Académique Documents
Professionnel Documents
Culture Documents
A SEMINAR REPORT ON
Shri. J Y KUDRIYAVAR
Submitted by
RAJAT KAWAL
________________________________________________
CERTIFICATE
Failures of the connecting rod oil hole is bonafied work carried out by
RAJAT KAWAL in partial fulfillment for the award of degree of Master of
Technology in Mechanical Engineering of the Visveswaraya Technological
University, Belgaum, during the year 2009-10. The seminar report has been
approved as it satisfies the academic requirements.
Abstract
This paper describes an application of ANSYS Probabilistic Design System
(PDS) as a superior method to efficiently identify the relative influences of random
input variables on fatigue stress and to optimize the chosen random input variable to
achieve the desired reliability.
The axial oil-hole of a connecting rod in a reciprocating air conditioning
compressor acts as a stress raiser and can lead to early fatigue failures. A parametric
3-D finite element model of the connecting rod with a hollow wrist pin was developed
to study the fatigue stress. Contact elements were placed between the rod and the pin.
The service stresses were simulated by pushing the pin against the connecting rod. A
high predicted fatigue stress at an observed failure origin verified the model.
A probabilistic analysis was carried out with three independent random
input variables: the wrist pin bearing ID, the wrist pin OD and the hollow wrist pin
bore ID. A macro file was created by a computer program to relocate selected nodes,
by which the dimensions defined by the random input variables were automatically
adjusted during each analysis loop.
The maximum fatigue stress at the failure location was the random output
parameter. This random output parameter was explicitly related to the random input
variables using the Response Surface Method. The sensitivity analysis revealed that
the wrist pin hollow ID had the most effect on the fatigue stress.
Additional PDS runs were performed to optimize the pin ID and tolerance
based on the estimated probabilities of failure using the Monte Carlo simulated
cumulative distributions of the maximum fatigue stress. The analysis showed that the
fatigue failures could be eliminated by reducing the wrist pin hollow ID, which was
also the most cost effective fix.
CONTENTS
1. Introduction……………………………………... 05
2. Procedure………………………………………... 06
3. Analysis………………………………………….. 07
5. Conclusion………………………………………. 15
6. References………………………………………. 16
1. Introduction
Connecting rod failures at the edge of the wrist pin bearing oil-hole early in
the life of the part initiated this FEA investigation. The purpose was to develop a
design variation that would remove this failure mode. Besides the application of
ANSYS PDS described in this paper, other design options that were also examined
included:
These design options had a very limited effect and were thus abandoned. The
analysis then focused on the wrist pin bearing ID and the hollow wrist pin OD and ID.
Modifications of these variables are more easily accommodated by Manufacturing
and they have less of an effect on cost. ANSYS PDS provides a very efficient way to
study these design variables simultaneously. More importantly, the traditional
deterministic analysis method tends to over-design. This can be more costly and could
have a negative effect on compressor performance due to parts that are heavier than
they need to be. Therefore, the ANSYS PDS was used to obtain an optimized design.
The wrist pin bearing ID, hollow wrist pin OD and ID were chosen as
random input variables to the ANSYS PDS. Examination of the finite element
generated stress distribution revealed that the maximum fatigue stress at the failure
location caused the oil-hole fatigue failure. To resolve the failure mode, therefore, the
maximum fatigue stress at the failure location was chosen as the random output
parameter.
Historic product reliability data indicated that connecting rods without an oil-
hole exhibit sufficient fatigue life. Therefore, a finite element model of the connecting
rod and wrist pin without the oil-hole was also analyzed; and the results were used as
the baseline for the allowable fatigue stress.
2. Procedure
The FEA model included the hollow wrist pin and the connecting rod. The
3-D solid model geometry of the connecting rod was extracted from Pro/E software.
All unimportant geometric details were deleted to save computation time. The hollow
wrist pin was modeled using the ANSYS preprocessor. The connecting rod and wrist
pin model was reduced to a quarter symmetric FEA model as shown in Figures 1 and
3.
Contact elements between the wrist pin bearing ID and the wrist pin OD
were used to simulate the hydrodynamic bearing oil film. A uniform pressure was
applied to pin to simulate the 1000 lb axial load transferred from the piston.
Inspection of failed connecting rods indicated the failures at the oil-hole were not
related to the high bearing friction due to bearing failures. Therefore, no
circumferential bearing friction force was included in the model since the friction of a
properly functioning bearing is generally low. Since the inertial loads were small, they
were neglected as well. The crank pin bearing end of the connecting rod was also
excluded from the FEA model. Instead, the connecting rod was supported against the
axial push load at the rod end of crank pin bearing.
