Engineering Failure Analysis 18 (2011) 2093–2103

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/engfailanal

Tribological failure analysis of journal bearings used in sugar mills

S.M. Muzakkir a, Harish Hirani a,⇑, G.D. Thakre b, M.R. Tyagi b
a
Department of Mechanical Engineering, Indian Institute of Technology, Delhi, India
b
Tribology Division, Indian Institute of Petroleum, Dehradun, India

a r t i c l e i n f o a b s t r a c t

Article history: In present paper the premature failure of journal bearings encountered in sugar mills has
Received 3 January 2011 been analyzed. The causes of bearing failure are identified by simulating the operating con-
Received in revised form 15 June 2011 ditions and conducting controlled experiments on a fully automated Journal Bearing Test
Accepted 16 June 2011
Rig with provisions for varied combination (i.e. load, speed, and lubricating oil) of operat-
Available online 7 July 2011
ing conditions. The results of performance behavior (i.e. coefficient of friction, change in
surface roughness and weight loss) of the bearings as observed in these experiments have
Keywords:
been reported. The theoretical and experimental results indicate the existence of boundary
Bearing failure
Failure analysis
lubrication conditions in sugar mill journal bearings. To mitigate the problem of relatively
Sugar plant high wear, lubricating oil with boundary additives have been tried and results are reported.
Simulation tests Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction

The cane juice is extracted using cane crushing mills, which consist of 4 or 6 rolls. The mill with four roller comprised of
three crushing rolls sitting on cheeks in triangular form. To optimize the extraction of juice these crushing rolls are designed
with high coefficient of friction [1] and very low rotational (4–5 rpm) speed. The journals of the rotating rollers rest on the
plain bearings which in turn rest on the cast steel headstocks rigidly grouted on the RCC pedestals. The top roller drives the
bottom rollers (i.e. feed and discharge roller) and exerts nearly constant load on the cane blanket through a system of
hydraulic ram and accumulator as shown in Fig. 1a. This system allows the top roll to float within 5–20 mm, depending upon
the thickness of the cane blanket passing through the rolls. With this kind of arrangement, the top half of the bearing
required to support the top roller bears all kinds of the reactive loads (as shown in Fig. 1b) induced in the system by virtue
of compression of the cane blanket and pinion forces caused by transmission of power from top roll to the bottom rollers.
The premature failure of top half of journal bearings employed to support top roller is a common problem in Indian Sugar
industries. Often these bearings fail within very first working season (4–6 months) of their installation. Failures of these
bearings impact on the production loss and maintenance costs are usually high. Typical failures of the bearing surface are
photographed in Fig. 2. Visual inspection of used bearings shows scratches in circumferential direction (Fig. 2), distinct
cracks (Fig. 2b and c) in axial direction, and marks of excessive generated heat at various locations (Fig. 2a). All these obser-
vations lead to conclude the failure of bearings caused by lubrication breakdown. Long cracks in bearings often create con-
fusion of ‘‘fatigue cracking’’, which occurs due to repetitive stress cycle. But the bearings in the present case experience
relative speed of only 5 rpm (bearing life <30,000 relative-rotations) and load does not fluctuate in rotation (fatigue cycles
are much lesser than 30,000). In other words very high load is responsible for failure of the bearings. To confirm this obser-
vation made on photographs of failed bearings, it is necessary to investigate the formation of fluid film between bearing and

⇑ Corresponding author.
E-mail address: hirani@mech.iitd.ac.in (H. Hirani).

1350-6307/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2011.06.016
2094 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103

Fig. 1a. Three rollers of sugar mill placed in triangular formation.

Fig. 1b. Direction of maximum force acting on top roller.

journal. The aims of the present research are to diagnosis the frequent failure of journal bearings and outline the means of
eliminating it or reducing its frequency.

2. Background

In the current article the problem related to the failure (as shown in Fig. 2) of journal bearing in a sugar mill installation
has been analyzed. The length and diameter of these bearings are 675 mm and 530 mm respectively. The material of the
bearing is solid tin bronze (Cu 84%, Sn 10%, Pb 3%, Zn 3%; Grade LTB-1 of IS 318). The maximum pressure that a tin bronze
bearing can withstand is 50 MPa. Standard sugar mill practise limits the bearing pressure to about 10 MPa [2].
All these bearings are lubricated with high pressure dual line system and lubricant used is generally SM 175 (78–88 cSt at
100 °C) or SM 85 (38–45 cSt at 100 °C). Bearings are equipped with water chambers for evacuating generated heat during
operation.
Liu et al. [3] investigated the influence of sea water in oil on the tribological performance of plain journal bearings. As per
their experimental study additions of sea water did not affect the bearing performance, but the rotational speeds and loads
had a great influence on the bearing performance. These observations are important from the present research point of view
 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103 2095

