AFAPL-TR-76-90

COMPUTER PROGRAM OPERATION MANUAL

ON "SHABERTH" A COMPUTER PROGRAM
0FOR THE ANALYSIS OF THE STEADY STATE
AND TRANSIENT THERMAL PERFORMANCE
OF SHAFT-BEARING SYSTEMS
RESEARCH LABORATORY
SKF INDUSTRIES, INC.
ENGINEERING & RESEARCH CENTER
KING OF PRUSSIA, PA.
.
As.
OCTOBER 1976
TECHNICAL REPORT AFAPL-TR-76-90
a-. Approved for public release; distribution unlimited.
C-0
.2 .AIR FORCE AERO PROPULSION
LABORATORY
AIR FORCE WRIGHT AERONAUTICAL
LABORATORIES
AIR FORCE SYSTEMS COMMAND
WRIGHT-PATTERSON AIR FORCE BASE, OHIO 45433
AND
NAVAL AIR PROPULSION TEST CENTER
TRENTON, NEW JERSEY 08628
NOTICE
When Government drawings, specifications, or other data
are used for any purpose other than in connection with a de-
finitely related Government procurement operation, the United
States Government thereby incurs no responsibility nor any obli-
gation 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 implica-
tion or otherwise as in any manner licensing the holder or any
other person or corporation, or conveying any rights or per-
mission to manufacture, use, or sell any patented invention
that may in any way be related thereto.
Publication of this report does not constitute Air Force
approval of the reports findings or conclusions. It is published
only for the exchange and stimulation of ideas.
This manual describes the work performed by SKF Industries,
Inc. at its Technology Center in King of Prussia, Pa. for the
United States Government, including the United States Air Force
Systems Command, Air Force Aero Propulsion Laboratory, Wright-
Patterson Air Force Base, Ohio and for the Naval Air Propulsion Test
Center, Trenton, N. J. Work was performed over a seven month period
starting in February 1976 under U. S. Air Force Contract No. F33615-
76-C-2061 and Navy M1PR No. N62376-76-MP-00005. Mr. John Schrand
administered the project for the Air Force and Mr. Raymond Valori
administered the project for the Navy.
The project was conducted at SKF under the direction of
Messrs. P. S. Given and T. E. Tallian. The SKF report designa-
tion is No. AL76PO30.
The manual contains the results of analytical modelling
and computer program development.
This report has been reviewed by the Information Office,
(ASD/OIP) and is releasable to the National Technical Informa-
tion Service (NTIS). At NTIS, it will be available to the
general public, including fcreign nations.
This technical report has been reviewed and is approved
for publication.
o chrand Raymqbd Valori
Pr ect Engineer Project Engineer
T -,CODIANDEFOR THlE OHMANDING OFFICER
/ 11owarF. Jole lokwoo-d,Supervisor
Chief, Lubrication Branch Lubricants and Power Drive
Fuels and Lubrication Division Systems Division, Naval Air
Air Force Aero Propulsion Lab. Propulsion Test Center
AIR FORCE/56780/10 August 1977 - 210
UNCLASST TID
SECURITY CLASSIFICATION OF THIS PAGE (When Dl Enteoes
'REPORT DOCUMENTATION
PAGE
BEAD DISTRU=ONOIS
i nL'-b~rw -' 'uml |. 30VT ACCESSln J-',.*-;.,: TI - - . .. "-
AFP R-76-9 /s
a--azA Tomputer Program Operation
Manual on-'i"'S-RBERTH' A Computer Program A Computer Program
for the .nalysis of the Steady State an Operation Manual
Transient Thermal Performance oT Shaft.O
XY AUTHOR:* u-a IaIAort OR GRANT NU We) . j. eelu7AF Contract No. F3615- ;w
W -.,./rece :ius~
"J./irvics - Navy MIPR No. M62376-76- P-00005
. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT, TASK
SKF Industries,
Inc., Technology Center
AREA & ORK UNIT NUMBERS
1100 First Ave.
King of Prussia, PA 19406
II. CONTROLLING OFFICE NAME AND ADDRESS -- EI flT
"
T
Air Force Aero Propulsion Laboratory i/l Jul 1276
Air Force Systems Command % 13. "UM-E-U . /
Wright-Patterson Air Force Base, Ohio IN___"
14. MONITORING AGENCY NAME A AODRESS(It different from Controllind Office) 15. SECURITY CLASS. (
Air Force Aero Propulsion Laboratory Unclassified
and Naval Air Propulsion Test Center
Trenton, New Jersey 08628 IS.. DECLASSIFICATION/DOWNGRADING
SCHEDULE
II. DISTRIBUTION STATEMENT (of Chie Report)
Approved for public release; distribution unlimited
17. DISTRIBUTION STATEMENT (of the abstract entered in Block 20, if different from. Report)
IS. SUPPLEMENTARY NOTES
It. KEY WORDS (Continue on ro erese eide if necessy and Identify by block number)
Ball and Cylindrical Roller Bearings, Shaft Bearing System Per-
formance, Deflections, Steady State and Transient Temperature Maps,
Rolling Element Bearing Lubrication; Elastohydrodynamics (EHD) and
Starvation, Hydrodynamics, Boundary, Friction, Hydrodynamic, EHD,
Asperity-Partial EHD, Cage and Rolling Element Dynamics, Shaft
20. ABSTRACT (Continue on "ero aide It neceeary end Identify by block number)
This report is a self contained manual describing an advanced
state-of-the-art analytical computer program (SHABERTH) for the
study of steady state and transient thermal performance of rolling
element bearing and shaft systems. This program embodies mathe-
matical models describing the role of the lubricant in bearing
behavior. Descriptions of these models are set forth along with
a detailed description of the program organization method of solu-
tion and convergence criteria. Input data Dreparation forms and
DD , v0 "1s 1473 EDITION OF INOV 65 IS OBSOLETE UNCLASSIFIED
SECURITY CLASSIFICATION OF THIS PAGE (When Date Entered
<CCoy/.
UNCLASSIFIED
SECURITY CLASSIFICA TION OF THIS PAGE(fthm Doe gntered)
19. Key Words
and Housing Fits and Bearing Clearance Change
20. Abstract
program output are discussed and examples are included.
.... .... ...
................
UNCLASSIFIED
SECURITY CLASSIFICATION OF THIS PAGcru"en Dome EIored)
TABLE OF CONTENTS
1. Introduction
Page
1
2. Problem Formulation and Solution 4
2.1 Temperature Calculations 5
2.2 Bearing Dimensional Change Analysis 17
2.3 Bearing Inner Ring Equilibrium 18
2.4 Bearing Quasi-dynamic Solution 20
3. Program Input 29
3.1 Types of Input Data 29
3.2 Data Set I - Title Cards 30
3.3 Data Set II - Bearing Data 32
3.4 Data Set III- Thermal Data 45
3.5 Data Set IV - Shaft Data 51
4. Computer Program Output 54
4.1 Introduction 55
4.2 Bearing Output 55
4.3 Rolling Element Output 59
4.4 Thermal Data 61
4.5 Shaft Data 61
4.6 Program Error Messages 61
S. Guides to Program Use 66
6. References 69
APPENDICES I MATHEMATICAL MODELS
Appendix I 1 Heat Transfer Information
Appendix I 2 Bearing Diametral Clearance Change Analysis
from Cold Unmounted to Mounted Operating
Conditions
Appendix 1 3 Elastic Shaft Analysis
Appendix I 4 Concentrated Contact Calculations
Appendix I 5 Lubricant Property and EHD Film Thickness
Models
Appendix I 6 Traction and Inlet Friction Calculations
Appendix I 7 Rolling Element Inertia Forces and Moments
Appendix I 8 Rolling Element Bearing Cage Model
Appendix I 9 Bearing Fatigue Life Calculations
LEi
APPENDICES II PROGRAM INFORMATION
Appendix II 1 SKF Computer Program Shaberth/SKYF Flow Chart
Appendix II 2 SKF Computer Program Shaberth/SKF Input Format Forms
Appendix II 3 SKF Computer Program Shaberth/SKF Sample Output
ii
I
FR_
LIST OF FIGURES
Fig. No.
2.1
Convective Heat Transfer
12
2.2
Divided Flow From Node i
13
2.3
Contact Geometry and Temperatures
16
2.4 Bearing Inertial (XYZ) and Rolling
22
Element (xyz) Coordinate Systems
2.5
Inner Ring-Cage Land Contact Geometry 24
2.6
Outer Ring-Cage Land Contact Geometry
25
2.7
Cage Pocket Normal and Friction Forces 27
Affecting Equilibrium
3.1
Angular Contact Ball Bearing Geometry 36
3.2
Split Inner Ring Ball Bearing Geometry 38
3.3
Roller-Raceway
Contact Geometry
40
iii
LIST OF TABLES
Table Title Page
No.
1 Properties of Four Lubricants 44
iv
- -- -~* ,.
LIST OF TABLES APPEARING IN THE APPENDICIES
Page
I 5.1 Lubricant Properties of Four Oils Used in I 5-3
Program SHABERTH
I 6.1 Tabulation of Constants for Four Oils I 6-9
I 6.2 Dimensionless Coefficients for the I 6-20
Calculation of Line Contact Inlet
Friction
y
LIST OF FIGURES APPEARING IN THE APPENDICIES
Fig. Title Page
No.
I 1.1 Parallel Conduction I 1-3
I 1.2 Series Conduction I 1-3
I 1.3 Heat Transfer Area I 1-8
I 2.1 Bearing Assembly Equivalent Sections I 2-3
I 2.2 Ring Radial Expansion Vs. Rotational Speed I 2-13
Squared
I 3.1 Shaft Coordinate System and Shaft Loading I 3-2
I 3.2 Shaft Schematic Showing Stepwise and Linear I 3-2
Diameter Variation
I 3.3 Schematic Shaft Supports I 3-3
1 4.1 Ball Bearing Geometry I 4-4
I 4.2 Ball Coordinate System Showing Ball Center I 4-5
Position Vectors
I 4.3 Roller Bearing Geometry and Roller Coordinate I 4-8
Systems
I 4.4 Ball Coordinate System Showing Ball-Race I 4-13
Contact Position Vectors
I 4.5 Ball Race Deformed Contact and Deformed I 4-14
Surface Radius
I 4.6 Calculation of Traction Force Components I 4-16
I 5.1 Film Geometry I 5-10
I 6.1 Auxiliary Function y* vs. x* I 6-2
I 6.2 Typical Traction Curves I 6-6
I 6.3 Notation for Rolling Sliding Point Contact I 6-11
1 6.4 Friction Forces on Sliding and/or Rolling Disks I 6-12
vi
. t
Fig. Title Page
No.
I 6.5 Variation of Fs with
1
l I 6-16
I 6.6 Variation of FR with the Dimensionless I 6-17
Meniscus Distance Pl
I 6.7 Configuration of Contacts 1 6-19
I 8.1 Cage and Rolling Element Speeds and Displace- I 8-6
ments
I 8.2 Cage Pocket Geometry I 8-8
I 8.3 Load Capacity Vs. Film Thickness for Hydro- I 8-12
dynamic and Elastohydrodynamic Operating
Regimes
vii
* *~ **"
NOMTiNC LATIO R E
Symbol Definit ion Units*
A a constant in Walther's equation (-)
A surface of contact between media (MM2
A cage-land surface area (mm or in.
2
)
A e area of outer cylindrical surface (mm
2
)
A area of inner cylindrical surface (mm
2 )
A
v
ball frontal area (mm 2)
B auxiliary variable (-)
B B = AX/a, a constant in Walther's (-3
equat ion
C a constant tabulation by Fresco (mm
2
]N or
in.2/lb)
C
0
a non dimensional fluid-geometry (-)
parameter
C specific heat at constant pressu're (W/kg-DegC)
Cr cage pocket clearance (mm)
C drag coefficient (-)
D ball or roller diameter (mm)
D a constant tabulated by Fresco (mm
2
/N or
in.2/lb)
constant tabulated by Fresco (mm
2
iN or
in. /lb)
El E2 Young's modulis for the contacting (N/mm
2
or p.i,
bodies
F
A
axial force (N or ib)
FnlI Fn
2
normal components of resultant force of (N or lb)
the inlet pressure distribution
'4herc multiple units are indicated, the first units given are
those associated with the computer program input and output.
viii
NOMLINCLATURE (CONTD)
Symbol Definition
Units*
FR sliding force acting on the ball (N or lb)
FR
1
, FR2 pumping forces acting on the ball (N or Ib)
tangential forces due to inlet rolling (N or lb)
FR
3
, FS
3
and shearing between hall and cage
F
S
shearing force acting on the ball (N or lb)
FSI, FS
2
inlet friction forces
(N or lb)
F F F force components in the x,y,z coordinate (N or Ib)
FxOFyFz system
F windage force or drag force (N or Ib)
w
F the vector of inertia and drag forces (N or lb)
F the vector sum of the hydrodynamic forces (N or Ib)
acting on the ball at the m-th contact
F b a vector of bearing loads and moments (N or lb mm-
N or in.-ib)
F si a vector of shaft loads and moments (N or lb. & min-
N or in.-lb)
G lubricant coefficient of thermal (I/DegC or
expansion 1/DegF)
If non dimensional film thickness parameter (-)
J moment of inertia of the ball kg-mm
2
Kf conductivity of the film (lb/egF-sec)
K.. the proportion of the heat flow from (-)
IJ node i going to node j.
K9,K10 constants in expression for heat transfer C-)
coefficient
L characteristic length (mm or in.)
10' full film fatigue life (hrs)
M c moment due to fluid friction between the (mm-N or in.- lb
cage and the ring land
*1here multiple units are indicated, the first units given are
those associated with the computer program input and output.
ix
NOMENCLATURE (CONTD)
Symbol Definition Units*
Mx,,MM
z
ball moment components in the x,y,z (mm-N or in.-ib)
coordinate system
M ball moment vector (mm-N or in.-lb)
N
u
Nusselt's number (-)
Pr Prandtl's number (-)
Pd diametral clearance (mm or in.)
PE bearing end play (mm or in.)
Pl,32 forces acting normal to the ball (N or Ib)
surface within the outer and inner
raceway contact ellipse
P3 ball-cage 'normal force (N or ib)
Q load (N or lb)
Qa average asperity borne load (N or lb)
Q the radial component of the minimum (N or Ib)
rolling element-race normal force
non dimensional load parameter (-)
&m the vector normal load per unit length (N/mm or Ib/in.)
of the contact ellipse
R radius of outer ring groove centers (mm)
Re Reynold's number (-)
RxRy effective iadii of curvature parallel (mm or in.)
and transverse
to the rolling direction
respectively
S coordinate along the contact in the (mm or in.)
direction perpendicular to rolling
friction
*111here muitiple units are indicated, the first units given are
those associated with the computer program input and output.
x
NOMENCLATURE (CONTD)
Tymbol Definition
Units*
Sd diametral play (mm or in.)
T time
(sec)
T long time duration
sQCI
Ts starting time (sec)
T1,T
2
traction forces (N or 1b)
T traction force vector at a general (N or lb)
location within the contact
U characteristic speed (m/sec or in./sec)
V fluid entrainment velocity at the (m/sec or in./sec)
contact center
V volume of the nodal element (m
3
or in.
3
V voltage
(volt)
V volume flow rate-through node i (M3/sec)
Vo voltage over long time duration (volt)
V
x
rolling velocity in x direction (m/sec or in./sec)
Vy rolling velocity in y direction (m/sec or in./sec)
X,Y,Z inertial coordinate system (-)
DCL diametral clearance (mm or in.)
EPSFIT user specified convergence criterion (-)
EPI, EP2 a user supplied convergence criterion (-)
EQ temperature equilibrium convergence (-)
criteria for Eq. (3-41)
NEQ number of equations in bearing solution(-)
XCAV volume fraction of lubricant in bearing(-)
cavity oil/air mixture
Wl Whre multiple units arc indicated, the first units given are
those associated with the computer program input and output.
xi
NOMENCLATURE (CONTD)
Symbol Definition Units*
a a constant coefficient in Nusselt's (-)
number
a contact ellipse semi-major axis (mm or in.)
a free convection temperature-exponent (-)
b an exponent in Nusselt's number (-)
b half the contact width (mm or in.)
c an exponent in Nusselt's number (-)
c coefficient of specific heat (W/kg-Deg C)
d exponent in free convection heat transfer (-)
equations
d cage-land diameter (mm)
dm bearing pitch diameter (mm or in.)
fm the vector of friction force per unit (N/mm)
length of the contact ellipse
g gravitational constant (m/sec
2
/ or
in./sec
2
)
h elastohydrodynamic film thickness (mm orp -in.)
hc critical value of film thickeness (mm orp -in.)
hf the film thickeness under fully flooded (mm orm -in.)
conditions
h
s
starved plateau thickness (mm orp -in.)
hA.C. film thickness calculated by Archard- (mm orp -in.)
Cowking formula
hD.H. film thickness calculated by Dowson- (mm orp -in.)
Higginson formula
i j indices of heat flow nodes (-)
*Where multiple units are indicated, the first units given are
this associated with the computer program input and output.
xii
A ~.. .... .. .. o -- ..
NOMENCLATURE (CONTD)
_,iybol Definition Units*
senaration distance between temperature (mm)
nodes
Ge contact length, or in the case of an (mm)
elliptical contact area, 0.8 times the
contact length
n number of rolling elements, total number (-)
of heat flow nodes
P
0
maximum contact pressure (N/mm
2
or psi)
q heat generation rate, net heat transfer (W)
qc heat generated by fluid shearing between (W)
the cage and land
qf fluid drag heat (W)
q, heat generated by shearing force in the (W)
hall-raceway and ball-cage inlet region
qi heat energy in the i-th nodal element (IV)
qT heat generated by traction in the contact (W)
zone
qti heat carried by mass flow from node i (W)
q gi heat generated at node i iV)
qoi heat flow from all neighboring nodes to (V)
node i
qRi,j the heat energy transferred by radiation (M)
Ri~j between nodes i and j
qci,j heat flow transferred by conduction from (M)
CiJ node i to node j
qui,j the heat flow between nodes i and j (M)
qvi'j heat flow transferred by free convection (M)
Vi,.)from node i to node j
*11here multiple unis are indicated, the first units given are
those associated with the computer program input and output.
xiii
NOMENCLATURE (CONTD)
Symbol Definition Units*
qwi,j heat flow by forced convection from (W)
node i to node j
r groove radius (mm or in.)
r
m
a vector from the rolling element center (mm or in.)
to the point of contact
r* meniscus distance from center of contact (mm or/-in.)
along direction of rolling
t temperature (Deg C or Deg K)
u sliding velocity at the contact center (m/sec or
in./sec)
u sliding velocity vector (m/sec or
in./sec)
Ul, u
2
surface velocity of bodies 1 and 2 (m/sec or
relative to the contact in./sec)
u
s
sliding speeci (m/sec or
in./sec)
us sliding speed at which traction coef- (m/sec or
ficient is a maximum in./sec)
x,y,z a local coordinate system established at (-)
each ball location
x sliding velocity scaled by us* (m/sec or
in./sec)
x
I
ball axial position relative to the outer (mm or in.)
race
x
m
maximum variation of x (mm)
Yl ball radial position relative to the (mm or in.)
outer race
Ym maximum variation of y (mm)
*Where multiple units are indicated, the first units given are
those associated with the computer program input and output.
xiv
Symbol Definition Units*
z
c
ball center-cage pocket offset (mm or in.)
& diametral clearance between cage and land (mm or in.)
4 shaft displacement at a bearing location (mm or in.)
ADCL change in bearing diametral clearance (mm or in.)
AT a small increment of time
(sec)
A. angular distance between rolling elements (deg)
Lb bearing deflection vector (mm or rad)
&C lubricant replenishment layer thickness (mm)
so') cumulative distribution function of C-)
standard normal distribution
resistance of heat flow (degC/W)
angular velocity (rad/sec)
c cage angular velocity (rad/sec)
res resultant resistance to heat flow
(degC/W)
contact angle (deg)
Cscaling factor in modified Newton-Raphson (-)
technique
pressure-viscosity index (in.
2
/ib)
inner race contact angle (deg)
outer race contact angle
(deg)
o auxiliary contact angle (deg)
v film coefficient of heat transfer by free (W/m
2
-degC
convection
*Where multiple units are indicated, the first units given are
those associated with the computer program input and output.
xv
NOMENCLATURE (CONTD)
Symbol Definition Units*
a W film coefficient of heat transfer by (W/m
2
-degC)
forced convection
'3
temperature-viscosity coefficient (i/degC)
ball speed vector pitch angle (deg)
the first variation (-)
-e elastic deformation
(mm)
Ex' SySz the linear deflection components of &b (mm)
r surface emissivity
(-)
C a small arbitrary constant
(-)
dynamic viscosity (centipoise or
lbf sec/in.
2
)
the angular deflection components of Ab (rad)
y z
A thermal conductivity (W/M-degC)
A a viscoelastic constant <oil parameter) (-)
traction coefficient (-)
I a coulomb friction coefficient (-)
r scaled byM (-)
EHD fluid traction coefficient (-)
M maximum EHD traction coefficient (-)
V kinematic viscosity (centistokes)
Vi, V2 Poisson's ratio for contacting bodies (-)
*Where multiple units are indicated, the first units given are
those associated with the computer program input and outputs.
xvi
NOMENCLATURE (CONTD)
Symbol Definition Units*
P density
(kg/m
3
)
P o density of the oil
(kg/m
3
)
P dimensionless meniscus distance (-)
Stefan-Boltzmann radiation constant W/m
2
-degK*)
RMS value of the distribution of (deg)
asperity slope angles
RMS value of surface roughness (micrometers)
aximuth angle (deg)
density function of standard normal (-)
distribution
4)s starvation reduction factor (-)
the film thickness reduction factor, (-)
due to heating
thermal diffusivity (mm
2
/sec)
Wc cage orbital velocity (rad/sec)
wo ball orbital velocity
(rad/sec)
x ball angular velocity component about (rad/sec)
the x axis
ball angular velocity component about (rad/sec)
the y axis
Wz ball angular velocity component about (rad/sec)
the z axis
o first derivative ofj)
o
with respect (rad/sec
2
)
to time
angular velocity of ball in x, y, z (rad/sec)
coordinate system
*Where multiple units are indicated, the first units given are
those associated with the computer program input and output.
xvii
SUBSCRIPTS
Symbol Definition
B refers to point where traction curve becomes
nonlinear
C refers to cage or conduction
N refers to current iteration
R refers to rolling or radiation
a, asp denotes asperity effect
f refers to fluid or flooded
i denotes the i-th ball, i-th node, inner ring
j denotes j-th node
k index denoting a specific time interval
m an index-denoting bearing component
o denotes outer ring
s refers to sliding, starvation effect, or shaft
t refers to thermal effect
v refers to free convection
w refers to forced convection
x,y,z denotes components of vector quantities with
respect to x, y, and z coordinates
1,2 refers to bodies 1 and 2
xviii
I. INTRODUCTION
The computer program described herein, SHABERTH, "A Compu-
ter Program for the Steady State and Transient Analyses of
Shaft Bearing Systems," is the third generation of S K F
Computer Program AE72Y003. Program AE72Y003 was developed by
Kellstrom (1) under U. S. Army Contract DAAD05-73-C-0011,
sponsored by the Ballistics Research Laboratory at Aberdeen
Proving Grounds. The original as well as the succeeding
generations of the program consists of the following major
subprograms.
The master program consists of the following major sub-
programs.
1) Bearing Analysis. These subprograms are largely
based upon the methods of Harris, (2,3).
2) Three Dimensional Shaft Deflection Analysis developed
by Norlander and Friedrichson. (See Appendix I 3).
3) Bearing Dimensional Change Analysis based on the
methods of Timoshenko, (4), and adapted to the shaft-
bearing-housing system by Crecelius, (5). (See
Appendix I 2).
4) Generalized Steady State and Transient Temperature
Mapping and Heat Dissipation Analyses based on the
methods of Harris. (6), Fernlund, (7) and Andreason,
(8).
Although the primary function of all three generati ns of
the program is to predict general bearing performance character-
istics, and the bulk of the coding reflects this emphasis,
the steady state and transient heat dissipation and temperature
mapping subprogram may be used on a stand alone basis to model
the thermal behavior of any system which can be represented by
discrete temperature nodes.
The differences between the successive generations of
the program reflect the development and installation of improved
bearing lubrication and friction models, improved analysis
of the bearing cage and improvements in the program structure
which increased the program versatility and solution procedures.
The first geieration of the program used the Newtonian
lubricated friction models developed by Harris (2, 3).
-_.
The second generation of the program which carries the
designation AT74YO01 was created by Crecelius, Liu and Chiu
under Air Force Contract No. F33615-72-C-1467 and Navy
MIPR No. M52376-3-000007 and is documented by McCool, et al
(9).* In that effort, with Program AE72Y003 as the basis,
the ball bearing subprogram was modified to include new models
as follows:
1) An EHD film thickness model that accounts for i)
thermal heating in the contact inlet using a
regression fit to results obtained by Cheng (10)
and ii) lubricant film starvation using theoretical
results derived by Chiu (11).
2) A new semi-empirical model for fluid traction in an
EHD contact (9), is combined with an asperity load
sharing model developed by Tallian (12) to yield a
model for traction in concentrated contacts that
reflects the state of lubrication as it varies from
dry, through partial EHD to the full EHD regime.
3) A model for the hydrodynamic rolling and shear forces
in the inlet zone of lubricated contacts accounting
for the degree of lubricant film starvation, (9).
4) Normal and friction forces between a ball and a cage
pocket are modelled in a way that accounts for the
transition between the hydrodynamic and elasto-
hydrodynamic regimes of lubrication (9).
5) A model for the effect on fatigue life of the ratio
of the EHD plateau film thickness to the composite
surface roughness, (9).
Additionally, models for temperature viscosity and pressure
viscosity variation as functions of temperature given by Walther
(13) and Fresco (14) respectively, were adopted.
Program AT74YO01 is capable of analyzing only a single
axially loaded ball bearing. The program cannot be used to
analyze a multi-bearing system. All other capabilities are
present however.
*Due to the similarities between major segments of SHABERTH
AT75Y004 and AT74YO01, many sections of (9) have been included in
this text, without modification.
2
The basis for the present program, SHABERTH, was AT74Y001.
The latent capability for the analysis of up to five ball and
roller bearings subjected to general, (5 degrees of freedom)
loading, has been utilized. The models added to AT74YO01 are
used in the calculation of both the ball and roller bearing
friction forces and frictional heat generation rates. The
present program also includes a new model for the hydrodynamic
rolling and slip forces in the inlet zone of lubricated line
contacts, based on the work of Chiu, Ref. (15). Additionally,
a cage model developed under NASA Contract No. NAS3-19739
Ref (15) has been added which allows the cage to move with up to
three degrees of freedom versus the one degree of freedom per-
mitted in Program AT74YO01. This cage model may be used in the
analysis of both ball and roller bearings.
Under Air Force Contract No. F33615-76-C-2061 and Navy
NAPTC MIPR No. N62376-76-MP-00005, the capabilities of SHABERTH
were expanded to solve the combined set of multi-rolling element
and cage quasidynamic equilibrium equations.
This exapnsion required changes in the concentrated contact
asperity friction model as well as changes in the cage-rolling
element and cage-ring interaction calculations. Additionally,
the mathematical definition of the range of permitted variable
values was made substantially more accurate.
SHABERTH is intended to be as general as possible with
the following limits on system size.
Number of bearings supporting the shaft - five (5) maximum
Number of rolling elements per bearing - thirty (30) maximum
Number of temperature nodes used to describe the system -
one hundred (100) maximum
The program structure is modular and has been designed to
permit substitution of new mathematical models and refinements
to the existing models as the needs and opportunities develop.
The third generation program, SHABERTH, exists as two
versions, SHABERTH/SKF, SKF Program No. AT75Y004 and SHABERTH/NASA,
SKF Program No. AT76YO01. The differences between the two
versions reside in the calculation of the elastohydrodynamic
(EHD) film thickness and traction forces which develop in the
rolling element-raceway and rolling element-cage concentrated
contacts. The calculation of these factors as performed in the
SKF version is detailed herein. The details of the calculations
performed by the NASA version are presented in Ref. (15).
3
2. PROBLEM FORMULATION AND SOLUTION
The purpose of the program is to provide a tool with which
the shaft-bearing system performance characteristics can be
determined as functions of system temperatures. These system
temperatures may be a function of steady state operation or a
function of time variant conditions brought on by a change
in the system steady state condition. Such a change would be
the termination of lubricant supply to the bearings and other
lubricated mechanical elements.
The program is structured with four nested, calculation
schemes as follows:
1. Thermal, steady state or traisient temperature cal-
culations which predict system temperatures at a given
operating state.
2. Bearing dimensional equilibrium which uses the bearing
temperatures predicted by the temperature mapping
subprograms and the rolling element raceway load dis-
tribution, predicted by the bearing subprograms, to cal-
culate bearing diametral clearance at a given operating
state.
3. Shaft-bearing system load equilibrium which calculates
bearing inner ring positions relative to the respective
outer rings such that the external loading applied to the
shaft is equilibrated by the rolling element loads
which develop at each bearing inner ring at a given
state.
4. Bearing rolling element and cage load equilibrium
which calculates the rolling element and cage equili-
brium positions and rotational speeds based upon the
relative inner-outer ring positions, inertia effects
and friction conditions, which if lubricated, are
temperature dependent.
The above program structure allows complete mathematical
simulation of the real physical system. The program has
been coded to allow various levels of program execution
which prove useful and economical in bearing design studies.
These levels of execution are explained fully in Sections
3, 4, and 5.
4
The structure of the program and the nesting of the
solution loops noted above can be seen clearly in the Program
Flow Chart which is discussed in Appendix II 1.
The sections below present the systems of field equations
which are solved in each of the nested calculation schemes.
A more detailed discussion is contained in (1, 9 and 15).
2.1 Temperature Calculations
Subsequent to each calculation of bearing generated heat
rates, either the steady state or transient temperature mapping
solution scheme may be executed. This set of sequential
calculations is terminated as follows:
1. For the steady state case, when each system tempera-
ture is within EPA Centigrade of its previously
predicted value, EPA is specified by the user. If
it is zero or left blank, a default value of 10 Centi-
grade is used. This criteria implies that the steady
state equilibrium conditions has been reached.
2. The transient calculation terminates when the user
specified time up is reached or when one of the
system temoeratures exceeds 600
0
C.
2.1.1 Steady State Temperature Map
The mechanical structure to be analyzed is thought of as
divided into a number of elements or nodes, each represented
by a temperature. The net heat flow to node i from the sur-
rounding nodes j, plus the heat generated at node i, must
numerically equal zero. This is true for each node i, i going
from 1 to n, n being the number of unknown temperatures.
