You are on page 1of 10

H.

McCALLION The Analysis of Thermal Effects in a


F. YOUSIF
T. LLOYD
Mechanical Engineering Department,
Full Journal Bearing
The University of Nottingham,
Nottingham, England
By means of numerical techniques, the behavior of a journal bearing of finite length
running under steady load, including thermal effects, is investigated with reference to the
two simpler models: the "isothermal" and the "adiabatic" cases.
Different bearing operating parameters are obtained, which in the latter case are
evaluated from the pressure and temperature distributions obtained from the simul-
taneous solution of the Reynolds and Energy equations. A simplified method, which
uncouples the Reynolds and the Energy equations in the oil by neglecting all the pressure
terms in the energy equation, is discussed and compared with the more exact numerical
method of solving the two equations simultaneously.
The case of heat being conducted through metal surfaces is described, with the assump-
tion of zero net heat flow to the journal. In order to shorten the calculations, the uncou-
pling technique investigated in the adiabatic case is again used and the results obtained
are compared with experimental findings.

Introduction the oil, the journal and bush being perfect insulators, and this is
referred to as the adiabatic case. The latter case involves the
A,
ITLMOST all previous theoretical solutions for the simultaneous solution of the Reynolds equation and an energy
operating parameters of finite-length steadily-loaded journal bear- balance equation and, as this involves a considerable amount of
ings, have been concerned with two ideal thermal situations. In computing time, the authors have investigated the accuracy of
o. ie the oil temperature has been assumed constant, the isothermal the magnitudes of the operating parameters evaluated on the
or isoviscous case. In the second it has been assumed that all basis of an approximation which decouples these two equations.
heat generated by the oil film friction has been carried away in Results based upon each of the assumptions are compared.
The interest of the authors in wishing to decouple the Reynolds
and energy equations was connected with the desire to solve for
the thermohydrodynamic case in which heat could be conducted
to the shaft and housing and could be lost by convection from the
Contributedjby the Lubrication Division and presented at the outer surface of the housing.
Fluids Engineering, Heat Transfer, and Lubrication Conference, De-
troit, Mich., May 24-27, 1970, of T H E AMERICAN SOCIETY OP M E - Other workers [1, 2 ] 1 have investigated this case theoretically
CHANICAL ENGINEERS. Manuscript received at A8ME Headquarters,
1
March 10, 1970. Paper No. 70-LubS-23. Numbers in brackets designate References at end of paper.

•Nomenclature-
c = radial clearance directions, respectively M,(c/R)/niUR
cc = convergence factor Km = thermal conductivity of lubri- n, nm = number of mesh points across
C — specific heat of lubricant cant oil film, and bush wall
D = dia of journal thickness
Kml = thermal conductivity of bush
e = eccentricity material N = number of mesh points in the
f(R/c) = friction parameter I number of iteration axial direction
h — film thickness P — pressure
L bearing length
P , = surrounding pressure
H = nondimensional film thickness number of mesh points around P = nondimensional pressure
(h/c) the bearing (P - P )
H* = nondimensional film thickness M, = Motional torque on journal — — (c/RY
at breakdown
per unit width P* = reduced pressure function
i, it h — mesh counts in the circum- M/ =
nondimensional friction
ferential, radial, and axial (.Continued on next page)
torque per unit width

578 / OCTOBER 197 0 Copyright © 1970 by ASME Transactions of the AS ME

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


for an infinitely long bearing, one assumed the shape of the tem-
perature distribution in the housing and the other assumed a two
lobed bearing under no load.
In this paper, results obtained on the basis of the approxima-
tion which uncouples the Reynolds and energy equations are
compared with the experimental findings of Dowson, et al. [3].
1 The Isothermal Cose. In this case it was assumed that the tem-
perature throughout the oil film, and hence the viscosity of the oil,
remained constant. The performance parameters were evaluated
from the pressure distribution which resulted from iterative
numerical, solution of the Reynolds equation

/ dP\ d f dP\
6E//i (1)
dx V i>x J dz\ &z /
where the various parameters in this equation are defined in the
Nomenclature and in Fig. 1.
In terms of the following nondimensional groups

