum
Deflection
Uniform Shaft
20
Maximum Deflection
Concentrated
Load
35
Concentrated
Load
Maximum
Deflection
20
Mass and Support Distribution 69

Next, consi der the stati c mode shapes di spl ayed i n Fi gs. 3-5 and 3-6 that
descri be the i nuence of movi ng the concentrated l oad from the mi dspan to the
end of the rotor. For the shaft si mpl y supported between beari ngs (Fi g. 3-5), the
mode shape returns back to the i ni ti al condi ti on (Fi g. 3-1). The addi ti onal l oad i s
di rectl y transmi tted to the ri ght hand support. Under thi s congurati on, the
deected mode shape i s dependent onl y on the shaft wei ght, but the support
l oads are cl earl y di fferent.
Fi nal l y, for the overhung case of Fi g. 3-6, the canti l evered l oad at the end of
the shaft resul ts i n an i ncrease i n the maxi mum deecti on. Thi s type of behavi or
certai nl y makes i ntui ti ve sense, and i t i s representati ve of real overhung
machi nes. I t shoul d al so be recogni zed that the appl i cati on of thi s l oad to the free
end of the rotor wi l l resul t i n a downward verti cal restrai ni ng force at the l eft
end support. Agai n, thi s i s consi stent wi th the forces and moments encountered
i n machi nes such as power turbi nes and overhung bl owers.
Overal l , i t i s recogni zed that the stati c support forces (i .e., at the beari ngs),
pl us the l ocati on and magni tude of the maxi mum deecti on are dependent upon
the support characteri sti cs. I t i s apparent that the addi ti on of el ements such as
i mpel l ers, coupl i ngs, bal ance pi stons, and spacers wi l l di rectl y i nuence the
resul tant support forces, and the associ ated maxi mum stati c deecti on.
Si mpl e mechani cal systems can often be model ed as a uni form wei ght di s-
tri buti on for the shaft, combi ned wi th concentrated l oads for the i mpel l ers. The
stati c deecti ons can be cal cul ated wi th beam theory, and the stati c beari ng
l oads determi ned by summati on of moments. Reference books such as

Roarks
Formulas for Stress and Strain

1



provi de characteri sti c equati ons for many typi -
cal mechani cal systems wi thout resorti ng to detai l ed beam cal cul ati ons.
For more compl ex rotors, i t i s necessary to di vi de the rotor i nto di screte and

Fig. 35

Simply Supported Shaft
Deected By Beam Weight And End Load

Fig. 36

Overhung Shaft Deected By
Beam Weight And End Load

1

Warren C. Young,

Roarks Formulas for Stress & Strain

, Si xth Edi ti on, (New York: McGraw-
Hi l l Book Company, 1989).
Concentrated
Load
Maximum Deflection
20
Maximum
Deflection
Concentrated
Load
50
70 Chapter-3

denabl e segments. Based upon the di mensi ons and materi al densi ty, i t i s possi -
bl e to cal cul ate the wei ght for each stati on. From thi s wei ght di stri buti on, the
stati c l oads at both beari ngs can be computed. Thi s approach i denti es the mass
di stri buti on al ong the rotor, pl us the resul tant beari ng forces i n a stati c posi ti on.



The wei ght of each secti on or porti on of a rotor i s dependent upon the phys-
i cal di mensi ons, pl us the densi ty of the shaft materi al . For exampl e, the hol l ow
ci rcul ar cyl i nder depi cted i n Fi g. 3-7 i s di mensi onal l y speci ed by an outer di am-
eter

D

o

, an i nner di ameter

D

i

, and an overal l l ength

L

. The shaft radi us to the
outer di ameter i s

R

o

, and the i nternal bore radi us i s i denti ed as

R

i

. Based on
these di mensi ons, a vari ety of necessary cal cul ati ons may be performed. For
i nstance, from pl ane geometry, the cross secti onal area of thi s annul us i s com-
puted by subtracti ng ci rcul ar areas i n the fol l owi ng manner:

(3-1)

The vol ume of the annul us may be determi ned by mul ti pl yi ng the cyl i nder
l ength

L

ti mes the cross secti onal area

A

annulus

as stated i n the next expressi on:

(3-2)

The wei ght of the annul us may be cal cul ated by mul ti pl yi ng the cyl i nder
materi al densi ty


ti mes the total vol ume

V

annulus

as shown i n equati on (3-3):

(3-3)

For reference purposes, the densi ti es of many common metal s have been
summari zed i n Appendi x B of thi s text. Thi s tabul ati on al so i ncl udes the modu-
l us of el asti ci ty, the shear modul us, and the coefci ent of thermal expansi on for
each metal . These fundamental properti es are referred to throughout thi s book,
and i t i s conveni ent to have typi cal val ues readi l y avai l abl e. I t shoul d be men-
ti oned that some materi al s do not have total l y uni que properti es. For cri ti cal cal -
cul ati ons the preci se physi cal properti es shoul d be obtai ned from a metal l urgi cal
reference source, or speci c tests of the metal .

Fig. 37

Hollow Circular Cylinder
With Physical Dimensions
Length (L)
O
u
t
e
r
D
i
a
m
e
t
e
r

(
D
o
)
I
n
n
e
r
D
i
a
m
.
(
D
i
)
D
o
=2 R
o
D
i
=2 R
i
A
annul us
R
o
2
R
i
2


4
--- D
o
2
D
i
2
( ) = =
V
annul us
L A
annul us

L
4
------------- D
o
2
D
i
2
( ) = =
W
annul us
V
annul us

L
4
----------------------- D
o
2
D
i
2
( ) = =
Mass and Support Distribution 71

For some cal cul ati ons, i t i s necessary to know the mass of the mechani cal
part. For the annul us under di scussi on, the mass i s easi l y determi ned by di vi d-
i ng the wei ght from equati on (3-4) by the accel erati on of gravi ty

G

as fol l ows:

(3-4)

I n many i nstances, machi nery shafts are not hol l ow, and they are fabri cated
of sol i d metal . I f the i nner di ameter

D

i

i s set equal to zero i n equati ons (3-1)
through (3-4), the fol l owi ng expressi ons for sol i d machi ne shafts wi th an outer
di ameter of

D

are easi l y devel oped:

(3-5)
(3-6)
(3-7)
(3-8)

I n order to be perfectl y cl ear on the di mensi onal aspects of equati ons (3-1)
to (3-8), the dened vari abl es and thei r respecti ve Engl i sh engi neeri ng uni ts are
summari zed as fol l ows:

L

= Cylinder Length (Inches)

D

= Solid Cylinder Outer Diameter (Inches)

D

o

= Hollow Cylinder Outer Diameter (Inches)

D

i

= Hollow Cylinder Inner Diameter (Inches)

R

o

= Hollow Cylinder Outer Radius (Inches)

R

i

= Hollow Cylinder Inner Radius (Inches)


= Cylinder Material Density (Pounds / Inch

3

)

A

= Cylinder Cross Sectional Area (Inches

2

)

V

= Cylinder Volume (Inches

3

)

W

= Cylinder Weight (Pounds)

M

= Cylinder Mass (Pounds-Second

2

/ Inch)

G

= Acceleration of Gravity (= 386.1 Inches / Second

2

)

These vari abl es and associ ated uni ts wi l l be used i n the next secti ons on
i nerti a, pl us throughout the remai nder of thi s text.
M
annul us
W
annul us
G
--------------------------
L
4 G
----------------------- D
o
2
D
i
2
( ) = =
A
sol i d
D
2

4
----------------- =
V
sol i d
L D
2

4
--------------------------- =
W
sol i d
L D
2

4
------------------------------------- =
M
sol i d
L D
2

4 G
------------------------------------- =
72 Chapter-3

Case History 3: Two Stage Compressor Rotor Weight Distribution

Consi der the compressor rotor depi cted i n Fi g. 3-8. Thi s hi gh speed rotati ng
assembl y consi sts of a shaft wi th vari ous di ameters, pl us two mi dspan i mpel l ers,
and a thrust col l ar. Si nce the i mpel l ers are mounted back to back, thi s parti cul ar
desi gn does not i ncl ude a bal ance pi ston. As i ndi cated on the di agram, the rotor
wei ghts 282 pounds, and there i s a moderate overhang on the thrust end of the
rotor. Normal operati ng speed vari es between 15,000 and 17,500 RPM. Thi s com-
pressor i s a dri ve through uni t wi th a stream turbi ne coupl ed to one end, and
another compressor coupl ed to the opposi te (thrust) end. Hi stori cal l y, the beari ng
on the dri ve end of the rotor sel dom exhi bi ted any damage, but the journal bear-
i ng l ocated at the thrust end was often found to be i n a di stressed or damaged
condi ti on. On other occasi ons, the beari ngs woul d fai l duri ng operati on, and the
rotor woul d be severel y damaged. Dependi ng on the severi ty of the fai l ure, the
shaft was often chrome pl ated. After severe fai l ures, the enti re rotati ng assembl y
was total l y repl aced.
Thi s machi ne operates i n a very di rty and corrosi ve envi ronment. The rotor
was constantl y subjected to l arge unbal ance forces due to the accumul ati on of
Fig. 38 Two Stage Cen-
trifugal Compressor Rotor
With Weight And Moment
Diagram
7.49"
17.96"
24.03"
27.24"
33.43"
40.68"
43.66"
45.92"
49.25"
53.06"
2.62"
1
.
1
1
#
4
.
8
3
#
1
0
.
5
8
#
3
6
.
4
1
#
1
9
.
2
1
#
3
7
.
1
8
#
3
2
.
4
0
#
3
7
.
1
8
#
1
7
.
3
7
#
3
2
.
6
7
#
1
0
.
4
0
#
9
.
1
3
#
1
7
.
5
1
#
1
3
.
8
4
#
2
.
0
9
#
S
e
a
l
Thrust End
Journal
Wa Wb
I
m
p
e
l
l
e
r
T
h
r
u
s
t

C
o
l
l
a
r
S
e
a
l
I
m
p
e
l
l
e
r
Drive End
Journal
Bearing
Journal
3.62"
Seal Seal
Thrust
Collar
Impellers
14.5"
Total Rotor Weight
282 Pounds
40.68"
14.51" 4.40"
21.17"
Mass and Support Distribution 73
materi al i n the i mpel l ers, pl us random corrosi on of the i mpel l er vanes. Duri ng a
typi cal run, the compressor woul d startup smoothl y after an overhaul . However,
synchronous runni ng speed vi brati on response woul d al ways deteri orate wi th
the passage of ti me.
An i ni ti al step i n the anal ysi s of thi s probl em requi red the determi nati on of
stati c beari ng l oads. To achi eve thi s goal , the rotor was di vi ded i nto fteen sec-
ti ons, and the wei ght of each secti on was cal cul ated. Si nce thi s was a sol i d shaft,
equati on (3-7) was used to compute the wei ght of each shaft secti on. The wei ght
of the thrust col l ar was determi ned wi th (3-3), and the two i mpel l ers were
wei ghed separatel y. Next, the thrust col l ar and wheel wei ghts were combi ned
wi th thei r respecti ve shaft secti on wei ghts, and the wei ght di stri buti on summa-
ri zed i n Fi g. 3-8.
Thi s i nformati on was then combi ned wi th the di stance from the center of
the dri ve end beari ng to the centroi d of each rotor segment. The compl ete array
of wei ghts and di stances are shown i n Fi g. 3-8. From thi s sketch, i t i s possi bl e to
perform a summati on of moments around the center or transverse axi s of the
dri ve end journal i n the fol l owi ng manner:
(3-9)
Si nce the total rotor wei ght i s 282 pounds, the next force bal ance appl i es:
Thus, the dri ve end beari ng has a 116 pound stati c l oad, and the thrust end
journal carri es 166 pounds. Al though the di fferenti al force i s onl y 50 pounds, i t i s
an appreci abl e percentage di fference. Ul ti matel y, i t was determi ned that the
beari ngs were onl y margi nal l y si zed to accommodate the rotor wei ght. However,
they were consi derabl y undersi zed when the addi ti onal unbal ance forces due to
forei gn objects were i ncl uded. I t was al so determi ned that the avai l abl e l oad
capaci ty for the dri ve end was barel y acceptabl e, whereas the l oad carryi ng capa-
bi l i ty for the thrust end journal beari ng was unacceptabl e. Based upon these
concl usi ons, both beari ngs were i ncreased i n si ze and l oad capaci ty.
The l arger beari ngs reduced the number of machi ne fai l ures per year, and
overal l rel i abi l i ty was substanti al l y i mproved. Further i mprovements i n
machi ne l ongevi ty woul d requi re changes i n the chemi cal pl ant process. Unfortu-
natel y, the requi red al terati ons to the processi ng scheme coul d not be economi -
cal l y justi ed. Hence, the compressor rotor was occasi onal l y sacri ced to meet
Moments
ccw
Moments
cw
=
17.54 Inch-Pounds W
b
40.68 Inches + 6 768.74 Inch-Pounds , =
W
b
40.68 Inches 6 751.20 Inch-Pounds , =
W
b
166 Pounds =
W
a
W
b
+ 282 Pounds =
W
a
166 Pounds + 282 Pounds =
W
a
116 Pounds =
74 Chapter-3
producti on quotas.
From thi s exampl e, i t i s cl ear that even a si mpl e anal ysi s of rotor wei ght
di stri buti on and beari ng stati c l oads may be beneci al . I n some cases thi s may
sol ve a probl em, or i t may provi de i nsi ght i nto prospecti ve sol uti ons. I t must al so
be recogni zed that the dynami cs of the rotati ng system must be consi dered. Thi s
i ncl udes the effects of mass unbal ance that serves to deform the mode shape,
pl us the effects of i nerti a, sti ffness, and dampi ng of machi ne el ements.
INERTIA CONSIDERATIONS AND CALCULATIONS
As shown i n the previ ous exampl e, the cal cul ati on of rotor wei ghts i s easi l y
achi eved for rotors that can be segmented i nto a seri es of cyl i nders. For a homo-
geneous materi al , the vol ume and wei ght of each secti on may be computed. The
summati on of i ndi vi dual cyl i ndri cal secti on wei ghts yi el ds the total rotor wei ght.
I n most i nstances, thi s i s an acceptabl e and achi evabl e computati onal procedure.
However, probl ems occur wi th machi nes that have non-symmetri c shafts, or are
composed of mul ti pl e materi al s. For i nstance, the rotor wei ght of el ectri c
machi nes, such as motors and generators, may be i ndetermi nate due to an
unknown combi nati on of i ron, i nsul ati on, and copper wi thi n the rotor wi ndi ngs.
I n these si tuati ons, the rotor assembl y can be wei ghed on a scal e, and the di stri -
buti on of wei ght between beari ng journal s esti mated. Any addi ti onal el ements
such as a cool i ng fan, ywheel , or overhung exci ter may then be added to deter-
mi ne the total rotati ng system wei ght. Hence, when accurate cal cul ati ons cannot
be performed, the machi nery di agnosti ci an can revert to the tradi ti onal experi -
mental techni que of wei ghi ng the rotor.
I n addi ti on to wei ght or mass, the i nerti a properti es of a rotati ng system
must al so be consi dered duri ng any dynami c anal ysi s. Due to the compl exi ty of
i nerti a cal cul ati ons, and the strong potenti al for confusi on between area and
mass moment of i nerti as, si gni cant errors are possi bl e. The physi cs deni ti on of
i nerti a i s the property of a body to resi st accel erati on. Thi s i ncl udes the tendency
of a body at rest to remai n at rest, or the tendency of a body i n moti on to stay i n
moti on. Thi s reference to a physi cal body i mpl i es mass, whi ch suggests the pres-
ence of a three-di mensi onal object. Hence, an area moment of i nerti a i s some-
what of a mi snomer si nce a pl ane area has no depth, and therefore no mass, nor
the potenti al for i nerti a. Neverthel ess, area moments of i nerti a abound i n the l i t-
erature, and for the sake of compl eteness, they wi l l be di scussed.
I t shoul d al so be menti oned that confusi on regardi ng i nerti a i s further com-
pounded by some of the techni cal l i terature that transposes pol ar wi th the trans-
verse i nerti as. I n other cases, the authors fai l to i denti fy the speci c type of
i nerti a cal cul ati ons, much l ess the reference i nerti a axi s. I n an effort to ci rcum-
vent these potenti al sources of confusi on, the fol l owi ng di scussi on of i nerti a i s
offered for consi derati on.
I t i s desi rabl e to begi n wi th a revi ew of the area moment of i nerti a. I n Fi g.
3-9, Sketch A shows an i rregul ar pl ane surface wi th axes x-x and y-y passi ng
through the area desi gnated as A. These rectangul ar axes may be i n any l ocati on
Inertia Considerations and Calculations 75
wi th respect to the area. I f the area was symmetri c (e.g., ci rcl e or rectangl e), and
i f the axes were axes of symmetry, they woul d be cal l ed the pri nci pal i nerti a
axes. However, the general sol uti on for i nerti a about each axi s i s determi ned
from strength of materi al s i ntegral s as:
(3-10)
(3-11)
The X and Y terms i n (3-10) and (3-11) are di stances from each respecti ve
axi s to the i ncremental area i denti ed as dA. Thi s deni ti on of area moment of
i nerti a i s consi stent wi th references such as Roark
2
, or Shi gl ey
3
. I n both equa-
ti ons, i nerti a i s cal cul ated by mul ti pl yi ng each area by the square of the di stance
to the respecti ve reference axi s. The area i s measured i n i nches squared, and thi s
i s mul ti pl i ed by the square of the di stance to the axi s i n i nches. Thi s product
yi el ds an area moment of i nerti a wi th Engl i sh engi neeri ng uni ts of I nches
4
.
I n Sketch B of Fi g. 3-9, a thi rd axi s z-z has been added. Thi s new axi s i s per-
pendi cul ar to the pl ane of the area A, and i t passes through the i ntersecti on of
axes x-x and y-y. The di stance from the i ntersecti on of axi s z-z wi th the area A, to
any i ncremental area dA i s i denti ed by the radi us R. The i nerti a about thi s axi s
i s termed the area pol ar moment of i nerti a J that i s gi ven by:
(3-12)
From pl ane geometry i t i s known that the radi us R, may be expressed i n
terms of X and Y coordi nates as:
Fig. 39 Area Moment Of Inertia For An Irregular Plane Surface
2
Warren C. Young, Roarks Formulas for Stress & Strain, Si xth Edi ti on, (New York: McGraw-
Hi l l Book Company, 1989), p. 59.
3
Joseph E. Shi gl ey and Charl es R. Mi schke, Standard Handbook of Machine Design, (New
York: McGraw-Hi l l Book Company, 1986), p. 9.13.
x
y
y
x x
y
y
x
y
0
y
0
X
Offset
z
x
y
y
x
z
Sketch A Sketch C Sketch B
Area (A)
Incremental
Area (dA)
At Coordinates
(X and Y)
Incremental
Area (dA)
At Radius (R)
from Axis z-z
I
xx
Y
2
A d

=
I
yy
X
2
A d

=
J
zz
R
2
A d

=
76 Chapter-3
(3-13)
Equati on (3-13) may be i nserted i nto (3-12), and expanded as fol l ows:
Substi tuti ng (3-10) and (3-11) i nto the above expressi on, the pol ar i nerti a
J
zz
i s equated to the summati on of the transverse area i nerti as I
xx
and I
yy
as:
(3-14)
For a symmetri cal area such as a ci rcl e, the transverse area i nerti as I
xx
and
I
yy
are equal . Thus, setti ng i nerti as I
xx
=I
yy
=I , equati on (3-14) may be restated i n
the fol l owi ng si mpl i ed format:
(3-15)
Si nce most rotati ng shafts are ci rcul ar, equati on (3-15) i s common wi thi n
the machi nery busi ness. The appl i cabi l i ty of thi s geometri c si mpl i cati on wi l l be
better appreci ated duri ng the forthcomi ng di scussi on of mass moment of i nerti a.
However, before addressi ng that topi c, the transl ati on of the i nerti a axi s shoul d
be revi ewed. Speci cal l y, Sketch C i n Fi g. 3-9 i denti es a new axi s y
o
-y
o
that i s
paral l el to the previ ousl y dened verti cal axi s y-y. The constant di stance between
the two axes i s i denti ed as X
offset
. I t can be shown that the area moment of i ner-
ti a about thi s new axi s i s gi ven by the expressi on:
(3-16)
The same argument may be made i n the perpendi cul ar pl ane for the area
pol ar moment of i nerti a. Consi der a new pol ar axi s z
o
-z
o
that i s paral l el to the
previ ousl y dened z-z axi s. I f the di stance between the two axes i s i denti ed as
R
offset
, the area moment of i nerti a about thi s new pol ar axi s i s gi ven by:
(3-17)
The moment of i nerti a of an area about a parti cul ar axi s may be converted
to a mass moment of i nerti a by i ncl udi ng the thi ckness, and the densi ty of the
body. Thi s converts a two-di mensi onal i nto a three-di mensi onal probl em, and the
compl exi ty of the associ ated equati on structure i ncreases proporti onal l y. For
expl anati on purposes, assume that the pl ane area shown i n Fi g. 3-9 i s expanded
by addi ng a ni te depth. I f a materi al densi ty i s appl i ed to the resul ti ng vol ume,
the mass shown i n Fi g. 3-10 evol ves. Sketch A i n thi s di agram reveal s that a two-
di mensi onal x-y coordi nate system i s i nadequate for l ocati ng a poi nt wi thi n the
R
2
X
2
Y
2
+ =
J
zz
X
2
Y
2
+ ( ) A d

=
J
zz
X
2
A d

Y
2
A d

+ =
J
zz
I
xx
I
yy
+ =
J
zz
I I + 2 I = =
I
y
o
y
o
I
yy
A X
offset
( )
2
+ =
J
z
o
z
o
J
zz
A R
offset
( )
2
+ =
Inertia Considerations and Calculations 77
mass M. Hence, the three-di mensi onal x-y-z coordi nate system i n Sketch B of
Fi g. 3-10 i s necessary to l ocate an i ncremental mass dM wi thi n the boundari es of
the object. The equati ons for mass moment of i nerti a around the three axes are
dened by sources such as Marks Handbook
4
as fol l ows:
(3-18)
(3-19)
(3-20)
The transl ati on of one axi s to another paral l el axi s al so appl i es to mass
i nerti a as wel l as to the previ ousl y di scussed area moment of i nerti a. For exam-
pl e, the pol ar mass moment of i nerti a around axi s z-z may be transl ated to the
paral l el axi s z
o
-z
o
wi th the fol l owi ng common expressi on:
(3-21)
Wi thi n thi s equati on, the di stance X
offset
represents the paral l el offset al ong
the x-x axi s as shown i n Sketch C of Fi g. 3-10. I f the offset of the new axi s was
further di spl aced from both the x-x and y-y axes, the pol ar moment of i nerti a
al ong the new axi s i denti ed as z
1
-z
1
woul d be gi ven by the fol l owi ng:
(3-22)
The di stance between the z-z axi s and the z
1
-z
1
axi s i s dened by the radi al
offset R
offset
. From (3-13) i t i s known that R
offset
2
= X
offset
2
+ Y
offset
2
. The val i di ty
Fig. 310 Mass Moment Of Inertia For An Irregular Solid Object
4
Eugene A. Aval l one and Theodore Baumei ster I I I , Marks Standard Handbook for Mechani-
cal Engineers, Tenth Edi ti on, (New York: McGraw-Hi l l , 1996), p. 3-9.
x
y
y
x x
y
y
x
z
0
z
0
X
Offset
z
x
y
y
x
z
Sketch A Sketch C Sketch B
z
z
Mass (M)
Incremental
Mass (dM)
At Coordinates
(X, Y and Z)
z
1
z
1
Y
Offset
I
xx
Y
2
Z +
2
( ) M d

=
I
yy
X
2
Z +
2
( ) M d

=
I
zz
X
2
Y +
2
( ) M d

=
J
z
o
z
o
J
zz
M X
offset
( )
2
+ =
J
z
1
z
1
J
zz
M R
offset
( )
2
+ =
78 Chapter-3
of thi s i s demonstrated by al l owi ng Y
offset
to equal zero, and noti ng that (3-22)
then reverts back to equati on (3-21). I n al l cases, (3-21) and (3-22) are anal ogous
to the paral l el axi s equati on (3-17) for a pl ane area. I n fact, thi s si mi l ari ty i n
mathemati cal formats i s the source for some of the confusi on associ ated wi th the
topi c of i nerti a. I n order to mai ntai n a proper di sti ncti on between area and mass
moment of i nerti a, the di agnosti ci an shoul d al ways pay cl ose attenti on to the
engi neeri ng uni ts. Recal l that the area moment of i nerti a carri es Engl i sh engi -
neeri ng uni ts of I nches
4
, and the mass i nerti a has uni ts of Pound-I nch-Second
2
.
The i ntegral s shown on the l ast few pages are i nteresti ng cal cul us topi cs,
but they do not hel p the el d di agnosti ci an unti l they are sol ved for speci c geo-
metri c shapes. For i nstance, i t i s meani ngful to devel op the equati ons for a ci rcu-
l ar shaft cross secti on. I n order to mai ntai n conti nui ty, the hol l ow ci rcul ar
cyl i nder from Fi g. 3-7 wi l l be reused. For the cal cul ati on of i nerti a, the pri nci pal
axes passi ng through the axi al centerl i ne o-o, and the mi dspan di ametral center-
l i ne b-b have been shown i n Fi g. 3-11. For thi s geometri c gure, the cross sec-
ti onal area moment of i nerti a I
area
al ong a di ameter of the hol l ow ci rcl e i s gi ven
by Spotts
5
, or Harri s
6
i n the fol l owi ng common format:
(3-23)
The area pol ar moment of i nerti a J
area
i s computed about an axi s that i s
perpendi cul ar to the ci rcul ar cross secti on. On Fi g. 3-11 thi s woul d be the center-
l i ne axi s o-o of the cyl i nder. The pol ar i nerti a of thi s ci rcul ar area i s equal to
twi ce the i nerti a al ong a di ameter (i .e., J =2xI ) as evi dent from (3-15). Mul ti pl y-
Fig. 311 Hollow Circular
Cylinder With Principal
Axes Identied
5
M.F. Spotts, Design of Machine Elements, 6th Edi ti on, (Engl ewood Cl i ffs, New Jersey: Pren-
ti ce-Hal l , I nc., 1985), p. 18.
6
Cyri l M. Harri s, Shock and Vibration Handbook, Fourth edi ti on, (New York: McGraw-Hi l l ,
1996), p. 1.12.
Length (L)
O
u
t
e
r
D
i
a
m
e
t
e
r

