Int J Adv Manuf Technot (19%) 11:232-239 1996 Springer-Verlag London Limited

The International Journal of

Rdvanced manufacturing Technolo!lu

Spindle Deflections in High-speed Machine Tools - Modelling and Simulation

Subhajit Chatterjee
Industrial Engineering Department, The University of Tennessee, Knoxville, USA

This research attempts to develop spindle deflection error models for high-speed machining systems. A model for determining total spindle deflection at the tool-end is presented. The model incorporates spindle bearing characteristics, shifts in ball contact angles, and centrifugal force and gyroscopic moment effects at high speeds. It uses the transfer matrix method to determine the total deflections at the tool-end based upon the point contact deformations at the individual balls of an angular contact ball-bearing assembly. A simulator is also developed for simulating spindle end deflections for various spindle rotational speeds. The results of the simulation show contact angle variations and peak deflections at particular spindle rotational speeds. Important research issues are also presented.

is organised as follows. The next section reviews pertinent literature in the area of high-speed machining and machine tool metrology. It is followed by a statement of the objectives of this study and presentation of a theoretical model relating spindle deflection to spindle and operational parameters. Initial results are then presented from simulation runs and finally potential research areas discussed.

2.

Literature Survey

Keywords: High speed machining; Spindle deflection; Spindle rotation simulation

1. Introduction
High-speed machining (speeds above 5000 r.p.m.) is emerging as a powerful tool for increasing productivity in finish machining [1]. Industrial studies in this area have evolved from early concepts in the 1920s and recent advances in the development of computer control systems have provided the capability for accurately controlling high-performance automatic machines. The evolution of these high-speed machines has been in parallel with progress in the field of spindle and machine tool design. However, research in highspeed machine tool design requires attention in many areas; particularly important are: the thermal growth problem, positional errors due to deflections from high-speed gyroscopic moments and centrifugal forces, and the changes in bearing stiffness characteristics at high speeds. This paper is the result of preliminary theoretical error modelling and simulation of high-speed spindle systems and

Various types of machine tools and machining centres exist and are classified for ease of specification. Some machining centre classification examples are: horizontal spindle fixed column; horizontal spindle moving column; vertical spindle fixed column; vertical column moving column; vertical spindle fixed bridge; vertical spindle travelling gantry; horizontal spindle travelling column with tilt rotary table, etc. The major motion directions are in the Cartesian directions and, additional degrees of freedom include table rotation and tilt. The literature survey will focus on research in four areas: 1. 2. 3. 4. Spindle deflections at high speeds. Error budgeting in machine tools. Errors due to thermal effects. New material possibilities in material structures.

2.1 Spindle Deflections at High Speeds

Preliminary work in bearing nonlinearity and spindle bearing stiffness analysis has been reported by Shin et al. [2]. They show analytic and experimental evidence of changes in stability zones for angular contact ball-bearing supported spindles at high speeds. According to their study, changes in bearing stiffness at high speeds resulted in spindle instability. To corroborate their theoretical study, they performed experimental machining tests at a maximum spindle speed of 8000 r.p.m. ANSI standard B5.54 refers to CNC machine tool performance evaluation (of which spindle error profile is one aspect). Researchers have also worked in this area for quite some time and individual work with respect to machine tool characterisation has been published [3,4].

Correspondenceand offprint requeststo: Subhajit Chatterjee, Industrial Engineering Department, The University of Tennessee, Knoxville, TN 37996-1506, USA.

