You are on page 1of 7

Engineering Failure Analysis 16 (2009) 2274–2280

Contents lists available at ScienceDirect

Engineering Failure Analysis


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

Fracture analysis of collapsed heavy-duty pulley in a long-distance


continuous conveyors application
J.A. Martins a,*, I. Kövesdy a, I. Ferreira b
a
Product Engineering Department, Material Buckling Machines, Metso, Brazil
b
Department of Materials Engineering, Faculty of Mechanical Engineering, University of Campinas, Brazil

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

Article history: In this paper we present a case study of a heavy-duty pulley that failed after 2.1  107
Received 15 December 2008 cycles. The pulley was installed as a take up on a continuous conveyor under a load of
Received in revised form 10 February 2009 350 kN transmitted by the counterweight tower. The failure was analyzed using (a) finite
Accepted 5 March 2009
element analysis by Ansys, (b) field data analysis, and (c) fracture macro examination. The
Available online 29 March 2009
prior FEA analysis showed that the pulley was designed according to its field application
and moreover when re-simulated, by applying higher loads, revealed high potential
Keywords:
towards its safety factor. The field examination showed that the conveyor did not suffer
Pulley
Finite element analysis
any overload that would contribute to the pulley collapse. The macro failure analysis
Failure revealed stress discontinuities and fatigue symptoms such as beach marks and even chev-
Welding ron marks within the welding region. In the meantime the welding region did not show
Fatigue evidence of thermal affected zone (TAZ), which is the key to its mechanical strength. The
case analysis showed that the pulley was designed in accordance with British Standard
[British Standards Institution, BS 5400:1980. Steel, concrete and composite bridges: Part
10: code of practice for fatigue; 1980] but a welding failure generated unexpected stresses
and weakness on the welding region which generated fatigue and the final fracture.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Pulleys are one of the main components used in a belt conveyor, supporting the loads generated from the belt tension
according to its wrapping angle. Design and manufacturing technologies are of primary importance to this structural com-
ponent’s performance. Failures—while not welcome—have to be faced as treasures to obtain consistent product improve-
ments. The application of design tools like finite element analysis (FEA), field data information completion and fracture
surface investigation brings rich knowledge to better understanding the product and its associated working behavior. Many
publications deal with the difficulty of fatigue/or manufacturing technique, especially for welded components (design of
weld seam, welding procedures), critical areas include the end disc, the shell of the pulley or the connection of different com-
ponents [1,2]. This paper focuses on the usage of powerful tools and proper techniques to determine the most probable root
cause of failure and then to give valuable information and resources for improvement to both the product and manufacturing
development areas.

* Corresponding author. Tel.: +55 15 21023576; fax: +55 15 21021694.


E-mail address: jairo.martins@metso.com (J.A. Martins).

1350-6307/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2009.03.022
J.A. Martins et al. / Engineering Failure Analysis 16 (2009) 2274–2280 2275

2. Methods

The techniques and tools used to determine, in a wider manner, the most probable failure root cause are data field anal-
ysis, finite element analysis (FEA), and macro examination of the failed surface [3]. All these methods are described below.

2.1. Field data

In order to have a better understanding of the environment where the pulley was used during its collapse, a monitoring
graph was obtained from the field production monitoring system to verify the occurrence of any overloading or other
mechanical impact, which could have happened at the time of the incident.

