AD-78, 319

HIGH-SPEED ROTOR DYNAMICS - AN ASSESSMENT

OF CURRENT TECHNOLOGY FOR SMALL TURBO-
SHAFT ENGINES
John M. Vance
Army Air Mobility Research and Development
Laboratory
Fort Eustis, Virginia
July 1974

DISTRIBUTED BY:

Krai
National Technical Information Service
U. S. DEPARTMENT OF COMMERCE
5285 Port Royal Road, Springfield Va. 22151
 Unclassified
SECURITY CLASSIFICATION OF THIS PAGE (Whm Dmtm Enltnd) gJZZSTB/Z
READ INSTRUCTIONS
REPORT DOCUMENTATION PAGE BEFORE COMPLETING FORM
1. REPORT NUMBER 2. SOVT ACCESSION NO 3. RECIPIENT'S CATALOS NUMBER

USAAMRDLTR-74-66
4. TITLE C'ntf Subl/tlaJ 5. TYPE OF REPORT * PERIOD COVERED

HIGH-SPEED ROTOR DYNAMICS - AN ASSESSMENT OF

CURRENT TECHNOLOGY FOR SMALL TURBOSHAFT
6. PERFORMING ORC. REPORT NUMBER
ENGINES
7. AUTMORf«.) 8 ~^NTRACT OR GRANT NUMBERr«)

John M. Vance DAHC04-72-A-0001

9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT, PROJECT, TASK
AREA t WORK UNIT NUMBERS
Eustis Directorate
U.S. Army Air Mobility R&D Laboratory Project 1G162207AA72
Fort Eustis, Virginia 23604
It. CONTROLLING OFFICE NAME AND ADDRESS ia, REPORT DATE

JtilY, 1974
ii" «UMBERTOF PAGES

I«. MONITORING AGENCY NAME » AODRESSf» dlfforwil Itam Conlroltlnt OHtct) IS. SECURITY CLASS, (ol IM« nporl)

it*. OECLASSIFIC VTION/DOWNCRADING

SCHEDULE

1». DISTRIBUTION STATEMENT fof (hi. R.por0

Approved for public release; distribution unlimited.

<7. DISTRIBUTION STATEMENT (ol IA« »btlfcl imlarad In Bloc» 10, II dlllmrtnl horn Rtport)

IB. SUPPLEMENTARY NOTES

\^ !
ii.iNAj ■ ! I;.;I

19. KEY WORDS (Contlnuu on nvmti» «Id* if n«c««*ary «nd Identlly by block nutnbtr)

Aerodynamics High velocity

Rotor
Dynamics Shafts (machine elements)
Excitation Turbojet engines
20. ABSTRACT (Conllnum on r*v»ra« aide II nocattmry and Identity by block numbm)

Problem areas connected with rotor dynamics which are peculiar to the special requirements of
rotor-bearing systems in the class of small turboshaft engines being developed for U. S. Army
helicopters are identified.

The present and past philosophy of design is discussed, including the impact of the demand foi
front drives.

^
DO 1 JAN 71 1473 EDITION OF I NOV «5 IS OBSOLETE • Unclassified
SECURITY CLASSIFICATION OF THIS PAGE (Wbtn Jmlm Enltitd)
 Unclassified
SECUWTY CLASSIFICATION OF THIS PAOI(»hw Dim Snlmnd)

Block 20. Abstract - continued.

Methods for critical speed prediction and high-speed balancing are reviewed. The trend to
higher speeds is seen to require consideration of new approaches to balancing through flexural
modes.

The major parameters available for control by the designer are shown to be the bearing
support properties, and recommendations are made for improving the accuracy of prediction
of these properties.

Nonsynchronous excitation is categorized according to the mechanisms producing the forces,

and a need is shown for better methods to identify the resulting whirling and vibration,
since several of these motions are potentially unstable.

Finally, reasons are given for the predominant use of rolling-element bearings in these engines,
and the potential for special applications of oil-film and gas bearings is discussed.

JiL
Unclassified
SCCURITV CLASSIFICATION OF THIS PASef*»«" DM« Bnltnd)

mam •'- ..^.-■■- .■„..-....■■.■■.^;... ...^

 DISCLAIMERS

The findings In this report are not to be construed as an official

Department of the Army position unles, so designated by other authorized
documents.

When Government drawings, specifications, or other data are used for any
purpose other than in connection with a definitely related Government
procurement operation, the United States Government thereby Incurs no
responsibility nor any obligation whatsoever; and the fact that the
Government may have formulated, furnished, or In any way supplied the
said drawings, specifications, or other data Is not to be regarded by
implication or otherwise as In any manner licensing the holder or any
other person or corporation, or conveying any rights or permission, to
manufacture, use, or sell any patented invention that may in any way be
related thereto.

Trade names cited in this report do not constitute an official endorse-

ment or approval of the use of such commercial hardware or software.

DISPOSITION INSTRUCTIONS

Destroy this report when no longer needed. Do not return it to the

originator.

«Ks:ir (ft
NTIt «rtitl SKtlM
DCC Iii(, 5-fi'lM Q

JIISTIriCAllOd

B»
liiSTRiainiOH/mawiun COOES
~ But. *i«IL ani/aTSKCiir

■rrr-f-y

it
 PREFACE

This report is the result of extensive literature review, experience of the author, and on-site
visits to more than a dozen engine contractors, research houses, and universities that are
active in the technical field of rotor dynamics. An attempt has been made to restrict the
field of coverage to those aspects of rotor dynamics that are directly relevant or applicable
to the class of small turboshaft engines under development for U.S. Army aircraft propulsion.

Many of the ideas expressed in the report were not original with the author, but were
offered to him in the course of the visits mentioned above. Due credit is not specifically
given for all of these in the report, due to proprietary constraints of the contributors, but a
list of the contributors is given under Acknowledgements.

As a technical report, this one is somewhat unique in its dearth of mathematical equations.
For these, the reader should consult the references noted throughout the report.

ACKNOWLEDGEMENTS

The generous cooperation of engine contractors, research houses (Government and private),
and universities in this investigation is greatly appreciated by the author and by the Eustis
Directorate, U.S. Army Air Mobility Research and Development Laboratory.

The contributing organizations are as follows:

1. AiResearch Manufacturing Company, Phoenix, Arizona

2. Avco/Lycoming Division, Stratford, Connecticut

3. Air Force Aero-Propulsion Laboratory, Wright-Patterson AFB, Ohio

4. U.S. Army Aeronautical Depot Maintenance Center (ARADMAC), Corpus Christi,

Texas

5. Bentley-Nevada Corporation, Minden, Nevada

6. General Motors Corporation, Detroit Diesel Allison Division, Indianapolis,

Indiana

7. Franklin Institute, Philadelphia, Pennsylvania

8. General Electric Company, Lynn, Massachusetts, and Evendale, Ohio

9. Mechanical Technology Incorporated, Latham, New York,

10. NASA-Lewis Research Center, Cleveland, Ohio

11. Pratt and Whitney Aircraft, West Palm Beach, Florida

12. Solar Division, International Harvester Company, San Diego, California

13. Teledyne CAE, Toledo, Ohio

14. University of Virginia, Department of Mechanical Engineering, Charlottesville,

Virginia

15. Williams Research Corporation, Walled Lake, Michigan

 TABLE OF CONTENTS

Page

PREFACE 1

ACKNOWLEDGEMENTS 1

LIST OF ILLUSTRATIONS 5

INTRODUCTION 6

SCALING FACTORS G

EFFECT OF FRONT-DRIVE REQUIREMENTS 9

Supercritical Power Turbine Shaft 10

Intershaft Bearings 10
Compressor Spool Design 12
Beryllium Power Turbine Shaft 12
Recommendations 12

PREDICTION OF CRITICAL SPEEDS 13

Design Philosophy 13
Published Analyses and Computer Programs 16
The "Rubber Disk Effect" 21
Effect of Shaft Joints 23
Recommendations 23

HIGHSPEED BALANCING 23

Multiplane Balancing 24
"In-Place" Balancing 25
Recommendations 26

PREDICTION AND CONTROL OF BEARING SUPPORT PROPERTIES 26

Squeeze Film Bearing Dampers 27

Mechanical Bearing Supports 30
Recommendations 31

NONSYNCHRONOUS EXCITATION 31

Backward Whirl 33
Friction-Induced Whirl 34
Aerodynamic-Induced Whirl 35
Overhung Rotors 35
Recommendations 36

3
 Page

HYDRODYNAMIC BEARINGS 37

Gas Bearings 37
Oil-Film Bearings 38
Recommendations 40

REFERENCES 41
\

LIST OF ILLUSTRATIONS

Figure Page

1 Scaled dimensions 7

2 Turbine engine cutaway 11

3 Cylindrical and conical whirl of a rigid rotor on flexible supports 15

4 Flexural whirl on rigid supports - 1st and 2d mode shapes 16

5 Jeffcott rotor 17

6 Response curve for Jeffcott rotor 17

7 Response curve for rigid rotor on flexible supports 18

8 Typical critical speed map 22

9 Disk deformation affecting gyroscopic moments 22

10 Squeeze film bearing damper 28

11 Depender ce of intershaft bearing load capacity on rotation direction of

both shafts 39
 INTRODUCTION

There are a number of special requirements connected with rotor dynamics that are peculiar
to rotor-bearing systems in the class of small turboshaft engines being developed for U.S.
Army aviation:

1. Increasingly higher shaft speeds. It is desired to increase airflow and power

output without increasing physical size.

2. Small physical size (especially in frontal area and wheel diameter) and light weight.

3. Front Drive. It is usually desired to locate the power takeoff shaft out through
the front of the compressor section.

4. Maintainability. Indi 'idual components making up the rotor-bearing assembly

should be easily replaceable.

5. Long Life. It is desired to increase the engine operating life from the presently
attained value of about 1500-2000 hours. Frequency of overhauls should be
reduced.

Taken on an individual basis, these requirements would generate significant problems for the
rotor-bearing design engineer. Taken simultaneously, these requirements are a severe
challenge. For example, the combination of requirements 1 and 2 has resulted in a blade
tip clearance problem in small engines. The very short blades require extremely small tip
to housing clearances to maintain high aerodynamic efficiency and high power output. This
in turn requires very small rotor shaft excursions to avoid blade-housing ir terferences, a
design condition that is incompatible with low dynamic bearing loads at high speeds.

From the standpoint of Army aviation, there are two broad objectives to be met in solving
rotor dynamics problems through a program of directed research. The first objective is to
reduce the magnitude and frequency of rotor dynamics-related failures and required redesign
efforts in Army propulsion development programs. The second objective is to improve the
reliability and maintainability of future Army aircraft engines in the field through reduction
of vibration and dynamic bearing loads.

SCALING FACTORS

An appreciation for the effect of physical size reduction on dynamics can be gained through
dimensional analysis of a rotor-bearing sytem. Consider a turbine or compressor wheel
assembly (disks, spacers, and blades) centrally located on a flexible shaft, as shown in Figure
1, and assume that it is desired to scale the system down in physical size without reducing
airflow rates.