Spindle Deflections in High-speed Machine Tools

233

Researchers [5-7] point out the importance of the stiffness of a machining system and the spindle in particular, to tool life and process stability. Particularly interesting is the observation of improved surface finish with an intentional reduction of tooling stiffness. Enhancement of chatter resistance through a reduction of stiffness has also been reported [5]. It should be noted that a majority of the reported test results are not in the high-speed regime. For high-speed machines, some important requirements are a very high degree of damping, very high accuracy (low radial and face runonts) and consideration of spindle and tooling structure deformations [6]. The effectiveness of tapers at high speeds is also important as increase in taper diameters at the front lead to changes in axial position of the spindle, and consequently, the cutting tool [6]. There is a need for formulating spindle error profiles and quasi-static deformation of the spindles by considering centrifugal forces and gyroscopic moments of spindle mounts. Additionally, the use of wear-resistant bearing types should also be considered in high-speed spindles. 2.2 Error Budgeting in Machine Tools Error budgeting is a systems analysis tool used for the prediction, control, and design of machine tools. An error budget can be used to control the individual subsystem errors given the total acceptable system error. The error budgeting scheme can use error sources and coupling mechanisms in conjunction with workpiece categories to relate the error source to workpiece errors [8]. In this study the error source is the deflection of the spindle and the coupling mechanism could be spindle properties. It is also important to determine the proportion of the spindle error in relation to other error sources such as servo error and displacement error. There is a significant body of research relating to quasistatic error formulation, detection and measurement, and compensation for machine tools [9-13]. The three common methods are: 1. Measurement of the twenty-one positional and angular error terms independently and compensation through interpolation. 2. Correlational models built on coordinate data and trigonometric relationships. 3. Formulation of error envelopes using rigid body kinematics. The rigid body kinematics technique uses rotation and translation of the links and joints of the machine tool to formulate the error envelope. Compensation for each error source is then accomplished by determining the rotational and translational coefficients through measurement and then predicting errors from known positions. However, all of the reviewed work in this area addresses the quasi-static errors; therefore, there is still a need to investigate error envelopes for dynamic errors due to spindle rotation and positioning at high speeds. 2.3 Errors due to Thermal Effects

of dimensional errors [14]. There are six sources of thermal errors: 1. 2. 3. 4. 5. 6. Heat generated from the cutting process. Heat generated by the machine. Heating or cooling provided by the cooling systems. Air-conditioning effects. People effects. Thermal memories from a previous environment.

It has been shown that the heat produced by spindle bearing friction in machine tools has a very serious effect and this effect has been investigated by many researchers [14]. Analytic and finite-element techniques have been used to formulate thermal effects and optimise machine tool structures for minimising thermal effects [15,16]. It is important to extend these concepts to high-speed machining and to examine temperature rises and resulting deformations at these conditions. Additionally, the wear of bearings at these high speeds and elevated temperatures should also be considered in these studies. 2.4 New Material Possibilities in Material Structures Advances in material development have prompted research efforts in examining their potential use in machine tools. Rahman et al. [17] showed that cementitious composites when used as lathe bed material exhibited significantly higher damping ratios and higher first natural frequencies than their conventional counterparts. Another study by Lee et al. [18] demonstrated the successful use of graphite epoxy composite as a spindle material in machine tools. Their analysis of the spindle bearing system showed about a 20% increase in damping of a graphite spindle system with the consequent higher natural frequencies (and delayed chatter) as compared to a conventional one. They also examined thermal stability of the spindle system and found that spindle expansion was almost negligible because of the almost zero coefficient of thermal expansion of the composite material. This is also advantageous because the bearing preload characteristics can be better maintained when the thermal expansion is negligible. However, the experimental work in both the above studies was conducted at cutting speeds less than 1500 r.p.m. Therefore, it is important to investigate new bearing spindle materials at high speeds. Figure 1 shows the highly interrelated overall

l[

ERROR
I

GEOMETRIC STATICLOADING THERMALEFFECTS ~

MACHINETOOLVIBRATION CHATTER SPINDLE VIBRATION 8E.ARING& SPINDLEDEFLECTION THERMALEFFECTS

The thermal error problem has been recognised for a long time and experts view this error as the largest single source

Fig. 1. Overall part error sources and error hierarchy.

234

S. Chatterjee

part error sources and the error hierarchy. The scope of this work is limited to dynamic error modelling of spindle end deflection arising from bearing deflections at the ball contact points.

