Documentos de Académico
Documentos de Profesional
Documentos de Cultura
By,
Thomas C. Grobaski, Dr. Bhavin Mehta*, Dr. Jay Gunasekera
Department of Mechanical Engineering
Ohio University, Athens, Ohio 45701
ABSTRACT
The goal of this research was to provide a preliminary step into developing a
complete forging die life model. The research involved analyzing the initial effects
of (1) friction, (2) work-piece temperature, (3) die temperature, and (4) forging
press stroke speed on effective die stresses, die surface temperatures, die/work-
piece sliding velocities, die/work-piece contact pressures, and die surface
temperatures were examined. To obtain the results the forging process was
modeled (SolidEdge 3D Solid Modeling Software), simulated (MSC.Superforge
Software), and statistically setup and examined using two-level full factorial
design of experiments (Analyzed with Minitab & MS. Excel). The product
reviewed was a 10inch diameter differential ring gear forged at the American
Axle Manufacturing, North Tonawanda, New York forging plant. The ring gear is
used in the rear differentials for Ford and GM trucks.
1.1 Forging Data and Outputs
The multiple metal failure prediction models were analyzed to identify the
variables required to estimate metal failure. The parameters required to estimate
die failure must be qualified and quantified to produce a valuable working
prediction model. By breaking each failure model down and regrouping the
variables can be organized into three main categories.
1. Material/Process Data
2. Process Outputs
3. Empirical Constants
Material data is comprised of independent variables based on the different
tooling parts including the dies, work-piece, lubricant, press and environment. For
each of the separate tooling features there are the mechanical, chemical,
electrical, and thermal properties. Material data are the input parameters into a
forging operation.
The various material properties were obtained from the abundant sources
of previously tested materials included in the ASTM Metals and Materials
Handbooks, matweb.com, independent research laboratory testing (ORNL),
Metal Processing Journal, etc. These sources, and many others, provide
assumed accurate data for various materials that are forged, also the effects of
temperature, humidity, and other environmental factors that lead to altered
physical properties. For the failure models, the required material data is
displayed in Figure 1.1.
Table 1.1: Material/Process Data and Process Outputs
Material & Process Data Forging Process Outputs
1. Die & Work-piece: 1. Material Sliding:
•Hardness •Distances
•Initial Temperature •Velocities
•Ductility 2. Heat Transferred:
•Ultimate Tensile Strength •Conduction (WP to Die)
•Modulus of Elasticity •Convection (WP & Die to Env.)
•Tempering Curves •Radiation (WP to Die & Env.)
•Yield Strength 3. Final Temperature of:
•Poisson’s Ratio •Die
•Tempering Curves •Work-piece
•Surface Roughness •Environment
•Toughness 4. Forging Load Force
•Geometry 5. Contact Pressures at:
•Thermo-physical Properties •Die/Work-piece Interfaces
2. Surface Coatings: 6. Effective Stress:
•Physical Properties •Ranges
3. Environment: •Amplitude
•Ambient Temperature 7. Total Strain Range From:
•Environmental Humidity •Elastic Strain
•Environmental Oxygen Content •Plastic Strain
•Preheat of Dies & Work-piece •Thermal Strain
4. Lubricant: 8. Strain Rate
•Thermo-physical Properties 9. Strain Hardening of:
•Friction Factor •Die Surfaces
•Coverage and Thickness •Work-piece
5. Forging Press
•Press Type
•Stroke Velocity Curve
•Stroke Acceleration Curve
•Press Rigidity
•Press Stroke Length
•Press Repeatability & Accuracy
6. Miscellaneous
•Operator Quality
•Cooling Time: WP & Die
•Contact Time: Die, WP & Lube
Process outputs are the results from forging process (stress, strain, final
surface temperature, etc.). They are dependent upon the material/process data
(load force, initial temperature, etc.), also known as independent variables [Table
1.1].
Once the effects of the independent variables upon the process outputs
are known, the number of cycles till failure of the dies, can be estimated using
current metal failure theories (Wohler or Basquin’s Stress Life Approach, Coffin &
Manson’s Strain Life Approach, Archard’s or Bayer’s Wear Model, Fracture
Mechanics, etc.).
The die failure models rely heavily upon the accuracy of empirically
determined constants. The constants and coefficients in Figure 1.2, are from the
failure models previously described.
After obtaining the material/process data and the process outputs and
empirically modeled constants are determined, the cycles till die failure can be
estimated. Using a multi-factor, two level full factorial design, the effects of the
independent variables on the die cycles till failure are estimated. From this
setup, a formula equivalent to the Taylor’s Tool Life [Appendix F] will be
constructed. This equation will be employed to quickly, efficiently, and accurately
determine the forging die life longevity.
