Application of Statistically capable Shot PeeningAutomotive Components Design.

Adam J. VahratianProduct EngineerPowertrain DivisionFord Motor CompanyLivonia, Michigan 48151 USA

ABSTRACT

Robert P. GaribayVice PresidentAdvanced MaterialProcess CorporationWayne, Michigan 48184 USA

Today's world automotive business climate is demanding morecost effective, higher strength/weight designs than everbefo~e. Engineering coverage factors are shrinking and, atthe same time, designs are becoming more robust.consequently, manufacturing processes without a quantifiedbenefit/cost ratio are being discarded. Those processes thatare utilized are required to statistically demonstrateacceptable product performance.

This paper will examine the application of the shot peenprocess to automotive components where the process designintent is dependant upon meeting a minimum fatigue strengthobjective. Specific automotive component applicationtestwork, process engineering, and Advanced Quality Planningnecessary to produce the statistical control limits resultingin acceptable product performance will be discussed as wellas production monitoring requirements for statisticalcapability. The benefit/cost ratio of such a process willthen be analyzed in light of conventional controlled shotpeening and other strength enhancement processes applied tothe same component.

KEYWORDS

coverage factors,' Advanced Quality Planning, statisticalcapability, benefit/cost ratio, parameter, variable, "sweetzone", strength/weight, time-to-manufacture.

INTRODUCTION

The following definitions are important to understanding thispaper:

Parameter: Any of a set of physical properties whose valuesdetermine the characteristics or behavior of theprocess. i.e, shot velocity, shot size, shotimpact angle, etc.

variables: A machine or media specific quantity that mayassume anyone of a set of values, thusinfluencing a parameter value. i. e, air pressure,shot breakdown rate, shot flow rate, etc.

10 MAT-TEC-93

The potential for fatigue strength increases of 25% - 50% dueto shot peening has been well documented (1) (2) (3) (4) (5)(6). It has also been established that there is a strongcorrelation between process parameter values and themagnitude of the fatigue strength increase induced by shotpeening. Numerous publications support the fact that uniqueoptimum peening process parameter values or ranges exist forspecific workpiece characteristics (6) (7) (8). As far backas 1943, John Almen identified the concept of a peeningintensity "sweet zone" (9).

If fatigue strength increases of 25% - 50% were obtainablefrom a particular technology, it seems logical that suchbenefits would be systematically included in product designunless process reliability and/or cost were a barrier.Numerous cases of high fatigue strength benefit produced intestwork followed by excessive fatigue strength variabilityonce in production, have characterized the history of theshot peen process. consequently, shot peening has become aprocess of last resort, rather than a process of choice.

Assuming relatively inclusion free material, when the fatiguestrength scatter of the peened population is significantlybroader than the nonpeened, the increase in variability canonly be attributed to shot peening variability. This isespecially significant, consideri~g the fact that fatiguestrength scatter of the shot peened population should be lessthan the nonpeened. This is due to shot peening's mitigatinginfluence on the contribution to fatigue strength variabilityof pre-peen manufacturing processes such as machining andheat treating, which can respectively produce surfaceresidual tensile stress and decarburization (10) (11) (12).Under these circumstances, the increase in fatigue strengthvariability can only be attributed to critical peeningprocess parameters fluctuating in and out of their "sweetzones" or optimum tolerances.

To further complicate matters, conventional controlled shotpeening has utilized the almen strip measurement system andvisual coverage as the primary process monitoring tools.The reasons this system is inadequate for producingacceptable fatigue strength variability have been welldocumented:

1. Almen intensity provides an indication of theaggregate amount of energy transfer to the workpiecewithout defining the individual contributing processparameters and variables (13) (14).

2. The effect of the cumulative tolerances within thealmen strip system itself will, in most cases, resultin undetected process variation that is too excessivefor acceptably reliable fatigue performance (15) (16)(17) .

SHOT PEENING, SURFACE TRAITEMENTS GRENAILLAGE... 11

3. Visual coverage is a qualitative attempt to relatealmen saturation to workpiece saturation (13). Theyoften do not occur simultaneously due to differencesin surface hardness and metallurgy. In addition tothe obvious weakness of inspector jUdgement, tracerdyes and coatings notwithstanding, it is impossiblefor a human to visually discern coverage levels above100% or between primary or secondary impactimpingement (13). Additionally, in many cases theoptimum workpiece saturation level is above 100%coverage (8). (See (13) for monitoring 100% + coveragelevels).

Finally, this almen strip/visual coverage measurement systemis the foundation upon which all industry and military shotpeening specifications rest. No matter how well one complieswith these specifications, if they are the sole governingcriteria, it has been statistically demonstrated thatunacceptable fatigue strength variability is a highly likelyresult. The specifications can be no more statisticallyreliable than the process measurement system used (16).

For the reasons described above, the conventional approachof engineering a shot peen process based solely upon almenintensity and coverage values (whether these values arederived from cursory testwork or from pUblished shot peenspecification recommendations) is highly unlikely to producestatistically reliable fatigue strength gains, and therefore,justifiable benefit/cost ratio results.