3. Modelling of the Spindle System

The study was undertaken with two initial objectives: 1. Formulation of the error analysis and deflection model for high speed spindle systems. 2. Examination through simulation, of the effect of different bearing preloads on high-speed spindle deflection. The present study attempts to model the spindle bearing system for deflections at the bearing surfaces by taking into account spindle and bearing material characteristics, rotational speeds, bearing ball centre migration, centrifugal forces and gyroscopic moments at high speeds. The deflections and evolved forces are then considered as input conditions in determining, by the transfer matrix method (TMM), the total spindle displacement at the spindle end (tool end). The spindle-bearing system will be analysed first and will be followed by the analysis of the TMM. Fig. 2. Angular position of rolling elements.

J=s

3.1 Analysis of the Spindle System

In this work, angular contact ball-bearings are used in the spindle-bearing system. In the modelling of the deflection at the ball-spindle interface, centrifugal forces and gyroscopic moments are considered in determining the ball centre migration, and consequently, the contact deformations. It has been shown that because of bearing preloading and high rotational speeds, a shift occurs in the radius of curvature of the inner race while the outer race is fixed [19,20]. Figure 2 shows the positions of the ball centre and raceways for a particular angular position of the ball. From this figure it is evident that a shift occurs in the inner raceway groove centre resulting in a change in the contact angles as shown in Fig. 3. Such changes in raceway contact angles lead to variation in load distributions on the balls. At high speeds, it has been shown that the outer contact angles are less than the nominal contact angle and the inner raceway contact angle is greater than the nominal contact angle [20]. Therefore, it is necessary to determine the instantaneous contact angles at each ball position for given rotational speeds. Once these angles are known the traditional force-deflection equations for pointcontact deformation can be used to determine the total deflection deformation of the ball-spindle interface and the forces evolved. The following equations, developed from Fig. 2, are used in determining the instantaneous contact angles and are based on the works in [19,20].
W z + V2 - ((FO (AF -

i
,
<

AF
- -

I t
Ball ~ n t ~ r ,

a.0.oo.
I

.!/,

-~q

Final ~ i f i o ~ a n e r r eway gr oov~ curvature eenladr

. . . . . . . .

Outerraeewa

r ~e

Fig. 3. Positions of ball centre and raceway groove curvatures at angular position ~, with and without applied load.
?to M V / D (FO +

Ko~ao .s W 0.5) D + 8o
(FI -

KiSl s ( A F - W ) - hi M ( R F - V ) / D

0.5) D + 8i

= 0

(3)

ko M V / D + KoS~os V ( F O - 0.5) D + 8o
Ki~I 5 ( R F - V ) + (FI -

0.5) D + ~o)2 = 0 V) 2 ( ( F I - 0.5) D +

(1) 802 = 0 (2)

W) 2 +

(RF -

k~M ( A F 0.5) D + 81

W)/D

CF=O

(4) (5)

cos(13o) =

V/((FO -

0.5) D + 8i)

Spindle Deflections in High-speed Machine Tools

235

cos(13i) = R F - W ( ( F I - 0.5) D + ~i) sin(13o) = W/((FO - 0.5) D + ~i) sin(13i) = A F - W / ( ( F I - 0.5) D + ~i)
CF = 2.095 (D/2) 3 p PW2262/1000 M = J(W1/o0 (W2Ao) 2 o~ z sin(F)

(6) (7) (8) (9) (10)

elastic modulus, the length, and the moment of inertia of the structure and the rotational speed. The following is a representation of the transfer matrix for deflections in the ydirections as used in this study:

The subscripts i and o refer to the outer and inner raceways. The first two equations are Pythagorean relationships between the instantaneous bal~ centre position and deflections of the inner and outer raceways. The third and fourth equations represent the sum of forces in two perpendicular directions (x and y). The solutions of the first four equations are used to provide the inner and outer contact angles found using equations (5)-(8). It is to be noted that these equations are nonlinear and the values of gyroscopic M, and the centrifugal force CF depend on the particular value of the variables. Also, A F and R F depend on axial misalignment of the spindle (if any), and the relative axial and radial displacement of the two bearing races set by bearing preload and assembly. These three parameters, along with initial guesses for the primary variables w, v, ~i, ~o are used as inputs for solving the system. This system of equations is to be solved for each of the ball positions to determine the deflection at the contact points and deflection stresses. Once this is done, force and moment equations shown below are applied, in conjunction with the contact deflections, to the entire spindle for determining more accurately the axial and radial displacements (initially input as guesses) for the applied preloads:
Fa
-

