Professional Documents
Culture Documents
A R T IC LE I N F O ABS TRA CT
Keywords: Connection rods are best known through their use in internal combustion piston engines. GM-
Engine failures GT26 locomotive uses a 3300 horse power V type engine with 16 cylinders that works at max-
Conrod failure imum crank speed of 900 RPM. The present study is concerned with the failure of a 645E3B
Experimental modal analysis engine connection rod with a reported catastrophic deformation due to unknown reasons. This
Stress analysis
rod has suffered from simultaneous bends in two surfaces with obvious marks of failure. A
Buckling
theoretical procedure, an experimental modal analysis technique and simulation by using finite
element engineering softwares are used for this study. It includes finding the natural frequencies
and the corresponding mode shapes for a new connection rod. This is then used to validate the
FEM simulated model. The critical loads and the buckling forces on the connection rod are then
calculated. The maximum loads are calculated by using the classical approaches and are verified
by simulation in ADAMS-Engine software. The dynamic modeling includes rigid and flexible
beam models and the torsional modes of vibrations are also considered. It is concluded that the
connection rod failure is due to buckling at the presence of hydrolock phenomenon. The outcome
of this research provides vital information for the proper operation and maintenance of heavy
duty Diesel engines.
1. Introduction
Connection rods are best known through their use in internal combustion piston engines, such as automotive engines. In re-
ciprocating engines, connection rod (Conrod) plays a vital role in conveying the power that is produced in the combustion chamber to
the crankshaft and consequently moving the vehicle. The new design of connection rods are of a distinctly different design from the
earlier ones that were used in steam engines and steam locomotives.
The role of Conrods in heavy duty Diesel engines of locomotives starts from turning the reciprocating motion of pistons into the
rotation of the crankshaft. In Diesel-electric locomotives and through this mechanism, the produced kinetic energy rotates the
coupled generator rotor and yields electrical power to be used in traction motors. Consequently, the wheels and axles rotate by the
coupling between the pinion and the gear.
GM-GT26 locomotive uses a 3300 horse power V type engine with 16 cylinders that works at maximum crank speed of 900 RPM.
The design of this engine goes back to 1967. The 645E3B engine uses two types of Conrods that have geometrical differences in their
designs. In both of these designs the small end is attached to the piston pin, gudgeon pin or wrist pin, which is constrained with a half
circular head. But the larger ends of the two types of Conrods are different. In this engine for minimizing the shear stresses on the
crankpin journal, the blade Conrod is at the center of the crankpin journal. But, the fork Conrod surrounds the blade Conrod and both
⁎
Corresponding author.
E-mail addresses: rezvani_ma@iust.ac.ir (M.A. Rezvani), Davood_javanmardi@rail.iust.ac.ir (D. Javanmardi), pedram_mostaghim@rail.iust.ac.ir (P. Mostaghim).
https://doi.org/10.1016/j.engfailanal.2018.05.005
Received 8 March 2017; Received in revised form 1 January 2018; Accepted 9 May 2018
1350-6307/ © 2018 Elsevier Ltd. All rights reserved.
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
of them are clamped with baskets to the crankpin journal of crankshaft. Fig. 1 shows the installation mechanisms of these two types of
Conrods and, Table 1 presents the design parameters of a blade Conrod.
Connection rods are subjected to inertial forces that originate from the reciprocating mass and gas pressure forces that are due to
the maximum ignition pressure. These forces result in the axial and bending stresses. Bending stresses in Conrods originate from the
eccentricities of the crankpin journal within the main shaft of the crankshaft, and the rotational mass forces. Therefore, a connection
rod must be capable of transmitting axial tension, axial compression, and bending stresses. Failures of connection rods are often
caused by the bending loads that act perpendicular to the axes of the two bearings.
In 1996, Lu presented an approach for shaping a Conrod that is subjected to a load cycle consisting of the ignition force, inertia
load, and the displacement. He calculated the stresses by a Finite Element Approach (FEA), and finally determined the fatigue life
based on the fracture mechanism [2]. Phad and Burande analyzed a Conrod for extracting its static and dynamic properties. The
results of this analysis showed that the tension stresses in the Conrod are higher than the compression stresses. The stresses decline
from the pin end to the crank end. The maximum stresses occur on the fillets of the small and the large ends of the crankshaft [3].