(
D
o
)
I
n
n
e
r
D
i
a
m
.
(
D
i
)
b
b
o
o
Diameter
Centerline
Axial
Centerline
D
o
=2 R
o
D
i
=2 R
i
I
area
annul us
D
o
4
D
i
4
( )
64
----------------------------------- =
Inertia Considerations and Calculations 79
i ng equati on (2-23) by two, the next expressi on for the area pol ar i nerti a evol ves:
(3-24)
The val i di ty of equati on (3-24) i s supported by Spotts, page 150, and others.
Note that the l ast two expressi ons are pl ane area moments of i nerti a (i .e., no
depth, no densi ty, and no wei ght). However, i n many machi nery cal cul ati ons i t i s
mandatory to consi der the mass of the el ement. When mass i s i ncl uded, the cal -
cul ati ons become more compl ex. Speci cal l y, i f the hol l ow cyl i nder i n Fi g. 3-11 i s
consi dered to be a rotati ng machi nery shaft, the mass pol ar moment of i nerti a
al ong the axi s of rotati on (axi s o-o) i s dened by Gi eck
7
and others. I f these equa-
ti ons are pl aced i nto the nomencl ature used i n Fi g. 3-11, the general expressi on
for the mass pol ar moment of i nerti a of the cyl i nder may be stated as:
(3-25)
Equati on (3-25) i s the mass i nerti a term of a rotati ng shaft that i s often
descri bed as the WR
2
of the rotor. I n actual i ty, thi s quanti ty shoul d be i denti ed
as the mass pol ar moment of i nerti a, but common nomencl ature someti mes
supersedes techni cal accuracy. Neverthel ess, i f the di ameters are used i nstead of
the radi i , and i f (3-4) for the annul us mass i s substi tuted i nto (3-25); the fol l ow-
i ng mani pul ati on may be performed to reach a common equati on used for cal cu-
l ati on of the mass pol ar moment of i nerti a for a hol l ow cyl i nder.
Si mpl i cati on of the l ast expressi on yi el ds the fol l owi ng equati on:
(3-26)
Thi s mass pol ar i nerti a i s as i mportant to a torsi onal anal ysi s as the mass
i s necessary for a l ateral anal ysi s. Speci cal l y, equati on (2-102) i denti es the uti -
l i zati on of the pol ar i nerti a to compute a torsi onal natural frequency i n the same
way that the mass i s used i n (2-44) to compute a l ateral natural frequency. As
shown i n Tabl e 2-1, the i nerti al term i n the general equati on of moti on i s gov-
erned by mass i n a l ateral system, and pol ar i nerti a i n a torsi onal system. Dur-
i ng the anal ysi s or model i ng of real machi nery both the pol ar i nerti a and the
7
Kurt Gi eck and Rei ner Gi eck, Engineering Formulas, 6th edi ti on, (New York: McGraw-Hi l l
I nc., 1990), p. M3.
J
area
annul us
D
o
4
D
i
4
( )
32
----------------------------------- =
J
mass
annul us
M
annul us
2
-------------------------- R
o
2
R
i
2
+ ( ) =
J
mass
annul us
M
annul us
2
-------------------------- R
o
2
R
i
2
+ ( )
M
annul us
2
--------------------------
D
o
2
4
-------
D
i
2
4
------- +
,

_
= =
J
mass
annul us
M
annul us
8
-------------------------- D
o
2
D
i
2
+ ( )
L
4 G 8
----------------------- D
o
2
D
i
2
( ) D
o
2
D
i
2
+ ( ) = =
J
mass
annul us
L
32 G
----------------------- D
o
4
D
i
4
( ) =
80 Chapter-3
transverse i nerti a are uti l i zed. Si nce (3-26) denes the mass pol ar moment of
i nerti a J
mass
through the axi al centerl i ne o-o, i t i s now reasonabl e to dene the
transverse mass I nerti a I
mass
of the hol l ow cyl i nder. Al though any dened axi s
may be used for the cal cul ati ons, the customary mi dspan di ameter axi s b-b
depi cted i n Fi g. 3-11 wi l l be used for the fol l owi ng exerci se. Agai n, extracti ng a
standard mass i nerti a equati on from Gi eck, and modi fyi ng the terms to be con-
si stent wi th Fi g. 3-11, the general expressi on for the mass moment of i nerti a on
the cyl i nder di ameter axi s b-b passi ng through the center of gravi ty i s as fol l ows:
(3-27)
I f di ameters are used i nstead of radi i , and i f the mass equati on (3-4) i s
i ncl uded, the transverse i nerti a equati on (3-27), may be modi ed as shown:
I f the mass pol ar moment of i nerti a from (3-26) i s substi tuted i nto the l ast
expressi on, the transverse i nerti a equati on may be si mpl i ed as fol l ows:
(3-28)
I f the l ength L of the annul us i s smal l compared to the outer di ameter, the
i nuence of the far ri ght hand term i n equati on (3-28) i s si gni cantl y di mi n-
i shed, and the fol l owi ng approxi mati on i s often used:
(3-29)
Thi s resul t i s consi stent wi th the previ ous rel ati onshi p descri bed by equa-
ti on (3-15) between the rati o of i nerti as for a pl ane ci rcul ar cross secti on. I n
actual practi ce duri ng the anal yti cal model i ng of a rotor system, the shaft sta-
ti on l engths are normal l y kept fai rl y short, and equati on (3-29) may be a good
I
mass
annul us
M
annul us
12
-------------------------- 3R
o
2
3R
i
2
L
2
+ + ( ) =
I
mass
annul us
M
annul us
12
-------------------------- 3R
o
2
3R
i
2
L
2
+ + ( )
M
annul us
12
--------------------------
3D
o
2
4
----------
3D
i
2
4
---------- L
2
+ +
,

_
= =
I
mass
annul us
M
annul us
16
-------------------------- D
o
2
D
i
2 4L
2
3
---------- + +
,
_
=
I
mass
annul us
M
annul us
16
-------------------------- D
o
2
D
i
2
+ ( )
M
annul us
12
-------------------------- L
2
+ =
I
mass
annul us
L
16 4 G
-------------------------- D
o
2
D
i
2
( ) D
o
2
D
i
2
+ ( )
L
12 4 G
-------------------------- D
o
2
D
i
2
( ) L
2
+ =
I
mass
annul us
L
64 G
----------------------- D
o
4
D
i
4
( )

' ;

L
3

48 G
-------------------------- D
o
2
D
i
2
( )

' ;

+ =
I
mass
annul us
J
mass
annul us
2
--------------------------------
L
3

48 G
-------------------------- D
o
2
D
i
2
( )

' ;

+ =
I
mass
annul us
J
mass
annul us
2
--------------------------------
Inertia Considerations and Calculations 81
approxi mati on. For l onger shaft secti ons, the more compl ex equati on (3-28) must
be used. Equati on (3-29) i s parti cul arl y useful for esti mati ng the transverse i ner-
ti a of wheel s that have a l arge di ameter, and a comparati vel y short l ength. For
exampl e, a centri fugal compressor i mpel l er may be 24 i nches i n di ameter, wi th a
di sk and a cover thi ckness of onl y 0.25 i nches each. I n thi s si tuati on, one hal f of
the mass pol ar moment of i nerti a J
mass
wi l l be very cl ose to the detai l ed trans-
verse mass moment of i nerti a I
mass
.
As menti oned earl i er i n thi s chapter, the di agnosti ci an must al ways be
aware of the potenti al di l emma i n the appl i cati on of i nerti a wi thi n techni cal doc-
uments, and computer programs. For i nstance, one set of software uses wei ght
i nerti a i nstead of mass i nerti a. The output i nerti a val ues from these programs
carry i nerti a uni ts of Pounds-I nches
2
i nstead of Pound-I nch-Second
2
. Al though
the di fference between the two i nerti a val ues i s onl y the accel erati on of gravi ty,
the resul ts can be confusi ng to the unprepared. I n al l cases, i t i s mandatory to be
compl etel y knowl edgeabl e of al l aspects of any i nerti a cal cul ati ons. Wi thi n thi s
text, i nerti as and thei r respecti ve Engl i sh engi neeri ng uni ts are as fol l ows:
I
area
= Area Moment of Inertia on Diameter (Inches
4
)
J
area
= Area Polar Moment of Inertia (Inches
4
)
I
mass
= Mass Transverse Moment of Inertia (Pound-Inch-Second
2
)
J
mass
= Mass Polar Moment of Inertia (Pound-Inch-Second
2
)
I n many i nstances, machi nery shafts are not hol l ow, and they are fabri cated
of sol i d metal . I f the i nner di ameter D
i
i s set equal to zero i n equati ons (3-23), (3-
24), (3-26), and (3-28), the fol l owi ng i nerti a expressi ons for sol i d shafts wi th a
di ameter of D are easi l y devel oped:
(3-30)
(3-31)
(3-32)
(3-33)
Once agai n these expressi ons may be veri ed from vari ous sources such as
Spotts, Marks, or Shi gl ey. Pl ease recal l that the devel oped equati ons are based
upon a ci rcul ar cross secti on. I f the cross secti onal area i s not ci rcul ar, then the
equati ons must be modi ed based upon the ori gi nal i ntegral s used to dene i ner-
ti a. Often thi s type of cal cul ati on i s not practi cal due to the compl exi ty of the
mechani cal part. I n these cases, the di agnosti ci an must resort to other tech-
I
area
sol i d
D
4

64
----------------- =
J
area
sol i d
D
4

32
----------------- =
J
mass
sol i d
L D
4

32 G
------------------------------------- =
I
mass
sol i d
J
mass
sol i d
2
-------------------------
L
3
D
2

48 G
---------------------------------------

' ;

+ =
82 Chapter-3
ni ques to determi ne the i nerti a properti es of the machi ne el ement.
I n the same way that a scal e may be used to determi ne the wei ght of a
rotor, there are experi mental techni ques that may be appl i ed to determi ne i ner-
ti a properti es. For exampl e, consi der the machi ne part shown i n Fi g. 3-12. Thi s
coul d be a bl ank for a bul l gear or a ywheel , or any other machi ne el ement for
whi ch a mass pol ar moment of i nerti a may be requi red. Due to the compl exi ty of
the part i t may not be feasi bl e to cal cul ate the i nerti a di rectl y, but i t i s possi bl e to
experi mental l y determi ne the i nerti a.
One techni que consi sts of suspendi ng the part from a hori zontal support as
shown i n Fi g. 3-12. I deal l y, thi s support member shoul d be a kni fe edge, but more
real i sti cal l y, i t wi l l probabl y be a sol i d ci rcul ar rod as depi cted. I f the mass pol ar
moment of i nerti a through the axi al centerl i ne of the el ement i s desi red, then the
di stance between that centerl i ne and the supporti ng pi vot poi nt X
offset
must be
accuratel y measured. I n addi ti on, the part shoul d be wei ghed to determi ne the
total wei ght W i n pounds. The machi ne el ement may now be rocked back and
forth as a compound pendul um. By measuri ng the peri od of the moti on, the i ner-
ti a about the pi vot poi nt may be determi ned. Thi s offset i nerti a may now be
transl ated back to the centerl i ne of the machi ne part by appl yi ng the paral l el
axi s equati ons previ ousl y devel oped.
I n actual practi ce, the I nerti a of thi s compound pendul um may be deter-
mi ned from Marks Handbook
8
. Extracti ng the appropri ate equati on, and pl aci ng
i t i n terms used wi thi n thi s text, the fol l owi ng equati on for overal l i nerti a about
the pi vot poi nt may be stated:
(3-34)
Fig. 312 Mechanical
Arrangement For Rocking
Test To Determine Mass
Polar Moment of Inertia
8
Eugene A. Aval l one and Theodore Baumei ster I I I , Marks Standard Handbook for Mechani-
cal Engineers, Tenth Edi ti on, (New York: McGraw-Hi l l , 1996), p. 3-16.
Element
Weight (W)
Rocking
Oscillation
40 to 60
Fixed and
Rigid
Support
X
0ffset
J
z
o
z
o
W X
offset
Peri od
2

4
2

----------------------------------------------------------- =
Inertia Considerations and Calculations 83
I f equati on (3-34) i s i nserted i nto the axi s transl ati on equati on (3-21), the
fol l owi ng i s obtai ned:
Factori ng out the common terms, the above may be si mpl i ed to:
(3-35)
Thi s i s a useful expressi on si nce al l of the vari abl es are easi l y determi ned.
Speci cal l y, the wei ght W of the machi ne part can be measured on a scal e, and
the di stance between the pi voti ng poi nt and the geometri c center of the el ement
X
offset
i s easi l y measured. The accel erati on of gravi ty G i s constant, and the
Period of the swi ngi ng moti on i s determi ned wi th a stopwatch i n seconds. Nor-
mal l y, a seri es of runs are made to determi ne the average peri od of osci l l ati on.
Experi mental techni ques are al ways more credi bl e i f the val i di ty of the
equati ons, and the experi mental procedures can be veri ed wi th a control l ed
test. Wi th an i nerti a experi ment, the object to be tested shoul d be of si mpl e
geometry such as the sol i d cyl i nder shown i n Fi g. 3-13. As i ndi cated on the di a-
gram, the average cyl i nder di ameter D i s equal to 6.312 I nches, and the average
l ength L i s 3.735 I nches. A shop scal e i ndi cated that the wei ght W of the speci -
Fig. 313 Rocking Test To
Determine Mass Polar
Moment Of Inertia Of A
Solid Brass Cylinder
J
z
o
z
o
J
zz
M X
offset
( )
2
+ =
J
zz
J
z
o
z
o
M X
offset
( )
2
=
J
zz
W X
offset
Peri od
2

4
2

----------------------------------------------------------- M X
offset
( )
2
=
J
zz
W X
offset
Peri od
2

4
2

-----------------------------------------------------------
W
G
----- X
offset
( )
2
=
J
zz
W X
offset

Peri od
2
-------------------
,
_
2 X
offset
G
------------------
,
_


' ;

=
Brass
Cylinder
Weight
(W=36 Lbs.)
Rocking
Oscillation
40 to 60
Fixed and
Rigid
Support

X
0
f
f
s
e
t

=

5
.
0
7
"
Length
(L =3.735")
Diameter
(D=6.312")
84 Chapter-3
men was 36.0 0.1 Pounds. Si nce the cyl i nder was sol i d brass, the densi ty was
found i n Appendi x B of thi s text to be 0.308 Pounds/I nch
3
. Two hol es were dri l l ed
and tapped i nto the si de of the cyl i nder, and two screw eyes were i nserted to pro-
vi de the pi vot arm. Si nce the mass pol ar moment of i nerti a al ong the axi al cen-
terl i ne i s desi red, the di stance between the pi vot poi nt and the cyl i nder center
l i ne X
offset
was measured to be 5.07 I nches.
Before starti ng the experi ment, i t i s desi rabl e to check the physi cal proper-
ti es of the brass cyl i nder. For exampl e, equati on (3-7) may be used to compute
the wei ght of the brass cyl i nder based upon the average di mensi ons, and the
densi ty of the materi al as fol l ows:
The cal cul ated val ue shows excel l ent agreement wi th the measured wei ght
of the cyl i nder, and 36.0 pounds may be condentl y used for the ensui ng i nerti a
measurements. Some i ndi vi dual s mi ght argue that thi s type of wei ght check i s
unnecessary. However, i f the materi al contai ned i ncl usi ons, or i f the densi ty was
wrong, or i f one or more of the di mensi ons were i ncorrect the wei ghts woul d
not match, and the experi mental accuracy woul d be i n jeopardy. So a si mpl e cal -
cul ati on such as thi s wei ght check i s desi rabl e to i nsure that the physi cal param-
eters are i n uni son. The other cal cul ati on that shoul d be made at thi s stage i s the
mass pol ar moment of i nerti a of the brass test speci men. From the previ ousl y
devel oped equati on (3-32) the pol ar moment of i nerti a of thi s brass cyl i nder i s
computed i n the fol l owi ng manner:
At thi s poi nt the test cal i brati on setup i s establ i shed, the nal answer i s
known, and the onl y remai ni ng vari abl e to be dened for equati on (3-35) i s the
peri od of the osci l l atory moti on. Several prel i mi nary swi ngs of the brass cyl i nder
reveal ed that the ti me for one compl ete cycl e was l ess than a second. I n addi ti on,
the fri cti on between the two eye bol ts and the support rod caused the osci l l ati ng
mass to gri nd to a stop after onl y a few cycl es. Thi s probl em was parti al l y reme-
di ed by putti ng a ti ght pl asti c sl eeve on the support rod, and then coveri ng thi s
surface wi th l i thi um grease. Thi s fri cti on reducti on effort was rewarded by a sub-
stanti al i ncrease i n the number of possi bl e osci l l atory cycl es. However, i t was
then di scovered that the hori zontal support rod was not qui te l evel , and the
W
sol i d
L D
2

4
------------------------------------- =
W
sol i d
3.735 Inches 0.308 Pounds/Inch
3
6.312 Inches ( )
2

4
------------------------------------------------------------------------------------------------------------------------------------ 35.997 Pounds = =
J
mass
sol i d
L D
4

32 G
------------------------------------- =
J
mass
sol i d
3.735 Inches 0.308 Pounds/Inch
3
6.312 Inches ( )
4

32 386.1 Inches/Second
2

------------------------------------------------------------------------------------------------------------------------------------ =
J
mass
sol i d
3.735 0.308 1 573.33 ,
32 386.1
----------------------------------------------------------------------- 0.464 Pound-Inch-Second
2
= =
Inertia Considerations and Calculations 85
brass cyl i nder had a tendency to wal k down the rod. Thi s probl em was corrected
by re-l evel i ng the support rod and the cyl i nder.
Fol l owi ng these test setup modi cati ons, the actual test was conducted. The
brass cyl i nder was di spl aced about 30 from the verti cal centerl i ne and rel eased.
The peak of the moti on at one extremi ty was vi sual l y si ghted, and a stopwatch
was used to measure the ti me requi red for mul ti pl e back and forth cycl es. The
nal measured test data i s summari zed as fol l ows:
Rocking Run #1 7.83 Seconds for 10 cycles
Rocking Run #2 7.95 Seconds for 10 cycles
Rocking Run #3 7.98 Seconds for 10 cycles
Rocking Run #4 8.09 Seconds for 10 cycles
Rocking Run #5 7.79 Seconds for 10 cycles
Rocking Run #6 7.90 Seconds for 10 cycles
Rocking Run #7 7.89 Seconds for 10 cycles
Rocking Run #8 7.94 Seconds for 10 cycles
Rocking Run #9 7.93 Seconds for 10 cycles
Rocking Run #10 7.89 Seconds for 10 cycles
Total Time = 79.19 Seconds for 100 cycles
Average Time for 1 Cycle = 0.7919 Seconds = Period
I t i s easy to l ose track of the cycl e count, or mi ss a ti mi ng poi nt, and negate
the accuracy of a data set. These types of errors are evi dent duri ng the data col -
l ecti on work, and erroneous ti mes are i denti ed and di scarded. For i nstance,
approxi matel y twenty runs were made to col l ect the data i n the above tabul ar
summary. Ten of the ti mi ng runs were not used due to obvi ous errors i n the data
accumul ati on. The ten acceptabl e test runs reveal an average peri od of 0.7919
Seconds. Thi s i s consi dered to be a consi stent val ue, and the experi mental mass
pol ar moment of i nerti a may now be computed from equati on (3-35).
The experi mental val ue for the pol ar i nerti a i s 0.502 versus the cal cul ated
val ue for thi s brass cyl i nder of 0.464 Pound-I nch-Seconds
2
. The di fference of
0.038 represents an 8% error of the experi mental versus the computed val ue. I n
some i nstances thi s l evel of devi ati on i s perfectl y acceptabl e. For exampl e, i f the
part under test was a coupl i ng hub that wi l l be mounted on a power turbi ne wi th
an i nerti a of 40.0 Pound-I nch-Seconds
2
, the smal l di fferenti al of 0.038 Pound-
I nch-Seconds
2
woul d be i nsi gni cant. However, i f the part under test was one of
J
zz
W X
offset

Peri od
2
-------------------
,
_
2 X
offset
G
------------------
,
_


' ;

=
J
zz
36.0 Pounds 5.07 Inches
0.7919 Seconds
2
------------------------------------
,
_
2
5.07 Inches
386.1 Inches/Second
2
------------------------------------------------
,
_


' ;

=
J
zz
182.52 Pound-Inches 0.01588 0.01313 ( )Second
2
0.502 Pound-Inch-Second
2
= =
86 Chapter-3
ei ght i mpel l ers to be mounted on a sl ender shaft, the cumul ati ve error may be
unacceptabl e. I n order to expl ai n thi s error, i t i s necessary to re-exami ne the
component equati ons used for the i nerti a test cal cul ati on. Speci cal l y, (3-34) for
the total i nerti a about the pi vot poi nt may be sol ved as fol l ows:
Substi tuti ng the test i nerti a from the above cal cul ati on back i nto equati on
(3-21), and performi ng the axi s transl ati on, the fol l owi ng resul t i s obtai ned:
As expected, thi s resul t i s i denti cal to the previ ous answer obtai ned by
usi ng the composi te equati on (3-35). The i nteresti ng poi nt of the above cal cul a-
ti ons i s that i nerti a due to the axi s transl ati on i s equal to 2.397, versus the over-
al l test i nerti a of 2.899 Pound-I nch-Seconds
2
. Hence, the cyl i nder i nerti a i s onl y
about 20% of the i nerti a due to the axi s transl ati on. Thi s i s not a desi rabl e condi -
ti on si nce the axi s transl ati on i s the domi nant term. The geometri cal congura-
ti on di spl ayed i n Fi g. 3-12 woul d not be as error prone si nce the X
Offset
di stance
resi des wi thi n the body of the el ement, and the axi s transl ati on term woul d not
domi nate the test. Hence, the di agnosti ci an shoul d al ways be concerned about
trusti ng thi s type of experi mental i nerti a test wi th a l ong X
Offset
di stance between
the pi vot axi s and the desi red pri nci pal pol ar moment of i nerti a axi s. The other
l esson to be l earned i s that si mpl i ed expressi ons such as (3-35) may not provi de
ful l vi si bi l i ty concerni ng the potenti al accuracy of the nal resul ts. I n some cases
i t i s necessary to revert back to the basi c equati ons, and reexami ne the enti re
cal cul ati on and/or experi mental test procedure.
Wi th respect to the brass cyl i nder, i t i s concl uded that i mprovement of the
test accuracy wi l l requi re a reducti on or el i mi nati on of the axi s transl ati on term.
Thi s coul d be accompl i shed wi th a test that consi sted of suspendi ng the mass
from cabl es, and then measuri ng the peri od as the mass osci l l ated i n a twi sti ng
manner (Fi g. 3-14). Si nce the axi s of osci l l ati on i s the axi al geometri c centerl i ne
of the el ement, there i s no axi s transl ati on i nvol ved, and test accuracy shoul d be
i mproved. Thi s type of i nerti a test i s i deal for machi ne parts such as compressor
i mpel l ers or turbi ne di sks that contai n compl ex geometri cal cross secti ons. For
J
z
o
z
o
W X
offset
Peri od
2

4
2

----------------------------------------------------------- =
J
z
o
z
o
36.0 Pounds 5.07 Inches 0.7919 Seconds ( )
2

4
2

-------------------------------------------------------------------------------------------------------------- 2.899 Pound-Inch-Second


2
= =
J
z
o
z
o
J
zz
M X
offset
( )
2
+ =
2.899 Pound-Inch-Second
2
J
zz
W
G
----- X
offset
( )
2
+ =
J
zz
2.899 Pound-Inch-Second
2 36.0 Pounds
386.1 Inches/Second
2
------------------------------------------------ 5.07 Inches ( )
2
=
J
zz
2.899 2.397 ( ) Pound-Inch-Second
2
0.504 Pound-Inch-Second
2
= =
Inertia Considerations and Calculations 87
i nstance, the i nternal star pattern shown i n Fi g. 3-14 mi ght be di fcul t to model
wi th an equi val ent i nner di ameter for the machi ne part.
For thi s procedure, the machi ne el ement i s suspended from three thi n
cabl es (3 poi nts determi ne a pl ane), spaced at 120 apart. The test pi ece must be
l evel ed as preci sel y as possi bl e. I f i t i s not l evel , then any i nduced twi sti ng osci l -
l ati ons wi l l cause the machi ne el ement to wobbl e duri ng the test. Thi s wobbl e
not onl y negates the test accuracy, i t can prove to be dangerous for parts wi th
any appreci abl e physi cal si ze and wei ght. After l evel i ng, the average suspensi on
cabl e l ength L
s-c
and the cabl e radi us R
s-c
are accuratel y measured and recorded.
For best resul ts, each of the suspensi on cabl e l engths shoul d be equal , and the
radi us for al l three cabl es shoul d be i denti cal . As before, the machi ne el ement to
be tested i s wei ghed on an accurate scal e i n Engl i sh uni ts of Pounds.
Duri ng executi on of thi s test, the machi ne el ement i s manual l y di spl aced i n
a twi sti ng manner, and rel eased. The machi ne part wi l l torsi onal l y twi st back
and forth, and the peri od of the twi sti ng osci l l ati ons wi l l be measured wi th a
stopwatch. Si nce fri cti on shoul d not be major probl em, the part wi l l osci l l ate
back and forth for many cycl es. I t i s not unusual to observe thi rty or more cycl es
resul ti ng from one i ni ti al di spl acement. Based upon these measured parameters
the mass pol ar moment of i nerti a may be computed as fol l ows:
(3-36)
The general form of (3-36) was extracted from the Shock & Vibration Hand-
book
9
,

and i t was converted to the nomencl ature used i n thi s text. As previ ousl y
Fig. 314 Mechanical
Arrangement For Twisting
Test To Determine Mass
Polar Moment of Inertia
9
Cyri l M. Harri s, Shock and Vibration Handbook, Fourth edi ti on, (New York: McGraw-Hi l l ,
1996), p. 38.5.
Element
Weight (W)


S
u
s
p
e
n
s
i
o
n

C
a
b
l
e
L
e
n
g
t
h

(
L
s
-
c
)
Element Axial
Centerline
Suspension
Cable
Twisting
Oscillation
30 to 40
Radius
(R
s-c
)
J
zz
W R
s c
2
Peri od
2

4
2
L
s c

------------------------------------------------------- =
88 Chapter-3
di scussed on the rocki ng i nerti a test, i t i s mandatory to val i date the test proce-
dure wi th an actual test on a known geometri c shape. For comparati ve purposes,
the sol i d brass cyl i nder used for the rocki ng test wi l l be used for thi s twi sti ng
i nerti a test as shown i n Fi g. 3-15. From thi s di agram i t i s noted that the average
suspensi on cabl e radi us R
s-c
was 3.00 I nches, and the average cabl e l ength L
s-c
was 33.73 I nches. As before, the total cyl i nder wei ght was 36.0 Pounds.
The rst test congurati on used 48 I nch l ong suspensi on cabl es i n an effort
to i ncrease the peri od of the osci l l ati on, and i mprove the ti me measurement
accuracy. Conceptual l y thi s was a good i dea, but i t turned out to be i mpracti cal
si nce the l ong cabl es had a tendency to wrap around each other. Thi s proved to be
an unmanageabl e si tuati on, and the support cabl e l engths were reduced to 33.73
I nches. Duri ng the acqui si ti on of test data, the cyl i nder was twi sted about 20
ci rcumstanti al l y from rest, and rel eased. The peak of the moti on at one extrem-
i ty was vi sual l y si ghted, and a stopwatch was used to measure the ti me requi red
for compl ete back and forth cycl es. The test data i s summari zed as fol l ows:
Twisting Run #1 13.83 Seconds for 10 cycles
Twisting Run #2 14.23 Seconds for 10 cycles
Twisting Run #3 14.15 Seconds for 10 cycles
Twisting Run #4 13.94 Seconds for 10 cycles
Twisting Run #5 14.02 Seconds for 10 cycles
Twisting Run #6 13.89 Seconds for 10 cycles
Twisting Run #7 14.01 Seconds for 10 cycles
Twisting Run #8 14.11 Seconds for 10 cycles
Twisting Run #9 14.05 Seconds for 10 cycles
Twisting Run #10 13.95 Seconds for 10 cycles
Total Time = 140.18 Seconds for 100 cycles
Average Time for 1 Cycle = 1.4018 Seconds = Period
Fig. 315 Twisting Test To
Determine Mass Polar
Moment Of Inertia Of A
Solid Brass Cylinder
Brass
Cylinder
Weight
(W=36.0 Lbs.)