My

Vy

llii '
1 0

l/E! 1

F/2EI 1

o~2m loj2m FoJ2m/2EI (13o~3m/6EI) + 1

MY i-1

v,
A similar matrix may be constructed for the deflections in the z-direction. Thus, by knowing conditions at the previous (i - 1) segment of the structure, the conditions at the next segment may be determined. It is assumed here that the spindle may deflect (bend) in the x, y and y, z planes leading to deflections along the y and z directions. The z-direction is vertical while the y-direction is along the axis of the spindle. The x-direction is perpendicular to the y and z directions. In summary, the inner and outer contact angles and the resulting deflections determined at each of the ball positions for the given material and operational conditions form input boundary conditions to the TMM for determining the total deflection at the spindle end. The next section presents initial results of the simulation.

X~ v (Qq sinaij - h 0 - ~ cosaj) = 0

Mq

F~ - "Z~ (Qq cosetij + hq

t~j) = 0

4.

Results

M - X u ((Qq sinaq - xij-~-) cosaq)Ri + hqMj)cost~j = 0

Mj

Here, the first two equations sum the forces for all the ball positions in the vertical and horizontal directions for the entire spindle whereas the third equation is the sum of the moments. The total deflection at the bearing is then given by: 10-5 {5 __FR = 2.53 D-TCOS[3

\ Z cosl3 + P

where P = K~ = contact deformation force and FR is the bearing preload and the deformation values are obtained by the solution of the above equations. These deflections form the input boundary conditions for the TMM method.

3.2 Analysis of the Deflection at the Spindle End

The TMM technique was used to determine the total deflection at the spindle end [21]. The TMM technique divides a structure into separate segments and from boundary conditions of the forces and deflections, the net deflection and forces at the free end are determined. The TMM technique requires as an input a matrix of coefficients whose elements are the

A computer program, based on the Hooke-Jeeves Search Technique (HJST) [22] was used to determine the optimum inner and outer contact angle values for given input conditions and axial deflections at the bearing end. The input conditions were the initial guesses for the contact deformations at the inner and outer races, the final ball centre position, the relative axial and radial deflection of the bearing, and the bearing misalignment (referred to in Fig. 3). The HJST searches the function space for the minimum function value based on initial guesses for the decision variables. In this case, the function value was the squared sum of equations (1)-(4) and was minimised to zero to yield corresponding values of inner and outer race deformations and ball centre migrations. These values were then used to determine the inner and outer race contact angles and the total tool end deflections for a given spindle system. The bearing and spindle system specifications used for the simulations are given below. Number of balls in bearing = 16 Nominal contact angle = 40 degrees Ball diameter = 22.23 mm Inner raceway diameter = 102.79 mm Outer raceway diameter = 147.73 mm Inner groove radius = 11.63 mm

236
60|

S. Chatterjee
~ ; ~ ~ .+___..4___+

+ - ,~,

CO 6050. ~ _ _ _ ~ ~ ~ ~ ~ ,-~ . . . . . ~

LLI

514o1--.. ~ ~ ~ ~ ~ ~ }~ ~. ~.-t~.. ~ ~::-~" ~ ~ : ~ ....

..,

45J 40 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

~ i .......................................................................................................................................................................................................................... b 25" --~-'-~--'~---~-.~-:~-"-FE--~'-_E---~F-~"--~Z-- ~--~Z--~y"-'.~-'-~F 2 0 ...........................................................................................................................................................................................................................

2 0 ......................................................................................................................................................

1 5 - ---~,,,

~4

:.4

:<

~.