Desai et al. carried out a numerical and experimental analysis to find the critical spots in the Conrod structure. In this study, the
stresses on different areas of the Conrod were measured in the laboratory by applying different loads. On the other hand, the under
investigation analysis was simulated for finding the tension and compression stresses on different parts of the Conrod. The results of
this study showed that the maximum stress happens at the larger end of the Conrod [4]. Shenoy conducted a dynamic analysis of the
loads and stresses in the connection rod component. They explained that the tensile load applied over 180° of the crank contact
surface with cosine distribution, whereas the compressive load applied as a uniformly distributed load over 120° of the crank contact
surface [5].
However, not many studies were carried out on the Conrod buckling. Shenoy and Fatemi performed an optimization including the
simple linear buckling analysis [6]. Recently, Lee et al. studied the buckling sensitivity of a Conrod to present a buckling evaluation
procedure via FEA. The study predicted the critical buckling stress through a classic analysis and FEA. This was followed by some
experimental procedures with a Conrod test rig for validation purposes. The results of this investigation demonstrated that when a
weight reduction of Conrod shank is attempted, buckling should be considered as an essential factor along with the other criteria such
Fig. 1. Installation schematic of the blade Conrod on 645E3B Diesel engine [1].
51
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
Table 1
Design parameters of the blade Conrod [1].
Design Parameters Initial Value (mm)
Frontal Profile
Effective length (L) 584.20
Diameter of small bore (D1) 93.35
Diameter of large bore (D2) 195.40
Shank Section
Fillet radius 1 (R1) 15.00
Fillet radius 2 (R2) 4.50
Total width (L1) 75.00
Total width (L2) 8.00
Total thickness (T1) 55.00
Middle thickness (T2) 10.00
In order to find out the root causes for the bending of the Conrod, four logical steps are considered. The methodology for this
problem solving is presented in Fig. 3. Initially, a modal analysis is performed on the subject by using ABAQUS engineering software.
52
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
Fig. 3. Process flowchart and the method for the Conrod failure analysis.
This assists in finding the natural frequencies and the mode shapes of the Conrod. It is also possible to compare such modes to the
failure modes of the Conrod. An experimental modal analysis is also performed on a new Conrod that facilitates finding its natural
frequencies and mode shapes. This is then used to validate the simulated model based on the results from ABAQUS. Modal analysis
provides an understanding of the Conrod circumstances in the engine hence facilitates the path for the diagnosis of the root causes for
its' failure.
Through the next steps, the critical loads and the bucking forces on the Conrod are calculated. This is performed by calculating the
loads and the stresses that appear in the Conrod in its' buckling mode.
Also, FEM analysis is carried out to find the Von misses stresses on the Conrod. This simulates the result of a static load on the
blade Conrod in the elastic range. The mechanical properties for the connection rod are presented in Table 2.
3. Modal testing
In order to measure the dynamic properties for the Conrod under investigation, including its natural frequencies and mode shapes,
an experimental modal analysis setup is prepared. The natural frequencies and the corresponding vibration modes for this object were
also worked out by using an FEM analysis [10]. In order to identify the connection rod structural characteristics, the test boundary
conditions need to have the least influence on the test results. Assuming that the stresses at the small end of the connecting rod and
the shaft section of the body are already determined, one remedy is to keep them at some certain levels during the optimization
process and apply them to the structural optimization design module as boundary conditions [11]. Fig. 4 presents the schematic of
the modal testing procedure. In the experiment, the connection rod is suspended by a soft rubber rope in order to provide the free
boundary conditions, Fig. 5.
The excitation force is applied by using an electronic hammer. This method of excitation is reliable for lower to moderately
damped structures.