Fig. 2. Photographs of failed journal bearing.

as there is always possibility of the presence of ‘‘extraneous matters’’ in the oil required for lubricating the cane sugar
bearings.
Burger et al. [4] investigated the failures of a magnetic drive pump journal bearing and found insufficient fluid film thick-
ness as root cause of bearing failure. They identified overload and relatively low lubricant viscosity as main contributing fac-
tors related to bearing failures.
In the present research work a systematic approach of first theoretical lubrication analysis and subsequent simulated
experimental testing were planned.

3. Theoretical analysis

To understand the working of journal bearings in roller mills a theoretical study on journal bearings was carried out. In
ideal case it can be assumed that bearing surface is separated from the journal surface by the lubricant film generated by the
journal rotation. The journal rotation causes the pumping of the lubricant flowing around the bearing in the rotation direc-
tion. The lubricant pressure distribution as function of journal speed, bearing geometry, operating clearance, and lubricant
viscosity can be described by Reynolds equation:
! !
3 3
@ h @P @ h @P 1 @ðU 2  U 1 Þh 1 @ðW 2  W 1 Þh
þ ¼ þ ðV 1  V 2 Þ þ
@x 12g @x @z 12g @z 2 @x 2 @z

In the above equation P is fluid pressure in N/mm2; x is coordinate in tangential direction in m; U2, V2, and W2 are velocity
components of bearing surface; U1, V1, and W1 are velocity components of journal surface; z is coordinate in axial direction in
m, g is fluid dynamic viscosity in N s/mm2, and h is gap between shaft and bearing surfaces. If bearing surface is stationary
and there is no motion other than motion of shaft in x direction, Reynolds equation can be reduced to [5]:
     2
@ @P @ @P R
ð1 þ e cos hÞ3 þ R2 ð1 þ e cos hÞ3 ¼ 6gx e sin h ð1Þ
@h @h @z @z C
In the above equation h (=x/R) is coordinate in circumferential direction in radians, e is eccentricity ratio, R is radius of journal
in m, x is angular velocity of shaft in rad/s, and C is radial clearance between journal and bearing in m.
The first step to solve this partial differential equation is the non dimensionalization. Assume
2096 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103

Table 1
Theoretical load capacity of sugar mill roller bearings.

S. no. Eccentricity ratio (e) Theoretical load capacity (N)

1 0.95 2.17  105
2 0.98 5.53  105
3 0.99 11.264  105

z  ¼ ð1 þ e cos hÞ ; p P
z ¼ ; h ¼  2
L 6gx CR
In the above equation L is length of bearing in m. Using these non-dimensional variables, Eq. (1) reduces to:
   2  

@ 3 @ p R 
@ 3 @ p
h þ h ¼ e sin h ð2Þ
@h @h L @z @z
Employing central difference method,
   2
2 h iþ1;j  hi1;j p i1;j 3 p
iþ1;j  p i1;j  2 p
iþ1;j þ p i;j R 3 p i;j1  2 p
i;jþ1 þ p i;j
3h i;j þ hi;j þ hi;j ¼ e sin hi;j ð3Þ
2 Dh 2 Dh ðDhÞ2 L ðDzÞ2
In the above equation i (in the range of 1 to N) and j (in the range of 1 to M) are identified as node number in circumferential
and axial directions respectively. Rearranging the above equation,
3  2 3
 Þþh
ðhiþ1;j  h   
hi;j  34 ðh  
e sin h ðDhÞ2
i1;j i;j iþ1;j  hi1;j Þ ðpi;jþ1 þ p
i;j1 Þ
i;j ¼ p
p iþ1;j 4  i1;j
þp  þ  þ4   2 5 ð4Þ
 1 þ ðRDh Þ2
2h 2h 1 þ RDh2 3 1 þ RDh2
2h 2 1 þ RLDDzh
i;j LDz i;j LDz i;j LDz

The above formulation with necessary boundary conditions (zero pressure at edge points, supply pressure at feed points) and
convergence criteria, as given in Eq. (5), were used to find the nodal pressures.

