Engineering Failure Analysis 16 (2009) 1542–1548

Contents lists available at ScienceDirect

Engineering Failure Analysis

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

Fatigue in engine connecting rod bolt due to forming laps

S. Griza *, F. Bertoni, G. Zanon, A. Reguly, T.R. Strohaecker
Departamento de Metalurgia, Universidade Federal do Rio Grande do Sul, Av. Osvaldo Aranha, 99/610, Porto Alegre 90035 190, Brazil

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

Article history: An analysis was performed to asses the failure root cause of an automotive diesel engine
Received 30 September 2008 which experienced collapse only 6 month after revision. The connecting rod bolts torque
Accepted 18 October 2008 disassembly was monitored and fractured parts were selected to laboratory fracture anal-
Available online 25 October 2008
ysis. It was verified with fatigue rupture of one of the fourth connecting rod bolt. Tensile
tests were performed in four of the remaining connecting rod bolts. During this procedure,
Keywords: it was verified another bolt with fatigue crack propagation an indication that the first fati-
Connecting rod bolt
gued bolt did not have suffer torque relaxation. A finite element analysis was performed in
Fatigue
Failure analysis
connection with an analytical fracture mechanics approach aiming to evaluate the relation
Design between tightening force and fatigue crack propagation in connecting rod bolts. The engine
Forming laps collapse occurred due to forming laps in the grooves of the bolt shank. Finally, some design
improvements were suggested for avoid future failures: a gap in the groove length at the
connecting rod cap interface, enough to avoid combination of forming laps and higher
stress amplitude; increase of the bolt torque assembly to reduce stress amplitude.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Failure of bolts can result in catastrophic events in an overall complex system [1]. Despite the wide understanding regard-
ing bolt mechanics, cases of failure are still reported. The most common factors which contribute to failure are low tighten-
ing force, design defects specially concerning thread root radius, material or fabrication defects which can be achieved in
material selection, heat or mechanical treatment [2–6]. In this paper, a failure analysis was performed in a 6.6 automotive
diesel engine which experienced collapse in service only 6 months after revision. It was reported that at the time of revision,
the engine was opened and some parts such as seals, bushings and connecting rod bolts were replaced. The failure analysis
methodology was evaluated, based on the experimental sequence suggested by the ASM [7]. For this analysis the engine dis-
assembly was performed and preliminary visual analysis of damaged parts was conducted to select the fractured parts that
could be the failure root cause. Laboratory fracture analysis, mechanical tests, metallographic analysis and numerical mod-
eling with an analytical fracture mechanics approach were performed.

2. Materials and methods

2.1. Disassembly of the engine

The first observation showed fracture of the engine block, due to impact of the fourth connecting rod, whose cap was dis-
assembled as a result of bolts failure. The engine head was then disassembled showing camshaft fracture in three parts and
some other resulting damages such as: piston marks due to valve contact, secondary cracks in the fourth piston, as well as

* Corresponding author. Tel.: +55 51 3308 4251; fax: +55 51 3308 3565.
E-mail address: [email protected] (S. Griza).

1350-6307/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2008.10.002
 S. Griza et al. / Engineering Failure Analysis 16 (2009) 1542–1548 1543

warping of valves, valve lifters and valve rods. No abnormal scratch marks in the cylinder liners were observed after block
and pistons disassembly. The fourth connecting rod bolts fractured at the plane of the cap interface. The bolts used were
Whitworth UNF7/16 with 20 threads per inch a 12 mm shank diameter and 68 mm tightening length. Those bolts had 18
shank grooves lengthwise (Fig. 1), which provide bolt assembly alignment with low contact surface, to enhance fretting
and corrosion resistance. The bolt manufacturer recommends 100 N m assembly torque. Disassembly of all connecting rods
was performed using a ratchet torquimeter as a way to verify the residual torque present at the bolts.

2.2. Fracture analysis, mechanical tests and metallographic analysis

Two of the fractured bolts, camshaft pieces as well as bolts from the remaining connecting rods were selected for labo-
ratory analysis. Those analyses consisted of low magnification and scanning electronic microscopy (SEM) observation of the
fracture surface, hardness, tensile tests and metallographic characterization. Hardness was performed over the flat shank of
some selected bolts using the Rockwell C scale (HRC). Tensile tests were performed with a MTS 810 (MTS Corporation, Eden
Prairie, MN). Metallographic analyses were done in two selected bolts: one broken during the failure, other during tensile
test.