It has been the practice of turboshaft design engineers, with one or two recent exceptions,
to maintain operating shaft speeds at least 20% below the first critical speed in shaft bending
(usually the third critical speed). If the bearings of this example are rigidly supported, the
bending critical speed is approximated by

V
■^ (1)
(M + | pdW

18d / E
V
P (0.5 + a? a2 )P

where E = Young's modulus

M = mass of wheel assembly
p = shaft mass per unit length
a, = d/D
a2 = //I
and d, D, /, L are defined by Figure 1.

d
w M
R
i
©■ B
T
L x/2
< ►4 ►4 ►

Figure 1. Scaled dimensions.

For a given material and geometric configuration, equation (1) can be rewritten as
^d
^cr =
^ ?T (2)
T
where C is a constant.

In addition to the variables defined above, other variables pertinent to a scaling analysis of
the rotor bearing system are
 W = shaft speed
R = maximum blade radius
v = average axial air velocity
Q = air volume flow rate

The most important dimensionless groups are

TT, = dj (d/D, f/L. etc.)

2
W W/
IT-, = ~ =r* ■
wcr Cd.

WR3 \AR
J
Q J

To preserve similitude for performance prediction, all of these TT groups must be held con-
stant when engine size is reduced. The first n group determines the "scale factor" n.

pj1- = pj2- requires (3)

D, d,
n
D2 d2 '

where the subscript 1 refers to the large engine and subscript 2 refers to the small engine.

The second n group preserves the same critical speed margin (say, 20%) in the small engine
as in the large engine.

W, iV W2 l22
requires
Cd! Cdi

W2 = nW, (4)

The third TT group requires the same velocity flow profiles in the small engine as in the
large engine. If, in addition, the same air volume flow rate is required (assuming equivalent
gas temperatures), the third TT group requires

W, R,3 W2R23
Q Q

W2 = n'W, (5)

Clearly, equations (4) and (5) are incompatible.

Engine aerodynamic performance requirements are usually given priority over dynamics
requirements. Thus, the speed of the small engine, as dictated by equation (5), will be much
higher than would be allowed by the critical speed margin, as dictated by equation (4). A
very small reduction in engine size can completely eliminate a substantial critical speed
margin.

8
For example, assume a 10% reduction in size of an engine with a 20% critical speed margin.
Equation (5) gives (n = 1.11).

W2 = 1.37 W,

Also,
Wcri = 1.25 W,

so that
W, = 0.8 X Wcri

Then
Wcr2 = 1.11 Wcri = 1.388 W,

and
Wj 1.37W,
= 0.99
Wcr2 1.39W,

The margin has been reduced to only 1%, simply by scaling the engine down 10% in size.

The analysis and example above are greatly simplified, and numerous other factors must be
considered in a realistic scaling analysis. Nevertheless, it illustrates one of the basic problems
of rotor dynamics in small engines. It is becoming ever more difficult to avoid supercritical
operation (shaft speeds through and above bending criticals) in modern small high-speed
turboshaft engines.

EFFECT OF FRONT-DRIVE REQUIREMENTS

Another set of design conditions that is difficult to meet in small high-speed engines is the
combination of the front-drive requirement and the maintainability requirement. Referring
again to Figure 1, the critical speed equation (1) is derivable from the more basic equation

(6)
Wer =yiv^

in which k is the shaft stiffness effective at the disk (wheel assembly), and m is the
s
effective mass at the disk. It is seen that maintaining a high critical speed depends on main
taining high shaft stiffness. The stiffness of a uniform shaft mounted on rigid supports,
relative to a centrally applied load, is

ks - «|1 17)

where I is the cross-sectional area moment of inertia; the other symbols were previously
defined. It is clear that the stiffness, and consequently the critical speed, is a very strong
inverse function of the distance ! between bearings.

In a front-drive turboshaft engine, either the distance between the bearings supporting the
power turbine shaft cannot be shorter than the length of the compressor spool, or an inter-
shaft bearing must be employed, since the power turbine shaft must pass through the inside

9
of the compressor spool to reach the front of the engine (see Figure 2). Intershaft bearings,
supporting relative rotation between the compressor spool and the power turbine shaft,
have a history of poor performance and maintainability.

SUPERCRITICAL POWER TURBINE SHAFT

As a result of the above considerations and constraints, the latest turboshaft engine under
development for Army helicopters (the T700 for UTTAS) will have its power turbine shaft
operating through and above a critical speed in shaft bending.

Safe operation of aircraft rotor-bearing systems at supercritical speeds can be obtained only
when at least one of two conditions is reliably met:

1. Accurate high-speed balancing of rotor shaft assemblies. Passage through bending

criticals may require multiplane balancing (more than two planes).

2. Provision for significant amounts of external damping.

Satisfaction of condition 1 allows passage through critical speeds without destructive whirl
amplitudes. Satisfaction of condition 2 also reduces whirl amplitudes, but additionally, it
suppresses several mechanisms of dynamic instability, such as nonsynchronous whirl induced
by internal friction, shaft stiffness asymmetry, or aerodynamic excitation. These phenomena
will be discussed in later sections.

The alternatives to a supercritical power turbine shaft in small high-speed front-drive engines
are

1. Intershaft bearings

2. Gas generator redesign to allow a large-diameter power turbine shaft

3. Use of a power turbine shaft material with a significantly higher stiffness/density

ratio than steel (e.g., beryllium)

INTERSHAFT BEARINGS

The most commonly used method for keeping power turbine shafts subcritical is the inter-
shaft bearing. To date, antifriction bearings have been used for this purpose, although a
design problem of radial space availability between the inner and outer shafts is often
encountered. These bearings have been a source of problems.

The most probable causes of problems with rolling element intershaft bearings are shaft
bowing from thermal effects, inacessibility for lubrication, and high DN values with
counterrotating shafts.

It is desirable from an aerodynamic design standpoint to have the gas generator shaft and
the power turbine shaft rotating in opposite directions, but this obviously increases the DN
value of the intershaft bearing. It is therefore a more common design practice to have
these shafts corotating (although usually at different speeds).

10
 i
ü

c
'5)
c
0)
0)
c

CM
a>
3

11
Intershaft bearings also appear to be a source of nonsynchronous vibration, since they can
transmit dynamic loads from one shaft to the other at any of the predominant frequencies.

In an effort to solve some of these problems the Air Force Aero-Propulsion Laboratory,
Lubrication Branch, has recently supported several programs (at Mechanical Technology
Incorporated and AiResearch Manufacturing Company) to develop gas foil bearings for
intershaft applications.

Another type of intershaft bearing that has potential merit is the oil-film bearing. At least
one engine manufacturer has attempted such an application without success. It should be
noted that the load capacity of a film bearing is increased when the bearing corotates with
the journal, and that film bearings have minimal radial space requirements. Of course,
oil-film bearings also have a much greater load capacity than gas bearings.

COMPRESSOR SPOOL DESIGN

The second alternative to supercritical shaft design involves either an increase of gas genera-
tor bearing diameter (with a consequent increase in bearing DN values) to accommodate a
larger-diameter power turbine shaft or a bias toward a purely centrifugal compressor design
(as in the Pratt and Whitney ST9) to allow a shorter shaft. Either of these options raises
the power turbine shaft critical speed, as desired, but they both tend to increase the cross-
sectional size of the engine.

It may be of interest to note that a large-diameter compressor spool bearing is compatible

with some of the unique requirements of gas-film bearings. For example, gas bearings
require large bearing areas and high journal velocities to generate significant load capacity.

BERYLLIUM POWER TURBINE SHAFT

The final alternative has been the subject of some preliminary studies at Williams Research
Corporation. Comparison of equations (1) and (2) shows that the shaft critical speed is
proportional to sj E/p . Most engineering metals in common use (e.g., steel, aluminum)
have almost identical E/p values, thus offering little selectivity with respect to critical
speed properties.

Engineers at Williams have pointed out that beryllium is an exception to this rule, with an
E/p value of about three times that of steel or aluminum. In designing a beryllium power
turbine shaft, problems to be overcome are the notch sensitivity and brittleness of the
material. If these problems could be overcome through proper design or through modifica-
tion of material properties, a power turbine shaft of this material could operate at signifi-
cantly higher speeds (perhaps 20%) without passing through resonance in bending.

RECOMMENDATIONS

The following subjects should be investigated in research and development programs in order
to optimize the front-dri/e design of small turboshaft engines:

12
 1. A thorough trade-off study should be made to compare the relative merits and
advantages of the several alternative means of achieving front drive in a small
high-speed turboshaft engine. A quantitative measure of relative merit should be
developed in this study and used to express the results. In contrast to a trade-off
study with performance and airflow characteristics as the sole criteria, this study
should consider rotor dynamics and its effects on bearing life, maintainability, and
reliability. The following design alternatives should be considered:

• Supercritical power turbine shaft

• Rolling element intershaft bearing

• Gas-film intershaft bearing

• Oil-film intershaft bearing

• Modified compressor spool design to accept larger-diameter power turbine

shaft

• Beryllium power turbine shaft

2. If the results of recommendation 1 warrant it, a study should be made of the

feasibility of using beryllium as a power turbine shaft material in order to raise
the shaft bending critical speed to keep engine operation subcritical.

PREDICTION OF CRITICAL SPEEDS

Cr tical speed analysis was historically and is currently the most important single rational
method for rotor-bearing-system design. It allows the designer to avoid resonant conditions
in the operating speed range of his machine.

DESIGN PHILOSOPHY

Equation (6) shows that the critical speeds of a rotor-bearing system are determined by the
effective stiffness, which may be either in the supports or in the shaft itself, or both, and by
the effective mass, which may be rotating with the shaft or vibrating with the bearing
support structure. Equation (6) applies strictly to a simple Jeffcott rotor1 '2 with a single
critical speed, but the concept of effective stiffness and mass may be carried out to much
more complex cases involving bending modes with distributed mass, in which these quantities
must be regarded as speed dependent.

1
H. H. Jeffcott, The Lateral Vibration of Loaded Shafts in the Neighborhood of a Whirling
Speed - The Effect of Want of Balance, Philosophical Magazine, Vol.37, No. 6, 1919, p. 304.

2
E. J. Gunter, Jr., Dynamic Stability of Rotor-Bearing Systems, NASA-SP 113, 1966.

13
I

In discussing »tiffness effects, the inverse of stiffness-flexibility—is a more convenient ter-

minology, since zero flexibility is more readily visualized and is more closely approached
in engineering structures than zero stiffness. If flexibility is used, it should be remembered
that the critical speeds are inversely proportional to the square root of both the effective
flexibility and the effective mass.

In rotor-bearing systems, the flexibility may be almost entirely in the bearing supports, in
which case we say we have a rigid rotor, or it may be almost entirely in the shaft, in which
case we say we have a flexible rotor. The corresponding critical speeds are called rigid-body
critical speeds and flexural or bending critical speeds. The reason for the terminology is
illustrated in Figures 3 and 4.

Figure 3 shows the two modes of rigid-body motion allowed by soft supports. The first
mode, in which the two ends of the rotor whirl in phase, is called cylindrical whirl. The
second mode, in which the two ends of the rotor whirl 180° out of phase, is called
conical whirl. Most modern turboshaft engines have bearing supports that are designed to
be very flexible relative to the shaft and therefore pass through both of these rigid-body
criticals at speeds below the operating range of the engine. Since most of the motion in
these modes takes place at the bearing-support locations, the resonance can be damped by
dissipating energy in specially designed dampers at the bearing supports.

