Professional Documents
Culture Documents
a r t i c l e
i n f o
Article history:
Received 8 June 2013
Received in revised form 13 September 2013
Accepted 20 September 2013
Available online 2 October 2013
Keywords:
Failure analysis
Agitator shaft
Finite element analysis
AISI 304L stainless steel
a b s t r a c t
Fracture failure analysis of an agitator shaft in a large vessel is investigated in the present
work. This analysis methodology focused on fracture surface examination and nite element method (FEM) simulation using Abaqus software for stress analysis. The results show
that the steel shaft failed due to inadequate llet radius size and more importantly marking
defects originated during machining on the shaft. In addition, after visual investigation of
the fracture surface, it is concluded that fracture occurred due to torsionalbending fatigue
during operation.
2013 Elsevier Ltd. All rights reserved.
1. Introduction
In order to mixing of uids, mechanically stirred vessels are widely used for variety of purposes such as homogenizing
single or multiple phases in terms of concentration of components, physical properties, and temperature. Processing during
mechanical mixing occurs under either laminar or turbulent ow conditions, depending on the impeller Reynolds number,
dened as Re = qND2/l. For Reynolds numbers below than about Re 6 10, the process is laminar which is called as creeping
ow. Fully turbulent conditions are achieved at Reynolds numbers higher than about Re P 104, and the ow which has a Reynolds number between these two regimes would be considered as transitional ow [1].
Typically, a large vessel consists of three main parts: agitator shaft with impeller, top structure with motor and gearbox
and xed vessel to foundation including anchor bolting. The agitator shaft consists of two parts, namely, upper shaft and
lower shaft. These two parts are tightly connected by means of a rigid coupling to construct the main shaft [2]. General view
of the investigated shaft is illustrated in Fig. 1. The agitator shaft was mainly made of an austenitic stainless steel to resist the
corrosive media in the vessel. Since shafts are subjected to uctuating loading of combined bending and torsion with various
degrees of stress concentration, the main problem would fundamentally be fatigue loading [3].
In general, shafts are an important component used for power transmission in machinery and mechanical equipment.
Failures of such components and structures have engaged scientists and engineers extensively in an attempt to nd their
main causes and thereby offer methods for their possible prevention [47]. In this study, failure investigation of an agitator
shaft of a typical large vessel was considered. The aim of this study is to understand the main reason of fracture and avoid the
loss of product and time due to shaft failure and happening of similar cases.
2. Experimental procedure
The failed shaft was inspected macroscopically and microscopically by means of optical and scanning electron microscopy (SEM) while a great care was taken to avoid damage of fracture surfaces. Atomic absorption spectrometry was
Corresponding author. Tel.: +98 918 385 3445.
E-mail address: Shzangeneh@aut.ac.ir (Sh. Zangeneh).
1350-6307/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engfailanal.2013.09.013
156
Electric motor
Top structure
Upper shaft
Flange connection
Lower shaft
Baffles
(4 each at 90)
Impeller (3 blades)
employed to determine chemical composition of the alloy. Room-temperature tensile and impact (Charpy) tests were performed in conformity with the ASTM E-8 and ASTM E-23 standards requirements, respectively. Test-samples for impact
(Charpy) toughness are made in compliance with the ASTM-E23 of the size 10 10 55 mm, with the U notch of 2 mm
depth and with the 1 mm radius on top of the notch. The tests were made on instrumented Charpy pendulum and the obtained results showed that in all the tested specimens, major fracture energy (85%) was used for crack initiation while its
minor portion (15%) was spent on the crack propagation. To determine the hardness of the alloy, Brinell hardness (HRB) measurements were carried out on several points from the outer surface to the central zone of the polished surface of the shaft.
Carbon
Manganese
Phosphorus
Sulfur
Silicon
Chromium
Nickel
Nitrogen
Iron
Failed shaft
304L stainless steel
0.029
0.03 max.
1.05
2.00 max.
0.035
0.045 max.
0.011
0.030 max.
0.44
0.75 max.
18.1
18.020.0
9
8.012.0
0.10 max.