P P P P

n m
 n m


i¼1 j¼1 pi;j  i¼1 j¼1 pi;j

P iteration k
iteration k1

P
6 0:0001 ð5Þ
n m

i¼1

j¼1 pi;j

iteration k

Using Eqs. (4) and (5) computer software was made and using that software theoretical load capacities of journal bearing
were evaluated using e equal to 0.95, 0.98, and 0.99. To evaluate the load numerical integration of pressure over bearing area
using Reynolds boundary condition has been used. The results of load capacities corresponding to the eccentricity ratios of
0.95, 0.98 and 0.99 are as listed in Table 1.
The results of Table 1 indicate that the actual operating load (3.3 MN) is much higher than the theoretical load capacity.
This means hydrodynamic lubrication does not exist between roller and bearing. Therefore piezo-viscous effect of lubricant
8
assuming lubricant viscosity as function of pressure (g ¼ g0 e10 P ) was considered. On employing this relation in Eq. (1),
reformulating Eqs. (1)–(5) and finally solving provides load results listed in Table 2.
Two important conclusions can be made from results of Table 2. First conclusion is that even eccentricity ratio equal to
0.99 is unable to generate required load capacity. This means contact among asperities cannot be avoided. Second conclusion
is that the value of maximum pressure increases very rapidly with increase in eccentricity ratio. In other words the maxi-
mum pressure may exceed 50 MPa, which is the limit of maximum pressure for tin bronze material used in cane sugar mill
bearings.
In-order to investigate the operative lubrication regime in the given problem, the specific film thickness, K, was esti-
mated using the following formula,

hmin
K ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
Rrms;shaft þ R2rms;bearing
2

ð6Þ
radial clearance  ð1  eÞ
K¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R2rms;shaft þ R2rms;bearing

Table 2
Theoretical load capacity of bearings considering piezo-viscous effect.

S. no. Eccentricity ratio (e) Load capacity (N) Maximum pressure (MPa)
1 0.95 2.187  105 2.8
2 0.98 5.718  105 11.4
3 0.99 12.527  105 36.8
 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103 2097

Fig. 3. Journal Bearing Test Rig.

The operating radial clearance of bearing of 530 mm diameter varies between 200 lm and 300 lm. If its upper limit of
300 lm is considered, then specific film thickness:

300ð1  eÞ
K ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7Þ
12 þ 52
For e = 0.99, K = 0.5883 which is lesser than 1.0. This mean boundary lubrication mechanism occurs in sugar mill bearings.
This establishes that the hydrodynamic lubrication does not exist in sugar mill bearings and metal to metal contact occurs.
Similar findings were observed by Hargreaves [6]. The boundary lubrication results in high friction, and consequently, rise in
contact temperature and causes bearing failures, as shown in Fig. 2. It is necessary to validate these theoretical findings with
experimental results, which are described in next session.

4. Experimental study

The experiments were conducted on a fully automated Journal Bearing Test Rig, shown in Fig. 3, to simulate the operating
conditions prevailing in the sugar mill roller bearings. The load was applied with the help of bellows receiving pressurized air
supply. The load arm was attached to ball bearing to ensure torque free motion of journal. The lubrication system comprises
of pump for continuous supply of lubricant and temperature controller to regulate the temperature. The belt drive was used
to rotate 50 mm diameter journal. The frictional torque was measured with the help of friction force load cell. Three identical
bearings of phosphorus bronze were fabricated for the experimental study. Phosphorus bronze allows double the maximum
pressure compared to tin bronze bearing. Three identical bearings (length 50 mm, diameter 50 mm), under identical load
and speed, were selected to try three different lubricants: SM 175, SM 85 and relatively low viscosity lubricating oil with
boundary additive (named as SM 120). As per theoretical study performed as part of present research, boundary additives
play more important role compared to the high-viscous oil. In other word, liquid lubricant may be used as carrier fluid so
that boundary additive completely covers the bearings surface and minimize bearing wear under boundary lubrication con-
ditions experienced by sugar mill roller bearings.
The surface roughness of the bearing was measured at four locations spaced at 30°, on either side from the bottom hole,
before and after the experiment to observe the changes in its values. These values are reported in Table 3. To conduct the
experiments, simulated conditions can be obtained using Sommerfeld number [7], expressed by
 2
gNr LD R
S¼ ð8Þ
W C

where Nr is rotational speed in rpm. However, in the present setup maximum load is restricted up to 10,000 N and minimum
rpm is restricted to 1 rpm. In load-speed combination load of 5000 N and rpm of 10 was possible. Therefore, this combination
was selected for all experiments.
At the start of every new experiment, lubricant reservoir, flow line and the test bearing were thoroughly rinsed with the
help of the solvent. After this the test bearing was thoroughly dried and weighed. Subsequent to this the reservoir was filled
2098 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103

Table 3
Measured surface roughness of fabricated bearings.