If the supports were made rigid, the first mode would look like Figure 4a, and the second
mode would look like Figure 4b. Since there is no deflection of the bearing-support
structure, and since the shaft whirls in a constant bowed shape, it is difficult to damp these
modes by dissipating energy in dampers. These flexural modes, with rigid supports, are
therefore extremely difficult to pass through safely, unless the shaft is precisely balanced
for the expected mode shapes.

At sufficiently high speeds, the rotor-bearing system of Figure 3 also displays the flexural
response of Figure 4. Thus, a rigid rotor changes into a flexible rotor just by an increase in
speed. This is because all real shafts have uome flexibility and will therefore have flexural
resonances at sufficiently high frequencies, even with soft supports. (If the supports had no
stiffness at all, these would be called free-free modes.)

In the design of turboshaft (also turbojet, turbofan) engines to be used as a primary source
of flight propulsion for aircraft, engineering design practice has almost always been to keep
the flexural critical speeds above the maximum operating speed of the engine. The safety
reasons for this should be obvious from the discussion above. The word "almost" is used
because the definitions of rigid-body and flexural critical speeds become hazy when the
support flexibility and shaft flexibility are of about the same magnitude. Also, it can
become very difficult to design a stiff power turbine shaft when the bearing span cannot be
made shorter than the length of the compressor spool.

In fact, the latest Army helicopter engine under development (the T700 for UTTAS) has a
supercritical power turbine shaft. This will be the first aircraft engine to fly as a primary
propulsion source with a turbine rotor shaft expressly designed to operate through and
above a flexural critical speed.

14
 f^L^ »^

a. Cylindrical whirl

/—s-^ ^

/ SHtflA.

b. Conical whirl

Figure 3. Cylindrical and conical whirl of a rigid rotor on flexible supports.

15
 r7T7T7
a. 1st mode shape

b. 2d mode shape

Figure 4. Flexural whirl on rigid supports - 1st and 2d mode shapes.

PUBLISHED ANALYSES AND COMPUTED PROGRAMS

The original work by Jeffcott1 was apparently the first published analysis of critical speed
response to correctly predict that shaft whirl amplitudes would come back down at speeds
above the critical speed. A typical response curve for the simple rotor (Figure 5) analyzed
by Jeffcott is shown in Figure 6. For this system the effective stiffness and mass can be
predicted quite easily, which allows accurate predictions of the critical speeds.

Note that the ordinate of the curve is shaft whirl amplitude, and the abscissa is shaft speed.
The most important information displayed by such a curve is the speed at which the peak
response occurs (the critical speed). The predicted whirl amplitude is less important because
it is usually inaccurate. (This is because it is determined by the magnitude of system
damping, which is extremely difficult to predict.)

16
 Figure 5. Jeffcott rotor.

5 OFFSET
CRITICAL SPEED

SPEED - rpm
Figure 6. Response curve for Jeffcott rotor.

The response curve of Figure 6 is also valid for the cylindrical rigid-body mode of Figure 3a.
However, if the response curve for the rigid rotor is extended out to a speed range that
includes the conical critical speed also, the curve will display both peaks as shown in Figure
7. (The shaft deflection measured in the conical mode would depend on the measurement
location along the shaft.)

As the speed range of the rotor is further increased, more peaks will be displayed on the
response curve, corresponding to the shaft flexural modes. The principal objective of all
critical speed analysis is to determine the speeds at which these peaks occur, so that they
can be adjusted outside of the operating speed range by proper design.

17
 CONICAL iODE

«st
N
"Mil CIITICAl M CIITICAl

SPEED - rpn
Figure 7. Response curve for rigid rotor on flexible supports.

Although the complexity of modern turboshaft rotor systems has provided an incentive for
development of critical speed analyses and computer programs with a high degree of mathe-
matical sophistication, the accuracy of the resultant predictions is completely dependent on
the accuracy of the stiffness, mass, and damping data used as input to the calculations.

In turboshaft engine design, one of the easiest rotor-bearing parameters for the engineer to
adjust is the stiffness of the bearing supports. Therefore, the critical speed analysis is often
used to generate curves like Figure 8 (for example, see reference 3).3 Since a large number
of critical speed calculations must be made to generate such a curve, it follows that the
speed of computation can be an important factor in choosing a method of analysis.

Five basic analytical methods (with ma^y variations of each) have been developed to calcu-
late critical speeds of rotor shafts. In their modern form, each has been adapted for use
with high-speed digital computers. All of these methods are applicable to rotor-bearing
systems much more complex than the Jeffcott rotor, but each has certain advantages and
disadvantages that make the choice of method dependent on the type of rotor, the speed
range of interest, and the accuracy required. The methods are briefly described as follows:

3
J. W. Lund, Stability and Damped Critical Speeds of a Flexible Rotor in Fluid Film
Bearings, ASME Paper No. 73-DET-103, 1973 ASME Design Technology Conference,
Cincinnati, Ohio, September 9-12, 1973.

18
 1. The Stodola method, now sometimes called the matrix iteration method.4's
The calculation begins with an assumption of the first mode shape, from which
the inertia loading due to whirling is calculated at an assumed critical speed. This
loading is used to calculate the shaft deflection curve, which is compared with the
assumed mode shape. If the agreement is not good the process is repeated using
the new calculated deflection curve and a properly adjusted value for the critical
speed. A surprisingly small number of iterations will converge to the true first
mode shape (eigenvector) and critical speed (eigenvalue). The calculation for
higher-order mode shapes is somewhat more complicated, but can be accomplished
up through several critical speeds.

This method is especially well adapted to the use of influence coefficients, which
can often be experimentally verified and thus add to confidence in the results.

The disadvantages of this method are a large requirement for computer storage
capacity and a loss of accuracy for the higher-order modes.

In graphical form, this was the first method used for turboshaft design analysis.

2. The Prohl-Myklest d method.6'7'8 This method is similar to the Hölzer method

for torsional vibrati in analysis; the shaft is divided into a number of sections with
the mass of each section concentrated at the ends. Beginning with the boundary
conditions at one end of the shaft, the moment, shear, and inertial loading for
each section are matched up with adjacent sections until the other end is reached.
If the required boundary conditions at this end are not met, then the calculation
is repeated for another value of shaft speed (which determines the inertia loading).
The critical speeds, which are the only speeds at which shaft deflection can exist
without external loads or unbalance, are deduced from a plot of the end boundary
conditions versus speed. The speeds at which the boundary conditions are met
are the critical speeds.

This is presently the most commonly used method for critical speed analysis in

4
J. P. Den Hartog, Mechanical Vibrations, New York, McGraw Hill Book Co., 1956,
pp. 155-159, 162-165.
5
J. V. LeGrow, Multiplane Balancing oj Flexible Rotors - A Method of Calculating Correc-
tion Weights. ASME Paper No. 71-VIBR-52, 1971 ASME Vibrations Conference, Toronto,
Canada, September 8 10, 1971.

6
M. A. Prohl, A General Method for Calculating Critical Speeds .)f Flexible Rotors. Trans-
actions of the ASME, Vol. 67, 1945, Journal of Applied Mech?nics, Vol. 12, pp. A 142 ■
A-148.
7
N. 0. Myklestad, A New Method of Calculating Natural Modes of Uncoupled Bending
Vibration of Airplane Wings and Other Types of Beams, Journal of the Aeronautical
Sciences, Vol. 11, No. 2, April 1944, pp. 153-162.
8
Den Hartog, pp. 229-232.

19
 the turboshaft engine industry. It has recently been modified to include damping
effects and to improve computational e f f i c i e n c y . 3 ' 9

The Rayleigh-Ritz method, 1 0 also k n o w n as the energy method. The maximum

strain energy in the rotor shaft is equated to the maximum kinetic energy due to
whirling. Since the kinetic energy is a function of shaft speed, the resulting
equation can be solved for the critical speed.

The main disadvantage of this method is that the calculations for strain and
kinetic energy require the shaft deflection shape in the desired mode, which is
generally not k n o w n and must therefore be assumed. However, the accuracy of
the method is not very sensitive to errors in this assumption, and ther • are
parametric variation methods available to minimize the error.

4. The characteristic equation m e t h o d . " Substitution of a general exponential solu

t i o n into the differential equations of m o t i o n yields a polynomial in the eigenvalues
(critical speeds), the roots of which are the critical speeds.

This method is not presently used much in the industry, probably because the
polynomials obtained for real systems are of high order and are therefore d i f f i c u l t
to solve. There has been some recent interest in applying modern algebraic
techniques to update this method, however. 1 2

5. The numerical integration method, also called the marching method. n ' 14 The
equations of dynamics for the rotor-bearing system are solved numerically,
marching out the m o t i o n f r o m the initial conditions for small steps of increasing
time.

•\. Fdkey and P. V. Chang, Avoiding Iterative Searches To Find Critical Speeds o] Rotating
With the Transfer Matrix Method, ASME Paper No. 71-VIBR-53, 1971 ASME Vibra
nons Conference, T o r o n t o , Canada, September 8 10, 1971.

1
Gen Hartog, pp. 141 147.

" R. L. Eshleman and R. A. Eubanks, On the Critical Speeds of a Continuous Rotor,

Transactions of the ASME, Journal of Engineering for Industry, November 1969,
pp. 1180-1188.
12
N. 0 . Myklestad, Seminar on Critical Speed Analysis, University of Florida, 1970.

13
J. M. Vance and H. R. Simmons, Computer Analysis of the Transient Dynamics oj Rigid
and Flexible Rotors With Bearing Dampers, Pratt and Whitney Aircraft F T D M 301,
September 15, 1969.

R. G. Kirk and E. J. Gunter, Transient Response of Rotor-Bearing Systems, ASME Paper

No. 73-DET-102, ASME Design Technology Conference, Cincinnati, Ohio, September 9-12,
1973.

20
 This method cons mes large amounts of computer time, since the critical speeds
are obtained by calculating steady-state amplitudes for a large number oi spt^ds
and plotting the results as in Figure 6 o. 7. Sufficient computer time must be
allowed at each speed for initial transients to die out.

This is the only method, however, that can simulate the nonlinear system. It is
therefore valuable for verification of the results from methods 1-4 and for investi-
gation of the effect of nonlinearities on critical speeds.

As shown above, critical speed analysis is now advanced to a highly sophisticated state, with
a number of techniques developed to handle practically any type of rotor-bearing system
The references noted give some description of the basic concept and theory for each method.

Each of the turboshaft engine manufacturers and industry-related research houses has its
own highly developed computer programs for critical speed analysis based on these methods.
Only a few of these programs are well documented in the published literature, • ow
ever, which tends to perpetuate a certain lack of comparability of results, even for similar
problems.
With one exception (described in the following section), critical speed analysis appears to be
sufficiently well developed to treat all foreseeable cases in turboshaft rotor-bearing design.
Any significant advances to be made in these analytical capabilities will most likely be to
improve computational efficiency and reduce computer time costs. The real need is tor
more accurate determination of rotor system parameters and characteristics to be used as
input to the various computer programs.