X z
Z =
R: fl

u e =—
e
R' c

(P - P.)
H = 1 + e-cos x; P
/UCO (-;)•
Fig. 1 Bearing geometry
and the substitution due to Vogelpohl [4]

P* = P-H'^
dP*
at the end of the pressure zone (X = 2*)
equation (1) becomes:

WP* &P* 3 P* ["/^ P* = 0 along the axial line representing an oil


+2H supply at x = 0
) -^\
Vs
P* = 0 throughout the cavitated region
= 6ff" (2)
dx For the solution of the Reynolds equation by numerical means
The boundary conditions used to solve this equation are: a rectangular mesh covering the developed surface of the bearing
was employed. The increments in the z-direction were constant
P* = 0 at the bearing sides and equal to Az. If the number of increments is N, then Az =

-Nomenclature-
Pe — nondimensional thermal con- u v a* = heat transfer coefficient
ductivity of the oil ponents; — and — (R/c), re- /3 = viscosity-temperature coef-
2 ficient
Cpuc spectively A = load on the journal per unit
R = i-adius of journal linear circumferential coordi- width
Rh = radius in the bush nate measured from maxi-
A ' = nondimensional load parame-
mum film thickness in the
Rbi, -ffito = the bush inner and outer radii ter A/M, V{R/CY
direction of rotation
Rb = nondimensional radius in the nondimensional circumferen- A / = radial and tangential compo-
(RJRH)
t, tb, ta = oil, bush, and ambient tem-
tial coordinate

angular coordinate
G) and nents of load parameters
e = eccentricity ratio
peratures respectively nondimensional circumferen- f = nondimensional viscosity tem-
(i = oil inlet temperature tial coordinate of oil film perature coefficient
breakdown
h h, h ~ nondimensional temperatures
V = radial coordinate through the Cp(c/Ry
AW film thickness measured
U = relative velocity between jour- from bush surface JJ, = lubricant viscosity
nal and bearing surface nondimensional radial coordi- y.i = lubricant viscosity at inlet
nate, (y/h)
u, v,w = velocity components in x-, y- jl = nondimensional viscosity ix/pi
axial coordinate measured
and z-directions, respec- from bearing centerline p = density of lubricant
tively nondimensional axial coordi- \p = load angle
u, v = nondimensional velocity com- nate (z/R) a) = angular velocity

Journal of Lubrication Technology OCTOBER 1 9 7 0 / 579

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


2(L/D) By a further substitution e f ' = 1 + f 6 analogous to that used
:—. The increments in the x-direction were not constant. by Zienkiewicz [8], equation (6) becomes
N
They were defined by the following relationship described by H.
McCallion, et al. [5]. be
s (1 +
|_2 12 ' bx _J dx Ll2
12 ^e) ^ bz
7r(i - 1) s 12 bz
Axi = 2-7T 11 + c'-e-cos (3) 1 H
2 X + (8)
- 1 ^ + ^ ( i + fe) l\bx) \bz)