A) Friction (NA)
WP Temp. (C)
D) Avg. Stroke
Yates Code
Friction (A)
(ABCD)
Run #
Run #
0 8.4 1700 300 0.2 0 -1 -1 -1 -1 1
1 8.4 1700 300 0.7 1 -1 -1 -1 1 A
2 8.4 1700 800 0.2 2 -1 -1 1 -1 B
3 8.4 1700 800 0.7 3 -1 -1 1 1 AB
4 8.4 2300 300 0.2 4 -1 1 -1 -1 C
5 8.4 2300 300 0.7 5 -1 1 -1 1 AC
6 8.4 2300 800 0.2 6 -1 1 1 -1 BC
7 8.4 2300 800 0.7 7 -1 1 1 1 ABC
8 11 1700 300 0.2 8 1 -1 -1 -1 D
9 11 1700 300 0.7 9 1 -1 -1 1 AD
10 11 1700 800 0.2 10 1 -1 1 -1 BD
11 11 1700 800 0.7 11 1 -1 1 1 ABD
12 11 2300 300 0.2 12 1 1 -1 -1 CD
13 11 2300 300 0.7 13 1 1 -1 1 ACD
14 11 2300 800 0.2 14 1 1 1 -1 BCD
15 11 2300 800 0.7 15 1 1 1 1 ABCD
The upper die and lower die were aligned with the work-piece [Figure 1.6].
Figure 1.6: Aligned Forging Process Setup:
By shortening the upper dies stroke length from the actual 14 inches to
3.58 inches simulation time is reduced. To do this the Positioner function in
Superforge moved the upper die down until it was touching the top of the work-
piece which was resting a top the bottom die as shown in Figure 1.6. The
properties and bounds of the forging process are added to insure the models
accuracy:
2.0 Results
The simulation, of each stage of the three stages required to forge the
AAM ring gear, were simulated in MSC.Superforge. The process outputs are
then analyzed using Microsoft Excel and Minitab Release 14. The results of the
main, secondary, tertiary, and fourth order effects on the process outputs were
calculated. To save space and time, the effects on the upper die of Blocker Stage
(Stage B) of the ring gear forging process is explained in detail, and the
remaining two stages’ results are summarized in the Appendix.
2.1 MSC.Superforge Buster Stage Outputs
The results of the MSC.Superforge simulations for the Buster Stage are in
Table 2.1.
Yates
Figure 2.3: 2D Simulation Maximum Final Temp. of Upper Die Surface {°F}
Figure 2.4: 2D Simulation of Net Energy Supplied by Upper Die {J}
Figures 2.2 through 2.7 show the output results obtained from
MSC.Superforge. Each image shows a colored contour image of the 2D blocker
stage of the work-piece, upper and lower dies.
The values for each output process were entered into Minitab Statistical
Analysis Software. Since the simulations were run using computer software,
there are a few assumptions that are made:
1. The run order of the simulations had no bearing on the out come of the
tests
2. The simulations are repeatable, with accurate repeatable results
3. No outside or environmental factors could affect the simulations
To display the results of the design of experiment and the simulations, the
data was formulated into “Normalized Plot of the Effects Charts”. These charts
display the data with the non-significant effects following a normal distribution.
Each chart is fitted with a linear trend-line and significant effects will vary from the
normal distribution and have a high residual from the trend line. The further the
point from the line the larger the effects on the response variable. These charts
are graphed with a Significance Level (α=0.25), which states any values that has
a t-value greater than the 0.25 significance level will be marked as a significant
effect.
“Yatesing” was employed for analyzing the data, and normalizing the plots
[4] the Yates code for each run is shown in Table 2.1.
The Yates difference between specified runs (d-1=d, df-f=d, etc.) was
performed to find the specific effect and then averaged. In Table 2.2, the yatesing
for the estimated main effect of stroke speed is shown.
The averaged estimated effects of stroke speed on each of the various
process outputs were graphed with results of the other main effects, secondary,
tertiary, and fourth order effects. The graph was setup with the effects as the
independent variable, and the percent value as the dependent variable.
Performed to determine which would produce the largest effect on the individual
process outputs. The normal probability plots are shown in Figure 2.8 through
Figure 2.15, and span the range of effects of the process outputs found in Table
5.3. These charts were obtained from Minitab, but the calculations were checked
using Microsoft Excel calculations.
D S troke S peed
60
50
40
30
20
10 A
5 BC
1
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Effect
Lenth's PSE = 0.084375
Figure 2.8: Normal Probability Plot of the Effects for Sliding Velocity {in /s}
3.0 Conclusion
This research is a preliminary step in developing a universal metal forging
die life equation. The point of researching (1) friction, (2) initial die temperature,
(3)work-piece temperature, and (4)forging press stroke speed effects on (1)die
stress, (2)net energy, (3)sliding velocity, (4)contact pressure, (5)final surface
temperature, and (6)load force is to ultimately predict the most likely mode of die
failure. Once the most probable mode of die failure is calculated, a forecast of the
forging press cycles until die failure can be determined.