THE "RUBBER DISK EFFECT"

Gyroscopic effects can be included in all of the critical speed analyses listed above; practi-
cally all of the industry computer programs do include these effects. This is usually done
b treating compressor and turbine wheels as rigid disks. Until recently this was a valid
assumption since these disks are normally very rigid compared to the shaft.

Some recent problems in accurately predicting critical speeds for extremely high-spee<d
(TM > 60 000 rpm) turbomachinery, u nd for compressor rotors with long thin blades, tend to
suqgest that there is a class of machines for which the rigid disk assumption -s not valid.
The extremely high out-of-plane forces developed on whirling disks at high speed aPPafe V
can bend the disk and thus modify the gyroscopic moments The type of aiggerted cJefom
tion is shown in Figure 9. The terminology "rubber disk effect was apparently coined by
Dr. J. P. Den Hartog during consultations with industry companies on this problem.

" J K Sevcik System Vibration and Static Analysis, ASME Paper No. 63-AHGT-57,
Aviation and Space, Hydraulic, and Gas Turbine Conference and Products Show, Los
Angeles, California, March 3-7, 1963.

" R. T. Bohm, Designing Complex Twrbo Rotor Systems With Controlled Vibration Charac-
teristics. SAE Paper No. 928B, National Transportation, Powerplant, and Fuels and Lubricants
Meeting, Baltimore, Maryland, October 19-23, 1964.

21
■

^C .RlVi'Ctf.
. ».I» KP V « X

^««iSii———
«/»

/ ^

BURlNG SUPPORT SllfFMbS • Ib/i«.

Figure 8. Typical critical speed map.

7
VI

RIGID DISK ON FIEXIIIE SHAFT

M
Ji
FUXIIIE DISK ON FIEXIIIE SHAFT

Figure 9. Disk deformation affecting gyros^opic inomer.tv

22
EFFECT OF SHAFT JOINTS

Turboshaft engine rotors are actually an assembly of shafts, spacers, disks, and blades. In
many cases the methods of attachment of these pieces are governed by maintainability con
siderations that require a disassembly capability, or by thermal expansion clearance requ're
ments that are often satisfied by the use of spline joints.

Critical speed calculations require knowledge of the bending stiffness of the rotor shafts in
the system at all points along the shaft. The effects on bending stiffness of the various
joints and discontinuities are largely unknown at the present time. Even with welded joints,
which are becoming more common in modern engines, the shaft stiffness often cannot be
accurately predicted in the design stages. In practice, engineering design analysts make
"educated guesses" based on experience for these stiffnesses and later improve the estimates
by component testing or by critical speed measurements.

The same type of problem exists for the prediction of stiffness for bearing support struc-
tures, but this important subject will be considered separately in a later section.

RECOMMENDATIONS

It is recommended that:

1. The rubber disk effect be investigated analytically to determine the range of

applicability for the rigid disk assumption and to properly modify critical
speed analysis to include the effects of disk flexibility. The results of the
analytical investigation should be verified by test. One possible test method
would be to install a very thin uniform disk on a series of test rig rotor shafts
with different shaft stiffnesses. The gyroscopic effect on the critical speed
could then be determined at the various critical speeds produced by the
different shafts.

2. A test program be conducted to experimentally determine the bending stiffness

of the various types of shaft joints, shoulders, and splines that are commonly
found in rmall turboshaft engine rotors. Parametric variation studies should be
made for those types of joints that are not likely to have constant stiffness
properties (for example, spline joint stiffness is probably dependent on the type
of spline, the clearance, the applied torque, and the state of lubrication).

HIGH-SPEED BALANCING

As turboshaft engines are made smaller and more powerful, it becomes ever more difficult
to keep critical speeds out of the operating range. With bearing supports designed to be
soft in order to keep the rigid rotor criticals below idle speed, the first flexural mode is
often very close to being a free-free mode. But even with the high frequencies usually
associated with such a mode, modern turboshaft engines are operating very close to, and in
one or two cases, above, a flexural critical speed. The power turbine shaft bearing span

23
required by passage through the compressor spool makes this problem even more severe.

In order u> operate saftly through or near tho flexura! critical speeds, a rotor shaft must
eithut tjfc well balanced, highly damped, or both damped and balanced to a sufficient
begreu in combination.

Since synchronous whirl due to unbalance does not produce significant rates of change of
shaft flexure, it is difficult or impossible to provide damping in the shaft itself. Support
damping may or may not be effective on flt-xural modes, depending on the bearing loca-
tions. Therefore, for at least some cases of supercritical* or near critical rotor design, a
highly refined state of shaft balance becomes mandatory for safe and smooth operation.

MULTIPLANE BALANCING

For a completely rigid rotor it has been shown that two balance planes (locations along
the shaft at which unbalance measurements and corrections are made) are both necessary
and sufficient for balancing to be effective at all speeds.17

Rotor shafts which flex obviously have a state of balance that changes with the magnitude
and shape of shaft deflection. There has been some controversy among experts as to
exactly how many balance planes are required for flexible rotors, but it is fairly certain
that more than two are required17 and that no more than N + 2 are needed,18 where N is
the number of critical speeds to be passed through.

The NASA-Lewis Research Center has supported a program of research and development at
Mechanical Technology Incorporated 19 ' 20 for high-speed balancing of flexible rotors.
This work is based on an influence coefficient method, in which the required correction
weights and locations are calculated from experimental data obtained with the rotor running
at various speeds with 3 known trial unbalance weight attached. References 19 and 20

"Henceforth, the word "supercritical" will refer to speeds above the first bending critical
speed
17
Den Hartog, pp. 232-246.
,8
W. Kellunoerger, Should J Flexible Rotor lie Balanced in /V or (\ + 2) Planes ', ASME
Paper No. 71-VIBR-55, 1971 ASME Vibrations Conference, Toronto, Canada, September
8 10, 1971.
19
J. M. Tessarzik, Flexible Rotor Balancing by the Exact Point-Speed Influence Coefficient
Method. NASA CR-72774, MTI-70TR59.
20
J. M. Tessarzik, R. H. Badgley, and W. J. Anderson, Flexible Rotor Balancing by the
Exact Point-Speed Influence Coefficient Method, ASME Paper No. 71-VIBR-91, 1971
ASME Vibrations Conference, Toronto, Canada, September 8-10, 1971.

24
report test verification of the theory, which is based on the work of Goodman,21 Rieger,2
and Lund and Tonnesen.23 Application of the method to a high-speed torbosh ift orpine Is
described in a publication by Rieger and Badgley.24

A similar method for flexible rotor balancing which uses influence coefficients triut are cai
culated from beam theory, rather than experimentally measured, is reported by LeGrow/

Another method for flexible rotor balancing is the modal method, 18 in which balance planes
are selected to have maximum influence on particular flexural modes, and the balancing is
done at the flexural critical speeds to minimize the effect of other modes.

Whatever method eventually proves to be best for small turboshaft engine applications, it
appears that balancing in more than two planes will become necessary as engine design
evolves further into the region where flexural whirl modes are significant.

Flexible rotor balancing requirements will also have their own effect on engine design. The
significant factor is that flexural modes must be balanced either in the engine or on
specially constructed bearing supports that simulate the engine bearing stiffnesses and loca-
tions. The latter approach might be simplified for some applications using very soft supports
(without squeeze film dampers) by mounting the rotor free-free in a specially designed
balance machine. In any case, however, balancing the rotor outside the engine would be
compromised by the necessity for disassembly and reassembly on installation, unless there
was a major change in design philosophy to allow installation of assembled rotors in engine
cases after balancing.

"IN-PLACE" BALANCING

"In-place" balancing means balancing complete assembled rotors in the engine. This would
also require a major change in design philosophy to provide integral probes for measurement
of rotor deflection and access ports for balance correction. Major changes always meet with
resistance in established industries, of course, and many in the turboshaft engine industry
believe that in-place balancing of small turboshaft engines is not a practical proposal. This
could be determined by a demonstrator program to convert an existing modern engine,
adding the necessary probes and access ports and attempting to apply the latest technology
for quick and accurate balancing.

21
T. P. Goodman, A Least-Squares Method for Computing Balance Corrections, AS ML
Paper No. 63 WA-295, 1963.
22
N. F. Rieger, Computer Program for Balancing Flexible Rotors, MTI Technical Report
67 TR68, NASA-Lewis Research Center Contract NA53-10926, September 1967.
23
J. W. Lund and J. Tonnesen, Analysis and Experiments on Multiplane Balancing of a
Flexible Rotor, ASME Paper No. 71-VIBR-74, 1971 ASME Vibrations Conference,
Toronto, Canada, September 8-10, 1971.
24
N. F. Rieger and R. H. Badgley, Flexible Rotor Balancing of a High-Speed lias Turbine
Engine. SAE Paper No. 720741, National Combined Farm, Construction, and Industrial
Machinery and Powerplant Meetings, September 11-14, 1972.

25
There is a potential side benefit from in-place balancing that could prove to be more valuable
than the balancing itself. This is the diagnostic capability provided by the integral
proximity probes. Manufacturers of high-speed compressors and other types of rotating
machinery have recently begun to use such probes as a routine source of information on
bearing condition, state of rotor balance, etc.

For turboshaft engine applications, proximity probes in the hot gas sections would be sub-
jected to an extremely severe environment in terms of temperature and erosion. At present,
probes are not available to survive this environment, but they are under active development
and should be available within 1 to 2 years from the date of this report.

RECOMMENDATIONS

It is recommended that the following subjects be pursued:

1. Multiplane balancing of power turbine shafts. Since these shafts generally are
designed in a way that makes balance corrections in more than one plane
exceptionally difficult, a conceptual and experimental study should be made to
determine if multiplane flexible balancing techniques can be practically applied
to such a shaft. The very difficult problem of balancing such a shaft in place
(inside the compressor spool) should be considered.

2. Demonstrator program for in-place balancing. A study should be made to deter-

mine if an existing engine can be modified with access ports and integral proximity
pr.^es to demonstrate application of the latest multiplane balancing techniques in
the engine. Possible impact on engine performance and reliability should be
considered. For example, balance correction would require either removal of
material in the engine or provision for attachment of balance weights, either of
which could affect reliability.

PREDICTION AND CONTROL OF BEARING SUPPORT PROPERTIES

The two bearing support properties of interest are stiffness and damping. Most modern
engines are designed with soft supports (low stiffness) to place the rigid rotor critical speeds
below the operating range. Support damping can reduce or eliminate the peaking synchro-
nous response to unbalance at the critical speeds (resonance), and can suppress many of the
nonsynchronous responses and associated instabilities.

Asymmetry and cross-coupling are important aspects of support properties, as they have a
profound effect on rotor dynamics, especially with regard to stability. An example of
asymmetry is different bearing support stiffnesses in the horizontal and vertical directions.
An example of cross-coupling is damping force generated in the horizontal direction by
motion in the vertical direction.

26
The influence of support properties on rotor shaft dynamics is well understood by experts
in the field, the subject having been extensively researched both analytically and experimen
tally.25 "3' The real problem at present is a lack of capability to reliably and accurately pre-
dict support characteristics from design data.

SQUEEZE FILM BEARING DAMPERS

The squeeze film bearing damper is probably the most significant development of the last
decade affecting high-speed rotor dynamics. Figure 10 illustrates the principle of operation.