In the authors opinion, obtaining statistical reliabilitybegins by relating specific levels of critical shot peenprocess parameters directly to the fatigue performance of theworkpiece. (See (6) (16) (18) (19) and (20) for a more in­depth treatment of the type of work involved). The dataderived from this testwork provides the peening processparameter optimum values and the process variable tolerancevalues for engineering a statistically reliable shot peenprocess. without this data, the likelihood that a specificshot peen process will produce unacceptable variability (whenprocess variable tolerances are too broad), suboptimumfatigue strength (when optimum process parameter values areunidentified), or unnecessary cost (when variable tolerancesare maintained to tighter than necessary values) is very high(8) (16).

Engineering a shot peen process by statistically relatingspecific levels of peening process parameters and variablesto workpiece fatigue strength performance is, in the authorsopinion, the way to reliably reproduce, in high productionvolumes, the high level of fatigue strength benefit the shotpeen process has historically been capable of developing inthe laboratory. The following will briefly summarize this'method. This will be followed by an example of how the FordMotor Company used this technique and the benefitjcost ratiothat resulted.

12

DISCUSSION

The engineering method

MAT-TEC-93

Typically a shot peen production process is engineered in thefollowing manner:

1. The component is shot peened to an almen intensity andcoverage value chosen from past internalspecifications or from published industry or militaryspecification / recommendations.

2. If the component passes the fatigue test, theproduction process is defined using almen intensityvalue, coverage and shot size. Also, in some casescompliance to a pUblished industry or militaryspecification is required.

3. If the component fails the fatigue tests, anotheralmen intensity value is tried or shot peening isabandoned for another solution.

Because of the reasons stated in the introduction, solereliance upon the almen strip process monitoring system andalmen strip based specifications is an unreliable approachto either determine if shot peening can solve a particularfatigue problem, or to implement a shot peening solution intoproduction. To do so will, in most cases , result inundetected process variation that is excessive. withoutstatistically defining each parameter optimum value andacceptable variable tolerance based upon the effect onfatigue strength magnitude and variability, one has noestablished reference points in the search for optimumbenefit/cost ratio during testwork, and similarly noassurance the reproduction of that sweet zone is occurringin production (16).

Table 1 lists the peening process parameters, and Table 2lists the peening process variables.

Table 1

Process parameters critical to fatigue strength

1.2.3.4.

Shot velocityShot diameterShot hardnessShot type/density

5.6.7.

Shot impact angleNozzle to workpiece position relationshipWorkpiece saturation

Table 3 provides an outline of the steps to statisticallyengineer a fatigue strength based shot peen process. "Thedefining element of this approach is that the ultimate goal

SHOT PEENING, SURFACE TRAITEMENTS GRENAILLAGE...

Table 2

Process variables critical to statistical reliability

13

1.2.3.4.

Shot size distributionShot shape distributionShot flow rateAir pressure

5.6.7.

Shot blast pauemExposure timeWorkpiece motion

shouldn't be to meet the requirements of any particularspecification or even any particular shot peening intensity.Rather it should be to achieve, within acceptable statisticalreliability, certain desired levels of process inducedbenefit on a statistically acceptable percentage ofproduction workpieces" (16). The amount of testwork requiredto accomplish this is not inexpensive or quick. However,once a database is established, the number of test iterationsnecessary to establish process parameters is reduced,shortening the time to engineer the process.

Table 3

Steps for statistically engineering a fatigue strength based shot peen process.

1. Analysis of workpiece failure mode via SEM, FEA. and/or baseline fatigue testing.2. Define optimum peening process parameter test matrix based upon above analysis and data base

using statistical design of experiments methodology.3. Determine optimum value for each critical process parameter with the corresponding fatigue

strength distribution utilizing testwork peening machinery whose control capability meets orexceeds Table 4.

4. Defme minimum fatigue strength requirements and the statistical indices necessary to assurecompliance.

5. Determine critical process variable tolerances based upon their cumulative effect on fatigue strengthrequirements.

6. Define process control limits based upon item 5.7. Obtain high production peening equipment with statistical capability within the defined process

control limits.8. Use statistics to monitor process variable trends in production.9. Statistically sample workpiece fatigue performance.

Properly performed, the resultant benefit/cost ratio of aproduction shot peen process so defined can beunprecendentedly high. Once such a capability is obtained,the strategic advantage afforded a~ automobile manufacturerin terms of design strength/weight, manufacturing costs,product performance, and time-to-manufacture, are just assignificant.

The following is a typical example of applying theseengineering principles.

14

An application

MAT-TEC-93

In 1986, Ford Motor Company introduced the Ford Taurus andMercury Sable four door sedans. These cars were equippedwith the newly designed AXOD (Automatic Transaxle withOverdrive) transmission, applied to a 3.0 liter (3.0L), V­6 engine capable of 135 horsepower. In 1989, thistransmission was scheduled to be applied to the new LincolnContinentai, equipped with a 3.8 liter (3.8L), V-6 enginecapable of 140 horsepower. Finally, in 1992, this sametransmission was called upon to handle the 220 horsepower,3.2 liter (3.2L) SHO (super high output), V-6 four valveengine application.