S
u
s
p
e
n
s
i
o
n

C
a
b
l
e

L
e
n
g
t
h
(
L
s
-
c
=
3
3
.
7
3
"
)
A
x
i
a
l

C
e
n
t
e
r
l
i
n
e
Suspension
Cable
Radius
Twisting
Oscillation
30 to 40
(R
s-c
=3.00")
Inertia Considerations and Calculations 89
Thi s i s a much smoother test than the rocki ng i nerti a previ ousl y di scussed.
The number of mi scounts and aborted runs were substanti al l y reduced, and
approxi matel y fteen runs were made to col l ect the data shown i n the above tab-
ul ar summary. Fi ve of the ti mi ng runs were not used due to obvi ous errors i n
data accumul ati on. The ten consi stent test runs reveal an average peri od of
1.4018 Seconds. Thi s was consi dered to be a consi stent val ue for thi s experi men-
tal procedure, and the mass pol ar moment of i nerti a may be computed from
equati on (3-36) as fol l ows:
The experi mental pol ar i nerti a from thi s twi sti ng procedure of 0.478 i s
qui te cl ose to the previ ousl y cal cul ated val ue of 0.464 Pound-I nch-Seconds
2
. The
3% devi ati on i s qui te acceptabl e for most rotor dynami cs cal cul ati ons. Thi s i s
parti cul ari ty true for smal l er components that are stacked on a shaft to achi eve
a nal rotor assembl y. I t shoul d be recogni zed that both the rocki ng and the
twi sti ng i nerti a tests have thei r own domai n of appl i cati on that i s dependent on
the si ze and geometry of the machi ne el ement.
Just as the wei ghts of i ndi vi dual components are summed up to determi ne
a total rotor wei ght, the i nerti a of the component pi eces may be added to deter-
mi ne the overal l rotor pol ar i nerti a. The ori gi n of the i nerti a val ues may be from
cal cul ati ons of dened geometri es, or from experi enti al l y determi ned i nerti a val -
ues. I n any case, as l ong as the engi neeri ng uni ts and the i nerti a axi s are com-
mon, the numeri c i nerti a val ues may be summed up to determi ne the mass pol ar
moment of i nerti a for the enti re rotati ng assembl y.
I n some i nstances, there i s mi ni mal opportuni ty to determi ne the i nerti a of
rotor components si nce the uni t cannot be di sassembl ed or unstacked. I n these
si tuati ons, the general i nerti a characteri sti cs may be esti mated based upon
avai l abl e di mensi ons and probabl e materi al s of constructi on. I n other cases, the
compl exi ty of the rotor may not al l ow a reasonabl e segmentati on and esti mati on
of i nerti a properti es. Thi s i s parti cul arl y true for rotors that are constructed of
mul ti pl e materi al s, pl us rotors that contai n compl i cated geometri c congura-
ti ons. I n these i nstances, another experi mental techni que may be empl oyed to
determi ne the overal l mass pol ar moment of i nerti a of the rotor.
Thi s techni que i s based upon the fami l i ar col l ege physi cs experi ment
depi cted i n Fi g. 3-16. I n thi s di agram, a cyl i nder or drum i s mounted i n ri gi d
beari ngs that al l ow rotati on of the cyl i nder, but restri ct any l ateral or transl ati on
movement of the cyl i nder. A cord i s wrapped around the cyl i nder at a shaft
radi us of R
s
. I t i s assumed that thi s cord i s of i nsi gni cant wei ght and di ameter,
and that i t wi l l not stretch wi th the appl i cati on of axi al tensi on. Next, a known
wei ght (mass M) i s attached to the end of the cord, and al l owed to free fal l . The
J
zz
W R
s c
2
Peri od
2

4
2
L
s c

------------------------------------------------------- =
J
zz
36.0 Pounds 3.00 Inches ( )
2
1.4018 Seconds ( )
2

4
2
33.73 Inches
---------------------------------------------------------------------------------------------------------------------- 0.478 Pound-Inch-Second
2
= =
90 Chapter-3
ti me T requi red to fal l a di stance D i s measured wi th a stop watch. The experi -
ment normal l y consi sts of determi ni ng the cyl i nder mass pol ar moment of i ner-
ti a J
mass
based on the four known quanti ti es of radi us R
s
, mass M, fal l di stance
D, and the average fal l ti me T.
Thi s basi c physi cs probl em may be sol ved by constructi ng free body di a-
grams of the cyl i nder and the fal l i ng mass, devel opi ng equati ons of moti on, and
then sol vi ng for the pol ar i nerti a term. Another way to achi eve the same resul t i s
by performi ng an energy bal ance on the system shown i n Fi g. 3-16. Conservati on
of energy requi res that the change i n potenti al energy i s equal to the change i n
ki neti c energy. I n thi s case, the change i n potenti al energy i s si mpl y the el eva-
ti on change i n the mass (M x G x D). The overal l change i n ki neti c energy i s com-
posed of a change i n transl ati onal energy of the fal l i ng mass (M x V
2
/2), pl us the
change i n the rotati onal ki neti c energy of the rotor (J
mass

2
/2). These tradi ti onal
physi cal concepts may be represented mathemati cal l y i n the fol l owi ng manner:
(3-37)
The vel oci ty V i n equati on (3-37) represents the average vel oci ty of the fal l -
i ng mass M. I n al l cases, the fal l i ng wei ght W i s equal to the mass M ti mes the
accel erati on of gravi ty G. Substi tuti ng the quanti ty W/ G for the mass M pro-
duces the fol l owi ng:
Fig. 316 Traditional
Mechanical Arrangement
For Polar Moment Of Iner-
tia Experiment Based On
Falling Mass Attached To
A Rotating Cylinder
R
a
d
i
u
s
(
R
s
)
Fall Distance (D)
during Fall Time (T)
Mass
(M)
Cylinder with Mass Polar
Moment of Inertia (J
mass
)
Angular
Velocity ()
M G D
M V
2

2
-------------------
J
mass

2

2
----------------------------- + =
Inertia Considerations and Calculations 91
Thi s may now be sol ved for the mass pol ar moment of i nerti a J
mass
as:
(3-38)
The average vel oci ty V of the mass M duri ng free fal l may be determi ned
from the fundamental equati ons of moti on for recti l i near moti on wi th constant
accel erati on. More speci cal l y, al l physi cs books agree that the average vel oci ty
mul ti pl i ed ti mes the drop ti me T wi l l determi ne the fal l di stance D as fol l ows:
(3-39)
I f the i ni ti al starti ng vel oci ty V
o
i s equal to zero, then equati on (3-39) may
be si mpl i ed to represent the vel oci ty V i n terms of the drop di stance D, and the
total el apsed drop ti me T as shown i n the next equati on:
(3-40)
The l ast conversi on to be performed consi sts of an expressi on for the angu-
l ar vel oci ty of the cyl i nder i n terms of the known experi mental parameters.
Si nce tangenti al vel oci ty V di vi ded by the radi us R
s
i s equal to the angul ar vel oc-
i ty , equati on (3-40) may be used to determi ne i n the fol l owi ng manner:
(3-41)
Equati on (3-38) for the pol ar i nerti a may now be cl ari ed i nto the known
experi mental data by substi tuti ng equati ons (3-40) and (3-41) back i nto equati on
(3-38), and performi ng the fol l owi ng si mpl i cati on of terms:
W D
W V
2

2 G
-------------------
J
mass

2

2
----------------------------- + =
2W D
W V
2

G
------------------- J
mass

2
+ =
J
mass

2
2W D
W V
2

G
------------------- =
J
mass
W

2
------ 2D
V
2
G
-------

' ;

=
D
V
o
V +
2
------------------
,
_
T =
V
2D
T
-------- =

V
R
s
-------
2D
R
s
T
----------------- = =
92 Chapter-3
The cancel l ati on of common terms resul ts i n equati on (3-42).
(3-42)
where: J
mass
= Mass Polar Moment of Inertia (Pounds-Inches-Seconds
2
)
W = Falling Weight (Pounds)
R
s
= Shaft Radius (Inches)
D = Drop Distance (Inches)
T = Drop Time (Seconds)
G = Acceleration of Gravity (= 386.1 Inches / Second
2
)
Wi th a few mi nor modi cati ons, thi s tradi ti onal physi cs experi ment may be
used to determi ne the pol ar i nerti a of a compl ete rotor assembl y. One i mpl emen-
tati on of thi s concept i s the techni que by Mi chael Cal i strat
10
that i s i l l ustrated
i n Fi g. 3-17. Thi s di agram depi cts a rotor resti ng i n the rol l ers of a two beari ng
shop bal anci ng machi ne. To mi ni mi ze errors, the rotor assembl y shoul d be prop-
erl y bal anced, and the rol l ers i n the bal anci ng machi ne shoul d be i n good condi -
ti on. As shown i n Fi g. 3-17, an overhead pul l ey i s supported by an external
structure, or the overhead crane that i s used to move rotors i n and out of the bal -
anci ng machi ne. The di ameter of the overhead pul l ey does not matter, but thi s
pul l ey shoul d be i n good condi ti on, and i t shoul d turn easi l y. The pul l ey redi rects
the gravi tati onal force to pul l upward on the rotor, but i t does not di mi ni sh the
effect of the fal l i ng mass. A pl asti c covered brai ded steel cabl e i s attached to the
shaft wi th Duct tape at a radi us of R
shaft
, and wrapped around the shaft for ten
or fteen turns. I ts probabl y a good i dea to keep the cabl e away from any of the
rotor beari ng journal s, or the rotor secti ons observed by proxi mi ty probes. The
thi ckness of thi s cabl e i s general l y smal l compared to the di ameter of the shaft.
Agai n, as shown i n Fi g. 3-17, the steel cabl e passes overhead, through the pul l ey,
and then i s secured wi th a pai r of verti cal wei ghts. The rst wei ght i s a tare
10
Mi chael M. Cal i strat, Flexible Couplings, their design selection and use, (Houston: Carol i ne
Publ i shi ng, 1994), p 464.
J
mass
W

2
------ 2D
V
2
G
-------

' ;

=
J
mass
W
2 D R
s
T ( )
2
------------------------------------------- 2D
2 D T ( )
2
G
------------------------------

' ;

=
J
mass
W R
s
2
T
2

4 D
2
-------------------------------- 2D
4 D
2
G T
2

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

' ;

=
J
mass
W R
s
2
2D
T
2
4 D
2

----------------
4 D
2
G T
2

-----------------
T
2
4 D
2

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

' ;

=
J
mass
W R
s
2

T
2
2D
--------
1
G
----

' ;

=
Inertia Considerations and Calculations 93
wei ght that i s used to counterbal ance the rotor, and keep the cabl e taunt. Thi s
tare wei ght shoul d just barel y al l ow the rotor to rotate, and overcome system
fri cti on.
The tare wei ght i s determi ned by tri al and error, and smal l changes i n thi s
wei ght wi l l have a bi g i nuence on the rotor. Si nce thi s work i s performed i n a
bal anci ng machi ne, i t makes sense to use bal anci ng cl ay to get a good tare
wei ght establ i shed. The second wei ght W i s the experi mental wei ght that actu-
al l y turns the rotor. I t may be compromi sed of bal anci ng cl ay, a stack of washers,
or any reasonabl e combi nati on thereof. I n al l cases, the di agnosti ci an must have
the abi l i ty to accuratel y wei gh thi s mass W at the end of the test.
As wi th the two previ ous i nerti a measurements, the majori ty of the ti me i s
spent i n setti ng up the test. The test executi on requi res mi ni mal ti me, and thi s
rotor i nerti a procedure i s no excepti on. Fol l owi ng the setup per Fi g. 3-17, the
cabl e i s wound up on the shaft, and the wei ght i s natural l y rai sed i n el evati on.
At thi s ti me the shaft i s rel eased, and the ti me T for the wei ght to fal l through a
predetermi ned di stance D i s measured wi th a stopwatch. The i nerti a cal cul ati on
descri bed by equati on (3-42) may be used di rectl y. However, i mproved accuracy
wi l l resul t by i ncl udi ng the cabl e radi us R
cable
wi th the shaft radi us R
shaft
as rec-
ommended by Cal i strat. Thi s addi ti on i s not requi red for very thi n cabl es, but i t
does i mprove accuracy for thi cker cabl es, si nce the moment arm i s real l y the sum
of the shaft pl us the cabl e radi us. Hence, (3-42) may be rewri tten as fol l ows:
(3-43)
The nomencl ature i n equati on (3-43) has been expanded to be more under-
standabl e, and yet mai ntai n consi stency wi th the general terms used wi thi n thi s
text. Agai n, the experi mental wei ght W i s i n pounds, the shaft radi us R
shaft
, the
Fig. 317 Mechanical Test
Arrangement For Polar
Moment Of Inertia Measure-
ment Of Full Rotor Assem-
bly Mounted In A Shop
Balancing Machine, Tech-
nique by M. M. Calistrat
M
achine R
otor M
ounted in
Tw
o B
earing B
alancing M
achine
Overhead Pulley
Shaft With
Radius (R
shaft
)
Weight (W)
Tare Weight
Axial Shaft
Centerline
Cable With
Radius (R
cable
)
J
mass
Wei ght R
shaft
R
cabl e
+ ( )
2

Ti me
2
2 Di s ce tan
-----------------------------------
1
G
----

' ;

=
94 Chapter-3
cabl e radi us R
cable
, and the fal l Distance are al l i n i nches. The measured Time for
the free fal l through the predetermi ned di stance i s dened as the ti me i n sec-
onds, and G i s the gravi tati onal constant.
As wi th any experi ment, the val i di ty shoul d be checked wi th a test mandrel
of known characteri sti cs. For thi s test, a secti on of 4140 steel was sel ected wi th a
densi ty of 0.283 Pound/I nch
3
. The l ength L of the speci men was 59.81 i nches, and
the ends were squared up i n a l athe. I n addi ti on, the outer di ameter D of the test
pi ece was cl eaned up to nomi nal l y 5.980 i nches across the l ength of the el ement.
Once more equati on (3-7) i s used to compute the mandrel wei ght:
Thi s cal cul ated wei ght of 475 pounds was l ower than the uncal i brated shop
scal e readi ng of 490 pounds. Al though a better agreement i n wei ghts woul d have
been comforti ng, the 15 pound devi ati on was consi dered to be wi thi n the mea-
surement accuracy of the shop scal e. Next, the mass pol ar moment of i nerti a of
thi s steel shaft may be computed wi th equati on (3-32) i n the fol l owi ng manner:
Pri or to testi ng, the runout (eccentri ci ty) al ong the l ength of the mandrel
was conrmed to be l ess than 1.0 Mi l . I n addi ti on, a check bal ance was per-
formed, and the resi dual unbal ance was mi nor. For the actual test runs, a fal l
di stance of 60 I nches was establ i shed on a verti cal reference stand. The shaft
radi us R
shaft
was 2.99 I nches, and the cabl e di ameter was 1/8, for a R
cable
of 0.06
i nches. The tare wei ght was found to be 0.696 Pounds (315.7 grams). The experi -
mental wei ght W was adjusted unti l the free fal l ti me was about 5 seconds. Thi s
requi red a total of 3.43 Pounds (1,556 grams). The ti med test data was obtai ned
by rel easi ng the shaft and measuri ng the ti me requi red for the wei ght to fal l 60
i nches. A stopwatch was used to measure the ti me, and the test data i s fol l ows:
Falling Weight Run #1 4.52 Seconds
Falling Weight Run #2 4.54 Seconds
Falling Weight Run #3 4.65 Seconds
Falling Weight Run #4 4.54 Seconds
Falling Weight Run #5 4.56 Seconds
Total Time = 22.81 Seconds for 5 Drops
Average Time for 1 Drop of 60 Inches = 4.562 Seconds
W
sol i d
L D
2

4
------------------------------------- =
W
sol i d
59.81 Inches 0.283 Pounds/Inch
3
5.98 Inches ( )
2

4
--------------------------------------------------------------------------------------------------------------------------------- 475 Pounds = =
J
mass
sol i d
L D
4

32 G
------------------------------------- =
J
mass
sol i d
59.81 In. 0.283 Pounds/In.
3
5.98 In. ( )
4

32 386.1 Inches/Second
2

------------------------------------------------------------------------------------------------------------- 5.504 Pound-Inch-Second


2
= =
Inertia Considerations and Calculations 95
Ei ght di fferent test runs were made, and the obvi ousl y i ncorrect ti mes were
di scarded. The above array was qui te consi stent, and there was good condence
i n both the val i di ty and the accuracy of thi s experi mental data. Hence, equati on
(3-43) was now appl i ed to compute the measured mass pol ar moment of i nerti a:
Note that there i s excel l ent agreement between the experi mental i nerti a of
5.450 and the cal cul ated pol ar i nerti a of thi s steel shaft of 5.504 Pound-I nch-
Second
2
. The actual devi ati on of 1% i s consi dered to be qui te acceptabl e accuracy
for thi s type of measurement. I t i s often desi rabl e to perform thi s test wi th di ffer-
ent wei ghts, and di fferent fal l di stances to veri fy the consi stency of the proce-
dure. I n al l cases, the test wei ght shoul d be substanti al l y l arger than the tare
wei ght, and pl ease recal l that the i ni ti al starti ng vel oci ty i s zero at the start
ti me. For bi gger and heavi er rotors that have a tendency to sag wi thi n the bal -
anci ng machi ne, consi derati on shoul d be gi ven to rol l i ng the rotors for several
hours before attempti ng thi s type of i nerti a test. I n addi ti on, di al i ndi cator mea-
surements shoul d be made at the rotor mi dspan before begi nni ng the i nerti a
tests, and after the concl usi on of the l ast run. Any appreci abl e sag to the shaft of
the rotor under test mi ght negate the test resul ts, and requi re a repeat of the
sl ow rol l and the test procedure.
Fi nal l y, i t shoul d be recogni zed that other experi mental procedures exi st for
determi nati on of mass moment of i nerti a. I n addi ti on, many computati on pro-
grams provi de the capabi l i ty for three-di mensi onal cal cul ati ons of the i nerti a
properti es of compl ex bodi es. As al ways, the machi nery di agnosti ci an shoul d be
aware of the engi neeri ng uni ts, and test cases of known geometri es shoul d
al ways be run to veri fy the techni que.
J
mass
Wei ght R
shaft
R
cabl e
+ ( )
2

Ti me
2
2 Di s ce tan
-----------------------------------
1
G
----

' ;

=
J
mass
3.43 Pounds 2.99 0.06 Inches + ( )
2

4.562 Seconds ( )
2
2 60 Inches
----------------------------------------
1
386.1 Inches/Sec
2
-----------------------------------------

' ;

=
J
mass
3.43 9.302 0.1734 0.0026 ( ) 5.450 Pound-Inch-Second
2
= =
96 Chapter-3
DAMPING INFLUENCE
Three basi c types of dampi ng occur i n a machi nery system. These dampi ng
types are commonl y referred to as viscous dampi ng, coulomb dampi ng, and solid
or structural dampi ng.
Viscous damping i s encountered by sol i d bodi es movi ng through a vi scous
ui d. I n thi s type of dampi ng, the resi stance force i s proporti onal to the vel oci ty
of the movi ng object. As an exampl e of vi scous dampi ng, consi der the si tuati on of
a cook sti rri ng a pot of soup versus a pot of mol asses. I t i s sel f-evi dent that sti r-
ri ng the mol asses i s consi derabl y more di fcul t due to the thi ckness and hi gher
vi scosi ty of the mol asses as compared to the thi n soup. The requi red force i s
di rectl y proporti onal to the vel oci ty of the sti rri ng spoon. I n most cases the cook
woul d sti r the mol asses at a much sl ower rate than he woul d the soup, si mpl y
because i t woul d take too much strength or energy to sti r i t rapi dl y.
The same type of physi cal property, i .e., vi scous dampi ng, i s encountered i n
the beari ngs and oi l seal s of l arge rotati ng machi nes. I n thi s case, the dampi ng i s
provi ded by the l ubri cati ng oi l , and the rotati ng shaft i s the ri gi d body movi ng
through the vi scous ui d. The process ui ds handl ed by the machi ne al so provi de
dampi ng to the rotor system. For l i qui d handl i ng machi nes such as pumps and
hydraul i c turbi nes, thi s i s si gni cant. However, for gas handl i ng machi nes such
as turbi nes or centri fugal compressors thi s i s a mi nor consi derati on.
Fig. 318 Calculated
Bode Plot Of Unbalance
Response For A Simple
Mechanical System With
Variations In Damping
180
150
120
90
60
30
0
P
h
a
s
e

L
a
g

(
D
e
g
r
e
e
s
)
=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
A
m
p
l
i
t
u
d
e