S. no. Ra before experiment (lm) Ra after experiment (lm) Rq before experiment (lm) Rq after experiment (lm)
Bearing 1 1–1 0.9664 0.2119 1.1847 0.31
1–2 0.9067 0.5367 1.1088 0.7728
1–3 1.0827 0.8833 1.3914 1.0789
1–4 0.8607 0.8179 1.0848 1.031
Bearing 2 2–1 0.9205 0.9622 1.1725 1.2297
2–2 0.6941 0.6624 0.9071 0.8512
2–3 0.8301 0.9393 1.0287 1.1646
2–4 0.9980 0.7458 1.1907 0.8754
Bearing 3 3–1 0.8308 0.9760 1.04 1.2525
3–2 0.7864 0.3843 0.9945 0.4956
3–3 0.6870 0.9037 0.9130 1.1026
3–4 1.0586 1.2003 1.3186 1.4854

Table 4
Determination of wear of fabricated bearings.

Bearing Weight before experiment Weight after experiment Reduction in weight Percentage reduction in Duration of the test
no. (gm) (gm) (gm) weight (h)
1 105.620 105.609 0.011 0.010 3.10
2 106.014 106.004 0.01 0.009 0.51
3 106.018 106.014 0.004 0.004 3.18

Fig. 4. Coefficient of friction of bearing 1.

with the lubricant to be tested and lubricant was circulated through the flow line. The lubricant temperature was regulated
at 35 ± 2 °C. After attaining the desired lubricant temperature the journal was rotated at the desired speed of 10 rpm, and
then test bearing was loaded using the pneumatic loading leaver. A load of 5000 N was applied and resistive friction torque
was recorded. Every experiment was carried out at least for 3 h duration. After the completion of the tests, the test bearing
was thoroughly cleaned with the help of the solvent, allowed it to dry and weighed again.
 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103 2099

Fig. 5. Coefficient of friction of bearing 2.

Fig. 6. Coefficient of friction of bearing 3.

4.1. Step wise procedure

1. Set the pneumatic pressure through the compressor to 6–8 bars.

2. Set the pneumatic bellow position as per the load requirement (small/big bellow).
2100 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103

Table 5
Radial clearance of manufactured bearings.

Bearing no. Diameter Radial clearance (mm) Radial clearance (lm)

8 50.4511 0.2255 225.550
3 50.1344 0.0672 67.200
5 50.0862 0.0431 43.100
1 50.0857 0.0429 42.850
9 50.0808 0.0404 40.400
6 50.0783 0.0391 39.150
7 50.0615 0.0308 30.750
10 50.0613 0.0307 30.650
12 50.0472 0.0236 23.600
2 50.0420 0.0210 21.000
11 50.0381 0.0191 19.050
4 50.0191 0.0096 9.550

3. Fill in the lubricant reservoir with 600–700 ml solvent (toluene).

4. Start the lubricant pump for approx. 10 min for the circulation of the solvent through the flow line. The solvent will
carry away the dirt and the traces of the lubricants if any present in the flow line.
5. Remove the solvent from the reservoir and the flow line thoroughly and re-rinse the system with the lubricant to be
tested according the above mentioned procedure.
6. Fill in the lubricant reservoir with the lubricant to be tested.
7. Weigh the bearing to the three decimal places of gms.
8. Assemble the test bearing in the bearing housing and mount the complete assembly on the journal.
9. Assemble the journal through the support bearings and lock its position with the help of lock nut and the belt pulley
arrangement.
10. Connect the temperature sensors with the control panel for the online monitoring of temperature within the bearing.
11. Switch on the oil pump and the heater.
12. Adjust the lubricant flow rate using the flow control valve.
13. Allow the lubricant temperature to reach the set value of 35 ± 2 °C.
14. Position the torque sensing lever and apply a static load on the test bearing for the check.
15. Start the motor and allow the journal to gain the desired speed of 10 rpm.
16. Apply a load of 5000 N on the test bearing with the help of pneumatic loading system.
17. Monitor and acquire the resistive friction torque with the help of the sensor.
18. Run the test for 3 h.
19. After completion of the test remove the bearing and clean it thoroughly with the solvent and weigh it.