The terminology "squeeze film" is descriptive of what takes place in the annular clearance
space, to which oil is continuously supplied. In a typical turboshaft application, the
clearance space is located between the damper housing bore and the outer race of a rolling
element bearing, which is therefore a loose fit in the damper housing. The rotating
unbalance of the rotor induces the bearing to orbit within the damper housing. In a prop
erly designed damper, metal to-metal contact is prevented by hydrodynamic support of the
oil film. The bearing race is normally constrained by a key to prevent rotation.

The oil-film support provides both low dynamic stiffness and high damping (energy dissipa-
tion). The results of a good design are greatly reduced dynamic bearing loads, elimination
of resonance, and in some cases even a reduction of whirling amplitudes. The latter result
is intuitively surprising to many; it is often difficult to convince a machine designer that he
needs to provide a loose bearing housing clearance in order to reduce whirling amplitudes.

2,5
E. J. Gunter, Influciuv oj i'hxihly Moiinted Rolling Ekmi'tit Ihwings on Rotor Responsf,
Part I - Linear Analysis, ASME Journal of Lubrication Technology, January 1970, pp. 59-75.
26
R. Williams, Jr., and R. Trent, Tfw Effects of Nonlinear Asymmetric Supports on Turbine
Engine Rotor Stahilitv, SAE Paper No. 700320, National Air Transportation Meeting,
New York, N. Y., April 20-23, 1970.
27
H. F Black and A. J. McTernan, Vibration of a Rotating Asymmetric Shaft Supported
in Asxnimetric Bearings, Journal of Mechanical Engineering Science, Vol. 10, No. 3, 1968,
p. 252. ' "
28
W. R. Foote, H. Poritsky, and J. J. Slade, Jr., Critical Speeds of a Rotor With Unequal
Shaft Flexibilities, Mounted in Bearings of Unequal Flexibility -1, ASME Journal of Applied
Mechanics, June 1943, Transactions, pp. A-77 - A-84.
29
E. E. Messal and R. J. Bonthron, Subharmonic Rotor Instability Due To Elastic Asymme-
try, ASME Paper No. 71-VIBR 57, 1971 ASME Vibrations Conference, Toronto, Canada,
September 8-10, 1971.
30
B. Sternlicht, Stability and Dynamics of Rotors Supported on Fluid-Film Bearings,
ASME Journal of Engineering for Power, October 1963, pp. 331-342.
iX
E. H. Hull, Shaft Whirling as Influenced by Stiffness Asymmetry, ASME Journal of
Engineering for Industry, May 1961, pp. 219 226.

27
 0)
a
E
(0
■D

O)

i
3
CT

g.

28
In the above description, the phrases "properly designed damper" and "good design' are
important. At present, it is not possible to predict damper performance from design data
with confidence. The good designs that have been obtained are largely based on empiri' at
information from earlier designs. The same can probably be said about the bad designs.

A squeeze film damper is basically an oil film bearing with zero rotation. The damper
force response should therefore be predictable from hydrodynamic bearing theory. For
example, Lund,: has defined eight coefficients that give the forces in terms of journal
position and motion in Cartesian coordinates. The problem is that these coefficients are
predicted differently by each of the various theoretical models based on certain assumption'
for analytical simplification; that is, "long bearing" theory," "short bearing" theory,14
and various assumptions about the circumferential location of the beginning and end of the
oil film (the boundary conditions).

Most experimental data for oil film bearings has been obtained for the case with pure
rotation (no orbiting), which is the antithesis of a squeeze film damper.

Whenever an attempt is made to design a damper on a scientific or analytical basis, it is

usually based on the work of Cooper, '■ who gives mostly qualitative results of a parametric
study made with an unbalanced rotor supported by a squeeze film. The most convincing
ind indisputable fact shown by Cooper's experiments is that any restriction of damper
orbiting by mechanical stops or bumpers produces a tremendous increase in bearing loads,
perpetuates resonance that would not occur without the bumper, and generally destroys the
good effects of the damper. Yet there are still designers in the industry who insist on
mechanical stops "to limit rotor excursions".

In 1973, Jones 3fi published the results of an experimental study of squeeze film hydro
dynamics. He reported a reasonable verification of short bearing theory except for cases
with large eccentricity (bearing journal far off center). The main shortcoming of this study
is that the apparatus motion was restricted to orbiting about the damper centerline, a con
dition rarely obtained in engines unless the damper is coupled with a stiff mechanical
support spring.

■,2 J. W. Lund and E. Saibel, Oil Whip Whirl Orbits of a Rotor in Sleeve Hearings, ASME
Journal of Engineering for Industry, November 1967, pp. 813-822.

" Y. Hori, A Theory of Oil Whip, ASME Journal of Applied Mechanics, June 1959,
pp. 189 198.
,4
G. 8. Dubois and F. W. Ocvirk, Analytical Derivation and Esi» rimental Evaluation of
Short-Hearing Approximation for Full Journal Bearings, Report 1157, National Advisory
Committee for Aeronautics, 1953. (Also see NACA TN 2808 and TN 2809, 1952.)
3(1
S. Cooper, Preliminary Investigation of Oil Films for the Control of Vibration, Institution
of Mechanical Engineers (England), Lubrication and Wear Convention, 1963.

•" M. G. Jones, An Experimental Investigation of Si/uceze Film llydroilynamies. Report No.

R.320, National Gas Turbine Establishment, Ministry of Defence (England), January 1973.

29
Since the squeeze film damper is a nonlinear device, the possibility exists that several shaft
motions can satisfy dynamic equilibrium. In fact, an analytical and experimental investigation
by White37 has confirmed the existence of jumps from one stable orbit to another of
different magnitude, a phenomenon originally suggested by Cooper's work.35

Some manufacturers and turbo machinery research houses have developed computer simula-
tions of the transient dynamics of engine rotors on squeeze film supports.13 • l4 Although
such simulations help one to understand the effects of changes in various design parameters,
little confidence can be placed in their quantitative predictions until the hydrodynamic
response portion of the analytical model is either verified or appropriately modified by
experimental studies.

MECHANICAL BEARING SUPPORTS

Understandably, some engineers in the industry prefer not to rely on the imprecisely known
characteristics of squeeze film dampers to provide the requirei support properties for con-
trol of rotor dynamics. Generally speaking, the stiffness of mechanical supports is easier to
control, even if not so easy to predict in the preliminary stages of design. The damping of
such supports is, of course, very predictable, usually being very low.

Even when squeeze films are used, they often are mounted in series or parallel with mechani-
cal spring supports. Mechanical bearing supports are usually designed to provide the minimum
stiffness practical while still maintaining the required strength and reliability. Most fall
roughly into one of three categories:

1. The squirrel cage, so named because of its appearance, made as a cylinder of

thin metal ribs.

2. Welded rod support. In this design the bearing mount is cantilevered on several
metal rods parallel to the shaft centerline.

3. Corrugated metal ring. A variety of these designs are ail constructed to fit snugly
around the outer bearing race and provide a mechanical cushion through deflec-
tion of small segments or protruding elements.

The stiffness properties of all these designs can be controlled with good accuracy by testing
and iterative redesign as required. The capability to predict these properties analytically in
preliminary design stages needs to be improved. One way of doing this would be to stan-
dardize some designs throughout the industry, although the practicality of accomplishing
this in a competitive environment is questionable.

Another type of mechanical bearing support used to provide both low stiffness and some
degree of damping is the common "Oring". Grooves in the bearing housing bore are made
to hold the "O-rings" so that they are compressed into an elliptical shape when the roller
bearing outer race is inserted. The "O-rings" are made of an elastomeric material, and
their fatigue properties and thermal degradation properties are not well known for this type

37
D.C. White, The Dynamics of a Rigid Rotor Supported on Sciueezc Film Bearings, Con
ference on Vibrations in Rotating Systems, London, Proceedings, Institution of Mechanical
Engineers, February 14-15, 1972.
30
of application. At present they are considered to be suitable only for sdut : f;1 appMcattor
in the cold (compressor) section of turboshaft engines. Further research and development
could extend the range of applications for "0 ring" supports; their greatest advantage is
that they provide damping along with stiffness properties that should be KcirMiMy p^e
dictable.

RECOMMENDATIONS

The following research is recommended:

1. Experimental determination of squeeze film bearing damper force response. The

eight hydrodynamic bearing coefficients defined by Lund,32 or their equivalent,
should be experimentally determined for damper configurations typically used in
turboshaft engines. An analytical model should be developed and verified to
predict these coefficients from design data. The study should include the case of
bearing orbits not about the damper centerline. Variables of potential interest
are clearance, effective bearing length, oil supply pressure, viscosity, orbit speed,
and leakage rate. The squeeze film forces should be separated from forces
generated by seals and/or parallel springs.

2. Prediction and control of mechanical bearing support stiffness. A few of the

spring mount designs commonly in use for turboshaft rotor bearing support
should be selected for a broad range of applications. These designs should then
be analyzed, and modified if necessary, so that their support stiffness properties
are predictable from data representing a minimum number of design variables.
The prediction capability thus generated might serve to standardize a basic
bearing support design throughout the industry.

NONSYNCHRONOUS EXCITATION

Rotor-bearing-system response to rotor unbalance is called synchronous whirl because it is

characterized by rotor orbiting at shaft speed. Synchronous whirl is usually in the same
direction as shaft rotation (forward), but can also be backward in direction. All other
motions executed by the rotor-bearing system can be classified either as nonsynchronous
whirl or as nonsynchronous vibration. These motions are often self-excited by rather
subtle mechanisms, and are sometimes associated with the possibility of dynamic instability.*
This latter characteristic makes an understanding of nonsynchronous excitation necessary to
insure safe designs for flight propulsion.

Nonsynchronous whirl has only recently been widely recognized in turboshaft engines. In
fact, some of the vibration and dynamic measurement techniques that have been in common
use by the industry were probably not capable of registering nonsynchronous response, so
intense was the focus of interest on synchronous response to unbalance.

'Dynamic instability is defined here as a motion that becomes unbounded (until system
limits are exceeded) either with time or with some normally variable parameter of the
system, following an initial perturbation.

31
The major sources of nonsynchronous excitation that have been identified and studied by
researchers to date are:

1. Asymmetric stiffness properties of shafts or bearing supports26'29'31'38'41

2. Internal friction in shafts or other rotating parts26, 42 46

"

3. Aerodynamic or gas flow excitation47- 48

38 H. D. Taylor, Critical Speed Behavior of Unsymmetrical Shafts, ASME Journal of Applied

Mechanics, June 1940, Transactions, pp. A-71 - A-79.

39
R. C. Arnold and E. E. Hoft, Stability of an Unsymmetrical Rotating Cantilever Shaft
Carrying an Unsymmetrical Rotor, ASME Paper No. 71-VIBR 58, 1971 ASME Vibrations
Conference, Toronto, Canada, September 8-10, 1971.

40
S. H. Crandall and P. J. Brosens, On the Stability of Rotation of a Rotor With
Rotationally Unsymmetric Inertia and Stiffness Properties, ASME Journal of Applied
Mechanics, December 1961, pp. 567-570.