R
a
t
i
o
Critical Speed Frequency Ratio
Low Damping, =0.1
=0.2
=0.5
=2.0
=1.0
()
Damping Influence 97
The next type of dampi ng i s coulomb damping, whi ch ari ses from the
sl i di ng of one dry surface upon another (rub condi ti on). The coul omb fri cti on
force i s nearl y constant, and i t depends on the nature of the sl i di ng surfaces, and
the perpendi cul ar pressure between the surfaces. Thi s type of force i s general l y
domi nant i n damped systems duri ng the nal stages of moti on when other types
of dampi ng become negl i gi bl e.
The thi rd category of dampi ng i s often referred to as sol i d or structural
damping. Thi s i s due to i nternal fri cti on wi thi n the materi al , and i t di ffers from
vi scous dampi ng i n that i t i s i ndependent of frequency, and proporti onal to the
maxi mum stress of the vi bratory cycl e. Si nce stress and strai n are proporti onal
i n the el asti c range, i t can be stated that sol i d dampi ng force i s proporti onal to
deecti on. Structural dampi ng i n rotati ng machi nery i s smal l when compared
wi th vi scous dampi ng, but i t does exi st.
The major contri buti on that posi ti ve dampi ng makes to a rotati ng machi n-
ery system i s the di ssi pati on of energy. Thi s i nuence i s most dramati cal l y i l l us-
trated when a mechani cal system passes through a resonance as i n Fi g. 3-18.
Thi s cal cul ated Bode di agram was dupl i cated from chapter 2. The unbal ance
response pl ot of frequency rati o versus ampl i tude rati o and phase of a damped
system provi des a good perspecti ve of the actual i nuence of dampi ng. The fam-
i l y of curves i n thi s di agram are pl otted wi th a dampi ng rati o (or dampi ng factor)
extendi ng from 0.1 to 2.0. Recal l that thi s rati o i s the actual dampi ng di vi ded
by the cri ti cal dampi ng. Note, that wi th a l i ghtl y damped system of =0.1, the
response at the resonance i s qui te hi gh. Thi s transl ates to the fact that there i s
l i ttl e energy di ssi pati on under thi s condi ti on. The system i s under damped, and
i t i s suscepti bl e to i nstabi l i ty due to a l ack of an energy di ssi pati on. Conversel y,
when the dampi ng factor i s l arge, =2.0, the system i s over damped, response
through the resonance i s restrai ned, and overal l stabi l i ty of the system i s hi gh.
I n many i nstances of free vi brati on, rotor i nstabi l i ty can be rel ated to a l ack
of dampi ng. The system dampi ng may be assessed by exami ni ng the cri ti cal
speed response on Bode pl ots (synchronous 1X vectors versus speed). Typi cal l y,
the ampl i cati on factor Q through the resonance i s used to quanti fy the severi ty
of the resonance, pl us the dampi ng rati o. A l arge ampl i cati on factor i s associ -
ated wi th a poorl y damped, hi gh ampl i tude resonance. Conversel y, a l ow ampl i -
cati on factor i s general l y associ ated wi th a wel l damped resonance, that di spl ays
smal l ampl i tudes at the peak of the resonance.
Extracti on of the ampl i cati on factor from the Bode pl ots may be performed
i n several di fferent ways. Unfortunatel y, there i s di sagreement wi thi n the tech-
ni cal communi ty as to the best manner to obtai n thi s i nformati on from the mea-
sured vi brati on data. For the sake of compl eteness, three separate approaches
for determi nati on of thi s di mensi onl ess ampl i cati on factor wi l l be presented.
The rst techni que consi sts of vi sual l y compari ng the measured vi brati on
response data wi th a set of cal cul ated curves, and esti mati ng a dampi ng rati o .
The rotor ampl i cati on factor for the speci c resonance i s then computed di rectl y
from the dampi ng rati o. As a practi cal exampl e of an i ndustri al machi ne, con-
si der the compressor Bode pl ot shown i n Fi g. 3-19 (same as Fi g. 2-21). Thi s vari -
abl e speed vi brati on data i s runout compensated, and i t exhi bi ts a cl ean
98 Chapter-3
transi ti on through a si ngl e cri ti cal speed. I f thi s data i s vi ewed i n conjuncti on
wi th the computed fami l y of curves shown i n Fi g. 3-18, a sui tabl e dampi ng rati o
may be sel ected. Speci cal l y, the shape of the ampl i tude versus speed, and the
phase versus speed are compared between the measured (Fi g. 3-19) and the cal -
cul ated pl ots (Fi g. 3-18). I t i s reasonabl e to concl ude that the cal cul ated pl ot wi th
a dampi ng rati o of 0.2 i s the cl osest match to the measured machi ne response
data. From the previ ousl y devel oped equati on (2-91), the ampl i cati on factor
(i .e., ampl i tude rati o) Q may be determi ned from the dampi ng rati o as fol l ows:
(3-44)
Substi tuti ng the previ ousl y i denti ed val ue for the dampi ng rati o of =0.2,
the fol l owi ng resul t i s obtai ned:
Thi s vi sual compari son between curve shapes provi des any easy way to
esti mate Q for a cl eanl y dened resonance. However, for more compl ex response
characteri sti cs, other techni ques are avai l abl e. For i nstance, the second
approach for computati on of the ampl i cati on factor i s deri ved from el ectri cal
engi neeri ng termi nol ogy. Thi s techni que has al so been adopted by vari ous
mechani cal standards organi zati ons such as the Ameri can Petrol eum I nsti tute.
I n thi s procedure, the center frequency of the resonance i s di vi ded by the reso-
nance bandwi dth at the Half Power Point i n accordance wi th the fol l owi ng:
(3-45)
Fig. 319 Measured Bode
Plot Of A Centrifugal Com-
pressor Startup
Q
Curve Fi t
Ampl i fi cati on Factor Ampl i tude Rati o
1
2
------------ = = =
Q
Curve Fi t
1
2
------------
1
2 0.2
----------------
1
0.4
------- 2.5 = = = =
Q
Hal f Power
Center Frequency
Frequency Bandwi dth @-3dB
--------------------------------------------------------------------------------------- =
Damping Influence 99
The hal f power poi nt i s equi val ent to the ampl i tude that i s -3 dB down from
the peak of the resonance. I n terms of l i near scal es, an ampl i tude change of -3 dB
i s equal to 0.707 ti mes the peak vi brati on ampl i tude. Thi s l evel denes the spe-
ci c poi nt on the resonance response curve where the frequency bandwi dth i s
measured. Al though thi s cal cul ati on procedure may sound compl i cated, the
actual executi on i s fai rl y strai ghtforward.
I f the Bode pl ot from the l ast exampl e i s reconsi dered, the ampl i cati on fac-
tor usi ng thi s hal f power approach may be computed as shown i n Fi g. 3-20.
Wi thi n thi s rendi ti on of the Bode, vari ous features have been i denti ed to
enhance thi s di scussi on. Speci cal l y, the center of the resonant peak (cri ti cal
speed) has been i denti ed at 6,100 RPM. The vi brati on ampl i tude at the peak of
thi s resonance i s 1.95 Mi l s,
p-p
. The hal f power poi nt i s determi ned by mul ti pl yi ng
0.707 ti mes the maxi mum of 1.95 Mi l s,
p-p
to obtai n a -3 dB poi nt of 1.38 Mi l s,
p-p
.
The response curve wi dth at thi s ampl i tude i s 900 RPM. Stated i n another way,
the frequency equal to an up sl ope ampl i tude of 1.38 Mi l s,
p-p
i s 5,800 RPM. The
frequency equal to a down sl ope ampl i tude of 1.38 Mi l s,
p-p
i s 6,700 RPM. The
bandwi dth at thi s -3 dB ampl i tude i s 6,700 mi nus 5,800, whi ch i s equal to 900
RPM. Substi tuti ng these val ues for center frequency and bandwi dth i nto the pre-
vi ous equati on (3-45) the next resul t i s obtai ned:
The cal cul ated ampl i cati on factor usi ng thi s second approach (6.8) i s con-
si derabl y hi gher than the rst method that produced a Q of 2.5. However, thi s
second method may be used for compl ex resonant response condi ti ons where
mul ti pl e cri ti cal speeds appear cl ose together. For i nstance, a machi ne that has a
spl i t cri ti cal (e.g., hori zontal fol l owed by a verti cal mode) woul d be di fcul t to
Fig. 320 Amplication
Factor From A Bode Plot
Using the Half Power Point
Calculation Technique
Q
Hal f Power
Center Frequency
Frequency Bandwi dth @-3dB
---------------------------------------------------------------------------------------
6 100 Rpm ,
900 Rpm
--------------------------- 6.8 = = =
100 Chapter-3
handl e wi th the rst approach. Whereas, the second scheme woul d al l ow a better
quanti cati on of the Q for each response peak.
One objecti on to the second approach i s that Q vari es wi th changes to the
center frequency of the resonance. For i nstance, i f the shape of the response
curve i s mai ntai ned, and the center frequency i s reduced to 3,000 RPM, the Q
drops to 3.3 (=3,000/900). By the same token, i f the cri ti cal speed occurs at
12,000 RPM, the Q now i ncreases to 13.3 (=12,000/900). Thi s change i n ampl i -
cati on factors wi th a constant shape to the response peak may be qui te confus-
i ng, as wel l as contradi ctory to the desi red deni ti on.
The thi rd approach for cal cul ati on of the ampl i cati on factor consi sts of
di vi di ng the ampl i tude at the resonance by the ampl i tude at a speed far above
the resonance. Agai n referri ng to the same Bode pl ot exampl e, thi s approach i s
i l l ustrated i n Fi g. 3-21. As before, the magni tude of 1.95 Mi l s,
p-p
at the transl a-
ti onal cri ti cal speed of 6,100 RPM i s i denti ed. Above thi s resonance, the phase
and ampl i tude enter a pl ateau regi on where the 1X vector remai ns reasonabl y
constant as speed i ncreases. Thi s i s normal behavi or, and the synchronous
response above a cri ti cal i s typi cal l y at unti l some external force, or a hi gher
order resonance i nuences the moti on. Wi thi n the context of thi s exampl e, the
regi on at 8,000 RPM i s sufci entl y removed from the cri ti cal speed, and the rota-
ti onal speed vi brati on ampl i tude i n thi s regi on i s 0.85 Mi l s,
p-p
. The ampl i cati on
factor determi ned wi th thi s thi rd techni que i s computed by si mpl y di vi di ng the
maxi mum vi brati on ampl i tude at the cri ti cal speed by the ampl i tude measured
wel l above the resonance i n accordance wi th the fol l owi ng expressi on:
(3-46)
Fig. 321 Amplication
Factor from a Bode plot
Using A Simple Amplitude
Ratio Between the Reso-
nance and a Speed Far
Above the Resonance
Q
Resonance/ Above
Ampl i tude
At Resonance
Ampl i tude
Above Resonance
--------------------------------------------------------------------------- =
Damping Influence 101
Usi ng the vi brati on ampl i tudes i denti ed on Fi g. 3-21, the ampl i cati on
factor determi ned by thi s method i s easi l y computed as:
Thi s val ue of 2.3 i s i ndi cati ve of an adequatel y damped machi ne, and a sta-
bl e mechani cal system. Due to the di fferences between machi nes, i t i s di fcul t to
preci sel y categori ze the vari ati ons between good and bad Q factors. However, i t
i s reasonabl e to i denti fy several categori es of ampl i cati on factors versus damp-
i ng and stabi l i ty characteri sti cs. Summary Tabl e 3-1 attempts to provi de some
real i sti c gui del i nes for these i nterrel ated parameters:
Machi nes that fal l i nto the l ast category may be potenti al l y dangerous, and
may not survi ve a mai den startup. Machi nes wi thi n thi s group often requi re
extensi ve modi cati ons to beari ngs and/or seal s to i ncrease system dampi ng. I n
some i nstances, modi cati ons such as the i nstal l ati on of squeeze l m damper
beari ngs may be requi red to provi de adequate dampi ng for the system.
I t shoul d al so be noted that the ampl i cati on factors referred to herei n are
associ ated wi th shaft vi brati on measurements where the oi l l m vi scous damp-
i ng i s domi nant (i .e., between the journal and beari ng). Casi ng vi brati on mea-
surements woul d typi cal l y be more recepti ve to structural dampi ng, and not so
sensi ti ve to vi scous dampi ng i n the beari ngs. Thus, casi ng measurements gener-
al l y di spl ay Qs that are much hi gher than shaft measurements, si mpl y due to
the l ack of dampi ng wi thi n the casi ng and the support structure.
Another consi derati on that must be appl i ed to any eval uati on of vari abl e
speed data i s the accel erati on rate, or more speci cal l y the rate of speed change
of the rotati ng system. Duri ng machi nery trai n coastdowns, there i s usual l y mi n-
i mal , i f any, control of the decel erati on. However, duri ng startup, the rate of rotor
accel erati on i s often control l abl e on vari abl e speed dri vers such as steam tur-
bi nes. On machi nes wi th ol der control systems, the startup rate i s often depen-
dent upon the ski l l and knowl edge of the operator handl i ng the tri p & throttl e
val ve. On newer speed control systems, the startup rate i s usual l y control l ed by
an el ectroni c governor wi th predetermi ned startup speed ramps. Unfortunatel y,
some el ectroni c governors suppl i ers are not wel l versed i n the acceptabl e startup
rates for vari ous types of machi nery trai ns. These vendors often set abnormal l y
Table 31 Shaft Amplication Factor Versus Damping And Anticipated Rotor Stability
Amplication Factor Damping Stability
Q = < 2 Wel l Damped Extremel y Stabl e
Q = 2 to 8 Adequatel y Damped Normal Stabi l i ty
Q = 8 to 15 Poorl y Damped Margi nal Stabi l i ty
Q = > 15 I nsufci ent Dampi ng I nherentl y Unstabl e
Q
Resonance/ Above
Ampl i tude
At Resonance
Ampl i tude
Above Resonance
---------------------------------------------------------------------------
1.95 Mils
p p
0.85 Mils
p p
------------------------------ 2.3 = = =
102 Chapter-3
fast startup rates i n an effort to snap the rotor(s) through a cri ti cal speed range.
Thi s tendency has resul ted i n machi nery damage, and more than one rotor fai l -
ure. Hence, the end user shoul d al ways veri fy that the proposed startup accel er-
ati on rates are reasonabl e for the machi nery i n questi on.
I ntui ti vel y, the passage of a rotor system through a cri ti cal speed regi on
shoul d be performed i n a di rect and knowl edgeabl e manner. I f the startup rate i s
i nordi natel y sl ow, the machi ne may hang up i n a resonance, and cause mechani -
cal damage due to the hi gh vi brati on l evel s. An exampl e of thi s type of occur-
rence i s bri ey di scussed i n the turbi ne generator case hi story 39 i n chapter 11.
On the other hand, i f the speed accel erati on rate through a resonance i s exces-
si ve, the machi ne may sel f-destruct after i t reaches operati ng speed and
attempts to rebal ance i tsel f about the mass center. Al though thi s type of occur-
rence i s rare, i t i s certai nl y avoi dabl e, and total l y unnecessary.
The speed transi ti on rate through a resonance wi l l al ter the characteri sti cs
of the vi brati on response data. A sl ow startup wi l l show a hi gher peak at the res-
onance, combi ned wi th a broader bandwi dth. Conversel y, a rapi d startup wi l l
produce a l ower peak ampl i tude at the cri ti cal speed, pl us a smal l er resonant
bandwi dth. Thi s attenuated response characteri sti c has erroneousl y l ed many
i ndi vi dual s i nto a fal se sense of securi ty by rampi ng through cri ti cal speeds at a
hi gh rate. Hence, the di agnosti ci an must be aware of the transi ti on rate through
the system cri ti cal speed(s), and any eval uati on of the resonant characteri sti cs
(e.g., Q) shoul d be wei ghed by thi s speed change rate.
I t i s i mpossi bl e to ful l y quanti fy proper startup accel erati on rates for al l
cl asses of machi nery operati ng wi th vari ous types of dri ves, and control systems.
However, to provi de some gui dance i n thi s area, Tabl e 3-2 of acceptabl e startup
rates i s offered for consi derati on. Thi s tabl e summari zes el d measurements on
a vari ety of vari abl e speed trai ns. I n al l cases, the general machi ne type i s i ndi -
cated, the rotor wei ght i s shown, and the maxi mum machi ne speed i s l i sted. The
l ast col umn provi des typi cal peak startup accel erati on rates (general l y through
Table 32 Various Machine Types Versus Acceptable Startup Acceleration Rates
Machine Type
Rotor Weight
(Pounds)
Maximum Speed
(RPM)
Acceleration Rate
(RPM/Second)
Expander/Pi ni on 60 34,000 550 to 750
Pi ni on/Compressor - Smal l 300 14,000 350 to 450
Pi ni on/Compressor - Medi um 1,500 11,000 80 to 100
Steam Turbi ne/Compressor 7,500 8,000 70 to 80
Gas Turbi ne/Compressor 25,000 5,400 20 to 30
Steam Turbi ne/Compressor 29,000 3,800 40 to 50
Steam Turbi ne/Compressor 54,000 4,000 35 to 40
Hydro Turbi ne/Generator 180,000 420 5 to 10
Damping Influence 103
resonances). I t shoul d be recogni zed that other porti ons of the startup sequence
often occur at much l ower rates. Thus, the compressor that exhi bi ts a 400 RPM
per Second ramp through a cri ti cal may be preceded by a regi on wi th a rate of
onl y 20 to 30 RPM per Second.
Tabl e 3-2 represents conservati ve speed accel erati on rates. The observed
trend of l i ght wei ght, hi gh speed rotors that accel erate rapi dl y i s consi dered to be
reasonabl e. I n addi ti on, l arge rotors that run at sl ow speeds exhi bi t understand-
abl y l ower rates. Thi s tabl e i s onl y provi ded for reference purposes based upon a
ni te set of measurements. For speci c recommendati ons of accel erati on rates
for dened machi nery trai ns, the Ori gi nal Equi pment Manufacturer (OEM)
must al ways be contacted.
Throughout thi s secti on on dampi ng i nuence, the emphasi s has been
pl aced on the effects of dampi ng upon a mechani cal system. The changes of rotor
ampl i cati on factor, and the changes i n stabi l i ty due to vari ati ons i n dampi ng
are general l y understood. For si mpl e cases, i t has been demonstrated that the
theory and the actual machi nery behavi or are qui te compati bl e. I n al l cases, the
machi nery di agnosti ci an i s encouraged to si mpl i fy probl ems to the greatest
extent possi bl e, and to stri ve towards strai ghtforward expl anati ons and sol u-
ti ons. Unfortunatel y, some mechani cal systems may not cooperate wi th thi s
approach, and the compl exi ty of the probl em pl us the potenti al sol uti on may very
wel l exceed the expl anati ons offered by si mpl e systems.
When faced wi th di fcul t techni cal probl ems, the tradi ti onal engi neeri ng
mental i ty often attempts to ful l y quanti fy and dene al l vari abl es associ ated
wi th the probl em. Wi th respect to dampi ng characteri sti cs, thi s i s di fcul t to
accompl i sh. For exampl e, John Vance
11

states that Accurate values for the damp-
ing coefcients are usually difcult to obtain. The author has found from experi-
ments that an all-steel rotor/ shaft assembly with tight ts and no gear backlash
will have about 1.52.0 percent of critical damping, not including friction between
the rotor and ground. Thi s statement rei nforces the di fcul ty i n determi nati on
of actual dampi ng coefci ents. I n many i nstances, the dampi ng characteri sti cs
are ei ther i gnored or determi ned experi mental l y.
There i s experi mental i nformati on publ i shed on speci c dampi ng materi al s
used for vi brati on and shock i sol ati on. Techni cal references such as the Shock
and Vibration Handbook
12
devote several chapters to i sol ators, and the charac-
teri sti cs and uti l i zati on of dampi ng materi al s. Thi s i nformati on i s appl i cabl e to
machi nes such as ai r bl owers and refri gerati on compressors that are i nstal l ed on
l arge bui l di ngs. For the comfort of the i nhabi tants of these bui l di ngs, the
machi nes are normal l y i sol ated from thei r respecti ve structures. However, on
l arge i ndustri al machi nes, the foundati ons and supporti ng structures are nor-
mal l y ri gi d, and external dampi ng materi al s are not used. On l arge gas handl i ng
process machi nes, most of the dampi ng i s provi ded by the beari ng l ubri cati ng oi l ,
11
John M. Vance, Rotordynamics of Turbomachinery, (New York: Wi l ey-I ntersci ence Publ i ca-
ti on, 1988), p 61.
12
Cyri l M. Harri s, Shock and Vibration Handbook, Fourth edi ti on, (New York: McGraw-Hi l l ,
1996), chapters 32, 33, 36 and 37.
104 Chapter-3
wi th onl y a mi nor contri buti on from the shaft oi l seal s. The vi scous dampi ng
associ ated wi th the beari ngs normal l y domi nates any coul omb and structural
dampi ng i n the system. The engi neeri ng val ues for beari ng dampi ng are usual l y
cal cul ated wi th anal yti cal programs, as di scussed i n chapter 5. Structural or
support dampi ng i s often determi ned experi mental l y, and thi s topi c i s addressed
i n the fol l owi ng chapter 4.
Al though quanti cati on of di screte dampi ng val ues i s necessary from an
i ni ti al desi gn standpoi nt, i t may not be parti cul arl y useful to the di agnosti ci an
attempti ng to sol ve a el d probl em. I n most cases, a worki ng knowl edge of the
i nuence of dampi ng upon the machi nery, pl us an understandi ng of the contri b-
uti ng parameters that are i nvol ved i n determi nati on of the dampi ng are the
most i mportant i tems. To thi s extent, i t i s mandatory for the di agnosti ci an to
understand that beari ng dampi ng vari es wi th speed, appl i ed l oad, beari ng geom-
etry, pl us the l ubri cant characteri sti cs. Thi s knowl edge may al l ow the devel op-
ment of a l ogi cal thought process duri ng the anal ysi s of a machi nery probl em.
For i nstance, i f a compressor exhi bi ts symptoms of decreased dampi ng, such as
hi gher vi brati on ampl i tudes passi ng through a cri ti cal speed, one mechani sm
that mi ght consi dered woul d be the i nuence of expanded beari ng cl earances
upon the dampi ng. Of course, i ncreased beari ng cl earances mi ght al so produce
si gni cant changes i n the beari ng sti ffness, whi ch woul d probabl y change the
center frequency of the measured cri ti cal speed. Hence, the i nuence of sti ffness
upon the machi nery behavi or must al so be addressed, and the fol l owi ng secti on
addresses the fundamental s of the sti ffness aspects of machi nery supports.
Stiffness Influence 105
STIFFNESS INFLUENCE
The sti ffness of a mechani cal system i s the spri ng l i ke qual i ty of mechani -
cal el ements to el asti cal l y deform under l oad. I n essence, the appl i cati on of a
force measured i n pounds wi l l produce a deecti on measured i n i nches. Thus, the
sti ffness of a mechani cal el ement carri es the tradi ti onal engi neeri ng uni ts of
Pounds per I nch. Mechani cal properti es such as the di mensi ons and wei ght of an
object have a physi cal meani ng for most i ndi vi dual s. However, a characteri sti c
such as sti ffness i s commonl y used, but i t i s sel dom associ ated wi th anythi ng of
real physi cal si gni cance. To provi de a sense of magni tude to sti ffness val ues,
consi der the fol l owi ng deri vati on of axi al , radi al , and torsi onal sti ffness for a
seri es of i denti cal cyl i nders.
The di agram presented i n Fi g. 3-22 descri bes a sol i d metal cyl i nder wi th a
di ameter D, and a l ength L. Assume that the bottom of thi s cyl i nder i s ri gi dl y
supported, and consi der the appl i cati on of a pre axi al compressi ve force F
ax
. Fur-
ther assume that the appl i ed force i s sufci ent to cause an el asti c deformati on of
the cyl i nder, and that the amount of thi s deformati on i s descri bed by the deec-
ti on L. From fundamental strength of materi al s i t i s known that:
(3-47)
(3-48)
I t i s al so known that wi thi n the el asti c regi on for the materi al , stress and
strai n are rel ated by the modul us of el asti ci ty E as fol l ows:
Fig. 322 Solid Cylinder
Mounted On Innitely
Rigid Base And Subjected
To Axial Compression
Diameter (D)
D
e
f
l
e
c
t
i
o
n

(

L
)
Axial Force
(F )
ax

L
e
n
g
t
h

(
L
)
Strai n
Defl ecti on
L ength
------------------------------
L
L
-------- = =
Stress
sol i d
Force
Area
sol i d
--------------------------
F
ax
D
2
4
------------------------
4 F
ax
D
2

------------------- = = =
106 Chapter-3
(3-49)
Substi tuti ng equati ons (3-47) and (3-48) i nto (3-49) produces the fol l owi ng:
(3-50)
I n accordance wi th the i ni ti al deni ti on for sti ffness, the axi al sti ffness of
thi s metal cyl i nder may be determi ned by di vi di ng the appl i ed force F
ax
by the
deecti on L. Hence, i t may be properl y stated that:
(3-51)
By combi ni ng (3-50) and (3-51), the axi al sti ffness may be i ncl uded as:
By rearrangi ng terms, (3-52) for axi al sti ffness of a sol i d cyl i nder evol ves:
(3-52)
Equati on (3-52) makes sense from the standpoi nt of the contri buti on of the
terms. Speci cal l y, a l arge di ameter cyl i nder wi l l be sti ffer than a smal l er di ame-
ter cyl i nder, and a short cyl i nder shoul d be sti ffer than a tal l one. To provi de a
physi cal representati on of the meani ng of vari ous sti ffness val ues, consi der sol v-
i ng the l ast expressi on for the di ameter as fol l ows:
(3-53)
I f the cyl i nder materi al i s steel , the modul us of el asti ci ty E woul d be i n the
vi ci ni ty of 30,000,000 Pounds/I nch
2
. I f the total l ength of the cyl i nder L i s
assumed to be 30 i nches, then (3-53) may be si mpl i ed to the fol l owi ng format:
Wi th thi s expressi on, the equi val ent cyl i nder di ameter may be computed
for vari ous sti ffness val ues. For i nstance, i f the axi al sti ffness of a sol i d cyl i nder
Modul us of El asti ci ty E
Stress
sol i d
Strai n
------------------------------ = =
E
4 F
ax
D
2

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

' ;

L
L
--------

' ;

--------------------------
4 F
ax
L
D
2
L
------------------------------- = =
K
ax
sol i d
F
ax
L
---------- =
E
4 L
D
2

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

' ;

F
ax
L
----------

' ;

4 L
D
2

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

' ;

K
ax
sol i d
= =
K
ax
sol i d
E D
2

4 L
--------------------------- =
D 2
L K
ax
sol i d

E
------------------------------ =
D 2
30 Inches K
ax
sol i d

30
6
10 Pounds/Inches
2

---------------------------------------------------------------
K
ax
sol i d
886.2
------------------------ = =
Stiffness Influence 107
i s 1,000,000 Pounds/I nch, the requi red di ameter i s cal cul ated as fol l ows:
I f thi s cal cul ati on i s repeated for a seri es of sti ffness val ues, the resul tant
di ameters may be pl otted agai nst sti ffness. Another way to vi ew thi s i nformati on
i s shown i n Fi g. 3-23. Wi thi n thi s di agram, a seri es of consecuti ve cyl i nders are
drawn that represent sti ffness val ues rangi ng from 1,000,000,000 Pounds/I nch
(i .e., 10
9
) wi th an associ ated di ameter of 35.682 i nches to a mi ni mum sti ffness
of 100,000 Pounds/I nch (i .e., 10
5
) wi th a computed di ameter of 0.3568 i nches.
Fi g. 3-23 i s a rendi ti on of a si mi l ar expl anati on of sti ffness presented by
F.L. Weaver
13
, i n hi s paper enti tl ed Rotor Design and Vibration Response.
Al though Weaver di d not present the method for computi ng the sti ffness of the
cyl i nders hi s di agram of spri ng gradi ents of 30 i nch tal l sol i d steel cyl i nders
on an i nni tel y sti ff support are equi val ent to the drawi ng i n Fi g. 3-23.
I f the cyl i nders subjected to axi al compressi on are hol l ow i nstead of sol i d,
the previ ousl y devel oped equati on structure may be easi l y modi ed by i ncorpo-
rati ng an i nner di ameter i nto equati on (3-48) as fol l ows:
(3-54)
Fig. 323 Axial Stiffness Of 30 Inch Tall Steel Cylinders Of Various Diameters
13
F.L. Weaver, Rotor Desi gn and Vi brati on Response, Proceedings of the First Turbomachin-
ery Symposium, Gas Turbi ne Laboratori es, Texas A&M Uni versi ty, Col l ege Stati on, Texas, (1972),
pp. 142-147.
D
K
ax
sol i d
886.2
------------------------
1
6
10
886.2
------------------- 1.1284 Inches = = =
11.284"
3.568"
1.128"
0.357"
C
y
l
i
n
d
e
r

W
e
i
g
h
t

=
8
,
4
8
9

P
o
u
n
d
s
8
4
9

P
o
u
n
d
s
8
5

P
o
u
n
d
s
8
.
5

P
o
u
n
d
s
0
.
8
5

P
o
u
n
d
s
Diameter = 35.682"

H
e
i
g
h
t

=

3
0
.
0
0
"
10
8 Lbs.
In.
10
7 Lbs.
In.
10
6 Lbs.
In.
10
5 Lbs.
In.
Axial Stiffness = 10
9 Pounds
Inch
Stress
annul us
Force
Area
annul us
-----------------------------------
F
ax

D
o
2
4
-------
D
i
2
4
-------
,

_

------------------------------------
4 F
ax
D
o
2
D
i
2
( )
----------------------------------- = = =
108 Chapter-3
I n equati on (3-54), the cyl i nder outer di ameter i s desi gnated by D
o
, and the
hol l ow i nner di ameter i s once more i denti ed as D
i
. The strai n equati on (3-47)
may now be combi ned wi th equati on (3-54) i n the tradi ti onal stressstrai n rel a-
ti onshi p descri bed by equati on (3-49) to produce the fol l owi ng resul t:
(3-55)
As before, the general expressi on for the spri ng constant (F
ax
/L) may be
i nserted i nto (3-55) to establ i sh an expressi on that i ncl udes the physi cal di men-
si ons, the modul us of el asti ci ty, pl us the axi al sti ffness of a hol l ow cyl i nder:
Thi s equati on may now be sol ved for the axi al sti ffness of a hol l ow cyl i nder:
(3-56)
Equati on (3-56) may be used to compute the axi al sti ffness of a hol l ow cyl i n-
der. I t i s cl ear that i f the i nternal di ameter i s set equal to zero, then (3-56)
reverts back to (3-52) for a sol i d cyl i nder. From thi s pai r of sti ffness equati ons, i t
i s noted that a hol l ow cyl i nder wi l l al ways have an axi al sti ffness that i s l ess
than an equi val ent sol i d cyl i nder of equal outer di ameter.
Thi s array of cyl i nders shown i n Fi g. 3-23 hel ps to vi sual i ze sti ffness val ues
for vari ous physi cal di mensi ons. However, i t shoul d be recal l ed that these val ues
are predi cated upon a cyl i nder pl aced i n axi al compressi on. I t i s al so meani ngful
to exami ne the radi al or l ateral sti ffness for thi s same group of steel cyl i nders.
For i nstance, assume that a cyl i nder i s posi ti oned hori zontal l y as shown i n Fi g.
3-24. Further assume that the cyl i nder i s si mpl y supported at each end, and that
a radi al mi dspan force F
rad
i s appl i ed. As before, the di mensi ons of the cyl i nder
are the di ameter D, and the l ength L. Furthermore, the maxi mum mi dspan
radi al deecti on R of the cyl i nder i s i denti ed on Fi g. 3-24.
The radi al cyl i nder sti ffness i s the appl i ed force F
rad
di vi ded by the deec-
ti on R. Even though Fi g. 3-24 depi cts a radi al i nstead of an axi al deecti on, the
same physi cal l aws appl y. Hence, i t may be properl y restated that the radi al
sti ffness i s equal to the appl i ed force F
rad
di vi ded by the deecti on R as shown:
(3-57)
E
Sress
annul us
Strai n
------------------------------------
4 F
ax
D
o
2
D
i
2
( )
-----------------------------------

' ;

L
L
--------

' ;

------------------------------------------
4 F
ax
L
D
o
2
D
i
2
( ) L
------------------------------------------------- = = =
E
4 L
D
o
2
D
i
2
( )
-----------------------------------

' ;

F
ax
L
----------

' ;

4 L
D
o
2
D
i
2
( )
-----------------------------------

' ;

K
ax
annul us
= =
K
ax
annul us
E D
o
2
D
i
2
( )
4 L
---------------------------------------------- =
K
rad
F
rad
R
------------- =
Stiffness Influence 109
The l ateral deecti on of thi s cyl i nder may be determi ned from si mpl y sup-
ported beam formul as. From Roark
14
, or Spotts
15
, the maxi mum mi dspan deec-
ti on may be computed from the fol l owi ng common expressi on:
(3-58)
Combi ni ng the previ ous equati ons (3-57) and (3-58), and sol vi ng for the l at-
eral (radi al ) cyl i nder sti ffness K
rad
, the next expressi on i s formed:
Thi s i s si mpl i ed to the fol l owi ng common format:
(3-59)
Equati on (3-59) i s often used for the cal cul ati on of the shaft sti ffness. Si nce
the el ement under l oad has a ci rcul ar cross secti on, the area moment of i nerti a
I
area
for a sol i d may be determi ned from (3-30), and i ncorporated i nto the sti ff-
ness equati on (3-59) to obtai n the fol l owi ng resul t:
Fig. 324 Simply Supported
Solid Cylinder Subjected To
A Midspan Radial Force
14
Warren C. Young, Roarks Formulas for Stress & Strain, Si xth Edi ti on, (New York: McGraw-
Hi l l Book Company, 1989), p. 101.
15
M.F. Spotts, Design of Machine Elements, 6th Edi ti on, (Engl ewood Cl i ffs, New Jersey: Pren-
ti ce-Hal l , I nc., 1985), p. 27.
D
i
a
m
e
t
e
r

(
D
)
Radial Deflection (R)
Radial Force
(F )
rad
Length (L)
R
F
rad
L
3

48 E I
area

------------------------------------- =
K
rad
F
rad
R
-------------
F
rad
F
rad
L
3

48 E I
area

-------------------------------------
,

_
------------------------------------------ = =
K
rad
48 E I
area

L
3
------------------------------------- =
K
rad
sol i d
48 E I
area
sol i d

L
3
----------------------------------------------
48 E
L
3
----------------
,
_
D
4

64
-----------------
,
_
= =
110 Chapter-3
Thi s si mpl i es to the next common equati on for l ateral shaft sti ffness:
(3-60)
I f the l ength L remai ns at 30 i nches, and the modul us of el asti ci ty for steel
i s 30,000,000 Pounds/I nch
2
, equati on (3-60) may be sol ved for the sti ffness of
vari ous cyl i nder di ameters. For consi stency, the previ ousl y cal cul ated cyl i nder
that had an axi al sti ffness of 10
6
, was computed to have a di ameter of 1.1284
i nches. The l ateral sti ffness for thi s same cyl i ndri cal el ement i s as fol l ows:
Repeati ng thi s cal cul ati on for cyl i nder di ameters of 35.682, 11.284, 3.5682,
and 0.3568 i nches respecti vel y, the summary di agram of Fi g. 3-25 may be con-
structed. I n thi s drawi ng, the cyl i nder di mensi ons are i denti cal to the val ues pre-
vi ousl y determi ned from the axi al l oad case. However, i t i s apparent that the
radi al sti ffness magni tudes have changed si gni cantl y from the axi al sti ffness
val ues shown i n Fi g. 3-23.
For the case of a hol l ow cyl i nder, the radi al sti ffness i s determi ned by
i nstal l i ng the area moment of i nerti a for an annul us, equati on (3-23), i nto the
general radi al sti ffness equati on (3-59) i n the fol l owi ng manner:
Fig. 325 Radial Stiffness Of 30 Long Steel Cylinders Of Various Diameters
K
rad
sol i d
3 E D
4