The system transfer function can be obtained by measuring the input excitation force and the response acceleration signals. In
order to identify the excitation and the response points, a grid plot is marked on the test specimen. An FEM model of the Conrod is
also drawn by using ABAQUS engineering software. The same grid lines and nodes are also marked on this FEM model, Fig. 5(a) and
(b). Taking the connection rod working condition into account, the selected frequency domain for the modal analysis is in the order of
0–6000 Hz. After exciting the Conrod with the hammer and acquiring the input and the output signals, data post processing is
Table 2
The material and the mechanical properties of 645E3B Diesel engine Conrod that is
made from 42CrMo4 [1].
Parameter Value Unit
A 1766.495 mm2
E 202 GPa
Iyy 1.284 × 106 mm4
K 0.7 –
L 584.200 mm
σy 415 MPa
σu 655 MPa
ω 7.850 gr/cm3
υ 0.290 –
53
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
Fig. 5. a) An FEM model of the blade connection rod including the grids and nodes. b) The blade connection rod hung from the roof and under the
test condition.
54
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
performed. The corresponding Frequency Response Functions (FRFs) in the logarithmic and linear scales are provided in Figs. 6 and
7, respectively. Fig. 6 presents the response of the accelerometer number 2, under the hammer excitation in the logarithmic scale.
Fig. 7 that is in the linear scale clearly offers the peak points in FRF that highlight the system natural frequencies. While going
through the resonances in the FRF the phase shifts of 180o are evident.
A comparison between the natural frequencies of the Conrod that are obtained from the FEM software and the test results are
presented in Fig. 8.
The proximity of the two sets of the results indicates the fact that the simulated model is reliable and it can be used for the Conrod
stress analysis. It is determined that the average differences between the two sets of the results are less than 6%.
The maximum loads on the engine, are calculated by using a theoretical approach. Authentication of the results is by performing a
simulation procedure in ADAMS-Engine software.
In order to calculate the loads on the Conrod by the theoretical approach some symptoms are also considered, including:
- The critical load on the Conrod happens due to the maximum pressure.
- The inertia forces due to the cylinder and Conrod weight have negative effects along with the ignition load.
- Due to the high volume of the lubricating oil between the piston rings and its liner, the friction force is neglected.
In the classical approach, some main parameters of the engine and its' Conrod that are given in Table 3 are used to find the
maximum pressure load at the ignition time in the cylinder chamber.
The total force on the small end of the Conrod is equal to the difference of the firing force and the inertia force as in Eqs. (1)–(3)
[14].
Ftotal = Ffiring − Finertia (1)
πD 2
Ffiring = Pmax
4 (2)
Multi-body engineering softwares such as ADAMS provide extended possibilities for the dynamic analysis of engineering subjects.
ADAMS is a multi-body software that facilitates simulation of motion and study of the dynamic behavior of three dimensional models
[15]. For the dynamic modeling of 645E3B engine in ADAMS, three methods including the rigid crankshaft (the rigid model), the
flexible beam model and the torsional model are used. For the beam model of the crankshaft, consideration of the torsional and
deflection stiffness adds to the accuracy of the model and pushes the results closer to the reality. Also, in this method calculation of
the normal stresses in webs is possible. Based on the studies, there is a difference of about 5% between the simulation results from the
flexible beam elements and the real case [16]. Therefore, in this research the flexible beam elements' method for the crankshaft is
used.
In this method the Timoshenko beam theory is used. In such a way that initially the crankshaft is meshed with beam elements and
then the whole crankshaft as a number of joined beams is loaded and analyzed. It is an indeterminate beam that will also be analyzed
55
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
Mode shape 5
Modal Test (Hz) FEM Analysis (Hz) Deviation (%)
5068 5163 1.22
Fig. 8. The natural modes of the Conrod, a comparison between the experimental and the FEM modeling results.
for lateral vibrations. The benefit for using the Timoshenko beam theory is that in this theory the effects of the shear strains are
considered and the assumption for the sustained perpendicularity of the surfaces normal to the neutral axis in the Euler-Bernoulli
beam theory is not needed. In fact in a Timoshenko beam, the shear deformations appear as the angular deviations of these surfaces
[17].