After each calculation of bearing generated heat, which
results from a solution of the shaft-bearing system portion
of the program, a set of system temperatures is determined
which satisfy the system of equations:
qi= qoi + q8 = 0 for all temperature
nodes i (2.1)
where qoi is the heat flow from all neighboring nodes to
node i
q8 i is the heat generated at node i. These values
may be input or calculated by the shaft bearing
program as bearing frictional heat
5
This scheme is solved with a modified Newton-Raphson
method which successfully terminates when either of two
conditions are met:
A ti
t EP2 for all nodes i (2.2)
t
i
where A t represents the Newton-Raphson correction to the
temperature t at a given iteration such that,
tN+l = tN + &t and N + 1, and N, refer to the
next and current iteration respectively.
EP2 is a user specified constant. If EP2 is left blank
or set to zero (0) a default value of 0.001 is used.
A second convergence criterion dependent upon EP2 is also
used. In the system of equations, q
0
+ qei = 0 for all nodes
i, absolute convergence would be obtained if the right hand
side (EQ) in fact reduced to zero (0). Usually a small residue
remains at each node, such that (qoi + qgi
)
= (EQ)i.
The second convergence criterion is satisfied if
n [(EQ) 2]
S
1
100 x EP2 (2.3)
where n = number of equations in bearing solution
2.1.2 Transient Temperatures
In the transient case, the net heat q. transferred to a
node i heats the element. It is thus necessary for heat
balance at node i that the following equations are satisfied.
dti
PCpi Vi dT = qi
(2.4)
6
where p = density
C = specific heat
= volume of the element
t = temperature
T = time
The temperatures, t o, at the time of initiation T - T
s
are
assumed to be knwon, that is
ti(T
s
) = toi i = 1, 2, ..., n (2.5)
The problem of calculating the transient temperature
distribution in a bearing arrangement thus becomes a problem
of solving a system of non-linear differential equations of
the first order with certain initial values given. The equa-
tions are non-linear since they contain terms of radiation
and free convection, which are non-linear with temperature as
will be shown later. The simplest and most economical way of
solving these equations is to calculate the rate of temperature
increase at the time T = Tk from equation 2.4 and then calcu-
late the temperatures at time Tk + AT from
tk+
1
= tk + dtk AT = tk + _7k AT (2.6)
aT eCp V
If the time step AT used as program input is chosen
too large, the temperatures will oscillate, and if it is chosen
too small the calcualtion will be costly. It is therefore
desirable to choose the largest possible time step that Coes
not give an oscillating solution. The program optionally
calculates such a time step. The step is obtained from the
condition, (16)
dti ,k-U
dtik - 0 i = 1, 2, ..., n (2.7)
If this derivative were negative, the implication would
be that the local temperature at node i has a negative effect
on its future value. This would be tantamount to asserting
that the hotter a region is now, the colder it will be after
an equal time interval. An oscillating soluticn would result.
7
Differentiating equation (2.6) for node i, one has as
condition (2.8),
dtik+l _ T
i
dq
i
dtik
- 1 + 1
- 0, i = 1, 2 .... n (2.8) dti,k + iiCpiVi dt
i
The derivative dqi/dt
i
is calculated numerically
dqi _ qi (ti + ti) - qi(ti) (2.9)
U[i a t.29
For each node, the value of AT. giving a value of zero
to the right hand side of Eqn. (2.8t is calculated.
A value of AT rounded off to one significant digit smaller
than the smallest of theAT
i
given by Eqn. (2.8) is used.
If the transient thermal scheme is being used interactively
with the bearing subprograms, the user must specify a small
enough time step between calls to the bearing subprograms in
order that the variation in bearing generated heats, with
time, accurately reflects the physical situation. At first,
a trial and error procedure will be required to effectively
use the program in its mode, however, experience will increase
the user's effectiveness.
2.1.3 Calculation of Heat Transfer Rate
The transfer of heat within a medium or between two media
can occur by conduction, convection, radiation and fluid flow.
All these types of heat transfer occur in a bearing appli-
cation as the following examples show.
1. Heat is transferred by conduction between inner ring
and shaft and between outer ring and housing.
2. Heat is transferred by convection between the surface
of the housing and the surrounding air.
3. Heat is transferred by radiation between the shaft
and the housing.
4. When the bearing is lubricated and cooled by cir-
culating oil, heat is transferred by fluid flow.
Therefore, in calculating the net flow to a node all the
above-mentioned modes of heat transfer will be considered.
2.1.3.1 Generated Heat
There may be a heat source at node i giving rise to a heat
flow to be added to the heat flowing from the neighboring nodes.
In the case that the heat source is a bearing, it may
either be considered to produce known amounts of power, in
which case constant numbers are entered as input to the program,
or the shaft-bearing program may be used to calculate the bearing
generated heat as a function of bearing temperatures.
2.1.3.2 Conduction
The heat flow qi %- which is transferred by conduction
from node i to node 3, 's proportional to the difference in
temperature (ti - t.) and the cross-sectional area A and is
inversely proportio~al to the distance t between the two points,
thus
_cij = A (ti_t )
(2.10)
where ? the thermal conductivity of the medium.
2.1.3.3. Free Convection
Between a solid medium such as a metallic body and a
liquid or gas, heat transfer is by free or forced convection.
Heat transfer by free convection is caused by the setting in
motion of the liquid or gas as a result of a change in density
arising from a temperature differential in the medium. With
free convection between a solid medium and air, the heat
energy q transferred between nodes i and j can be cal-
culatedqYr6g the equation, (2.11)
qvi,j
=
(vA Iti-t i d SIGN (ti-t
j
) (2.11)
where ov = the film coefficient of heat transfer by
free convection
A = the surface area of contact between the media
d = is an exponent, usually = 1.25, but any value
can be specified as input to the program

if
t.
a
t.
SIGN
f
-
if
tk:<
t:
9
The last factor is included to give the expression qvi,j
a correct sign.
The value of(V can be calculated for various cases, see
Jacob and Hawkins, (16).
2.1.3.4 Forced Convection
Heat transfer by forced convection takes place when liquid
or gas moves around a solid body, for example, when the liquid
is forced to flow by means of a pump or when the solid body is
moved through the liquid or gas. The heat flow qwi j transferred
by forced convection can be obtained from the following equation.
9wi,j
=
w a(ti - tj) (2.12)
wherecw is the film coefficient of heat transfer during
forced convection. This value is dependent on
the actual shape, the surface condition of the body,
the difference in speed, as well as the properties
of the liquid or gas.
In most cases, it is possible to calculate the coefficient
of forced convection from a general relationship of the form,
Nu = aRebprc (2.13)
where a, b, and c are constants obtained from handbooks,
such as (17). R and P are dimensionless numbers
defined
by
e
r
Nu = Nusselt's number = o. L/A
L= characteristic lengt
A = conductivity of the fluid
Re = Reynold's number = UL p/V
U = characteristic speed
p = density of the fluid
= dynamic viscosity of the fluid
Pr = Prandtl's number = qCp/A
C = specific heat
P
10
The program can use a value of the coefficient of convec-
tion, or let it vary with actual temperatures, the variation
being determined by how the viscosity varies. Input can be
given in one of four ways, for each coefficient.
Constant viscosity
1. Values of the parameters of equation (2.13) are
given as input and a constant value of o w is cal-
culated by the program.
Temperature dependent viscosity
2. The coefficient Ow for turbulent flow and heating
of petroleum oils is given by
= k
9
* {)(t)} klo
(2.14)
where k
9
and kl
0
are given as input together with
viscosity at two different temperatures.
3. Values of the parameters of equation (2.13) are given
as input. Viscosity is given at two different
temperatures.
2.1.3.5 Radiation
If two flat parallel, similar surfaces are placed close
together and have the same surface area A, the heat energy
transferred by radiation between nodes i and j representing
those bodies will be,
qRi,j CaAfti
+
273)4 _ (t. + 273]
4
(2.15)
where E is the surface emissivity. The value of the
coefficient E is an input variable and varies
between 1 for a completely black surface and 0
for an absolutely clean surface. In addition,
avis Stefan-Boltzmann's radiation constant
which has the value 5.76 x 10-8 watts/m
2
-(OK)
4
and t
i
and tj are the temperatures at points i
and j.
Heat transfer by radiation under other conditions can also
be calculated, (16). The following equation, for instance, applies
i11
between two concentric cylindrical surfaces
aAi [(t
i
+ 273)
4
- (tj + 273)4(
qRi,j = _________________ (2.16)
1 + (l-E) (Ai/Ae)
where A
i
is the area of the inner cylindrical surface
Ae is the area of the outer cylindrical surface
2.1.3.6 Fluid Flow
Between nodes established in fluids, heat is transferred
by transport of the fluid itself and the heat it contains.
' jq i '
Figure 2.1 Convective Heat Transfer
Figure 2.1 shows nodes i and j at the midpoints of consecu-
tive segments established in a stream of flowing fluid.
The heat flow quij through the boundary between nodes
i and j can be calcu a ed as the sum of the heat flow qfi through
the middle of the element i, and half the heat flow qoi trans-
ferred to node i by other means, such as convection.
The heat carried by mass flow is,
qfi = pi CPi Vi ti = Kiti (2.17)
where V. = the volume flow rate through node i
12
12
The heat input to node i is the sum of theheat generated
at node i (if any) and the sum over all other nodes of the
heat transferred to node i by conduction, radiation, free and
forced convection.
q oi= qG,i + (qci,j + qvi,j + qwi,j + qRi,j) (2.18)
The heat flow between the nodes of Fig. 2.2 is then,
qui,j = qfi + qoi/2 (2.19)
If the flow is dividing between node i and j, Figure 2-2
then the heat flow is calculated from
qui,j = Kij (qfi + qoi/2) (2.20)
where Kij = the proportion of the flow at i going to
the node j, o Kij 1. Kij is specified
at input.
k k eki
kki-
IiO ki
FIGURE 2.2 1
DIVIDED FLUID FLOW FROM NODE i
Figure 2.2 Divided Fluid Flow from Node i
2.1.3.7 Total Heat Transferred
The net heat flow rate to node i can be expressed as,
qi =q G,i + (qci,j + q ui,j + q vij + q wi,j + q Ri (2.21)
13
The summation should include all nodes j, both with un-
known temperatures
as well as boundary nodes, at which the
temperature is known so long as they have a direct heat exchange
with node i.
This expression is a non-linear function of temperatures
because of the terms qw and q . Therefore, the equations to be
solved for a steady state soldtion are non-linear. The sub-
program SOLVXX for solving non-linear simultaneous equations
is used for this purpose.
2.1.4 Conduction Through a Bearing
As described in Section 2.1.3.2, the conduction between
two nodes is governed by the thermal conductivity parameter A
of the medium through which conduction takes place. The
value of ?i is specified at input.
An exception is when one of the nodes represents a bearing
ring and the other a set of rolling elements. In this case,
the conduction
is separately calculated
using the principles
described below.
2.1.4.1 Thermal Resistance
It is assumed that the rolling speeds of the rolling
elements
are so high that the bulk temperature
of the rolling
elements
are the same at both the inner and outer races, except
in a volume close to the surface.
The resistance
to heat flow
can then be calculated
as the sum of the resistance
across
the surface
and the resistance
of the material
close
to the
surface.
The resistance
A is defined
implicitly
by
4t = A-q
(2.22)
where
&t is temperature
difference
q is heat flow
The resistance
due to conduction
through
the EHD film is
calculated
as
= (h/A) - A
(2. 24)
14
14
where h is taken to be the calculated plateau film
thickness
A is the Hertzian contact area at the specific
rolling element-ring contact under consideration.
Ais the conductivity of the oil.
The geometry is shown in Figure 2.3(a). Asperity con-
duction is not considered.
So far, a constant temperature difference between the
surfaces has been assumed. But during the time period of
contact, the difference will decrease because of the finite
thermal diffusivity of the material near the surface, Fig. 2.3(b).
To points at a distance from the surface this phenomenon
will have the same effect as an additional resistanceA
2
acting
in series with Ji"
This resistance was estimated in (18) as,
= 1 (_V
(2.24)
ALre,i
2biV
where Ir = contact length, or in the case of an
elliptical
contact area, 0.8 times the
major axis
A = heat conductivity
= thermal diffusivity p
= density
Cp = specific
heat
b = half the contact width
V = rolling speed
The resultant resistance
is
res 1 + 12
(2.25)
15
* - *, -.. .
AVERAGE
-
FILM
THlICKNESS -h
TEPIPERATURF ti l
(a) Schematic Concentrated Contact
i
LL
i
direction of rolling
(b) Temperature Distribution at Rolling,
Concentrated Contact Surfaces
FIGURE 2.3
CONTACT GEOMETRY AND TEMPERATURES
16
There is one such resistance at each rolling element.
They all act in parallel. The resultant resistance, tres'
is thus obtained from
n
1 r 1
1 - 1(2.26)
ares i=l firesi
2.2 Bearing Dimensional Change Analysis
The program calculates the changes in bearing diametral
clearances according to the analysis described originally in
(5) and herein in Appendix I 2, and expressed in generalized
equation form as,
ADCL = f ((Fits)m, t
i ,
jjm' (Qr)ml , m = 1, 2 for inner and (2.27)
outer rings
respectively
i = 1,2,3,4,5 for
shaft, inner ring,
outer ring, housing
and rolling element
respectively
17
where: ADCL is the change in bearing diametral clearance
Fits are the cold mounted shaft and housing fits
t
i
are the component temperatures
d
m
refers to the ring rotational speeds
r refers to the radial component of the minimum
rolling element-race normal force.
A bearing clearance change criterion is satisfied when
the change in bearing diametral clearance remains within a
narrow, user specified range, for two successive iterations
as follows:
I(ADCL)N - (ADCL)N-I 1 EPSFIT for all bearings (2.28)
D
where: N denotes the most recent iteration and
N-1 denotes the previous iteration
D denotes the ball or roller diameter and
EPSFIT is a user specified value = .0001D
It should be noted that although ring rotational speeds,
and initial, i.e. cold, shaft and housing fits are considered
in the clearance change analysis, these two factors are fixed
at input and remain constant through the entire solution.
Although component temperatures may change as a consequence of
the thermal solution, temperatures remain constant through a
complete set of clearance change iterations. As a result, only
the change in bearing load distribution affects the change in
bearing clearance within a set of clearance change iterations.
2.3 Bearing Inner Ring Equilibrium
The bearing inner ring equilibrium solution is obtained
by solving the system:
(Fb)i - (Fs)
i
= 0 for all bearings, i (2.29)
where: Fb denotes a vector of bearing loads and
moments resulting from rolling element/
race forces and moments.
18
Fbxi
Fbyi Forces
Fbzi
(Fb)i -
(2.30)
Mby
Mbyi Moments
Mbzi
If the bearing solution considers friction, Fb is com-
prised of the rolling-element race friction forces as well as
the normal forces.
If the bearing solution is, at the user's option, friction-
less, Fb is comprised only of rolling element/race normal con-
tact forces.
(Fs)
i
denotes a similar vector of loads, .xerted on the
inner ring by the shaft. The calculation of (Fs)i is presented
in Appendix I 3.
Fsxi
Fsyi Forces
F Fszi
(2.31)
(Fs)
-- --
Msyi
Moments
Mszi
The variables in this system of equations are the bearing
inner ring deflections db and the shaft displacements A at
all bearing locations. The bearing loads may be expressed as
a function of the inner ring deflections.
Fb = Fb (
6
'b) (2.32)
The dcflectionAb of a bearing is described by two radial
deflections
6 and 4, two angular deflections
e and e
z
and one
axial deflectyon6
x. The axial deflection
is asumed to be
the same for all bearings on a shaft, i.e.
19
The solution scheme is ended when
4 EPSI (frictionless)
6(&)ij < (2.37)
( )ij EPS2 (friction)
i = 1,... (Number of bearings)
j = 1,5 - for the 3 linear and two
angular deflections at
each bearing
If for some i or j, (
6
)ij = 0, Eq. (2.38) is used in place of
(2.37)
+
(EPSI (frictionless)
() ij . (2.38)
(0.001 x NBRG) LEPS2 (friction)
NBRG denotes the number of bearings in the system.
EPSI and EPS2 are Used depending on whether the bearing
solutions are frictionless or include friction, respectively. If
the bearing deflections are extremely small, computer-generated
numerical inaccuracies may prevent convergence according to
the above criteria although a perfectly good solution has
been obtained. To overcome this problem, the iteration is ter-
minated if all angular deflections are less than 2 x 10-6 radians
and all linear deflections are less than 5 x 10-8 inches. Any
one of the above criteria imply that inner ring equilibrium is
satisfied.
2.4 Bearing Quasi-Dynamic Solution
The bearing quasi-dynamic solution is obtained through
a two-step process:
1. Elastic Solution - considering rolling element
contrifugal force.
2. Elastic and Quasi-dynamic Solution*
*Quasi-dynamic equilibrium is used to connotc that the true
dynamic equilibrium terms containing first derivatives of the
ball rotational speed vectors and the second derivatives of
rolling element position vectors with respect to time are
replaced by numerical expressions which are position rather
than time dependent.
20
..- .
2.4.1 Rolling Element Equilibrium Equations
The equations which define rolling element quasi-dynamic
force equilibrium take the form
[f am . . * m = 1-3 refers to the
(Qm + fm)dt + Fm
1
+F=0 outer, inner and cage
m _am rolling element contacts (2.39)
respectively
where: Qm is the vector normal load per unit length
of the contact. See Appendix I 4.
f is the vector of friction force per unit length
of the contact. See Appendices I 5 and I 6.
F is the vector of inertia and drag forces.
See App. I 6 and I 7
t is a coordinate along the contact, perpen-
dicular to the direction of rolling (usually
the major axi,
a i3 half the contact length. See Ref. (1).
F
m
is the vector sum of the hydrodynamic forces
acting on the rolling element at the m-th
contact. See Appendix I 6.
Rolling element moment equilibrium is defined by:
[
m
x ( + f)dt + x M
=
0
(2.40)
-am
m
Qm,fmFmamp
and t are defined above, M is a vector
of inertia moments. For te definition of
MI, refer to Appendix I 7.
rm is a vector from the rolling element
center to the point of contact, see Appendix
I 4.
The solution to the equation sets represented (2.39) and 2.40)
generate the necessary data to calculate bearing fatigue life.
See Appendix I 9.
In the frictionless
elastic solution Fm and fm
=
0.
Additionally,
the only rolling element inertia term considered
in the frictionless
solution
is centrifugal
force. As a con-
sequence, only the axial and radial force equilibrium
equa-
tions are solved for each ball. For each roller, the radial
force equilibrium
and the tilting moment about the Z axis of
Fig. 2.4 is solved.
A dummy equation
for axial force equilibrium
is included
in the solution
matrix which keeps the roller centered
21
FIGURE 2.4
Bearing Inertial XYZ) and Rolling Element (xyz),
Coordinate
Systems
z
Al
x
CAGE POCKET CENTER
FOR BALL NO. I
CENTER BALL NO. I
22
with respect to the outer race. The cylindrical roller bearing
considered by the program cannot carry axial loading.
The friction solution determines ball quasi-dynamic equili-
brium for six degrees of freedom. A roller is permitted four
degrees of freedom. The rolling-element variables in this
solution are x,, Yl' U)x, IJy, W z, and W o.
where x
I
is the rolling element axial position relative to
the outer race. For a roller, this is a dummy variable.
Yl is the rolling element radial position relative to
the outer race,
(Jx,c.y, Lz are the orthogonal rolling element rotational speeds
relative to the cage speed, about the x, y and z
axes and W o is the rolling element orbital speed.
For the roller, Wz is a dummy variable.
The variables x
I
and y are the ball variables in the friction-
less solution. The variables in the rol ler frictionless solu-
tion are x
I
, a dummy, Y
1
and 0z = tan (L.Oy/Lhix).
2.4.2 Cage Eguilibrium Equations
The cage equations and cage-rolling element interactions
are not considered when the friction forces are omitted from
the rolling element equilibrium equations.
The number of degrees of freedom given to the cage is one,
if the cage will tend to rotate concentrically with respect to
the ring on which it is riding. This condition is determined
as a function of the rolling element orbital speed variation and
prevails with most roller bearings and with ball bearings
subjected only to axial loading. In both cases, orbital speed
variation is often inconsequential. Also, single degree of
freedom is allowed when the cage is rolling element riding.
The single degree of freedom corresponds to a small angular
rotation about the bearing axis, measured with respect to rolling
element 1. The angular displacement is converted to a linear
dimension by a multiplication by the bearing pitch diameter and
is noted in Fig. 2.4 as I . When a single degree of freedom
is permitted, the sum of moments acting on the cage about the
bearing x axis is required to be zero. This moment equation
considers the cage-rolling element normal and friction forces
as well as the torque generated at the cage-ring surface.
If there is significant rolling element orbital speed
variation, the cage is permitted to move to an eccentric posi-
tion with respect to the land on which it is piloted. Two
additional degrees of freedom are required to describe the
eccentric pcsition. These are the cage center of mass radial
displacement, e, and the angular displacement of the center of
mass, with respect to the bearing Y axis, 0 c', see Figures 2-5
and 2-6. These radial and angular displacement variables are
determined when the sum of forces acting on the cage, resolved
along the bearing Y and Z axes, reduce to zero. The rolling
element-cage normal and friction forces as well as the pressure
23
FIGURE 2-5
Inner Ring-Cage Land Contact Geometry
5,f
24
FIGURE
2-6
outer Ring-Cage
Land Contact
Geomuetry
if.
ec- e9
25
'.0 g.. ..
LAND
buildup between the cage and its piloting surface are considered
in the equilibrium equations. The effect of the cage mass is
neglected.
Figure 2-7 depicts the cage pocket normal and friction
forces acting on a rolling element which are considered in
the cage equibrium solution. These forces are functions of
the rolling speedswx andy and the contact geometry are cal-
culated in the x,y,z frame. The forces exerted on the cage
due to the i-th rolling element are, in the XYZ frame of
reference:
Mxi = - (Fyl + Fy
2
)ir + (P
1
- P2)i Rm
Fyi = - (Fy
I
- Fy
2
)
i
cos Oi - (PI - P2)i sin 5i (2.41)
Fzi = (Fy
I
- Fy
2
)
i
sin 1i + (P
1
- P2)i cos Oi
when the forces of Eq. (2.41) are summed over all of the rolling
elements, and the total added to the cage land contact forces,
the cage equilibrium equations for the three degree of freedom
model are obtained as:
M
x
= 0 = (Mxi) + McX
Fy = 0 = (Fyi) + Fcy
(2.42)
F
z
= 0 = (Fzi) + FcZ
where MXi' F and Fzi are defined for each rolling element
by Eq. (2.411, Mcx is the cage land friction torque, Fcy and
FcZ are the cage land hydrodynamic forces.
Within SHABERTH, Eq. (2.42) which defines cage equili-
brium, are solved simultaneously with the set of all ball or
roller quasidynamic equilibrium equations.
Details for calculating the rolling element/cage pocket
forces and the cage land/ring land forces are presented in
Appendix I 9.
The ball bearing friction solution is thus obtained by
solving 6Z+(l or 3) equations where Z is the number of rolling
elements. The ball bearing frictionless solution is obtained
by solving 1, (Z/2) (Z/2+l) or Z sets of 2 equations, depending
upon the number of rolling elements in the bearing and the
degree of load symmetry which prevails. The various symmetry
conditions are explained below.
The roller bearing friction solution contains 4Z+(l or 3)
equations and the frictionless
solution contains Z/2, Z/2+1
or Z sets of three equations again depending upon the number
of rolling elements and whether or not load symmetry exists.
26
FIGURE 2-7
Cage Pocket Normal and Friction Forces
Affecting Equilibrium
2
27
2.4.3 Load Symmetry Conditions
The various load symmetry conditions are as follows.
Axial symmetry is utilized if, for a ball bearing the load
is axial only, then only one set of two equations is solved for
the frictionless case. Six ball and one cage equilibrium equations
are solved when friction is included. All balls are assumed to
behave identically.
Radial load symmetry is utilized if the non-axial shaft
loading is comprised of only radial components parallel to the
Y axis and moment components parallel to the Z axis and the
position of the first rolling element is on the Y axis, then
symmetry exists, only half the rolling elements need be con-
side-ed if the number of rolling elements is even and one half
plus one need be considered if the number is odd. Because
of inertia terms, radial load symmetry can only be utilized in
the frictionless solution.
If load symmetry is not present, then Z sets of two (ball
bearing) or Z sets of three (roller bearing) equations must be
solved to obtain the frictionless solution.
2.4.4 Bearing Quasidynamic Solution Criteria
As with the steady state temperature mapping scheme, the
Newton-Raphson scheme in subprogram SOLVXX is used to solve
the sets of equations for each bearing. The iteration scheme
terminates when either:
J hXiN I ( [EPS1 frictionless (2.43)
XiN-I -- EPS2 friction
i=l..n
or
EPS1 frictionless
n 2
(2.44)
1 EQi EPS2 friction
n J
______ j_ S 100 X I
Experience has shown that the second criteria is usually
responsible for terminating the solution. However, when rolling
element loads are extremely large, on the order of 105 Newtons,
it becomes difficult to reduce the equation residues to less
than 10 Newtons. In those instances, the first criteria usually
terminates the iteration scheme.
28
3. PROGRAM INPUT
3.1 Types of Input Data
A complete set of input data comprises data of four
distinct categories. Within these categories, cards which
convey specific kinds of information are referred to as card
types. Depending on the complexity of the problem, the input
data set may contain none, one or several cards of a given type.
The categories are listed below.
I. Title Cards
A title card plus a second card which provides the
program control information for the shaft-bearing
solution.
II. Bearing Data Cards
A set of up to sixteen (16) card types, each set des-
cribing one bearing in the assembly. All bearings must
be so described. The card sets must be input sequentially
in order of increasing distance from a selected end of
the shaft.
III. Thermal Data Cards
A set of up to nine (9) card types to describe the
thermal model of the assembly.
IV. Shaft Data Cards
A set of three (3) card types to describe the shaft
geometry, bearing locations on the shaft and shaft
loading
If the program is being used to predict the performance
of a bearing assembly, cards from all four sets must be included
in the runstream. If the program is being used to thermally
model a mechanical system wherein no bearing heat generation
rates are required, and therefore, no bearing calculations
need be performed, the cards from sets II and IV are omitted.
The review of required input information which follows is
broken into the four sets of data categories given above, with
special emphasis on program control data.
The input data instructions are given in Appendix II 2,
and are for the nfost part self explanatory. They are laid out
in the format of an eighty column data card. A description
of the variables is given in the input instruction forms.
29
The units used for input data are as follows:
Linear Dimensions - (mm)
Angles - (degrees)
Surface Roughness - (microns)
Bearing Angular Mounting Errors - (radians)
Rotational Speeds - (RPM)
Force - (Newtons) (N)
Moments - (N-mm)
Pressure, Elastic Modulus - (N/mm
2
)
Density - (gm/cm3)
Kinematic Viscosity - (cs)
Temperature - (degrees centigrade) (C)
Coefficient of Thermal Expansion - (C-
1
)
Thermal Conductivity - (Watts/m/OC)
3.2 Data Set I - Title Cards
3.2.1 Title Card 1
This card should contain the computer run title and any
information which might prove useful for future identification.
The full eighty (80) columns are available for this purpose.
The title will appear at the top of each page of Program output.
3.2.2 Title Card 2
This card provides the control information for the shaft
bearing solution.
Item 1: Shaft Speed in rpm, GOV (1). All bearings have the
same shaft, i.e. inner ring speed.
Item 2: Number of Bearings on the shaft (NBRG), a minimum
of zero is permitted if no bearing solution is being sought.
A maximum of five is permitted.
Item 3: Print Flag (NPRINT), NPRINT equal to zero is normal
and will result in no intermediate or debug output. With a value
of one, a low level intermediate print is obtained at the end
of each shaft bearing iteration. The values of the variables,
the inner ring displacements (DEL), and the equation residues
are printed.
At the end of each bearing iteration, wherein the rolling
element and cage equilibrium equations are solved, an error
parameter is printed which has the value:
30
k-'
Error Parameter = XN/XNl
XN is the change in the variable X specified
at iteration N.
XNl is the value of the variable specified
at the previous iteration.
The Error Paremeter is calculated for each of the bearing
variables, but only the largest one is printed.
Additionally, at the end of each Clearance Change itera-
tion, the clearance change error parameter is printed. This
error is defined by Eq. 2.28.
If NPRINT is set at 2, all of the above information is
printed. Additionally, the variable values and residue values
are printed for each iteration of the rolling element and cage
equilibrium solution.
Item 4: ITFIT controls the number of iterations allowed
to satisfy the bearing clearance change iteration scheme. If
ITFIT is set to zero (0), or left blank, the clearance change
portion of the program is not executed. If a position integer
is input, the clearance change scheme is utilized with a maximum
iteration limit of five (5). If a negative interger is input,
the scheme is used with a maximum iteration limit equal to
the absolute value of the negative integer.
Item 5: ITMAIN limits the number of iterations attempted
during the solution of the shaft and bearing inner ring equili-
brium problems, i.e., establishing the equilibrium of bearing
reactions and applied shaft loads. If ITMAIN is left blank,
set to zero, or to a positive integer, then (15) iterations
are premitted. If ITMAIN is set to a negative integer, the
number of iterations is limited to the absolute value of that
integer.
31
.... .. .. ..'l - ..... .... .. . . .. . . . . . . ... .. ... ... ..... .. .. .
Item 6: GOV(2) or EPSFIT is the convergence criterion
for the diametral clearance change portion of the analysis.
As mentioned under item 3 above, this error parameter is
defined by Eq. 2.28.
The iteration scheme is terminated when the error para-
meter is less than the input value of EPSFIT. If EPSFIT is
left balnk or is set to zero (0), the program default value
of 0.0001 is used.
Items 7 & 8: Main loop accuracy for frictionless elastic
(EPSl) and friction solution (EPS2). These accuracy values
control the accuracy of the shaft bearing deflection solution
as well as the quasi-dynamic solution of the component dynamics
(cf. Section 3). If EPSI and EPS2 are left blank or set to
zero (0), default values of 0.001 and 0.0001 respectively are
used.
Item 9: IMT, if set to 1, the Material properties for
both bearing rings and the rolling elements are to be input
on card types B 11 through B 19. If IMT is zero or blank,
the rings and rolling elements are assumed to be steel.
Card types B 11 through B 14 are required only if the change
in bearing diametral clearance is to be calculated.
Item 10: NPASS controls the level of the bearing solution
0 Elastic Contact Forces are calculated. No lubrication
or friction effects are considered.
1 Elastic Contact Forces are calculated. Lubrication
and friction effects are considered using raceway control
(ball bearing) or epicyclic (roller bearing) assumptions
to estimate rolling element and cage speeds.
2 Inner Equilibrium is satisfied considering only the
Elastic Contact Forces. Using the inner ring positions
thus obtained, rolling element and cage equilibrium
are determined considering friction.
3 Complete Solution. The inner ring, rolling element
and cage equilibrium is determined considering all
elastic and friction forces.
3.3 Data Set II - Bearing Data
Most of the input instructions are self-explanatory.
Where certain items are deemed to require more explanation
than given in the input data format instructions, they are
treated on an individual basis by card type and item number.
Most of the bearing input data is read into a two dimen-
sional array named "BD," which has the dimensions (1830, 5).