In finite difference form, employing a similar mesh configuration


This concentrated a large number of nodes in the region of mini- to t h a t used for the isothermal case, equation (5) becomes:
mum film thickness.
In finite difference form equation (2) becomes: P*i,u = « i / P V u + 6,-,*'P*v_i,*
+ di,h'(P*i,t+i + P*i.k-i) + gilk' (9)
P\.k = asP*i+1,k + biP*i-uK + di(P*i,k+l + P*i,k^) + gt (4)
Thus for a particular viscosity distribution, evaluated from an
The system of equations typified by equation (4) was solved by
initially assumed or previously calculated temperature distribu-
the extrapolated Gauss-Seidel iteration procedure, vising an opti-
tion, equation (9) was solved for the boundary conditions outlined
mum over-relaxation factor for the particular mesh configuration
earlier to yield a corresponding pressure distribution.
as determined by H. McCallion and T. Lloyd [6].
The procedure for the solution of the energy equation differed
2 The Adiabatic Case. In this case the heat generated by the oil
somewhat from that of the Reynolds equation, in that the line
film friction was assumed to raise the temperature of the oil only,
x = 0 was the only position at which t was known initially, where
and that heat could only leave the bearing in the hot oil. The
actually I = 0, and thus 9 = 0.
temperature through the oil film was assumed to be constant.
The method of solution involved a step by step integration
The variation of the lubricant viscosity with temperature only
(marching) of equation (8). For this purpose, the method and
was taken into account, and its influence on the performance
the modified method of Euler were used. These are basically
parameters was estimated from a combined solution of the
first and second order methods, respectively, and equation (8)
Reynolds and energy balance equations. In terms of the follow-
was rewritten in a form suitable for such a method of solution, as
ing nondimensional groups
follows:
(P - P.)
be 1
AW bx wl - Af,u + re)]
p . f f ' A . M~ be
The Reynolds equation reduces to
M3(l re) — + M4(I + re)2 + M6 ao
bz J
d 2 P* d 2 P* bH For a particular pressure distribution, the procedure of solution
—— 2 -| h A - P * = 6-ff" V» p'/. (5) was to start with the bearing center line and march around each
bx bz2 dx
circumferential line in turn, thus evaluating a distribution in 9 ,
where by means of which the distribution in I and /Z were evaluated.
The method employed here for solving equations (8) and (10)
simultaneously, was to start with the bearing centerline and
evaluate the pressures and temperatures at consecutive mesh
points; and to repeat this for the adjacent circumferential line,
and so on. This procedure was continued a number of times
until a convergence limit of 10 ~3 was satisfied, between the pres-
sure distributions for two successive iterations. This method
proved to be more stable than the one which alternates between
solving separately the pressure and then the temperature dis-
Following the procedure used by Hakansson [7], the following tribution over the whole field.
nondimensional energy balance equation was obtained 3 The Simplified Method for the Adiabatic Case. Neglecting the
effects of pressure in the axial direction, equation (6) may be
/H H°_ bP\ c» H*_ /bP\ bl rewritten as:
\ 2 ~~ 12/t bx ) bx~ 12/Z \bz) bz ' H
bl 2/x 1 + 35 2 "
EL /bP\ (spy bx ~ ~H* ' . 1 - & .
(11)
(6)
+ 12/1 \bx) + \bz)
where
where 2
bP
8 ^
= Cp 6/ I bx
/tlCO
Provided that 5 <<C 1, the pressure terms may be omitted from
The viscosity variation with temperature was represented by equation (11), thus uncoupling the energy and the Reynolds
the following exponential relationship equations.
The term 8 is much smaller than unity, for bearings of small
length-to-diameter ratio, operating at low eccentricity ratios.
/!=£- = e -««-M (7) This may be demonstrated by considering the approximate iso-
Mi
thermal pressure distribution at the bearing centre line, appro-
where priate to these conditions