Balance
Balance
157
Elongation (%)
Hardness (HRB)
Failed shaft
274 15
595 10
95 3
110 8
116 3
To calculate Dr, a simple geometry such as the one shown in Fig. 6a was chosen i.e. a cylindrical shaft with a surface crack
with elliptical shape, in a plane perpendicular to the shaft axis (direction of loading). The crack shape is dened by means of
the lengths a and b representing the semi-axes of the ellipse (Fig. 6b). The stress intensity factor for the geometry and mode
of loading is:
DK Y ast
a
p
Dr pa
D
2
Y Da
158
Fig. 3. (a) Macroscopic examination of the exterior surface of the shaft revealed a number of machining grooves and (b) microscopic examination shows
micro cracks on the surface of failed shaft.
Y
a
a 2 12 a a 2 14
:
0:473 3:286
14:797
D
D
D
D
D
a
The asterisk is used to indicate only one geometric parameter (the crack depth a) is required, i.e., it is a simplied approach in which the aspect ratio a/b and the curvilinear coordinate s are not considered. This was taken, since a representative geometry (the cylinder with a straight-fronted edge crack) is employed, and the global character of the fracture
criterion is achieved, respectively [13]. Therefore, striation fatigue based on SEM observation shown in Fig. 5b was taken
nearly 2 mm of shaft surface. So, a value is nearly 2 mm, and also the diameter of the failed shaft under investigation was
100 mm. As a result, Y ast Da and axial stress range Dr would be 4.14 and 311 MPa, respectively.Final fracture zone has a
diameter of about 25 mm. The size of this zone provides information on how big the load. That is, the smaller the size is,
the lower the load applied. As it can be clearly seen in Fig. 2, when it comes to the comparison, the size of nal fracture zone
is smaller than the shaft diameter. Thus, the nominal applied loads are fairly low and the shaft material is highly ductile and
tough. Note that because of a distance between the shaft center and zone center, if the shaft is subjected to bending accompanied by torsion, the nal fracture zone relocates from center to edge of the shaft. Under higher magnication, one can see
the typical dimple features (Fig. 7a). Carbide/matrix interface which is shown in (Fig. 7b), was very weak contributed to
interfacial separation. In this case, carbides were responsible for the void nucleation.
3.4. Calculation of torque acting on the shaft
Torque calculation on the shaft during process determined through delivered power to a given uid at a constant rotational speed [2]. The power consumed by a mixer can be obtained by the following equation:
Np qn3 D5
gc
!
4
159
Ratchet Marks
Fig. 4. (a) Cracks initiation zones on ange connection and (b) ratchet marks.
Fig. 5. (a) Striation fatigues at fracture surface, some attened surface indicated by red arrow and (b) striation spacing. (For interpretation of the references
to color in this gure legend, the reader is referred to the web version of this article.)
160
a
D
b
a
2b
Fig. 6. Geometry of analysis: (a) 3D view of the shaft and (b) 2D sketch of the cracked section.
where q is density of uid (lb/ft3), Da is impeller diameter (ft), n is the speed of impeller (rps), g is Newtons law conversion
factor and Np is power number which depends on Reynolds number (NRe), Froude number (NFr) and impeller type based on
following equation:
nD2a q n2 Da
Np /
;
; S1 ; . . . ; S9
l
g
!
5
As given in Eq. (5) the power number, Np, also is a function of ratio of tank to impeller diameter (T/D), height of impeller
above vessel oor to impeller diameter (Z/D), length of impeller blade to impeller diameter (L/D), impeller diameter to width
of blade (D/W), tank diameter to bafe width (T/B), liquid depth in vessel to impeller diameter (H/D), number of impeller
blades (n), pitch/angle (Degree) and number of bafes (n), all of which included to the above equation as shape factors
S1,. . .,S9 respectively. In general, if Reynolds number goes over 104, based on available information [2] Np will be constant
and independent of liquid viscosity. Furthermore, changes in NFr have not any signicant effect on Np. Therefore, Eqs. (4)(5) can be rewritten as independent of NRe and NFr.
Np
Pg c
n3 D5a q
/S1 ; . . . ; S9
Regarding the available information [1], power number for three at blades is 2.58. For pitched blade turbines, changing
the blade angle h would change the power number by Np (sinh)2 [2]. As a result, when blades make a 45 angle with shaft,
Np is 0.4 times the value when blades are parallel with a shaft, which means the power number will be changed to 1.032
based on S8 shape factor. Some necessary data taken from the specication sheet of vessel are given in Table 3. The torque
can be calculated 372 NM in accordance to P = xT = 2pnT.