2.3. Numerical modeling

The assembly between connecting rod and bolts was finite element modeled, in order to analyze the correlation be-
tween the tightening force, the external in service loads and the resulting stress magnitude in the bolt shank. The net
generation, boundary conditions as well as numerical analysis were done using commercial software. The assembly con-
sists of bolts, connecting rod body, connecting rod cap and a rigid shell used to distribute external load at the connecting
rod cap. The connecting rod has 36 mm width and 80 mm crankshaft diameter. The assembly was simplified utilizing two
planes of symmetry, to facilitate the net generation and decrease calculation time. Fig. 2 shows a quarter of assembly,
including boundary conditions and net generated. The bolt net was refined at fracture plane, which is the site in analysis.
Nut and head interface sites also were refined. The parts were considered isotropic and homogeneous and all simulations
were in the linear elastic regime. Table 1 lists net characteristics for each part. Constitutive properties assumed for all
elements, excluding the rigid shell, were 200 GPa Young modulus and 0.3 Poisson coefficient. Tightening force was sim-
ulated according to interference technique, which is present in the software package. It consists in assess the strains in
the bolt; taking into account force, area and elastic module. Tightening force is that normal to the bolt axis as result of
the tightening torque. The torque applied in the bolt (T) is related to the tightening force (F), the bolt nominal diameter
(d) and a nut factor or torque coefficient (k) according to Eq. (1). The literature indicates a nut factor k = 0.2 is a general
good estimative [8,9].
T ¼ F dk ð1Þ
The displacement due to the force is imposed as interference in the simulated assembly. The faces in interference are
automatically adjusted pondering the stiffness differences between the bolt and the connecting rod, approaching the real
situation. Considering 100 N m torque, 11.1 mm nominal bolt diameter and 0.2 torque coefficient, the resulting tightening
force (Fa) is 45 kN. Table 2 shows three magnitudes of simulated forces (Fa) and the respective interferences. They were sim-
ulated to demonstrate the expected decrease of stress amplitude with increasing tightening force. The highest magnitude
60 kN shall be near the yield strength of the bolt. From this first analysis it is possible to obtain the normal stress due to bolt

Fig. 1. General view of the bolts collected for mechanical tests. Whitworth UNF7/16 20 threads per inch with 12 mm diameter shank and grooves
lengthwise.
1544 S. Griza et al. / Engineering Failure Analysis 16 (2009) 1542–1548

Fig. 2. Connecting rod–bolt assembly, boundary conditions and net generated.

Table 1
Net characteristics of four parts of the model.

Part No of elements No of nodes Element type

Bolt 3672 5044 Linear hexaedric
Connecting rod cap 50,698 10,217 Linear tetraedric
Connecting rod body 25,305 5239 Linear tetraedric
Rigid shell 713 768 Linear quadractic rigid

Table 2
Tightening force simulated and the respective interferences.

Model Thightening force (kN) Interference (mm)

1 30 0.09
2 50 0.15
3 100 0.30

tightening. As a second step, external loads of 15, 30 and 60 kN were applied in the axis of symmetry, normal to the shell, in
order to separate the cap of the connecting rod. This procedure increases the stress level at the bolt. The difference between
stresses due to second and first steps results in the stress amplitude imposed to the bolt.

2.4. Fracture mechanics analysis

For the connecting rod bolt stress intensity factor calculation, it was adopted the procedure described in the standard BS
7910:2000 [10]. The section M.6.2 of the standard is dedicated to geometric factor for cylindrical bars with semi-elliptic
cracks and direction of propagation perpendicular to the axis. The stress intensity factor (KI) is related to crack length (a),
stress amplitude (r) and geometrical factor (Y) according to Eq. (2). The geometrical factor (Y), was obtained from Eq. (3),
the stress intensity magnification factor (Mn) was calculated from Eq. (4), while the geometrical factor (g) was taken from
Eq. (5). These equations are valid for (a/2r) < 0.6, where (r) is the radius of the cylindrical bar.
 S. Griza et al. / Engineering Failure Analysis 16 (2009) 1542–1548 1545

K I ¼ Y rðpaÞ1=2 ð2Þ
Y r ¼ Mm rmax ð3Þ
 a n pao3 
Mm ¼ g 0:752 þ 2:02 þ 0:37 1  sen ð4Þ
4r 4r
1:84
p ftanðp4raÞ=ðp4raÞg1=2
g¼ ð5Þ
cosðp4raÞ

3. Results

3.1. Fracture analysis

The camshaft fractured due to overload, without signals of fatigue as can be seen in Fig. 3. This fracture was a conse-
quence of impact between valves and pistons. One of the fourth connecting rod bolts failed due to overload as well, with
evident distinct fibrous and shear zones (Fig. 3). The other bolt of the fourth connecting rod shows a flat fatigue fracture sur-
face with clear indication of beach marks (Fig. 4). The fatigue nucleated at the top of one of the grooves (Fig. 5) and prop-
agated along most of the section until final fracture. Only one nucleation site and a large propagation area are evidences
of low applied stresses. The fracture initiation site located close do the crankshaft. The SEM analysis highlighted stable crack
propagation up to 2/3 of the cross section when dimples characteristics of instable propagation can be observed (Fig. 4).

