Sunteți pe pagina 1din 9

Engineering Failure Analysis, Vol 2, No~ 4 pp.

287-295, 1995
Pergamon Copyright © 1995 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
1350-6307/95 $9.50 + 0.00

1350-6307(95)00023-2

FAILURE ANALYSIS AND DESIGN MODIFICATION OF A


BRIQUETTING MACHINE ROLLER

SOON-BOK LEE
Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology,
Science Town, Taejon, 305-701, Korea
(Received 28 June 1995)

Abstract--In a steel-making process, a briquetting roller failed prematurely. A failure analysis


was performed and design modifications were suggested. The roller and shaft were assembled
by a shrink fitting process. The residual stress due to the shrink fit and the rolling pressure
combined to produce a fatigue loading on the roller. Cracks initiated in the water channel at
the bore of the roller due to the corrosion fatigue.

1. I N T R O D U C T I O N

In a steel-making process, steel dust was gathered between two briquetting rolls and
formed into gravel-shaped steel balls. The steel balls are used as raw materials and as
a t e m p e r a t u r e reduction agent in the steel-making process. One of the briquetting
rolls fractured after 43 days of operation. The B / Q machine was operated for an
average of 20 h per day. The roller speed was 10 rpm. The briquette volume was
20 cm 3, and the throughput capacity was 17.4 ton h -1. The roller gap was 2.4 m m . The
total pressing force was 3080 kN, and the specific pressing force was 140 k N c m -~.
The nominal operating pressure was 300 bar (30 MPa). The temperatures of the B / Q
roller machine were measured with a non-contacting temperature-measuring device.
The dust t e m p e r a t u r e was 600 °C. Cooling water at 20 °C was passed through a water
channel machined into the bore of the roller. The surface of the roller was at 152 °C
and the shaft beneath the roller was at 61 °C during operation.

1.1. Roller assembly


Figure 1 shows the roller and the roller assembly. The outer roller was heated to a
uniform t e m p e r a t u r e of 210 °C: the shaft was then inserted into the roller. The inner
diameter of the roller was 460 m m , and the outer diameter of the shaft was 460.8 m m .
The diameter of the press roller was 750 m m .

1.2. Material evaluation


The chemical composition of the roller material was evaluated and c o m p a r e d with
two other materials as shown in Table 1. The roller material has the acceptable
composition of STF4 steel. The hardness of the roller material was also tested and
formed to be 49 H R c , which is within the specification range for roller steels. The
ultimate tensile strength (Su) of STF4 steel is approximately 1352 MPa.

2. F A I L U R E O B S E R V A T I O N
2.1. Failure morphology
Figure 2 shows the failed roller. The fractured region has been cut out for failure
analysis and the working surface of the roller is shown in Fig. 2(a). The fractured
288 SOON-BOK LEE

(a)

..... t ....
320 ~
880 "l IT

4 1570

(b)
Fig. l. Briquening machine roller and shaft: (a) roller, (b) roller assembly. Dimensions in
mm.

surface was hard to read due to oxidation in air after the failure. Therefore the
surface was cleaned by the acid pickling process. Figure 2(b) shows the fracture
surface of the roller, which has been acid-pickled. By careful observation of the
fracture surface, a fracture morphology was made. Figure 3 shows the fracture
morphology. In Fig. 3 the shaded area indicates the corrosion fatigue region and the
lines indicate beach marks produced by fatigue. Two distinct failure modes exist on
the failure surface. They are the smooth fatigue fracture on the upper portion of the
surface, and the rough corrosion fatigue fracture on the lower portion of the surface.
Most of the beach marks in the corroded area were damaged due to corrosion,
except in regions A and E. Severe corrosion was identified around the water channel
e~

e~

Table 1. Chemical composition of roller materials (wt%)