The results of this research show that the forging process die outputs
responses are affected significantly by only a small number of main effects, or
interaction of those effects. The significance level for this research was set high
(α=.25) showing that any effect that had an effect greater than 25%, above or
below the zero effect was deemed significant. For instance, analysis of the load
force [Figure 2.12] revealed that the only significant main effect was the change
of work-piece temperature. As research on this subject is furthered, and actual
effect of input processes on the cycles till failure is calculated, the significance
level of these primary studies can be altered to optimize the final equation.
3.1.2 Finale
Just as the forging die life prediction equation itself will be a valuable tool
in the cost reduction, design and optimization of forging dies for the forging
industry, the research preformed for this thesis will be a valuable primary step in
developing the die life equation.
APPENDICES
40
30
Upper Die 47 strokes per min
Upper Die 36 strokes per min
20
Velocity of Upper Die {in/s}
10
Stage A Figure F.2
BDC BDC
0
0.0178 0.0245
0 0.005 0.01 0.015 0.02 0.025 0.03
-10
Figure F.3
-20
-30
-40
Time (s)
Analysis of Variance for Z-Load Force Up Die {lbs}, using Adjusted SS for Tests
Source DF Seq SS Adj SS Adj MS F P
Friction 1 737452 737452 737452 0.16 0.698
Die Temp 1 4111770 4111770 4111770 0.89 0.367
WP Temp 1 93629814 93629814 93629814 20.19 0.001
Stroke Speed 1 234983 234983 234983 0.05 0.826
Error 11 51017288 51017288 4637935
Total 15 149731306
S = 2153.59 R-Sq = 65.93% R-Sq(adj) = 53.54%
Analysis of Variance for Net Energy {J}, using Adjusted SS for Tests
Source DF Seq SS Adj SS Adj MS F P
Friction 1 1464210225 1464210225 1464210225 7.07 0.022
Die Temp 1 435348225 435348225 435348225 2.10 0.175
WP Temp 1 10581208225 10581208225 10581208225 51.06 0.000
Stroke Speed 1 18190225 18190225 18190225 0.09 0.773
Error 11 2279581675 2279581675 207234698
Total 15 14778538575
S = 14395.6 R-Sq = 84.58% R-Sq(adj) = 78.97%
Analysis of Variance for Z-Force Up (Cht) {lbs}, using Adjusted SS for Tests
Source DF Seq SS Adj SS Adj MS F P
Friction 1 1.35387E+11 1.35387E+11 1.35387E+11 1.91 0.194
Die Temp 1 1.61765E+11 1.61765E+11 1.61765E+11 2.29 0.159
WP Temp 1 22710490000 22710490000 22710490000 0.32 0.582
Stroke Speed 1 38239802500 38239802500 38239802500 0.54 0.478
Error 11 7.77996E+11 7.77996E+11 70726936591
Total 15 1.13610E+12
S = 265945 R-Sq = 31.52% R-Sq(adj) = 6.62%
Analysis of Variance for Effective Stress Up {psi}, using Adjusted SS for Tests
Source DF Seq SS Adj SS Adj MS F P
Friction 1 1690000 1690000 1690000 0.55 0.476
Die Temp 1 39601 39601 39601 0.01 0.912
WP Temp 1 19395216 19395216 19395216 6.26 0.029
Stroke Speed 1 1229881 1229881 1229881 0.40 0.542
Error 11 34092486 34092486 3099317
Total 15 56447184
S = 1760.49 R-Sq = 39.60% R-Sq(adj) = 17.64%
Analysis of Variance for Contact Pressure {Psi}, using Adjusted SS for Tests
Source DF Seq SS Adj SS Adj MS F P
Friction 1 4225 4225 4225 0.00 0.971
Die Temp 1 4536900 4536900 4536900 1.49 0.248
WP Temp 1 370177600 370177600 370177600 121.51 0.000
Stroke Speed 1 902500 902500 902500 0.30 0.597
Error 11 33510950 33510950 3046450
Total 15 409132175
S = 1745.41 R-Sq = 91.81% R-Sq(adj) = 88.83%
ACKNOWLEDGEMENTS:
I would like to thank American Axle Manufacturing and there North Tonawanda Forge
Facility for opening there doors and allowing access to there forge facility. A special
thanks to Dr. Bamidele Oyekanmi and Matt Gersley for there time and cooperation.
REFERENCES