3.2. Mechanical tests

Hardness test was performed on the fatigued bolt while hardness and tensile tests were performed in another four
selected bolts. At the first tensile test, it was noted fracture at a low load (61.8 kN). The fracture surface analysis of

Fig. 3. Overload camshaft fracture (left) and one of the fourth connecting rod bolt (right).

Fig. 4. Fatigue fracture of the fourth connecting rod bolt (left). High magnification in detail showing dimples formation (right) which covering the instable
fracture surface.
1546 S. Griza et al. / Engineering Failure Analysis 16 (2009) 1542–1548

Fig. 5. Fatigue fracture initiation point at the groove top.

the bolt indicated that it has already experienced fatigue crack propagation (Fig. 6). The fatigue did not occur as a result
of tightening relaxation, since 100 N m was verified prior its disassembly. Tensile tests of the three remaining bolts
resulted in 1213 (101 standard deviation) MPa maximal tensile stress. Hardness tests resulted in 39 (1 standard devia-
tion) HRC.

3.3. Microstructural analysis

A cross section metallographic sample adjacent to the fracture surface was taken from the fatigued bolt. It was observed
forming laps at the top of several grooves, including those where fatigue initiated (Fig. 7). The other bolts were examined and
the same features were observed. Bolts microstructure consisted of low temperature tempered martensite and other than
the forming laps no microstructural defect was observed. The bolts were classified as 12.9 class according microstructure
and mechanical tests.

Fig. 6. Fracture surface of a bolt that rupture at lower load in the tensile test. Prior fatigue crack propagation can be observed.
 S. Griza et al. / Engineering Failure Analysis 16 (2009) 1542–1548 1547

Fig. 7. Forming lap at the top of a groove where fatigue nucleated. Tempered martensite. Etchant: Nital 2%.

3.4. Numerical simulation and fracture mechanics analysis

Fig. 8 shows a correlation of normal bolt stress amplitude and tightening force of the three external load models. The high
stress field corresponds to interface between two connecting rod parts as expected due to combined normal and bending
stresses. It is possible to observe increasing stresses with tightening force as well as external load. At higher external loads,
nonlinearity between stress and tightening force is highlighted. The correlation among stress intensity factor, crack length,
as well as stress amplitude, is described by Eq. (2). In a generalization, it can be applied to explain the threshold stress inten-
sity factor (DKth), which is a fatigue endurance limit of the bolt assuming some crack length and stress amplitude.

4. Discussion

The preliminary analyses of the damaged engine indicated fatigue is one of the fourth connecting rod bolt. Bolts manu-
facture sequence includes head, threads and grooves forming from a commercial bar and heat treating, producing a class 12.9
bolts. Furthermore, grooves are grinding to adjust dimensional tolerance. The grooves are formed to allow the adjustment to
the connecting rod hole decreasing the chance of fretting or corrosion. The fracture occurred at the interface cap plane.
Fatigue nucleated at a restrict site of higher tensile stresses, at the side of crankshaft, and propagated up to 2/3 of the cross

500
450 15kN
Stress variation (MPa)

400 30kN
350 60kN
300
250
200
150
100
50
0
20 35 50 65 80 95 110
Tightening force (kN)
Fig. 8. Bolt stress amplitude correlated to tightening forces for three external loads.
1548 S. Griza et al. / Engineering Failure Analysis 16 (2009) 1542–1548