161
Fig. 7. (a) Final fracture zone of the failed shaft shows the typical dimple features and (b) interfacial separation of carbide/matrix interface.
Table 3
Necessary data extracted from the specication sheet of vessel.
Denition
Impeller diameter
Viscosity
Density
Acceleration of gravity
Speed of impeller
Values
3.44 ft
69 lb/ft3
32.17 ft/s2
1.17 rps
162
Fig. 8. (a) 3D geometry model of the agitator shaft and (b) intake ange connection.
Fig. 9. Stress contours at different stages of the analysis during the start-up of the shaft calculated according to the Von Mises criterion in the incremental
times of (a) 104, (b) 3 104, (c) 7 104 and (d) 103 s, respectively.
ment simulations. The mesh is graded in a way such that there is a higher mesh density at shaft to ange connection. This
improves the accuracy of the solution around the shaft without tremendously increasing the computational time. Mesh in
the all parts generated by sweep method. Abaqus/CAE creates swept meshes by internally generating the mesh on an edge or
163
Fig. 10. Stress contours at different stages of the analysis during the start-up of the shaft with a single machining groove with a depth of 0.2 mm calculated
according to the Von Mises criterion in the incremental times of (a) 104, (b) 3 104, (c) 7 104 and (d) 103 s, respectively.
face and then sweeping that mesh along a sweep path. Normally, a higher mesh density provides for higher accuracy but also
increases the computational time, therefore, a trade-off between time and accuracy becomes crucial. In this case, three different mesh densities were investigated. The shaft were initially meshed with 7855 elements with a higher mesh density
closer to the shaft to ange and then mesh density changed from 10,165 to 11,705 in order to reach successful convergent.
After obtaining the torque (372 NM), it was applied to a reference point 2 (Rp-2) which was coupled to lower position of the
shaft (impeller location). Also, constant rotating speed of 70 rpm dened to the reference point 1 (Rp-1) in the upper position
of the model (electric motor location). In addition, some constraints were applied for conformity to the real shaft operation.
Dynamic analysis in Abaqus/Standard uses implicit time integration method to calculate the transient dynamic response
of a system. These time integration operators are implicit, which means that the operator matrix must be inverted and a set
of simultaneous nonlinear dynamic equilibrium equations must be solved at each time increment. This solution is done iteratively using Newtons method [14]. Fig. 9ad shows stress contours at different stages of the analysis during the start-up of
the shaft calculated according to the Von Mises criterion in the incremental times of 104, 3 104, 7 104 and 103 s,
respectively. As can be seen in Fig. 9ad, stress waves propagated into the upper shaft in a very short period of time, however
due to abrupt change in the cross section of shaft to ange, stress raised at shaft/ange connection until the ow stress
reached to the bolts and then the stress relaxed by owing through the bolts to the lower shaft caused the shaft to rotate.
Although stress highly increased up to 101 MPa at the location of crack initiation at shaft to ange connection in a short period of time (7 104 s), this can be neglected by considering mechanical properties of the shaft. In general, fracture of the
shaft originates at points of stress concentration either inherent in design or introduced during fabrication or operation. In
this case, the design features that cause concentrated stress is inadequate llet radius size. Furthermore, stress concentration
produced during fabrication as a result of machining grooves.
Depending on the variables of machining technique and the quality of the cutting tool, machining can lead to fairly sharp
grooves. To understand the effect of machining grooves on stress concentration, a single machining mark with a depth of
0.2 mm was considered based on roughness measurements. Surface roughness used here was based on the difference between the highest peak and the lowest one. Fig. 10ad shows stress contours at different stages of the analysis along with
a single machining mark with a depth of 0.2 mm at start-up of the shaft calculated according to the Von Mises criterion in
the incremental times of 104, 3 104, 7 104 and 103 s, respectively. Fig. 11ab shows the effect of 0.2 mm machining
groove on stress concentration during the start-up of the shaft at the time of 7 104 s at higher magnication. Also, it can
164
Fig. 11. Effect of a single machining groove with a depth of 0.2 mm in a start-up of the shaft at the time of 7 104 s (a) side view and (b) root of machining
mark.