For the dynamic simulation of the 16 cylinders engine under investigation, all particulates corresponding to the masses and
dimensions of the components such as the engine block, crankshaft, flywheel, cylinders, pistons and bearings are considered. Also,
general specifics of the engine such as the number of cylinders, the combustion sequences, the piston diameter and stroke, the type of
the lubricating oil, the gas pressure in a combustion cycle, etc. are included. The simulated model for the crankshaft based on the
flexible beam method is presented in Fig. 9.
By simulating the crankshaft and its' components in ADAMS-Engine software, the force versus crank angle on the small end of the
Conrod is presented in Fig. 10.
56
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
Table 3
Main parameters of 645E3B engine and the calculated loads [13].
Parameter Value Unit
Fig. 9. The simulated model for the EMD645 Diesel engine crankshaft.
Fig. 10. Conrod pin bearing force and combustion pressure Vs.Crank Angle for the 645E3B Engine during 2 cycles (output from ADAMS).
From the results it becomes clear that 7° after firing in each cylinder, at the top dead center point of the piston, the maximum peak
pressure applies to the piston. This means that the entire applied load on the piston is applied directly to the piston pin bearing and
the Conrod small end.
As depicted, the maximum value of the applied force to the blade Conrod small end is equal to 381.5 kN. This is less than 0.8%
different from the theoretical total force (Ftotal). This can be considered as a credit to the accuracy of the simulated model.
Merchant-Rankine is a precise formula for predicting the critical buckling load of structures [18]. Unfortunately, this formula
cannot be applied to the Conrod because of its difficulty in defining the proper conditions for the two ends of the rod. The same issue
also exists for defining the geometry changes in the shank. Therefore, a special technique is used for predicting the critical load of the
57
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
Conrod by using the FEM method. Based on this method, the critical force of buckling in the elastic ranges is equal to the multi-
plication of the calculated eigenvalues and the maximum compressive load [19]. The calculated first and the second modes of Conrod
buckling with eigenvalues of 13.102 and 20 are presented in Fig. 11.
The buckling behavior of a column or a thick rectangular plate that has a constant cross section follows Euler's equation based on
the theories of solid mechanics and elasticity [18]. The critical buckling load and the critical stress in the elastic range are expressed
as in Eqs. (4) & (5), respectively.
π 2EIyy
Pcre =
(KL)2 (4)
π 2E
σcre =
(KL/ ry )2 (5)
Euler's equation considers only the elastic stiffness and the moment of inertia cannot be used. In order to overcome this re-
striction, Merchant-Rankine formula includes the plastic part to Euler's buckling equation, and the critical elastoplastic buckling load
is given as in Eq. (6) [18]. For the evaluation purposes the buckling possibility BSF is defined as in Eq. (7).
−1 −1 −1
2
1 1 ⎡ π EIyy ⎞ ⎤
Pcr = ⎡ e + p ⎤ = ⎢ ⎜⎛ ⎟ + (Aσy )−1⎥
⎢ Pcr P ⎥ (KL)2 ⎠
⎣ cr ⎦ ⎣⎝ ⎦ (6)
Pcr
BSF =
Ftotal (7)
By calculating the loads on Conrod and the critical buckling loads and stresses in both elastic and plastic ranges, the calculated
results are provided in Table 4. According to these results, the deviation of the predicted critical load and the stress in between the
classical approach and FEM is not more than 8.5% and 4.3%, respectively. This confirms the validity of the results from the cal-
culations. The last parameter to be calculated is the buckling safety factor (BSF), which in this case, on the average is equal to 1.74.
A Finite Element analysis (FEM) is carried out to determine the Von Mises stresses and the maximum stresses on the loaded
Conrod. The Conrod material properties are the same as in Table 2. To define the constraints on the Conrod, the crank end is fixed in
all directions and the pin is allowed only for translational motion in the loading direction. The analysis generated 100,684 quadratic
tetrahedral elements of type C3D10. The Von Mises stresses on the Conrod due to the applied maximum compressive load of 384.4 kN
at 900 rpm are presented in Fig. 12. The results define the connection rod shank as its weakest part.