4 L
3

------------------------------------- =
K
rad
sol i d
3 30
6
10 Pounds/Inch
2
1.1284 Inches ( )
4

4 30 Inches ( )
3

------------------------------------------------------------------------------------------------------------------ 4 240 Pounds/Inch , = =


1
1
.
2
8
4
"
Cylinder Weight=
8,489 Pounds
849 Pounds

D
i
a
m
e
t
e
r

=

3
5
.
6
8
2
"
Length = 30.00"
4.24 x 10
7 Pounds
Inch
Radial Stiffness =
4.24 x 10
9 Pounds
Inch
3.568"
1.128"
0.357"
85 Pounds
8.5 Pounds
0.85 Pounds
4.24 x 10
5 Pounds
Inch
4,240
Pounds
Inch
42.4
Pounds
Inch
Stiffness Influence 111
Si mpl i fyi ng and combi ni ng terms, the expressi on (3-61) for radi al sti ffness
of a hol l ow cyl i nder may be easi l y produced:
(3-61)
Thi s i s si mi l ar to equati on (3-60) for a sol i d shaft. I f the i nner di ameter of
equati on (3-61) i s set equal to zero, the above expressi on wi l l revert back to equa-
ti on (3-60). Agai n, i t i s apparent that a sol i d shaft i s al ways l ateral l y sti ffer than
a comparabl e hol l ow shaft of equal outer di ameter.
I n addi ti on to the radi al and axi al di recti ons, machi nery shafts are al so
subjected to twi sti ng forces. For exampl e, the normal movement of reci procati ng
machi nes produces synchronous torsi onal exci tati ons i n rotati ng el ements such
as crankshafts. Al though centri fugal machi nes are l ess suscepti bl e to torsi onal
moti on fai l ures do occur, and the torsi onal aspects of centri fugal machi nery
must be consi dered. For ei ther cl ass of machi ne, the torque on one end of a shaft
may resul t i n an angul ar deecti on al ong the shaft l ength. The amount of twi st
i s di rectl y rel ated to the appl i ed torque, and the torsi onal sti ffness.
I n general , torsi onal sti ffness K
tor
i s anal ogous to the axi al or l ateral sti ff-
ness. However, the force i s repl aced by torque, and the l i near di spl acement i s
repl aced by an angul ar deecti on . Performi ng thi s parameter substi tuti on i n
(3-51), the fol l owi ng general deni ti on for torsi onal sti ffness i s obtai ned:
(3-62)
A typi cal condi ti on for a cyl i nder subjected to twi sti ng deformati on i s
depi cted i n Fi g. 3-26. From strength of materi al s (Roark
16
, or Spotts
17
), i t i s
Fig. 326 Solid Cylinder,
Fixed At One End, And
Subjected To Torque
16
Warren C. Young, Roarks Formulas for Stress & Strain, Si xth Edi ti on, (New York: McGraw-
Hi l l Book Company, 1989), p. 346.
K
rad
annul us
48 E I
area
annul us

L
3
----------------------------------------------------
48 E
L
3
----------------
,
_
D
o
4
D
i
4
( )
64
-----------------------------------
,

_
= =
K
rad
annul us
3 E D
o
4
D
i
4
( )
4 L
3

------------------------------------------------------- =
D
i
a
m
e
t
e
r

(
D
)
Torque
Length (L)

Angular Deflection
K
tor
Torque

--------------------- =
112 Chapter-3
known that the angl e of twi st i s rel ated to the appl i ed Torque, the member
l ength L, the area pol ar moment of i nerti a J
area
, and the materi al shear modul us
G
shear
i n the fol l owi ng manner:
(3-63)
Note that the shear modul us and the accel erati on of gravi ty are both tradi -
ti onal l y descri bed by the l etter G. To di sti ngui sh between these two common
terms, the shear modul us wi l l be i denti ed by G
shear
throughout thi s text. I t
shoul d al so be menti oned that the shear modul us i s al so known as the modul us
of ri gi di ty, the modul us of el asti ci ty i n shear, and the torsi on modul us. I n any
case, combi ni ng equati ons (3-62) and(3-64), the fol l owi ng general expressi on for
torsi onal sti ffness i s obtai ned:
(3-64)
The resul tant J G/ L expressi on for torsi onal sti ffness i s common wi thi n the
l i terature. For i nstance, thi s same equati on structure i s used i n the ASM Hand-
book of Engineering Mathematics
18
. Wi thi n (3-64), the area pol ar moment of
i nerti a J
area
i s the val ue cal cul ated through the center of the member (i .e., al ong
the rotati onal axi s). I f the cyl i nder cross secti on i s an annul us (hol l ow shaft) wi th
an outer di ameter of D
o
, and an i nner di ameter of D
i
, the area pol ar moment of
i nerti a J
area
was previ ousl y dened by equati on (3-24). Substi tuti ng (3-24) i nto
equati on (3-64), the fol l owi ng torsi onal sti ffness expressi on i s formed:
Thi s i s combi ned i nto the fol l owi ng common equati on for the torsi onal sti ff-
ness of a hol l ow shaft (cyl i nder):
(3-65)
The val i di ty of equati on (3-65) i s agai n supported by i denti cal expressi ons
for torsi onal sti ffness of hol l ow ci rcul ar cyl i nders as shown i n the Shock and
Vibration Handbook
19
. I f the shaft cross secti on i s sol i d, then the i nner di ameter
17
M.F. Spotts, Design of Machine Elements, 6th Edi ti on, (Engl ewood Cl i ffs, New Jersey: Pren-
ti ce-Hal l , I nc., 1985), p. 150.
18
Wi l l i am G. Bel di ng and others, ASM Handbook of Engineering Mathematics, 4th pri nti ng,
(Metal s Park, Ohi o: Ameri can Soci ety for Metal s, 1989), p. 319.
19
Cyri l M. Harri s, Shock and Vibration Handbook, Fourth edi ti on, (New York: McGraw-Hi l l ,
1996), p. 1.11.

Torque L
J
area
G
shear

--------------------------------------- =
K
tor
Torque

---------------------
Torque
Torque L
J
area
G
shear

---------------------------------------
----------------------------------------
J
area
G
shear

L
--------------------------------------- = = =
K
tor
annul us
G
shear
L
------------------

' ;

D
o
4
D
i
4
( )
32
-----------------------------------

' ;

=
K
tor
annul us
G
shear
D
o
4
D
i
4
( )
32 L
------------------------------------------------------------ =
Stiffness Influence 113
D
i
i s equal to zero, and equati on (3-65) may be si mpl i ed to the fol l owi ng expres-
si on descri bi ng the torsi onal sti ffness of a sol i d ci rcul ar shaft:
(3-66)
Once more, thi s expressi on for torsi onal sti ffness i s common wi thi n the
techni cal l i terature previ ousl y ci ted. From (3-66), i t i s evi dent that the torsi onal
sti ffness i s a functi on of the cyl i nder di ameter D, the cyl i nder l ength L, and the
shear modul us G
shear
. I f the physi cal di mensi ons, and G
shear
are known, then the
torsi onal sti ffness may be computed. However, i n cases when the shear modul us
i s unknown, i t may be computed from the modul us of el asti ci ty E, and Poi ssons
rati o . Thi s rel ati onshi p i s dened by Shi gl ey
20
as:
(3-67)
I f the steel cyl i nder di mensi ons previ ousl y used for computati on of axi al
and radi al sti ffness are used for the cal cul ati on of torsi onal sti ffness the rst
step i s to determi ne the shear modul us. As previ ousl y noted, the modul us of el as-
ti ci ty E for these steel cyl i nders was assumed to be 30,000,000 Pounds/I nch
2
. I f
Poi ssons rati o for thi s materi al i s equal to 0.3, then G
shear
may be determi ned
from equati on (3-67):
For a sol i d steel cyl i nder l ength L of 30 i nches, and a di ameter D equal to
1.1284 i nches, the computed shear modul us may now be used to cal cul ate the
torsi onal sti ffness from equati on (3-66) i n the fol l owi ng manner:
Note that the di rect engi neeri ng uni ts for thi s cal cul ati on are I nch-Pounds
but the non-di mensi onal angul ar uni ts of Radi ans have been i ncl uded. Thi s i s
i nferred by the previ ous equati on (3-62) of Torque/Angl e and i t i s customary
nomencl ature for torsi onal sti ffness. For numeri cal compari son purposes, the tor-
si onal sti ffness cal cul ati on i s repeated for cyl i nder di ameters of 35.682, 11.284,
3.5682, and 0.3568 i nches respecti vel y. The sel ected cyl i nder di ameters are i den-
ti cal to the val ues previ ousl y determi ned from the axi al l oad case. The axi al ,
radi al , and torsi onal sti ffness val ues for each of the ve cyl i nder di ameters are
now summari zed i n Tabl e 3-3.
20
Joseph E. Shi gl ey and Charl es R. Mi schke, Standard Handbook of Machine Design, (New
York: McGraw-Hi l l Book Company, 1986), pp. 10.6.
K
tor
sol i d
G
shear
D
4

32 L
----------------------------------------- =
E 2 G
shear
1 + ( ) =
G
shear
E
2 1 + ( )
--------------------------
30
6
10 Pounds/Inch
2
2 1 0.3 + ( )
-------------------------------------------------- 11.5
6
10 Pounds/Inch
2
= = =
K
tor
sol i d
11.5
6
10 Pounds/Inch
2
1.1284 Inches ( )
4

32 30 Inches
-------------------------------------------------------------------------------------------------------------- =
K
tor
sol i d
58 573 344 , ,
960
------------------------------ Inch-Pounds 61 010
Inch-Pound
Radian
---------------------------- , = =
114 Chapter-3
Cl earl y, the steel cyl i nder sti ffness vari es consi derabl y from the axi al to the
radi al di recti on. These val ues are i ndependent of l oad, and they are a functi on of
cyl i nder di mensi ons, and the modul us of el asti ci ty. The si gni cant poi nt of thi s
exerci se i s that i ndi vi dual machi ne el ements sel dom have si ngul ar el ements that
exceed the range of sti ffness val ues presented i n these physi cal exampl es. That
i s, machi ne parts wi l l probabl y not have a sti ffness much greater than 10
8
or 10
9
Pounds per I nch. These are enormousl y ri gi d el ements, and sti ffness val ues of
10
8
are sel dom encountered i n rotati ng machi nery. At the other end of the scal e,
the ski nny cyl i nder wi th a di ameter of 0.36 I nches, and a hei ght of 30 i nches, di s-
pl ays an axi al sti ffness of onl y 100,000 Pounds per I nch. Agai n, i t i s hard to
i magi ne many l oad carryi ng members wi thi n a rotati ng machi ne wi th thi s type
of aspect rati o. Thus, i t i s l ogi cal to concl ude that thi s sti ffness of 10
5
Pounds per
I nch i s a real i sti c mi ni mum val ue. For reference purposes, Tabl e 3-4 of typi cal
sti ffness ranges (i n Pounds/I nch) for vari ous machi ne el ements i s presented.
The concept of sti ffness i s appl i cabl e to shafts, beari ngs, cases, support
structures, and foundati ons. I n a real machi ne al l of these sti ffness parameters
pl ay a rol e i n determi nati on of the nal vi brati on response characteri sti cs. The
rel ati onshi p between effecti ve system sti ffness, and each of the i ndi vi dual el e-
Table 33 Comparison of Calculated Axial, Radial, and Torsional Stiffness of 30 Inch Long
Solid Steel Cylinders of Various Diameters
Cylinder Diameter
(Inches)
Axial Stiffness
(Pounds/Inch)
Radial Stiffness
(Pounds/Inch)
Torsional Stiffness
(Inch-Pounds/Radian)
35.682 1.00 x 10
9
4.24 x 10
9
6.10 x 10
10
11.284 1.00 x 10
8
4.24 x 10
7
6.10 x 10
8
3.5682 1.00 x 10
7
4.24 x 10
5
6.10 x 10
6
1.1284 1.00 x 10
6
4.24 x 10
3
6.10 x 10
4
0.3568 1.00 x 10
5
4.24 x 10
1
6.10 x 10
2
Table 34 Summary of Typical Machinery Element Stiffness Values
Mechanical Element
Typical Stiffness Range
(Pounds/Inch)
Oi l Fi l m Beari ngs 300,000 to 2,000,000
Rol l i ng El ement Beari ngs 1,000,000 to 4,000,000
Beari ng Housi ng Support - Hori zontal 300,000 to 4,000,000
Beari ng Housi ng Support - Verti cal 400,000 to 6,000,000
Shaft 1 to 4 Di ameter 100,000 to 4,000,000
Shaft 6 to 15 Di ameter 400,000 to 20,000,000
Stiffness Influence 115
ments i s that the reci procal of the overal l system sti ffness i s equal to the sum of
the reci procal s of each i ndi vi dual el ement. Thus, the change i n any el ement (e.g.,
soft beari ng support) can i nuence the enti re system. I t must al so be recogni zed
that the weakest (softest) member i n the rotor support system wi l l be the domi -
nant el ement i n establ i shi ng overal l or effecti ve system sti ffness.
The previ ous di scussi on has centered around the vari ati ons i n sti ffness due
to changes i n geometry and materi al s of constructi on. Al though thi s i s an aca-
demi cal l y i nteresti ng topi c, i t i s not parti cul arl y useful unti l i t i s appl i ed wi thi n
the real m of process machi nery. One of the obvi ous i nuences of sti ffness i s the
effect upon the natural frequency of machi ne el ements. For di scussi on purposes,
consi der the axi al , torsi onal , and l ateral natural frequency of an undamped hol -
l ow cyl i nder based upon the previ ousl y devel oped equati ons. Speci cal l y, the
axi al natural frequency may be determi ned from equati on (2-45). The requi red
mass of the cyl i nder may be speci ed by equati on (3-4), and the axi al sti ffness of
thi s annul us may be determi ned from equati on (3-56). Combi ni ng these three
expressi ons, the fol l owi ng resul t i s obtai ned:
Thi s equati on for the axi al natural frequency may be further si mpl i ed to:
(3-68)
Thi s i s an i nteresti ng resul t. The di ameters have cancel ed out, and the onl y
remai ni ng physi cal di mensi on i s the l ength L of the cyl i nder. Hence, the hol l ow
cyl i nder di ameters do not i nuence the axi al natural frequency. Based on the
materi al properti es used for the steel cyl i nder exampl es, the axi al natural fre-
quency for a 30.0 I nch l ong member may be computed from equati on (3-68) as:
The computed val ue of 64,400 CPM i s a rel ati vel y hi gh frequency, and
woul d probabl y fal l outsi de of the operati ng speed range for most machi nes. Si m-
i l arl y, i f the shaft l ength was extended to 60 I nches, the axi al resonant frequency
woul d be 32,200 CPM, whi ch i s sti l l qui te a hi gh frequency for a 5 foot l ong shaft.
Due to thi s general rel ati onshi p, the axi al natural frequency of a shaft i s sel dom
wi thi n the operati ng speed range. Hence, there i s mi ni mal opportuni ty to exci te
F
c
axi al
1
2
------
K
ax
annul us
M
annul us
--------------------------
1
2
------
E
4 L
-------------- D
o
2
D
i
2
( )

' ;

L
4 G
----------------------- D
o
2
D
i
2
( )

' ;

--------------------------------------------------------------
1
2
------
E
4 L
--------------
4 G
L
----------------------- = = =
F
c
axi al
1
2
------
E G
L
2

---------------- =
F
c
axi al
1
2
------
30
6
10 Pounds/Inch
2
386.1 Inches/Second
2

0.283 Pounds/Inches
3
30 Inches ( )
2

--------------------------------------------------------------------------------------------------------- 1 073
Cycles
Second
---------------- , = =
F
c
axi al
1 073
Cycles
Second
----------------- 60
Seconds
Minute
------------------- , 64 400
Cycles
Minute
----------------- , = =
116 Chapter-3
thi s resonance. Al though axi al vi brati on does occur i n most machi nes, there i s
l i ttl e evi dence to suggest that the shaft axi al resonant frequenci es are commonl y
exci ted.
As a cauti onary note, the di agnosti ci an shoul d not i gnore other l ongi tudi nal
or axi al resonances wi thi n the rotor system. For i nstance, exi bl e di sc or metal l i c
membrane coupl i ngs wi l l often have a l ower frequency axi al resonance that may
be i n the vi ci ni ty of the operati ng speed range. Thi s i s general l y referred to as
the Natural Axi al Resonant Frequency (NARF) of the coupl i ng assembl y. I t i s
common to have NARF val ues between 5,000 and 15,000 CPM for many congu-
rati ons of axi al l y compl i ant (soft) coupl i ngs. The computati on of these frequen-
ci es becomes qui te di fcul t due to the compl ex di aphragm sti ffness, and the
determi nati on of the effecti ve mass. Hence, the si mpl i ed format represented by
equati on (3-70) wi l l probabl y not be adequate for cal cul ati ng a coupl i ng NARF.
Furthermore, other machi ne el ements such as wheel s or thrust col l ars al so have
axi al resonant frequenci es that may have to be consi dered duri ng a machi nery
anal ysi s.
Movi ng on to the torsi onal behavi or, the fundamental undamped torsi onal
resonance of a hol l ow cyl i nder may be determi ned based upon equati on (2-103).
The mass pol ar moment of i nerti a may be speci ed by equati on (3-26), and the
torsi onal sti ffness of thi s annul us may be determi ned from equati on (3-65). Com-
bi ni ng these three expressi ons, the fol l owi ng resul t i s obtai ned:
The general expressi on for the cyl i nder torsi onal natural resonant fre-
quency may be further si mpl i ed as fol l ows:
(3-69)
I nteresti ngl y enough, the torsi onal resonance equati on (3-69) has the same
format as the axi al resonance equati on (3-68). The onl y di fference i s that the tor-
si onal equati on uses the shear modul us G
shear
, and the axi al equati on uses the
modul us of el asti ci ty E. Once more the cyl i nder di ameter i s cancel ed out, and
equati on (3-69) reveal s that the torsi onal natural resonant frequency of the hol -
l ow cyl i nder i s a functi on of the l ength L and the materi al properti es. For com-
parati ve purposes, the torsi onal resonant frequency for the 30.0 I nch l ong steel
cyl i nders may be computed as fol l ows:
F
c
tor
1
2
------
K
tor
annul us
J
mass
annul us
--------------------------------
1
2
------
G
shear

32 L
---------------------------- D
o
4
D
i
4
( )

' ;

L
32 G
----------------------- D
o
4
D
i
4
( )

' ;

-------------------------------------------------------------
1
2
------
G
shear

32 L
----------------------------
32 G
L
----------------------- = = =
F
c
tor
1
2
------
G
shear
G
L
2

----------------------------- =
Stiffness Influence 117
Si nce the shear modul us i s smal l er than the el asti c modul us, the torsi onal
resonant frequency i s l ower than the previ ousl y computed axi al resonant fre-
quency for the 30.0 i nch l ong cyl i nder. Thi s torsi onal frequency i s probabl y out-
si de of the normal operati ng speed range, and the torsi onal natural resonance of
the shaft by i tsel f i s not the pri mary area of concern. For most machi nery, the
torsi onal resonance(s) of the system are governed by the torsi onal sti ffness of the
coupl i ngs, and the overal l mass pol ar moment of i nerti a of the enti re rotor
assembl y (i .e., not just the shaft). When these i tems are combi ned, the actual tor-
si onal resonance(s) may fal l wi thi n the operati ng speed range. An exampl e of
thi s type of torsi onal resonance si tuati on i s presented i n case hi story 11 of the
torsi onal anal ysi s of a power turbi ne and pump.
The thi rd type of undamped radi al or l ateral resonance of a hol l ow cyl i nder
may be determi ned wi th equati on (2-45). The mass may be speci ed by equati on
(3-4), and the sti ffness of thi s annul us may be determi ned from equati on (3-61).
Combi ni ng these three expressi ons, the fol l owi ng resul t i s obtai ned:
The expressi on for the cyl i nder radi al natural resonant frequency may be
further si mpl i ed as fol l ows:
(3-70)
The l ateral natural resonance equati on (3-70) i ncl udes the compl ete array
of cyl i nder di mensi ons, pl us the materi al constants. I t i s cl ear that the l ateral
resonant frequency wi l l change i n accordance wi th di ameter changes. Obvi ousl y,
thi s evol ves as a probl em of greater mechani cal compl exi ty than ei ther the axi al
or the torsi onal natural resonance of the hol l ow cyl i nder. For exampl e, assume
F
c
tor
1
2
------
11.5
6
10 Pounds/Inch
2
386.1 Inches/Second
2

0.283 Pounds/Inches
3
30 Inches ( )
2

-------------------------------------------------------------------------------------------------------------- 664.5
Cycles
Second
---------------- = =
F
c
tor
664.5
Cycles
Second
----------------- 60
Seconds
Minute
------------------- 39 870
Cycles
Minute
----------------- , = =
F
c
rad
1
2
------
K
rad
annul us
M
annul us
-----------------------------
1
2
------
3 E
4 L
3

----------------------- D
o
4
D
i
4
( )

' ;

L
4 G
----------------------- D
o
2
D
i
2
( )

' ;

-------------------------------------------------------------- = =
F
c
rad
1
2
------
3 E
4 L
3

----------------------- D
o
2
D
i
2
+ ( )

' ;

L
4 G
-----------------------

' ;

--------------------------------------------------------------
1
2
------
3 E
4 L
3

-----------------------
4 G
L
----------------------- D
o
2
D
i
2
+ ( ) = =
F
c
rad
1
2
------
3 E G D
o
2
D
i
2
+ ( )
L
4
--------------------------------------------------------- =
118 Chapter-3
that the cyl i nder i nner di ameter D
i
i s zero, and the outer di ameter i s D
o
i s
3.5682 i nches (compati bl e wi th previ ous cal cul ati ons). The resul tant radi al reso-
nant frequency for the 30.0 i nch l ong steel cyl i nders may be computed as fol l ows:
Thi s frequency of 13,270 CPM i s l ow enough to be a potenti al threat to the
operati ng speed range of the machi ne. I n actual i ty, the l ateral cri ti cal speed of an
enti re rotor wi th thi s type of shaft geometry (i .e., 30.0 i nches l ong, and 3.5682
i nches i n di ameter), woul d be even l ower due to the i nuence of the attached
wheel s, thrust col l ars, and coupl i ngs. Hence, based upon thi s si mpl e exampl e, i t
i s cl ear that a machi ne general l y has a greater propensi ty towards l ateral cri ti -
cal speed probl ems, rather than axi al or torsi onal resonant probl ems. Thi s con-
cl usi on i s strongl y supported by actual i ndustri al mal functi ons where resonant
probl ems are usual l y observed i n the l ateral di recti ons, and i nfrequentl y encoun-
tered i n the axi al or torsi onal di recti ons.
From the previ ousl y di scussed stati c deecti on curves i n Fi gs. 3-1 through
3-6, the effect of mass di stri buti on and support l ocati on shoul d be appreci ated.
Now when the magni tude of the restrai nt or sti ffness i s i ncl uded i nto the di scus-
si on, the dynami c response of the machi ne can be subjected to further eval ua-
ti on. One way to envi si on the i nuence of sti ffness upon vi brati on ampl i tudes i s
to consi der the rel ati onshi p of parameters wi th respect to an appl i ed force. Spe-
ci cal l y, the appl i ed Force i s equal to the Response (vi brati on) ti mes the Restraint
(sti ffness). Thi s concept i s often stated i n the fol l owi ng manner:
(3-71)
The l arger the restrai nt, the smal l er the response for a gi ven uni t force
i nput. Cl earl y, vi bratory moti on can be suppressed by i ncreasi ng the sti ffness,
but there are penal ti es to be pai d. For exampl e, a rol l i ng el ement beari ng i s
sti ffer than a ui d l m beari ng. However, the rol l i ng el ement beari ng may
adversel y al ter the shaft mode shape. I n addi ti on, hi gh sti ffness di ctates l ow di s-
pl acement, whi ch general l y means smal l vi scous dampi ng. Under thi s case, the
rotor may not make i t through the cri ti cal speed range, and the stabi l i ty wi l l cer-
tai nl y suffer. Therefore, the machi ne that performed wel l for many years wi th
sl eeve beari ngs mi ght sel f destruct wi th the addi ti on of sti ff bal l beari ngs. The
same argument often appl i es to ti l ti ng pad beari ngs. Over the years, many End
Users have suffered the consequences of an i nappropri ate change from sl eeve
beari ngs to a ti l t pad assembl y wi th hi gher sti ffness, and l ower dampi ng.
Another vi ew of the sti ffness i nuence i s presented i n Fi g. 3-27 i l l ustrati ng
rotor mode shapes at three di fferent cri ti cal speeds. The mode shapes on the l eft
descri be the rotor moti on wi th compl i ant (soft) beari ngs. For compari son, the
F
c
rad
1
2
------
3 30
6
10 Pounds/Inch
2
386.1 Inches/Second
2
3.5682 Inches ( )
2