For each of the five bearings permitted on a shaft, a total
of 1830 pieces of data may be stored. Denoting BD(I,J), I
represents a specific piece of bearing data, J represents
the bearing number. The bearing input data of Data Set II
occupies the first 106 locations of the 1830 allocated. On
the input data format sheets, the designation BD(I) where
I-i...106, denotes the location within the BD array where
each piece of input data is stored.
3.3.1 Card Type 1 - Bearing Type and Material Designations
Item 1: Bearing type, columns 1-10 must be specified,
left justified, i.e., "B" or "C" in column 1. This format
must be followed since the Program recognition of bearing
type, (ball bearing or cylindrical roller bearing), is derived
from reading the "B" or "C" in the first column of this card.
Item 2 & 3: Columns 11-30 and 31-50, "Steel designations,"
inner and outer rings, respectively. The alphameric-literaI
description of the steel types such as "M-50" or "AISI 52100"
is input.
Items 4 & 5: Columns 51-60 and 61-70, the numbers input
for items 4 (inner ring) and 5 (outer ring) are used to account
for improved materials and multiply the raceway fatigue lies
as determined by Lundberg-Palmgren methods. Typical life
factor values for modern steels are in the neighborhood of 2.0
to 3.0. If the ASME Publication Life Adjustment Factors for
Ball and Roller Bearings, is referenced by the user, the Material
Factor D and the Material Process Factor E should be used
multiplicatively as inputs for items 4 and 5. Additionally, if
the user is accustomed to using a lubricant life multiplier
he must also multiply the material factor by the lubricant life
multiplier. The program considers EHD film thickness and RMS
surface roughness but generates a life multiplier having
a maximum value of 1 and a minimum of 0.479, i.e. Lube-Life Factor
Programmed only serves to reduce predicted Fatigue Life.
Item 6: Columns 71-78, "Orientation angle of the first
rolling element." (01) (degrees). Refer to Fig. (2.4). The
quasi-dynamic rolling element bearing problem has an infinite
number of solutions which fall within a narrow envelope having
a periodic shape. The solution obtained is a function of
the rolling elem6nt positions relative to the bearing system
coordinate axes. 01 = 0, places a rolling element on the Y
axis and is the choise customarily made. 01 can be desig-
33
360
nated as any value 0O 360-where Z is the number of rolling
elements. For each different value assigned to 01, a differ-
ent, although similar, bearing solution will be obtained. To
take advantage of bearing symmetry and the computer time savings
which result, 01 must be specified as zero or left blank.
Item 7: Column 80, a signal, termed the crown drop flag,
which specifies for a cylindrical roller bearing, whether the
roller-race crown drops will be calculated, or read directly.
If item 7 is blank or zero, the crown drops are calculated
based on the roller-race crown radius, and effective flat
length input information. If the crown drop flag is other than
zero or balnk, the non-uniform separation of the roller and
raceway must be specified at the center of each slice into
which the roller-raceway effective contact length is divided.
The slice widths are identical. The number of slices is
input as item 7 card type B4. The non-uniform roller-raceway
separation is input on card types B5 and B6.
3.3.2 Card Type B2 - Bearing Geometry and Outer Ring Speed
Item 1, 2 and 5 need no explanation, however, items 3
and 4 require substantial explanation as they apply to the
various types of ball bearings.
3.3.2.1 Ball Bearing Geometry
Through the proper specification of the diametral clearance
and contact angle, the Program can properly handle deep grove,
split inner, and angular contact ball bearings.
The deep groove ball bearing requires the specification
of zero contact angle and either the operating diametral clear-
ance Pd or the off-the-shelf diametral clearance, if the dimen-
sional change analysis is utilized.
The angular contact bearing is fully described through
specification of the contact angle which obtains under a gauge,
axial load. However, this method of input does not accurately
define the system if there is more than one angular contact
supporting the shaft and at least one of those bearings has
its grooves offset in the direction opposite to the other
bearings and if the shaft is capable of axial and/or radial
play. In other words, if what are known as angular contact
ball bearings are mounted such that some diametral shaft
play is permitted, an auxilliary angle as well as the diametral
play must be specified at input. The angle input is not the
34
manufacturer's designated contact angle,oe , but an auxilliary
angle,c( 0, the calculation for which shall be demonstrated.
Refer to Figure 3.1. The manufacturer's contact angle is
calculated as follows:
= cos-l [ ] (3.1)
A =r
0
ri - D (3.2)
where: ro and r. are the outer and inner raceway groove
1 radii respectively
D is the ball diameter
Under a gauge axial loada< is obtained at both inner and
outer raceways for each ball. Under this condition, the outer
and inner raceways are axially offset an amount s,
sa = A Sino (3.3)
When angular contact ball bearings are mounted with some
diametral play, the grooves are offset an amount Sx
o
such that
Sxo< So . The diametral play which obtains at this condition
is Sd. This diametral play is usually known by the engineer
or designer and is usually required to allow some forgiveness
when thermal gradients are encountered. Assuming that the user
has the values foro , r
o ,
ri, D and So< then:
o

tan
-1
So
where: Pd and A may be calculated from Eqs. (3.1) and (3.2).
The manufacturer might be able to provide the value of
Sac at the Sd value of interest. If not, the following
equation may be solved foro.
35
-- '.-
~0
0J 4 M
u 0
44-
S C3
4
36
Sd -APd0
2-- C 2 A - 2 o- 0 (3.5)
Note: Sat
o
should be less than Sa and o
0
should be less thano..
Equation (3.6) is derived by developing two expressions
for the radial separation (A
r
) of the outer and inner raceway
curvature centers.
Ar = A - Pd
Ar = (A - Sdcos<o) Cos
O
(3.6)
2
Pd _ o 2
A - A Cost
o
- 5 CX
T 2 0
In order that the Program properly handle split inner
ring ball bearings an auxilliary angle and diametral play must
be input. Referring to Figure 3.2, the auxilliary angleao_ and
diametral play Sd must be determined and input. Typically,
the values of D, ro,o s and Sd' are known. Pd and Sd may be
calculated as follows:
Pd = Sd' + (2ri-D) (l-Coss) (3.7)
Sd = [Pd - 2A (l-Cosoc
0
) /Cos
2
O
0
(3.8)
The unloaded half of the inner ring must be removed from con-
sideration and the ball moved such that its center lies on the
line connecting the origins of r
i
and r and positioned such
that the auxilliary clearance Sd 4 exists at both the inner
and outer raceways. The auxilliary angle is
= Tan
1
(r
i
- D/2) Sina
s
t
(3.9)
r
o
D/2 - Sd'/2 + (r
i
- D/2) Cos.
s
37
,-.,,,,d
1-4r
4
1
038
The angle associated with each ball bearing must be
specified with the correct sign. A positive contact angle
allows the bearing to accept a positively directed axial
load transmitted by the shaft.
3.3.3 Card Type B3 - Rolling Element Geometry
These data are self explanatory. Although space has
been set aside for the specification of roller end radius and
roller included angle, this has been done for future considera-
tions and are not used by the program. The items may be left
blank.
3.3.4 Card Type B4 - Raceway and Roller Raceway Contact Geometry
3.3.4.1 Ball Bearing
Items one and two refer to the outer and inner raceway
curvatures respectively where curvature is defined as the
cross groove radius divided by the ball diameter. Typical
values range from 0.515 to 0.57.
3.3.4.2 Roller Bearing Contact Geometry Data
All items are used to define the roller-race contact
geometry, see Fig. 3.3 "Flat length" and "Crown Radius" are
used to calculate roller-race separation along the roller
profile if this information is not specifically input. See
Item 7 of the Bearing Data title card and Bearing Data Cards
B5 and B6.
Items 1 and 4 "Effective Contact Length" refer to the
longest possible length which can obtain at a roller-race
contact. Typically, this is the roller total length less the
corner radii. If, however, the raceway undercuts are excep-
tionally large so that the tract width is smaller than the
effective roller length then the tract width should be input.
Item 7 refers to the number of slices into which the
roller raceway contact may be divided. A maximum value of
twenty is permitted a default value of two is used if Item 7
is blank or zero. Each slice is the same width.
3.3.5 Roller-Raceway Non-Uniform Profile Definition
3.3.5.1 Card Type B5 - Inner Roller Raceway Contact
These cards are used to input the inner and outer race
roller-race separation along the roller profile. With the
high points of the roller and race in contact, i.e. with all
clearance between roller and raceway removed. These cards
must be omitted if item 7 of the Bearing Data Title card is
zero or blank.
39
got-
V)
0
0 u ____
-r44
Cd 0
u u
"-4
-4
0
40
3.3.6 Card Type B6 - Outer Roller Raceway Contact
Same as Card Type B5.
3.3.7 Card Type B7 - Raceway-Rolling Element Surface Data
Items 1 through 6 define the statistical surface micro-
geometry parameters of the rollers and raceways. Items 1 through
3 require the input of center line average CLA surface rough-
ness. Within the program CLA values are converted to RMS by
multiplying by 0.9.
Items 4 through 6 are RMS values of the slopes measured
in degrees of the surface asperities as measured in a traverse
across the groove for rings, longitudinally for rollers and
in any arbitrary direction for balls. Typical values for
raceway and rolling element surfaces are 1 to 2 degrees.
This card is omitted if item 10 of title card 2 is zero or
blank.
3.3.8 Card Type B8 - Cage Data
This card is omitted if item 10 of the title card 2 is
zero or blank. These data are self explanatory. Note that the
cage weight is an input item. The weight, however, is not
used in any calculation. It is included only for future con-
sideration of cage stability predictions.
3.3.9 Card Type B9 - Shaft and Housing Fit Dimensions
These cards are to be included only if the change in
bearing diametral clearance with operating conditions is to
be calculated, i.e. if item 4 ITFIT on the Bearing Title Card 2
is non-zero. On Card Type B9, tight interference fits bear a
positive sign and loose fits, a negative sign.
Items 3 and 6 on Card No. 9 are termed the shaft and
housing effective widths, respectively. The value specified
for these effective widths may be as great as twice the ring
width.
Use of an effective width is an attempt to account for
the greater radial rigidity of a shaft than the ring that is
pressed on to it, owing to the fact that the shaft deflects
over a distance that extends beyond the ring width. In the
proaram, the calculated internal pressure on the ring due to
41
its interference fit with the shaft, is distributed over the
shaft effective width and this (lower) pressure is used in
computing the shaft deflection. Using double the actual width
as the effective width is customary.
3.3.10 Card Type B10 - Shaft Housing Fit Dimensions
These items are self explanatory, and are used to describe
equivalent ring sections, see Fig. I 2.1.
3.3.11 Card Type Bil
This card defires the elastic modulus for the shaft, inner
ring, rolling element, outer ring, and housing, respectively.
This card is to be included only if the change in bearing
diametral clearance with operating conditions is to be calcu-
lated, i.e., if item 4 ITFIT on the Bearing Title Card is non-
zero.
3.3.12 Card Type B12
This card defines the Poisson's ratio for the shaft, inner
ring, rolling element, outer ring, and housing, respectively.
This card is to be included only if the change in bearing
diametral clearance with operating conditions is to be calcu-
lated, i.e., if item 4 ITFIT on the Bearing Title Card is non-
zero.
3.3.13 Card Type B13
This card defines the density for the shaft, inner ring,
rolling element, outer ring, and housing, respectively. This
card is to be included only if the change in bearing diametral
clearance with operating conditions is to be calculated, i.e.,
if item 4 ITFIT on the Bearing Title Card is non-zero.
3.3.14 Card Type B14
This card defines the coefficient of thermal expansion
for the shaft, inner ring, rolling element, outer ring, and
housing, respectively. This card is to be included only if
the change in bearing diametral clearance with operating con-
ditions is to be calculated, i.e., if item 4 ITFIT on the Bearing
Title Card is non-zero.
3.3.15 Card Type B15 - Lubrication and Friction Data
This card is omitted if item 10 of title card 2 is zero
or blank.
42
~IL
Items 1 and 2
These items are the amounts by which the combined thickness
of C'e lubricant film on the rolling track and rolling element is
increased during the time interval between the passage of successive
rolling elements, from whatever replenishment mechanisms are
operative. Items 1 applies to the outer and Item 2 to the inner
race-rolling contacts, respectively. If Item 1 is zero or
blank, the mode of friction is assumed to be dry.
At the present time, the magnitude of the inner and
outer raceway replenishment layers has not been correlated to
lubricant flow rate, lubricant application methods and bearing
size and speed factors. At this point then, the user is
forced to establish proper values for the replenishment layer
thickness. As a rough guide, the following suggestions are made.
1. To avoid starvation, the replenishment layer thick-
nesses should be one or two times the EHD film thick-
ness which develops in the rolling element raceway
contacts.
2. Because of centrifugal force effects, intuition suggests
that the outer raceway replenishment layer should be
several times thicker than that prescribed at the inner
raceway.
Item 3, XCAV, describes the percentage of the bearing
cavity, estimated by the user to be occupied by the lubricant.
OSXCAV6I00.
As with the replenishment layer thicknesses, the amount
of free lubricant should be able to be correlated with lubricant
flow rate, lubricant application methods and bearing size and
speed factors. At this time, such correlations do not exist.
XCAV values of approximately one percent are recommended at
this point.
Item 4 is the coefficient of coulomb friction applicable
for the contact of asperities. If Items 1 and 2 are zero,
then Item 4 serves as the coulomb friction coefficient which
prevails in all contacts.
3.3.16 Card Type B16
This card is omitted if Item 10 title card 2 is zero or
blank or if Item 1 card 15 is zero or balnk which implies dry
friction.
This card specifies the lubricant type. If Item 1, NCODE
is 1, 2, 3, or 4, the Program uses preprogrammed lubricant
properties as presented in Table 1, and no further information
is required.
43
kL .... .. .- , . - ," .
PLLA
[: ~ fn~ CD -WN
E-i 0 '-O
H
E-4 0N -4L r4
E-4 Z -:
0
U)~L
E-4 E- 0-m
H DP r'0 :! O Ln r-4
u)E ULA t 0 ) O ON (Nq 0
U Z .0".',
H [W L 0o 0 0 ; r4
Dz 0 44H
'E) [: -4cOD
E-4 0
3
E1 u >4
H E-4 U
E-4 HU 0
0
o ~0 NO
41 [U .0 C) N0 LA
4 ZU N r- H4 %D(4 N C14
E-1
0 %D
co m
H N 4
H4 >44
.1-1 H 1.4-) H
44
NCODE Lubricant
1 A specific mineral oil
2 A MIL-L-7808G
3 C-Ether
4 A MIL-L-23699
NCODE may also be specified as negative (-1 to -4), in
which case, the traction characteristics of the respective
lubricant NCODE noted above are used but the actual properties
specified by Items 2 through 7 override the hard coded data.
This option is most useful in specifying various mineral oils
i.e. NCODE = -1.
3.4 Data Set III - Thermal Model Data
Appendix I 1 has been included to aid the user in data
preparation and calculation of heat transfer coefficients
required at input.
3.4.1 Card Type 1
Card type 1 is a control card. If no temperature map
is to be calculated, this card is to be included as a blank
card followed by a Type 2 card for each bearing on the shaft.
Card Type 1 contains control input for both steady state and
transient thermal analyses. It is not intended, however,
that both analyses be executed with the same run.
Item 1: The highest node number (M). The temperature
nodes must be numbered consecutively from one (1) to the
highest node number. The highest node number must not exceed
one hundred (100).
Item 2: Node Number of the Highest Unknown Temperature
Node (N). This number should equal the total number of
unknown node temperatures. It is required that all nodes
with unknown temperatures be assigned the lowest node numbers.
The nodes which have known temperatures are assigned the
highest numbers.
Item 3: Common Initial Temperature (TEMP)C: The tempera-
ture soution iteration scheme requires a starting point, i.e.,
guesses of the equilibrium temperatures. Card Type 3 allows
the user to input guesses of individual node temperatures.
When a node is not given a specific initial temperature, the
temperature specified as Item 3 of Card Type 1 is assigned.
Item 4: Punch Flag (IPUNCH): If the Punch Flag is not
zero (0) or blank, the system equilibrium temperature- along
with the respective node numbers will be punched accordinq
to the format of Card T3. This option is useful if, for
instance, the user makes a steady state run with lubrication,
and then wishes to use the resultant temperature as the initia-
tion point for a transient dry friction run in order to assess
the consequence of lubricant flow termination.
45
Item 5: "Output Flag" (IUB). If the "Output Flag" is
not zero, the bearing program output and a temperature map
will be printed after each call to the shaft bearing solution
scheme. This printout will allow the user to observe the flow
of the solution and to note the interactive effects of system
temperatures and bearing heat generation rates. Since the
temperature solution is not mathematically coupled to the bearing
solution, the possibility exists that the solution may diverge
or oscillate. In such a case, study of the intermediate output
produced by the "Output Flag" option may provide the user with
better initial temperature guesses that will effect a steady
state solution.
Item 6: "Maximum Number of Calls to the Shaft Bearing
Program" (IT1). IT1 is the limit on the number of Thermal-
Shaft-Bearing iterations, i.e., the external temperature
equilibrium calculation. The user must input a non-zero
integer such as 5 or 10 in order for the Program to iterate
to an equilibrium condition. If ITI is left blank or set to
zero (0) or 1, shaft bearing performance will be based on the
initially guessed temperatures of the system. The tempera-
tures printed out will be based on the bearing generated
heats. It is unlikely that an acceptable equilibrium condition
will be achieved. However, the temperatures which result may
provide better initial guesses, for a subsequent run, than
those specified by the user.
IT1 also serves as a limit on the transient temperature
solution scheme, by limiting the number of times the shaft-
bearing solution scheme is called. Each call to the shaft-
bearing scheme will input a new set of bearing heats to the
transient temperature scheme until a steady state condition
is approached or until the transient solution time up limit
is reached.
Item 7: "Absolute Accuracy of Temperatures for the
External Thermal Solution" (EPl). In the steady state thermal
solution scheme, each calculation of system temperatures occurs
after a call to the shaft-bearing scheme which produces bearing
generated heats. After the system temperatures have been
calculated for each iteration, using the internal temperature
solution scheme, each node temperature is checked against the
nodal temperature at the previous iteration.
If [t(Nli - t(N-Il< EPI for all nodes i then equilibrium
has been achieved and he iteration process stops.
46
Item 8: "Iteration Limit for the Internal Thermal Solution"
(IT2). After each call to the shaft-bearing program, the
internal temperature calculation scheme is used to determine
the steady state equilibrium temperatures based on the calculated
set of bearing heat generation rates. If the program is used
to calculate the temperature distribution of a non bearing
system, it is the internal temperature scheme which is employed.
If IT2 is left blank or set to zero, the number of internal
iterations is limited to twenty (20).
Item 9: "Accuracy for Internal Thermal Solution" (EP2).
The use of EP2 is explained in Section 2.1.1. If EP2 is left
blank or set to zero (9), a default value of 0.001 is used.
Item 10: "Starting Time" (START) is a time T
s
at which
the transient solution begins; usually set to zero (0).
Item 11: "Stopping Time" (STOP) is the time in seconds at
which the transient solution terminates, Tf. The transient
solution will generate a history of the system performance
which will encompass a total elapsed time of
(Tf - T
s
) seconds
Item 12: "Calculation Time Step" (STEPIN). The transient
internal solution scheme solves the system of equations
tk+l = tk + 1k 6T
PCpV
(3.10)
T = STEPIN
The user may specify STEPIN. If left blank or set to zero (0),
the Program calculates an appropriate value for STEPIN using
the procedure described in Section 2.1.2.
Item 13: "Time Interval Between Printed Temperature Maps"
(TTIME) seconds. The user must specify the length of time
which will elapse between each printing of the temperature
map. The interval will always be at least as large as the
"calculation
timestep"
(STEPIN).
Item 14: "Time interval Between Calls of the Shaft Bearing
Portion of the Program" (BTIME). BTIME will always have a value
larger than or equal to (STEPIN) even if the user inadvertently
inputs a shorter interval. Computational time savings result
if BTIME is greater than STEPIN, however, accuracy might be lost.
47
3.4.2 Card Type 2
Card Type T2 is required, one card for each bearing if no
thermal analysis is being performed. The temperature data is
used within the shaft-bearing analysis portion of the program
to fix temperature dependent properties of the lubricant in
which case the inner race, outer race and lubricant bulk
cavity temperatures are used. The assembly component tempera-
tures at each bearing location are used in the analysis which
calculates the change in bearing diametral clearance from
"off the shelf" to operating conditions.
Item 9: "Flange" temperature is not currently used in the
analysis. It simply provides for future consideration of
tapered roller bearings.
3.4.3 Card Type 3
In the steady state analysis this card is used to input
initial guesses of individual nodal temperatures for unknown
nodes as well as the constant temperatures for known nodes,
such as ambient air and/or an oil sump.
In the transient analysis, Card Type T3 is used to input
the nodal teymperatures of all nodes at (START) = T
s
i.e.,
at the initia.tion of the transient solution.
3.4.4 Card Type 4
With this card, node numbers are assigned to the components
of each bearing, one card per bearing. With this information, the
proper system temperatures are carried into each respective
bearing analysis. The inner race and inner ring node numbers
may or may not be the same at the user's discretion. Similarly,
the outer race and outer ring node numbers may or may not be
the same.
3.4.5 Card Type 5
The shaft bearing system analysis accounts for frictional
heat generated at four locations in the bearing, i.e., at the
inner race, the outer race, between the cage rail and ring
land, and in the bulk lubricant due to drag. The heat genera-
ted at the hydrodynamic cage-rolling element contact is added
to the bulk lubricant. Heat generated at the flange is not
presently considered. This card allows the heat generated to
be distributed equally to two nodes. For instance, the heat
generated at the inner race-rolling element contact should be
48
distributed half to the rolling element and half to the inner
race. The heat developed between the cage and inner ring
land may be distributed half to the inner ring and half to
the cage if a cage node has been defined, otherwise, half to
the bulk lubricant.
3.4.6 Card Type 6
This card specifies the node numbers and the heat genera-
tion rate for those nodes where heat is generated at a constant
rate such as at rubbing seals or gear contacts.
3.4.7 Card Type 7
This card type is used to input the numerical values
of the various heat transfer coefficients which appear in the
equations for heat transfer by conductivity, free convection,
forced convection, radiation and fluid flow. Up to ten
coefficients of each type may be used. Separate values of
each type of coefficient are assigned an index number via
card T7 and in describing heat flow paths (Card Type T8 below)
it is necessary only to list the index number by which heat
transfers between node pairs.
Incides 1-10 are reserved for the conduction coefficient
, 11-20 for the free convection parameters, 21-30 for forced
convection, 31-40 for emissivity and 41-50 for fluid flow
(product of specific heat, density and volume flow rate).
As an example, for heat transfer by conduction with
coefficient A of 53.7 watts/MOC one could prepare a card type
T7 with the digit 1 punched in column 10 and the value 53.7
punched in the field corresponding to card columns 11-20.
If a conduction coefficient of 46.7 were applicable for
certain other nodes in the system, one could punch an additional
card assigning index No. 2 to the valueA = 46.7 by punching
a "2" in card column 10 and 46.7 anywhere within card columns
11-20.
Rather than inputting constant forced convection coefficients,
optionally, these coefficients can be calculated by the program
in oen of three ways. If the calculation option is exercised
a pair of cards is used in place of a single card containing
a fixed value ofo . The contents of the pair of cards depends
upon which of the three optional methods are used.
Option 1) is independent of temperature but is calculated
as a function of the Nusselt number which in
turn is a function of the Reynolds number Re,
the Prandtl number Pr as follows, (cf. l7j)
49
= oil/LNu (3.11)
N
u
= aRbpc (3.12) ue r
where Aoilis the lubricant conductivity, L is a
characteristic length (with a unit of meters)
and K, a, b and c are constants.
Option 2) 0 is a function only of fluid dynamic viscosity
and viscosity is temperature dependent.
= c
d
(3.13)
where c and d are constants
Option 3) Mis a function of the Nusselt, Reynolds and
Prandtl numbers, and viscosity is temperature
dependent.
3.4.8 Card Type 8
Thic card defines the heat flow paths between pairs of
nodes. Every node must be connected to at least one other
node, i.e., two or more independent node systems may not be
solved with a single Program execution.
The calculation of heat transfer areas is based on lengths,
L
1
and L
2
input using Card Type T8. Additionally, the
type of surface for which the area is being calculated is
indicated by the sign assigned to the heat transfer coefficient
index. If the surface is cylindrical or circular, the index
should be positive, if the surface is rectangular the index
should be input as a negative integer.
In the case of radiation between concentric axially
symmetric bodies, L3 is the radius of the larger body. For
radiation between two parallel flat surfaces or for conduction
between nodes, L
3
is the distance between them.
Fluid flow heat transfer accounts for the energy which
the fluid transports across a node boundary. Along a fluid
node at which convection is taking place, the temperature
varies. The nodaa temperature which is output is the average
of the fluid temperature at the output and input boundaries.
If the emerging temperature of the fluid is of interest, it
is necessary to have a fluid node at the fluid outlet. At
this axiliary node, only fluid flow heat transfer occurs and
the fluid temperature would be constant throughout the node.
Thus, the true fluid outlet temperature will be obtained.
50
Conduction of heat through a bearing is controlled by
index 51. The actual heat transfer coefficient which contains
a conductivity, area and a path length term is calculated in
the bearing portion of the program. The term is based upon
an average outer race and inner race rolling element contact.
3.4.9 Card Type 9
This card inputs data required to calculate the heat
capacity of each node in the system. This card type is required
only for a transient analysis.
3.5 Data Set IV - Shaft Input Data
The Shaft-bearing analysis requires all loading to be
applied to the shaft. The loads applied to each bearing are
a product of the shaft-bearing solution. There is no need
for the user to solve the statically determinant or indeter-
minant system for bearing loads. Even if a single bearing is
being analyzed, with the applied load acting through the center
of the bearing, data for a dummy shaft must be supplied.
In the analysis, the housing is assumed to be rigid.
Provision has been allowed to input data for housing radial
and angular spring characteristics. However, this has
been done for future consideration of an elastic housing and
is therefore currently unavailable.
The shaft input data consists of three card types:
1. Shaft Geometry and Elastic Modulas Data
2. Bearing Position and Mounting Error Data
3. Shaft Load Data
3.5.1 Card Type 1
This card type is used to describe shaft geometry at up
to twenty locations along the shaft. The user must place his
shaft in a cartesian coordinate system with the end of the
shaft at the origin and with the shaft lying along the X-axis.
The shaft may have stepwise and linear diameter variations.
The stepwise variations require a single card which specifies
different diameters immediately to the left and right of the
relevant X shaft coordinate. The shaft analysis assumes a
linear diameter variation if on two successive cards, i.e.
two successive X coordinates, the diameters to the right of the
location differ from the diameters to the left of the location
51
I
of the following card. Complex shaft geometries may be
approximated with a set of linear diameter variations spaced
at close intervals.
If an Elastic Modulus is not specified at the designated
input location, the modulus of steel is assumed, 204083 N/mm2.
3.5.2 Card Type 2
This card type locates the bearing inner ring on the
shaft in the X-Y and X-Z planes. For a ball bearing, the
X coordinate specified locates the inner ring center of
curvature. For cylindrical roller bearings, the X coordinate
locates the center of the inner race roller path.
In addition to specifying bearing location, the Type 2
card is also used to specify housing radial and angular
mounting errors. As mentioned previously, space has been
reserved for inputting housing radial and angular spring
characteristics, however, these characteristics are not used
in the system analysis.
Two sets of Type 2 cards may be required. The first set
is always required and defines housing alignment errors in the
shaft X-Y plane. The second set defines the housing align-
ment errors in the shaft X-Z plane and is required only if
non zero errors exist for the particular bearing in question.
The first set of Type 2 cards must contain a card for
each bearing.
3.5.3 Card Type 3
Type 3 cards are used to specify shaft loading at a
given X cooidinate. Loading may be applied in the x-y and
x-z planes, thus requiring two distinct sets of Type 3
cards. Applied loads may have the form of concentrated radial
forces, concentrated moments, linearly distiibuted radial
forces and concentrated axial loads which may be eccentrically
applied. If an axial load is eccentrically applied, the moment
which results must not be separately calculated and input as
a concentrated moment.
Variations in distributed radial loads are handled at
input just as shaft linear diameter variations are handled.
Note that each set of Type 3 cards must be followed by
a blank card.
52
Also, note that in order for symmetry conditions
(see section 2.4.2) to be considered the second type 3
card must be void of any loading data.
53
4. COMPUTER PROGRAM OUTPUT
4.1 Introduction
The Program Output is intended to provide the engineer or
designer with a complete picture of the shaft-bearing system
performance.
In addition to the calculated output data, the input data
is listed, thus producing a complete record of the computer
run.
A sample set of program output is included for reference
as Appendix II 3 and represents an NPASS=2 solution for a two
bearing system comprised of a 209 size cylindrical roller
bearing, a single 110 mm bore angular contact ball bearing
operating at 10,000 rpm under a thrust load of 2,000 lbs.
(8,896 Newtons) with MIL-L-23699 lubricant and a 6220 size
split inner ring angular contact ball bearing operating at
15000 RPM under shaft loads of 8896 Newtons axial, 2248 Newtons
radial and a moment load of 4000 Newton millimeters. The
bearings are lubricated with an MIL-L-7808G lubricant.
The first seven pages of output essentially consists of
a summary of the input data categorized into bearing, cage,
steel, lubricant, fit temperature and shaft geometry and
loading data.
For four specific lubricants, see Table 1, the relevant
lubricant data has been coded into the Program. In this case,
the lubricant input information consists only of a single number
which designates the particular lubricant but the relevant
information for the lubricant is printed in the input data list.
Except as just noted, the actual results of calculations
are printed under the headings "Bearing Output" and "Rolling
Element Output."
Key output items are discussed briefly below.
54
4.2 Bearing Output
4.2.1.1 Linear and Angular Deflections
These deflections refer to the bearing inner ring relative
to the outer ring and are defined in the inertial coordinate
system of Figure 2.4. The bearing deflections are not necessarily
equal to the shaft displacements since the bearing outer ring
radial or angular mounting errors may be specified as non-zero
input.
4.2.1.2 Reaction Forces and Moments
These values reflect bearing reactions to shaft applied
loading and outer ring mounting errors.
When the bearing inner ring has achieved an equilibrium
position, the summation of all bearing reaction loads should
numerically equal the shaft applied loading. When the level
of solution indicated by "NPASS" = 2 is employed, as discussed
in Section 5, differences between shaft applied and bearing
reaction loads will exist but will typically be less than 10%.
This difference is a consequence of friction forces con-
tributing to the reaction loads whereas the inner ring equilibrium
position has been determined considering elastic contact forces
only.
4.2.2 Fatigue Life Data
The L
1 0
fatigue life of the outer and inner raceways as
well as the bearings are presented. The bearing life represents
the statistical combination of the two raceway lives. These
lives reflect the combined effects of the lubricant film thick-
ness and material life factors. The lubricant film thickness
life factor is described in detail in Section 3.
4.2.2.1 h/sigma
The ratio h/q, also referred to asA , is printed for the
most heavily loaded rolling element. The variable h, represents
the EHD plateau film thickness with thermal and starvation
effects considered. The variableW, represents the composite
root mean square surface roughness of the rolling element and
the relevant raceway.