2.2. Pulley design – finite element analysis (FEA)

Based on the original simulation by FEA, an additional simulation was performed to determine the load needed to reach
the allowable stress in the pulley’s components. It is well known that the allowable stress is far different from the ultimate
stress and takes also into account the cycling stresses with the time, as well as factors based on the component environment
application [4]. The boundary condition of this simulation was a load applied on the shaft representative of the pulley load-
ing reaction. The software package Inventor 2009 [5] was used to generate the solid and Ansys 2009 [6] to simulate the stres-
ses. The shaft was simulated as a solid with mesh relevance setting and converged mesh with 44,906 nodes and 27,245
elements. The materials in the simulation were assumed to be homogeneous and isotropic, the Poisson ratio to be 0.3
and Young Modulus to be 2.1  105 MPa.
The shaft calculation of infinite fatigue lifetime was based on ANSI/CEMA B105.111992 [7], Shigley [8] and Dieter [3]. The
estimated fatigue limit to the pulley follows:
32  FS  M b
Sf ¼  ðka  kb  kc  kd  ke  kf Þ  ð0:5:SUT Þ ð1Þ
p  D3
where FS is the safety factor, Mb the bending moment, D the shaft diameter, SUT the material tensile strength, and ki (i = a–f)
are the modifying factors related to the surface condition, size, load, temperature, reliability, and miscellaneous effects,
respectively. It was used, in the calculation, FS = 1.5, ka = 0.8, kb = 0.59, kc = 0.897, kd = 1.0, ke = 1.0 (reliability of 50%), and
kf = 1.0. The stress concentration factor was included in the stress, according to Shigley [8].
The allowable higher stress disc/shell welding calculation was calculated according to British Standard [9] (see Eq. (2))
based on an assumed lifetime of 107 cycles.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m K  D
d
0
rr ¼ ð2Þ
N
where rr is the stress range, D the anti-log reciprocal of the standard deviation of log N, d the number of standard deviation
below the mean-line, N the number of cycles estimated to failure to a stress range of rr, K0 the constant term related to the
mean-line of the statistical analysis results, and m is the inverse slope of the mean-line log rr–log N curve.
The weld type number 3.4 class F was chosen based on British Standard [9] to the Pulley calculus once it is driven com-
ponent and the main stress applied on it is bending. The crack shown in the Standard to this welding type is perpendicular to
the pulley longitudinal and similar to the failure characteristic found. British Standard [9] also says ‘‘if the transverse mem-
ber is relatively stiff (i.e. its width is at least 1.5 times the width of the first member) and the full penetration butt weld is
used in accordance with the recommendations of type 3.5, the classification in this particular case may be considered to be
effectively F2 with the stress concentration factor of unity. Otherwise the class shall be F with the appropriate stress con-
centration factor”. The stress concentration factor is used unless third member is plate or has continuity plating [9]. The type
number 3.4 description reinforce that if the backing strip is fillet or tacked within 10 mm of the member edge it won’t be
reduced below Class F, unless permanently tacked within 10 mm of the member edge, in which case it will be class G (type
2.11). The worse scenario class W is related to welds primarily applied to all fillet or partial penetration butt welds where
bending action across the throat does not occur, which again is not the case due to the full penetration welding type applied
on the pulley design and the load type associated.
Theoretically—and usually in application—fatigue cracks nucleate perpendicularly to the highest tensile stress state [3].
Briefly the factors which contribute to this weld class choice are due (a) the welding is a full penetration type and (b) there
is a permanent back strip beneath the welding, see Fig. 1.

2.3. Mapping the fractured surface

The fracture surface was carefully cleaned in order to facilitate the analysis of the fracture surface. The cleaning and the
surface analysis were done at the same field where the failure took place. The surface was mapped by dividing the pulley
circumference in eight regions covering 360° of the failed surface. Moreover, the distance between the pulley surface and
2276 J.A. Martins et al. / Engineering Failure Analysis 16 (2009) 2274–2280

Fig. 1. Sketch of the pulley fractured mapping.

the failed surface was measured in order to improve the morphology of the fracture across the pulley cross-section path
tracking (distance Z, Fig. 1).

3. Results & considerations

3.1. Field analysis parameters

The curves in Fig. 2 reveal the data gathered from the production monitoring system on the days on which the collapse of
the heavy-duty pulley happened. According to the monitored data, the belt conveyor capacity was 5.5 tons per hour on aver-
age and the motor electrical current indicated 29.95 A. The graph in Fig. 2 is almost flat with no abnormality indicated by the
monitored parameters. This could reveal the presence of overload on the pulley. The vertical line close to the date reveals the
exact moment of collapse: October 18th at 10:54:16 p.m. The total pulley lifetime since its start up was 2.1  107 cycles.