For probabilistic analysis, the FEA model has to be set up parametrically so
that it can change its dimensions in response to the three random input variables for
PDS calculation loops. The solid model was meshed with a sufficiently fine mesh at
the failure location. All nodes located on the surfaces of the wrist pin bearing ID,
hollow wrist pin OD and ID were selected and saved to an ANSYS text file. A
computer program was developed to read in the text file and create an ANSYS macro
file that relocated the selected nodes according to the three random input variables.
Note that special attention was given to the elements associated with the selected
nodes so that they remained good shapes after the relocations of the selected nodes.
All random input variables were declared as independent random variables.
There were no correlations defined between the random input variables. The
variations of random input variables were characterized by uniform distributions. The
descriptions of random input variables and the random output parameter are
summarized in Table 1. As an example, the cumulative distribution function of the
hollow wrist pin ID is illustrated in Figure 5. Another macro file was written to
retrieve the random output parameter from the ANSYS result file for each PDS loop.
3. Analysis
(a) Original Pro/E rod model (b) Simplified rod solid model
(c) Quarterly symmetric FEA rod / pin model with an oil hole
(d ) The finite element meshes near the oil hole. The volume at the bearing contact surface
was selected to show FEA results in Fig 4
Figure 1. Rod/pin finite element model with an axial oil hole
(c) 2nd principal stress (S2) (d) 3rd principal stress (S3)
Figure 2. Stress distributions for the model with an axial oil hole (One half of the wrist pin
bearing surface contacting the wrist pin is showing)
(c) Von Mises stress (Seqv) (d) 1st principal stress (S1)
Figure 4. Stress distributions for the model without an axial oil hole (One eighth of the
The PDS results were used to fit the response surface that was an
approximation describing a random output parameter as an explicit function of
random input variables. Monte Carlo Simulations were conducted based on the
response surface to obtain probabilistic results for statistical post-processing.
The initial PSD run was performed with a reduced hollow wrist pin ID for
the connecting rod and wrist pin model with the oil-hole. The cumulative distribution
function of the initial PDS run is shown in Figure 7. The rod/pin model without an
oil-hole was also analyzed as the baseline. The cumulative distribution function of the
baseline is shown in Figure 8. Comparing Figures 7 and 8, it can be seen that the
maximum fatigue stresses of the initial PDS run became even lower than the ones of
the baseline due to the reduction of pin ID.
Figure 7. The results of the initial PDS run. The histogram and cumulative distribution
function of random output parameter (maximum fatigue stress at the failure location) for
the model with an oil hole and a reduced pin ID.
Additional PDS runs were performed to reach the objective that the rod with
an oil-hole would have approximately the same reliability as the rod without an oil-
hole by reducing the wrist pin ID. This objective was realized based on the
cumulative distribution functions. First, the allowable fatigue stress was read from the
cumulative distribution function of the baseline in Figure 8 at 99.9% probability,
corresponding to a probability of failure of 0.1%. After the determination of the
allowable fatigue stress, additional PDS runs were performed with different ID values
of the wrist pin. The wrist pin ID dimensions were finally determined when the
maximum fatigue stress read from PDS generated cumulative distribution function at
0.1% probability of failure was about the same as the allowable fatigue stress. The
cumulative distribution function for the final design is given in Figure 9.
Figure 8. The results of the baseline PDS run. The histogram and cumulative distribution
function of random output parameter (maximum fatigue stress at the failure location) for the
model without an oil hole.
Figure 9. The results of the final design PDS run. The histogram and cumulative distribution
function of random output parameter (maximum fatigue stress at the failure location) for the
model with an oil hole and a reduced pin ID.
Bench strain gage tests were conducted to confirm the FEA analysis. Strain
gages were laid inside the oil hole and close to the edge where the failure initiated.
The strain gage tests confirmed the major FEA result: a stiffer pin led to a lower
fatigue stress at the failure location.
5. Conclusion
successfully duplicated the failure mode by predicting a high fatigue stress at the
observed failure location.
The sensitivity analysis revealed that the hollow wrist pin ID had the
greatest influence on the maximum fatigue stress. Reducing the wrist pin ID was the
most effective and economic method to remove the failure mode.
According to the PDS generated cumulative distributions of the maximum
fatigue stresses at the failure location, with an 11% reduction of the wrist pin ID, the
connecting rod with an oil hole can achieve approximately the same reliability as the
connecting rod without an oil-hole.
6. References
[2] Kenneth J. Rasche, “Probabilistic Study of a Refrigeration Steel Heat Loop Tube
Joint”, ANSYS 2002 Conference.
[3] Andreas Vlahinos, Subhash Kelkar, Stefan Reh, Robert SeCaur, & Steve Pilz,
“Reliability Based Optimization within the CAD Environment”, ANSYS 2002
Conference.
[4] Adrian Rispler & John Raju, “Optimization of an Aircraft Control Surface”,
ANSYS 2002 Conference.