55
4.2.2.2 Life Multipliers
4.2.2.2.1 Lubrication - This life multiplier is a function of
h/1 at each concentrated contact and is in the form of a
derating factor. Its value ranges from 0.479 for h/v- = 0 to 1.0
at h/q Z4. Since the lubricant life multiplier is decremental
the normal multiple of 3 used for thick film lubrication must
be multiplied by the material life factor normally used and
this product should be specified at input. This subject is
covered in more detail in Section 3.3.1.
4.2.2.2.2 Material - This output simply reflects the input
value. Again ,it is covered in Section 3.3.1.
4.2.3 Temperatures Relevant to Bearing Performance
These temperatures flilly describe the temperature condi-
tions which affect the performance of a given bearing. If one
of the temperature mapping options is used, the temperatures
printed reflect the results of the particular option. If,
neither temperature option was used, the list is simply a repeat
of the input data. Note that there are separate temperatures
for outer and inner raceway and ring temperatures. The raceway
temperature is used to determine lubricant properties. The
ring temperatures are used in the bearing dimension change
analysis. The raceway and ring temperatures may be the same
value.
4.2.4 Frictional Heat Generation Rate and Bearing Friction Torque
4.2.4.1 Frictional Heat Generation Rate
The various sources of frictional heat generated within
the bearings are listed. The values printed for "OUTER RACE,
OUTER RINGS, INNER RACE, INNER RINGS, R.E. DRAG AND R.E.-CAGE"
represent the sum of the generated heats for all rolling ele-
ments. Additionally, the heats printed for the outer and inner
raceways plus the rolling element-cage, reflect the friction
developed outside the concentrated contacts, i.e., the HD friction
as well as the EHD friction developed within the concentrated
contacts. The raceway data also include any heat generated
as a consequence of asperity contacts. "R.E. DRAG" should be
interpreted as the heat resulting from lubricant churning as
the rolling elements plow through the air-oil mixture. Items
2 and 4 relevant to rolling element-flange contacts are present
for future program expansion.
56
4.2.4.2 Torque
The torque value is calculated as a function of the total
generated neat and the sum of the inner and outer ring rotational
speeds. The intent is to present a realistic value of the torque
required to drive the bearing. Under conditions of inner ring
rotation, the torque value reflects the torque required to drive
the inner ring. The inner ring torque includes that fraction
of torque required to impart an angular velocity to the lubri-
cant in the bearing. A considerable portion of the lubricant
will come to rest within the housing and not at the outer ring.
Thus, the measured outer ring torque may not equal the torque
at the inner ring.
4.2.5 EHD Film and Heat Transfer Data
4.2.5.1 EHD Film Thickness
These values refer to the calculated EHD plateau film
thickness at both contacts of the most heavily loaded rolling
element and include the effects of the thermal and starvation
reduction factors.
4.2.5.2 Starvation Reduction Factor
These factors give for the inner and outer ring contacts,
the reduction in EHD film thickness ascribable to lubricant
film starvation according to the methods of Chiu, (11).
These factors pertain to the EHD film thickness for both
the inner and outer race contacts of the most heavily loaded
rolling elements, but are applied to the respective inner and
outer race film thickness for each rolling element in the bear-
ing.
4.2.5.3 Thermal Reduction Factor
These factors are calculated according to the methods of
Cheng, (10) and pertain to the EHD film thickness for both
the inner and outer race contacts of the most heavily loaded
rolling elements, but are applied to the respective inner and
outer race film thickness for each rolling element in the
bearing.
4.2.5.4 Meniscus Distance
These factors are calculated according to the methods of
Chiu, (11) and pertain to the EHD film thickness for both
the inner and outer race contacts of the most heavily loaded
57
rolling elements, but are applied to the respective inner and
outer race film thickness for each rolling element in the
bearing.
4.2.5.5. Raceway-Rolling Element Conductivity
These data reflect the amount of heat transfer between
rolling element and raceway for each degree centigrade differ-
ence between the two components. These data reflect the average
of all outer and inner contacts, respectively.
4.2.6 Fit and Dimensional Change Data
4.2.6.1 Fit Pressures
These data refer to the pressures built up as a conse-
quence of interference fits between shaft and inner ring and
housing and outer ring. Pressures are presented both for the
standard cold-static condition (16
0
C) and at operating
conditions.
4.2.6.2 Speed Giving Zero Fit Pressure (Between the Shaft
and Inner Ring)
This is a calculated value based upon operating conditions
and provides a measure of the adequacy of the initial shaft
fit.
4.2.6.3 Clearances
"Original" refers to cold unmounted clearance which is
specified at input if the diametral clearance change analysis
is executed. "Change" refers to the change in diametral
clearance at operating conditions relative to the cold unmounted
condition. A minus sign indicates a decrease in clearance.
"Operating" refers to the clearance at operating conditions.
For all types of ball bearings, the decrease in clearnace can
be combined with the initial diametral clearance, and the free
operating contact angle at operating conditions may be
calculated. Note that the change in clearance should be com-
pared against the diametral play of the split inner ring ball
bearing in order to determine if the possibility for three
point contact exists. The Program does not account for three
point contacts even though the change in clearance might suggest
that three point contact is obtained.
58
4.2.7 Lubricant Temperatures and Physical Properties
The lubricant properties, particularly the dynamic vis-
cosity and to a lesser degree, the pressure viscosity coefficient,
are heavily temperature dependent. These factors enter the EHD
film thickness calculation and the HD and EHD friction models.
The lubricant is assumed to be at the same temperature as the
relevant raceway. As noted elsewhere, these temperatures may
be either input directly or calculated by the Program.
The physical properties printed are self-explanatory.
The units are enumerated.
4.2.8 Cage Data
4.2.8.1 Cage-Land Interface
The cage data indicates the performance parameters at the
interface between the cage rail and the ring land on which the
cage is guided. The torque, heat rate and separating force
require no explanation. The eccentricity ratio defines the
degree to which the cage approaches the ring on which it is
guided at the point of nearest approach. The radial displace-
ment of the cage relative to the bearing axis is divided by one
half the cage-land diametral clearnace. An eccentricity ratio
of one indicates cage-land contact. A ratio of zero indicates
that the cage rotation is concentric with the bearing axis.
Only the cage-land and rolling element pocket forces are
considered in determining the cage eccentricity. The cage
weight and centrifugal force which result from the eccentricity,
although available, are not considered in the analysis. The
omission of these considerations helps reduce convergence
problems.
4.2.8.2 Cage Speed Data
Cage speed data presents the comparison between the cage
speed calculated based upon the quasidynamic equilibrium con-
siderations and the speeds calculated with raceway control
theory for ball bearings and the epicyclic speeds of the
roller bearing components.
4.3 R olling Element Output
4.3.1 Rolling Element Kinematics
59
4.3.1.1 Rolling Element Speeds
All of the rolling element speeds tend to vary from
position to position when the bearing is subjected to
combined loading.
The total rolling element speed is with reference to the
cage and represents the vector sum of the three orthogonal
components.
4.3.1.2 Speed Vector Angles
The rolling element speed vector angles, Arctan (4.,y/wx)
and Arctan (4z/wx) are presented in order to show a clearer
picture of the predicted ball kinematics. The ball speed
vector tends to become parallel with the bearing X axis with
increasing shaft speed and decreasing contact friction.
4.3.2 Rolling Element Raceway Loading
4.3.2.1 Normal Forces
The norMal forces acting on each rolling element are
printed. The rolling element race normal forces are self-
explanatory. The cage force is calculated only when the
friction solution is employed and is always directed along the
rolling element Z axis. If the rolling element orbital speed
is positive, a positive cage force indicates the the cage is
pushing the rolling element, tending to accelerate it. Cage
force is a function of rolling element position within the
cage pocket. Its magnitude is derived using hydrodynamic lubri-
cation assumptions, when the distance between the rolling
element and cage web is large, and EHD assumptions when the
separation is of the order of the EHD film thickness or when
rolling element-cage web interference exists.
4.3.2.2 Hertz Stress
The stress printed represents the maximum normal stress
at the center of each ball race contact or at the most heavily
loaded slice of the roller raceway contact.
4.3.2.3 Load Ratio Qasp/Qtot
If the EHD film thickness is small compared to the RMS
composite rolling element-race surface roughness, the rolling
element-race normal load will be shared by the EHD film and
asperity contacts. The load ratio reflects the portion of the
total load carried by the asperities.
60
4.3.2.4 Contact Angles
A ball bearing, subject to axial loading, misalignment or
mounted such that the inner ring is always displaced axially
relative to the outer rings, (i.e., a duplex set of angluar
contact ball bearings) will have non-zero contact angles.
At low ball orbital speeds, the inner and outer race angles
will be substantially the same. At high speeds, ball centrifugal
force will cause the outer race contact angle to be less
than the inner race angle.
4.4 Thermal Data
As in the case for bearing output, all of the input data is
printed. The calculated output data is presented in the form
of a temperature map in which a node number and the respective
node temperature appear. The appearance of the steady state
and transient temperature maps are identical. The transient
temperature map also includes the time (T) at which the
temperature calculations were made.
4.5 Shaft Data
These data simply reflect the input information.
4.6 SHABERLH Error Messages
4.6.1 Introduction
For various reasons, SHABERTH execution may terminate
before the desired results have been achieved. This section
is intended to give insight to the user as to the nature of
the problem which caused termination.
In some instances, error messages are printed and exe-
cution proceeds. These messages indicate that in one of the
internal iterative loops, a particular solution failed to
converge to the desired accuracy. These messages should be
taken as caution signals to the user to check the results
carefully. In particular, compare the calculated bearing reaction
forces against the applied shaft loading. If these results check
tc within 10 percent with an NPASS = 2 solution and to within
percent with an NPASS = 3 solution, the solutions should be
sifficiently accurate.
Additional means of evaluating solution accuracy are
,.,nted in section 4.6.10.
61
4.6.2 From ALLT - Message: "STEADY STATE SOLUTION WITH CEPi)
DEGREES ACCURACY WAS NOT OBTAINED AFTER CITI) ITERATIONS."
Explanation: This message pertains to the thermal equi-
librium solution in which bearing generated heat and
system temperatures must be consistent.
4.6.3 From AXLBOJ - Message: "ERROR MESSAGE, KX = (IER)
SINGULAR SET OF SHAFT EQUATIONS." Explanation: This
message indicates an error in the input data which
describes the shaft.
4.6.4 From DAMPCO - Message: "TIlE NUMBER OF EQUATIONS CALCU-
LATED BY SUMMING THE NUMBER OF EQUATIONS IN EACH SUBSET
IS (NTOT). THIS DOES NOT EQUAL THE TOTAL NUMBER OF
EQUATIONS SPECIFIED (N) AN ERROR EXISTS AND EXECUTION
TERMINATES." Explanation: If the nonlinear equations
are comprised of M independent subsets, then N must
equal the summation of NSIZE(K) = 1,M.
K=M
NTOT = E NSIZE(K) (4.1)
K=l
In SHABERTH M is always 1 and NSIZE(l) is N. This
message should thus never be written.
4.6.5 From EHDSKF - Message: "AN IMPROPER LUBRICANT TYPE CODE
HAS BEEN PASSED TO EHD SKF. EXECUTION TERMINATES."
Explanation: Coming into EHD SKF N must have an integer
value 1, 2, 3 or 4, a test has shown that l>N>4.
4.6.6 From INTFIT - Message: "SINGULAR MATRIX ON TIGHT SHAFT
FIT." Message: "SINGULAR MATRIX ON LOOSE SHAFT FIT."
Explanation: These messages indicate bad data entering
INTFIT.
4.6.7 From SHABE - Message: "AFTER (ITFIT) ITERATIONS, ERRMAX =
(ERRMAX) WHILE THE REQUIRED FIT ACCURACY WAS (ERFIT)."
Explanation: The bearing diametral clearance change
analysis did not converge in ITFIT iterations. Either
increase the number of iterations or set the number to
-2 for a good approximate solution.
4.6.8 From SONRI - Message: "SINGULAR SET OF EQUATIONS, NPASS =
(NPASS)." Explanation: This message pertains to shaft
equilibrium solution and has never been known to occur.
4.6.3.1 From SONRI - Message: "THE RELATIVE ACCURACY EPS HAS
NOT BEEN OBTAINED AFTER IT ITERATIONS IN ROUTINE SONRI."
Explanation: This message indicates that shaft bearing
inner ring equilibrium has not been achieved within the
specified number of iterations. This may occur under
light loading.
62
mili,
I
4.6.9 From STARFC - Message: "***IN STARFC/ROOTI*** ROOT
OF F(X) DOES NOT EXIST BELOW HO." Explanation: The
iterative solution for the meniscus distance for a
rolling element raceway contact has not converged.
This occurs only when the specified replenishment layers
are extremely thin.
4.6.10 From SOLVXX
The majority of the error messages printed by SHABERTH
will be printed front SOLVXX, indicating that SOLVXX has been
unable to fully solve a particular set of nonlinear equations.
Within SHABERTH failure has never occurred during the solution
of a set of steady state thermal equations. Failure does occur,
however, in the solution of the bearing quasidynamic set of
rolling element and cage equations. The major portion of this
section should be read with this in mind.
4.6.10.1 Message: "ONLY (NDER) EQUATIONS WERE FOUND TO VARY
WITH X(J), ND(J) WERE EXPECTED." THE DIFFERENCES
EQ (X+DX) - EQ(X) + DIFF(I)" IF DIFF(I) IS LESS THAN
SF8*EQ(I), THEN EQ(I) IS NOT CONSIDERED TO VARY WITH
X SF8 = (SF8)." "FOR THE FOLLOWING EQUATIONS THE
DIFFERENCES ARE BIG ENOUGH" CC(N*J-N+I), I = 1, COUNT.
Explanation: The matrix of partial derivatives cal-
culated within SHABERTH must have at least N, nonzero
diagonal elements. It is possible to specify a mini-
mum number of nonzero elements greater than one, for
each variable with the use of the array ND. If that
minimum number is not obtained, the above set of
error messages is printed. In SHABERTH, only the
diagonal elements are required.
4.6.10.2 Message: " ** ERROR MESSAGE FROM THE EQUATION
SOLVING ROUTINE AT ITERATION LOOP (LOOP) ****."
One of the four following situations has occurred.
Situations 1. and 2. have never been known to occur;
3. and 4., however, have, and are explained.
1. "SINGULAR SET OF EQUATIONS," IER = 1.
2. "DIVERGENCE HAS OCCURRED 10 CONSECUTIVE ITERA-
TIONS," IER = 2.
3. "THE LIMIT FOR NUMBER OF ITERATIONS IS REACHED,"
IER = 3.
This message (No. 3) indicates that the solution accuracy is not Ps
good as desired. To achieve the required accuracy, the problem
might be rerun with the iteration limit increased. This is accom-
plished at solution level NPASS = 2, through changing the "20" in
CALL BEAR statement in "SHABE" from 20 to a larger number. At
solution level NPASS = 3, the "30" in the CALL BEARC statement in
SONRI must be increased to a larger number.
Prior to making these Program changes, however, the magni-
tudes of the equation residues should be examined in the same
manner a suggfsted above. The solution may be sufficiently good
as discussed in 4 below, so that further computations are unnecessary.
63
4. "THIS IS THE BEST WE CAN DO. IT MAY BE USABLE."
IER = 4.
This message indicates that computation has stopped before
the desired results have been obtained. Fifty attempts have been
made to increment the variable values without finding a set of
increments which would reduce the equation residues.
After failing at these numerous attempts to improve the
solution, it is concluded that the best solution has been
achieved and that further changes to the variables will serve
only to increase the equation residues. (In this discussion
equation values and residue values are synonomous.) It is believed
that this situation arises when large changes in variable values
introduce small changes in equation values, i.e. when the force
versus variable function has a very shallow slope.
When the differences in equation values are of the same
order of magnitude as the numerical accuracy of the particular
computer being used, these kinds of convergence problems can be
expected.
It is possible that when this "BEST WE CAN DO" message is
printed, that even though the solution is not as accurate as
desired, it may be sufficiently accurate to be usable. The accur-
acy can be assessed by comparing the magnitude of the equation
residue to the magnitude of the individual terms which comprise
the equation. Since most convergence problems arise in the
quasi-dynamic equilibrium solution of the rolling element and
cage equations, the method of assessing the accuracy of this set
of equations shall be addressed.
As noted earlier, the set of equations used to define the
quasidynamic problem is comprised of (6Z + Mcage) equations where
Z is the number of rolling elements and Mcage is the number of
degrees of freedom (1 or 3) assigned to the cage. To assess the
accuracy of the solution, the magnitude of the equation residue
should be compared to the magnitude of the components which make
up the equation. The residue values are printed under the heading
"CORRESPONDING EQ-VALUES."
'
The residues are printed in a six per
line format such that the residues pertaining to a given element
are all on one line. These six values represent the following
six equilibrium equations:
1) ZF
x
= 0 2) EFy = 0 3) ZF
z
= 0
4)
x
= 0 5) EM= 0 6)
Z
= 0
Equations 1 and 2 are dominated by the normal raceway contact
forces and the rolling element centrifugal force. Compared to
the magnitude of these forces the residues of EF, and EFy are
usually very small and thus acceptable. The remaining four
equilibrium equations, however, have as their major terms, various
components of the friction forces which act upon the element. Large
values of these residue values are a manifestation of an unstable
64
operating condition, where the instability is defined qualita-
tively in terms of large change in component and raceway relative
speeds producing small changes in internal bearing forces. The
shallow slopes of these "Force vs. Relative Speed Functions"
promote dynamic and numerical instabilities.
The EFZ equation should be examined for each rolling
element. Typically this may be done by inspection with close
comparison and calculations made for only one or two elements.
The three cage equilibrium equations are:
1) zMx = 0 2) EFy = 0 3) EFz = 0
If only one cage equation is considered, it is equation 1); if
more than one is considered, all three are used.
The components of these equations are the rolling element
cage normal and friction forces as well as the cage-ring normal
and friction forces. Magnitudes of the components of these
forces and moments are printed as part of the output. Thus,
comparison of components against residues is straight forward
after converting to a consistent set of units. Residues are in
English while the output is in metric units.
It should be noted that in those solutions in which the
cage has only one degree of freedom, that the cage interaction
with the rolling elements has only minor impact upon the rolling
element dynamics. Therefore, a relatively large residue for
the cage equation is not terribly significant. The rolling
element ZFz equations should be the basis for the judgement as
to whether a solution is good enough.
Although the message "THIS IS THE BEST WE CAN DO. IT MAY BE
USABLE" may be written during a steady state temperature calcu-
lation scheme, numerical instabilities in those schemes are rather
uncommon. None have been experienced with this program after
three years of operation.
The following data are printed subsequent to printing messages
1 through 4 above.
RELATIVE ACCURACY (ERREL) ITERATION.
LIMIT (ITEND), NUMBER OF UNKNOWNS (N) ABSOLUTE ACCURACIES EXA(J).
DAMPING FACTORS 1-5 OTHER STEP FACTORS 6-10 (SF), MAXIMUM STEP
FACTORS (SMX(J), J=I,N).
CORRECTIONS OF THE X-ES FROM SMQ D(J), J=I,N.
NUMBER OF DERIVATIVES EXPECTED FOR EACH X ND(J), J=I,N
X-VALUE X X(J), J=l,N.
CORRESPONDING EQ-VALUES EQ(J), J-I,N.
65
5. GUIDES TO PROGRAM USE
The Computer Program is a tool. As with any tool, the
results obtained are at least partially dependent upon the
skill of the user. Certainly, the economics of the Program
usage are highly dependent upon the user's technical need
and discriminate use of Program options.
Some general guides for efficient use of the Program are
listed below:
1. Attempt to use the lowest level of solution possible.
For instance, if the prime object of a given run is to
obtain bearing fatigue lives, execute only the elastic
solution (NPASS = 0). If an estimate of bearing
frictional heat is required, execute the low level
friction evaluation (NPASS = 1). Execute the friction
solution (NPASS = 2) only if rolling element and cage
kinematics are of interest. Execute the highest level
of solution (NPASS = 3) if kinematics are of interest
and the bearing reaction loads deviate substantially
from the shaft applied loading, i.e., a deviation
greater than ten percent.
2. Attempt to input bearing operating diametral clearance
rather than calculate it. Or, execute the diametral
clearance change analysis once for a group of similar
runs and use the output from the first run as input
to the subsequent runs omitting the clearance change
analysis.
3. Attempt to input accurate operating temperatures
rather than calculate them.
4. The more non-linear the problem, the more computer time
required to solve it. In the bearing friction solution,
large coefficients of friction seem to increase
the degree of non-linearity. In the thermal solutions,
if possible, eliminate non-linearities by omitting
radiation terms and by using constant rather than
temperature dependent free and forced convection
coefficients.
5. In the transient thermal solution, space the calls to
the shaft-bearing solution (BTIME) to as large an
interval as prudently possible. Be careful, however,
too long an interval will produce large errors in heat
rate predictions.
66
6. In the steady state thermal analysis, attempt to
estimate nodal temperatures on a node-by-node
basis. Nodes which are heat sources should have
higher temperatures than the surrounding nodes.
The above suggestions are intended to encourage the use
of the Program on a cost-effective basis. The intent is not
to discourage the use of important program capabilities, but
to emphasize how the program should be most effectively used.
It is suggested that the user take a simple, axially
loaded ball bearing problem and execute the program through
the full range of options beginning with a frictionless solu-
tion proceeding to the three levels of friction solution with
a low (0.01) and high (0.1) friction coefficient. The dia-
metral clearance change analysis and the thermal solutions
should also be executed on an experimental basis. This exer-
cise will provide the user with some insight into economics of
the Program usage on his computer as well as the results obtained
from various levels of solution of the same problem.
It is also suggested that a constant user of the program
should study the hierarchical Program flow chart, Appendix II 1,
along withe the Program listing to gain an appreciation of the
program complexity and the flow of the problem solution. The
Program is comprised of many small functional subroutines.
Knowledge of these small elements may allow the user to more
easily piece together the philosophy of the total problem
solution.
SHABERTH is intended to be used for the analysis of a
multi-bearing system. It may, however, be used to analyze
single bearings, mounted on dummy shafts under certain conditions
of limited applied loading. These loading conditions are
outlined and explained below as they apply separately for
ball and cylindrical roller bearings.
Ball Bearings
Of major value is the capability of SHABERTH to treat, in
a simple, economic manner, ball bearings subjected to axial
loading only. Use of the program in this manner is recommended.
SHABERTH is not recommended for the solution of a single
ball bearing subjected to radial load only. The Program
attempts to satisfy axial and angular equilibrium as well as
radial. A ball bearing is elastically very soft in those
directions which causes mathematical instabilities during the
solution scheme. This makes the Program uneconomical for this
particular situation.
67
Whereas the use of SI.ABFRTII to solve single radially loaded
ball bearing problems is not recommended, because of economics,
the solution of a single, radially and axially loaded problem
is impossible. The impossibility arises because a moment reac-
tion will develop when a ball bearing is subjected to both radial
and axial loading. In order for the user to solve this problem
he must specify at input the bearing reaction moment. The user
must know the answer to a portion of the problem before he can
begin to solve it.
Cylindrical Roller Bearing
A single, cylindrical roller bearing may be subjected to
radial loading or combinations of radial and moment loading.
When SHABERTH is used in this manner, it is important that
bearing misalignments be specified indirectly through specifi-
cation of a non-zero applied moment. If a radial load and an
initial outer ring misalignment are specified along with a
zero applied amount, SHABERTH will attempt but will be unable
to solve the problem since it will be impossible to equilibrate
the non-zero reaction moment, resulting from the offset, against
the zero applied moment.
The cylindrical roller bearing cannot accept applied
axial loading and thus when a single cylindrical roller bear-
ing is being examined, the applied axial load must be speci-
fied to be zero.
When a single cylindrical roller bearing is being examined
all loading should be referred to the X-Y plane in order to
take advantage of the symmetry of load distribution among
rolling elements.
68
LIST OF REFERENCES
1. Kellstrom, M.. "A Computer Program for Elastic and Thermal
Analysis of Shaft Bearing Systems," S K F Report No.
AL74PO14, submitted to Vulnerability Laboratory, U. S.
Army Ballistic Research Laboratories, Aberdeen Proving
Ground, MD, under Army Contract DAADO5-73-C-0011.
2. Harris, T., "An Analytical Method to Predict Skidding in
High Speed Roller Bearing," ASLE Transactions, 9, 1966,
pp. 229-441.
3. Harris, T. A., "An Analytical Method to Predict Skidding
in Thrust Loaded Angular Contact Ball Bearings," Journal
of Lubrication Technology, Trans. ASME, Series F, Vol. 93,
No. 1, 1971, pp. 17-24.
4. Timoshenko, Strength of Materials Part II Advanced Theory
and Problems, 3rd Edition, D. VanNostrand Co., Inc., 1958.
5. Crecelius, W. J. and Milke, D., "Dynamic and Thermal Analysis
of High Speed Tapered Roller Bearings Under Combined Loading,"
Technical Report NASA CR 121207.
6. Harris, T. A., "How to Predict Bearing Temperature Increases
in Rolling Bearings," Product Engineering, pp. 89-98,
9th December 1963.
7. Fernlund, I. and Andreason, S., "Bearing Temperatures Cal-
culated by Computer," The Ball Bearing Journal No. 156,
March 1969.
8. Andreason, S., "Computer Calculation of Transient Tempera-
tures," The Ball Bearing Journal, No. 160, March 1970.
9. McCool, J. I., et al, "Technical Report AFAPL-TR-75-25,
"Influence of Elastohydrodynamic Lubrication on the Life
and Operation of Turbine Engine Ball Bearings - Bearing
Design Manual," S K F Report No. AL75P014 submitted to
AFAPL and NAPTC under AF Contract No. F33615-72-C-1467,
Navy MIPR No. M62376-3-000007, May 1975.
10. Cheng, H. S., "Calculation of EHD Film Thickness in High
Speed Rolling and Sliding Contacts," MTI Report 67-TR-24
(1967).
11. Chiu, Y. P., "An Analysis and Prediction of Lubricant
Film Starvation in Rolling Contact Systems," ASLE Trans-
actions, 17, pp 23-35 (1974).
69
12. Tallian, T. E., "The Theory of Partial Elastohydrodynamic
Contacts," Wear, 21, pp 49-101 (1972).
13. McGrew, J. M., et al, "Elastohydrodynamic Lubrication Pre-
liminary Design Manual," Technical Report AFAPL-TR-70-27,
pp 20-21, November 1970.
14. Fresco, G. P., et al, "Measurement and Prediction of Vis-
cosity-Pressure Characteristics of Liquids," A Thesis in
Chemical Engineering Report No. PRL-3-66, Department of
Chemical Engineering College of Engineering, The Penn-
sylvania State University, University Park, Pennsylvania.
15. Crecelius, W. J., Heller, S., and Chiu, Y. P., "Improved
Flexible Shaft-Bearing Thermal Analysis with NASA Friction
Models and Cage Effects," S K F Report No. AL76P003,
February 1976.
16. Jakob, M. and Hawkins, G. A., "Elements of Heat Transfer,"
3rd Ed., John Wiley & Sons, Inc., 1957.
17. Kent's Mechanical Engineering Handbook-Power Volume, John
Wiley & Sons, Inc., 12th Ed., 1960, Chapter 3, p. 20.
18. Burton, R. A., and Staph, H. E., "Thermally Activated
Seizure of Angular Contact Bearings," ASLE Trans. 10,
pp. 408-417 (1967).
19. Liu, J. Y., Tallian, T. E., McCool, J. I., "Dependence
of Bearing Fatigue Life on Film Thickness to Surface
Roughness Ratio," ASLE Preprint 74AM-7B-I (1974).
20. Archard, J. and Cowking, E., "Elastohydrodynamic Lubrication
at Point Contact," Proc. Inst. Mech. Eng., Vol. 180, Part
3B, 1965-1966, pp. 47-56.
21. Dowson, D. and Higginson, G., "Theory of Roller Bearing
Lubrication and Deformation," Proc. Inst. Mech. Eng.,
London, Vol. 177, 1963, pp. 58-69.
22. Loewenthal, S. H., Parker, R. J., and Zaretsky, E. V.,
"Correlation of Elastohydrodynamic Film Thickness Measure-
ments for Fluorocarbon, Type II Ester and Polyphenyl Ether
Lubricants," NASA Technical Note D-7825, National Aero-
nautics and Space Administration, Washington, D.C.,
November 19, 1974.
70
23. McCool, J. I., et al, "Interim Technical Report on Influence
of Elastohydrodynamic Lubrication on the Life and Operation
of Turbine Engine Ball Bearings," S K F Report No. AL73PO14,
submitted to AFAPL and NAPTC under AF Contract No. F33615-
72-C-1467, Navy MIPR No. M62376-3-000007, October, 1973.
24. Allen, C. W., Townsend, D. P., and Zaretsky, E. V., "New
Generalized Rheological Model for Lubrication of a Ball
Spinning in a Nonconforming Groove," NASA Technical Note
D-7280, National Aeronautics and Space Administration,
Washington, D.C., May 1973.
25. Lundberg, G. "Cylinder Compressed Between Two Plane Bodies,"
SKF Internal Report 1949-08-02.
26. Nayak, P. R., "Random Process Model of Rough Surfaces,"
Journal of Lubrication Technology, 93, pp. 398-407 (1971).
27. McCool, J. I., et al, "Interim Technical Report on Influence
of Elastohydrodynamic Lubrication on the Life and Operation
of Turbine Engine Ball Bearings," SKF Report No. AL73P014,
submitted to AFAPL and NAPTC under AF Contract No. F33615-
72-C-1467, Navy, MIPR No. M62376-3-000007, October 1973.
28. Johnson, K. L., and Cameron, R., "Shear Behavior of EHD Oil
Film at High Rolling Contact Pressure," The Institution of
Mech. Engr. Tribology Group, Westminster, London WEI (1968).
29. Smith, R., et al, "Research on Elastohydrodynamic Lubrication
of High Speed Rolling-Sliding Contacts, Air Force Aero Pro-
pulsion Laboratory, Wright Patterson AFB, Ohio Technical
Report AFADC-TR-71-54 (1971).
30. Floberg, L., "Lubrication of Two Cylindrical Surfaces Consid-
ering Cavitation," Report No. 14 Chalmers University of Tech-
nology, Gothenburg, (1966).
31. Snidle, R. W. and Archard, J. K., "Lubrication at Elliptical
Contacts," Symposium on Tribology Convention, Gothenburg
Institute of Mechanical Engineering, 1969.
32. Pinkus, 0., and Sternlicht, B., "Theory of Hydrodynamic Lubri-
cation," McGraw-Hill Book Company, Inc., New York, N.Y. 1961
p. 48.
33. Lundberg, G.' and Palmgren, A., "Dynamic Capacity of Rolling
Bearings" Acta Polytechnica, Mechanical Engineering Series 1.
Proceedings of the Royal Swedish Academy of Engineering,
Vol. 7, No. 3, 1947.