~4

;'. - - - : ~ . -

8
22.5 67.5 45 0TA=0UTER R~,CANGLE INA=Ii'~ERRACE ANGLE P~LOA~ IN KN I~ OKN, OTA
]

O 112.5. 157.5 202.5 247.5 292.5 337.5 90 185 180 225 270 315 860 BALL LOCATION ANGLE, DEGREES
+ OKN, INA ~ 10KN, O T A |

~o~z~,

'.....e75'
45

'~k5'~5_
90 135

_ ~ , 5 ' 2 0 ~_ . 5 ' ~ 4 k_5_' ~ 0 ~ - ~ ' ~ '_ ,~ _'

180 225 270 315 360

OTA = OUTER CONTAGT ANGLE

BALL ~,-',o,~,r.,,,.~r_~o,.,~, DEG REEB

IN,k=iNNER GONTA~T AN~LE PRELOADINK-,N ~ -~20KN, O T A

1 0 K N , INA ~

L [] IOKN, INA

20KN, OTA - ~ - 20KN, INA

~" 20KN, tNA

5KN, OTA

+
~

10KN, OTA
5 K N , INA

Fig. 4. Contact angle variation at 5000 r.p.m. Pitch diameter of ball circle = 125 mm Bearing stiffness = 375350 N / m m ~~ Mass moment of ball = 2.21163 kg/mm e Spindle length = 304.8 mm Elasticity modulus = 200000 N / m m 2 Moment of inertia of shaft = 39612.55 m m 4 Spindle diameter = 29.27 m m Assumed shaft assembly misalignment = 0.001 radian The results of this simulation study are now presented. It is seen from Figs 4 - 7 that the ball contact angle at a circumferential position varies with rotational speeds. The inner and outer contact angles vary, from the mean contact angle of 40 degrees, the inner race contact angle being higher and the outer race contact angle being lower than the mean contact angle. The patterns of ball excursions are very similar at different preloads although at lower preloads the contact angle variations at the races are larger than at higher preloads. With increasing speeds the contact angles tend to exhibit higher excursions from the nominal. This can be seen by comparing Figs 4 and 7 for spindle speeds of 5000 and 30000 r.p.m.
O3 60rr 50 4540' <
2 5 ...........................................................................................................................................................................................................................

Fig. 6. Contact angle variation at 15000 r.p.m.

CO 60" [3_ ~--.E~--.-.E~,~I:3 - u LJ ~ ~ "-'~' [::3' ~ CZ'--'--~ ~ 55 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5 o , L ~ - - . ~ - . . . . ~ - - - - -~ILl c-', 45 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

40 ...........................................................................................................
z ~ < F-35, ~ 0 .............................................................................................................................

o<

2 5 ~ - " --- - - - - '

-"

---''

-'

-'

-'- - - - - '

'

- +'

-'

'

-J~-------'--'~-

20 ...................................................................................................................................................

o 152~.5 . 67. . 5 . . 112.5 . 15~'.5' 20~.5' 24~.5' 292.5' 8a7.5~45 90 OTA= OUTER CONTACT ANGLE INA=INNER CONTACT ANGLE PRELOAD IN KN mE 0 K N , O T A 0KN,INA 135 180 225 270 B A L L POSITION, D E G R E E S 315 360

~" 1 0 K N , O T A + IOKN,INA .A

20KN,OTA 20KN,INA

Fig. 7. Contact angle variation at 30000 r.p.m.

18o-r z

-0.5

160 .{i-4~ ] ~. ~ ; _;~ ,.................. _+___:. ...~.__ '-:---+ , ...,l~=__ ,, ~L._.~_.......~ ; .t,. 4-- ~ t,, ~,--->~.---~ t,, i ~ 0.4 140 ......................................................................................................................................................................................... '-

12o I ..................................................................................................................................................................... = LL 1004.................................................................................................................................................................................... 03 z [ .I

8 0 .................................................................................................................................................................................. z

LL
I

50~ .................................................................................................................................................................................... 0.2 "'~O 40 ............................................................................................................................................................................................... 20 ............................................................................................................................................................................................. 0.1

~ ~ - '

[]

~L....~.......~o..