41
D. E. IMewland, On the Stability of Rotation of a Rotor With Rotationally Unsymmetric
Inertia and Stiffness Properties (Brief Note), ASME Journal of Applied Mechanics,
December 1964, pp. 723-724. '

42
F. F. Ehrich, Shaft Whirl Induced by Rotor Internal Damping, ASME Journal of Applied
Mechanics, June 1964, pp. 279-282.

43
A. Seirig, Whirling of Shafts in Geared Systems, ASME Journal of Engineering for Indus-
try, May 1967, pp. 278-283.

44
E. J. Gunter, Jr., The Influence of Internal Friction on the Stability of High Speed Rotors,
ASME Journal of Engineering for Industry, November 1967, pp. 683-688.

4
5 E. J. Gunter, Jr., and P. R. Trumpler, The Influence of Internal Friction on the Stability
of High Speed Rotors With Anisotropie Supports, ASME Journal of Engineering for Indus-
try, November 1969, pp. 1105-1113.

46
J. M. Vance and J. Lee, Stability of High Speed Rotors With Internal Friction, ASME
Paper No. 73-DET-127, ASME Design Technology Conference, Cincinnati, Ohio, September
9-12, 1973.

47
F. F. Ehrich, An Aeroelastic Whirl Phenomenon in Turbo machinery Rotors, ASME Paper
No. 73-DET-97, ASME Design Technology Conference, Cincinnati, Ohio, September 9-12,1973.

48
J. S. Alford, Protecting Turbomachinery From Self-Excited Whirl, ASME Journal of
Engineering for Power, October 1965, pp. 333-344.

32
 4. Excitation from fluids trapped within rotor shafts49 51

5. Rubbing friction between radial surfaces of rotors and stator housings52's3

6. Excitations from other rotating components in the same engine structure

(e.g., excitation of a power turbine rotor from unbalance of a compressor
rotor)

7. Excitations from gears, at the gear tooth mesh frequencies

BACKWARD WHIRL

Backward whirl is one of the least-understood phenomena of rotor dynamics. Den Hartog54
states that he doubted the possibility of its existence until he finally observed it in a model.
Although Den Hartog concluded that backward whirl is only of minor importance, more
recent experience in the turboshaft engine industry indicates otherwise. Large amplitudes
of backward whirl have been observed in engines, with the source of excitation traced to
the ball bearing effects originally identified by Yamamoto.55

49 F. F, Ehrich, The Influence of Trapped Fluids on High Speed Rotor Vibration, ASME
Journal of Engineering for Industry, November 1967, pp. 806-812.
50
R. J. Fritz, The Effects of an Annular Fluid on the Vibrations of a Long Rotor,
Part 1 • Theory, ASME Paper No. 70-FE-30, Fluids Engineering, Heat Transfer, and Lubri-
cation Conference, Detroit, Michigan, May 24-27, 1970.
51
R. J. Fritz, The Effects of an Annubr Fluid on the Vibrations of a Long Rotor. Part 2 -
Test. ASME Paper No. 70 FE-31, Fluids Engineering, Heat Transfer, and Lubrication
Conference, Detroit, Michigan, May 24-27, 1970.

"I.C. Begg, Friction Induced Rotor Whirl - A Study in Stability, ASME Paper No.
73-DET-106, ASME Design Technology Conference, Cincinnati, Ohio, September 9-12,
1973.
53
F. F. Ehrich, The Dynamic Stability of Rotor/Shaft Radial Rubs in Rotating Machinery,
ASME Journal of Engineering for Industry, November 1969, pp. 1025-1028.
54
Den Hartog, p. 265.
55
T. Yamamoto, On Critical Speeds of a Shaft Supported by Ball Bearings, ASME Journal
of Applied Mechanics, June 1959, pp. 199-204.

33
A number of critical speed studies3, " • 3 1 ' 5 6 have shown that backward whirl ir un eigr:1
value (natural frequency) of rotor bearing systems. Since any shaft unbalance always
rotates forward with the shaft, backward whirl would not seem to be excited by
unbalance. However, it has been shown that it is excited when the bearing support stiff
ness is asymmetric,57 and it is also believed to be associated with gyroscopic moment-,.^

References 2 and 56 disagree as to the possibility of the existence of backward whirl wh<
damping is present in the system.

Radial rub between rotors and stator housing is the most easiiy visualized mechanism
driving backward whirl, and this may actually be the source of most cases occurring in
turboshaft engines. Inspection of Army helicopter engines being rebuilt at the U. S. Army
Aeronautical Depot Maintenance Center provides proof that such rubs are frequent and
pronounced in these engines.

FRICTION-INDUCED WHIRL

The practical realization of design advantages to be gained in small turboshaft engines by

operation at supercritical speeds has been prevented by (1) the necessity to pass through
the resonances at the critical speeds and (2) the instabilities of self-excited whirl associated
with high-speed operation.

If it is assumed that the first problem can be overcome through refined balancing techniques
and the provision of external damping, self-excited whirl remains as a potentially more
serious problem that can destroy a rotor-bearing system whenever regions of instability are
encountered.

Nonsynchronous whirl induced by internal friction is probably the most common type of
self-excited whirl in turboshaft engines. The nature and cause of this phenomenon have been
rigorously analyzed by Gunter44 and Ehrich.42

A subsequent investigation by Gunter and Trumpler4s shows that the threshold speed of
instability can be increased by asymmetric bearing supports. These analyses also confirm
earlier experimental findings that both increased bearing support flexibility and external
damping raise the threshold speed of instability. Since squeeze film dampers provide both
of these effects simultaneously, it is to be expected that friction-induced whirl could be
suppressed by their use.

The author of this report has investigated the effect of rotor unbalance, shaft stiffness
asymmetry, and the location of external damping in the system on friction-induced whirl. 4 '

S6
T. R. Kane, An Addition to the Theory of Whirling, ASME Journal of Applied Mechanic.
September 1961, pp. 383-386.
57
W. T. Thomson, Vibration Theory and Applications, Prentice Hall, Englewood Cliffs, N.
1965, pp. 84-86.
58
R. 8. Greene, Gyroscopic Effects on the Critical Speeds of Flexible Rotors, ASME Journal
of Applied Mechanics, Transactions, Vol. 70, 1948, pp. 369-376.

34
I;
wes found that aerodynamic drag can be an important source of external damping to
suppress friction-induced whirl, whenever stiff bearing mounts are used, because bearing
supports cannot dissipate significant amounts of energy unless they are flexible enough
to move.

The most important parameter affecting the threshold speed of instability for friction-
induced whirl is the ratio of internal friction to external damping. For large values of
this parameter, rotor-bearing operation is unstable at all speeds above the critical speed
associated with the motion producing the internal friction. For values of about unity
(external damping equal to internal friction), the threshold speed is about twice the
critical speed. Thus, it is seen that with a reliable and effective mechanism for external
aamping, rotor-bearinq systems could be safely operated at speeds up to about 80% above
i x- cmical speed.

The problem in turboshatt engine design is that neither external damping nor internal
friction can be reliably predicted from design data, or even after hardware already exists.
At present, these variables can only be measured indirectly from their dynamic effects
on the system.

There are numerous potential sources of internal friction in a typical turboshaft engine
rotor assembly. In addition to hysteresis of the material itself, shaft splines, shrink fits,
and bolted connections are all capable of generating friction forces as the rotor-shaft
assembly flexes. Some experimental data giving at least the relative magnitudes of
friction generated by these mechanisms could be very helpful to the rotor dynamics
engineer for preliminary design analysis.

AERODYNAMIC INDUCED WHIRL

Rotor whirling induced by aerodynamic forces on blades and seals is such a complex
phenomenon that research to date has been confined mostly to hypothesizing qualitative
theories to explain it. It can be surmised that an analogy to "oil whip" in hydrodynamic
bearings might exist for bladed disks in cylindrical housings with small tip clearances, or
that a similar analogy to "propeller whirl flutter" might exist for compressor stages with
long blades.

Combined experimental nuasuremonts of rotor motion and circumferential pressure

distributions are needed to establish the relationships that exist between disk whirling
and aerodynamic forces.

OVERHUNG ROTORS

This subject is isolated here because there has been a long history of dynamics-related
problems associated with overhung rotor disks which are almost certainly not associated
with synchronous response to unbalance. Figure 9 illustrates the overhung rotor disk
configuration that is characterized by a disk having shaft bearings on only one side.

35
The problems with overhung rotor disks have been recognized by numerous investigators,
and there is a considerable body of literature addressing the subject.39- 40' '!8'59 Yet the
problems seem to persist.

An exaggerated example of such problems is the spin pit test rig configuration used by
most turboshaft engine manufacturers to spin disks up to high speed for burst tests or for
proof tests. The disk is suspended vertically in an underground pit (to minimize danger
to surroundings) from an extremely flexible quill shaft to insure supercritical operation and
thus minimize whirling. Occasionally, and yet persistently, the test is ruined by premature
failure of the rotor-shaft system from violent shaft whirling. These failures seem to be
unaffected by rotor balance, and would therefore appear to result from some type of
dynamic instability.

In turboshaft engines, especially in first compressor stages, or fans, backward whirl and
nonsynchronous whirl appear to occur more prevalently when the disks are overhung.
Apparently then, either the results of the research in the literature cited are not being consis-
tently applied to engine design, or a still unknown mechanism is producing these effects.
The former proposition is more likely.

RECOMMENDATIONS

It is recommended that the following subjects be researched;

1. Causes of backward whirl. A combined analytical and experimental study should

be made to identify all possible conditions and excitations capable of producing
backward whirl. The analytical study should include the effects of damping and
gyroscopic moments. Predicted eigenvalues for backward whirl should be evalua-
ted in terms of likelihood of excitation. Dynamic stability should be investigated
for all possible cases. The experimental study should be designed to verify the
conditions producing backward whirl as predicted by analysis, including prediction-,
that are currently in the literature but that have never been experimentally
verified. A special study should be made of rotor friction rubs as a source of
excitation, Some variables of interest for this study are rotor-stator clearance,
coefficient of friction, and speed.

2. Mechanisms of internal friction. A primarily experimental study should be made

to quantitatively assess the magnitudes of internal friction that can be produced
by shaft splines, press fits, bolted connections, and any other types of joints
commonly used in turboshaft rotor assemblies. The measured internal friction
should be compared with the magnitude of external damping estimated from
measurements of vibratory response.

3. Aerodynamic excitation of rotor whirl. Circumferential pressure distributions and

rotor whirl should be measured simultaneously in an experimental compressor
rig with adjustable parameters such as blade tip clearance, stator configuration.

*9T. C. Huang and F. C. C. Huang, On Precession and Critical Speeds of Two-Bearing

Machines With Overhung Weight, ASME Paper No. 67-VIBR-19, 1967 ASME Vibrations
Conference, Boston, Massachusetts, March 29-31, 1967. (Also published in the ASME
Journal of Engineering for Industry.)

36
 and inlet pressure. If configurations are found which produce aerodynamic-
induced whirl, an attempt should be made to analytically simulate the phenomenor
and develop a prediction capability.

Dynamic stability of overhung (cantilevered) rotors. An analytical study should

be made to determine possible causes of the seemingly inordinate number of
problems with overhung rotors in high-speed systems. The study should be done
with access to a large bank of experience with such systems, and should begin
with a classification of rotors having a history of failures or problems in order to
identify critical parameters. One such critical parameter might be the ratio of
polar moment of inertia to transverse moment of inertia of the overhung disk. It
the analytical study reveals likely causes of dynamic instability, they should be
verified by experiment.