0.283 Pounds/Inches
3
30 Inches ( )
4

------------------------------------------------------------------------------------------------------------------------------------------------------------------ =
F
c
rad
221.1
Cycles
Second
----------------- 60
Seconds
Minute
------------------- 13 270
Cycles
Minute
----------------- , = =
Response
Force
Restrai nt
---------------------------- =
Stiffness Influence 119
mode shapes on the ri ght si de of Fi g. 3-27 depi ct the anti ci pated mode shapes
wi th ri gi d rotor supports. Thi s ri gi d support condi ti on i mpl i es that the beari ng
cl amps down the shaft moti on, and produces a nodal poi nt at each beari ng. I t i s
i mportant to recogni ze that a l ateral nodal poi nt i s a l ocati on of negl i gi bl e moti on
al ong the shaft axi s. Thi s condi ti on i s not total l y deni ti ve of a node, and i t nec-
essary for the shaft moti on on each si de of a node to be nomi nal l y 180 out of
phase. Thi s means that the shaft on one si de of the node i s movi ng i n the oppo-
si te di recti on of the shaft on the other si de of the node. I n other words, a true
zero axi s crossi ng nodal poi nt i s dened by a rocki ng moti on.
Note that the exi stence of a nodal poi nt wi thi n an oi l l m beari ng may be
i ndi cati ve of a seri ous si tuati on. Speci cal l y, a nodal poi nt i s a l ocati on of mi ni -
mal moti on. From previ ous di scussi ons, l ow shaft vi brati on ampl i tudes wi thi n a
beari ng may mani fest as l ow vi scous dampi ng, and thi s may resul t i n l ow stabi l -
i ty margi ns. Hence, a good rotor and beari ng desi gn may be rendered i noperati ve
due to the coi nci dence of shaft nodal poi nts and journal beari ngs.
Referri ng back to the rotor mode shapes shown i n Fi g. 3-27, i t i s noted that
the machi ne descri bes a cyl i ndri cal mode at the rst cri ti cal speed wi th soft sup-
ports. Rel ati ve moti on at the beari ngs are si gni cant, and phase measurements
woul d i ndi cate si mi l ar val ues between beari ngs. Wi th a hard support, a nodal
poi nt occurs at each beari ng, and shaft vi brati on measurements at the beari ngs
woul d be ni l . Furthermore, one can be fool ed by taki ng si mul taneous phase mea-
surements i nboard of one beari ng, and outboard of the other beari ng. Thi s data
woul d i ndi cate a nomi nal 180 phase di fference, and the di agnosti ci an mi ght
bel i eve that the uni t was goi ng through a second mode, rather than a rst mode
wi th hard supports.
The second cri ti cal descri bes the shaft mode shapes at the pi votal bal ance
Fig. 327 Effect Of Radial
Bearing Stiffness On
Rotor Mode Shapes For
The First Three Criticals Brg. Brg. Brg.
Hard Supports Soft Supports
3rd
Mode
2nd
Mode
1st
Mode
Brg.
120 Chapter-3
resonance. Thi s i s the cri ti cal speed duri ng whi ch shaft moti on pi vots through
the centerl i ne of the rotor produci ng a zero axi s crossi ng nodal poi nt wi thi n the
rotor span. Wi th a soft support sti ffness, a pure coni cal mode i s observed. I t i s
evi dent that i ncreasi ng sti ffness wi l l cl amp down at the beari ng l ocati ons, and
cause the formati on of two more nodal poi nts. Agai n, the vi brati on measurement
pl anes are extremel y i mportant, and a machi ne coul d be i nterpreted as passi ng
through a rst cri ti cal , when i t i s real l y transcendi ng through a wel l restrai ned
second mode. As wi th any type of modal anal ysi s, the use of transducers at addi -
ti onal l ateral l ocati ons al ong the shaft may be qui te i nformati ve, and i n some
cases absol utel y necessary.
The thi rd cri ti cal speed wi th soft beari ngs i s si mi l ar i n shape to a transl a-
ti onal rst mode wi th hard supports. Agai n, the di agnosti ci an i s cauti oned about
premature mode shape specul ati ons based upon parti al data. The thi rd mode
wi th hard beari ngs exhi bi ts a predi ctabl e shape wi th four nodal poi nts. For many
rotors thi s thi rd cri ti cal speed i s a rotor bendi ng mode, and shaft al terati ons are
usual l y requi red to appreci abl y change thi s mode shape or the associ ated natu-
ral frequency. Conversel y, the rst two modes are normal l y beari ng dependent.
The mode shapes and natural frequenci es of the rst and second cri ti cal s can
often be al tered by changi ng the sti ffness characteri sti cs of the radi al journal
beari ngs and/or the sti ffness of the beari ng housi ngs.
CRITICAL SPEED TRANSITION
The nomi nal 180 phase change across a shaft nodal poi nt i s general l y
understandabl e. The concept of a shaft reversi ng di recti on (i .e., pi voti ng) across a
nodal poi nt makes physi cal sense, and extensi ve techni cal expl anati ons are not
requi red. However, the ampl i tude change, and the i deal 180 phase shi ft associ -
ated wi th a speed transi ti on through a cri ti cal speed (bal ance resonance) do not
l end themsel ves to di rect i ntui ti ve l ogi c. Certai nl y the anal yti cal model s exhi bi t
these characteri sti cs (chapters 2 and 5), and the el d vi brati on measurements
(chapters 7, 8, 11, etc.) al so di spl ay vector changes across a rotor resonance.
However, an i nsti ncti ve, and uni versal understandi ng of thi s behavi or by
mechani cal l y i ncl i ned i ndi vi dual s i s not a common trai t.
Thi s i s not a new topi c, and references date back to the post ci vi l war era.
For i nstance, the 1869 techni cal paper by W.A. Ranki n
21
i ni ti al l y addressed thi s
subject. By 1882, Carl Gustaf de Laval of Sweden i ntroduced a exi bl e shaft, si n-
gl e stage steam turbi ne. The German teacher and researcher August Fppl dem-
onstrated i n 1895 that a rotor coul d operate above a cri ti cal speed. I n 1919, the
Engl i sh i nvesti gator H.H. Jeffcott
22
publ i shed hi s cl assi c paper The Lateral
Vibration of Loaded Shafts in the Neighborhood of a Whirling Speed: The Effect
of Want of Balance. Fi fteen years l ater, McGraw Hi l l Book Company began pub-
21
W.A. Ranki n, On the Centri fugal Force of Rotati ng Shafts, Engineer (London): 27, (1869).
22
H.H. Jeffcott, Lateral Vi brati on of Loaded Shafts i n the Nei ghborhood of a Whi rl i ng Speed -
The Effect of Want of Bal ance, Philosophical Magazine, Vol 37, (1919), pp. 304-314.
Critical Speed Transition 121
l i shi ng the J.P. Den Hartog
23
textbook Mechanical Vibrations. I n 1948, Prenti ce-
Hal l publ i shed the rst edi ti on of the W.T. Thomson
24
textbook al so ti tl ed
Mechanical Vibrations. The academi c expl anati ons conti nue wi th many papers
and textbooks such as the 1988 Rotordynamics of Turbomachinery by John
Vance
25
, fol l owed by the 1993 Turbomachinery Rotordynamics, Phenomena,
Modeling, and Analysis by Dara Chi l ds
26
. Obvi ousl y, these expl anati ons of fun-
damental rotor behavi or have gone on for wel l over a century.
I nteresti ngl y enough, Jeffcott corrected the earl i er work of Ranki n. Some
hi stori ans bel i eve that the Jeffcott work shoul d be credi ted to de Laval or Fppl .
I t appears that Den Hartog, and Thomson both used the Jeffcott model , but they
apparentl y negl ected to reference the earl i er work of Jeffcott. The books by Vance
and Chi l ds do recogni ze Jeffcott, and hi s contri buti ons to the el d of rotordynam-
i cs. However, many good machi nery engi neers have expressed the opi ni on to the
seni or author that the physical understanding of the critical speed phenomena
still seems to be clouded by the lack of a direct physical explanation of the ampli-
tude change and the phase shift through the resonance.
Si nce thi s i s such an i mportant concept to the el d of rotati ng machi nery,
another expl anati on wi l l be attempted on the fol l owi ng pages. Thi s i nterpreta-
ti on of the phenomena i s based upon the Jeffcott model , and comments by Den
Hartog and Thomson. Thi s expl anati on wi l l not be as mathemati cal l y ri gorous or
as extensi ve as the di scussi ons by ei ther Vance or Chi l ds. However, the deri va-
ti on presented herei n i s si mpl y di rected at a physi cal expl anati on of rotor behav-
i or through a cri ti cal speed regi on. I t i s hoped that the fol l owi ng di scussi on does
make i ntui ti ve sense to the machi nery di agnosti ci an.
To begi n thi s di scussi on, i t i s appropri ate to consi der the tradi ti onal di a-
gram of a Jeffcott rotor as depi cted i n Fi g. 3-28. The beari ngs i n thi s model are
mounted at the ends of the shaft, and they are consi dered to be ri gi d, and fri c-
ti onl ess. I ni ti al l y, the dampi ng wi l l i ntenti onal l y be set to zero, and i nuences
from any ui ds or other sources wi l l be i gnored. The shaft i s uni form, massl ess,
and exi bl e (i .e., el asti c), and the shaft wi l l have a uni form radi al sti ffness
dened by K.
At the mi dspan of the shaft, a thi n, at di sk i s mounted. Thi s di sk wi l l have
a resi dual , or concentrated mass unbal ance that i s i denti ed as M. Thi s si ngl e
mass unbal ance i s l ocated on the mass centerl i ne i denti ed as G (not to be con-
fused wi th the accel erati on of gravi ty, or the modul us of ri gi di ty). As shown i n
Fi g. 3-28, the beari ng centerl i ne B, and the shaft and di sk centerl i ne S are i den-
ti ed. When the rotor i s at rest, the beari ng, shaft, and di sk centerl i nes are coi n-
ci dent. The di stance between the shaft and di sk centerl i ne S, and the mass
centerl i ne G, wi l l be dened as the eccentri ci ty e. Thi s i s not the journal offset
23
J.P. Den Hartog, Mechanical Vibrations, (New York: McGraw-Hi l l Book Company, 1934).
24
Wi l l i am Tyrrel l Thomson, Mechanical Vibrations, (Engl ewood Cl i ffs, New Jersey: Prenti ce-
Hal l , I nc., 1948).
25
John M. Vance, Rotordynamics of Turbomachinery (New York: Wi l ey-I ntersci ence Publ i ca-
ti on, 1988).
26
Dara Chi l ds, Turbomachinery Rotordynamics Phenomena, Modeling, and Analysis, (New
York: Wi l ey-I ntersci ence Publ i cati on, 1993).
122 Chapter-3
wi thi n a beari ng, or an eccentri c surface. I t i s merel y the di stance between the
center of the shaft and di sk S, and the unbal ance mass M (and associ ated mass
centerl i ne G). Throughout thi s di scussi on, the geometry of the di sk and the mass
remai ns xed, therefore the eccentri ci ty di stance e remai ns constant.
At very sl ow speeds, unbal ance forces are negl i gi bl e. The shaft turns
around the beari ng centerl i ne B, and al l rotati ng el ements are concentri c. Thi s
condi ti on i s depi cted i n the end vi ew of Fi g. 3-29. A di al i ndi cator or a proxi mi ty
probe mounted anywhere al ong the l ength of the shaft wi l l i ndi cate zero moti on.
As rotor speed i ncreases, the strai ght shaft wi l l deect i nto the predi ctabl e
mode shape shown i n Fi g. 3-30. The transl ati onal mode shapes wi thi n thi s di s-
cussi on exhi bi t zero moti on at the beari ngs, and a maxi mum deecti on at the
shaft mi dspan. The onl y dri vi ng force i n the system i s the centri fugal force due to
the unbal ance mass M. The fol l owi ng seri es of di agrams are consi stentl y anno-
tated wi th the previ ous centerl i ne desi gnati ons (B, S, and G), and the mass
eccentri ci ty e (offset) from the deected shaft center. I n ti me honored tradi ti on,
the maxi mum bendi ng deecti on of the shaft (di stance between B and S) i s i den-
ti ed by r. Furthermore, the rotati onal speed i s i ndi cated by i n al l of the di a-
grams and equati ons.
By i nspecti on of these di agrams, i t i s cl ear that the shaft and di sk are rotat-
i ng at the operati ng speed . Si mul taneousl y, the deected shaft i s whi rl i ng i n
the beari ngs at thi s speed, and i t carri es the di sk al ong wi th i t as i t moves. Thi s
moti on i s often referred to as synchronous whi rl . The mechani sm dri vi ng thi s
whi rl i s the centri fugal force generated by the eccentri c mass on the di sk. As
Fig. 328 Jeffcott Rotor Conguration
Fig. 329 Jeffcott Rotor At Very
Low Shaft Rotational Speed
Unbalance
Mass (M)
Uniform, Massless Shaft,
Flexible with Stiffness (K)
Rigid, Frictionless
Bearings, No Damping
S B
G
e
G
B S
G
Shaft & Disk
Center
Mass
Center
Bearing
Center

e
Critical Speed Transition 123
rotor speed i ncreases, the outward force i ncreases i n accordance wi th the normal
centri fugal force equati on:
(3-72)
I n thi s expressi on, the total radi us of the mass unbal ance M i s composed of
the shaft bendi ng r, pl us the eccentri ci ty of the mass wi th respect to the shaft
centerl i ne e. I f the di sk was perfectl y bal anced (i .e., M=0) there woul d be no
deecti on of the shaft, and no resul tant whi rl . Wi th respect to real machi nes,
there i s al ways some amount of shaft bendi ng, and some l evel of resi dual unbal -
ance i n the rotor that produces synchronous whi rl . The physi cal i nuence of the
centri fugal force dened i n equati on (3-72) may be graphi cal l y depi cted i n Fi gs.
3-30 and 3-31. The rst drawi ng (Fi g. 3-30) descri bes a moderate speed bel ow the
bal ance resonance (cri ti cal ) speed. The condi ti on descri bed i n Fi g. 3-31 depi cts a
hi gher speed that i s approachi ng the cri ti cal speed. Note that the two drawi ngs
are i denti cal except for the i ncreased mi dspan deecti on r caused by the hi gher
operati ng speed, and the associ ated l arger centri fugal force.
The end vi ews of the mi dspan di sk i n Fi gs. 3-30 and 3-31 show the beari ng
center B, shaft center S, and mass G. Note that the centri fugal force i s i n the
same l i ne as B, S, and G. These poi nts are col l i near. That i s, they l i e al ong a com-
mon strai ght l i ne. The l ogi c behi nd thi s statement i s based on a force bal ance.
Speci cal l y, the centri fugal force acts outward from the beari ng center B through
the mass center G. The onl y opposi ti on to thi s radi al centri fugal force i s the
Fig. 330 Jeffcott Rotor At Moderate
Speed Well Below the Balance Resonance
Fig. 331 Jeffcott Rotor Approaching The
Critical Balance Resonance Speed
B B
S S
G G
r
e
B
S
G
Centrifugal
Force
e
r
Shaft & Disk
Center
Mass
Center
Bearing Center

Restoring
Force
Midspan Orbit
B B
S S
G G
r

e
S
G
Centrifugal
Force
e
r
Shaft & Disk
Center
Mass
Center

Restoring
Force
B
Bearing
Center
Midspan Orbit
Centri fugal Force M r e + ( )
2
=
124 Chapter-3
restori ng force of the shaft. The restori ng force i s si mpl y the shaft spri ng con-
stant K mul ti pl i ed by the shaft deecti on r as descri bed i n the next expressi on:
(3-73)
The restori ng force i s the el asti c pul l of the shaft that attempts to
strai ghten the shaft, and resi st any deecti on. I t acts from the shaft center S,
back to the beari ng center B. I n order for the centri fugal and the restori ng forces
to be i n equi l i bri um, they must be equal i n magni tude, and opposi te i n di recti on.
Hence, for the si mpl e case, poi nts B, G, and S must be col l i near. Based upon thi s
l ogi c, i t i s reasonabl e to equate the two opposi ng forces as fol l ows:
(3-74)
Substi tuti ng (3-72) and (3-73) i nto equati on (3-74) the fol l owi ng i s obtai ned:
(3-75)
From thi s expressi on, a mechani cal di l emma i s i mmedi atel y apparent. The
l eft si de of the expressi on that represents the restori ng force i s l i near wi th shaft
deecti on. However, the ri ght si de of thi s expressi on that denes centri fugal
force, vari es as the speed squared. Hence, at some rotati onal speed, the shaft di s-
pl acement must become i nni te to restrai n the di sk and attached unbal ance
mass. Thi s poi nt of i nni te di spl acement i s aptl y termed the cri ti cal speed, or the
bal ance resonance speed of the rotati ng assembl y.
Another way to demonstrate thi s concept of i nni te ampl i tude i s based
upon an expansi on, and re-congurati on of (3-75) i n the fol l owi ng manner:
Thi s expressi on may now be sol ved for the bendi ng deecti on r as fol l ows:
(3-76)
From equati on (2-41), i t was shown that (K/ M) i s equal to the undamped
natural frequency
c
(cri ti cal speed) squared. Substi tuti ng thi s expressi on i nto
equati on (3-76), the deecti on r may now be restated as:
(3-77)
The eccentri ci ty e between the shaft center S, and the mass centerl i ne G
remai ns constant. I n addi ti on, the natural frequency
c
remai ns xed due to an
Restori ng Force K r =
Restori ng Force Centri fugal Force =
K r M r e + ( )
2
=
K r M r
2
M e
2
+ =
K r M r
2
M e
2
=
r K M
2
( ) M e
2
=
r
M e
2

K M
2

------------------------------
e
2

K M ( )
2

-------------------------------- = =
r
e
2

c
2

------------------- =
Critical Speed Transition 125
i nvari abl e mass M and spri ng constant K. Hence, the shaft deecti on r i s a func-
ti on of the operati ng speed . Furthermore, i f the numerator and denomi nator of
(3-77) are di vi ded by the cri ti cal speed squared
c
2
, the fol l owi ng i s obtai ned:
(3-78)
I n an effort to further si mpl i fy thi s expressi on, recal l that the undamped
cri ti cal speed frequency rati o was dened i n equati on (2-86) as:
Substi tuti ng thi s speed rati o i nto (3-78) yi el ds the next rel ati onshi p:
(3-79)
As the shaft rotati onal speed i ncreases from some sl ow rol l condi ti on, the
val ue of the cri ti cal speed rati o i ncreases, and the mi d span shaft deecti on r
i ncreases. I f the eccentri ci ty e between the mass centerl i ne G, and the shaft cen-
terl i ne S i s assumed to be some real i sti c val ue such as +2, Fi g. 3-32 may be pl ot-
ted. As expected, the mi dspan deecti on becomes qui te l arge as the cri ti cal speed
(=1) i s approached.
Stated i n another way, as the rati o of /
c
approaches uni ty, the denomi na-
tor of equati on (3-79) becomes smal l er, and the val ue of r becomes i ncreasi ngl y
l arge. When the shaft speed i s equal to the cri ti cal speed
c
, the =
c
/
c
=1,
and the denomi nator of equati on (3-79) becomes zero. Obvi ousl y, di vi si on by zero
wi l l resul t i n i nni ty. Thi s i s consi stent wi th the previ ous l ogi c, and the deni -
ti on of an undamped system.
The response pl ot i n Fi g. 3-32 was constructed between val ues of 0 and
0.94 to descri be the ampl i tude characteri sti cs bel ow the resonance. I f the pl ot
Fig. 332 Plot Of Jeffcott
Rotor Approaching The
Critical Balance Reso-
nance Speed of =1
r
e
c
( )
2

1
c
( )
2

------------------------------- =
Cri ti cal Speed Frequency Rati o
c
= =
r
e
2

1
2

---------------- =
0
2
4
6
8
10
12
14
16
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
S
h
a
f
t

C
e
n
t
e
r

D
e
f
l
e
c
t
i
o
n

(
r
)
Frequency Ratio (=/
c
)
r
e
e
r
=

= +
=

2
2
2
2
1
2
1
and for :
2
126 Chapter-3
range i s extended from 0 to 3.0, the di agram shown i n Fi g. 3-33 evol ves. Agai n, a
numeri c val ue of +2 was sel ected for the eccentri ci ty e, and i t i s noted that the
mi dspan shaft deecti on i ndeed moves off towards i nni ty. I n a real machi ne,
i nni tel y l arge shaft deecti ons are not possi bl e. Onl y two possi bi l i ti es for the
Jeffcott (or any other) rotor are feasi bl e. The rst opti on i s for the mi dspan
ampl i tudes to i ncrease to the poi nt where the machi ne destroys i tsel f (onl y
choi ce for an undamped system). The other al ternati ve for a real machi ne i s to
have the di spl acement ampl i tudes at the cri ti cal speed restrai ned by dampi ng.
I n accordance wi th the earl i er di scussi on wi thi n thi s chapter, posi ti ve
dampi ng i s an energy di ssi pater. I t wi l l l i mi t the vi brati on ampl i tude through
the resonance. For a system wi th l ow dampi ng, such as a structural resonance,
the ampl i tude at the resonance wi l l be hi gh, and the resonance bandwi dth wi l l
be smal l (hi gh Q). Conversel y, for a system wi th l arge dampi ng, such as a ui d
l m beari ng wi th vi scous dampi ng, the resonance peak wi l l be l ower, and the
bandwi dth wi l l be wi der (smal l Q). Thus, wi thout dampi ng i n the system, the
machi ne coul d not survi ve a resonance. Thi s di scussi on al so i denti es the l ogi c
associ ated wi th the ampl i tude i ncrease at the cri ti cal speed. I t i s hopeful l y cl ear
from the precedi ng expl anati on, and the general equati ons, that the di spl ace-
ment at the resonance must i ncrease. Furthermore, the amount of the vi brati on
i ncrease at the cri ti cal i s dependent upon the avai l abl e system dampi ng.
Unfortunatel y, the characteri sti cs of the ampl i tude response through the
resonance are not compl etel y dened, because one other pecul i ari ty must be rec-
onci l ed. Speci cal l y, Fi g. 3-33 reveal s a mi dspan ampl i tude r that mi grates off
towards posi ti ve i nni ty (+) as the resonance i s approached. I n true mathemat-
i cal fashi on, the ampl i tude above the cri ti cal returns from negati ve i nni ty (-).
That i s parti cul arl y di sturbi ng when i t i s al so recogni zed that the ampl i tudes
above the resonance are al l negati ve. For exampl e, i f e=2, and =2, then by equa-
ti on (3-79), r i s computed to be mi nus 2.67. Hence, the pl otted curve i s mathe-
mati cal l y correct, but i t does not descri be a true physi cal si tuati on. That i s, a
negati ve vi brati on ampl i tude i s i ncomprehensi bl e.
Fig. 333 Plot Of Jeffcott
Rotor Passing Through
The Critical Balance Res-
onance Region of =1
-60
-40
-20
0
20
40
60
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
S
h
a
f
t

C
e
n
t
e
r

D
e
f
l
e
c
t
i
o
n

(
r
)
Frequency Ratio
From Minus
To Plus
(=/
c
)
r
e
e
r
=

= +
=

2
2
2
2
1
2
1
and for :
2
Critical Speed Transition 127
I n retrospect, equati on (3-79) used for computati on of thi s deecti on r onl y
contai ns two terms. The eccentri ci ty e, pl us the frequency rati o between the run-
ni ng speed and the natural frequency =/
c
. Above the resonant frequency
c
i t i s obvi ous that thi s speed rati o must be a posi ti ve number that i s greater than
one. Hence, there i s no vari ati on of the frequency rati o that woul d reverse the
si gn of the mi dspan deecti on r.
The onl y other al ternati ve resi des wi th the eccentri ci ty e. As previ ousl y
sti pul ated, the magni tude of e i s xed by the geometry of the di sk, and the l oca-
ti on of the mass on the di sk. However, the ori gi nal deni ti on of the Jeffcott rotor
di d not restri ct the di recti on (i .e., the si gn) of thi s eccentri ci ty e. Hence, i f the
di recti on of e was reversed, the pl otted response curve shoul d i p over i nto the
posi ti ve domai n.
I n fact, that i s exactl y what happens wi th a real machi ne, as wel l as the Jef-
fcott model . To prove thi s poi nt, the response data above the resonance wi l l be
repl otted i n Fi g. 3-34, and e=+2 wi l l be repl aced by e=-2. I n order to mai ntai n
reasonabl e ampl i tudes, a frequency rati o range extendi ng from 1.08 to 3.0 wi l l be
used. I t i s noted that the resul tant pl ot i l l ustrates the proper posi ti ve deecti on
r, and i t al so i s i ndi cati ve of normal behavi or on the back sl ope of a resonance.
The si gn reversal of the eccentri ci ty e i s physi cal l y equi val ent to a reversal
i n the posi ti ons of the shaft S, and mass G centerl i nes. For i nstance, compare the
Jeffcott rotor runni ng bel ow the cri ti cal i n Fi gs. 3-30 and 3-31, wi th the rotor
operati ng above the cri ti cal as shown i n Fi gs. 3-35 and 3-36. Bel ow the cri ti cal ,
the mass M was on the outsi de of the rotor. Whereas, above the cri ti cal speed, the
mass M i s tucked away underneath the curvature of the shaft and the l i ght
si de of the rotor i s now on the outsi de. Thi s reversal of heavy side out to light side
out i s the mechani sm behi nd the nomi nal 180 phase shi ft across a resonance.
Another i nteresti ng poi nt from the response pl ot i n Fi g. 3-34 i s that the
val ue of r steadi l y di mi ni shes wi th an i ncreasi ng frequency rati o . At hi gh shaft
speeds, the magni tude of r approaches the magni tude of e. Physi cal l y, thi s char-
acteri sti c i s descri bed i n Fi g. 3-36 of the Jeffcott rotor at a hi gh speed condi ti on.
Fig. 334 Plot Of Jeffcott
Rotor Leaving Critical Bal-
ance Resonance Speed
Region of =1 with e=-2
0
2
4
6
8
10
12
14
16
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
S
h
a
f
t

C
e
n
t
e
r

D
e
f
l
e
c
t
i
o
n

(
r
)
Frequency Ratio
At High Speeds "r" approaches "e"
(=/
c
)
r
e
e
r
=

=
=

( )


2
2
2
2
2
2 2
1
2
1 1
2
1 1
and for the condition of :
-2 2
128 Chapter-3
I n thi s state (i .e., wel l above the cri ti cal speed), the val ues of e and r are essen-
ti al l y equal . Si nce e carri es a negati ve si gn (above the cri ti cal ), the centerl i ne for
the mass G i s now coi nci dent wi th the beari ng centerl i ne B. Thi s di recti on rever-
sal el i mi nates the centri fugal force, si nce the radi us to the mass i s now equal to,
or very cl ose to zero, and the rotor has sel f bal anced i tsel f.
Thi s sel f-bal anci ng characteri sti c i s the mechani sm that al l ows so many
machi nes to operate successful l y at speeds i n excess of the rotor bal ance reso-
nance, or cri ti cal speed. I n thi s hi gh speed condi ti on, the system i s i n equi l i b-
ri um, and the shaft actual l y rotates around the mass center M, whi ch i s
equi val ent to the centerl i ne of gravi ty G for the rotor. For further di scussi ons
wi thi n thi s chapter, and especi al l y the bal anci ng chapter 11, reference wi l l be
made to rotati on about the geometri c axi s bel ow the cri ti cal speed, and rotati on
about the mass or i nerti al centerl i ne above the cri ti cal . These conti nui ng refer-
ences are based upon the fundamental behavi or and understandi ng of the tradi -
ti onal Jeffcott rotor.
At the begi nni ng of thi s di scussi on, i t was stated that the dampi ng was
i ntenti onal l y set at zero. That was a conveni ent and necessary presumpti on to
mai ntai n si mpl i ci ty of the anal yti cal model . The i ncl usi on of dampi ng i nto the
system does substanti al l y compl i cate the response of the Jeffcott rotor. For
i nstance, consi der Fi g. 3-37 of thi s rotor wi th dampi ng. Note that the centerl i nes
B, S, and G are no l onger col l i near. An angul ar phase change has been i nvoked
upon the mass centerl i ne G. Thi s i mpl i es that the centri fugal force, and the shaft
Fig. 335 Jeffcott Rotor Operating Slightly
Above The Critical Balance Resonance
Speed
Fig. 336 Jeffcott Rotor Operating In A
Self Balanced Condition Well Above The
Critical Balance Resonance Speed
B B
S S
G G
r
e
S
G
Centrifugal
Force
e
r
Shaft & Disk
Center
Mass
Center

Restoring
Force
B
Bearing
Center
Midspan Orbit

B
S S
G
r

e
B
S
G
e
Shaft & Disk
Center
Mass
Center
Bearing
Center
r
Midspan Orbit
Critical Speed Transition 129
spri ng restori ng force are no l onger equal and opposi te. Actual l y, a new force
must now be i ncl uded to achi eve a force bal ance. Thi s new force wi l l be a damp-
i ng term that may be tangenti al to the di sk, and may vary i n magni tude wi th the
shaft surface vel oci ty.
The addi ti on of thi s dampi ng does not i nval i date the previ ous di scussi on,
but i t does compl i cate the scenari o. For si mpl i ci ty, the dampi ng force mi ght be
assumed to be vi scous, and the resul tant force equal to the tangenti al vel oci ty
ti mes the dampi ng coefci ent. Thi s type of force coul d be i ncl uded, and the previ -
ous anal ysi s repeated. However, thi s i ncl usi on woul d not necessari l y i mprove the
overal l understandi ng of the cri ti cal speed phenomena. Furthermore, dampi ng
characteri sti cs of real machi nes are not a si mpl e l i near functi on of rotor vel oci ty,
and a proper anal ysi s woul d be substanti al l y more i ntri cate.
I n essence, i t must be recogni zed that al l rotors have some amount of damp-
i ng. The oi l l m i n the beari ngs, oi l i n the shaft seal s, or the process ui d i tsel f
may provi de the dampi ng. I n al l cases, the presence of dampi ng wi l l i nuence
the shaft behavi or. There wi l l be posi ti ve contri buti ons to the addi ti on of damp-
i ng, such as l ower vi brati on ampl i tudes through the cri ti cal speed regi on. How-
ever, dampi ng wi l l al so open the door to a vari ety of mechani sms such as non-
synchronous whi rl . I n al l cases, the real worl d i s al ways more compl i cated than
the model s that we can bui l d to expl ai n physi cal events. Fortunatel y, the
undamped Jeffcott model may be used to expl ai n the fundamental characteri s-
ti cs associ ated wi th the bal ance resonance or cri ti cal speed phenomena.
Fig. 337 Jeffcott Rotor With Damping
B
S
G
Shaft & Disk
Center
Mass
Center
Bearing
Center