f Cp \R P' = 3(L/Dy- """ ~


(1 + e-cos x)3

580 / OCTOBER 1 970 Transactions of the AS ME

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


and and for the cavitated region
dP' It cos x + e2 cos2 x + 3e2 sin 2 x
r dx • dz
*" " 3(L/D)2
-( ^ (L V£)J 0 J,* &
giving The Friction Parameter
i? 2 d P ' (L/Z>)2 / £ cos x + e2 cos2 S + 3e2 sin 2 x> M,
8' = F(R/c) =
6 dx ff2
Table 1 gives some values of e', for a range of length-to-diameter Note in the foregoing t h a t a correction for the shear stress at the
and eccentricity ratios, at three circumferential locations around journal surface in the cavitated region was included.
the bearing. The various integrals in the above were evaluated numerically
by Simpson's rule.
Table 1 4 The Thermohydrodynamic Case. A rigorous analysis which
attempts to consider the effect of heat flow to the shaft and the
L/D e x = 0 * = T/2 X = 7T
housing for a finite length bearing must consider the simultaneous
0.50 0.01 0.02 -0.03 solution of the two dimensional Reynolds equation, three dimen-
0.250 0.80 0.01 0.06 -0.13
0.90 0.01 0.08 -0.28 sional energy equation in the oil film, and a three dimensional
0.50 0.04 0.09 -0.13 heat conduction equation in the solid bodies. This would be ex-
0.50 0.80 0.06 0.24 -0.50 pensive to compute, but the experimental observations of Dow-
0.90 0.06 0.30 <-0.5 son, et al. [3] suggest that the following simplifying assumptions
0.50 0.17 0.38 -0.50 may be made:
1.0 0.80 0.22 >0.5 <-0.5
0.90 0.24 >0.5 <-0.5 1 The bush heat flow in the axial direction is negligible; only
radial and circumferential heat flows need be considered.
This table confirms the validity of neglecting S in equation (11) 2 The cyclic variation of the journal surface temperature is
at the lower end of the range of eccentricit}? and length-to- small and it can be treated as an isothermal component whose
diameter ratios considered. mean surface temperature is close to the mean temperature of the
This approximation removed the necessity for solving the bush inner surface.
Reynolds and the energy equations simultaneously. The method In addition to the above experimentally based simplifications,
of solution was the following simplification was also made:
3 The small effect of the pressure on the temperature distribu-
1 To solve the energy equation to give the temperature dis- tion again makes possible the uncoupling of the energy and the
tribution and hence the corresponding viscosity distribution Reynolds equations.
2 To solve the Reynolds equation with this viscosity distribu- The temperature within the oil film which resulted from the
tion to give the pressure. When 8 is neglected, equation (11) viscous dissipation was defined by the following iiondimensional
can be solved analytically energy equation
1
, 2f e + cos x \ m /5M V
ln HH (13)
7 i + jmij* 1 + e-cos5 %) dx
e-sin x- \ / l — e where
(12)
(1 + e-cos x)
Pe =
Performance Parameters. Having established pressure, tempera- Cpcoc2
ture and viscosity distributions, the following performance pa- and
rameters in dimensionless form were evaluated. The Load
Parameter _ !Z- _ -it
Mi
A' = — ( - V = [A/ 2 + A('2]V*

where JOURNAL BUSH

A/ = p •sin x dx-dz
(Z7^)J0 J0
and
1 fL/D Clir
A/ = — — - P • cos x dx • dz

The load Angle

* = tan- (fj)
The friction Torque

M/ = JL
f
nJJR k (I) - "»'
where for the pressure zone
ie
die • dz
Fig. 2 Heat flow situation at oil bush interface
(L/D) J 0 J0 U 2 bxj
Journal of Lubrication Technology OCTOBER 1 9 7 0 / 581

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


JOURNAL (13) and (14) transform into the following finite difference equa-
tions
-AiJuj + Bi,Ah,i+l + k,j^] + Ei./k^j
+ FijU^.j + Gi.j = 0 (18)
and
(lb)i,j = aai.jQ^i+Lj + bbi,j(lb)i^,j + ddi,j(li)i,J+i
+ ggi,j(hhj-i (19)
BUSH respectively. The foregoing coefficients are also defined in the
Fig. 3 Mesh representation of transformed lubricating gap
Appendix.
An initial temperature distribution on the bearing surface was
assumed, and in order to solve the energy equation the tempera-
tures through the film thickness at x — 0 must be known, and a
parabolic distribution having a minimum corresponding to 1 = 0
was assumed. A marching procedure employed by Dowson and
Hudson [9] was used here, by means of which the temperature
distribution within the oil film was calculated, and the tempera-
ture gradients at the solid surfaces were found.
When solving the conduction equation in the bush, the asso-
ciated boundary conditions were expressed in difference form.
Referring to Fig. 4, equations (15) and (16) become

R\ fK„,
(?6),-._i = (kU + 2ARb 1
<V \Kml
1 (dl
X (20)

and
or
\h)i'iim+l — {t'bJi^nm-l — 2ZAxt(, •R[ 1 4-

X [(4)i,„„, - U (21)
respectively. A relaxation procedure was used to solve equation
(19) for the bush temperatures making use of the above two equa-
tions as boundary conditions.
The method used for solving the energy and heat conduction
equations simultaneously was as follows:
1 Assume an initial temperature distribution throughout the
Fig. 4 Mesh representation of the bush bush and on the shaft surface.
2 Solve the energy equation for the temperatures in the oil
The derivation of the foregoing equation is given in the Appendix. film; and thus evaluate the temperature gradients at the bearing
The temperatures in the bush are defined by the Laplace equa- surfaces, then correct these for the presence of air in the cavitated
tion; which in nondimensional polar coordinates is as follows: region.
3 Estimate the new bush temperatures from equation (19)
+ (14) by sweeping, once, all the mesh points.
Rb dRb Rb* dx2 ~ 4 Equate the shaft surface temperature to the mean tempera-
where ture of the bush inner surface.
5 Repeat steps (2)-(4), and leave the loop after every ten
Rh Rh/Rbi iterations or so to evaluate the oil film breakdown boundary b y
At the bush inner surface, the heat flow situation was defined solving the Reynolds equation.
with reference to the coordinates of Pig. 2 by the following: 6 Stop if the following convergence criterion is satisfied:

R
X 1 + X X
y^Rjsbi \K,„i Hj \dlj)y = 0 (22)
< cc
(15)
(-)'
and at the bush outer surface
\zyji
7 From the resultant temperature distribution within the oil
1+
ZRy )*-[(£)*-( i)_- (4o - l„) (16) film, evaluate the pressure distribution by solving the Reynolds
equation, and hence the performance parameters.
8 Evaluate the net heat flow to the shaft by integrating the
The form of the Reynolds equation employed was as equation (5)
product of the thermal conductivity of the oil and the tempera-
except for jX which in this case was assumed to correspond to t h a t
ture gradients in the oil film at the shaft surface around the bear-
at the mean cross film temperature lm where
ing, and if the net heat is not zero, then repeat the above proce-
i rh dure with a slightly modified shaft surface temperature from that
(17) of the mean inner bush temperature.
"Jo ^ Usually three estimates of the shaft surface temperature were
With reference to Figs. 3 and 4 for the mesh configuration in the found to be necessary before the net heat flow to the shaft was
transformed lubricating gap and the bush, respectively, equations sufficiently small.

582 / OCTOBER 1970 Transactions of the ASME

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


ISOTHERMAL ISOTHERMAL .
• UNC.ADIABAT1C UNC. ADIABATIC
• EXACT AD1ABATIC EXACT ADIA.

10
oK 10

LU
r-
a LU

<
<
a.
z
UJ 1.0 O
H 1.0
O
CC
u_

C1
0.2 0/ 0.6 0.8 1.0 0.1
ECCENTRICITY RATIO £ 0 0.2 0A 0.6 0.8 1.0
ECCENTRICITY RATIO s
Fig* 5 Comparison of adiabatic and uncoupled adiabalic (load pa-
rameter) Fig. 7 Comparison of adiabatic and uncoupled adiabatic (friction p a -
rameter)

. ISOTHERMAL
• . UNC. ADIABATIC
EXACT ADIABATIC EXPERIMENTAL
THEORETICAL Z - 0.5

10 60

< 55

<

<
O

INLET TEMPERATURE
01
0.2 0A 0.6 0.8 1.0
ECCENTRICITY RATIO £
0 1U2 TT 3TT/2 2 it
Fig. 6 Comparison of adiabatic and uncoupled adiabatic (load pa-
ANGULAR POSITION FROM INLET _ rad.
rameter)
Fig. 8 Comparison with experiment (inner bush surface temperature)

Discussion of Numerical Results


The Adiabatic Case. The adiabatic solutions presented graph- mated by testing with finer meshes, that the numerical errors in
ically give solutions up to an eccentricity ratio of 0.950. At the the load parameters evaluated at the highest eccentricity ratio
higher eccentricity ratios, some accuracy was lost because the were about 10 percent. It should be mentioned that when high
marching technique used in the calculation of the temperature eccentricity ratios were combined with a relatively high value of
was dictated by stability considerations. f, these solutions became unstable.
With 64 mesh points in the circumferential direction and with a The uncoupling technique which has been tested in this paper
first order method of marching out the temperatures, it was esti- over the practical range of parameters, provides an efficient and