3.2. Finite element method (FEA)

The maximum equivalent stress of 138 MPa is found when the pulley shaft region is calculated by finite element analysis
(FEA) to a load three times higher the project normal regime (1400 kN against 450 kN). This maximum should be doubled
once the region passes through tensile to compressive stress state during its operation cycle. The range of 276 MPa conse-
quently had to be maximum 210 MPa according to AISI/CEMA [7]; it generates a negative safety factor of 25% to the mate-
rial L-20 (alloyed steel by Metso) on the shaft radius region. When the maximum range (according to British Standard [9]) is
reached, the stress at the shaft radii exceeds the calculated Shaft Infinite Lifetime [7]. Therefore, it is understood that the
weak region in this pulley is the shaft radii, instead of the welding, because the load of 1400 kN on the shaft has 66 MPa
beyond the limit and the welding. This value is very close to the limited value of 40 MPa by British Standard [9].
When analyzing the disc/shell connection to the same loading (3 regime), the stress range level of 31 MPa is verified.
The stress range calculated according to British Standard [9] is 40 MPa, as seen in the Table 1. The type of welding connection
done in the pulley, type number 3.4 class F weld, was the basis for calculated the stress limit at this region; see Section 2.2.
These calculations are also shown in Table 1.
In terms of stresses on the working regime, it was verified that the actual, calculated limit of shaft stress (108 MPa) was
far lower than the limit indicated by FEA (221 MPa), while the welding presented a stress of 12 MPa (tensile plus compres-
sive – see Figs. 3 and 4) rather than the 40 MPa indicated according to the Standard [9]. Comparing the safety factors, there
was a safety factor of 167% in the welding region against a factor of 105% in the shaft. Again, this indicates that the weakness
J.A. Martins et al. / Engineering Failure Analysis 16 (2009) 2274–2280 2277

Fig. 2. Monitoring the production field parameters.

Fig. 3. Equivalent stress on the pulley shaft (MPa).


2278 J.A. Martins et al. / Engineering Failure Analysis 16 (2009) 2274–2280

Fig. 4. Equivalent stress on the disc/shell welding region (MPa).

Table 1
Pulley description of materials and calculations.

Loading Drum Material Load Weld Stress range Yielding Ultimate Shaft infinite Stress range Shaft safety Weld safety
part (kN) type (MPa) limit strength fatigue life (MPa) weld (F) factor factor
*
[FEA] (MPa) (MPa) (MPa) AISI/CEMA British [9] British [9] (%) British* [9] (%)
*

3 Regime Disc/shaft L-20 1400 n/a 276 410 620 221 n/a 25
shell ASTM A36 1400 F 31 210 450 40 29
Regime Disc/shaft L-20 450 n/a 108 410 620 221 n/a 105
shell ASTM A36 450 F 12 210 450 40 167

in the pulley design is in the shaft radii rather than the welding region. To give the reader another idea of this disparity, the
calculated infinite fatigue life stress to the shaft material (L-20) is 1/3rd of its tensile strength, while the stress to the pulley
shell made in material ASTM A 36 less than 1/12th.
In terms of design, the welding is assumed as to have the same mechanical properties as the softer base material, ASTM A
36. The pulley is welded first using pre-heating and afterwards passing through a stress relieve heat treatment, a fact that
lends support to this assumption.

3.3. Mapping the fractured surface

As mentioned in Section 2.3, the fracture surface was mapped towards its path along the pulley’s failed cross-section, ref-
erenced in Fig. 1, afterwards following path indicated in Table 2 and Fig. 5. Noticeably, the fractured surface occurred pri-
marily within the welding region, between the disc and the shell, although in certain regions the shell is partially sliced,
Fig. 5. From this figure, it is seen that at points A, B, C, and H, the shell is sliced, while at the remaining points D, E, F,
and G the failure is contained within the welding region, being within the 10 mm described in the Fig. 1.
J.A. Martins et al. / Engineering Failure Analysis 16 (2009) 2274–2280 2279

Table 2
Failure path variation.