0 -20 _ , , -, , , . . . . . . . . . -0 22.5 67.5 112.5 157,5 202.5 247.5 292.5 337.5 M=MOMENT 45 90 185_ 180 225 270 315 860 CF=CENTRIFUGALFORCE BALL LOCATION, DEGREES PRELOADIN KN [ ~ _
/

~-~*-~:
~4

fL~-~-`L~`.--~ L - - L - ~ . ` L - ~
)< ..'." .~I x ."<~

`---~e--~.L--~.~-~~. x ;-I" x

~ ~-~),(-.-~

OF, 0KN

CF, 1oKN- - - ~ - CF, 20KN 1 M, 20KN

20.
15. ~.~-~`~:~.~-~-~-~'~-~-~-~'~--~.~:: ...............................

--E~-- M, 0KN

--x-- M, 10KN ~

O 22.5 45 67.5 90 157.5 202.5 247.5 292.5 337.5 135 180 225 270 815 360 B A L L LOCATION ANGLE, DEGREES 112.5

Fig. 8. Centrifugal force and gyroscopic moment variations at 5000 r.p.m. The variations of centrifugal forces and moments are shown in Figs 8 and 9. Both the parameters vary with ball location and gyroscopic moments are noted to be similar at 5000 and 30000 r.p.m. The deflections at the bearings are shown in Figs 10-12. It is seen that the deformation is maximum at certain preloads.

OTA==3UT~RRACE ~tNGLE (NA=INNER RACE ANGLE FRELOAO IN KN -.m-- 0 K N , O T A

'~"

OKN,INA

:~

10KN,OTA

10KN,INA

- - x - - 2 0 K N , O T A --~1~-- 2 0 K N , I N A

Fig. 5. Contact angle variation at 10000 r.p.m.

Spindle Deflections in High-speed Machine Tools

0,11L6 0 5000
0 4000 i

237

~o
......................................................................................................................................

,$

0.1.
0,~ I

--0.2 LO.4 5" L-o.6 Lu

.........................................................................................................................

0,08 0.07'

2000 E:: ,~ 1000 Z LU 0 "" = 0

-1000-

...............................................................................................................................

-,.0.8 0 .....
67,5
. . . . . T , i i 7 ~ ' ~

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

0.06

~r~,-.-1
'-1.2

0.05 0.04

22.5

45 PRELOADIN KN CF=CENTRIFUGALFORCE M=MOMENT [--~

90

112.5 157.5 202,5 247.5 292.5 ,337.5 135 180 225 270 815 360 BALL LOCATION, DEGREES CF, 10KN --~4-- CF, 20KN

lb
PRELOAD, KN

2b

CF, OKN +

z~DELFECTION -'+-- Y-DEFLECTION

--~"- M, 0KN

--x-- M, 10KN * - i - M, 20KN

Fig. 12. Deflection at bearing end at 15000 r.p.m.

Fig. 9. Centrifugal force and gyroscopic moment variations at 30000 r.p.m.

0,06[
}

0.45
0.4
0135

,51 .......................................................................... 00,1 ............................................................. 0.o`31 ............................................. 0021 ........................... '1I ..................

..................................................

0,3

.................................. ...................... ...................... ..................................

(3 0.25 0 uJ LU a 0.2

o.15
0.1

0,05

5 PRELOAD, KN

10

20

PRELOAD, KN

[
1.3-

Z-DEFLECTION - - + - Y-DEFLECTION

I~

Z DEFLECTION - + - - Y DEFLECTION ]

Fig. 13. Deflection at spindle end at 5000 r.p.m.