Journal of Lubrication Technology OCTOBER 1 9 7 0 / 583

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


a reasonably accurate way of predicting the journal bearing per- perimental measurements and overestimate it in the high load
formance. The accuracy of this technique was checked by com- region. This may be explained as follows:
paring its performance parameters with those of the exact Two factors which have opposing effects are neglected in the
adiabatic solutions. Figs. 5 and 6 show the comparison for present analysis: these are, the mixing of the hot recirculating
length-to-diameter ratios of 0.250, 0.50, and 1.0 at values of f of oil with the cool inlet oil, and the cooling effect of the leakage oil.
0.05 and 0.10, indicating differences between the two methods for The net effect of these two factors may well be; to raise the tem-
the load parameter. These figures show t h a t the technique perature levels in the low load region, and to lower the tempera-
works for a rather greater range than might be expected from ture levels in the higher load region.
consideration of Table 1. For f = 0.05, the difference is less than A good correlation is also obtained with the same source of ex-
10 percent, provided that the load parameter A ' is kept below 2.5. perimental results for the bush temperature isotherms as shown
Also, for f = 0.10, the difference is less than 10 percent, pro- in Figs. 10 and 10(a).
vided that the load parameter A' is kept below 2.0. This allows
eccentricity ratios up to 0.6 for length-to-diameter ratio of 1.0, up The dimensionless variables introduced in the thermohydrody-
to 0.8 for length-to-diameter ratio of 0.50, and up to 0.90 for namic case, not applicable in the adiabatic case are, Pe, I J
length-to-diameter ratio of 0.250.
la, and those related to the bush geometry.
The agreement between the two methods when considering the The importance of the foregoing parameters was investigated
friction parameter is very good as shown in Fig. 7, since this is for a bearing of similar dimensions in the radial direction as that
the ratio of the journal friction torque to the load parameter, both of Dowson, et al. [3]. The shaft and bush dia were 4 and 9 in,,
of which increase in the uncoupled case. respectively, the clearance ratio was 0.00125, and the bush
The saving in computing time obtained by employing the un- material was phosphor-bronze.
coupling technique is considerable; the more exact solution re- Results were obtained for which ta was based on inlet and am-
quires between 4 and 20 times that for the uncoupled case, de- bient temperatures of 40 and 20 deg C, respectively. These were
pending upon the eccentricity ratio. compared with results for la = 0 and showed differences not ex-
The Thermohydrodynamie Case. The thermohydrodynamic ceeding 3 percent for the load parameter, shaft temperature, and
analysis discussed here attempts to extend the normal thermal maximum bush temperatures. All the results presented (Figs. 11
analysis of journal bearings, by taking into consideration the in- and 12) are based on the assumption that la = 0.
fluence of heat conduction to the solids. A comparison of bush The influence of the dimensionless variable "Pe" was found to
temperatures was made between the present theoretical analysis be relatively small; for example, a change of a factor of 7 produced
for the case where the oil inlet was situated on the load line, and a change of only 20 percent in the load parameter at an eccentricity
the measurements reported in [3] for a similarly situated oil inlet. ratio of 0.80, and length to dia ratio of 0.50.
Fig. 8 shows a comparison for the bush inner surface tempera-
tures, it also shows that a good quantitative agreement was ob- A change of ( —— 1 by a factor of 4, equivalent to increasing the
tained. Fig. 9 shows the comparison for the maximum and \-K-ml/
minimum bush temperature over a range of load, where in the thermal conductivity of bush material from t h a t of phosphor-
low load region, the theoretical results underestimate the ex- bronze to aluminum produced a change in the load parameter of
about 3 percent at 0.8 eccentricity ratio and 0.5 length-to-dia
ratio. The reason for the insensitivity of the results to the value

70 1 1 1 1 1 1 1 i i i—"T 1 i i 1 1 1 r of I -—- I is that the amount of heat conducted to the bush and
\Kml/
EXPERIMENTAL that returned to the oil by the bush were very much smaller than
THEORETICAL- the heat circulating through the shaft. The assumption of an in-
sulating bush was made on this basis, with zero temperature
gradients within the oil film at bush surface. This resulted in a
60 - reduction in the load parameter of about 17 percent, 0.5 deg dif-
ference in the load angle, and less than 1 percent change in the
friction parameter, from those for full solution of the aluminum
MAXIMUM BUSH TEMPERATURE bush, at an eccentricity ratio of 0.80 and length-to-dia ratio of
0.50. The important conclusion to be drawn from this is that the
bearing geometry and material properties affect only slightly the
50 - bearing performance parameters. This makes it possible to pro-
duce general design curves for the full journal bearing.
The values of load parameter for the thermohydrodynamic
case, as shown in Fig. 11 do not lie between those from the iso-
thermal and coupled adiabatic solutions, as expected; this may
be explained as follows:
Qc 1,0 -
LLl Some of the heat conducted to the bush, and all that gained by
^ - Inlet Temperature the shaft in the hotter regions of the system, is returned to the oil
uj in the cooler region after the oil inlet position. This raises the oil
temperatures sharply in that region, thus reducing the oil vis-
cosity and hence the load parameter. Fig. 12 shows that the in-
fluence of f is a most important one, giving a change of about 33
30 -
percent on load parameter for a change of f of a factor of 2 at
eccentricity ratio of 0.80, and length-to-diameter ratio of 0.50.