Fig. 12. Stress contours at steady state operation of the shaft calculated according to the Von Mises criterion.
165
be seen that such a circumferential groove increased stress concentration to as much as 50% the smooth one at start-up time.
The number of such machining marks with the size of more than 0.2 mm originated during fabrication process (Fig. 3a), so
each one play a crucial role in decreasing fatigue life and eventually contributed to failure of the shaft. In general, the major
defects that originate during machining are machining grooves, grinding burns and cracks. Machining grooves are extremely
dangerous in critical rotating parts, with life limited by fatigue. Because of their stress-raising effect, machining defects are
the preferred sites for fatigue crack initiation. A large number of service failures occur due to improper machining.
Static analysis in Abaqus/Standard was carried out to understand stress distribution during steady state operation of the
shaft. Fig. 12 depicts stress contours in the assembled agitator shaft. As it is clear due to inadequate llet radius size, stress
raised at shaft/ange connection, but the obtained maximum stress, i.e. 6 MPa, is not comparable with the one occur during
startup. Therefore, an increase in stress at shaft/ange connection might occur mostly in the service during startup or other
transient conditions.
4. Conclusion
Experimental and numerical simulation of the agitator AISI 304L stainless steel shaft was studied. Results showed that
mechanical properties (uniaxial tension, impact and hardness tests) and chemical composition of the failed shaft were in
acceptable range. Calculations showed that torque applied to the shaft during mixing operation at constant speed of
70 rpm was approximately 372 NM. Stress analysis based on the boundaries conditions mentioned earlier showed that in
the smooth shaft at 7 104 s of starting time, stress raised up to 101 MPa within shaft to ange connection. This value
in the shaft with a single machining groove with depth of 0.2 mm increases as much as 50%. In addition, Static analysis
showed that stress rose at ange/shaft connection up to 6 MPa in steady state operation. On the basis of all the above stated,
it can be concluded that inadequate llet radius size and more importantly machining grooves on the shaft surface was the
main cause of the shaft failure.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
Perry RH. Chemical engineering handbook. 7th ed. New York: McGraw-Hill; 1997.
Paul EL, Atiemo-Obeng VA, KrestaSM. Handbook industrial mixing science and practice. Hoboken (New Jersey): John Wiley & Sons; 2004.
Vincent L, Roux JL, Taheri S. On the high cycle fatigue behavior of a type 304L stainless steel at room temperature. Int J Fatigue 2010;38:8491.
Lancha AM, Lapena J, Serrano M, Gorrochategui I. Metallurgical failure analysis of a BWR recirculation pump shaft. Eng Fail Anal 2000;7:33346.
Bhaumik SK, Rangaraju R, Parameswara MA, Venkataswamy MA, Bhaskaran TA, Krishnan RV. Fatigue failure of a hollowpower transmission shaft. Eng
Fail Anal 2002;9:45767.
Sattari-Far I, Moalemi M. Failure of stainless digester shafts in a paper production plant. Eng Fail Anal 2003;10:67582.
Fahir Arisoy C, Basman Gokhan, KelamiSesen M. Failure of a 17-4 PH stainless steel sailboat propeller shaft. Eng Fail Anal 2003;10:7117.
ASTM A 27692, Specication for stainless and heat-resisting steel and shapes.
Courtney TH. Mechanical behavior of materials. 2nd ed. NewYork: McGraw-Hill; 2000.
Jian Ping J, Guang M. Investigation on the failure of the gear shaft connected to extruder. Eng Fail Anal 2008;15:4209.
Dieter GE. Mechanical metallurgy. 3nd ed. New York: McGraw-Hill; 2001.
Bates RC, Clark WG. Fractography and fracture mechanics. Trans ASM 1969;62:3809.
Athanassiadis A, Boissenot JM, Brevet P, Francois D, Raharinaivo A. Linear elastic fracture mechanics computations of cracked cylindrical tensioned
bodies. Int J Fract 1981;17:55366.
Harewood FJ, McHugh PE. Comparison of the implicit and explicitnite element methods using crystal plasticity. Comput Mater Sci 2007:48194.