O

C Si Mn P S Cr Mo Ni V

Tested 0.53 0.265 0.788 0.0075 0.0011 0.844 0.308 1.59 0.131 O
56NiCrMoV7 (DIN) 0.50-0.60 0.10-0.40 0.65-0.95 1.00-1.20 0.45-0.55 1.50-1.80 0.07-0.12
STF4 (KS) 0.50-0.60 < 0.35 0.6-1.0 < 0.03 < 0.03 0.70-1.00 0.20-0.50 1.30-2.00 < 0.20
O"
e~

=0

O0
290 SOON-BOK LEE

CRACK CUT

(a)
TIHIII Ill[lTr - ~"

: ; o4

(b) ""

Fig. 2. Photographs of the failed roller: (a) working surface of the roller cut near the fractured
surface. (b) a fractured surface of the roller that has been acid-pickled.

in regions C and D. These areas seem to be the corrosion fatigue initiation sites. The
corrosion fatigue cracks propagated progressively upwards. The corroded depth
decreased gradually from the water inlet side to the water outlet side.
Once the crack depth increased the stress intensity factor increased and, conse-
quently, the crack growth rate increased. In this region fatigue cracking is dominant,
and corrosion has not taken place due to the lack of time before the final fracture•
The distance between beach marks in this region is considerably greater than in the
corrosion fatigue region. The dotted lines indicate the crack branching traces during
crack propagation. In Fig. 3 region G shows beach marks, while the roller surface on
the left shows some shear lips, indicating the final stage of fracture.
Failure analysis and design modificationof a briquetting machine roller 291
gorier o u t e r surface
Temp. High (about 500 °C)
shear mode

A • • o
coohng water out severe corrosion area coohng water m (20 C)

•oHer inner surface


Temp. Low

Fig. 3. Fracture morphology of the fractured surface shown in Fig. 2(b). Shaded area indicates
the corrosion fatigue region. Lines indicate beach marks by fatigue.

!.2. Failure process prediction


From the failure morphology, it is clear that corrosion fatigue cracks initiated near
he water channel and propagated upwards. Final fracture occurred on the left outer
;urface of the roller. The plane of the failure surface is normal to the circumferential
lirection. Therefore, the hoop stress o00 is the most significant failure-producing
)arameter.

3. F A T I G U E A N A L Y S I S

Since the crack surface of the roller is along the shaft axis and normal to the
:ircumferential direction, circumferential stresses at the inner surface of the roller are
lsed in the fatigue analysis. The S - N diagram of the roller material has been
;stimated by considering all the factors, including corrosion and the surface condition
)f the roller. The alternating circumferential stress at the inner surface of the roller
las been predicted from the estimated S - N curve and the actual failure life of the
'oiler.

~.1. S - N curve estimation


The generalized S - N curve for steels under axial loading can be obtained by
~stimating the fatigue strength at 103 to be 0.75Su and the fatigue limit S" to be
).50S. at 106 cycles [1-3]. Figure 4 shows the generalized S - N curve for the steels
ruder axial loading. The endurance limit should be modified as [2]
S n = S;aCsCLCD, (1)
~here Cs is the surface factor, CL is the load factor (0.9 for an axial load), and CD is
292 SOON-BOK LEE

S(Iog)
0.75S u

~n
I I
103 106
N(log)
Fig. 4. Generalized S - N curve for steels under axial loading on log-log plot.

the size factor. Figure 5 shows the surface factor vs ultimate tensile strength under
various surface conditions for steel parts [2]. In Fig. 5, C s - - 0 . 2 (corroded in
water) x 0.62 (machined) for Su = 1352 MPa (about 200 ksi). The size factor can be
assumed to be [3]
C D = 1 . 1 8 9 d -0'097, (2)
with d = 460 mm, CD = 0.656. Then, from Eqn (1), the endurance limit of the roller
material becomes
Sn = (0.5 x 1352 SPa)(0.2 x 0.62)(0.9)(0.656) = 49.5 MPa. (3)
For this specific application the S - N curve can be estimated as following Basquin's
equation between 103 and 106 cycles, giving
S = 20,772(N) -°'437, (4)

where S is the fully reversed alternating stress and N is the number of cycles to
failure.