34. Lundberg, G. and Palmgren, A. "Dynamic Capacity of Roller
Bearings" Proceedings of the Royal Swedish Academy of
Engineering Vol. 2, No. 4, 1952.
71
35. Knudsen, J. G., and Katz, D. L., Fluid Dynamics and Heat
Transfer, McGraw-Hill Book Company, New York, New7 York
72
APPENDIX I 1
HEAT TRANSFER INFORMATION
I 1-1
APPENDIX I 1
HEAT TRANSFER INFORMATION
I 1.1 BACKGROUND
The temperature portion of SHABERTH is designed to
produce temperature maps for an axisymmetric mechanical system
of any geometrical shape. The mechanical system is first appro-
ximated by an equivalent system comprising a number of elements
of simple geometries. Each element is then represented by a node
point having either a known or an unknown temperature. The
environment surrounding the system is also represented by one or
more nodes. With the node points properly selected, the heat
balance equations can be set up accordingly for the nodes of unknown
temperature. These equations become non-linear when there is
convection and/or radiation between two or more of the node points
considered. The problem is, therefore, reduced to solving a set
of linear and/or non-linear equations for the same number of
unknown nodal temperatures. It is obvious that the success of
the approach depends largely on the physical subdivision of
the system. If the subdivision is too fine, there will be a
large number of equations to be solved; on the other hand, if the
subdivision is too crude, the results may not be reliable.
In a system consisting of rolling bearings, for the sake
of simplicity, the elements considered are usually axially
symmetrical, e.g., each of the bearing rings can be taken as an
element of uniform temperature. For an element which is not
axially symmetrical, its temperature is also assumed to be
uniform and its presence is assumed not to distort the uniformity
in termpature of a neighboring element which is axially symmetri-
cal. That is, the non-symmetrical element is represented by
an equilvalent axially symmetrical element with approximately
the same surface area and material volume. This kind of approx-
imation may seem to be somewhat unrealistic, but with properly
devised equivalent systems, it can be used to solve complicated
problems with results satisfying some of the important engineering
requirements.
The computer program can solve the heat-balance equations
for either the steady state or the transient state conditions
and produce temperature maps for the mechanical system when the
input data are properly prepared.
I 1.2 BASIC EQUATIONS
I 1.2.1 Heat Conduction
The rate of heat flow qcij(W) that is conducted from node
i to node j may be expressed by,
I 1-2
L (t
i
- t.) I 1.1)
t
i
and tj are the temperatures at i and j, respectively, Aij
the area normal to the heat flow, (m
2
) Li. the distance
(m) and Aij the thermal conductivity between i and j, (W/mC).
Assuming that the structure between point i and j is
composed of different materials, an equivalent heat conductivity
may be calculated as follows:
A
2
3 Fig. I 1.1 Parallel Conduction
A
1
A X
2
A
2
A
ij
A..
A ij
A A
1
+ A
2
Fig. I 1.2 Series Conduction
L . Su.IL Jt
1
+
2
The calculation of the areas will be discussed in Section I 1.2.5
I 1.2.2 Convection
The rate of heat flow that is transferred between a solid
structure and air by free convection may be expressed by
I 1-3
1.25
qvi, j = i,j Ai'j ti - tj SIGN (t
i
tj) (1 1.2)
where
1, if (t
i
- t.) . 0
SIGN =f3
, if (tj - ti)< 0
in which
2 .5 10-2 W/m
2
- (degC)
1 "2 5
for hot surfaces facing upward
and cold surfaces facing downward
oij = 1.4 10-2 W/m
2
- (degC)
1 -25
for hot surfaces facing downward
and cold surfaces facing upward
1.8 10
-
2 W/m
2
- (degC)
1 -2 5
for vertical surfaces
For other special conditions, ij must be estimated by referring
to heat transfer literature.
The rate of heat flow that is transferred between a solid
structure and a fluid by forced convection may be expressed by
Ini,j = i,j Ai'j (ti - tj) (I 1.3)
in which ij is the heat transfer coefficient.
Now, withc = ij, introduce the Nusselt number
Nu = L (I 1.4)
the Reynolds number
Re L (I 1.5)
and the Prandtl number
=r PlCp (I 1.6)
A
where
L is a characteristic length which is equal to the diameter
in the case of a cylindrical surface and is equal to the
plate length in case of a flat surface.
I1-4
U is a characteristic velocity which is equal to the difference
between the fluid velocity at some distance from the surface
and the surface velocity (m/sec)
A is the fluid thermal conductivity (W/MC)
V is the fluid kinematic viscosity (M
2
/sec)
is the fluid density (kg/m
3
)
c is the fluid specific heat (J/kgC)
p
For given values of Re and Pr, the Nusselt number Nu and
thus, the heat transfer coefficient may be estimated from one
of the following expressions:
Laminar flow along a flat plate: Re < 2300
Nu = 0.323 VRe " - (I 1.7)
Laminar flow of a liquid in a pipe:
Nu = 1.36 'e Pr ) (I 1.8)
L
where D is the pipe diameter and L the pipe length
Turbulent flow of a liquid in a pipe:
N
u
0.027 RO
"8
.p (I 1.9)
e r
Gas flow inside and outside a tube:
N = 0.3 RO
" 5 7
(I 1.10)
u * e
Liquid flow outside a tube:
Nu = 0.6 RO
5
P
0
.
3 1
(I 1.)
Forced free convection from the outer surface of a
rotating shaft
Nu =0.11 [0.5 Re
2
Pr] 0.35 (I 1.12)
where the Reynolds number Re is developed by the shaft
rotation.
Re = W,1D
2
(I 1.13)
I 1-5
aim
in which W is the angular velocity (rad/sec)
D is the roll diameter (i)
The average coefficient of forced convection to the lubri-
cating oil within a rolling contact bearing may be approximated
by,
0.0986 {! L dm JJ A (Pr) (I 1.14)
using + for outer ring rotation
- for inner ring rotation
in which N is the bearing operating speed (rpm)
D is the diameter of the rolling elements (mm)
d
m
is the bearing pitch diameter (mm)
O is the bearing contact angle (degrees)
I 1.2.3 Fluid Flow
The rate of heat flow that is transferred from fluid node
i to fluid node j by fluid flow is
qfij = Vij Cp (ti - tj) (I 1.15)
V.- is the volume rate of flow from i to j. It must be observed
tAt the continuity of mass requires the following equation to
be satisfied
IVij = 0 (I 1.16)
provided the fluid density is constant. The summation should
be extended over all nodes i within the fluid which have heat
exchange with node j by fluid flow.
I 1.2.4 Heat Radiation
The rate of heat flow that is radiated to node j from
node i is expressed by
qRi,j
=
i,j Q(ti+273)4-(tj+273)4} (I 1.17)
where
Tj = tj + 273.16
T
i
= t
i
+ 273.16
I 1-6
- 4m
and the value of the coefficient depends on the geometry
and the emissivity or the absorptiviiy of the bodies involved.
For radiation between large, parallel and adjacent surfaces
of equal area, A
i
and emissivity, i, i. is obtained from
the equationij ij
0i,j = Ai. OAj (I 1.18)
whereCr, the Stefan-Boltzmann constant, is
e = 5.76
10
8
W/m 2/(degK)
4
For radiation between concentric spheres and coaxial
cylinders of equal emissivity, .i, Ji is given by the
equation
dij &ij A. (I 1.19)
1+ (I- E )L i.
A,j
where e is as above A is the area of the enclosed body and
A* . is the area of t surrounding body, i.e. A.A.
i
A*
123 iVj i,j
Expressions for0"i,j that are valid for more complicated
geometries or for different emissivities may be found in the
heat transfer literature.
I 1.2.5 Calculation of Areas
In the case of heat transfer in the axial direction A is
given by the equation (I 1.3)
A
i
j = 2'Wr Ar (I 1.20)
i m
Referring to the input instructions, Section 5, but recalling
L must be input in mm not m.
r +r
L =r = 1 2 (I 1.21)
m 2
L2 r r 2 - r (I 1.22)
I 1-7
.. .
In the case of heat transfer in the radial direction, Aij
is obtained from the expression
A
i
- 21W rm H; L
1
rm; L
2
H
and similarly for the radiation term above
A'i,j =
2
w r*mH
L
3
= r*m
L
2
= 2H
in which H is the length of the cylindrical surface; where heat
is conducted between i and j, r
m
is given by the same equation
as above (Fig. I 1.3(d)); where heat is convected between i and 1,
rm is the radius of the cylindrical surface (Fig. I 1.3(c)); where
heat is radiated between i and j, r
m
is the radius of the enclosed
cylindrical surface and rm * the radius of the surrounding
cylindrical surface (Fig. I 1.3(d)).
H
Fig. I 1.3 Ca) Fig.I 1.3 (b)
H
H OP
0ii
Te
Fig. I I.
3
(c)
Fig. I 1. 3(d)
I 1-8
I 1.3 Transient Analysis
For the transient analysis, all of the data pertaining to
the node-to-node heat transfer coefficients must be provided by
the input. Additionally, the volume and the specific heat at
each node is required. For metal nodes, this input is striaght-
forward. However, when fluid flow is being considered, there is
no easy way to approximate the fluid nodal volume in a free
space such as the bearing cavity. However, through use of the
Program, the user's ability to make appropriate estimates will
improve.
I 1-9
APPENDIX I 2
BEARING DIAMETRAL CLEARANCE CHANGE ANALYSIS, FROM COLD UNMOUNTED
TO MOUNTED OPERATING CONDITIONS
I 2-1
APPENDIX I 2
BEARING DIMIETRAL CLEARANCE CHANGE ANALYSIS, FROM COLD UNMOUNTED
TO MOUNTED OPERATING CONDITIONS*
I 2.1 INTRODUCTION
As a bearing is taken from the shelf, mounted in a housing
and on a shaft, turned up to speed and subjected to operating
loads and temperature, the diametral clearance of the bearing
will change. To accurately analyze the performance of a bearing,
its operating clearance must be known. An analysis has been
developed to account for the following effects:
1. Temperature changes and gradients.
2. Initial and operating shaft and housing fits.
3. Rotation induced, ring radial growth.
4. Uniform radial components of the rolling element-
raceway normal loads.
The basis for the major portion of this analysis is taken
from Timoshenko, i 4 1. The bearing rings are treated as thick
walled circular cylinders of constant wall thickness subjected
to the action of uniformly distributed internal and external
pressure.
The external pressure arises in the case of the outer ring
from a press fit into the bearing housing. The internal
pressure on the outer ring arises from the discrete rolling element loads
which are regarded as uniform internal pressure acting on the
outer ring. Similarly for the inner ring the press fit on the
shaft provides a uniform internal pressure and the rolling element loads
are regarded as a uniform external pressure. Fig. I 2.1 show the
idealized sections used in the analysis.
*This Appendix is based upon the original work L53
I 2-2
02
zz
- 7;0
Vl)
1 2-3
L_ _ _ _ _ _ _ _ -
I 2.2 BEARING SHAFT AND HOUSING EQUIVALENT SECTIONS
Timoshenko's analysis requires that a ring be fully defined
by specification of only internal and external radii. If two
concentric rings are pressed together the analysis assumes that
the rings are of equal width. Perhaps the outside surface of the
actual inner ring section cannot be defined by a single radius.
However, an equivalent radius may be found such that the cross sec-
tional areas of the real and equivalent rings are equal.
A second significant dimension associated with both the outer
and inner rings is the radius of the rolling element path at each
section. Ultimately, it is the change in these path radii, in
addition to a change in the rolling element diameter, which permit
a change in the bearing diametral clearance.
The bearing diametral clearance change problem requires the
solution to similar problems for both the inner ring-shaft (shaft)
and the outer ring-housing (housing) sections. To help eliminate
repetition and cumbersome subscripting, only the shaft section
will be covered thoroughly. Differences in analysis which occur
at the housing section will be noted.
I 2.3 INTERNAL AND EXTERNAL EQUIVALENT PRESSURES
Both the shaft and housing sections are represented by two rings
The rings may or may not be in intimate contact. Intimate contact
occurs when due to any of several factors, the outside radius of
the inside ring is greater than the inside radius of the outer
ring. A pressure is developed at that interface, which tends to
expand the outside ring and collapse the inside ring. Although
the pressure acting on the actual interfering surfaces is iden-
tical, the use of equivalent geometric sections requires the
I 2-4
use of equivalent pressure, such that the calculated pressure
acting to collapse the inside ring will not be identical to the
pressure acting to expand the outside ring.
The equivalent pressure concept allows the effect of unequal
component widths to be considered.
The effect of unequal inner ring and shaft width is taken into
account by the factor X which applies to th. pressure where,
referring to Fig. 1 2.1
. ( Z
(1 2.1)
(I
2.2)
Where W denotes the element width and the subscripts I, 0, S
and Ii refer to inner, outer, shaft and housing respectively.
The equivalent pressures acting on the shaft and housing are
thus
so (1 2.3)
H HX POO (1 2.4)
denotes the pressure on the shaft O.D. as related to the
pressure on the inner ring %,P1C andPO* are similarly defined.
The pressures (P) in Eq. (I 2.3 and 1 2.4) are non-zero only if an
interference fit exists at the section in question. An inter-
ference fit may exist under cold mounted conditions but due to thermal
gradients and high speed rotation, clearance may develop at operating
conditions.
1 2- 5
To account for the elastic effect of the uniform radial
components of the rolling element loads on the ring dimensions,
an equivalent inner ring external pressure and outer ring internal
pressure are calculated based on the minimum radial component of
the rolling element loads, such that for the inner ring;
-Pt CosQ Cki o~
zr- C W
(I 2.5)
where Z is the total number of rolling elements
C(Qxosa ) is the smallest rolling element raceway
Kii radial load component at the inner
raceway
W, is the inner ring total width, or one half the total width
if the inner ring is split.
An expression corresponding to Eq. (1 2.5) can be developed
for the internal pressure acting to expand the outer ring. In fact
?Io will exist only when the applied load is predominantly axial or
when the bearing is preloaded, but the internal pressure acting on
the outer ring, Pot, will always exist to some extent, as a result
of rolling element centrifugal force.
I 2.4 TEMPERATURE EFFECTS
Temperature effects at operating conditions are considered in
two ways. A bulk effect based on the radii to the rolling paths (R)
is calculated from Eq. (I 2.6)
g 2 Z X!(I 2.6)
where
w-
is the change
in rolling
path
radius
- is the component coefficient of thermal expansion,
S- is the difference between the component effective
temperature and 68 F.
I 2-6
A similar expression may be written for the change in roller
diameter.
The second effect of temperatures is the change in interfer-
ence fit from cold to operating conditions such that the inter-
ference fit as a result of the initial fit and the temperature
difference between the two components can be calculated. Cold
(293K, (68
0
F)), shaft and housing fits C*LA)are input to the
analysis. A positive value reflects an interference fit. The
analysis neglects asperity crushing. The fit(Fho~ti)at operating
temperature is given by
(F11 ) ( CA A) z'6 ksts LtI)(I 2.7)
hotis the fit considering operating temperatures.
I 2.5 RING RADIAL DISPLACEMENT CONSIDERING SURFACE PRESSURES
At this point we consider the problem of evaluating thick
walled cylinder, (ring) radial displacements as a function of
the ring radial dimensions, physical properties, and uniform
internal and external pressures. Timoshenko develops Eq. (I 2.R)
for determining the ring/radial displacement (g), as a function
of a general radius (r) to any point within the ring when consid-
erations of ring rotation are omitted.
0 1 (1 2.8)
The general solution of this equation is
~ ~ 4(1 2.9)
The expression for normal stress in the ring is given by:
+ ~(1 2.10)
I 2-7
* ;~* ~-*
The constants of integration Kand V can be determined from
Eqs. (I 2.11 and I 2.12) which make use of the boundary conditions
at the inner and outer surfaces of the ring. Note that the dimen-
sions of inner ring section are used in Eq. (I 2.11 and I 2.14).
These equations, however, are general expressions and are valid
for all rings,
-"-M = Z I
"C . (1 2.11)
i _p r6&- (1 2.12)
Where E and ) are the modulus of elasticity and Poisson's
ratio respectively. The negative sign associated with 'Pzand
PX. reflect the sign convention, wherein a positive normal
stress indicates tension. Solving Eqs. (I 2.11-12) for K, and
k and substituting the results into Eqs. (I 2.9 and I 2.10)
gives the general expressions for normal stress and radial deflec-
tion as functions of the internal and external pressures acting on
the ring:
S ., - _7-6 % P_ (1 2.13)
CI - aI r. 16 C%- I.
)
__ 6___ C
2
O fV**t~~
vi-L"P )
+
(L (c%
-
262
.
Using the appropriate dimensions and pressures, Eq. (I 2.14)
can express the inward displacement of the shaft and the outward
displacement of the inner ring when assembled with an interference
fit FIIOT.
Erb - Es
o
(I 2.15)
"') ( -c L , L' 'a-F" (I 2.16)
Ef - 6-Z __6 - 1' l (I 2.17)
I 2-8
s andPi, are assumed zero. Using Eq. (1 2.5) to obtain ?Z.,
the following expression may be developed for Pxi which accounts
for all variables in the interference fit problem except for the
effect of high speed rotation.
R t( 2.18)
6 .
. 6X
Note, P may not be negative, i.e. the effect of fitting one
ring on another cannot result in placing the common surfaces in
tension.
Having determined _t. and r. Eq. (I 2.14) may be used to
determine !t,~ by replacing the general radius (r) with the
radius to the roller path, R. ce(Ftj4is defined as the change in
rolling path radius resulting from:
F The initial fit
t* The change in fit resulting from a temperature gradient
The effect of the rolling element load.
I 2.6 RING ROTATION
We must now examine the effect on the rolling path which
results from high speed rotation. If the rotational speed is
less than 100 rpm, the rolling path radius is assumed to be
unaffected.
Timoskenko presents Eq. (I 2.19) to define ring displace-
ment 9 in terms of a general radius (r), the weight density of the
ring material, ( ), and the ring angular velocity (SL)
I 2-9
Using the notation;
N -0
(I 2.20)
the general solution to Eq. (I 2.19) nay be written:
--. , Y. (1 2.21)
The general expression for the normal stress is given by:
r N K, l(I 2.22)
Using Eqs. (I 2.21 and I 2.22) and the principle of super-
position, the effects of ring rotation can be considered. In this
case, superposition allows a set of integration constants
K, and KL to be calculated for a ring based on a change in
pressure (P*) at the internal and external surfaces. Eq.
(I 2.22) may be written for the four specific surfaces of the
shaft section as follows;
Inner Ring External Surface
-P W z 0: (1 2.23)
Inner Ring Internal Surface
ET 0 (-1'Vr-- (1 2.24)
Shaft External Surface
Shaft Internal Surface
is (I 2.26)
From Eq. (I 2.3)
s(1 2.27)
An additional useful relationship derived from Eq. (I 2.21)
defines the difference between the inner ring and shaft
I 2-10
displacement at their common surface, as a result of their
rotational speed, in terms of the four integration constants of
Eq. (I 2.23 - I 2.26).
64 21S N,.e - ts Ns(1 2.28)
10 8 6
Under p.ress fit conditions 9,-
We will now examine the application of Eq. (I 2.23 - I 2.26)
in determining the ring behavior as a function of rotational
speed.
The following conditions might be encountered:
1. A tight fit remains tight
2. A tight fit loosens
3. A loose fit remains loose
4. A loose fit tightens
For all four conditions, u.4Oand '*(.O , i.e. it is assumed
that no change in pressure occurs at the internal surface of the
shaft or the external surface of the inner ring resulting from ring
rotation. AlsoS2
Z
and
S
, are identical. All four of the integra-
tion constants Kz , 11 , Kl, and 4x are unknowns. Also, either
the change in pressure PZf or the rotational speed S. embodied in
NE and N% are additional unknowns.
Now the formulations of Eq. (I 2.23 - I 2.28) are presented
which are required to solve each of the four conditions.
I 2.6.1 SITUATION 1 AND 2, INITIALLY TIGHT FIT
In addition to the integration constants,'Pt is the
unknown, where is the change in inner ring internal pressure
resulting from the rotational speed. If ?zL is less than ?:
from Eq. (I 2.18), situation 1 is realized. The operating
fit pressure is the difference
between w and' and the fit
remains tight. The resulting radial displacement at the inner
ring rolling path is given by:
. - (I 2.29)
I -L
and the total change in the rolling path radius at the inner ring
is
"'+ ",L 4 . (I 2.30)
is the change in rolling path radius which occurs as a
result of a temperature change from 68 F.
If is greater than ?ZL the solution to the problem
requires several additional steps. Eq. (I 2.23), must be re-solved
to deternine the rotational speed, St.. at which the change in
fit pressure is just equal to the initial fit pressure and VO
This rotational speed is termed the tight fit speed limit and is
subscripted with the letter (TL). Using the integration constants
thus determined and Eq. (I 2.21), -rL.. may be determined.
Note, the subscripts in parenthesis should be interpreted
as follows: The first subscript refers to the state of the
initial fit (T-tight) or (L-loose), the second subscript refers
to the rotational speeds (T-Total) or (L-Limit). For instance,
LUr. refers to the change in inner ring rolling path radius due
to the rotational speed at which the initially tight shaft fit is
lost.
After determining UL.yEqs. (I 2.23 - I 2.26) are resolved
twice for the four integration constants, with Ifset to zero.
First, for the full rotational speeds, yielding UT-and then
for the tight fit speed limit, yielding ULL . This is accom-
plished by changing the values of Nz and Ns through a change
in the value of SL in Eq. (I 2.20). Utilizing super-position,
the change in rolling path radius resulting from rotation is given
in Eq. (I 2.31).
9A -l. 9TL
(1
" "
[ 2.31)
Eq. (1 2.31) is presented graphically in Fig. I 2.2. Then
is calculated from Eq. (I 2.30).
I 2-12
FIGURE I 2.2
Ring Radial Expansion vs. Rotational Speed Squared
ULT-
UJ
ULL
z
P Q
P.=o
a4
IL
TL
0
-
0
2 n
LT
ROTATIONAL SPEED SQUARED S
2
1 2-13
I 2.6.2 SITUATIONS 3 AND 4 INITIALLY LOOSE FIT
The integration constants KY , zz, , and %$.s are
unknowns. , , L and are all set to zero, and
only Eqs. (I 2.23 - I 2.26) are solved. Using the integration
constants thus obtained, Eq. (I 2.28) is solved for S . If
e IT rrI where To.m is negative, the loose fit remains
loose at operating conditions. The constants and K are
used to calculate U&_ . Using Eq. (I 2.30), Uior is
calculated. If FAoc', the shaft has expanded further
under the effects of rotation than the inner ring, and the
initially loose fit becomes tight. This requires additional
solutions to Eqs. (I 2.23 - I 2.26).
S is set equal tor-Wi.rand the equations are solved
for the integration constants plus the speed at which the init-
ially loose fit becomes tight EL.. Using the integration
constants L0T and kg7 along with Eq. (I 2.32), 96L. is
determined. Eq. (I 2.23 - I 2.28) are again resolved twice at
-. and 61 for P,. and the integration constants, after
setting :* to zero. 6T, and S',r can be obtained by from
eq. (I 2.29) using the two sets of integrati1 constants . The
operating fit pressure is then given by;
1 -rT
(1 2.32)
and Eq. (I 2.30) is used to determine gor. Where
represents the total change in the radius to the rolling path
of the section as the bearing is taken off the shelf at 68
0
F,
mounted on the shaft subjected to load and rotated up to
operating speed.
I 2-14
I 2.7 BEARING DIAMETRAL CLEARANCE
The preceeding sections dealt primarily with the shaft-
inner ring section. However, the equations presented are valid
for all four of the assembly sections. For both sections, a value
is deternined for 9T, cold fit pressure, operating fit pressure
and the speed at which an initially tight fit loosens.
As noted earlier, Eq. (I 2.6) can be-used to determine this
change in rolling element diameter such that:
eg" =(1 2.34)
The change in bearing diametral clearance is given by
- - (1 2.35)
The bearing operating diametral clearance equals the initial
clearance plus the change in clearance PI
1 2-15
APPENDIX
I 3
ELASTIC SHAFT ANALYSIS
I 3-1
APPENDIX 1 3
ELASTIC SHAFT ANALYSIS *
1 3.1 COORDINATE SYSTEM, LOAD AN) GEOMETRY CONSIDERATIONS
The shaft subprogram calculates the deflection characteristics of
a general shaft in two planes, one plane at a time, by solving
the differential equation for the deflection curve of
the shaft,
= _ M
(1 c 3.1)
where M = moment
E = modulus of elasticity
I = moment of inertia
The coordinates X and Y are shown in Fig. I 3.]
Figure I 3.1 Shaft Coordinate System and Shaft Loading
The shaft load may consist of concentrated loads, moments, or
linearly varying distributed loads, Fig. I 3.1. The shaft may
be hollow, and the inner and outer diameters may vary stepwise
or linearly, Fig. I 3.2
Figure I 3.2 Shaft Schematic Showing Stepwise and Linear
Dianeter Variation
*This Appendix is based upon the original work (1i
1 3-2
The shaft may be supported by bearings at up to five locations.
The bearings may take force and/or moments and they may be initially
displaced ( yo )
Figure 1 3. Schematic-Shaft Supports
The bearing subprograms calculate the reaction forces and moments
on the shaft from the bearings as functions of bearing deflections.
The bearing deflections can be looked upon as shaft support displace-
ments , (S , )o
I 3.2 SIAFT BEARING EQUATIONS
The shaft reaction at any location i is calculated as the shaft
reaction, FO; , at i when all additional displacements of the shaft
supports are zero, plus the additional reactions at i., caused by all
additional displacements, or bearing deflections:
I* F . -* I- &,r -~ (1 3.2)
The reactions Fo are calculated using the shaft program with the
given initial displacements and spring constants. The constant
derivatives J/j and cl/< are calculated by introducing
one initial displacement at a tine. If the bearing mounts are
rigid, i.e. spring constants are zero, then the reaction at loca-
tion i depends only on the deflections at i-l, i,and i+l. The
other derivatives are zero. The present program uses the shaft
program this way, and has not yet been extended to include all
derivatives. Thus the influence of housing elasticity is not
taken into account.
I 3-3
-- _ , " .. . .. ,d - ' : _.. - _, i -_,.. " .
I 3.3 METHlOD OF SOLUTION
The object is to calculate the shaft deflections and slopes
at the bearing locations along the shaft. The shaft may be arbi-
trarily loaded and have arbitrarily varying cross-section. The
supports may take force or moment load, they can be initially
displaced. These deflections and slopes are found by going through
the following steps:
Step 1. Take away all supports (support and bearing are
synonomous) except two. In the following, these two sup-
ports are called the initial supports. In the steps to
follow, slopes, deflections and initial displacements
refer to a coordinate system with the X-axis along the
center-line of the unloaded shaft when it is supported
by the initial supports. The relation between this co-
ordinate system and the original system is given by the
coordinates ; andp of the left end of the shaft as de-
fined in the original coordinate system.
The problem is now statically determinate and the reac-
tions of the initial supports are calculated.
Step 2. Define a new coordinate system. Fix the left
end of the shaft in that system so that deflection and
slope are zero at that end. Let the load acting on the
shaft still be the external load plus the reactions from
the initial supports as calculated in Step 1.
Since these reactions were obtained from equilibrium
equations of the shaft, there is no reaction at the
fixed end. Therefore, all shaft loads are known and the
moment M of the shaft can be calculated at any section X
I 3-4
from an equation of equilibrium of the shaft to the left
of the section. The slopes and deflections are calculated
from the Moment-Area theorems by numerical integra-
tion of the following equations.
Oc) =CI 3.3)
Step 3. Put the shaft back on the initial supports and
correct the slopes and deflections accordingly. The
resulting calculated slopes and deflections are in the
following denoted by &C.
Step 4. The reaction forces and moments, here called
Vi .at all supports except the initial supports are now
introduced. Their magnitude is determined from the condi-
tion that the deflection at each support location goes
back to zero + initial displacement,
... "" 3.5)
where
are influence coefficients
(is the deflection or slope at location i from a
unit force or moment at location j.
The influence coefficients are obtained by introducing
a unit force or moment at one support at a time and then
calculating the deflections and slopes starting from
Step 2.
There is one equation for each of the n bearings except
the two initial supports. The linear system of equations
I 3-5
* !
can be written
(I 3.6)
where A is a matrix Nb x Nb
the vector of unknown reactions
= - - (1 3.7)
The reactions are obtained by solving this system of equations.
Step 5. Calculate the reactions at the initial supports
again, this time with both the external loads and the cal-
culated reactions acting on the shaft.
Step 6. The deflections and slopes are calculated again
as in Step 2, this time with external loads and all reac-
tions acting on the shaft.
Step 7. The deflections are expressed in the original co-
ordinate system by addition of SVC to the lateral deflec-
tion and IO to the slopes.
I 3-6
APPENDIX I 4
CONCENTRATED CONTACT CALCULATIONS
I4-1
APPENDIX I 4
CONCENTRATED CONTACT CALCULATIONS *
I 4.1 INTRODUCTION
This section deals with the calculation of the forces which
develop in rolling element-raceway contacts in ball and cylindrical
roller bearings.
I 4.2 COORDINATE SYSTEMS
Consider a plane through the bearing axis and the center of a
rolling element Fig. 2.4. The position of the center of the rolling
element relative to the outer ring is described as a coordinate system
fixed to the rolling element center, having its x-axis parallel
to the bearing X-axis and y-axis radially outward from the bearing axis.
I 4.3 BALL RACEWAY FORCES
In a ball bearing, the outer and inner ring groove curvature
centers EC and IC, are defined by the vectors 9, and
respectively, Fig. 4.1 and 4.2.
The normal force Q is obtained from the Hertz theory for bodies
in contact under load.
Q K
1
s- el3 (1 4.1)
where K
1
= a function of the initial geometry and of the material
properties.
Sel = the elastic deflection
*This Appendix is based upon the original work (11
1 4-2
I -.. ... I 1 1 lT i .rl'.... . -- . .
The elastic deflection at one ring is obtained from the distance
between the ball center and the groove curvature center, that is
1.9,1 I$ . When the elastic deflection is zero, this distance is
r A- o.d5 where rg is the groove curvature radius. When there is
elastic deflection, this distance is greater by the amount Seq of
the elastic deflection. Thus,
The dimensions of the contact ellipse are obtained from the same
theory,
a = K2 Q1/3 (I 4.3)
b = K
3
QI/3 (I 4.4)
where a = the major half-axis
b = the minor half-axis
K
2
and K
3
= functions of the initial geometry and of the
material properties
4XI-
1 4-3
FIGURE 1 4.1
Ball Bearing Geometry
_____ _center__ of ute
rac cona4
-. - 4. - ~ ~ ' LC
FIGURE I 4.2
Ball Coordinate System Showing Ball Center Position Vectors
Final position
inner rinO groove
curvatire center
ic
original position
EC X1
outer ring
groove cur ature
ce n t er
1 4-5
where = a vector from ED to the original position of IC
9= a vector between IC
o
and IC
where (I 4.6)
the total clearance at one ball in the direction of
C is the displacement of the inner ring groove curvature center
at a ball from original position to loaded position in the bearing
coordinate system. S)'A is the vector Q in the rolling element
coordinate system.
