You are on page 1of 11

Engineering Failure Analysis 9 (2002) 457–467

www.elsevier.com/locate/engfailanal

Fatigue failure of a hollow power transmission shaft


S.K. Bhaumik*, R. Rangaraju, M.A. Parameswara, M.A. Venkataswamy,
T.A. Bhaskaran, R.V. Krishnan
Failure Analysis Group, Materials Science Division, National Aerospace Laboratories, Bangalore 560 017, India

Received 10 August 2001; accepted 22 August 2001

Abstract
A micro/hairline crack was noticed on a low speed, hollow shaft of a single stage helical gearbox during service.
Though this had not resulted in a catastrophic failure, the shaft was withdrawn from service because of leakage of oil.
Subsequent investigation revealed that the crack had initiated by fatigue at one of the keyway edges and progressed
about 3/4 of the shaft periphery in a helical manner but had not given rise to final fracture of the shaft. The fatigue
crack initiation was due to stress concentration arising from a depression mark at the keyway end surface. The problem
was further aggravated due to inadequate radius at the keyway edges and rough machining marks. An analysis of the
failure, together with recommendations for failure prevention, is presented in this paper. # 2002 Elsevier Science Ltd.
All rights reserved.
Keywords: Fatigue; Shafts; Stress concentrations; Surface finish; Keyways

1. Introduction

Shafts with circular cross-sections are used for transmission of power. Fatigue is among the most com-
mon cause of failure of such shafts. Fatigue failures start at the most vulnerable point in a dynamically
stressed area particularly where there is a stress raiser [1]. The stress raiser may be mechanical or metal-
lurgical in nature, or sometimes a combination of the two. In general, the shafts in power plant systems run
with a steady torsional load superimposed with bending stress either due to the shaft’s self-weight or pos-
sible misalignment between journal bearings [2]. Thus, these rotating components are susceptible to fatigue
by the nature of their operation and the fatigue failures are generally of the rotating-bending type [1]. The
common sites on shafts where fatigue cracks may initiate are the stress raisers occurring at the keyway root
radius and where sharp changes in cross-sectional area of the shaft occur. In spite of the preventive
measures taken during the design stage, fatigue failure can still occur due to either defects introduced
during fabrication and/or degradation of shafts during service [3]. In this paper, a case study on the fatigue
failure of a hollow power transmission shaft is described.

* Corresponding author. Fax: +91-80-527-0098.


E-mail address: subir@css.cmmacs.ernet.in (S.K. Bhaumik).

1350-6307/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S1350-6307(01)00033-4
458 S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467

2. Background

There was failure of a low-speed hollow shaft of a single stage helical gearbox during service. The low-speed
shaft of the gearbox was connected to a hydro-turbine runner shaft through a coupling. The coupling on the
shaft was fixed through a shrink fit along with a key. Oil was found leaking from the shaft near the coupling
end. On strip examination, a crack was found to have developed circumferentially through the keyway.
The cracked portion of the shaft (Fig. 1) was cut and sent to the laboratory for investigation. The details of
the damaged shaft are given below:

Material specification
Material : alloy steel—EN 24
Hardness : 276 BHN ( RC 29)
Yield stress (Sy) : 6500 kg cm2
Ultimate tensile strength (Sut) : 8600 kg cm2

Shaft dimensions
Inner diameter : 135 mm (uniform bore)
Outer diameter : 225 mm (where the shaft has failed)
Operating speed : 158 rpm
Power transmitted : 1400 kW
Number of hours in service : 550
Number of start-ups : 75

Fig. 1. Photograph showing a portion of the failed shaft containing the crack. The shaft has an outer diameter of 225 mm (where the
shaft has failed) and inner diameter of 135 mm (uniform bore).
S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467 459

3. Observations

3.1. Visual and stereo-binocular

Visual observations showed the presence of a circumferential crack at the coupling end of the shaft
running through the keyway. Fluorescent dye penetrant inspection revealed that the crack had propagated
circumferentially covering about 3=4 of the shaft periphery. The crack was through the thickness and had
propagated in a helical manner (Fig. 2).
The crack was opened for observation by making a cut of suitable length opposite the crack surface (to
reduce the net section area) and then pulling the crack surfaces apart. One of the resulting fracture surfaces
is shown in Fig. 3.
Visual examination of the fracture surfaces revealed a half-moon shaped region at the keyway along the
thickness of the shaft. In the half-moon shaped region, concentric ring type beach marks, typical of delayed
fatigue failure, could be seen (Fig. 4). These well delineated beach marks were seen throughout the fracture
surface. The fatigue crack had propagated in a helical manner along the circumference of the hollow shaft.
The fracture surface was cleaned with acetone and observed under a stereo-binocular microscope to
determine the fracture origin. By tracing back the beach marks, it was found that the fracture coincided
with one of the edges of the key way, as shown by an arrow in Fig. 4.
The keyway at the fracture origin was examined to look for any possible abnormalities. Careful exam-
ination revealed a depression like mark at the keyway run out radius region close to the crack origin
(Fig. 5). The surface of the depression mark had distinctive machining marks (Fig. 6), which indicate that
this has resulted during machining. Moreover, at this location, the surface was covered with rust suggesting
no contact with the key. The crack was found to originate at this point along the edge immediately fol-
lowing the depression mark (shown by an arrow in Fig. 5). It was also observed that the keyway had very
rough machining marks (Fig. 7). The radius at the ends of the keyway edges was measured and found to be
0.4 mm.