section, an indication of low stress level. At the nucleation site, over the groove top, forming laps were observed. This sort of
lap formation frequently occurs due to symmetrical metal flow during forming. Subsequently it was found fatigue in another
connecting rod bolt. Since its disassembly torque was around 100 N m, we are able to suppose that the root fatigued bolt did
not experience torque relaxation. Then we conclude that fatigue takes place even when a bolt was coupled in this particular
torque, since it has a critical defect placed in a higher tensile stress site.
In this paper, a finite element modeling was performed to evaluate the stress amplitude. Some important model simplifi-
cations include the oversight of stress concentration of the grooves, since the bolt shank modeled was flat, cylindrical, without
grooves. The choice for simulation instead of classical mechanical analysis, as reported in mechanical handbooks, was decided
due to the difficulty to select more realistic stiffness relation between the connecting rod and bolt. Finite element is suggested
for some papers as an alternative to avoid member stiffness nonlinearity found in the classical linear bolt theory [11,12]. Fur-
thermore, combined axial and bending stresses can be a concern in the connecting rod bolt, and it is easily analyzed by means
of finite element. As can be noted in our results, increasing tightening moment provide nonlinear decrease at the stress ampli-
tude experienced for the bolt shank, and increasing external load increase the stress amplitude and the nonlinearity.
Fracture mechanic approach also was evaluated. Once again, a simplification was adopted; the longitudinal forming lap
was considered a transverse semi-elliptical crack. Moreover, the fracture mechanics approach accomplished by finite ele-
ment allows insights concerning critical forming lap length and material threshold (DKth). The yield strength of bolt class
12.9 is 900 MPa in average, and the resistant area of this bolt at the fracture section is 113 mm2. From tensile test result
of the pre-fatigued bolt and the resistant section of the fatigued bolt, it was assumed a 30 kN external load. Furthermore,
a 45 kN tightening force was predicted applying the 100 N m assembly torque, Whitworth UNF7/16 20 threads per inch
resistant section and 0.2 torque coefficient in Eq. (1). Then, according to numerical simulation, for 30 kN external load
and 45 kN tightening force, the stress amplitude in the bolt shank should be around 100 MPa. Considering a 0.5 mm lap
pffiffiffiffiffi
length, for instance, the threshold found is DK th ¼ 2:65 MPa m, which is an accepted value for high strength bolts material.
However, a short defect, undetectable in conventional non destructive methods is sufficient to achieve fatigue propagation, if
material toughness has been lower.
We suggest recall of all elements of this manufacture lot. For avoid future failures, some decisions can be done. Mainte-
nance of excess metal in the groove top enough for the laps are excluded during grinding is not a reliable concern, since laps
control is difficult. The use of a higher toughness material can be impracticable due to increasing costs. Producing a flat bolt
shank at the cap interface region, maintaining a gap lengthwise between grooves can be an interesting design improvement
to reduce fatigue incidences. Increasing bolt assembly torque also can be an alternative approach, since this procedure
decrease the stress amplitude driving the stress intensity to a threshold [8]. According to Eq. (1), it is possible to apply a
200 N m torque still remaining below the axial yield strength.
Finally, as a critical aspect regarding the methodology adopted in this failure analysis case, we suggest some non destruc-
tive tests should be employed in the bolts before destructive ones. Indeed, if the second event of fatigue unfortunately was
not detected in tensile test, one could be argue the hypothesis of clamping force relaxation of individual bolt as the root
cause of engine failure.

5. Conclusions

The engine collapse as a result of forming laps at the groove tops of on connecting rod bolt. To avoid future failures, some
design improvements were suggested:

– design a flat bolt shank at the cap interface region to reduce stress concentration;
– increase of the assembly torque to reduce stress amplitude.

Acknowledgement

The authors would like to acknowledge the financial support of CNPq.

References

[1] Nishida S. Failure analysis in engineering applications. Great Britain: Butterworth-Heinemann; 1992.
[2] Milan MT, Spinelli D, Bose Filho WW, Montezuma MFV, Tita V. Failure analysis of a SAE 4340 steel locking bolt. Eng Fail Anal 2004;11:915–24.
[3] Baggerly RG. Hydrogen-assisted stress cracking of high-strength wheel bolts. Eng Fail Anal 1996;3(4):231–40.
[4] Yu Z, Xu X. Failure analysis of connecting bolts and location pins assembled on the plate of main-shaft used in a locomotive turbochanger. Eng Fail Anal
2008;15:471–9.
[5] Chen Hsing-Sung, Tseng Pi-Tang, Hwang Shun-Fa. Failure analysis of bolts on an end flange of a steam pipe. Eng Fail Anal 2006;13:656–68.
[6] Rabb R. Fatigue failure of a connecting rod. Eng Fail Anal 1996;3(1):13–28.
[7] American society for metals, metals handbook, 9th ed. Failure analyses and prevention, vol.11. Ohio: Metals Park; 1992.
[8] Bickford JH. An introduction to the design and behavior of bolted joints. New York: Marcel Dekker; 1995. 951.
[9] Shigley JE, Mischke CR. Mech Eng Des. 5th ed. Singapore: Mc Graw-hill; 1989.
[10] BS 7910:1999: Guide on methods for assessing the acceptability of flaws in metallic structures, British Standards Institution BSI. London, October 2000.
[11] Grosse IR, Mitchell LD. Nonlinear axial stiffness characteristics of axisymmetric bolted joints. J Mech Des Trans ASME 1990;112:442–9.
[12] Wileman J, Choudhury M, Green I. Computation of member stiffness in bolted connections. J Mech Des Trans ASME 1991;113:432–7.