(I 4.7)
The vector is the projection of on the x-y plane, i.e.
with the z-component equal to zero.
I 4-6
Sis the rotational transformation matrix which transforms
forces and moments described in the bearing coordinate system to
their equivalents in the rolling element coordinate system.
0 0
(i
4.8)
where 0 is the rolling element aximuth angle described in Fig. 2.4.
The vector Z is obtained from the bearing linear deflection vector
and angular deflection vectors
where 1 (1 4.9)
1 4.4 ROLLER RACEWAY FORCES
In a roller bearing the position x
1
0 and y, = 0 for the roller
center is defined as the position of the roller center when it is
centrally located in relation to the raceway in the x-direction, and
when the roller under no load and no misalignment touches the outer
raceway, Fig. I 4.3.
I 4-7
FIGURE I 4.3
Roller Bearing Geometry and Roller Coordinate System
I 4 RCK
1 4-8
To obtain the normal forces on the roller, it is thought of as
divided into YIK slices. The force on each individual slice is
calculated separately, as though the entire length of the roller
was deflected to the amount which obtains at the actual slice. Shear
forces between slices are neglected. The relation between deflection
and force is given by the following formula, J
2 5
K
4
'.~ t- . .I (1 4.10)
where Q = roller load
Iq= roller length
Yt
=
elastic deflection
The load per unit length of the contact is q.
Q
q(I 4.11)
The slice load, Qk is obtained from:
~ .% i ~re = ai.L A r. K re ' ZeW (1 4.12)
At the outer race, the deflection ,Eat slice k is the difference
between the normal approach 9 and the crown drop - . See
Fig. 3.3.
9 e 1K
- jat(
(I 4.13)
I 4-9
where sum of roller and race crown drops
g~= ~(1 4.14)
A
y = a unit vector in the y-direction
0
The vector Irv, from the rolling element center to the center of
slice k in the roller raceway contact is defined as follows:
(I 4.15)
At the inner race, the deflection at slice k is:
+ Ovc K X (1 4.16)
where the vector from the inner ring center to the contact at
the inner race, - is
I 4-10
Since the roller rotates only about its geometric axis, the
angular alignment of the roller may be described in terms of the
orthogonal components of the roller rotational speed (W) as follows:
I 4.5 SUMMATION OF CONTACT TRACTION FORCES
Of all the forces which develop within various concentrated
contacts, only the elastic Hertz force in a ball raceway contact
may be treated as a point force. The elastic forces which develop
in a roller raceway contact and both ball and roller traction forces
cannot be treated as point forces since they are not necessarily
symmetric. Within the extremely small areas which develop in
concentrated contacts, the variation in surface pressure is extreme.
Despite the small area, significant variation in rolling element-
raceway, relative surface velocities may occur within the contact.
The acute pressure variations plus the changes in surface veloci-
ties necessitate that concentrated contact areas be subdivided into
smaller elemental areas. Over the latter the assumptions of a
circular pressure distribution and constant relative sliding
velocity closely approach the physical situation.
For the roller raceway contact, the same contact sub-areas used
to calculate the elastic force distribution are subsequently used
to evaluate the traction force distribution. For the ball
raceway contact, the sub-areas consist of slices perpendicular
to the contact major axis.
1 4-11
1 4.5. I V[CTOR TO 'TiIl CON'ACT (AiN'IR FkOM TlE ROLLINC EL..MENT CENTER
For roller bearings the vector i has been given previously.
In a ball bearing the vector i- is required to calculate the
ball raceway friction forces and is given by its components n and t.
See Fig. I 4.4.
,. t. (1 4.19)
the upper and lower signs apply for outer and inner ring contacts
respectively.
The radius of the deformed surface, Fig. I 4.5, is given by
Eq. 1 4.20.
- " ' (1 4.20)
(I 4.21)
0 t-A ((sL ~lci a
(1 4.22)
The coordinates t and n are used with the other components of the ball
rotational speed to calculate the ball surface velocity components.
1 4.5.2 SIMMINC THE TRACTION FORCE OVER A CONCENTRATED CONTACT AREA
In computing the traction force acting in a contact, the high
pressure contact region is divided into a number of slices parallel
to the minor axis. In a roller raceway contact, these slices are
the same as those used to calculate the normal force. In a ball
raceway contact the total contact length as determined by the Hertz
analysis is divided into approximately 21 slices. The tractive
force is computed for each lice and then summed to give the total.
1 4-12
FIGURE 1 4.4
Ball Coordinate System Showing Ball-Race Contact
Position Vectors
AI
Ole"
1 4-13
FIGURE I 4-S
Rail Race Deformed Contact and Deformed Surface Radius
origin of radius of
original groove siirface
1 4-14
Figure I 4.6a shows a Llertzian contact area with the semi-
elliptical distribution of pressure due to elastic deformation
(effects of the lubricant on the pressure distribution are
neglected) and a local coordinate system established at the
contact center (t-O ).
By considering a sufficient number of slices the variation of
prevsure in the x-direction over a slice width may be neglected,
i.e. each slice is regarded as the contact zone due to a cylindrical
disk (without edge effects).
Sliding velocities at a typical race slice are shown in Fig.
I 4.6b. A sliding velocity in the x-direction results if the ball
rotational vector has a component LJ The sliding velocity u
5
s is
always equal for all slices.
Because of groove curvature the sliding velocity component u
will vary from slice to slice across the contact ellipse.
Figure I 4.6c shows the traction components T
x
and T
z
on the
slice. T indicates the resultant for the given slice. The forces
T and T are computed for each slice and summed to give the
components of the total traction force acting at the contact.
Additionally, during the integrations, friction moments of the
tractions about the rolling element center are obtained, using the
radius vector coordinates n and t.
I 4-15
FIGURE I 4.6
Calculation of Traction Force Components
Tz
T
:TX
c. TRACTION COHPONENTS AND THEIR RESULTANTS AT
A TYPICAL SLICE
sZ
S?
UsX
b. SLIDING VELOCITIES ON TYPICAL SLICE
z
I-9'x
a. CONTACT ELLIPSE AND PRESSIIRI DISTRIBUTION
1 4-16
APPENDIX IS
LUBRICANT PROPERTY AND EHD FILM THICKNESS MODELS
1 5-1
IL-
APPENDIX I 5
LUBRICANT PROPERTY AND EHD FILM THICKNESS MODELS *
I 5.1 LUBRICANT PROPERTY MODELS
Many of the calculations performed by computer program SHABEPTH
require that the viscosity 7 and the pressure-viscosity coefficient
of, be known at a given temperature.
Accordingly, the program employs subroutines which, when given
lubricant kinematic viscosity V at 1000F (37.78C) and 210F
(98.89
0
C), density ' at 60'F (15.560C) and the thermal coefficient
of expansion, determine the lubricant density, viscosity,
and the pressure-viscosity coefficient at any temperature required.
The kinematic viscosity V (cs) at atmospheric pressure is
calculated at a given temperature t (OF) from Walther's relation 113
log
1 0
logl
0
(V1+0.6) = A - B log
10
(t + 460) (I 5-1)
where A and B are constants determined by substituting the known
values of V at t = 100F and t = 2100F into Eq. (I 5-1) and
solving the two equations which result for A and B.
Having calculated at a specific t, I is comnuted as
(T S-2)
* This Appendix is basedpon the original work [93
I 5-2
r44 0
rm 0 4.
C0
(A4~ 0
-4 U - .
0 ) 0) C) C
-4 - 14 1-
C 44
4 -
*~0
cla -4 41
'0 4 1.0 0
0 0 0
C 0C>
C)) 4C
00- C -
CC
'0 m0 -4 '0)
Cid
oJ U r4-n C
0 .4
tfld
-4 -
U -
M A000C
0'0)
-4
-4
.10
1 5-3
where the lubricant density at temnerature t is given by,
Ct C6o.) - Ci-o (1 53)
where G is the lubricant coefficient of thermal expansion.
The pressure-viscosity index o( is defined implicitly by the
relation
(I 5-4)
where I (p) denotes the viscosity at pressure p at an arbitrary
temperature.
The value of a( itself varies with pressure, The appropriate
value of c7 to use in the film thickness prediction equations
wherein it appears, is the value applicable at the inleti.e.,at
atmospheric pressure.
The value ofCo( at a given temperature and at atmospheric
pressure is calculated by the relation developed by Fresco, j14
(560)
= (2.303) 10
"4
[C+Dlogl0 + E(lOgl, )21(t + 460)(in.2/lb) ( 5-5)
wherein V is evaluated at temperature t (OF) and C, D, and E are
constants tabulated by Fresco as a function of S* 0.2B where B is
the coefficient in Walther's equation.
I 5-4
Another lubricant property that appears in the model for film
thickness reduction due to inlet heating is the temperature viscosity
coefficient , that appears in Reynolds' exponential temperature
viscosity relationship.
is computed from the viscosity values at t = 100F
and 210 F as follows,
0.00909 ln jj(t = 100F) (I 5-6}
0 (t = 2100F)
Relevant lubricant properties for the oils whose properties have
been preprogrammed in SHABERTH are listed in Table I 5-1. These
property values have been supplied by the manufacturers.
I 5.2 LUBRICANT FILM THICKNESS
The elastohydrodynamic (EHD) film thickness,h, at each contact
is computed as the product of the film thickness predicted by the
Archard-Cowking 20 (point contact) or by the Dowson-Higginson t2j (line
contact) formulas and two reduction factors 0t and s The factors
ot and 0s account respectively for the reduction in film thickness
due to heating in the contact inlet and the decrease in film due
to lubricant starvation, i.e.,due to the finiteness of the distance
between the contact zone and the inlet oil meniscus. In equation
form,
h t O s hA.C.3 (I 5-7)
D.H.
I$-S
The Archard-Cowking and Dowson Higginson film thickness formulas
take the following forms respectively.
2 3 -0.93 0.740 0.407 -0.074
hA.C7 2.04 1 + R( IV) R (Q/E') (I 5-8)
Pyo
~~~~
0. 13
V
5 9
Rx, Ry - effective radii of curvature parallel and transverse
to the rolling direction respectively
- pressure viscosity coefficient
V - lubricant entrainment velocity
R - = + Ry11 -1
q = maximum load per unit length
Q load -1
E= 2 1 + 21
E
1
E
2
absolute viscosity
EIE
2
= Young's modulus for the contacting bodies
= Piosson's ratio for the contacting bodies
I 5-6
1 5.2.1 Inlet hleating Factor 0t
A Grubin type inlet film thickness analysis considering full
thermal effects was developed for line contact by Cheng 110 .
Results presented in J10) covering wide ranges of loads, speeds, and
lubricant parameters were used to develop regression formulas for
the thermal reduction factor $t" Based on 28 sets of data, each
containing 15 data points, the regression formulas obtained take the
following form:
x
t =e 0
t (I 5-10)
where
(1) x = -0.3011 - 0.00432 In (p
0
/E') - 0.03469 In (1 + S)
-0.16423 in Qm- 0.01728 (lnQm) 2 + 0.00389 l'I
- 0.06316
in
for 0 < Qm< 0. 1, ' 11.5 and 0< Q< 0,4,
(2) x = -1.119304 - 0.16192 in (p /E') - 0.0895 In (1+S)
0
- 0.29 In Qm - 0.04572 (lnQ M) 2 + 0.13615 In o '
- 0.31615
In
for 0.14 Qm <.1 ' 11.5 and 0.4 Q 1, 1 , ' 11.5
(3) and x

= -3.66426 -0.48511 In (p
0
/E') + 0.00568 * ,
- 0.05491 ' - 0.1678 In (1+S) - 0.19573 lnQm
2
- 0.09392 (lnQ ) + 0.20908 - In O
for Qm > 1
where, o ' =
Po
I 5-7
S = (u
2
-u
1
)/2V
Qm
=
2 V
2
/Kf to
0(- pressure viscosity coefficient, in
2
/lb.
b - half of Hertzian width in the rolling direction, in
Po- maximum Hertz pressure, lb/in
2
- temperature viscosity coefficient, oR I computed via
Eq. (4-6)
t - ambient temperature, oR
2
- ambient viscosity, (lb.sec/in
2
)
U, 2 - surface velocity of bodies No. 1 and 2, (relative to the
contact) in/sec
Kf- conductivity of the film (lb/F.sec)
V - (ul+ u
2
)/2, lubricant entrainment velocity Cin/sec)
It is noted that for siall values of Qm' t computed from
Eq. (I 5-10) may be larger than unity. Ot = 1.0 is used whenever
the value computed using Eq. (I 5-10) is larger than 1.0.
In evaluating 0t for the elliptical point contacts in a ball
bearing, p
0
is taken to be the maximum of the Hertzian pressure
ellipse.
The point contact is thus treated as if it were a line contact
having a maximum contact pressure p
0
along its entire length of
contact. This is a conservative approximation inasmuch as it will
tend to underestimate 0t and hence underestimate film thickness.
I 5-8
The magnitude of the error resulting from this approximation is
small as p
0
is not a highly influential variable in the expression
for $t"
I. 5.2.2 Starvation Reduction Factor 0s
A hydrodynamic analysis of an elliptical point contact having
two equivalent principal radii of curvature R and R (parallel and
x y
transverse to the rolling direction, respectively) is used for
calculating
0
s" Finite thickness of the half films (h
1
,
1
and h
1
,
2
)
upstream from the inlet are set. The flow rate in the center plane
at the meniscus line is set equal to the incoming flow rate at the
contact centerline. The complete analysis is given in jllj and
Figure I 5-1 shows the geometry considered. The starved
plateau film thickness is 6 3 and the meniscus distance from the
contact center along the direction of rolling is r *. The ambient
film layers move toward the contact zone with velocities u
1
and u
2
.
The analysis shows h
s
and r* to be related to hi, the sum
of the upstream ambient film layer thicknesses, (h
1
= h
1
,
1
+ h
1 2
)
through the following two equations:
S.5V~ R - 2V r* 1.0 (I 5-11)
hs3/
2
(3+2k) (3+2k) (hs+ r*
2
/2R x
)
2
5 x
and h
1
= 2(24k) (hs )+ kr*
2
(I 5-12)
3+2k (3+2k)R
x
I 5-9
FIGURE I S-1
Film Geometry
EXIT INLET
__-w x ROLLING DIRECTION
Yr*/
Ul
2
15-10
where k R x /Ry, J is the viscosity of oil at the contact inlet,
OK the pressure-viscosity coefficient and V 3 ( u
1
+ u
2
) is the
entrairment velocity in the x-direction.
The simultaneous solution of Eqs. (I 5-11) and (I 5-12) for
given values of Rx, Ry, V, , o( and hl yields the associated
values h and r*. Subroutine STARFC performs this calculation.
s
As the meniscus distance r* increases, the film thickness h
s
increases, asymptotically approaching a value hf as r*-P al,
Therefore, hf is the film thickness under fully flooded conditions.
On letting r*--aocand hs-- hf in Eq. (I 5-11), the second
term on the left hand side vanishes and one may solve for hf as
2/3
112
hf S.5V o( (Rx ) (I 5-12)
(3t2k)
Note that for line contacts, the contact curvature ratio k is
set equal to a small value.
K
=
R = 0.01.
This k value results in a contact length to width ratio of 18 : 1.
The lubricant meniscus distance in the rolling element-cage
contact is assumed to be proportional to the rolling element radius.
r* = 0.25 r
I 5-11
Subroutine STARFC evaluates hf from Eq. (I 5-13) and then
calculates the ratio Os = hs/hf" As given by Eq. (I 5-13), it
does not indicate a dependence upon load. This is characteristic
of film thickness formulas derived from a Grubin type assumption
applied to rigid bodies. It is considered preferable to use the
Archard-Cowking and Dowson Higginson formulasfor the unstarved cases
rather than hf since these formulas better describe the dependence
of film thickness upon the influential physical variables. The
onl-y role played by hf is to scale h
s
to yield the ratio which
is applied as a multiplicative factor on the Archard-Cowking and
Dowson Higginson predictions of unstarved film thickness.
I 5.3 Film Replenishment
As noted above, it is necessary to know the combined oil layer
thickness h
1
to calculate 0s and r*.
As a rolling element passes a point on the inner or outer raceway
of a bearing, a very thin lubricant film remains on each of the
components and is of the same order of magnitude as half the
plateau film thickness in the EHD contact. Replenishment of the
lubricant layer on the raceway is required in order to assure
sufficient lubricant in the inlet region of the succeeding contact,
so that the EHD film thickness will be the same as in the preceding
contact. If replenishment fails to occur, each successive rolling
element pass would have a thinner EHD film and steady state operation
with EHD lubrication would not be possible.
I 5-12
;L6.............. ..
Many mechanisms serve to replenish the lubricant in the track
of a high speed bearing. Of prime concern is replenishment at the
inner race in high speed ball bearings since centrifugal force tends
to direct free fluid away from that surface. Seven replenishment
mechanisms have been identified:
(1) Centrifugal flinging of the lubricant from the ball.
(2) Centrifugal travel of oil along the surface.
(3) Random splashing of lubricant in the bearing cavity.
(4) Direct deposition from a jet.
(5) Back flow along the surface into the track, from its
edges resulting from lubricant surface tension.
(6) Carrying into the contact, of lubricant adhering to
the ball.
(7) Back flow into the gap behind the contact exit due to
vacuum in the cavitated area.
A model exists for item (5) above. In view of the other,
possibly more influential sources of replenishment enumerated above,
this model has not been adopted in this program. Instead, it is
assumed that an externally supplied replenishment, amount
adds to the plateau film thickness to yield h
1
.
As a simplification, since A is usually much larger than h,
the approximation h vh is used so that,
s
h.h s + A(I-S-13)
Subroutine STARFC uses Eq. (I 5-13) in solving Eqs. (I 5-10) and
(I 5-11), with a user specifified & value.
Criteria for estimation of 4 I are presented in Sect. 3.3.15.
I 5-13
APPENDIX
I 6
TRACTION
AND INLET
FRICTION
CALCULATIONS
1 6-1
APPENDIX I 6
TRACTION AND INLET FRICTION CALCULATIONS
I 6.1 INTRODUCTION
The traction model developed for use in program SHABERTH
is applicable to the partial EHD regime in which the lubricant
film separating the contacting surfaces may be small enough to
permit some degree of asperity contact. The model computes the
traction coefficient as a function of the ratio (h/gj, film
thickness h to composite surface roughness Ir. For small values
of h/ir (h/r-<6(LO the model represents dry friction. For large
values of h/ (h/T> 3) the model becomes a non-Newtonian semi-
empirical fluid film model in which the traction coefficient,
IEHD' depends upon sliding rate, as well as the load, rolling
speed and lubricant properties at operating temperture. For
intermediate values of h/q the model is a combination dry and
fluid film friction. Essential features of the model are described
below.
A single, two dimensional, functional relationship is used
to model aspects of both wet and dry friction. The function is
plotted in Fig. I 6-1 and is valid over the range.
O '-
o
and has the following characteristics.
L (1 6-1)
j.-
. IS
*Yb
V,
X ") (I 6-2)
where:
x to7) - xL
defines the value of y below which y'is a linear
function of X.
defines the value of X* below which y
is a linear functionof X I 6-2
6-34
Note that y* increases monotonically with x* and that
is continuous at x* = xB.
Both wet and dry friction coefficients will be defined in
terms of Eqs. (I 6.1 and I 6.2).
I 6.2 ASPERITY TRACTION MODEL
If Q is the total load applied to a concentrated contact (or
to a suitable sub-element of a contact), a portion of this load,
designated Qa will be carried by elastically deformed asperities
and the remaining, Q-Qat will be carried by the EID film. The
traction force T
s
then is,
Ts =4a Qa +,'
4
EHD
(
Q - Qa)
(I 6.3)
Under the assumption that the rough surface consists of two-
dimensional ridges of random height and slope angle, the average
asperity borne load Qa is the following function of the ratio h/q.
a I (h/- A(I 6.4)
where
F
1
,E 2 Young's moduli of the contacting bodies
A = contact area
To = RS value of the distribution of asperity slope
angles (radians)
I(h/V, a)= function defined in (121 of the filn parameter h/q"
and a statistical micro-geometry parameter V defined
in Nayak (261.
It is shown in f271 that O = 2 is a reasonable value to use
for rolling bearing surfaces. The following polynomial fit to the
function I(h/q,o< =2.0) is used ini computation:
I(h/q,2)=2.3le
-
l.
8 4
h/q
"
+0.ll75(h/1-f.4)
0
"
6
(2-h/q )2 (I 6.5)
0.4 h/q 2
I(h/q,2) = 17e -2.84h/,- + 1.44 x 10
- 4
(h/T-2)
1
.
1
(4-h/r)7.8
2 ch 1"
When h/V4O.4 the asperities are assumed to carry the entire
contact load.
I 6-4
The asperity friction coefficient (ja) is calculated as
follows:
4
a /1* y*
(I 6.6)
* is supplied by the user, 0.1 to 0.2 is the recommended
range of values.
where y* is calculated from Eq. (I 6.1 or I 6.2) in which:
YB is 0.66
XB is 0.005
x* - U-UI) calculated, local
U V+Uj slide to roll ratio
U
2
and U
1
are defined in Fig. I 6.4
This relationship is believed to better approximate the
asperity friction phenomena than the coulomb friction model.
This relationship causes the friction coefficient to be a
function cf the small tangential displacement of the surface of
one body with respect to the other. The slide to roll ratio
provides a measure of the displacement.
In the numerical solutions being employed, it is important
that the functional relationships produce a unique set of forces
for a given set of rolling element rotational speeds. The use of
the above asperity friction model helps guarantee this uniqueness.
The value = 0.1 was recommended in (277 as being consis-
tent with values deduced in traction measurements in the partial
EiHD regime.
I 6.3 FLUID TRACTION COEFFICIENT EHD
The general behavior of the fluid traction coefficient
1
E11D
as a function of sliding rate is illustrated by the curves in
Fig. (I 6.2). (Throughout this section the subscript on EHID will
be omitted.)
I 6-5
I.
In the curve in Fig. (I 6.2a) the traction coefficient
increases linearly, at low sliding speeds, reaches a maximum/-/*
at speed u
s
= us*, and decreases thereafater.
In the curve in Fig. (I 6.2b) the traction coefficient
increases linearly at low sliding speeds and then approaches an
asymptotic valueA*.
Both types of traction curves have been experimentally
observed. Both signify a departure from isothermal Newtonian
fluid behavior since for this situation the traction coefficient
increases linearly with sliding speed.
For the curve of Fig. (I 6.2b),/A* denotes the asymptotic
traction coefficient. Define us* = 3 x usc, where usc is the
sliding speed at which the lineM= *intersects the extended linear
portion of the curve /Mvs. u
s
.
It has been found that/A* for either type of curve increases
with the contact pressure and decreases with rolling velocity
and lubricant viscosity (and hence ambient temperature) /(s* on
the other hand decreases with pressure and increases with rolling
velocity and viscosity. This joint variation has been found to
result in a scale change in the two axes but not in a substantial
change in the character of the traction curve.
This means that traction curves obtained under widely dif-
ferent conditions of pressure, rolling velocity and temperature
when plotted on a grid with coordinates/Ar =/ k* and x - us/us,
yield substantially the same curve ,L'.
A characteristic of the curves of the type of Fig. (I 6.2a)
is that /Ur decreases indefinitely with large x. Curves of this
type can cause convergence difficulties in a bearing computer
program because there are two sliding speeds associated with
each value of the traction coefficient. The solution may cycle
I 6- 6
..........
FIGURE 1 6.2
Typical Traction Curves
LU
U
SLIDING VELOCITY, u
s
a) A CURVE EXII1BITING A PEAK
LL.
z
0
uc SLIDING VELOCITY, u
s
b) A CURVE EXITING ASYMPTOTIC BEHAVIOR
1 6-7
between two values in seeking an equilibrium condition.
An approach to avoiding this difficulty is to use the
previously mentioned monotonically increasing curve Eq. (I 6.1 and
I 6.2) that is a good match to the actual curve over the increasing
portion of that curve. If the program then converges at a sliding
speed that is within the range where the curves match, the solu-
tion is valid.
In the present context, let yo and x * from Eq. (I 6.1 and I 6.2)
correspond to the relative traction coefficient (,MR) and the
sliding speed LIS , respectively.
If we now define /A as the asymtotic EHD traction coefficient
the acatual coefficien kEHD is:
/14 EHD /IA' * +/A
R
(I 6.8)
For a given oil, the maximum traction coefficient/U* has
been found J273 to vary with pressure, viscosity, rolling velocity,
and film thickness in the following manner:
4C'P")' ob 14)C (0,59)v(o4
0 '
a
'
6
'
)kk (-u,4 (I 6.9)
where
2
Po
=
the maximum contact pressure (lb/in )
h = plateau EHD film thickness (in)
f(po) = a function governing the dependence of viscosity on
pressure po (oil parameter)
= a visco-elastic constant (oil parameter)
Similarly, us* exhibits the following dependence on Po'Iop
Van . ,(-o, o (-L -0.09) , (I 6.10)
LA
5
A (V ? j Vh)(16.10
The quantities# *,,s * and h can be measured experimentally
for a set of given values of po, 1, and V. For the oils thusfar
examined, the function f(po) has been found to follow a law of
the form,
f%(PO/p1)
A
4
PO4P,
At Pa. " 1P
I 6-8 (I 6.11)
where A
1
, A
2
and p
1
are lubricant dependent constants.
Values of A
1
, A
2
and P1 for four oils are listed in Table (1 6.1).
The procedure for determining these values for other oils is given
in j23,2 .
Making use of Eq. (I 6.11), Eqs. (1 6.9) and (I 6.10) can be
expressed in the following form on introducing C
1
and C
2
as propor-
tionality constants.
,y. 1.-,,4 AL. ,.o (o. ') (o.-,,5o"
f.I . V h (I 6.12)
, - 0.1o 4 . ,A (1 O DA -0.09). l -
FO~ '' v (I 6.13)
The values of C
1
and C
2
are evaluated by substituting measured
*, Us * and h values for a specific test condition. Then knowing
values of C
1
, C2, Pl, Ai and h (starved), it is possible to pre-
dict values of/* and us* as functions of the operating parameters
po, V, and by Eqs. (I 6.12) and (I 6.13). Values for C
1
and C
2
thus calculated for four oils are tabulated in Table (I 6.1). The
units in Eqs. (I 6.12) and (I 6.13) are V (in/sec), I (cp), Po (ksi)
and h (microinches). Also shown in Table (I 6.1) are the values of
( Ar)B and (us)B used for the four oils. These data have been
preprogramned into SItABERTH.
In summary, the calculation of /4EIID at a given sliding velocity
u
s
and for a given pressure po, film thickness h, rolling velocity
V and temperature t, proceeds as follows:
1. Calculate viscosity I at temperture t.
2. Using appropriate constants from Table (I 6.1), calcu-
late/,4s* and/A* for given po, h, V and
3. Calculate x - u
s
/ U *
I 6- 9
0D 0 0
00
I 10 C', C'
64 *4
0 ) -4 1
0 -
U) LA 0
0 .,4
00 a 0
E--4 H ",
ora
04
E-ot
10; C',
IC' ~0 0
m 0 C%44 4
4I 00 Ch 0&0
o en0
>1 0
V 1 44 4 44
E4- w~.
0 :0 0 0) 0-)
- 0 rn 0
0~ C',' .- 4 M
0 0
06 10
LL6
4. Use Eq. (1 6.1 and I 6.2) with values of (/,r)B and
(US)B from Table I 6.1 to calculate /
r
associated with
the value of x calculated in Step 3.
5. Compute #EID =MA r ./A*
I 6.4 INLET REGION HYDRODYNAMIC FRICTION FORCES
I 6.4.1 ELLIPSOIDAL CONTACT
The contacts between ball and race and between ball and
cage pockets are "point" contacts. Under lubricated conditions,
the surfaces are separated by a fluid film and there is a
pressure build-up around the contact caused by the sweeping-in
motion of the surfaces. This pressure build-up contributes to
the friction in rolling, Tangential surface forces are required to
pump the oil into the high pressure zone.
The pumping forces are due to rolling, (FR), and sliding (FS).
Expressions as a function of a finite meniscus distance r*
have been found for these forces. The complete analytical develop-
ment is contained in ill and 27j .
Since these forces arise in the contact inlet region,
elastic deformation is not considered to have a significnat effect
and the analysis invokes a rigid body assumption.
Figure (I 6-3 ) shows the relevant geometry. Two rigid bodies
are shown in nominal point contact separated by an oil film and
undergoing relative rolling and sliding. A local cartesian co-
ordinate system is established with the x-y plane parallel to
the tangent plane of the two bodies and with the origin coincident
with the surface of body 2. The coordinate system remains fixed
in the contact as the surfaces of the two bodies move. The prin-
cipal radii of curvature of the two bodies are (Rx)
i
and (Ry)i
,
i - 1,2. The eauivalent radii are
p , + l. e (1 6.14)
1 6-11
FIGURE 1 6.3
Notation for Rolling Sliding Point Contact
z
BODY 1
41
u V-1
FIGURE 1 6.4
Friction Forces on Sliding and/or Rolling Disks
FORCES ON SOLID BODIES
y
u
1
< U
2
( FORCES ON FLUID
I
F
1
FR*.4
F RF\
2F
1 6-13
The normal separation of points on the two bodies near the
origin is given by
h = h + + V2 (I 6.15)
0 7x
My
where h is the minimum film thickness at the origin. This
separation function is applicable if the width of the tlertzian
flat region in the contact is negligible compared to the
relevant x and y dimensions in the inlet and outlet.
The surfaces are assumed to be moving with velocities u.
1
and v
i
(i = 1,2) relative to the origin in the x and y directins.
The rolling velocities in the x and y directions are defined
respectively as V
x
= (u
1
and u2)/2 and V. = (v
I
and v2)/2 and
sliding velocities in the respective directions are usx u l-u
2
and usy = v
1
-v
2
"
) is the portion of the normal load that is being supported
by hydrodynamic forces in the inlet. For elastic ball race con-
tact 0 is considered negligible when compared to the load sup-
ported over the Hlertzian contact zone.
The forces F., F
s
and Fn are displayed in Fig. (I 6.4). For
clarity the special case of contact between two disks is illustrated.
In this case, R, is infinite and the forces are directed along the
x axis. In the general point contact case the forces have both x
and v components.
The force FR acts in the same direction on both contacting
bodies and opposite to the direction of motion. F
s
acts in
opposite directions on the two bodies in such a way as to tend to
increase the speed of the slower body and to decrease the speed of
the faster body.
The forces F
1
and F
2
are the resultants of the hydrodynamic
pressure distribution in the inlet and act through the centers of
the two bodies. The component of these forces in the y direction
1 6-14
represents the (synail) portion of the total load supported
hydrodynamically. The components Fn adFn acting in the x
direction contribute to the force balance in the rollinxi direc-
tion.
The magnitudes of F R, FS and Fn depend upon the meniscus
location r*. The calculation of r* for the rolling element race
and rolling element cage contacts is discussed in Appendix I S.