Fig. 2. Photograph showing the crack after fluorescent dye penetrant test.
460 S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467

Fig. 3. Photograph showing one of the mating fracture surfaces. It can be seen that the crack has propagated through the thickness
for about 3=4 of the shaft periphery.

Fig. 4. Macrofractograph showing the half moon shaped region at the keyway. The crack origin is shown by an arrow.
S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467 461

Fig. 5. Photograph showing the depression like mark on the keyway end surface. The crack origin is shown by an arrow.

Fig. 6. Photograph showing the machining marks on the surface of the depression. Note the rust in this area.
462 S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467

3.2. Scanning electron fractography

A small sample containing the fracture origin was cut, cleaned ultrasonically in acetone, and examined in
a scanning electron microscope (SEM, LEO 440I). The fracture features in the half-moon shaped region

Fig. 7. Photograph showing the machining marks on the bottom surface of the keyway.

Fig. 8. SEM fractograph at the half moon shaped region. The fracture surface is covered with a thick layer of oxides. No meaningful
fractographic features could be observed.
S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467 463

are shown in Fig. 8. It was seen that the fracture surface was covered with a thick oxide layer and there was
no discernible fracture feature. Further cleaning of the fracture surface was therefore undertaken. But, in
spite of repeated stripping of acetate replicating tape followed by ultrasonic cleaning in acetone, the oxide
layer persisted. This obviated the determination of any meaningful micro-fractographic features. More-
over, the fracture surface was mechanically damaged due to the rubbing action between the two mating
fracture surfaces during crack propagation.

3.3. Metallography

A small sample was cut from the shaft close to the fracture origin region, metallographically prepared
and observed in an optical microscope (Neophot-2) in both unetched and etched conditions. In the unet-
ched condition, stringer type MnS inclusions were seen (Fig. 9). Some of the MnS inclusions were as large
as 50 mm in length. These inclusions were aligned in the direction of the shaft axis. After etching with 2%
Nital solution (2% HNO3 in ethanol), the material showed a tempered martensitic structure (Fig. 10).
There was no abnormality in the microstructure.

3.4. Compositional analysis

Semiquantitative chemical analysis was carried out by EDX attached to SEM. All the elements in the
steel were found to conform to specification of EN24 steel except sulphur (Table 1). The sulphur content
was found to be higher (0.15 wt.%) than that of the specified value (0.05 wt.% max.).

Fig. 9. SEM photograph showing MnS inclusions. These are aligned in the direction of the shaft axis (arrow). MnS inclusions as large
as 50 mm were observed.
464 S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467

Fig. 10. Optical micrograph showing the microstructure of the shaft material. It consists of tempered martensite. 630.

Table 1
Semiquantitative EDX analysis of the shaft material

Element Specification of EDX analysis of failed


EN24 (wt.%) shaft (wt.%)
a
C 0.35–0.45
Si 0.10–0.35 0.26
Mn 0.45–0.70 0.49
S 0.050 max 0.15
P 0.050 max –
Ni 1.30–1.80 1.84
Cr 0.90–1.40 1.32
Mo 0.20–0.35 0.27
Fe Balance Balance
a
Carbon cannot be determined accurately by EDX in SEM.

3.5. Hardness measurements

Hardness measurements were carried out on a polished specimen using a Shimadzu HSV-20 Vickers
microhardness tester at a load of 500 g. The hardness value measured was HV 318 ( RC 32), which was
within specified limits.

4. Stress analysis

Though the shaft is subjected to combined bending, torsion and axial loads, in the absence of axial and
bending load data from the designer/manufacturer, only torsional load on the shaft is considered for stress
calculation.
The maximum shear stress at the surface of the shaft due to torsion is expressed as [4]

max ¼ 16Td0 = d40  d4i :
S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467 465

where,  max=maximum shear stress, d0=outer diameter of the shaft, di=inner diameter of the shaft,
T=torsional moment at critical cross-section.
Substituting appropriate data in the expression, one gets  max=437 kg cm2.
As per the ASME code, the permissible shear stress,  p=0.3 Sy or  p=0.18 Sut, whichever is less. Using
the yield stress and ultimate tensile stress for EN-24 steel,  p=1548 kg cm2.
Though the shear stress developed on the shaft appears to be much less than the permissible value, if one
accounts for the axial and bending loads on the shaft, the difference might get reduced.