Conclusions
20 I ' ' i—J I i 1 i 1 i 1 1 1 1 1 1 1 i— 1 The uncoupling technique provides an efficient and accurate
0 BOO 2030 3000 way of predicting the journal bearing performance up to an ec-
LOAD _ ibf. centricity ratio of 0.80 and a length-to-dia of 0.50.
Fig. 9 Comparison with experiment (maximum and minimum inner bush 2 The thermohydrodynamic model considered here agrees
lemperafure) well with the experimental observations, and the bearing

584 / O C T O B E R 1970 Transactions of the ASME

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


o

II

u
SI
a

01

Journal of Lubrication Technology OCTOBER 1970/ 585

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


1 • i
ISOTHERMAL Where the dissipation function <j> is defined as follows:
UNCOUaED ADIABATIC
2 /bvV /,bw\2 bu bvy
- ~ THERMOHYDRODYNAMIC Pe=0.15
(9' by bx)
'bv dwy
\bz + £)'
- When compared to the velocity gradients —, and — ; all the
by by
/ > other velocity gradients in the foregoing are relatively small, and
UJ
can be neglected. Similarly the conduction terms in the x and z-
UJ L/D=0.5 / / /
directions are relatively small when compared to that in the y
2 / / / direction and may also be neglected. In addition, if all the pres-
1-0 -
sure terms are omitted, equation (23) reduces to
2 : / / / ^ '-
bt bt bU buY
// / :
Q
<
o
/
Cp u— + v
bx t]
by
Ku
by2
(24)

Where

1
-

1
u = U
01 -
/

III
By the use of the following dimensionless expressions
- 1 i

0 02 0.4 &6 0-8 LO u


ECCENTRICITY RATIO e u =
U' u
Fig. 11 Comparison of isothermal, uncoupled adiabalic, and thermo-
hydrodynamic (load parameter) X )X

R Mi

Pe =0.15 (t - U
Cp L
-
'(i)'
10 K6,
1 1 1 1 1 1 1 1

Pe =
Cpwc*

and the substitution


: L
1/ y
'o= a5 fcoos//
i

(Id

1.0 - which transforms the variable height lubricating gap to a rectan-


< gular one; equation (24) transforms to
-
Q
< / H1
bt
u — + utj
e-sm x bt
1
v bt
- = Pe
O bx H by H by _ bhj*

+P (25)

0.1 : / The velocity component in the y direction is usually evaluated
by integrating the continuity equation, but as continuity is
i i i i t -
violated when all the pressure terms are omitted, an estimation of
0.?. 0.4 0.6 0.8 1.0
ECCENTRICITY RATIO e v is made with the aid of Fig. 13 where it is assumed that the in-
clination to the horizontal of u varies linearly with y,
Fig. 12 Effect of f on load parameter
y bh
housing geometry and material properties affect only slightly the 7i>
h bx
bearing performance parameters. This makes it possible to
produce design curves for the full journal bearing. and

v = u- sin 7 = u-y
APPENDIX
V bh
Derivation of the Thermohydrodynamic Energy Equation u- h bx
The temperature of an elemental volume within the oil film is
defined by the following energy balance equation
U \cj bx
Cp
bt bt
u — + v —
bx by 1 VbH b'H bH~\
= — it-Y-e-sinx

H^+^+c?J + ^ (23)
Tjr This implies that the second and third terms of the bracketed
r.h.s. of equation (25) are equal in magnitude and of opposite