Fig. 10. D e f l e c t i o n at b e a r i n g end at 5000 r.p.m.
&Off 0.070,06i 0.05" O 0.040.080.02. 0,010 5000 10000 PRELOAD, KN 20000 1.21,1" 1-

O 0.9FO 0,8uJ
uJ 3 0.6-

0,7-

0.5. 0.4

ib
PRELOAD, KN

2'0

-m--Z

D E F L E C T I O N ' - + - Y DEFLECTION

]
5.

l+

Z-DEFLECTION ~

Y-DEFLECTION 1

Fig. 14. Deflection at spindle end at 15 000 r,p.m,

Fig. 11. Deflection at bearing end at 10000 r.p.m. The deformation is for two directions perpendicular to the shaft centre-line. The reasons for the peaking of the deflection at particular preloads cannot yet be determined. The total deflections at the spindle end, obtained by the TMM, are shown in Figs 13 and 14 and are slightly higher than those at the bearing end. The deflections increase with rotational speeds for the simulated conditions. In this study the bearing stiffness was assumed to be constant. This, however, may not be the case as observed by others [2]; changes in bearing stiffness can be taken into consideration in the process.

Discussion

This simulation has shown variations in the contact angles owing to high rotational speeds resulting in subsequent deflections at the bearing and the spindle ends. A t low speeds the contact angles are close to the nominal; however the variations increase with rotational speeds. A n increase in the preload seems to limit the angular excursions somewhat, but variations are still significant. Deflections at the spindle end are more than those at the bearing end for the spindle system simulated. It is therefore necessary to simulate different

238

S. Chatterjee
2. Y. C. Shin, K. W. Wang and C. H. Chen, "Dynamic analysis of a high speed spindle system", Transactions of the NAMRI/SME, pp. 298-304, 1990. 3. Y. C. Shin, H. Chin and M. J. Brink, "Characterization of CNC machining centers", Journal of Manufacturing Systems, 10(5), pp. 407-421, 1991. 4. R. J. Hocken, Technology of Machine Tools, vol. 5, Lawrence Livermore Laboratory CA, UCRL-52960-5, 1980. 5. E. Rivin, "Trends in tooling for CNC machine tools: machine system stiffness", Manufacturing Review, 4(4), pp. 257-263, 1991. 6. E. Rivin, "Trends in tooling for CNC machine tools: tool-spindle interfaces", Manufacturing Review, 4(4), pp. 264-274, 1991. 7. G. Chryssolouris, "Effects of machine-tool-workpiece stiffness on the wear behavior of superhard cutting materials", Annals of the CIRP, 31(1), pp. 65-69, 1982. 8. R. R. Donaldson, "Error budgets", Technology of Machine Tools, vol. 5, Lawrence Livermore Laboratory CA, UCRL52960-5, pp. 9.14-1-9.14-14, 1980. 9. G. Zhang, R. Ouyang, B. Lu, R. Hocken, R. Veale and A. Donmez, "A displacement method for machine geometry calibration", Annals of the C1RP, 37(1), pp. 515-518, 1988. 10. R. Hocken, J. A. Simpson, B. Borchardt, L. Lazar, C. Reeve and P. Stein, "Three dimensional metrology", Annals of the CIRF, 26(2), pp. 403-408, 1977. 11. P. M. Ferreira and C. R. Liu, "A method for estimating and compensating quasistatic errors of machine tools", Journal of Engineering for .Industry, 115, pp. 149-159, 1993. 12. R. Schultschik, "The components of the volumetric accuracy", Annals of the CIRP, 25(1), pp. 223-228, 1977. 13. J. Mou and C. R. Liu, "A method for enhancing the accuracy of CNC machine tools for on-line inspection", Journal of Manufacturing Systems, 11(4), pp. 229-237, i992. 14. J. Bryan, "International status of thermal research", Annals of the CIRP, 39(2), pp. 645-656, 1990. 15. J. Jedrzejewski, J. Kaczmarek and Z. Kowal, "Numerical optimization of thermal behavior of machine tools", Annals of the CIRP, 39(1), pp. 379-382, 1990. 16. R. Venugopal and M. Barash, "Thermal effects on the accuracy of numerically controlled machine tools", Annals of the CIRP, 35(1), pp. 255-258, 1986. 17. M. Rahman, M. A. Mansur and K. H. Chua, "Evaluation of advanced cementitious composities for machine-tool structures", Annals of the CIRP, 37(1), pp. 373-376, 1988. 18. D. G. Lee, H.-C. Sin and N. P. Suh, "Manufacturing of a graphite epoxy composite spindle for a machine tool", Annals of the CIRP, 34(1), pp. 365-369, 1985. 19. A. B. Jones, "A general theory for elastically constrained ball and radial roller bearings under arbitrary load and speed conditions", Journal of Basic Engineering, pp. 309-320, June 1960. 20. T. A. Harris, Rolling Bearing Analysis, 3rd edn, John Wiley and Sons, 1991. 21. A. Bhattacharya, Principles of Machine Tools, 2rid edn, New Central Book Agency, Calcutta, India, pp. 605-609, 1975. 22. B. S. Gottfried and J. Weismann, Introduction to Optimization Theory, Prentice-Hall, Englewoods Cliffs, NJ, pp. 113-130, 1973.