HYDRODYNAMIC BEARINGS

Hydrodynamic bearings provide rotor shaft support through pressure in a thin fluid film
between the shaft journal and bearing. The fluid may be either a gas (usually air) or a
liquid (usually oil). The pressure may be self-generated from rotation of the journal (wedge
effect) or may be supplied externally from an auxiliary pump or compressor. The chief
distinguishing characteristic of hydrodynamic bearings, as opposed to boundary-lubricated
bearings, is that metal-to-metal contact between the journal and bearing does (should) not
occur except possibly durino startup or shutdown.

Rolling-element bearings are firmly entrenched in aircraft turboshaft engine design, and the
author knows of only one case in which a hydrodynamic bearing lias been successfully used
for this application. The two principal reasons for this are that (1) a rolling-element
bearing usually fails in a gradual way, which gives warning time before aircraft power is
lost, and (2) rolling-element bearings reject less heat to the lubricant, thus allowing a smaller
heat exchanger for cooling.

The latter comparison does not apply to gas bearings. This, and the reliable availability of
lubricant for a gas bearing, has provided considerable incentive for research and develop-
ment of gas bearings for turboshaft applications.

The chief advantage that all hydrodynamic bearings offer is long life. Gas bearings also
offer extremely low friction coefficients, although they have a much smaller load capacity
for their size than fluid film bearings.

GAS BEARINGS

Reference 60 describes a feasibility study performed by Mechanical Technology Incorpora-

ted (MTI) and Pratt and Whitney Aircraft for the Eustis Directorate, U.S. Army Air Mobility

60
P. W. Curwen, Feasibility of Gas Bearings for Small High-Performance Aircraft Gas
Turbines, USAAVLABS Technical Report 68-87, U. S. Army Aviation Materiel Labora-
tories, Fort Eustis, Virginia, March 1969, AD 684956.

37
 and inlet pressure. If configurations are found which produce aerodynamic-
induced whirl, an attempt should be made to analytically simulate the phenomenor
and develop a prediction capability.

4. Dynamic stability of overhung (cantilevered) rotors. An analytical study should

be made to determine possible causes of the seemingly inordinate number of
problems with overhung rotors in high-speed systems. The study should be done
with access to a large bank of experience with such systems, and should begin
with a classification of rotors having a history of failures or problems in order to
identify critical parameters. One such critical parameter might be the ratio of
polar moment of inertia to transverse moment of inertia of the overhung disk. If
the analytical study reveals likely causes of dynamic instability, they should be
verified by experiment.

HYDRODYNAMIC BEARINGS

Hydrodynamic bearings provide rotor shaft support through pressure in a thin fluid film
between the shaft journal and bearing. The fluid may be either a gas (usually air) or a
liquid (usually oil). The pressure may be self-generated from rotation of the journal (wedge
effect) or may be supplied externally from an auxiliary pump or compressor. The chief
distinguishing characteristic of hydrodynamic bearings, as opposed to boundary-lubricated
bearings, is that metal-to-metal contact between the journal and bearing does (should) not
occur except possibly during startup or shutdown.

Rolling-element bearings are firmly entrenched in aircraft turboshaft engine design, and the
author knows of only one case in which a hydrodynamic bearing has been successfully used
for this application. The two principal reasons for this are that (1) a rolling-element
bearing usually fails in a gradual way, which gives warning time before aircraft power is
lost, and (2) rolling-element bearings reject less heat to the lubricant, thus allowing a smaller
heat exchanger for cooling.

The latter comparison does not apply to gas bearings. This, and the reliable availability of
lubricant for a gas bearing, has provided consideraule incentive for research and develop-
ment of gas bearings for turboshaft applications.

The chief advantage that all hydrodynamic bearings offer is long life. Gas bearings also
offer extremely low friction coefficients, although they have a much smaller load capacity
for their size than fluid film bearings.

GAS BEARINGS

Reference 60 describes a feasibility study performed by Mechanical Technology Incorpora-

ted (MTI) and Pratt and Whitney Aircraft for the Eustis Directorate, U.S. Army Air Mobility

60
P. W. Curwen, Feasibility of Gas Bearings for Small High-Performance Aircraft Gas
Turbines, USAAVLABS Technical Report 68-87, U. S. Army Aviation Materiel Labora
tories, Fort Eustis, Virginia, March 1969, AD 684956.

37
Research ui:d Development Laboratory, to apply gas bearings to a small turboshaft engine.
Oyndm.c siabtnty f/(i,blems were encountered in the attempted application, which is a
cor;»moii occurrence with gas bearings.

More recfently the Air Force Aero-Propulsion Laboratory has supported work at AiResearch
Corporation and MTI to develop gas bearings for intershaft support between a power turbine
shaft and compressor spool. This work is not complete at this date.

There is a severe radial space problem in the design of a front-drive engine with the power
turbine shaft inside the compressor spool. If the compressor rotor could be mounted on
gas bearings, its shaft diameter could be increased to accommodate a larger and stiffer power
turbine shaft. The large-diameter compressor spool would be compatible with the need for
large gas bearings to support the load. This appears to be the one unexplored, but logical,
potential application of gas bearings to small turboshaft engines.

OIL FILM BEARINGS

Although rolling element bearings are called "antifriction" bearings, they do not necessarily
have a lower friction coefficient than oil-film bearings. Furthermore, in some stationary
applications, oil-film bearings have demonstrated extremely long life capabilities.

In the one successful turboshaft application of an oil-film bearing, mentioned above, the
incentive was a limitation of radial space.

The required heat exchanger size and flight safety considerations probably preclude broad
application of these bearings in turboshaft aircraft engines. Nevertheless, it is likely that
there will be special applications where long life or radial space is a problem that can best
be solved by an oil-film bearing.

Problems of dynamic instability with oil-film bearings, such as "oil whip",33 have been
effectively solved through research and development of new bearing designs. An example
is the "tilting pad bearing", in which the cylindrical bearing sleeve is replaced with several
pivoted blocks around the circumference, each one supporting the journal over a segment
of the cylindrical surface. This type of bearing is more stable and can also accept greater
shaft misalignment.

Due to limitations of radial space between the compressor spool and power turbine shaft,
oil-film bearings may find useful application in this location as intershaft bearings. This
has been attempted in at least one engine development program. First tests were not suc-
cessful, and the idea was abandoned without determining the cause of failure.

Intershaft oil-film bearings require corotating shafts (same direction), since the hydrodynamic
support is greatly reduced in counterrotating shafts and completely disappears if the shafts
rotate at equal speeds in opposite directions. Figure 11 illustrates how the wedge effect to
generate support pressure depends on rotation direction when both the journal and the
bearing are rotating.

Whenever a hydrodynamic bearing is used in aircraft propulsion machinery, the possibility

of lubricant supply interruption or failure must be considered in the design. Both military
specifications and FAA requirements demand some operating time after lubricant supply

38
 z o •
o as IM
IM

g£ jg^ «/»
IM - IMo
** O o.o
a •* IM
3B oIM
«^ 2 3^ ^
■^ 21 Qg g

< 2o IM
O
>
OC o o O re
2) oc a
O O o re
u
w

ndence of intershaft bearing load

on direction of both shafts.
IM ^^ 0) '*-
O <^ |s
o o Q 2
g •
^ • ^ ■• 2j MJ ""
«f O */» IA
Hi
5 o 2 2?

il
3

►«• o IM

5 O _ JE «
_j
r* 5
oe * — •**
IM M. oe
^ IM «J ^
5 SS «A
S
OC
o «^
IM
o o

39
interruption. One way to approach this problem is to design lubricant reservoirs near the
bearings.

Hybrid bearings, in which a hydrodynamic bearing is located inside a rolling element bear-
ing, are under development at the NASA-Lewis Research Center."1 The hybrid bearing
may prove to combine some of the best features of both types. Since the film bearing and
rolling element bearing are in series, the result is a corotating journal and bearing for the
former and a smaller DN value for the latter.

RECOMMENDATIONS

It is recommended that the following subjects be investigated:

1. Feasibility of gas bearing support for compressor spools, to allow more room for
a stiffer power turbine shaft in front-drive engines. This should begin with a
preliminary study to determine if the concept is sound. It may be that the
large-diameter gas bearing and spool shaft would require an increase in the overall
cross-sectional size of the engine, thus producing an undesirable trade-off.

2. Feasibility of oil-film intershaft bearings for midspan power turbine shaft support.
There is little question that this proposal is conceptually sound, except for a
possible lubricant supply problem and excluding counterrotating shafts. The
most severe problem is likely to be shaft misalignment and deflection, which can
be attacked using recently developed bearing design technology. The possibility
of longer bearing life and higher critical speeds for the power turbine shaft makes
this study especially attractive.

3. Application of hybrid bearings to aircraft turboshaft engines. Progress at

NASA-Lewis Research Center should be monitored on a continuing basis to
ascertain when this type of bearing may be ready and suitable for application
to rotor-shaft support in military engines. This may be the only way to get the
higher shaft speeds and longer life that will be required in the future.

61
L. J. Nypan, H. W. Scibbe, and B. J. Hanrock, Optimal Speed Sharing Characteristics of
a Series-Hybrid Bearing, ASME Journal of Lubrication Technology, January 1973,
pp. 76-81.

40
 REFERENCES

1. Jeffcott, H. H., The Lateral Vibration of Loaded Shafts in the Neighborhood of a

Whirling Speed - The Effect of Want of Balance. Philosophical Magazine, Vol. 37,
No. 6, 1919, p. 304.

2. Gunter, E. J., Jr., Dynamic Stability of Rotor-Bearing Systems, NASA-SP-113, 1966.

3. Lund, J. W., Stability and Damped Critical Speeds of a Flexible Rotor in Fluid Film
Bearings. ASME Paper No. 73-DET-103, 1973 ASME Design Technology Conference,
Cincinnati, Ohio, September 9-12, 1973.

4. Den Hartog, J. P., Mechanical Vibrations. New York, McGraw-Hill Book Co., 1956,
pp. 155-159, 162-165.

5. LeGrow, J. V., Multiplane Balancing of Flexible Rotors - A Method of Calculating

Correction Weights, ASME Paper No. 71-VIBR-52, 1971 ASME Vibrations Conference,
Toronto, Canada, September 8-10, 1971.

6. Prohl, M. A., A General Method for Calculating Critical Speeds of Flexible Rotors.
Transactions of the ASME, Vol. 67, 1945, Journal of Applied Mechanics, Vol. 12,
pp. A-142 - A-148.

7. Mylkestad, N. O., A New Method of Calculating Natural Modes of Uncoupled Bending

Vibration of Airplane Wings and Other Types of Beams, Journal of the Aeronautical
Sciences, Vol. 11, No. 2, April 1944, pp. 153-162.

8. Den Hartog, J. P., Mechanical Vibrations. McGraw-Hill Book Co., New York, 1956,
pp. 229-232.

9. Pilkey, W., and Chang, P. Y., Avoiding Iterative Searches To Find Critical Speeds of
Rotating Slwfts With the Transfer Matrix Method. ASME Paper No. 71VIBR-53,
1971 ASME Vibrations Conference, Toronto, Canada, September 8-10, 1971.