130 Chapter-3
MODE SHAPE MEASUREMENT
Previ ous porti ons of thi s chapter have bri ey covered the deecti on of stati c
beams wi th vari ati ons i n wei ght di stri buti on, and support l ocati on. I n addi ti on,
the i nuence of dampi ng, and the effect of rotor support sti ffness have been
addressed. Al though these are i mportant concepts that di rectl y i nuence the
behavi or of rotati ng machi nery, there i s a strong argument to be made for di rect
observati on of the deected rotor mode shapes.
A vi sual observati on of the rotor mode shapes woul d provi de substanti ati on
of the previ ousl y di scussed theory, and al l ow further study i nto the compl exi ti es
and subtl eti es of rotor dynami cs. Fortunatel y, a method exi sts to provi de thi s
vi sual i zati on by the appl i cati on of proxi mi ty probe transducers, and el ectroni c
di spl ay i nstrumentati on. For exampl e, consi der Fi g. 3-38 of a si mpl e rotor that
consi sts of a shaft pl us a center mounted di sk. Thi s i s si mi l ar to the Jeffcott
rotor. For thi s model , assume that the rotor and di sk are i ni ti al l y wel l bal anced,
and that system dampi ng i s mi ni mal . Further presume that a si ngl e concen-
trated mass unbal ance i s pl aced on the di sk.
The outer di ameter of the di sk i s observed by a verti cal proxi mi ty di spl ace-
ment probe that measures the di stance between the probe ti p and the di sk outer
surface. I n many ways, thi s type of vi brati on transducer may be consi dered as
the el ectroni c equi val ent of a mechani cal di al i ndi cator. Detai l ed di scussi ons of
the characteri sti cs of thi s type of probe are presented i n chapters 6 and 7. How-
ever, for the current di scussi on, the reader shoul d accept that thi s probe wi l l
accuratel y measure the di stance between the probe ti p and the observed surface.
The condi ti oned output from thi s proxi mi ty probe transducer system i s a
vol tage sensi ti ve si gnal that may be observed on an osci l l oscope. The typi cal
scal e factor for thi s type of transducer i s 200 mi l l i vol ts per Mi l . Thus, a 1 Mi l
(0.001 I nch) change i n di stance between the probe and the target surface wi l l
produce an el ectroni c si gnal equal to 200 mi l l i vol ts. I f the di stance between the
rotor surface and the probe remai ns constant at al l angul ar posi ti ons, the osci l l o-
scope wi l l di spl ay a strai ght l i ne as the shaft i s rotated. However, i f the observed
surface i s eccentri c (for whatever reason), then the probe wi l l di spl ay a si ne wave
on the osci l l oscope. The frequency of thi s si ne wave wi l l be equal to the speed of
Fig. 338 Simple Rotor With
Concentrated Mass Unbalance
And Midspan Radial Proximity
Probe
225
270
315 0
90
180
270
Mass Unbalance
C
C
W R
o
t
n
Proximity Probe
Mode Shape Measurement 131
rotati on, and the ampl i tude of the resul tant si ne wave wi l l be dependent upon
the magni tude of the eccentri ci ty.
An angul ar coordi nate system must be establ i shed and mai ntai ned. Thi s
angul ar coordi nate system must al l ow a deni ti ve and repeatabl e rel ati onshi p
between the rotati ng system, and a stati onary reference poi nt. Normal l y, thi s i s
achi eved by another proxi mi ty probe that observes a notch (keyway) or projec-
ti on (key) on the rotati ng shaft. Duri ng shaft rotati on, thi s ti mi ng or Keypha-
sortransducer produces a synchronous pul se as descri bed i n l ater chapters of
thi s text. The angul ar l ocati on of thi s pul se i s determi ned by stoppi ng the
machi ne, and physi cal l y l i ni ng up the tri gger poi nt on the notch or projecti on
wi th the ti mi ng probe. When thi s physi cal al i gnment occurs at zero speed, i t i s
equi val ent to the tri gger poi nt of a pul se si gnal duri ng rotati on.
Agai n, speci c detai l s of thi s tri gger arrangement are di scussed i n chapters
6, 7, 8, and 11. For the purposes of thi s current di scussi on, accept the fact that a
tri gger/reference arrangement does exi st. Furthermore, the vi brati on sensi ng
probe i s al ways l ocated at zero degrees for al l phase measurements. I n the exam-
pl e rotor shown i n Fi g. 3-38, the vi brati on sensi ng proxi mi ty probe i s shown at
the top verti cal posi ti on above the di sk. The rotati on i s speci ed as countercl ock-
wi se, the angul ar coordi nate system on the di sk begi ns at 0 at the probe, and
the angl es i ncrease i n a cl ockwi se di recti on. Thi s same l ogi c wi l l be used
throughout thi s text.
Al so, i t must be understood that the proxi mi ty probe cannot di rectl y i den-
ti fy the angul ar l ocati on of an effecti ve mass unbal ance. The proxi mi ty probe can
onl y measure di stances. That i s, the probe can measure the change i n di stance
between transducer and rotor around the enti re ci rcumference, but i t cannot
di rectl y i denti fy the l ocati on of a mass unbal ance. Thi s effecti ve, or equi val ent, or
l umped mass unbal ance l ocati on i s normal l y i denti ed as the Heavy Spot for the
di sk or rotor system. The ci rcumferenti al poi nt that the proxi mi ty probe does
i denti fy i s the hi gh poi nt of the observed surface. Thi s i s the poi nt of peak vi bra-
ti on that i s i denti ed by the synchronous vector phase angl e. Thi s physi cal l oca-
ti on i s general l y referred to as the High Spot.
Consi der a cl oser exami nati on of the wheel on the si mpl e rotor descri bed i n
Fi g. 3-38. The di agram presented i n Fi g. 3-39 represents the moti on of thi s wheel
at l ow rotati ve speeds. For purposes of di scussi on, assume that the rotor i s oper-
ati ng at a speed that i s wel l bel ow the rst cri ti cal speed (i .e., transl ati onal bal -
ance resonance) of the system. I n thi s drawi ng, the center of rotati on i s shown to
be coi nci dent wi th the geometri c center. A resi dual or effecti ve mass unbal ance i s
shown as the Heavy Spot.
I f thi s rotor was perfectl y bal anced, then the Mass Center woul d be i denti -
cal to the Geometric Center. However, i f a l umped unbal ance i s pl aced on the
di sk, i t i s cl ear that a shi ft i n the Mass Center must occur. I t i s al so l ogi cal to rec-
ogni ze that the new Mass Center must resi de on a l i ne between the Geometric
Center and the Heavy Spot. Furthermore, as the rotor turns, centri fugal force
wi l l cause the di sk to deect radi al l y i n the di recti on of the Heavy Spot. Under a
l ow speed condi ti on, the High Spot occurs at the same angul ar l ocati on as the
Heavy Spot. Stati ng i t another way, the mi ni mum di stance between the di sk and
132 Chapter-3
the probe occurs at the High Spot. Cl earl y, thi s i s coi nci dent wi th the Heavy Spot
at speeds wel l bel ow the cri ti cal speed (thi s i s exactl y true onl y i n the compl ete
absence of dampi ng).
I f rotati on of thi s exampl e system i s countercl ockwi se, the angl es woul d be
measured agai nst rotati on, or cl ockwi se from the probe. Thus, the Heavy-High
Spot angul ar posi ti on i s shown at a physi cal l ocati on of 110 i n Fi g. 3-39. I f the
shaft i s turni ng, the phase angl e obtai ned from a Vector Fi l ter woul d be 110. On
the vi brati on waveform, thi s woul d be equi val ent to the ti me l ag between the
peak of the vi brati on si gnal , and the tri gger poi nt of the Keyphasorpul se. I f the
shaft i s not rotati ng, and the ti mi ng notch i s posi ti oned under the Keyphasor
probe, then a counter rotati on angl e of 110 from the verti cal vi brati on probe
woul d l ocate the High Spot, and the associ ated Heavy Spot. Bal anci ng of thi s
rotor woul d requi re removi ng wei ght at 110 or addi ng wei ght at 290. Thi s i s
the l ogi c behi nd the tradi ti onal proxi mi ty probe bal anci ng rul e of:
At speeds well below the critical,
remove weight at the phase angle,
or add weight at the phase angle plus 180
Next, consi der what happens at rotati onal speeds wel l above the rst cri ti -
cal . Thi s condi ti on i s descri bed i n Fi g. 3-40. I n thi s drawi ng, the center of rota-
ti on i s now coi nci dent wi th the actual mass center of the rotor. That i s, rotati on
occurs around the mass center, or pri nci pal i nerti a axi s i nstead of the geometri c
axi s. Thi s i s i denti cal to the behavi or descri bed by the Jeffcott rotor i n the sel f
bal anced condi ti on i n Fi g. 3-36.
The resi dual mass unbal ance or Heavy Spot remai ns i n the same angul ar
Fig. 339 Simple Rotor Operating Well
Below The Shaft Critical Speed
Fig. 340 Simple Rotor Operating Well
Above The Shaft Critical Speed
Probe
@ 0
90
180
270
Mass
Center
Geometric
Center
Center of
Rotation
Heavy
Spot High
Spot
C
CWRotation
Phase
= 110
Probe
@ 0
90
180
270
Mass
Center
Geometric
Center
Center of
Rotation
Heavy
Spot
C
C
W
Rotation
Phase
= 290
High
Spot
Mode Shape Measurement 133
l ocati on as the sl ow speed case. I ntui ti vel y, thi s must be true. That i s, the rel a-
ti onshi p between the probe and the unbal ance must remai n constant, or the
rotor coul d never be bal anced. For i nstance, an unbal anced and l oose i mpel l er on
a shaft woul d have a di fferent phase angl e duri ng every runup. Thi s wheel coul d
not be bal anced because the angl e between the probe and the unbal ance changes.
A rotor can onl y be bal anced when the angl e between the stati onary probe and
the Heavy Spot on the rotor remai ns constant from run to run.
At hi gh speed condi ti ons, the new center of rotati on mani fests as a new
High Spot. The rotor has sel f-bal anced i tsel f through the cri ti cal speed regi on,
and the new center of rotati on about the mass center produces an eccentri c rota-
ti on axi s. Thi s eccentri ci ty of rotati on about the mass center resul ts i n a new
High Spot that i s 180 away from the Heavy Spot. Stati ng i t another way, the
mi ni mum di stance between the di sk and the proxi mi ty probe occurs at the i den-
ti ed High Spot. Thi s l ocati on i s exactl y opposi te the Heavy Spot at rotati onal
speeds wel l above the cri ti cal (compl etel y true i n the case of no dampi ng).
As previ ousl y menti oned, the Heavy Spot remai ns at 110, and the new
l ocati on of the High Spot i s shown at a physi cal l ocati on of 290. I f the shaft i s
turni ng, the phase angl e obtai ned from a DVF woul d be 290. On the vi brati on
waveform, thi s woul d be equi val ent to the l ag between the peak of the vi brati on
si gnal , and the Keyphasorpul se. I f the shaft i s not turni ng, and the ti mi ng
notch i s posi ti oned under the Keyphasorprobe, then a counter rotati on angl e of
290 woul d l ocate the High Spot. The Heavy Spot woul d remai n at 290 mi nus
180 or at 110. Bal anci ng of thi s rotor woul d requi re addi ng wei ght at 290 or
removi ng wei ght at 110. Thi s i s the l ogi c behi nd the proxi mi ty probe bal anci ng
rul e of:
At speeds well above the critical,
add weight at the phase angle,
or remove weight at the phase angle minus 180
Obvi ousl y the wei ght correcti ons requi red bel ow and above the cri ti cal
speed woul d be performed i n exactl y the same angul ar l ocati ons. Thi s i s the
property that al l ows sl ow speed bal anci ng machi nes to correct the mass unbal -
ance characteri sti cs of hi gh speed exi bl e rotors operati ng over one or more cri ti -
cal speeds. However, more i mportant to the current topi c of rotor mode shapes; i t
i s cl ear that techni ques exi st to measure dynami c shaft moti on, and to i denti fy
shaft cri ti cal speeds. I t shoul d al so be menti oned that thi s exampl e was pre-
sented as a rotor wi th mi ni mal dampi ng. As previ ousl y menti oned, dampi ng wi l l
i nuence the response, and the i denti cati on of the true l ocati on of the resi dual
rotor unbal ance, i .e., the Heavy Spot. Thi s determi nati on i s usual l y made by the
computati on of bal ance sensi ti vi ty vectors as presented i n chapter 10. I n a hi ghl y
damped mechani cal system, the true verti cal and hori zontal vi brati on response
characteri sti cs must be careful l y exami ned, and the previ ousl y stated bal anci ng
rules of thumb judi ci al l y appl i ed.
Up to thi s poi nt, the di scussi on has consi dered vi bratory moti on onl y i n the
verti cal pl ane. Thi s i s acceptabl e i f the hori zontal support characteri sti cs are
i denti cal to the verti cal . However, i n many cases there i s a denabl e di fference
134 Chapter-3
between verti cal and hori zontal rotor support. Thi s di fference may be attri but-
abl e to vari ati ons i n oi l l m characteri sti cs, vari ati ons i n beari ng housi ng sti ff-
ness from verti cal to hori zontal , or a combi nati on of both. I n order to
accommodate thi s asymmetry, i t i s customary to i nstal l mutual l y perpendi cul ar
probes to observe both the verti cal and hori zontal vi brati on response character-
i sti cs. The i nstal l ati on of orthogonal proxi mi ty probe transducers provi des the
capabi l i ty to measure total moti on of the shaft wi thi n the beari ng.
I n addi ti on, the measurement of shaft mode shapes shoul d be consi dered as
a three-di mensi onal proposi ti on. Thi s i s due to the di fferences between hori zon-
tal and verti cal restrai nts, dampi ng, and exci tati ons al ong the l ength of the rotor.
An i nformati ve approach towards quanti cati on of three-di mensi onal mode
shapes consi sts of measuri ng shaft orbi ts at vari ous poi nts al ong the rotor. These
orbi ts may be combi ned i nto an i sometri c vi ew of the shaft wi th the respecti ve
orbi ts constructed at each measurement l ocati on.
The di agram i n Fi g. 3-41 i l l ustrates such a presentati on. I n thi s exampl e, a
3/8 i nch di ameter shaft Rotor Ki t i s congured wi th two sl eeve beari ngs pl us a
mi dspan mass. Rel ati ve shaft vi brati on i s observed wi th X-Y proxi mi ty probes
mounted at si x di fferent l ateral posi ti ons. The shaft orbi ts are pl otted at each
probe l ocati on, and the Keyphasordots are connected to descri be the deformed
rotor mode shape at a speci c speed.
Thi s drawi ng depi cts the rotor ki t operati ng at the verti cal transl ati onal
cri ti cal speed of 3,500 RPM. These smal l rotor ki ts are qui te sensi ti ve to support
characteri sti cs. Hence, i t i s normal to observe a hori zontal cri ti cal speed, fol -
l owed by a verti cal cri ti cal wi th a si mi l ar mode shape. At the measured cri ti cal
speed of 3,500 RPM, the center of the shaft i s consi derabl y more deformed than
at the beari ngs, and a typi cal rst transl ati onal resonance for thi s type of
machi ne i s noted.
Usi ng the same rotor ki t, and changi ng the sl eeve beari ngs to bal l beari ngs
(hi gher sti ffness), the resul tant orbi tal patterns and associ ated mode shapes are
presented i n Fi g. 3-42. The system resonant frequency has i ncreased to 6,300
RPM due to the hi gher support sti ffness, and the mi dpoi nt shaft deecti on has
greatl y i ncreased. I t i s al so noted that a zero axi s crossi ng nodal poi nt appears at
Fig. 341 Mode Shape Of
Rotor Kit With Soft Sleeve
Bearings And Center Mass
Soft Sleeve Bearings
1st Critical Speed = 3,500 RPM
Bearing Journal
Mass Between Bearings
Shaft
Mode
Shape
Mode Shape Measurement 135
both beari ngs. Thi s i s anal ogous to the hard support case presented i n Fi g. 3-27
where a deni te phase change occurs, and the orbi ts are compl etel y reversed
passi ng through a nodal poi nt.
Si nce the shaft moti on at the beari ngs i s qui te smal l , i t must be recogni zed
that beari ng dampi ng i s mi ni mal wi th thi s congurati on. The absence of damp-
i ng i ncreases the ampl i cati on factor through the cri ti cal , and i ncreases the sus-
cepti bi l i ty of the machi ne to a vari ety of i nstabi l i ty mechani sms. For sti ff
beari ngs, i t shoul d al so be noted that the unbal ance energy of the rotor i s cl osel y
coupl ed to the supporti ng structure. I n thi s congurati on, casi ng or structural
vi brati on i s l i kel y to be hi gher than the shaft rel ati ve moti on due to the di rect
transmi ssi bi l i ty of energy from the rotati ng shaft to the stati onary casi ng.
On l arge turbi nes or compressors that contai ns very sti ff beari ngs, the
desi gners often i ncl ude a ci rcumferenti al squeeze l m damper to provi de addi -
ti onal dampi ng. These squeeze l m dampers consi st of a non-rotati ng, l oose t-
ti ng annul us, around the beari ng outer di ameter. The cavi ty between the beari ng
outer di ameter and the i nner di ameter of the damper i s l l ed wi th l ube oi l . The
mi nor moti on (vel oci ty) of the beari ng housi ng wi th respect to the stati onary
damper provi des a means of energy di ssi pati on through the vi scous oi l l m.
Hence, thi s type of devi ce can be used to provi de more dampi ng for a poorl y
damped system.
Another popul ar machi ne congurati on consi sts of an overhung wheel wi th
a short i nboard beari ng span. Thi s arrangement i s used by machi nes such as
overhung bl owers, or power turbi nes on dual shaft gas turbi nes. The behavi or of
thi s type of mechani cal system i s domi nated by the mass and the gyroscopi cs of
the overhung wheel . To demonstrate the modal di fferences of a mass between
beari ngs versus an overhung mass, the rotor ki t was re-congured to the
arrangement shown i n Fi g. 3-43. Hard bal l beari ngs were agai n used, and orbi tal
measurements were obtai ned at 5 l ocati ons. Note that sti ff bal l beari ngs produce
nodal poi nts wi th a 180phase change across each node (beari ng). Al so observe
Fig. 342 Mode Shape Of
Rotor Kit With Hard Ball
Bearings And Center Mass
Bearing Journal
Hard Ball Bearings
1st Critical Speed = 6,300 RPM
Mass Between Bearings
Shaft
Mode
Shape
136 Chapter-3
the l arge excursi ons at the outboard end of the rotor, and the new cri ti cal speed
of 7,000 RPM. Cl earl y, the dynami c moti on outboard of the beari ngs descri bes
the anti ci pated coni cal mode shape. Thi s resul t i s to be expected, and i t i s ful l y
consi stent wi th the stati c beam di agrams presented i n Fi gs. 3-2, 3-4, and 3-6.
Al though the orbi tal techni que i s extremel y i nformati ve, i t i s someti mes
graphi cal l y di fcul t to draw a meani ngful i sometri c di agram of the rotor system.
Thi s i s often combi ned wi th the di fcul ty of i nstal l i ng mutual l y perpendi cul ar
probes at each desi red l ocati on. An al ternate approach consi sts of i nstal l i ng a
si ngl e proxi mi ty probe at each l ocati on. Thi s i nformati on provi des a si ngl e vector
quanti ty at each measurement poi nt, whi ch coul d be pl otted i n an i sometri c
manner as previ ousl y achi eved wi th the orbi ts. However, thi s agai n can be di f-
cul t from a graphi cs standpoi nt.
Another techni que for handl i ng the vector data i s to resol ve i t i nto Carte-
si an coordi nates and pl ot the si ne or cosi ne component at each measurement
posi ti on. Thi s approach wi l l provi de a si ngl e quanti ty, ei ther posi ti ve or negati ve,
as a functi on of posi ti on al ong the rotor. Thi s techni que provi des acceptabl e mode
shapes, and the mechani cs of pl otti ng the two-di mensi onal data i s usual l y qui te
strai ghtforward. I t shoul d al so be menti oned that thi s techni que, as wel l as the
other proxi mi ty probe measurements, must be corrected for sl ow rol l runout at
each probe l ocati on. Fai l ure to perform thi s basi c runout compensati on may eas-
i l y destroy the val i di ty of the dynami c shaft mode shape measurements.
Fig. 343 Mode Shape Of
Rotor Kit With Hard Ball
Bearings And Overhung
Mass
Bearing Journals
Hard Ball Bearings
1st Critical Speed = 7,000 RPM
Overhung Mass
Shaft
Mode
Shape
Mode Shape Measurement 137
Case History 4: Vertical Generator Mode Shape
Appl i cati on of any measurement techni que i s al ways predi cated upon the
avai l abi l i ty of accessi bl e measurement pl anes. Wi thi n the reni ng and petro-
chemi cal i ndustri es, rotati ng shafts are al most total l y encl osed, and the i nstal l a-
ti on of addi ti onal vi brati on transducers i s di fcul t i f not i mpossi bl e. However,
wi thi n the power generati on i ndustry, the mechani cal equi pment i s si gni cantl y
l arger, and exposed shaft surfaces are often avai l abl e.
Thi s accessi bi l i ty to the rotati ng el ements al l ows the di rect appl i cati on of
the previ ousl y di scussed measurement techni que. For exampl e, consi der the ver-
ti cal generator rotor di spl ayed i n Fi g. 3-44. Thi s generator i s dri ven by a hydro
turbi ne at a normal operati ng speed of 277 RPM. Physi cal l y, the turbi ne extends
downward from the generator, and thi s di agram pri mari l y depi cts the three
gui de beari ngs pl us the generator shaft. The wei ght of the enti re rotati ng system
approaches 100 tons, and the verti cal span between upper and l ower beari ngs i s
approxi matel y 27 feet. Al though the majori ty of the hydro turbi ne rotor i s under
water l arge porti ons of the generator shaft are external l y vi si bl e, and al l three
radi al gui de beari ngs are accessi bl e.
Fig. 344 Machinery And
Vibration Transducer
Arrangement For A Verti-
cal Generator Mounted On
Top Of A Hydro Turbine
Thrust
Bearing
Lower
Guide
Bearing
Upper
Guide
Bearing
Turbine
Guide
Bearing
CW
Rotation
Transducers and Rotation
Are Viewed From the Top
of the Generator Looking Down
2Y
2X
3X
3Y
4Y
4X
5Y
5X
Down To Kaplan Turbine
1Y
1X &
K
139"
30"
67"
92"
Elevation 1
Elevation 2
Elevation 3
Elevation 4
Elevation 5
Generator
90
90
90
90
90
138 Chapter-3
The machi ne i s normal l y moni tored wi th a pai r of proxi mi ty probes
mounted bel ow the l ower generator gui de beari ng. For the purposes of thi s test,
rel ati ve X-Y proxi mi ty probes were temporari l y i nstal l ed at ve di fferent el eva-
ti ons as shown i n the machi nery arrangement di agram, Fi g. 3-44. The di spl ace-
ment probes mounted at el evati ons 1, 2, 3, and 5 were attached to each
respecti ve beari ng housi ng wi th ri gi d brackets. The X-Y probes i nstal l ed above
the coupl i ng assembl y (El evati on 4) were qui te di fcul t to mount si nce the
machi ne had no stati onary el ements i n the vi ci ni ty of the coupl i ng. To address
thi s si tuati on, a uni -strut frame was constructed wi thi n the connes of the tur-
bi ne pi t. Thi s al umi num structure was attached to the wal l s, and i t was rei n-
forced wi th many cross braces to provi de a ri gi d support for the vi brati on
transducers observi ng the generator shaft above the coupl i ng assembl y.
At al l ve el evati ons, the structural moti on of the beari ng housi ngs and the
uni -strut frame provi ded a potenti al error source for the shaft di spl acement mea-
surements. I n order to quanti fy the magni tude, frequency content, and ti mi ng
rel ati onshi ps of the structural vi brati on, a seri es of ten l ow frequency accel erom-
eters were i nstal l ed on the machi ne. One accel erometer was mounted adjacent to
each radi al proxi mi ty probe, and the rel ati ve shaft moti on from the di spl acement
probe was acqui red si mul taneousl y wi th the casi ng absol ute vi brati on sensed by
the accel erometer. I n most cases, the casi ng accel erati on si gnal s were i nsi gni -
cant, and the majori ty of the vi brati on was rel ati ve moti on between the shaft
and the casi ng. However, under some test condi ti ons, the casi ng moti on had to be
i ncl uded, and the techni ques di scussed i n chapter 6 were appl i ed to determi ne
the overal l , or absol ute shaft vi brati on.
Under most test condi ti ons, the l argest error source was attri buted to sur-
face i mperfecti ons i n the observed shaft surfaces. Al l of the shaft surfaces were
pi tted and/or rusted, and the radi al vi brati on si gnal s had to be runout compen-
sated to mi ni mi ze the shaft surface i mperfecti ons. The normal operati ng speed of
277 RPM requi red that sl ow rol l data be obtai ned at a much l ower speed, and a
sl ow speed of 40 RPM was sel ected. Al though thi s sl ow rol l speed was a substan-
ti al percentage of the normal runni ng speed, i t was observed that mi ni mal
changes occurred between 60 and 20 RPM duri ng coastdown. Hence, 40 RPM
was sel ected as the nomi nal sl ow rol l speed si nce the Di gi tal Vector Fi l ter (DVF)
used for thi s data coul d correctl y process thi s l ow speed val ue.
Thi s parti cul ar machi nery trai n al ways experi enced a speed i ncrease dur-
i ng l oad rejecti on. Dependi ng on the actual generator l oad at the ti me of the tri p,
the rotor speed of the uni t mi ght accel erate from the normal val ue of 277 to as
hi gh as 450 RPM, and then coastdown to a stop. Dynami c vi brati on data coul d be
acqui red throughout the speed range, and sl ow rol l runout vectors may be sub-
tracted from the database to obtai n runout compensated synchronous 1X shaft
di spl acement. For exampl e, consi der the array at runout compensated 1X vectors
at 410 RPM presented i n Tabl e 3-5.
These compensated vectors carry the uni ts of Mi l s,
p-p
, and the phase angl es
are referenced to each i ndi vi dual probe l ocati on. By i nspecti on of thi s tabul ar
summary two i mmedi ate concl usi ons may be drawn. Fi rst, the moti on at each
Mode Shape Measurement 139
el evati on i s forward, and essenti al l y ci rcul ar. Second, i t i s apparent that the bot-
tom four el evati ons are general l y i n phase, and the top el evati on i s out of phase
wi th respect to the other four measurement pl anes. Based on these observati ons,
i t i s reasonabl e to compute an average radi al moti on at each el evati on (basi cal l y
ci rcul ar response). Al so recogni ze that a si gn change (i ndi cati ve of a phase
change) must occur between the top two pl anes (El evati ons 1 and 2).
The vector array from Tabl e 3-5 may be mani pul ated i n vari ous ways to
determi ne an average val ue for the general l y ci rcul ar orbi ts. One way woul d be
to rotate the i ni ti al Y-Axi s probe angl es by 90 to be i n the same angul ar refer-
ence posi ti on as the X-Axi s probes. Performi ng thi s si mpl e addi ti on of 90 to each
of the Y-Axi s probe angl es, Tabl e 3-6 i s easi l y generated. The I n-Phase magni -
tudes of each runout compensated shaft di spl acement vector may now be deter-
mi ned by mul ti pl yi ng each ampl i tude by the cosi ne of the associ ated angl e (i .e.,
I n-Phase=A cos ). Performi ng thi s mani pul ati on on the vectors i n Tabl e 3-6, and
cal cul ati ng a si mpl e ari thmeti c average, Tabl e 3-7 may be produced.
The average I n-Phase ampl i tudes from Tabl e 3-7 may now be combi ned i nto
a rotor mode shape. I f these respecti ve pl us or mi nus magni tudes are pl otted at
each el evati on, and i f thi s i s performed on an overl ay of the rotor drawi ng, the
mode shape shown i n Fi g. 3-45 evol ves. Further embel l i shment to the graphi cs
was performed by sweepi ng thi s shape over a compl ete.
Table 35 Vertical Generator - Summary Of Initial Runout Compensated X-Y Shaft Vibration
Vectors At Maximum Load Rejection Speed Of 410 RPM
Location Y-Probe X-Probe
El evati on 1 10.25 Mi l s,
p-p
@ 138 7.54 Mi l s,
p-p
@ 215
El evati on 2 4.48 Mi l s,
p-p
@ 292 5.55 Mi l s,
p-p
@ 337
El evati on 3 10.81 Mi l s,
p-p
@ 275 8.37 Mi l s,
p-p
@ 12
El evati on 4 21.73 Mi l s,
p-p
@ 271 27.41 Mi l s,
p-p
@ 357
El evati on 5 10.10 Mi l s,
p-p
@ 255 11.09 Mi l s,
p-p
@ 350
Table 36 Vertical Generator - Summary Of Initial Runout Compensated X-Y Shaft Vibration
Vectors With Common Angular Reference At Load Rejection Speed Of 410 RPM
Location Y-Probe X-Probe
El evati on 1 10.25 Mi l s,
p-p
@ 228 7.54 Mi l s,
p-p
@ 215
El evati on 2 4.48 Mi l s,
p-p
@ 22 5.55 Mi l s,
p-p
@ 337
El evati on 3 10.81 Mi l s,
p-p
@ 5 8.37 Mi l s,
p-p
@ 12
El evati on 4 21.73 Mi l s,
p-p
@ 1 27.41 Mi l s,
p-p
@ 357
El evati on 5 10.10 Mi l s,
p-p
@ 345 11.09 Mi l s,
p-p
@ 350
140 Chapter-3
From Tabl e 3-7 and Fi g. 3-45, i t i s noted that the bottom four I n-Phase com-
ponents are posi ti ve, and the top el evati on 1 carri es a negati ve si gn. Thi s change
i n si gn si gni es a zero axi s crossi ng between the upper gui de beari ng, and the
l ower gui de beari ng. The resul tant nodal poi nt i s vi si bl e di rectl y above the mi d-
dl e of the generator i n Fi g. 3-45. Thi s same general mode shape was ori gi nal l y
evi dent duri ng l oad rejecti on, and at the normal operati ng speed of 277 RPM.
Table 37 Vertical Generator - Summary Of In-Phase Components At 410 RPM
Location Y-Probe In-Phase X-Probe In-Phase Average In-Phase
El evati on 1 -6.86 Mi l s,
p-p
-6.18 Mi l s,
p-p
-6.52 Mi l s,
p-p
El evati on 2 +4.15 Mi l s,
p-p
+5.11 Mi l s,
p-p
+4.63 Mi l s,
p-p
El evati on 3 +10.77 Mi l s,
p-p
+8.19 Mi l s,
p-p
+9.48 Mi l s,
p-p
El evati on 4 +21.73 Mi l s,
p-p
+27.37 Mi l s,
p-p
+24.55 Mi l s,
p-p
El evati on 5 +9.76 Mi l s,
p-p
+10.92 Mi l s,
p-p
+10.34 Mi l s,
p-p
Fig. 345 Measured Mode Shape For Ver-
tical Generator At The Maximum Load
Rejection Speed Of 410 RPM
Fig. 346 Measured Mode Shape For Ver-
tical Generator At The Normal Machine
Operating Speed Of 277 RPM
Lower
Guide
Bearing
Upper
Guide
Bearing
Turbine
Guide
Bearing
Average In-Phase (A*Cos )
Motion At Each Elevation
139"
30"
67"
92"
(-6.52 Mils,
p-p
)
+4.63 Mils,
p-p
+9.48 Mils,
p-p
+24.55 Mils,
p-p
+10.34 Mils,
p-p
Lower
Guide
Bearing
Upper
Guide
Bearing
Turbine
Guide
Bearing
Average In-Phase (A*Cos )
Motion At Each Elevation
139"
30"
67"
92"
(-2.25 Mils,
p-p
)
+0.67 Mils,
p-p
(-0.76 Mils,
p-p
)
(-2.76 Mils,
p-p
)
(-1.48 Mils,
p-p
)
Mode Shape Measurement 141
After some del i berati on, i t was nal l y concl uded that a constant mode shape
under al l operati ng condi ti ons was unusual . Thi s was combi ned wi th abnormal
temperature characteri sti cs of the beari ng pads at each gui de beari ng. I t was
wel l documented that the upper generator gui de beari ng and the turbi ne gui de
beari ng were functi oni ng at temperatures that were nomi nal l y 20F hi gher than
the l ower generator gui de beari ng. Thi s i nconsi stency appeared for many years,
and no expl anati on was avai l abl e for thi s behavi or.
Eventual l y, the machi nery trai n was rebui l t by mechani cs that were
empl oyees of the uti l i ty company i nstead of contractors hi red by the OEM. Dur-
i ng reassembl y of thi s uni t, the uti l i ty company mechani cs di scovered that the
OEM procedure for setti ng the l ower generator gui de beari ng cl earances pro-
duced excessi ve radi al cl earances. Thi s assembl y procedure was modi ed by the
uti l i ty company mechani cs, and the nal resul ts were veri ed wi th swi ng checks
of the verti cal rotor. The ensui ng test run was very successful , and beari ng tem-
peratures were now consi stent at al l three gui de beari ngs. I t was concl uded that
the l ower generator gui de beari ng had been essenti al l y i neffecti ve i n restrai ni ng
the mi ddl e of the generator rotor for many years. Thi s change i n restrai nt at the
generator l ower gui de beari ng was cl earl y reected i n the shaft mode shapes.
Speci cal l y, the runout compensated I n-Phase components at the normal operat-
i ng speed of 277 RPM are summari zed i n Tabl e 3-8.
The average X and Y-Axi s I n-Phase components are used to pl ot the shaft
mode shape shown i n Fi g. 3-46. The i nuence of an acti ve generator l ower gui de
beari ng i s qui te apparent, and a new shaft nodal poi nt was cl earl y i ntroduced
bel ow thi s mi dspan beari ng. I t shoul d al so be menti oned that the synchronous
1X vectors were reduced, and the vi brati on ampl i tude scal i ng on Fi g. 3-46 at 277
RPM i s ve ti mes l arger than the adjacent Fi g. 3-45 at 410 RPM. Thi s change i n
scal i ng was consi dered to be appropri ate to al l ow ful l vi si bi l i ty of the modi ed
mode shape at 277 RPM.
I t was concl uded that the excessi ve beari ng cl earance associ ated wi th the
i ncorrect OEM assembl y procedure contri buted to many of the mechani cal prob-
l ems on thi s machi nery trai n. The val i di ty of the end user modi ed beari ng cl ear-
ance adjustment was demonstrated by a new consi stency i n gui de beari ng
temperatures, a reducti on i n synchronous vi brati on ampl i tudes, and l ogi cal
Table 38 Vertical Generator - Summary Of In-Phase Components At 277 RPM
Location Y-Probe In-Phase X-Probe In-Phase Average In-Phase
El evati on 1 -2.14 Mi l s,
p-p
-2.35 Mi l s,
p-p
-2.25 Mi l s,
p-p
El evati on 2 +0.62 Mi l s,
p-p
+0.72 Mi l s,
p-p
+0.67 Mi l s,
p-p
El evati on 3 -0.78 Mi l s,
p-p
-0.73 Mi l s,
p-p
-0.76 Mi l s,
p-p
El evati on 4 -2.68 Mi l s,
p-p
-2.85 Mi l s,
p-p
-2.76 Mi l s,
p-p
El evati on 5 -1.39 Mi l s,
p-p
-1.56 Mi l s,
p-p
-1.48 Mi l s,
p-p
142 Chapter-3
changes i n shaft mode shapes. I n addi ti on, the overal l vi brati on characteri sti cs
were l ess suscepti bl e to changes i n el ectri cal l oad.
Al though the presented hydro turbi ne generator mode shape data was di s-
cussed at onl y two speeds, the enti re 1X response duri ng startup or coastdown
shoul d be checked. Thi s data shoul d be exami ned for the presence of rotor and
structural resonances. The number of cri ti cal speeds, combi ned wi th the general
rotor congurati on, and the generi c beari ng type (i .e., soft versus hard) wi l l go a
l ong way towards deni ti on of the expected mode shapes. I n addi ti on, the exami -
nati on of runout compensated pol ar pl ots across a machi ne shoul d conrm the
i nphase behavi or of a transl ati onal resonance, and the out-of-phase moti on of a
pi votal mode. Anal ysi s of thi s type of data wi l l be di scussed i n greater detai l i n
subsequent chapters of thi s text.
Overal l , i t i s reasonabl e to concl ude that there are techni ques avai l abl e for
measurement and presentati on of the dynami c shaft mode shapes. Some compa-
ni es appl y the techni ques descri bed i n thi s secti on to produce on-l i ne three-
di mensi onal mode shapes that are conti nuousl y di spl ayed and moni tored by both
operati ons and mai ntenance personnel . Thi s i s parti cul arl y true i n the hydro
power generati on i ndustry where mul ti pl e access poi nts to the rotati ng shaft are
readi l y avai l abl e. On other machi nes, i t i s i mpossi bl e to mount probes al l al ong
the l ength of the shaft. However, i n some cases i t i s possi bl e to attach proxi mi ty
probes i nboard and outboard of the beari ngs, and outboard of the shaft seal s.
Further i nformati on may be obtai ned from probes i n the coupl i ng area. Thi s i s
parti cul arl y true for machi nes wi th sol i d (i nexi bl e) coupl i ngs. Based on thi s
type of addi ti onal shaft moti on data, i t i s often possi bl e to obtai n a real i sti c esti -
mate of the deformed shaft mode shape. Agai n vari abl e speed data must be con-
si dered, and the di agnosti ci an must have an awareness of the expected type of
mode shape for the machi ne under exami nati on.
ANALYTICAL RESULTS
There are si tuati ons when di rect mode shape measurements are not practi -
cal due to a l ack of probe l ocati ons, or duri ng the i ni ti al desi gn stages of a
machi ne. I n these cases, mathemati cal model i ng of the rotor system i s not onl y
an extremel y i nformati ve tool , i t may be the onl y avai l abl e opti on. These cal cul a-
ti ons provi de the abi l i ty to previ ew machi ne response before constructi on of the
machi ne, and i t al l ows exami nati on of many parameters associ ated wi th the sys-
tem behavi or. I n addi ti on, i f probl ems do devel op duri ng operati on, the computer
model s can provi de consi derabl e i nsi ght i nto the anti ci pated operati ng mode
shape of the system. I n addi ti on, mode shape changes due to vari ous exci tati ons,
or mechani cal abnormal i ti es can be model ed and exami ned i n the computer.
There are three fundamental types of shaft mode shape cal cul ati ons. The
si mpl est form consi sts of undamped cri ti cal speed cal cul ati ons, and the associ -
ated mode shapes. Thi s type of anal ysi s i s based upon mass and sti ffness proper-
ti es of the system. The undamped computati ons yi el d cal cul ated natural
frequenci es, and di mensi onl ess mode shapes. The second type of mode shape cal -
Analytical Results 143
cul ati on evol ves from a damped stabi l i ty anal ysi s of the mechani cal system. The
i ncl usi on of dampi ng al l ows the computati on of stabi l i ty, as wel l as the enti re
array of forward and reverse modes. Thi s damped stabi l i ty anal ysi s al so pro-
duces di mensi onl ess shaft mode shapes.
The thi rd major type of computed mode shape evol ves from a forced
response anal ysi s of the system. Whereas the previ ous two types of cal cul ati ons
do not i ncl ude synchronous forci ng functi ons (e.g., unbal ance), the forced
response does i ncorporate i nput forces, and i t produces di mensi onal mode
shapes. Hence, rotor di spl acement ampl i tudes at any l ocati on and any speed
may be cal cul ated. Chapter 5 on anal yti cal model i ng wi l l address these three
types of mode shape cal cul ati ons i n greater detai l . However, for the purposes of
thi s current chapter on rotor mode shapes, the di scussi on wi l l remai n at the fun-
damental l evel of undamped modes.
Case hi story 3 at the begi nni ng of thi s chapter on the two i mpel l er rotor
forms the basi s of performi ng undamped cri ti cal speed cal cul ati ons. I n al l cases,
the rotor model consi sts of a seri es of ci rcul ar cyl i nders wi th parti cul ar i nner and
outer di ameters, l ength, densi ty, and modul us of el asti ci ty. Lumped wei ghts and
i nerti as are i ncl uded for el ements such as i mpel l ers or coupl i ngs. The support
sti ffness characteri sti cs are al so i ncl uded as part of the i nput. Thi s type of ana-
l yti cal program for undamped cri ti cal speeds performs an i terati ve sol uti on for
each of the natural frequenci es. I t al so provi des a si mpl e mode shape for each
cri ti cal speed. The speci c resul ts that can be obtai ned wi th thi s type of anal yti -
cal computati on are best i l l ustrated wi th the fol l owi ng exampl e of a exi bl e com-
pressor rotor.
Case History 5: Eight Stage Compressor Mode Shape Change
For thi s case study, consi der the compressor rotor descri bed i n Fi g. 3-47.
Thi s i s an ei ght stage centri fugal compressor operati ng at 10,500 RPM, and con-
sumi ng 7,000 horsepower. The outboard i s at the thrust end of the compressor,
and a l arge bal ance pi ston i s mounted at the outboard di scharge end of the
machi ne. Thi s parti cul ar machi ne was operated i n surge for an extended peri od
of ti me. When the uni t was returned to normal operati on, i t was di scovered that
probe gap vol tages had i ncreased, and radi al vi brati on had i ncreased to l evel s
between 3.5 and 4.0 Mi l s,
p-p
. Attempts to sl ow the machi ne down were met wi th
even hi gher vi brati on ampl i tudes, and operati ons personnel were understand-
abl y perpl exed. I t i s common practi ce wi thi n many pl ants to sl ow down a
Fig. 347 Eight Stage
Centrifugal Compressor
Rotor Conguration
144 Chapter-3
machi ne when i t starts to vi brate excessi vel y. I n some compani es, thi s i s a nor-
mal i nstructi on gi ven to al l compressor and turbi ne operators. Hence, when thi s
ti me honored approach of sl owi ng down the machi ne to get away from a poi nt of
hi gh vi brati on resul ts i n even greater vi brati on ampl i tudes, the operati ons per-
sonnel nd that thei r opti ons are severel y l i mi ted.
Exami nati on of the ful l speed vi brati on data reveal ed that the machi ne was
apparentl y operati ng i n a pi votal mode, that became more severe as the speed
was decreased. Thi s was consi dered to be unusual si nce the compressor normal l y
operated above the rst, but bel ow the second cri ti cal speed. I n addi ti on, the
probe gap vol tages i ndi cated a measurabl e babbi tt l oss at both journal beari ngs
(i .e., i ncreased cl earances). Fortunatel y, an undamped cri ti cal speed anal ysi s has
been performed, and thi s i nformati on was avai l abl e for exami nati on i n conjunc-
ti on wi th the vi brati on data. Fi g. 3-48 di spl ays the cal cul ated mode shapes at the
transl ati onal rst cri ti cal wi th three di fferent beari ng sti ffness val ues. Note that
at a sti ffness of 200,000 Pounds/I nch (top pl ot), a typi cal cyl i ndri cal mode i s pre-
di cted wi th a cri ti cal speed of 3,350 RPM. On the mi ddl e pl ot, the sti ffness was
i ncreased to 500,000 Pounds/I nch, and thi s change has l i ttl e modal effect, but i t
Fig. 348 Mode Shapes For Compressor
Rotor Translational First Critical Speed
Fig. 349 Mode Shapes For Compressor
Rotor Pivotal Second Critical Speed
Analytical Results 145
does i ncrease the rst cri ti cal speed to 4,270 RPM. A further jump i n beari ng
sti ffness to 1,000,000 Pounds/I nch reduces moti on (vi brati on) at the beari ngs,
and rai ses the cri ti cal speed to 4,770 RPM. Al so note the si mi l ari ty between
these pl ots and the previ ousl y di scussed Fi g. 3-27.
The next set of mode shape di agrams i n Fi g. 3-49 descri bes the pi votal bal -
ance resonance, or second cri ti cal speed. Agai n, the top pl ot shows the 200,000
Pounds/I nch sti ffness case, and an expected coni cal mode at a cal cul ated cri ti cal
speed of 7,160 RPM. The mi ddl e pl ot reveal s the change i n mode shape as the
sti ffness i s i ncreased to 500,000 Pounds/I nch. At thi s sti ffness the cri ti cal has
been i ncreased to 10,730 RPM, and the compressor coul d be operati ng on the
front sl ope of thi s second mode. A nal step of i ncreasi ng sti ffness to 1,000,000
Pounds/I nch has an addi ti onal cl ampi ng i nuence upon the mode shape, and the
cri ti cal has been rai sed to 13,810 RPM.
Due to the abnormal operati on of thi s compressor i n surge, i t i s probabl e
that some mechani cal damage was i ni cted on the beari ngs. I t was postul ated
that the potenti al beari ng damage resul ted i n a reducti on i n beari ng sti ffness,
and thi s al l owed the second cri ti cal speed to drop bel ow operati ng speed. I n
essence, the uni t was presumed to be runni ng above the second cri ti cal speed.
For the sake of compl eteness, the thi rd cri ti cal was al so exami ned, and the
resul tant mode shapes are presented i n Fi g. 3-50 as beari ng sti ffness i s
i ncreased through the three previ ous steps. Note that changes i n beari ng sti ff-
ness has mi ni mal effect on the deected shaft mode shapes. As expected, the fre-
Fig. 350 Mode Shapes
For Compressor Rotor
Bending Third Critical
Speed
146 Chapter-3
quency conti nues to i ncrease wi th a thi rd cri ti cal speed at 14,970 RPM at
200,000 Pounds/I nch, and 22,960 RPM for the sti ffest case of 1,000,000 Pounds/
I nch. I t i s al so worthwhi l e to note that the thi rd cri ti cal i s a sensi ti ve area from a
rotor fati gue standpoi nt. Thi s i s due to the reverse exure acti ng wi thi n the
rotor. Thi s mode i s sensi ti ve to outboard masses. For i nstance, the i nstal l ed cou-
pl i ng must be extremel y wel l bal anced to al l ow the machi ne any success of sur-
vi val i n thi s mode. I n actual i ty, most compressors of thi s general congurati on
woul d be desi gned to avoi d thi s mode. Wi th respect to the current mechani cal
probl em on thi s rotor, there i s no evi dence to suggest any i nuence from thi s
thi rd cri ti cal speed upon the measured hi gh shaft vi brati on.
The i nformati on from a seri es of undamped cri ti cal speed cal cul ati ons i s
often summari zed i nto a fami l y of curves that i s typi cal l y cal l ed a cri ti cal speed
map. More speci cal l y, thi s type of data array i s dened as an undamped cri ti cal
speed map. I n thi s type of presentati on the combi ned effects of beari ng sti ffness,
cri ti cal speed, and shaft mode are di spl ayed on a l og-l og scal e. The format of thi s
di agram faci l i tates the exami nati on of a l arge amount of i nformati on i n a rea-
sonabl y conci se manner.
For i nstance, i f the cal cul ated cri ti cal speeds from the ei ght stage compres-
sor rotor are pl otted agai nst beari ng sti ffness, the undamped cri ti cal speed map
di spl ayed i n Fi g. 3-51 may be constructed. I n order to provi de i mproved deni -
ti on of the curve shapes, two addi ti onal data poi nts at 100,000 and 2,000,000
Pounds/I nch have been added. Hence, each curve consi sts of ve cal cul ated cri ti -
cal speeds correspondi ng to ve di fferent sti ffness val ues. I t i s si gni cant to note
that changes i n sti ffness al one can determi ne the mode shape of the system. The
l ower order modes are general l y control l ed by beari ng sti ffness, and the hi gher
order modes are typi cal l y governed by shaft sti ffness. The control l i ng parameter
i s normal l y determi ned by the di stri buti on of strai n energy wi thi n the
undamped cri ti cal speed cal cul ati ons.
Fig. 351 Undamped Critical Speed Map For Eight Stage Compressor Rotor
1,000
10,000
30,000
100,000 1,000,000 2,000,000
F
r
e
q
u
e
n
c
y