spindle and bearing systems that impart more rigidity to the system and examine the robustness of the simulator. A t this stage one of the primary concerns is to establish the validity of this modelling process and the simulator with experimental data. Efforts are presently underway for obtaining industrial data and/or designing the experimental set-up. Specially built motor-integrated spindles are necessary at speeds of 10000r.p.m. and above. Additionally, the sensitivity of the model output with respect to the input conditions is also currently being examined. The variation of the contact angles and the subsequent contact deformations and deflections are important for highspeed machining. Specifically, the spindle deflection errors can be compared with other error sources (servo errors, thermal errors, etc.) and used in the development of machine tool error budgets. Another important factor is the effect of the variability of the dimension of each ball on the overall spindle deflection at high speeds. This may have an effect on the overall process capability. The effect on total deflection, through simulation, of ball material differences on contact deformation and alternative materials, such as composites, by incorporating the material properties in the simulation needs to be examined. The present model does not incorporate bearing heat effects. It is intended in the future to include heat effects in the model.

6.

Conclusions

This research has attempted to model spindle error profiles for high rotational speeds. Bearing contact angle variations, variations in centrifugal forces and moments, total tool end deflections have been modelled. Based on the research some important further research areas in high-speed machining are: 1. Consideration of alternative materials in spindle structures. 2. Examination of the effect of variations in ball dimensions on the deflections. 3. Development of error budget processes in machine tools. 4. Examination of the effects of various spindle tapers on deflection at high speeds. 5. Examination of effects of stiffer bearing materials on deflections. It is hoped that such investigations will lead to a better understanding of the high-speed machining process and better design of machine tools.

Nomenclature

Acknowledgement
The financial support provided by the Office of Research Administration, The University of Tennessee, Knoxville, TN, is gratefully acknowledged.

AF RF
W V D ro r~ M

References
1. R. I. King (ed.), Handbook of High-Speed Machining Technology, Chapman and Hall, New York, pp. 1-26, 1985.

final position, inner raceway groove centre initial position, inner raceway groove centre final position of ball centre initial position of ball centre ball diameter, mm inner raceway groove radius, mm inner raceway groove radius, mm gyroscopic moment, N-mm

FO t71
P Ko

ro/D ri/o
bearing pitch diameter, mm outer race load-deflection constant, N/ram 1.5

Spindle Deflections in High-speed Machine Tools

239

K~
CF J l E I Y
Z

M V m P

8o

inner race load-deflection constant, N/mm 1'5 centrifugal force, N mass moment of inertia, N.mm 2 length of spindle, mm modulus of elasticity, N/mm 2 moment of inertia of spindle, mm 4 deflection of spindle along y-direction, mm deflection of spindle along z-direction, mm moment at spindle end, N.mm shear force at spindle end, N spindle mass, kg material density outer race contact angle

~o
F W1 W2

inner race contact angle nominal contact angle inner race deformation outer race deformation angle between ball centre of rotation and the horizontal mis-alignment (in degrees) of shaft assembly measured in a plane perpendicular to shaft axis (x-direction) ball and raceway angular raceway velocity ratio for outer raceway control ball orbital and angular raceway velocity ratio for rotating inner raceway and outer raceway control circumferential ball position raceway control parameter