1.0
IV~drror - olished
0.9 .. Fme"-ground or I h y,~ e d~ "i ~ ' '
commerciolly polls
0.8 ' ~

O, 7 -~....
C
'\ ~ ~ ~
0.6 ~
~0.5 x
"\.?\.
0.3 ".. .

0.1 tapwo[er-~ 1 |l J "~--~


J Corroded in salt water , /
0 l l I I I I
60 80 IO0 120 140 160 180 200 220 240 260
Tensile strength Su. ksi
Fig. 5. Surface factor vs ultimate tensile strength under various surface conditions of steel
parts [2].
Failure analysis and design modification of a briquetting machine roller 293

3.2. Alternating stress estimation


The roller is subject to residual stresses due to the shrink fit assembly, operating
stresses, and thermal stresses due to the temperature difference between the bore and
the outer circumference of the shaft. The residual stress and the thermal stress are
both axisymmetric during the briquetting operation, and are the static circumferential
stress components. However, during the briquetting, the operating load produces
compression and relieves some of the hoop stress at the bore of the roller. Since the
roller is rotating at 10 rpm, this generates a fatigue loading.
Since the briquetting machine was stopped from time to time, the stress range
between the operating and rest conditions should also be considered in the fatigue
analysis. In the rest period only the residual stress remains. Since the roller was
subjected to the alternating stresses $1 (related to the operating period) and $2
(related to the rest period), Miner's linear damage rule can be applied as follows:
n l + n2 1, (5)
N1 N2
where nl is cycles of operation and n2 is cycles of rest. N1 and N2 are number of
cycles to failure corresponding to S~ and $2, respectively. Since the roller was in
service for an average of 20 h per day at 10 rpm, nl/n2 = 20(60)(10) = 12,000. This
cycle ratio strongly implies that most of the fatigue damage occurred during the
briquetting operation.
The alternating stress was estimated as S = 66.23 MPa from Eqn (4) with N =
(43 days)(20 h)(60 min)(10 rpm) = 516,000 cycles. The 66.23 MPa is the fully reversed
alternating stress at the bore of the roller.

4. DISCUSSION

In this failure analysis it was more important to find the method of improving the
service life of the roller than to find the exact stress values. Therefore the stress
analysis of the briquetting roller is left for a further study.

4.1. Estimation of the operating stress


The roller was subjected to an alternating stress Sa = (Sm~x- Smin)/2 and a mean
stress Sm= Smax- Sa. A constant life line can be constructed for the mean and
alternating stresses even though the exact values are unknown. Figure 6 shows a
Goodman line constructed with S = 66.23 MPa and Su = 1352 MPa. One possible
combination of the mean and alternating stresses at the bore of the roller for the life
of 43 days is shown by dotted lines.

86

4a

365 1352
Sm Su
Mean stress, S m (MPa)

Fig. 6. Goodman line for constant life of mean and alternating stresses.
294 SOON-BOK LEE

4.2. Design modification for life extension


4.2.1. Residual stress. The tensile residual stress generated by the shrink fit was
reduced by reducing the radial interference. However, an excessive reduction in the
interference may be harmful to the integrity of the roller assembly. In machining the
roller and the shaft, special care should be taken so that the radial displacements are
uniform throughout the shrink-on area.

4.2.2. Thermal stress. Reduce the temperature difference between the outer cir-
cumference and the bore of the roller by using a material with a higher thermal
conductivity. A 10% reduction in temperature difference produces a 10% reduction in
erG0 [4]. However, changing the material may not be allowable for other reasons.

4.2.3. Operating stress. Reducing the alternating stress is much more effective than
reducing the maximum stress. However, reducing the operating load may degrade the
performance to an unacceptable extent.

4.3. Fatigue strength improvement


The most effective method of life extension can be achieved through improving the
endurance limit by changing the factors affecting the endurance limit of the roller
material.