586 / OCTOBER 1 970 Transactions of the AS ME

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


where
aau = 2lli,1-/{(Rb)i*Axu.1(Axi + Axi+1)}
bbs.j = 2lli,j/[(Rb)SAxi{Axi + Axi+1)}
1 0.50 "I
ddi.j = lh,j +
ARb* (B,i)yAJfJ

1 0.50 "I
ggi.i = Ui.j _ARh* ~ (Rb),ARbj

and

0.50-AitS
lh,i = I + ARS/KRJSAxiAxi-J

where in the above


BUSH
(RM - 1)
Fig. 13 Estimation of velocity component v ARh

and
signs; and cancel each other resulting in the following equation
(Rb)j = 1 + j-ARb 0 <j < nm
Hhi — = Pe + (26)
dx by2 '&)"
Acknowledgments
Finite Difference Representation The authors wish to thank Professor A. G. Smith for placing
By approximating the temperature derivatives to their finite the facilities of the Mechanical Engineering Department, Uni-
difference equivalents, the following finite difference equations versity of Nottingham, at their disposal and the Science Research
were obtained. Council for the use of Chilton, Atlas Computer.
The Energy Equation. In finite difference form, equation (25) be-
comes References
— Ai,jh,j + Bi,j[li,j+i + li,j-i] 1 Dowson, D., and March, C. N., "A Thermohydrodynamic
Analysis of Journal Bearings," Proceedings, Institution of Mechanical
+ Ei,jU-l,j + Fi,j~k-Z,j + Oi.j = 0 Engineers, Vol. 181, Part 30, 1966-67, p. 117.
2 Czeguhn, K., "Variability With Time of Bearing Clearance Due
where to Temperature and Pressure," Paper 26, p. 136, ibid.
(2Axt + Axi_0 3 Dowson, D., Hudson, J. D., Hunter, B., and March, C. N.,
Af.j 2Pe(Axi + A f c j ) + UjHS-Ay1- "An Experimental Investigation of the Thermal Equilibrium of
Axu Steadily Loaded Journal Bearings," Paper 3, Proceedings, Institution
. X Axi-Ax^i of Mechanical Engineers, 1966-67, Vol. 181 (Part 3B), p. 70.
4 Vogelpohl, G., "Contribution to the Knowledge of Bearing
Bi.j = Pe- AxiAxi_i(Axi + As<_i) Friction," VDI Forschungsheft, No. 386, 1937, p. 28.
5 Lloyd, T., Horsnell, R., and McCallion, H., "An Investigation
Ei.j 2
= uj-HS-Ay -(Axi + As,-_i)2 into the Performance of Dynamically Loaded Journal Bearings:
Theory," Proceedings, Institution of Mechanical Engineers, 1966-
Fi.s = —a-Hi*Ag*Axt* 1967, Vol. 181, Part 3B, Paper 6, p. 49.
6 Lloyd, T., and McCallion, H., "Recent Developments in Fluid
Film Lubrication Theory," Proceedings, Institution of Mechanical
Oi.j = /Zi, r Aj/ 2 -AxiAs,_i(Axi + Ax, Engineers, Vol. 182, Part 3A, Paper 3, p. 36.
"'(!)' 7 Hakansson, B., "The Journal Bearing Considering Variable
Viscosity," Report No. 25 from the Institution of Machine Elements,
where in the foregoing Chalmers University of Technology, Gothenburg, 1964.
8 Zienkiewicz, O. C , "Temperature Distribution Within Lubri-
i cating Films Between Parallel Bearing Surfaces and its Effect on the
Ay = Pressure Developed," 1957, Proceedings, Conference on Lubrication
and Wear, p. 135 (Institution of Mechanical Engineers).
9 Dowson, D., and Hudson, J. D., "Thermohydrodynamic
The Heat Conduction Equation. Equation (4), in finite difference Analysis of the Infinite Slider-Bearing, Part 1, The Plane-Inclined
form transforms to the following: Slider-Bearing," Paper 4, Proceedings, First Lubrication and Wear
Convention, Bournemouth 1963, p. 34, Institution of Mechanical
(h)i.i = aat,j(lb)i+ij + bbi,j(lb)i_i,j + ddi,jCh)i,i+i + ffff«.;Wi.i-i Engineers.

Journal of Lubrication Technology OCTOBER 1970/ 587

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like