Expressions for the x and y components of the forces F R,
F sand Fn are given below in terms of the dimensionless quanti-
ties F T and Fs.
Pumping Forces
Roli-nF component
Fv = 11C F RCos It (1 6.16)
Fx= 11 C
0
FR (sinV) (Rv/Rx)il (1 6.17)
Sliding component
F V FS usx ( RR )i (1 6.18)
F T= F
5
.u
5
s (IRR) (1 6.19)
sxxy
Normal Forces (on ball)
F C o* (1 6.20)
ny Co R r o~
Fn =C FR - . (sinys) (R/Rx)" (1 6.21)
(For a ring, r is replaced by the raceway radius) where
2 -1 2( . 2
Co Vx (TRXV) (.3,2k)- + (Vx/v )2 (3+2k yk13 ( .2
ta-1 3 +2k X(6.3
=~~ ~ tn **k(3+2k-)
t y
4 I6.3
k
R /Rx
Ix
)=absolute ambient viscosity~
I6-15
The ouantities 7R and 7
s
are dimensionless and depend upon
two further dimensionless parameters 9 and 1 1 is a
dimensionless meniscus distance defined as
[r*/(2hoRx) ][(cos2i + (l/k)sin
2
6) ] (I 6.24)
h
o
for ball-race contact is taken as the plateau EHD film
thickness.
For ball-cage contact h
o
is as calculated in Appendix I 8.
r* is the distance of the oil meniscus from the contact
center along the rolling direction.
is the product of the pressure viscosity coefficient of
the lubricant and the maximum fluid pressure qmax
that prevails if the lubricant is isoviscous.
assumes values between 0 and 1 with f= 0 indicative
of purely hydrodynanic and = 1 of purely elasto-
hydrodynamic conditions. a = 1 is taken for ball-race
and = 0 for ball-cage pocket calculations.
Plots of s and FRas a function of ?l for various values
of 3 are given as Figs. (I 6.5) and (I 6.6) taken from 123
For the EIID ball race contacts the following expressions
have been fit to the FR VS. 1 curve in Fig. (I 6.6)
for = 1.
FR = 23.59 in l - 10.1; ?1A 5 (I 6.25)
FR = 36.57 in R1 - 22.85; ,1 5
F
s
is not considered for a ball-race contact because the
amount of sliding is so small.
For the predominantly hydrodynamic contacts, (= 0) which
arise between the ball and the cage web, both pumping and
sliding friction are considered. The following equations were
fit to thea = 0 curves in Fig. (I 6.5) and (I 6.6).
I 6-16
...............
FIGURE 1 6.5
Variation of F
5
with P
so
40
s
30
20
10
110
102
I 6-17
FIGURE I 6.6
Variation Of FR With the Dimensionless Meniscus Distance p
1
0.5
ci*1.0
120
0.8
100
80
r
60
aq max~
40
20
01 p p 3 11I 1 11
110 102
1 6-18
FR = 8.53 x In-" tln(1000 ,1.70 4 5.4 (I 6.26)
FR = 36.576 In 1 - 29.32; ?
1
S.4
TS = 1.71 x 10-9 [Ln (l OO t j
1
) - nl 1
'
2 1 2
- (I 6.27)
Ts = 10.115 (lntl) .965 + 1.5; '
1
k 2
In applying the above results to ball-cage web and ball-
race contacts, it is necessary to interpret the geometrical
parameters of the general configuration of Fig. (I 6.7) in terms
of the appropriate bearing dimensions.
Figure I 6.7 shows the relevant geometry for the two contact
types as well as the direction of the various force components.
In this figure r denotes the ball radius,r' the cage pocket
radius, rg the outer ring groove radius, R the radius to the
center of the contact ellipse and 0( the outer ring contact angle.
I 6.4.2 INLET REGION IIYDRODYNAMIC FRICTION FORCES LINE
CONTACTS
A nodel has been developed for calculating hydrodynamic
pumping and sliding friction forces in the inlet regions of
roller bearing race contacts wherein film variable viscosity
and starvation effects are included. The contacting bodies are
assumed to be rigid. The model was developed by Dr. Y. P. Chill
of S K F Industries.
Consider two rigid cylinders of radius R y and Ry
2
moving
at surface velocity UL
1
and LA2 as shown in Figure I 6-4. Floberg
ino3 has obtained expressions for the fluid pressure distribution
P,. normal rolling friction forces F and Fn2 and tangential
friction forces FR as functions of a dimensionless meniscus
distance X.1 = r*/( R-ho r Isoviscosity is assumed. h
o
is the
minimum film thickness and R = (1/RyI + 1/Ry
2
)
'1
I 6- 19
FIGURE 1 6-7 Configuration of Contacts
(a) Ball-Cage Contact
(b) Ball-Race Contact (Outer Ring)
INERNN
AXISR
R~r ~ (/../) A Ir j-
R I.R.COC/)
t -2
'-... ".
Spec if i cal ly', tlie ,ax i)irnur fluid pressure Pmax is given by
nax Po* (2-3 u!)h
0
/2
(I 6.29)
where Po* is a function of X
0 1
as shown in Table I 6. 2. The
normal friction forces acting through the centers of the two
cylinders of length I are
The pumping forces are
FR = 0.5 (Fnll + F n
2
) (I 6.31)
where u = (u
2
+ ul)/2
The dimensionless coefficients Pxo1 A, B and C, given in
Tablel 6.2below, are functions of X.1 and were calculated by
Floherg
Table I 6.2
.Dimensionless Coefiicients for the Calculation of
Line Contact Inlet Friction
X
01
OXo A C B
- 4.6 3.5 4.485 3.41
-2.245 0.81 2.72 0.877 2.94
-1.380 0.25 2.27 0.322 2.92
-0.9S4 0.11 2.02 0.176 3.22
For the case that the fluid viscosity increases exponentially
with pressure, i.e. -1=ae it is desireable to use the Archard-
Snidle approximation 1311 that the minimum film thickness h
o
in
an EI) contact is deterniend from Eq. (1 6-25) by setting p max
-1
equal to the reciprocal
of the pressure viscosity coefficient
(o( ).
Usinp P max 1, one then has,
Pa 0 a-"/(zI
12,/,t )
(I 6.32)
and (t,\(r 2)
-t /(rt~
1) (1 6.33)
kL ,.,A .
(I 6 .34)
where C .P X o X i )AP.)V
I b-21 (I 6.35)
The values of C and B, as functions of X
0 1
, are tabulated in
the last two columns of the above table.
For the case where one has pure rolling suach that k a 1.0,
Eq. (I 6.30) yields,
F~c ~
t
'e/Q.) (1 6.36)
The forces F
n
exerted on the roller race contacts are,
[ F (outer ring) R1/3
(innering)
C
(
6.37)
n ~RI R
where Rm = R- r = R
i
+ r = roller pitch radius.
Chiu I has determined in his analysis that contact load has
a negligible effect on the above pumping forces, and has obtained
good agreement with experiment for his rigid body assumption.
Sliding friction has been determined to be U,
/~~
~~ ~ (o teb
o j C - Z /
SS(i.v e=tr rl'i.a )) ('4 j( :fl1 "L . j L(I 6.38)
where u = entrainment velocity = (u
2
+ ui)/2 and V = sliding
velocity = u
2
- u
1
I 6.4.3 HEAT GENERAT
T
)N RATES
In the ball-race and ball-cage inlet regions, the heat
generated due to the sliding force F
s
and sliding force FR is
calculated as,
= 2F RV + Fsu (I 6.28)
where V = fluid entrainment velocity at the contact center
u = sliding velocity at the contact center
I 6.5 BALL DRAG FORCE IN BULK LUBRICANT
In[ 33the following form of "churning friction force" is
cited, to account for all friction losses on the ball other than
EHY) sliding traction in the ball/race contacts:
I 6- 22
F
w
= Av Cv (dmiAo)- (I 6.39
8g
where F
w
is the drag force
A : the ball frontal area
Cv: a drag coefficient given int353as a function
of the Reynolds number
diM: the bearing pitch diameter
W0: the ball orbital angular velocity
g: the gravitational constant
the density of the air-oil mixture in the bearing
cavity
= XCAV . (I 6.40)
XCAV: the fractional amount of lubricant assumed to
be in the bearing cavity
o: the density
of the oil
In the present model, three hydrodynamic force components at
each point contact on a ball have been defined.
These components tend to retard ball motion as would Fw.
Since two race contacts and a cage contact exist for each ball,
15 force components have been made explicit. After accounting
for all contact friction forces, there is left a residual loss
due to "windage" or "drag" acting on a ball as it moves through
the air-oil mixture in the' bearing cavity. Eq. (I 6.37) has been
used to model this windage force., Although the effect of the drag
force is less significant than calculated in 3 , it remains impor-
tant.
XCAV values of one percent or less are recommended. In actu-
ality,XCAV is a function of lubricant supply rate, method of
supply, speed and bearing and bearing cavity geometry.
I 6-23
APPENDIX I 7
ROLLING ELEMENT INERTIA FORCES AND IOPENTS
I 7-1
APPENDIX 1 7
ROLLING ELEMENT INERTIA FORCES AND MOMENTS
A rolling element, Fig 2-4, traveling between azimuth locations,
is forced to undergo changes in its rotational velocity components
(x' 9 y' and 1z as well as in its orbital velocity W0
o .
The
forces which must act on an element to produce time variations in
its rotational and orbital velocities may be deduced from Newton's
Laws of Motion as follows:
F Fyy-W0(R+y) (1 7.1)
where F
x
, F and F are the components of the forces in the rotating
coordinate system attached to the element, m is the ball mass, x and
y are the element center displacements shown in Fig. 2-4, and R is
the radius of outer ring groove centers,
A rough estimation assuming stable operation yields that the
term x is smaller than & 02 .(R+y) by a factor of the order of
x /R, where xm is the maximum variation of x. Similarly, the
m
terms y and 2 4oy are smaller than 0
R
by a factor in the order
of y /R where ym is the maximum variation of y. Note that both
x m/R and ym/R are very small in magnitude. The second derivatives
with respect to time of x and y are thus neglected as is the
Coriolis term 2 4 o. The term d) is expressible as follows:
I 7-2
= dl
o
d d do dSO (I 7.2)
ko 0 -- (1 .2
dT d dT do
The term db0 is approximated for ball i as follows:
d
CdA) 1_ (6o) i + 1 (Aio)i - 1 (I 7.3)
dO 2 A
where At is the angular distance between rolling elements.
The moments necessary to cause the elemient velocity to change are
as follows:
x
0 0
0 13 0
&
(I 7.5)
00 0J
Z
r a ball:
2
* I - ,1 = *D2/10 (I 7.6)
, V 7
I 7-3
for a roller:
Jx = mD
2
/8
J = Jz = m/12 (3/4 D2 + Xre
2
) (I 7.7)
yz
m is the element mass
D is the element diameter
ire is the element length
The time variation of the rotational velocity components
4,3 y and uz are approximated in the same manner as 4o e.g.,
45 &[ ( i+l (x i=l (I 7.8)
Using D'Alembert's principle, forces -F and moments -M
calculated as described above are imposed on the element along with
the other forces and moments due to friction and elastic contact.
The combined system of forces is then regarded as being in static
equilibrium.
I 7-4
Because the time rates of change of 6o' 0&x',Jy andAM
z
are
included by approximation as described above, the analytical treat-
ment is considered to be quasi-dynamic as distinct from analyses
wherein these terns are neglected and only the centrigufal force
m 4)
2
(R+y) and gyratory moments JW o. W Q and -JW) W are considered.
The description "quasi-static" has been applied to slutions of
this type.
I 7-5
APPENDIX
I 8
ROLLING ELEMENT BEARING CAGE MODEL
I 8-1
APPENDIX 1 8
ROLLING ELEMENT BEARING CAGE MODEL*
I 8.1 INTRODUCTION
The cage is driven by normal and friction forces which act
at the interfaces between balls or rollers and cage pockets, and
at the cage rail(s) and ring land(s). These forces are calculated
as functions of the separation and speeds of the interfacing mem-
bers. In this analysis it is assumed that:
normal forces exerted by the rolling element on the
cage pocket act in the plane of cage rotation which
is coincident with the cage axial midplane.
friction forces exerted by the rolling element on
the cage pocket act orthogonal to a corresponding
normal force and at the normal force point of appli-
cation.
the only friction force components considered in the
cage equilibrium equations are those which lie in the
plane of cage rotation. It is assumed that each rolling
element is axially centered within its pocket.
cage rail normal forces act at the cage midplane and
pass through the axis of the cage. These forces are
coplanar with the rolling element normal forces.
cage-land friction forces act in the cage midplane such
that any resulting torque tends to drive or retard the
cage rotation.
*This Appendix is based upon the original work G53
I 8-2
The analysis is used to determine the normal and traction forces
at each rolling element and ring land on the basis of hydrodynamic,
elastohydrodynamic and Hertzian concentrated contact theory.
I 8.2 GEOMETRY
Figure 2.4 shows a coordinate system (XYZ) with the origin on
the outer ring axis in the plane of the outer raceway centers. A
local coordinate system (x,y,z) is established at the center of each
rolling element. The azimuth angle 0 defined in the (X,Y,Z) coord-
inate frame locates the x axis penetration through the Y-Z plane.
The x axis is parallel to X. The y direction is radially outward
and the z direction is tangent to the direction of rolling.
A local coordinate z c is also defined for each cage pocket,
wherein the origin is located on the cage pitch circle. We wish
to determine the position of each rolling element center with respect
to the cage pocket center along zc' in terms of the rolling element
orbital speeds Wo, the cage rotational speed C and the cage
rotational and translational displacement components.
I 8.3 CAGE MOTIONS
The equilibrium solution considers that the cage operates
in one of three modes:
(1) The cage is outer ring land riding such that radial
and small circumferential motions of the cage with
respect to the rolling elements are resisted by hydro-
namic fluid film forces that develop between the cage
rail land outside diametral surface and the bearing outer
ring outside diametral surface, Three degrees of freedom
1 8-3
apply to the cage motion. These are the circumferential
position of the cage relative to the rolling elements ( )
and two components of radial displacement (AY and AZ
in rectangular coordinates, or e and *'c in polar coord-
inates.) When the bearing is subjected to axial load
only, or when the rolling element speed variation is in-
consequential, the radial degrees of freedom are neglected.
(2) The cage is inner ring land riding when motions are re-
sisted by hydrodynamic forces which develop at the cage
inside surface and the bearing inner ring outside surface.
Three or one degrees of freedom also apply.
(3) The cage is ball or roller riding in which case there
are no net radial fluid film forces between the bearing
rings and the cage, and consequently, no radial motion
of the cage relative to the bearing axis of rotation.
Angular motion of the cage relative to the roiling
elements is the only applicable degree of freedom.
The circumferential displacement of rolling element No. 1 at
azimuth location = 1 relative to its cage pocket center is
designated
I 8.4 ROLLING ELEMENT MOTIONS
Returning to Fig. 2-4, the velocity with which the moving
coordinate system rotates about the X axis is designated o0 and
is also assumed to be a function of azimuth angle, i.e. to =0 o
(
).
I 8-4
The rolling element is assumed to rotate relative to each of
the axes in the moving system of coordinates. The angular
velocities about each of the axes x, y, z are denoted 0, o
,
and
4z respectively and are shown as the orthogonal components of the
rotational velocity vector )in Fig. 2-4
The value of the ball center-cage pocket center offset z
c
applicable at other ball positions is deduced relative to ball
position No. 1, which remains fixed at its aximuth position.
The cage is assigned a rotation , so that the offset of cage
pocket no. 1 relative to bal no. 1 is
g ---- ge (1 8.1)
In so doing, it is assumed that a rolling element orbital
velocity remains constant as it traverses the distance corresponding
to one half of the pitch spacing on either side of the nominal
azimuth position. As a rolling element enters the azimuth location
of the next adjacent rolling element the orbital speed undergoes a
step change. This is illustrated in Fig. I 8.1 for ball Nos. 1 and
2. The top half of Fig. I 8.1 is a plot of the assumed variation
of orbital velocity with respect to ball position.
The cage orbital velocity is denoted bycj3 and is assumed
uniform and equal to the average of the ball orbital velocities, i.e.
M jL& (1 8.2)
where n is the number of rolling elements.
I 8-5
Figure 1 8-1
Cage and Rolling Element Speeds and Displacements
UO
z
II
0.
0)
.4*400
*f
U
1 8-6
The distance between ball positions is the quotient of the
circumference 11 d
m
of the locus of rolling element centers
(neglecting small excursions) and the number of rolling elements.
The time 4 T for the cage to traverse this distance is then
ATz-
(1 8.3)
In this time period the center of rolling element No. 1 moves
a circumferential distance of
Z., A (1 2..8.4)
The circumferential distance between the rolling element and cage
pocket center at position No. 2 is obtained as the difference
between the rolling element travel and cage travel in time A T,
less the initial offset of the cage pocket center at rolling element
position 1 with respect to the center of rolling element 1,(Y), see
Fig. I 8.1, plus the components of the radial eccentricity
(6Z and4Y) of the cage axis with respect to the bearing axis. Then
(I 8.S)
Letting AT = then Eq. (I 8.5) becomes for the general
i-rolling element;
I 8-7
FIGURE I 8-2
Cage Pocket Geometry
1 8-8
(I 8.6)
I 8.5 CALCULATION OF CAGE POCKET NORMAL FORCES
A means for calculating the cage driving forces due to the
balls was developed in J23 I. The analysis has been extended to
include roller-cage pocket,line contacts, by approximating the
line contact to be an elliptical contact that has a large
curvature ratio (e.g.: a*/b* = 18 ).
The analysis is applied to determine the normal forces acting
at two diametrically opposite points on a rolling element, i.e., the
points of nearest and furthest approach of the ball or roller
relative to the cage. The net normal force acting on the rolling
element is the resultant of these two forces. The discussion
below considers ball-cage; but it applies to roller contact.
The typical ball geometry is shown in Fig. I 8.2. z c denotes
the offset between the ball and cage pocket centers in the direction
of rolling. Wx and W y denote the components of the ball rotational
velocity vector that result in relative surface speeds of the
ball and cage pocket.
The closest approach h
0
is the minimum film thickness when the
cage is lubricated.
I 8-9
Given the ball cage pocket eccentricity the associated cage
pocket load P can be calculated. When z c is small the load is
small and borne hydrodynamically by the lubricant film, which then
has minimum thickness h
0
= r' - r - z c In this regime, the load for
a given value of the ball and the cage pocket clearance is that
supported by a hydrodynamic contact of minimum thickness h
0
.
Elastic deformation is negligible in this regime, As z c increases,
h decreases until it reaches a critical value h below which a
further increase in z c results in elastic deformation hut no further
decrease in film thickness, In this regime
h =h (I 8.7)
and the elastic deformation is calculated from,
e = Zc + hc - Cr (I 8.8)
where Cr is the cage pocket radial clearance (r' - r).
The load P in this case is assumed to be the sum of the load P
c
hydrodynamically related to the film thickness h c' and an
additional load Pe associated with the elastic deformation through
the Hertzian and flexural equations of contact elasticity.
An analysis was performed as described in t23j of the relation-
ship between normal load P and minimum film thickness h
0
in a
lubricated point contact between two rigid bodies, each having two
I 8-10
principal radii of curvature, assuming, that the luhricant
viscosity increases exponentially with pressure. The analysis
yielded a relationship between the nondimensional load parameter (7
and the nondimensional film thickness parameter IT, as shown by the
solid curve in Figure I 8.3. These nondimensional variables are
defined as:
= P. ( Ry /C 2 1/3 (RxRy
- i
1/2 P D (I 8.9)
H= h Rx (Co R x- o 0 B (I 8.10)
C
r
where:
C
0
= o o Vy (RXRY) 1/2k (I 8.11)
-2V-l22 -l
k = (3+2k)
-2
+ (3+2k- )
- 2
k
- 1
) J (I 8.12)
V
y
k Ry /R
x
(I 8.13)
D ,( yC21/3 -1/2
D (P( R IC ) (Rx R y (I 8.14)
B . C P (C R )2/3 (I 8.15)
r x o x
R ( 1) - (I 8.16)
x r rV
C
r
P. = r
y
C r r' r, cag- pocket radial clearance
r
I 8-11
FIGURE I 8-3
Load Capacity Vs. Film Thickness for Hydrodynamic
and Elastohydrodynamic Operating Regimes
so
ELASTIC
OR
I EHD REGION
40
Tc
30
I NTERMIEDIlATE
B REGION
~20
RIGID
II SOVI SCOUS
5 REGION
RII
0
loISOICU
0
0 Hc 10 20 30
DIMENSIONLESS FILM THICKNESS,
I 8-12
V = 1/2 W.y r
= bulk viscosity
Vy = -1/2L x * r
It has been found that the relationship between'Q andH for an
unstarved point contact can be approximated by the following formula:
Q = 53.3 (H)-1/2 + 163 H)
-
' ( I 8,17)
provided that:
&-e c= 37.6
I 8.5.1 ELASTOHYDRODYNAMIC (EHD) CONTACT
For Q 37.6, the film thickness is independent of load and the
nondimensional parameter H remains constant at H . Operation in this
case is in the EHD region.
H = H = 3.122 (I 8.18)
c
for EHD Contact
= 37.6 (I 8.19)
Equations (I 8.18) and (I 8.19) result in the following for the EHD
region of operation,
= c = 37.6/D (I 8.20)
h
o
= h
c
3.122 Cr/B (1 8,21)
1 8-13
The -lastic deformation e is given by,
= z + h Cr (I 8.22)
Pe (9e) was originally calculated according to Hertz Theory.
This model has been changed, however, in an attempt to reduce the
nonlinearity of rolling element-cage load displacement relationships.
This was done through the assumption that the cage will respond
to large rolling element loads through flexing as well as
through the local contact displacements.
It was assumed that 95 percent of -e would be acc'mmodated
by cage flexing and the remainder would be accommodated by the
Hertz deformation. The Hertz calculations are made based on the
assumptions of a 9:1 major to minor contact axis ratios for a
ball-cage contact and an 18.2:1 ratio for the roller-cage contact.
The cage flexure deflection is calculated from
Pel(fex) = (0.9Se) R, 13500 (I 8.23)
The spring constant 13500 R where R is the cage rail radius,
was derived using circular ring theory, considering the cage
material to be steel. The total ball-cage contact load P is
thus
P = Pc + PHZ ( .e) + Bflex) (0. 9qSe) (I 8.24)
I 8.5.2 HYDRODYNAMIIC (HD) CONTACT
If the contact film thickness h is greater than the critical
0
value hc, the contact is assumed to be hydrodynamic:
H Hc = 3.122 for HD contact
The minimum film thickness for this case is given in terms of the
ball-cage clearance and eccentricity zc as:
I 8-14
h= C - z
c
(I 8.25)
the calculation procedure is,
(a) Calculate h as above
(b) Evaluate H = h
o
R
x
(C Rx)-j
(c) For H from (b) find Q = 53.3 ()
-1
'
/ 2
+ 163 (H)
3
(d) Calculate cage-ball load P as Q/D
The procedure for determining the ball-cage normal load is
performed for the points of nearest and further approach (h
0
min
and h max). The net normal load acting on the ball is given by:
ZP = P (h
0
minimum) - P (h
0
maximum) (1 8.26)
I 8.5.3 DRY CONTACT
Normal cage-pocket rolling element contact forces for con-
ditions of dry contact are calculated for the h
0
(minimum) contact
only. A continuous force displacement function is assumed for
this calculation. A soft spring and low force values occur for
4 0.99, where Z = 2 7c/tr'-r); such that,
P = 10 . r . e for e 4- 0.99 (I 8.27)
A hard spring function is assumed for Z > 0.99, such that
- = 0.99 + 0.01 ( - 0.99)/;*
e1
e
1
+ 1.01
P ~ 00506rA rl -(l/ ) I( . 1 2) 3 ( .8
8[1-e15p (e/2) " "1 p
1
(I 8.28)
1 8-15
I 8.6 CAGE POCKET/ROLLING ELEMENT FRICTION FORCES
Friction forces which arise in the rolling element/cage
pocket contacts are calculated according to Appendix I 6 for
wet friction. Dry friction forces are calculated with a Coulomb
model.
I 8.7 CALCULATION OF CAGE LAND NORMAL FORCES AND FRICTION MOMENT
The lubricant forces which develop between a cage rail and
its supporting ring surface are obtained using the hydrodynamic
solution for self-acting short-journal bearings. According to[34
the resultant of the pressure distribution on the cage has orth-
ogonal force components, one of which lies along the cage lin
of centers (the line which passes through the cage center and its
point of closest approach to the ring). Both components pass
through the cage center.
I 8-16
Figure 2.5 depicts the geometric and operating parameters
for the inner land riding situation, and Figure 2.6 the outer
land riding situation. ui, u
0
, uc are the surface speeds of
the inner ring, outer ring, and cage land, respectively. The
cage undergoes a displacement in the bearing XYZ frame of a
magnitude e and direction ' C. An xyz frame is attached to
the cage, such that the y axis passes through the point of
minimum film thickness. The y axis is rotated 1c from the
ring Y axis. The short bearing solution for an isoviscous,
Newtonian fluid gives the magnitude of the normal first terms
in Eq. (I 8.29) through Eq. (I 8.31). The second terms account
for cage elastic flexure.
Wy= Jo
U L 3
2 50
c (1 8.29)
3
w = J23/2
-475 S-
(I 8.30)
and of the friction torque as,
Mc =0 VR
2
L
8.1
C (1-2
1 /
E I7 VI (1 8.31)
where ;el = E 0.999 C
C = radial clearance, (in.)
0 = viscosity, (lb-sec/in.2)
L = cage ring width, (in.)
R = cage ring radius, (in.)
U = entrainment velocity, )(in/sec)
V = (u
i
+ uc) for inner land riding cage
V = (u
0
+ uc) for outer land riding cage
V = (ui, - uc) sliding velocity
= eccentricty ratio, e/C
I 8-17
W = cage land normal force component along line of centers, (lb)
y
W
z
= cage land normal force component normal to line centers,(lb)
z(
Mc = cage land friction torque, (in-lb)
In using Eqs.(I 8.29) and (I 8.30),the upper sign applies to an
inner ring riding cage and the lower to an outer.
Subroutine CGWET makes the calculations.
For dry contact a load displacement relationship is assumed
which has the following form.
W = XK a : = + F (I 8.32)
y - y
XK = L3/C
2
forC_60.9 (I 8.33)
XK = 0.2111 L 31C
2
C /(1- E
2
) >0,9
Wz = , -- +F z JI 8.34)
MC = WZ* R (I 8.35)
Subroutine CGDRY makes the calculations.
In order to insert the values for Wy, W
z
and M
c
into the cage
equilibrium equations, the following transformations are made for
inner and outer rings:
Mcx 1 0 0 Mc
Fcy , 0 cosO -sinO
c
F
=
c c y (I 8.36)
Fcz
0 sinO
c
cos9
c z
where tanOc (-A Z/-AY) for an inner ring riding cage and tanOc
AZ/ j Y for an outer ring riding cage.
I 8-18
I 8.7 FRICTION HEAT GENERATION RATES
The heat generated by fluid shearing between the cage and land
is calculated as the product of the cage friction moment and
rotational speed, i.e.,
qc := Mc'. 1 0 'tc I (1 8. 3 7)
where M
c
is calculated according to Equation 18.31 and - 4-cliS
the absolute value of the difference between the cage speed and the
speed of the ring that guides the cage.
1 8-19
APPENDIX I 9
'ING FATIGUE LIFE CALCULATIONS
I 9-1
. . . . .. . .... b.
APPENI)TX 1 9
BEARING FATIG[E LIFE CALCULATIONS
I 9.1 INTROI)IJCTION
Within SIIABERTt, ball and roller bearing raceway fatigue life
is calculated with the methods of Lundberg Palmgren f33jandJ343.
The life thus calculated is modified by multiplicative factors which
account for material and lubrication effects.
I 9.2 BALL BEARING RACEWAY LIFE
Bearing raceway L
1 0
fatigue life in millions of revolutions as
determined by Lundberg-Palmgreni,{33
3is expressed by
Qc I
3
(I 9.1)
loi \em )
Qcm is the raceway dynamic capacity,
the load for which the bearing
raceway will have 90 percent assurance of surviving 1 million
revolutions. From ref.133J.
2 fm1
0 4 1
(1+
0m
l
'm
3 1.8 0
cm 7140
2
fm- )Z (1
m m I
0 " 3 3
cos.m
(I 9.2)
where:
f groove curvature, raceway radius/ball diameter
(rm/D)
D Cos 61
1
/dm
I)
=
Ball diameter
*C= Raceway contact angle
dm = Bearing pitch diameter
Z = Nunber of rolling elements
m
=
is a subscript, it is 1 for the outer raceway and 2 for
the inner raceway.
I 9-2
The upper sign is used for the outer race, the lower for the inner
race.
Qem is the raceway equivalent load,
. (1 9.3)
where 0 is the individual ball contact load, and L- or C-- 3
depending respectively upon whether the applied load rotates or is
stationary with respect to the raceway in question.
1 9.3 ROLLER BEARING RACEIWAY LIFE
To account for non symmetrical load distributions across a
line contact, the roller and raceways are thought of as being
comprised of a number of sliced discs. Raceway L
1 0
fatigue life,
in millions of revolutions at a given slice as determined by Lundberg-
Palmgren, {34 is expressed by
LlOmk (Qm;k ( 9.4)
Qcmk is the dynamic capacity of a raceway slice, defined as the
load for which the slice will have a 90 percent assurance of
surviving 1 million revolutions. n refers to raceway, k refers
to slice, n is the index of the last slice, from 343
1 6 0 ,1 7 f . Z Z 2
QCW sA- 0)I -I. "- e .a. Ar -e' Cc.
The upper sign is used for the outer race, the lower sign refers to
the inner race.
0
mek is the equivalent load for the slice.
Onek =( ok) (I 9.6)
I 9-3
0
mkj is the individual roller contact load on the k-th slice and
= 4.0 or C = 4.5 depending respectively upon whether the
applied load rotates or is stationary with respect to the raceway
in question.
The L life of a raceway is given by
1,10 CLZt'= CLJ Lo M (1 9.7)
where e is the Weibull slope exponent here taken to be 9/8
for roller bearings and 10/9 for ball bearings
a2 is a life improvement factor to account for improved
materials.
a
3
is a life improvement factor to account for full film lubri-
cation which
a*
3
is less than 1 when full film lubrication is not obtained.
See I 9,5.
1 9.4 BEARING LIFE
The L
1 0
life of the bearing considering both raceways is:
,
1 0
(LlOm) e - (I 9.8)
m=l
I 9.5 BEARING LIFE REDUCTION DUE TO ASPERITY INTERACTION
IntlSland 1 the form of a reduction factor accounting for
the effect of surface asperity interaction was deduced and its
parameters were set to best fit to a large body of rolling con-
tact life tc..t data.