Based on the results from the modal analysis and the simulation results from ABAQUS engineering software, the second vibration
mode is similar to the mode at failure in the Conrod that appeared at the frequency of 1131 Hz. By considering the maximum
crankshaft rotation speed of 900 rpm (15HZ), resonance as the cause of the failure in Conrod is ruled out. The engine protective
systems limit the crankshaft rotational speed to a maximum of 910 rpm beyond it the engine will shut down. Therefore, it is con-
cluded that the design of the Conrod from the viewpoint of its' vibration behavior is safe. Consequently, one needs to search for other
causes of failure in Conrod.
On the other hand, the first and the second modes of buckling are exactly similar to the modes of the failure in Conrod. Buckling
Fig. 11. The first and the second modes of Conrod buckling.
58
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
Table 4
The results by the classical and FEM approaches.
Parameter Value Unit
Classical FEM
Fig. 12. The Von Mises stresses on Conrod based on the FEM analysis.
becomes a possible suspect for Conrod failure. This mode of failure is known as HYDROLOCK or hydrostatic lock that occurs in the
internal combustion engines. This can happen when a volume of liquid greater than the available volume of the cylinder enters the
cylinder. This can happen at the end of the piston stroke when the remaining cylinder volume is at its minimum. Since liquids are
almost incompressible, under hydrostatic lock, the piston cannot complete its travel; therefor the engine must stop rotating or a
mechanical failure must occur.
Due to the Diesel engines' high compression ratio for self-ignition, probability of hydrolock phenomenon is then much higher than
the gasoline engines. This leads Diesel engines combustion chamber to need less liquid to apply the high value of forces on the Conrod
that cause hydrolock. The high value of pressure at the combustion chamber is the result of ignition or compressing incompressible
liquids like water. It then requires strong components with higher durability and a higher buckling safety factor.
Detection of buckling in the Conrod requires special measuring and testing methods, and it cannot be easily detected. A buckled
Conrod will eventually bring up some signs of the symptom. Wear appeared on the crankpin bearing and crank end of a failed Conrod
are presented in Fig. 13. This figure illustrates the fact that at the presence of the buckled Conrod, the piston was tilted and
Fig. 13. Wear on the crank end of the Conrod due to hydrolock phenomenon.
59
M.A. Rezvani et al. Engineering Failure Analysis 92 (2018) 50–60
consequently wears on the crankpin bearing and crank end of the Conrod occurred.
Abnormal pressure on pistons can damage bearings and even crankshaft in addition to the Conrod. There are no offers available
for practical approaches to detect this type of failures, yet. With some luck, the produced abnormal pressure in combustion chamber
can cause seizure on liner wall and also piston body which cause blow-by. Consequently, under such circumstances the engine will
shut-down as a result of the increasing crankcase pressure.
8. Conclusions
This research involved the study of a bent connection rod of a heavy duty Diesel engine. The analyses included experimental
modal analysis and simulation by using FEM approaches. The search for the natural frequencies and the corresponding mode shapes
through modal analysis proved that the connection rod failure does not relate to its dynamic properties. The fiasco is not due to the
excitation of the rod vibration resonances.
It was then necessary to check the status of the rod under engine loads and to study its' mechanical stresses distribution. The
search for the causes of the failure then continued by using FEM approaches and the evaluation of the buckling safety factor.
Connection rod transmits combustion energy from piston top to the crankshaft. Engine loads on connection rod are calculated
theoretically and then verified by using ADAMS engineering software for dynamic simulation. The results of the loading are then used
for simulation by using an FEM software. The buckling safety factor for the Conrod and its buckling modes examined. The results
confirm that the connection rod failure is due to low BSF. The causes of this failure are obviously hydrolock phenomenon that is due
to water leaking into the combustion chamber and the presence of high compression ratio. The outcome of this research provides vital
information for the proper operation and maintenance of heavy duty Diesel engines. Through proper monitoring of the engine and its
performance it will be feasible to block water from leaking into the combustion chamber of the engine and to prevent destructive
hydrolock to happen.
References
60