5. Analysis of results

Gross fractographic features indicate fatigue as the cause of failure of the shaft. After the crack has
initiated, it has propagated under torsional load over about 3=4 of the shaft periphery. This indicates that
fatigue is of high cycle, low stress type under low nominal stress. The beach marks on the fracture surface
were well delineated. Tracing back the beach marks, the fracture origin was determined and was found to
be at the edge of the keyway (Fig. 4). After initiation, the crack has propagated along the thickness and
exited inside the hollow shaft. The resulting crack fronts have then propagated progressively in either
direction along the circumference of the shaft. The crack has propagated in a helical manner, which sug-
gests that cracking is by torsional fatigue.
The fact that the crack has originated at the keyway edges, indicates stress concentration at this point.
From the observations, there could be several reasons for stress concentration at the crack origin. These
are: (i) depression mark at the edge of the keyway due to faulty machining; (ii) inadequate radius at the key
way run out radius; (iii) rough machining marks; and (iv) elongated MnS inclusions. One or more of the
above factors could have produced a notch or stress concentrator of sufficient magnitude to cause initia-
tion of a fatigue crack. Since the fracture surface was covered with a thick oxide layer and was also
damaged due to rubbing action of the mating fracture surfaces, it was not possible to examine the fracture
origin region in greater detail to arrive at the exact cause/causes of crack initiation. However, the effects of
the above mentioned observations are discussed individually and collectively in the following section to
arrive at the most probable cause of failure.

5.1. Depression mark

Schematically, the keyway of the failed shaft is shown in Fig. 11. The locations of the depression mark,
crack and crack origin are shown. When such a depression is present at the end side of the keyway, stress
concentration is expected at the contact points with the key at either sides of the depression mark.

5.2. Radius at the keyway end edge

The sharp corners at the keyway end edges are stress concentrators and these are the most vulnerable
sites on a rotating shaft for fatigue crack initiation. Generally, to avoid stress concentrations at these
points, a generous radius or a cut out is provided. But, in the failed shaft in question, the radius is found
to be very sharp (0.4 mm) which is much smaller compared to the specified values from design data
handbook [4]. The problem is further aggravated by the presence of the depression mark at the keyway
end side.

5.3. Rough machining marks

These also act as stress raisers and in the present case can be an additive factor.
466 S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467

Fig. 11. Schematic showing the depression mark on the keyway end side. The locations of stress concentrations and crack origin are
also shown (not to scale).

5.4. Elongated MnS inclusions

MnS inclusions are normally elongated in the working direction and have sharp end points (Fig. 9).
These end points are stress concentrators and if they are located in critical regions as mentioned above, can
further aggravate stress concentration effects and facilitate fatigue crack nucleation.
From stress analysis, it is seen that the shear stress developed on the shaft is well within the permissible
limit. However, it is possible that due to the presence of one or more of the above mentioned stress raisers
together with superimposed axial and bending loads, the actual shear stress at the crack origin would
exceed the critical value necessary for fatigue crack initiation.

6. Conclusions
1. The shaft has failed by fatigue. The fatigue crack has initiated at one of the keyway end edges.
2. The fatigue crack has initiated because of stress concentrations resulting from the combined effect of
improper machining (depression mark) and inadequate radius at the keyway end edges.
3. Presence of MnS inclusions might have contributed as an additive factor.

7. Recommendations
1. Care should be taken during machining to avoid any machining defects at the keyway. A better sur-
face finish will reduce stress concentrations and thereby reduce the possibility of fatigue crack initia-
tion.
2. Providing a generous radius at the keyway run out region to minimise stress concentrations.
3. Use cleaner material.

Acknowledgements

The authors thank Mr. C.R Kannan, Ms. Kalavathi and Mr. K. Basker for NDT, scanning electron
microscopic and metallographic support. The authors are thankful to the Director, NAL, for permission to
publish this work.
S.K. Bhaumik et al. / Engineering Failure Analysis 9 (2002) 457–467 467

References

[1] Metals handbook, failures of shafts, vol. 10: failure analysis and prevention. American Society for Metals, 1975 (8th ed., pp. 373–
97).
[2] Mannuel da F. Stress intensity factors for semi-eliptical surface cracks in round bars under bending and torsion. International
Journal of Fatigue 1999;21:457–63.
[3] Berndt F, van Bennekom A. Pump-shaft failures—a compendium of case studies. Engineering Failure Analysis 2001;8:135–44.
[4] Handbook of engineering fundamentals. Wiley Engineering Handbook Series, John Wiley & Sons, NY, 1975 (pp. 537–41).

You might also like