Region (R) Drawing distance (D) Z K (ZD)


A 171 180 9
B 220 49
C 180 9
D 165 6
E 165 6
F 165 6
G 170 1
H 185 14

Fig. 5. Fracture path variation.

Fig. 6. Region C (K = 9 mm/transition shell/welding).

Fig. 6 reveals a hole at the disc-shoulder surface, which generated a detrimental stress raiser that resulted in a crack
nucleation and propagation. The beach marks in the same Figure are common fatigue characteristics. The shorter distance
near the discontinuity reveals that this was the first place where a nucleation took place and initiated propagation (see
Fig. 7).
Non conforming manufacturing welding processes may possibly have influenced the thermal affected zone (TAZ) zone
strength. This structural component depends strictly on the welding integrity and the ZTA strength (thermal affected zone).

4. Conclusions

No overload or mechanical impact was observed that would justify the pulley collapse. The project loads on the pulley,
when simulated by FEA generated low stresses when compared with the used materials’ mechanical properties, the standard
2280 J.A. Martins et al. / Engineering Failure Analysis 16 (2009) 2274–2280

Fig. 7. Mapping the failure mechanics.

allowable stresses or even the fatigue calculation to an infinite life. The maximum load based on the design allowable stres-
ses is (fatigue limit) 1400 kN, which is significantly higher than the project requirement of the 450 kN and either of the val-
ues found in the field 350 kN. The pulley is overdesigned especially with respect to welding requirements, which create a
significant safety factor when compared with the project requirement based on FEA. If a safety concern was present at
the time when the pulley was designed, it should have been addressed in the design of the shaft radius, which is the region
with the highest stresses, rather than in the welding requirements.
Despite of selecting appropriate type of weld and class it was not able to predict the interaction between the detrimental
factors of welding poor penetration (TAZ) and the high stress concentration factor (Kt) generated by the hole during the
manufacturing.
This discontinuity increased the stresses locally in a manner which exceeded the material fatigue strength limit and
which culminated in fatigue crack nucleation. The failure phenomenon was high cycle fatigue. The thermal affected zone
(TAZ) was not evidenced along the failure surface and has in great part contributed to the final failure once the pulley con-
nection did not have enough strength to support the load.
As a result of this analysis, in critical cases, the pulley should periodically be inspected for cracks, Ultra-sound inspection
is strongly advised. If the stress concentration factor was not present in the pulley and the TAZ was properly done the prod-
uct’s lifetime would be much longer.

Acknowledgments

The acknowledgements are addressed to CNPq (National Council for Scientific and Technological Development – Brazil),
Mr. Carlos Ivan Equi, Systems Executive Manager, Mining Operations at Metso Minerals, and Mr. Rubens Costa, Vice Presi-
dent, Mining Operations, South America at Metso Brazil and Dr. Thomas Jackson from Rona Consulting Group.

References

[1] Jones DRH. Fatigue of welded conveyor drums. Eng Fail Anal 1995;2:59–69.
[2] Affolter Ch et al. Fatigue in the shell of a conveyor drum. Eng Fail Anal 2007;14:1038–52.
[3] Dieter GE. Mechanical metallurgy SI metric edition. McGraw-Hill Book Company; 1988.
[4] Provenza F. Projetista de Máquinas, 71th ed. Editora F. Provenza; 1996.
[5] Autodesk Inventor Professional. Version 2009, theory manual; 2008.
[6] Autodesk Ansys Technology. Version 2009, theory manual; 2008.
[7] AINSI/CEMA B105.1-1992. Specification for welded steel conveyor pulleys with compression types hubs. Conveyor Equipment Manufacturers
Association; 1992.
[8] Shigley JE, Mischke CR, Budynas RG. Mechanical engineering design. 7th ed. McGraw-Hill; 2004.
[9] British Standards Institution, BS 5400:1980. Steel, concrete and composite bridges: Part 10: code of practice for fatigue; 1980.

You might also like