10. Den Hartog, J. P., Mechanical Vibrations. McGraw-Hill Book Co., New York, 1956,
pp. 141-147.

11. Eshleman, R. L, and Eubanks, R. A., On the Critical Speeds of a Continuous Rotor.
Transactions of the ASME, Journal of Engineering for Industry, November 1969,
pp. 1180-1188.

12. Myklestad, N. O., Seminar on Critical Speed Analysis. University of Florida, 1970.

13. Vance, J. M., and Simmons, H. R., Computer Analysis of the Transient Dynamics of
Rigid and Flexible Rotors With Bearing Dampers, Pratt and Whitney Aircraft FTDM-
301, September 15, 1969.

41
14. Kirk, R. G., and Gunter, E. J.. Transient Respond of Rotor-Bearing Systen^.
ASME Paper No. 73 DET 102, ASME Design Techno**, >-« -
September 9 12, 1973.

I- Stivci* j. K., System Vibration and Statu Analysis, MOWIE. r-JO^ - ' -
Aviation and Space, Hydraulic, and Gas Turbine Commence and Prouucu aho...
Los Angeles, California, March 3-7, 1963.
16 Bohm, R. T., Designing Complex Turbo Rotor Systems With Contmllc! 'br..r »
Characteristics, SAE Paper No. 928B, National Transportation, Fowerplam an:. Fee -
and Lubricants Meeting, Baltimore, Maryland, October 19 23, 1964.

17 Den Hartog, J. P., Mechanical Vibrations, McGraw-Hill Book Cc New Vert- m*':
pp. 232-246.
18. Kellenberger, W., Should a Flexible Rotor Be Balanced in X or (N + 2) Planes:. ^
ASME Paper No. 71-VIBR-55, 1971 ASME Vibrations Conference, Toronto, -anbCj
September 8-10, 1971.

19. Tessarzik, J. M., Flexible Rotor Balancing by the Exact Point-Speed Influena
Coefficient Method, NASA CR-72774, MTI-70TR59.

20. Tessarzik, J. M., Badgley, R. H., and Anderson, 1W J. Flexible

Exact Point-Speed Influence Coefficient Method, ASME Paper No 71 VI BR 9 ,
ASME Vibrations Conference, Toronto, Canada, September 8 10, 1971.

21. Goodman, T. P., A Least-Squares Method for Computing Balance Corrections,

ASME Paper No. 63-WA-295, 1963.

22. Rieger, N. F., Computer Program for Balancing of Flexible * » ' u ^ ^ J ' c , e c t , f " - a l . .
Report 67 TR68, NASA-Lewis Research Center Contract NA53-10926, Septemuc.

23 Lund J. W., and Tonnesen, J., Analysis and Experiments on Multiplane Balancing >J
Flexible Rotor. ASME Paper No. 71-VIBR-74, 1971 ASME Vibrations Confertr c.
Toronto, Canada, September 8-10, 1971.
?4 Rieger, N. F.. and Badgley. R. H., Flexible Rotor Balancing of a llivh-Srer-1 Ha*
Engine, SAE Paper No. 720741, National Combined Farm, Construction, ana ino^i
Machinery and Powerplant Meetings, September 11-14, 1972.

25 Gunter, E. J., Influence of Flexibly Mounted Rolling Element Bearings on Rotor

Response. Part I - Linear Analysis. ASME Journal of Lubrication Technology. Jara-i-y
1970, pp. 59-75.
26. Williams. R., Jr., and Trent, R., The Effects of Nonlinear Asymmetric Supports on
Turbine Engine Rotor Stability. SAE Paper No. 700320, National Air Transportation
Meeting, New York, N. Y., April 20-23, 1970.
27 Black, H. F., and McTernan, A. J., Vibration of a Rotating Asymmetric Shajt SU;J, ..r: .
in Asymmetric Bearings. Journal of Mechanical Engineering Science, Vol. 10, No. ->
1968, p. 252.

42
?•" F'to't '•*' Porit«kw M and Slgde, J. J., Jr, Critical Speeds of a Rotor With Un-
equal Shaft Flexibilities Mounted in Bearings of Unequal Flexibility -1, ASME
of Annijmi Merhanics, June 1943, Transactions, pp. A-77 - A-84.

Mesial. F. £ . 'nd Bonthron, R. J., Subharmonic Rotor Instability Due To Elastic

K. A«5VIE Paper No. 71-VIBR-57, 1971 ASME Vibrations Conference,
Toronto. Canada, September 8-10, 1971.

30 S ::rrlic*t 9., Stability and Dvnanrics of Rotors Supported on Fluid-Film Bearings,

ASME Journal of Engineering for Power, October 1963, pp. 331-342.

31. Hull, fc. H. Shaft Whirling as Influenced by Stiffness Asymmetry, ASME Journal of
Engineering for Industry. May 1961, pp. 219-226.

3? Lund, J. W , and Saibel, E., Oil Whip Whirl Orbits of a Rotor in Sleeve Bearings,
ASML 'Jur.ia^qf Engineering for Industry, November 1967, pp. 813-822.

I? Ho»\ Y. A Pieorv of OH Whip, ASME Journal of Applied Mechanics, June 1959.

po. 189 198.

34. D'ibois. G. B., ?nd Ocvirk, F. W., Analytical Derivation and Experimental Evaluation
oj Stmt t-Bearing Approximation for Full Journal Bearings, Report 1157, National
Advisory Committee for Aeronautics, 1953. (Also see NACA TN 2808 and TN 2809,
19F2.)

Cooper S., Preliminary Investigation of Oil Films for the Control of Vibration,
Institution of Mechanical Engineers (England), Lubrication and Wear Convention,
1963.

36. Jones. M. G., An Experimental Investigation of Squeeze Film Hydrodynamics,

Report No. R.320, National Gas Turbine Establishment, Ministry of Defence (England)
January 19^3.

?7 White, 0. C., The Dynamics of a Rigid Rotor Supported on Squeeze Film Bearings,
Conference on Vibrations in Rotating Systems, London, Proceedings, Institution of
Mechanical Engineers, February 14-15, 1972.

3S Tjylor, H. 0.. Critical Speed Behavior of UnsymmetricaI Shafts, ASME Journal of

Applied Mechanics, June 1940, Transactions, pp. A-71 - A-79.

39. Arnold, R. C., and Hoft, E. E., Stability of an Unsymmetrical Rotating Cantilever
Shaft Carning an Unsymmetrical Rotor, ASME Paper No. 71-VIBR-58, 1971 ASME
Vibrations Conference, Toronto, Canada, September 8-10, 1971.

40 Crandall. S. H., and Brosens, P. J., On the Stability of Rotation of a Rotor With
Rotationallv Unsymmetric Inertia and Stiffness Properties, ASME Journal of Applied
Mechanics, December 1961, pp. 567-570.

41 Newland. D. E.. On the Stability of Rotation of a Rotor With Rotationallv Unsymme-

tric Inertia and Stiffness Properties (Brief Note), ASME Journal of Applied Mechanics,
December 1964, pp. 723-724.

43
42. Ehrich, F. F., Shaft Whirl Induced by Rotor Internal Damping. ASME Journal of
Applied Mechanics, June 1964, pp. 279-282.

43. Seirig, A., Whirling of Shafts in Geared S\ terns, ASME Journal of Engineering for
Industry, May 1967, pp. 278-283.

44. Gunter, E. J., Jr., The Influence of Internal Friction on the Stability of High Speed
Rotors, ASME. Journal of Engineering for Industry, November 1967, pp. 683-688.

45. Gunter, E. J., Jr., and Trumpler, P. R., The Influence of Internal Friction on the
Stability of High Speed Rotors With Anisotropie Supports, ASME Journal of
Engineering for Industry, November 1969, pp. 1105-1113.

46. Vance, J. M., and Lee, J., Stability of High Speed Rotors With Internal Friction.
ASME Paper No. 73-DET-127, ASME Design Technology Conference, Cincinnati,
Ohio, September 9-12, 1973.

47. Ehrich, F. F., An Aeroelastic Whirl Phenomenon in Turbo machinery Rotors. ASME
Paper No. 73-DET-97, ASME Design Technology Conference, Cincinnati, Ohio,
September 9-12, 1973.

48. Alford, J. S., Protecting Turbo machinery From Self-Excited Whirl. ASME Journal of
Engineering for Power, October 1965, pp. 333-344.

49. Ehrich, F. F., The Influence of Trapped Fluids on High Speed Rotor Vibration.
ASME Journal of Engineering for Industry, November 1967, pp. 806-812.

50. Fritz, R. J., The Effects of an Annular Fluid on the Vibrations of a Long Rotor.
Part I ■ Theory, ASME Paper No. 70-FE-30, Fluids Engineering, Heat Transfer, and
Lubrication Conference, Detroit, Michigan, May 24-27, 1970.

51. Fritz, R. J., The Effects of an Annular Fluid on the Vibrations of a Long Rotor.
Part 2 - Test, ASME Paper No. 70-FE-31, Fluids Engineering,Heat Transfer, and
Lubrication Conference, Detroit, Michigan, May 24-27, 1970.

52. Begg, I. C, Friction Induced Rotor Whirl - A Study in Stability, ASME Paper No.
73-DET-106, ASME Design Technology Conference, Cincinnati, Ohio, September 9 12,
1973.

53. Ehrich, F. F., The Dynamic Stability of Rotor/Shaft Radial Rubs in Rotating
Machinery. ASME Journal of Engineering for Industry, November 1969,
pp. 1025-1028.

54. Den Hartog, Mechanical Vibrations, McGraw Hill Book Co., New York, 1956,
p. 265.

55. Yamamoto, T., On Critical Speeds of a Shaft Supported by Ball Bearings, ASME
Journal of Applied Mechanics, June 1959, pp. '99-204.

44
56. Kane, T. R., An Addition to the Theory of Whirling, ASME Journal of Applied
Mechanics, September 1961, pp. 383-386.

57. Thomson, W. T., Vibration Theory and Applications, Prentice Hall, Englewood Cliffs,
N. J., 1965, pp. 84-86.

58. Greene, R. B., Gyroscopic Effects on the Critical Speeds of Flexible Rotors, ASME
Journal of Applied Mechanics, Transactions,Vol. 70, 1948, pp. 369-376.

59. Huang, T. C, and Huang, F. C. C, On Precession and Critical Speeds of Two-Bearing

Machines With Overhung Weight, ASME Paper No. 67-VIBR-19, 1967 ASME Vibrations
Conference, Boston, Massachusetts, March 29-31, 1967. (Also published in the ASME
Journal of Engineering for Industry.)

60. Curwen, P. W., Feasibility of Gas Bearings for Small High-Perfornuince Aircraft Gas
Turbines, USAAVLABS Technical Report 68-87, U. S. Army Aviation Materiel Labora-
tories, Fort Eustis, Virginia, March 1969, AD 684956.

61. Nypan, L. J., Sctbbc, H. W., and Hanrock, B. J., Optimal Speed Sharing Characteristics
of a Series-Hybrid Bearing, ASME Journal of Lubrication Technology, January 1973,
pp. 76-81.

H7'MI-74
45