(
C
y
c
l
e
s
/
M
i
n
u
t
e
)
Stiffness (Pounds/Inch)
Translational - 1st Mode
Pivotal
2nd Mode
Rotor Bending - 3rd Mode
Good
Bearings
Damaged
Bearings
Analytical Results 147
However, the sl ope of the vari ous cri ti cal speed curves may al so provi de an
i ndi cati on of the control l i ng physi cal parameters. For i nstance, i f the cri ti cal
speed changes si gni cantl y wi th sti ffness vari ati ons, then the rotor response i s
probabl y dependent upon the beari ng sti ffness. Under thi s condi ti on, changes i n
the journal beari ngs woul d change the resul tant shaft cri ti cal . However, i f the
cri ti cal speed curve i s fai rl y at, then shaft sti ffness i s probabl y the control l i ng
parameter. I n that case, the rotor woul d have to be physi cal l y modi ed to al ter
the shaft cri ti cal speed and/or associ ated mode shape.
Returni ng to the case hi story under di scussi on, i t was concl uded from the
cri ti cal speed map (Fi g. 3-51), that operati on at the normal speed of 10,500 RPM
coul d resul t i n the compressor experi enci ng a change i n operati ng modes due to a
reducti on i n beari ng or support sti ffness. Hence, the machi ne that ran above the
rst cri ti cal speed may be al tered i nto operati ng above the second mode. Any
speed reducti on woul d then rai se shaft vi brati on ampl i tudes due to the rotor
enteri ng the pi votal bal ance resonance. Thi s predi cted behavi or was certai nl y
consi stent wi th the previ ous experi ences of the operati ng personnel . Thi s concl u-
si on i s al so consi stent wi th the strong potenti al of beari ng damage mani festi ng
as a reducti on i n support sti ffness and effecti ve beari ng dampi ng.
I n the nal scenari o, the uni t was tri pped, and i t di d pass through both a
pi votal and a transl ati onal resonance duri ng coastdown. Both journal beari ngs
were found to be damaged wi th expanded cl earances, and the shaft seal s
reected the beari ng damage. Fortunatel y, the rotor journal s were not scored,
and al l beari ngs and seal s were repl aced. The post overhaul compressor startup
was normal , and the ful l speed vi bratory characteri sti cs returned to previ ousl y
documented val ues. Hence, the damaged beari ngs (due to extended operati on of
compressor i n surge) proved to be responsi bl e for the i ncreased vi brati on ampl i -
tudes, and the change i n operati ng shaft mode shapes.
I n summary, thi s mode shape anal ysi s, and the associ ated cri ti cal speed
map i s i nformati ve duri ng the desi gn stages of a machi ne. Thi s i nformati on i s
al so qui te useful duri ng troubl eshooti ng of a el d probl em. I n most si tuati ons,
the el d probl em sol vi ng exerci se does not al l ow sufci ent ti me to perform thi s
type of anal ysi s. Hence, i t i s al ways desi rabl e to perform thi s type of detai l ed
cri ti cal speed and mode shape anal ysi s before the machi ne starts to shake. I t i s
al so desi rabl e to appl y some of the more advanced techni ques di scussed i n the
fol l owi ng chapters to exami ne the i nuence of beari ng characteri sti cs, torsi onal
resonances, damped anal ysi s, pl us di mensi onal si mul ati on of synchronous
response characteri sti cs.
148 Chapter-3
BIBLIOGRAPHY
1. Aval l one, Eugene A. and Theodore Baumei ster I I I , Marks Standard Handbook for
Mechanical Engineers, Tenth Edi ti on, pp. 3-9, 3-16, New York: McGraw-Hi l l , 1996.
2. Bel di ng, Wi l l i am G. and others, ASM Handbook of Engineering Mathematics, 4th
pri nti ng, p. 319, Metal s Park, Ohi o: Ameri can Soci ety for Metal s, 1989.
3. Cal i strat, Mi chael M., Flexible Couplings, their design selection and use, p. 464,
Houston: Carol i ne Publ i shi ng, 1994.
4. Chi l ds, Dara, Turbomachinery Rotordynamics Phenomena, Modeling, and Analy-
sis, New York: Wi l ey-I ntersci ence Publ i cati on, 1993.
5. Den Hartog, J.P., Mechanical Vibrations, New York: McGraw-Hi l l Book Company,
1934.
6. Gi eck, Kurt and Rei ner Gi eck, Engineering Formulas, 6th edi ti on, p. M3, New York:
McGraw-Hi l l I nc., 1990.
7. Harri s, Cyri l M., Shock and Vibration Handbook, fourth edi ti on, pp. 1-11, 1-12, 38-5,
and chap. 32, 33, 36, and 37, New York: McGraw-Hi l l , 1996.
8. Jeffcott, H.H., Lateral Vi brati on of Loaded Shafts i n the Nei ghborhood of a Whi rl -
i ng Speed - The Effect of Want of Bal ance, Philosophical Magazine, Vol . 37 (1919),
pp. 304-314.
9. Ranki n, W.A., On the Centri fugal Force of Rotati ng Shafts, Engineer (London): 27,
1869.
10. Shi gl ey, Joseph E. and Charl es R. Mi schke, Standard Handbook of Machine Design,
pp. 9.13, and 10.6, New York: McGraw-Hi l l Book Company, 1986.
11. Spotts, M.F., Design of Machine Elements, 6th Edi ti on, pp. 18, 27, and 150, Engl e-
wood Cl i ffs, New Jersey: Prenti ce-Hal l , I nc., 1985.
12. Thomson, Wi l l i am Tyrel l , Mechanical Vibrations, Engl ewood Cl i ffs, New Jersey:
Prenti ce Hal l , I nc., 1948.
13. Vance, John M., Rotordynamics of Turbomachinery, p 61, New York: Wi l ey-I nter-
sci ence Publ i cati on, 1988.
14. Weaver, F.L., Rotor Desi gn and Vi brati on Response, Proceedings of the First Turbo-
machinery Symposium, Gas Turbi ne Laboratori es, Texas A&M Uni versi ty, Col l ege
Stati on, Texas (1972), pp. 142-147.
15. Young, Warren C., Roarks Formulas for Stress & Strain, Si xth Edi ti on, pp. 59, 101,
and 346, New York: McGraw-Hi l l Book Co., 1989.

Das könnte Ihnen auch gefallen