4.3.1. Surface factor. Among the factors influencing the endurance limit given in
Eqn (1), the surface factor Cs is the most obvious source of improvement. As shown
in Fig. 5, Cs increases from 0.62 to 0.84 at Su -- 1352 MPa (about 200 ksi) if the bore
is commercially polished. The 35% increase in surface factor causes a 35% increase in
the endurance limit. The Basquin relation is changed to
S = 15,333(N) -°'3932. (6)
At the same stress fatigue life is 1,031,960 cycles, giving a 100% increase in life.
The surface factor can be improved further by using a steel of lower tensile strength
or better corrosion resistance. The reduction in the hardness of the working surface of
the roller can be compensated for by using a surface-hardening procedure such as
carburizing.

4.3.2. Size factor. A reduction in the diameter of the shaft of 10% will produce
only a 1% increase in Co from Eqn (2), which is not significant. However, the water
channel should be smooth without abrupt changes in shape and sharp corners to
minimise stress concentration factors.

4.4. Material change for corrosion resistance


The surface factor discussed in Section 4.3.1 can be improved further by using a
steel with a better corrosion resistance. A 50% increase in Cs due to the corrosion
factor, coupled with surface polishing, gives an endurance limit of 100.6 MPa. For this
case the Basquin relation becomes
S = 10,223(N) 0.3345. (7)

The life of the roller becomes 3,488,763 cycles, giving a 576% increase in life.
Corrosion fatigue can be reduced by: (a) use of a more corrosion-resistant material
such as 56NiCrMoV7 steel, (b) reduction of corrosive action by cathodic protection,
inhibitors, or protective coatings, or (c) introduction of compressive residual stresses.
A summary of the life improvements given by various design modifications is shown
in Table 2.
Failure analysis and design modification of a briquetting machine roller 295

Table 2. Summary of life improvements by design modifications

S
Design modification (MPa) N/No* S - N curve

Original roller 66.23 1 Equation (4)


Cs = 0.2 x 0.84t 66.23 2.0 Equation (6)
Cs = 0.3 × 0.84* 66.23 6.76 Equation (7)

*No = the original failure life of 516,000 cycles.


*Surface factor by polishing.
*Surface factor by polishing and corrosion resistance treatments.

5. CONCLUSIONS

The failure of a roller of a briquetting machine has been analyzed. The failed
surface was carefully examined and the failure-generating parameters were identified.
Fhe following conclusions were made:
Ia) The shrink fit assembly of the roller and the shaft produced a tensile residual
stress in the roller at the water channel. The temperature difference between the
circumference and bore of the roller also produced a tensile stress at the inner
surface of the roller. The load from the briquetting operation produced the st/'ess
fluctuation.
~,b) The generalized S - N curve for steels was represented with a Basquin equation
which included the corrosion effect. The fluctuating stress was obtained from the
failure life and the S - N curve.
~c) The roller failed due to corrosion fatigue: cracks initiated from the water channel
or the bore of the roller and propagated towards the roller surface.
id) The effectiveness of life extension by various design modifications was shown.
Reducing the interference for the shrink fit increased the service life. Reducing
the operating load increased the service life. Doubling the surface factor for the
endurance limit produced a 576% increase in the service life.
ie) The most effective method for the life extension of the roller is increasing the
surface factor by surface treatment to resist corrosion damage. A corrosion-
resistant material, such as 56NiCrMoV7 steel, was recommended for the roller.

4cknowledgement--Dr J.-R. Park of the Technical Research Laboratories, Pohang Steel Co., provided the
;ervice history and the fracture morphology of the failed roller part.

REFERENCES

. H. O. Fuchs and R. I. Stephens, Metal Fatigue in Engineering, Wiley, New York (1980).
!. R. C. Juvinall, Engineering Considerations of Stress, Strain, and Strength, McGraw-Hill, New York
(1967).
L J. E. Shigley and L. D. Mitchell, Mechanical Engineering Design (4th edn), McGraw-Hill, New York
(1983).
L A. P. Boresi, R. J. Schmidt and O. M. Sidebottom, Advanced Mechanics of Materials (5th edn), Wiley,
New York (1993).