The weight of the bearing before and after the experiments was measured to estimate the amount of wear. The results are
listed in Table 4.
The results of Table 4 indicate severe wear for bearing number 2. This bearing was lubricated with SM 85 oil. Due to
excessive heating test setup was stopped before completion of 1 h test. Bearing number 1, which was lubricated with SM
175 oil, could tolerate operating conditions, but showed excessive wear compared to bearing number 3. Wear even in the
case of bearing 3 was observed, but at relatively low rate. In other words it can be stated that it was difficult to operate bear-
ing which means existence of metal to metal contact. Such conclusion can also be made based on the plots of friction coef-
ficient shown in Figs. 4–6.
To ensure the reputability of the test, twelve news bearings were manufactured. The bearings were manufactured
with intention of radial clearance ranging between 25 lm and 40 lm. The detail of radial clearances is provided in
Table 5.
Two bearings (bearing numbers 8 and 3) were rejected due to excessive radial clearance and one bearing (bearing number
4) was rejected due to very low radial clearance. Experiments were performed on remaining nine bearings. First three bear-
ings: bearings numbers 12, 2, and 11 were experimented using SM 175, SM 120 and SM 85 oil respectively. Experimental
results of coefficient of friction for all three bearings are shown in Fig. 7, which confer the earlier made conclusion that
SM 85 is the least preferred oil compared to other two (SM 120 and SM 175) oils. Therefore in further experiments only
two oils SM 120 and SM 175 were used. Results of experiments performed on bearing numbers 7 and 10 are shown in
Fig. 8, which indicate that performance of oil SM 175 is improved with increase in radial clearance. To ensure this, the exper-
iments on bearings 9 and 6 were performed and results of which are illustrated in Fig. 9. These results indicate that coeffi-
cient of friction reduces with increase in clearance from 30 lm to 40 lm. Performance of SM 175 and SM 120 are
comparable. Finally experiments were performed on bearings 5 and 1, and results are presented in Fig. 10. These results
of coefficient of friction are higher compared to results of Fig. 9. The weight loss of bearings 1 and 5 are listed in Table 6.
 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103 2101

Fig. 7. Comparison in coefficient of friction for three oils.

Fig. 8. Coefficient of friction for bearings 7 and 10.

The scratches on bearing working surfaces are exposed in Figs. 11 and 12. These results indicate an increase in coefficient
of friction and wear of larger clearance bearings lubricated with both (SM 120 and SM 175) lubricants.
2102 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103

Fig. 9. Coefficient of friction for bearings 9 and 6.

SM-175

SM-120

Fig. 10. Coefficient of friction vs. time for bearing no. 1 (SM-120) and 5 (SM-175).

Table 6
Weight loss of bearings.

Sl. no. Bearing no. Wt. before test (gm) Wt. after test (gm) Wt. difference (gm)
01 01 109.477 109.471 0.006
02 05 110.866 110.861 0.005
 S.M. Muzakkir et al. / Engineering Failure Analysis 18 (2011) 2093–2103 2103

Fig. 11. Bearing no. 1 after the test showing loaded region (lower portion).

Fig. 12. Bearing no. 5 after the test showing loaded region (lower portion).

5. Conclusions

 The theoretical study of the sugar mill journal bearings has indicated the existence of boundary/mixed lubrication con-
ditions. The possibility of hydrodynamic conditions for which the journal bearing is being used is difficult to achieve. All
these conditions are responsible for the premature failure of the said bearing.
 The experiments were conducted on the fully automated Journal Bearing Test Rig to simulate the conditions prevailing in
the sugar mill roller bearings during its operation. The experiments with different lubricating oils indicated that the use of
lubricating oil with EP additives significantly reduce the coefficient of friction and amount of wear (weight loss).
 The experiments performed on bearings having radial clearance in the range of 0.075–0.17% of bearing radius. Relatively
low friction and low wear were observed for bearings having clearance in the range of 0.1–0.15% of bearing radius.

References

[1] Oliver G, Buchanan V, Cooke K. A tribological study for an increased coefficient of friction in the extraction of sugarcane juice. Tribol Trans
2007;50:198–204.
[2] http://www.sugartech.co.za/extraction/millbearings/index.php. [ accessed. 31.12.10].
[3] Liu H, Xu H, Zhang Y, Ji H, Ge Y, Ellison P, et al. The influence of sea water in oil emulsion on bearing performance. Proc IMechE Part J: J Eng Tribol
2009;223:457–68.
[4] Burger NDL, Ghela PN, Mulder AO, Nkambule PM, Schoombee A. Tribological failure analysis of a magnetic drive pump. Eng Fail Anal 2009;16:1501–9.
[5] Hirani H, Rao TVVLN, Athre K, Biswas S. Rapid performance evaluation of journal bearings. Tribol Int 1997;30:825–34.
[6] Hargreaves DJ. The lubrication of mill brasses. Proc Aust Soc Sugar Cane Technol 1984:191–7.
[7] Khonsari MM, Booser ER. Proper film thickness key to bearing survival. Mach Des 2006;14.