As employed in Program SHtABERTH, the reduced tenth percen-
tile life L10 is calculated as follows,
a
+ 4 2) h/_
- /
(I 9 .9)
I 9-4
where
Y J (h/q) = .(h/q)
(I 9.10)
1- (h/-)
S(') = density function of standard
normal distribution
(') = cumulative distribution function
of standard normal distribution
= ratio of plateau film thickness
to surface roughness for most
heavily loaded ball
(L
1 0
)= the full film life
L = a3* (Llo)f
(1 9.11)
The term (Llo)ao is calculated using the principles of
Lundberg-Palmgren
and multiplying by the user supplied product
of two factors which represent by Industry practice the life
improvement due to the type of material from which the bearings
is fabricated and the life improvement due to full EHD film
conditions.
(L
1 0
)O is then down-rated to actual film conditions
by Eq. (I 9.11).
1 9-5
:. i* .
APPENDIX - 11 1
S K F COMPUTER PROGRAM AT15Y004 "SHABERTH/SKF"
HIERARCHIAL FLOW CHART
AIPIiNDIX II 1
S K F COMPUTIR PROGRAM AT75YOO4 FLOW CHART
Flow Chart
The hierarchical flow chart presents the program
structure, listing the program elements in the order in which
they would be called to solve the shaft-bearing dynamic, as
well as steady state and transient temperature distribution
problems. The various solution loops are indicated, as well
as notes which indicate the functions of various subroutine
groupings.
Each line in the flow chart represents a program element,
subroutine, function or the main program ALWAYS. The call of
one subroutine by another is dencted by indenting the called
subroutine relative to the routine doing the calling. As an
example, subroutine SKF calls subroutines FLAGS, TYPE, PROPST,
LUBPROP, LUBCON, DATOT, CNVRT, CONS and SPRING.
Subroutine CONS calls CONST, CONST calls BCON and CRCON and BCON
calls ABDEL.
The first mention of a subroutine within the flow chart
includes the entire list of subordinate program elements.
At subsequent calls to that subroutine the list of subordinate
elements is omitted. As an example the first call to subroutine
AXLBOJ is followed by the subordinate elements JHVIKT, SNITHT,
NUMILOS, DUBSIM, HEIE, MEIL and SIMQ. After the call of AXLBOJ
from INDEL, the subordinate elements are not listed but are,
nevertheless, employed. The list of subordinate program ele-
ments are omitted in repeated calls of subordinate GUESS, BEAR,
SOLV13 and DELIV3 as well as AXLBOJ.
As noted earlier, rolling equilibrium is calculated, first
without, then if required, with friction forces included.
Whether or not friction is considered is highlighted with
the words Frictionless or Friction beside subroutine BEAREQ.
If the Program is too large to fit in its entirety on
the user's computer, segments of the program may be "overlaid".
For this purpose the Program is subdivided into ten (10) modules
which can be sequentially "overlaid". The contents of the ten
modules are listed below.
The Program segments SKF, TEMPIN, SHAFT and GUESS all
perform initiation functions and with the exception of GUESS,
are called only once per program execution.
The real problem solving portion of the program is
embodied in segment ALLT. Within this segment the shaft bear-
ing solution is obtained through the call to SIIABE, then the
steady state or transient temperature distributions are obtained.
II 1-2
This scheme is repeated until the end objective, steady state
thermal equilibrium or time up for the transient scheme, is
realized.
The nonlinear equation solver SOLV13 is central to the
program and deserves special discussion as related to the flow
chart. The first call to SOLV.3 is from BEAR. Only for this
first call are all of the SOLV13 subordinate subroutines
listed as noted earlier. These include INSOLV, EQS, PARDER,
SIHQ, EOCIIEK, ERWRIT and ERCHEK. In the subsequent call to
SOLV13 in which the steady state temperatures are being calcu-
lated, the above listed subroutines are again called but
these calls with the exception of EQS are not listed on the
flow chart.
EQS is the name given by SOLV13 to a subroutine which sets
up the system of equations to be solved. EQS is brought into
SOLV13 through the argument list. When the bearing equations
are being solved, subroutine BRGGEQ is brought into SOLV13
and within SOLV13 is referenced by the name EQS. When the heat
transfer equations are being solved as a consequence of the
call of SOLV13 from ALLT, NET is brought into SOLV13 and is
referenced as EQS.
II 1-3
NO. AT75YO04
PLOW CHART
ALWAYS
IKF Road and set bearing and bearing solution
LAGS control data
TYPE_/
PROPST Set lubricant properties and calculate
LUPROP constants for temperature dependency calculations
LUBCON
DATOT
ITLE Write bearing input and hardcoded. preset data
IT-AT
CNVRT
CONS
r BCON Calculate bearing related constanLa
ADDEL
CRCON
SPRING_
TENPIN
INDUN Read and write thermla and thermal solution
RWHTC control data and calculate heat transfer
RWG coefficients
THAp
SHAFT
ARRANG Read and write shaft gometry, loading
ORDERR and bearing position data
AXLBOJ
IJNVIXT Calculate shaft deflection constants
ISNi m
UNULOS flake initial guesses of bearing reaction loads
DUBSIN and displacements
IFIEE
RACT
INDEL
AXLBOJ-
PART
PAR . Calculate shaft influence coefficients
UESSXLBOJ
GGBRG
GROLL Guess values of bearing variables
GBRG
GBALL
VARRDC
GUESCG Begin the solution of the steady state or transient
ALLT __ thermal and temperature dependent shaft-bearing
SHABE anslyses
FIT'
INTFIT
Calculate bearing diametral clearance
SIHEQ
SONRI
3BEAKC
EAR
PREPAR
1INITX
Establish iteration scheme to satisfy
UNLOAD
inner ring equilibrium
X14IN
SOLVI5
INSOLV
EQS - BRGGEQ
EARFO (FRICTIONLESS)
GCIRL
I ALLIN
I IALLEQ
ROLLR
I
IOLLEQ
I TNOR1.
PARDER Calculate rolling element raceway normal
EQS - BRGGEQ forces
SI ?4FQ
EQCHEK
IEQS - BRGGEO
IERWE IT
IDAHPCO.
ERCIIEK- Sum the rolling element forces and moments
pSUNK acting on the inner ring
LIFE Calculate bearing fatigue life
BFILL Add bearing inner ring forces and moments to the
SHAPA
FILL8
shaft equilibrium equations and predict new shaft
SIMEQ
displacements
BEAR_ CElculate rolling element raceway normal loads
GUESS _:ith the new shaft displacements and guess
VISC0- bearing component speeds
ALPHAO
DRAGNO Calculate temperature dependent lubricant
STCON
. properties
EVALUT (IF NPASS 1) Begin the calculation of bearing friction with
BRCGEQ the guessed component speeds
BEAREQ (WITH FRICTION)
iGCTRL-
BALLIN
BALLEQ
B INT Calculate ball-race film thickness plus the
THERFC hydrodynamic and concentrated contact
TARPC frictiom forces
shalt .,III) t Iq3r
iun3
ISLMEQL
displacements
gEAR _Calculate rolling element raceway normal loads
GUESS with the new shaft displacements and guess
VISCO2 bearing component speeds
ALPHAO Calculate temperature dependent lubricant
DRAGNO properties
STCON
EVALUT (IF NPASS 1) Begin the calculation of bearing friction with
PREPAR the guessed component speeds
BRCGEQ
BEAREQ (WITH FRICTION)
BGCTRL
BALLIN
BALLEQ
FMIX
TINT Calculate ball-race film thickness plus the
THERFC hydrodynamic and concentrated contact
STARFC friction forces
HOHI
HDFRIC
ASLOAD
FRINT
EHIDSKF
- FRICTN
ROLLIN
ROLLEQ
TNORN
I4XR
HDFRIC
THERFC Calculate the raceway normal and all friction
STARFC forces acting on each roller
HOHI
ASLOAD
EHDSKF
FRICTN
CAGESP Calculate the bearing cage speed
CAGGEQ Calculate the forces acting on the cage
CGLAND
CGDRY Calculate the cage-ring land forces and moments
ICGWCET.
CGRE
CGBALI
CGNRHW
1 CGEHDP Calculate the ball/cage normal and friction
CGHDP forces for the ball in question
CGFRN
ICGNOF
JCGEHi
EHDSKF
J FRICTN
i CGNRMD
CGFRD, D
E SUNRE
Calculate the cage equilibrium
equations
BRAX
Calculate the rolling element inertia terms
LIFE Calculate bearing fatigue life and bearing
LRHS) heat transfer coefficients
BEAR IF (NPASS - 2)""
PREPAR Calculate component equilibrium using the inner
4ISOLVXX ring positions determined with elastic rolling
EQS - BRGGEQ element-raceway forces
II BEAREQ (WITH FRICTION)
SONRI IF (NPASS - 3)'
EBARC Calculate inner ring and component equilibrium
EAR using friction as well as elastic component
PREPAR forces
14 ISOLVXX
III -BRGGEQ
SuMFBEAREQ (WITH FRICTION)
DELPV3
TITLE
RITE Write bearing output
REOUT3
RITE2
FILLGT
SOLVXX
BQS T NET Calculate the steady state temperature distribu-
DELIV3 tion and write results
STEPNA
NETEET
DELIV3
NET Calculate transient temperature distribution
NETEET and write results
TNAP
DELIV3
TAP
Solution Loops
1. Steady State and Transient Thermal
2. Change in Clearance
S. Shaft-Inner Ring Equilibrium
4. Rolling Element and Cage Equilibrium
S. Temporntsm i g
41
hru
II 1-4
APPENDIX 11I 2
S K F COMPUTER PROGRAM "SILABERTH/SKF"
INPUT FORMAT FORMS
11 2-1
- S S
-4
-4-J
4j
-4
4*
D
5 0
to l -4
1 -
44
1.4
0' o
II2-
IPASS arn IuinJ iJfl
I4T i MitRAL ROPERTY FLAG
20
A-A
a
v" A U
110
a - a
o 0
A
0 .
u do
.. 0
.4.
16
cd
:~~* a :
kIa -
0
r. -
H .o~C
cd~S
0
4-4
as
a)a
04 4
03 on
u o
uj >1 c
-0 Ci
2-5.
ItS
I t
2L
.1 - '
1 0 Is
u rg
0 -U-
co
On ,
I1
~~
II
~i U
mlI 2-6
I-
1 ___
V -
I -
4, 4,
*1
I.- a-
4, 4,
C C
41 -
-i
C, S
3 C-
-3
S
3
if 0
4 4 *4,.
C. C-
4,~
- 4,
-, 1 @5
'C - - -c
aCi----- C
MI
.51
,CJ S.-
4, @1
- r
- -0
- Mt5
U
C, S
0 6
C- t
g
U ~ 4,
I I..
- 5-
-. 5
Sm
-5'
< ~ 0 9
S S
0 I *
C, .. I. a -
0~. * :3
(.~ >-, .,- S * a *
Z H - U -
- 4 4* 4 - 4'
~ ma S S
~-* *~M
4
- - *4 -
~U - - 0~
* a
~ Urn - 6 - u
II 2-7
0~ f
2
4 4
3~ 3.
0
3-4
2
9-
o ~, 03 -
0
C *~ ~
2
3
3 -
3~ 3- 03
-0 03
3-. 3
-~ a
0
.7 -~
03 03
03 7 3..
e. 3- 9
0 .0 * -
4 4* -
4 - U *
* C S
4 S 9
-3
03 .3
3-3 4* -
C 9
0 -
* 33.
4 9 9
-, 3- 0-5
* *
* 4.
3. 9
N =
A. C
U
3-, 3~
N N
0 p.
S
U
0 9
03 ..- 3
- 03 4I
* I. - 41
* 9
3- 9- -
- 33- t -
I * .3-
4.* CI C
- 94. - C
-j J
- -
0) 0 *fr N
- 94.
- -~ 5 03.
9
0 -
03 43
4?
F-' ~ .. -
.71: 3- 0,11
* 9* 3. q~i
0) . 33.
CI 9 9 -
= U. . 9-5
C.~ >.. - 93. 4 5
? H: :
- 4 .39 43*
C~ ~ - .9 NN -
- -. 3 - - .4
< ~ 0 9 0~ 03 3-
- 5 * 033
~ c~)
II 2-8
.~ ~
A
i
A4
A
41
4 -
C C
4, -
CC
aq co to -
73 4
z
4
I1 1--
Cdd
-4
4-j
o c0
o0
La
11 2 1
CU S
440
0 - 0
Sm
o o-
2-1
AA
C131
u 75
u A
4J.
0a
C13
-. -BL-- w
CO
pq u
M ri 4)
-4 Q. 46 ft
3"'K .003
oi a a - via
u am
- -
11 2-1
2r1
-d
Cd
-A-
A
A
-Im
CUX
-2.)o
=L-- I- -J
l) 3a T- .
0121
Cu
-o
s-4
41~
.0
P-I MU
at A
*1- -
11 2-1
PL I
92
0 A0
0
-i-
0 OW
,-4
C C
zU
4) .4. .n U
0
-4-
10 -co
U2,0
U3 ci
52 15
IR
'-4
41) - .
4-J
- 4
uHI :I
-4
-4 'I
0
H -4.
A .3
4j
0
~11 2-16
4-J
0
W
.
Ulo~
0
a
-2-1
4) a
4)-.
II -1"
w -
-I-
*
$
o o
En
4- ,
.! -1
- -U
z - -
a -
t . _,~
* _-
- U
-- U
P-4
.3
II2-20
CS
s
C-)
112 2
Ii: 112
CIS{
CI!
II
0 r_
um
ce . u
CU -IIa-1B
0 ii
4J "N i -,|
>2 -- "
, -; .a -- -- -44
'C .. I C I !'C
!
r r-2
CU2-22
.4
u.1 2;~ - N
V Iall
I 'U
44j!
~ E~
A ~ ;- ' m
A.L
4
- i.
It
o
u'
AA
1122
4.
cJ~ U
z
z -Ii
- .4
E-. ~up
U p.,
p.. 3
'I
U
4.
z -
.4 -
I-I 4. 4.
i _______
z
H
A
o
~.L. U
U
VP
z
0
- .5
U -I I
4 p..
-3
- -a
U
S
H
4 -3
w
- U
z
2 -
o G)VP 31
z *~ - - U -
o *~ -- a
~ 2 ~ -
H ~- *
U *4~U
~ U *- .~ - C,..
.5.- Se.
* p. .- U US U
515 - .5 3 -- 'U
i
- 0 - S U U.- 4p ~ 0.
~L4 - S *~SS Us-~ U
~ -- *.
- h--U U0Uh-SS
H H ~.3p..UPU~-U
TI 2-24
b.0O
4 4
0 to a
- .
co 0--
-'a -..
Ls) @10c
Cdi
* -
- 4
*n 0a4 Ur
**5 2-25
1
o
o
X
0
0
"44
o
1
0 .i
0 - - Nli:
0 0
m!lI l
z c
co to!
w tntAll
I
z o -_ _H -26
o -m
UT -Z
- I.... ... . .
0 0
-1 0
00
UU
C - I
1.4 0.-
V. (n [C
-11 2-2
.4m
a).t
(-.
4-4
3.-0
4 ')~
CU 1 * .
a) :i0.( .
4. 4
C,,
.4- 4. . .
-I 2--
CU * L
- s- - .,
I as
z
0 u
N-C -,
- _ -
u al
C44
0U
-
Cw. . .
r. S - Z
0
z "C
* o: -.-
Cd
go
1.14
.... D.... ...
z " 8 U 3
t--+ ,4 e
. w _____
E-, u
"-r CU "
CtJ C'
o j
Li.
= - ., - ..-
T T - +
APPENDIX 11I 3
COMPUTER PROGRAM SHABERTH/SKF
SAMPLE OUTPUT
I1 3-1
101
I . I
NoI
a-A
4n
3,
X 0
W: W
I-- N
U. c
P4I z
UK P- "
IL'
kL- L
U.0
Ge
1.- a 2I. M
In .U I x .)I
0 ) 0
In J In C
U) ) 0'l w . j.
-jz ., InUO
v "W
.u..a z CD W
Iii In I- U 00i
I-' 0"0
40 z3 O'D
1F 4W . .-41 (..
CU GII-t % I;c
1. 101 C. 0
j. -C i4 4
(L I-
z 14.1 111W X00
w I W 0 NC
" Wz ZW no of 1-0 j- d 2
I L I- 1- * 0 0. F
I- I- Wi HIn~ 1-h U64O5
o) I-. X 0
0-4 1-..JZ
m 0- 4 1-O. I of 0
43 Q L9 6-Z
w Z/-I x w C
0 V)- i 1-1 xZ Or
hi fiC)a w 110
m C4 w -
w 0 L
C , 1.1-
w i mus1 w
U. 00) hi%
a. zi
s-o
I.j v Inf0 Lu I-4U
.JU)III 0- 1.- .. 2. 0
x I 111 0 U)a mm j
f- ) *~L I- i 00 11 IN z I
IL m C... 0. - 1 LI 4 z m
I- 0 ~I 2I-
H I W U
44
.9
I- Il
m I tnl & z MIMc
U. 04 0Z .2
-j l " .0
4e CL D.HU 0 0-w 0 U% *
.9 In -l mo .4(. W
I-. U. Z.1 - 'a -8 41 .44E' of &
* I. r x I.- 0 4 02 a. -
* (A -X0 wOWZC
- P. U ce wI 3-3
w I!
Itt
K.7 I=
.4,
99 1.
Q0 G
Wi .41 o e
"J z Z %L '
w w a
0., -4 W. p.4~~
C*4 a U
vp 0 '
I) LII -9 A. *
(A Z A j34
.494 a w I I
i Ix
0 i0 4
a~ ~~' i i ia2
. 1 a l a Hm 0 D
o3 -1 H i WW a t
z ~ ~ c c~~.. a C, W C
I ~ ~ ' hi z-0 .h
C. I- = OZ ca 1. 31' cmi
* i IV 3 ,w M
w d* a 5 I
~~~a x wi -4 z.z aa
4.) 3~X- OW ;r ~ - .
U P. w 1 I .
Of -j Ui4K P
H04 IC a
*0 I- C3 .
_j * W W CS M d U
M Zi
C.:Ih s a ' a h
H IL .11 h
itH~i IA. * '
Iz 0
59
!'
i
wz 1W
ag m 43
w~~~ pU'
c~w w
2 ~ ~ ~ I 00%- 2 2
I.
w
0
. H I. 0
,QOe
ILI
w a~
W
II 'N
I~. z
C9',
z 0
z *'t u
1 2 z DC
IL
Ni z
43x w
1 24 U -%H
'Il ON- w
.1I
x
"-
+~2
u113 0 6 LI.
~~~~~>
- 0:9. Z -'
IV if% 00 w
C2 4
ZHz
z
I)
IA 4
IL IL b K
20
2
W '4
49 " 1
31
.
141
U --
F I
I I I
I~0 I I I
co %D. 4 g av
C2 cmc 3a
10 C3 r I.3 0 Cai)
ca a w- vo %.4 I4 0 %.4
C3 c C3 *~cca
00t~ U% I C.... 3 I -
Iii ca
Ia o
0. T: w .4 0; C34
z w pq
V)iE 3 i; C I..
4 1
) in 4 O 1m
" if. hA 0i .O,. 'CD 40VC
,'4 _4, U
15 0, * * ~ ,
0 1 o OW lz P;
F
ae a' .9 .. 0 L3
at of Ci
-4 ca ca. %2 * '
r. 1 CN 141 W 1 4 -0 t
1-44 Ii) 0 , ;W 'wN~1.. Z I).
~ h A. ha , a 'z fw * cu * I
w a. . C i sC2JtvC
) U CX *4
15W o
!, em ca 0- I ( -
"U 9- C3 W 00 N0 41 NL Z 'I
I 9- U% 0 1-4 C a #IL 0 C3 C2. I. * C3 . 14 ca I
v w 5 '0 ha 0 i c:m ) 0 . 1.
IX I-d mm ,a , a' . .4
U K I63- L 0
U. IA ei. 0 mf
2 2 -a- 0 a- It a,4 .I ,
tu w " I. w w -1 "I 9-.
214 p- f-
WO 0 4
4 O 0 . al of2F 0 haI V ,I&. 9- w
a 1 Ii Ii 09: Of6 31J a-~I U. ii; I I I .'F
IL, I, I xI - LX ,r LXL
I In
+ C2
44 I
~.inII of
C3-
tjI
w -9
.,
A w
j
le w
3.. z ,
0 -1 -C, o wi
xa a. " L
z z Wn Z
H in Id
vi -9 Int
x 0 '~J1
H 0 4.
x a.
-9 -L - I
U, V) 0% Ino4Nr - n 3" 40 .4. 04 0g
CJ z - 0 UN4H C* U4 4 CD 34 " ., tU % aG
iaw -a. IA Q0 o .4 1.44.4 4 I
1 0 A ..J Q- f .*11 r- Um% % 4 **140 4Da
u c ~ cm C C3- ooaa 4243= W.,ccpap mU0"
in z 0 2 D 1; ; 1: 4 14 ; *; 0 C; C; *; 4 *3 . . . .
IA,
1.- 0 do in WI1- Co , , C I C!C:C :C IlC: C C. 1:mma
" j V, I*~ga cm mom emaaac00C
,a INC Ja. In W * W C F
co O
in Ia- I- m * r-t- %
In
A
N
I
1 -
F.C.I
U.
01 -C ". I
.0 II
hi .4 .4 .4 *4 F .
" zI w i
ce o
b- 0 I
%16I
H i It %*-J w
0 0y
.I In
x I I
64J .U 4 w C, w . f
I- j IL
SAN U% 0 I M 4
kf x -
0)4 24 Co ft6 0 *S a a m a a m ~ a U 4 aa
r 0 - 40' tC..4
CI* 1. 0 hib at ~ m m a~o a U)UU)
z i 0 4 b-ly @ 9 9 9 9 .. * e . . ..
z*ZIZ 6: t:i' C: 4: 1: Chi 43eA~ftje . C: lt l! It
z *j IM .4-4aca4 D .0c a ade pc oa
w IS
4D w SAJ Z 0 I . . . . . . . . * @
No H 7 A x H I UA4 @ i aC343 2 c c I
0a 0- 06 1 - o@a a~ n P)m UN #Aa0
IL 0- II. CA. 9 9 . I9 * * . I
* IL -4 h 1 i I
.4 rY M. t4. 0 a@ U 000 m3 aaa-4 @I3pw% o 4 c m.
4 4 04 -0 04 HZ 949.9 9 ** 99 9 v4- 9*yt
1I 3 -
I. I
I I I
i 1
4
NI W I
W . 1.- WI
0U
14 v4~ 9-9* c
.04 I 1
.34 IL .4 -K M, *Q 9 .
bi I ~ IL
I- ". U. Ii . ,a
I. - II , . I . 3Ar 10 I
0...I w 4* w
z 5 @4 IX O0f N I%4
=1 = I 3 1* I I I
IQ t-1 C.3 .
9-, %a .04.3) Z ..j
1
4@
01 0 . IL II
9- 0 .44 4 w Of L;
P. - C2 %ml Jul in e 11 IL
L K 4 O0 ly 9 4 *
o K - 43. 4
104 C- ' 3 co w2 W
w , W, " 44u 0
C. I- N* tw.w 0o UN I 4.
C) 9-go 4 N3- c at I- , a
W~ ~~ ~~ 9- a. -l D fl9
*~~ 33 w2 9 - .
4W. ' 4' 36 -C in W I"
)-- a tAcc 0 ") W e
03 .0 (Y . J .4 Z
m 3p q a C2 ) W * %E Ch! * .
4 w IuP 1 3 4 4EI.-
0 z0. .03 14 w &
K4 U% Ma W W 9
at 0 14 Z WW N04 W .
w WW: ~ U 4 W 9
0 p .. z.
o @ 9
0 4 w K . -
.0 4 lLm (Vt (
0 t, w
0
w t3
0 Ie
ul -.
$01 4-
0 u 'a * 1 " N
I.4- Q 0 U
z 1
u -cc xd 4%14H
0 ca m4~ 0 0
o o H ..
~ IL
* Li Id0 0m
D4 W ! 9 4 0
ce -Wu I p
C3 N o-%
1,4 (4 0~ IL . 0Z 11
z4 a4 In In I
H : 4 IA 0 -3
0U
H:
4 -I. 2
.9 w 4 0 a
Inz xI I.-~ :
z Id 0 li
U x I .. - 0 B,
xn 4- -8 1 Wa Fn0. In
2: LL 0 Q 0% 0 1 11 1
4- 1- 0%0%&)
IEI.- 0. z
x 14 CL zL '
j 1. 0 . .(U 20
I.-
3;-310<.0
w w~
I, I. P) t
~t 'tVi N ':
w~ Ill Il w Ii
0C. WWc a
(n~ ,t-.
I~~~~ UU IIl
CAz
*~~ ~ 4'. 0~~l
U% I2 caW
8-4 :04 EL V)r~
us iI I I %' I4. = cm
U.. H I I
1- W 1 I &~ c.c
1
f
6- ). H4
LJ O
0~ I-.
w in 1-
z w ,
zl 44 fn ~
0 b. w 0 I -4 .. 1
)- 0 I I *9 * w
Uw 4.12 C3 * 'm L
e * C' CO C3 Z z .
ww 4 4 . t- I.-I
4.4 IL1 41- .40c
I~ w 1
CID 1- .j
Z 044
0-4 1- WWI W ww 0 ~z N
H Q 4 ' w 0 -
dig M 04
,; LA *I N
14I
* 0 I IIT 3-l1!
I0 z
w -x
ia~ I(%am ammam maCJ
.1 's 14 6.4 4.4 4 IA.4.4. 44. 4.
iCi
M F,9 ( n 3U
04 z 0 %' a ' %'aa maw m ,D amISIS1
H~6 coc ON *m .40 w %c4p..4
0~ j
z )qpL 293rDC Cl 10t = calO
2b w
*t z P4 2C wC 3 Dc *c 24
0 1-
z In
I z a m a a m ' a
2
2i ,I C2 4 0 (y waa 3 40 10a a 0
0% zmo mammam% % % r "Mmmmm
-* *- f I tt 1?- rM-
HZ -I 4AI
I. w t . to .4aP) 4 amm Naaa
of 0
I tJgimD N~g.QM~i'0Q
w -. N hFiNU
1- 9 j. 4* 4 m w
IDJt.I~
K.0
116~a N4a~~
2112
I I
m ca mo xc *C 3 2C
**4 2c 3c 0c m ca eODOOO2oOOO ,
CL 0. o0. . . * . 0 . . . I
Zy C3 - CC, C2 C
cy 0 GI QUI 0 0
Pi- U% UN0 t0 300
Is,(40 C C, a000
, -J C,
044
.4 4- 0 N i
0. w. 04 UN= 0N4 0W tN0
of
w
oU
41
N
I n n e ' o O D 0 m a m 0 0 0 . 4
@000 *3 6 0
(4 4; C: w dmo A 42caca a 4
z hi
10 C3 .11
0. - 0K* ~ @ 'i
Z~ ~ In N In 1- er0 .It.p.
U I U 0 14
inL IL
I 1u 321
I
od I qI I 4 q 1
ii
.
4 do WNnf-
kaN 16z .,NN o oU. &
upW $- O P4DWd O D ;4 4 ;0:6 I ;0:Za
U N N N N N N N N N I
4.Z C9ZZZ
I-I
(* I-.
Z4 0' -W adi ca % U% 4 go.4
of ~ 1- 40 10 % P N .4 4.4 W Ut0 l.
Wn -J
0 N....% * . . *: P: C: i . I %
b w
a I
A. z I~ I
1,- 0- 3. 4 W 4 4 M * - 4
oz 0. -4 -t C f0D% 0 M .O340I
o4 .4 114,,4 .4 .4 . .4 V4 f .4 * l *. . .q1
0 io m wN M% M, N AU% NP.- m
ad*N 4P aP .4 mA* %U%Pl.
w N.0 % N N 0% W)w4 f .3,% s.4 N" t N r
le( **** *..
Sal ~ ~0 -a'a0 'a [email protected] %0.D
NNNNm''AN I N EJ
0-~
PI eg e e e g e s g
'D - . .t .4 . * .4 04 N N 4 N F. W)
w 1.
. 3.01 . t* **
(A f -l)N ofn
Q U- T 4 * v* N N N 0%4 N,
4~r cu LA %agN P
N~ I- UN I . r-c A
Iw LA4 4 nU,'
C3 C, C, ~r.Q*C,,
z wi w ;j N LA %Nr U, mC N C C4 aQ
ou 2u: z
w 14 4. j- . 44N t.4 4 f CN0MC4 4 4 .4 m
14 CN *
z W t jc 41 N 40V N~~
0 $.- I - I
6-4. (4 Id 1; v: * . .
(/. 4o I z %'OAJ N N Co oC3 CU "~ UN MNc o
m- P4 RI t in UU
~~~~~~ Id 44-.. v.CJ ' 4Q .4 4 OC .
M . .4 U,4 M4 t1%Lf - 4 .4 .4 4
1: : t UN ON *0*6ca4O1N0
o- " N .mQ1P
I z *** 99 99 9 *9999494 v
z~~~~~ L)V ;I, : C
0 * N 4V- 0 4 4Lf 4N C
C 4 I.. I- 4. o9 %0 000P ,NWO PO 4
Vl %o P, N. 10 ( mP f 1r)- n' 0
.4 ..J .4 .4 .. 4 m C D C3 43
0 z
4 csc
4aP
* I
Co In
'l 94* 444 1
CL . In V)
0 c
Coo, W 0 z IV N U 0* 0
.4 0 39 1A 1. - Go
& Q2 NO . 1- 0 f n.. 0 .4 1- 'IW
-%) V1 CIO a.. 4-1 * 61 c,, WuI w
Q la - .4 e* r41. 446
-1N Ad* 4:1 " x0 1 a I Z, % %
=. 0 op. N . 8 .1 * Z .22 .
OF' 02 A, *: #A If)l Na Wa x A.. 6(1C00WW e
on .4 .k IS)-C In 001 1* 0). -jn 0 W
c0 sa uA* U, LC2 US us .1 AA W , 11A Ul kf 0 bow46, NO .
z 4*9e .. q~q I. .. .jW 4 (W@N 00 C
20 -A 62 *j *j -j 0 WA,- 016 a *4 K *4N'o 4
xJ -j .L) 14 LO20 01: O)_ I* *WJ *0 .C 1)~~4 .1 z IuN %, Q4
*CJ01L. * 1 A.,2t9- I. 1- SW1zs on W-Won "OWsIs9.
0.E S. 4.11 "-. ~ * M .4U 0 ~ CA.
ix. a .J f/I W.~J ~ 6 a0 1-00 041.0. I4 In
Orm or CA m a~~a u4 i. . a.. 3-.CJ 4 $ f 1 C On 0.8 ^r1 UH E U . 00 Co *4 1
'. A "at P. x . W . 6.- 0: *@ mu(o0 IO ss1@50 Pi 0 aamammaaminCmaa%*
-j I O .1 U, W 0 0 l44.u .- 44..46.4..6. g*1,.u'5sa C; Ot(atAA Ag (~ 43 *4
4 ~ ~ Z alC NNN N N NJJAP9JN N.~ 9JNJJNe~~5J*J44 CL 0' %-1
*U.SGWveWOWn PrintinguI,- O1f.- 0 7 101 1--01