A consideration for the first upgrade of the AXODtransmission for the Lincoln continental, and sUbsequentlyfor the second upgrade for the Taurus SHO, was the finaldrive planetary pinion gearset. The purpose for the gearsetis to mUltiply the torque between the primary geartrain andthe vehicle axle.

The workpiece was a helical pinion machined from 5130 steeland carburized to a case hardness of 58 - 62 Rc. The failuremode, as defined by Finite Element Analysis (FEA) , was geartooth bending fatigue. This was confirmed with dynamometerfatigue testing and Scanning Electron Microscopy (SEM)failure analysis.

The requirement for validating the transmission for theLincoln continental (3.8L) application was a 9.25 hour B-10life, at a predetermined dynamometer low gear torque output.(Note, due to the proprietary nature of this testwork, theactual torque requirements will not be disclosed) .

99

90

50

20

10

Figure 1.

Weibull 00\0

~~J~hl ~.k~ 1~~

I

10 0 = failure 100

life ~ = suspension

Weibull probability plot of gear life at3.8L loads without shotpeen.

The transmissionas applied to the3.8L engine had aB-10 life ofapproximately 5.5hours as shown inFig. 1.

A peening processparameter testmatrix wasdeveloped with theobjective ofgenerating a 68%increase in B-10life. Referringto Table 1, theprocess parametersinitially studiedwere shot velocity,

SHOT PEENING, SURFACE TRAITEMENTS GRENAILLAGE... 15

shot diameter, and shot impact angle. The shot used was caststeel with a density of 7 gms./ml. and hardness of 55 - 60Rc. The size distribution was maintained to the range of0.0139 inch diameter to 0.0165 inch diameter. All variableslisted in Table 2 were maintained per Table 4.

The optimum process parameter values were determined fromfatigue life data gathered from three months of dynamometertestwork. Fig. 2, which illustrates the final results ofthis testwork, demonstrates a 165% increase in B-10 life.

Table 4

Typical example of precision peening equipment process control capability.

Process Parameter

1. Air pressure

2. Shot size distribution

3. Shot shape

4. Shot impact angle

5. Shot flow rate

+/- 0.75 psi

95 % between 0.0138"dia. and 0.0165' dia.

< 5% broken

+/- 2 degrees

+/- 2 oz.l~.

_"'__II

x

x

6. Nozzle to workpieceposition relationship:via workpiece fixture position repeatabilityvia workpiece height fiber optic eyevia workpiece rpm sensor

7. Exposure time:minimum via timermaximum via computer

+/- 0.001"·0 + 0.062'+/- 2%

+/- 0.25 sec.+/-0.25 ~.

x

xx

xxx

10 -L.

.. = maintained by the machine II = monitored electronically

')0

Weibull Dolo

~~&%61 ~L~o~~

50

20

-~-~--f~.,----.·-·t-·I-r"1-'

10 <) = failur:-e 100Lire > = suspension

Figur:-e 2. Weibull probability plot of gear:- lifepotential at 3.8L loads using shotpeen.

This testworkshowed thepotential fatigues t r eng t himprovementobtainable throughoptimized shotpeening and wouldbecome valuable forthe next upgradeoft histransmission,

The next phase oftestwork, definingminimum fatiguestrength

16 MAT-TEC-93

requirements and thetolerances necessary to4 & 5) required another

o

99

90

""~ 50

c:"u" 20n.

"" 10

j

Weibull DoloSlope B 10 Life4.89912 9.65263

o

corresponding process variableassure compliance (see Table 3 stepfour months of testwork and produced

the results shownin Fig. 3. Thisdata represents thelowest cost processnecessary tostatisticallyassure compliancewith the B-10 liferequirement of 9.25hours at thevalidation torquerequirement.

equipment process control requirementsTable 3 steps 6 & 7).

From this data andthe backgrounddata, the processcontrol limits weredefined and thecorresponding highproduction peening

were determined (see

100o '" failure> '" suspension

10

LifeFigure 3. Weibull probability plot of gear life at

3.8L loads with lowest cost shotpeen.

since implementing into production in September 1988, theproduction process has continued to produce final driveplanetary pinions that meet the requirements forstatistically acceptable product performance (Table 3 steps8 & 9). See (21) & (22) for the type of statistical measuresthe authors suggest employing to monitor a shot peen process.

The testwork for upgrading the AXOD transmission for theTaurus SHO (3. 2L, 4 valve) application was performed insimilar fashion. The dynamometer validation torquerequirement was increased 12.5% over the 3. 8L validationrequirement with the same B-10 life requirement of 9.25hours.

Testwork indicated that the current production shot peeningprocess produced inadequate fatigue strength at the highertorque levels. The decision was made to upgrade the gearsteel from 5130 to 8620 as well as to implement the previousshot peen optimization testwork process utilized during the3.8L upgrade testwork, shown in Fig. 2. This data indicatedgreater fatigue strength results at higher peening energytransfer levels.

Fig. 4 illustrates the testwork results after completingsteps 1 through 3 in Table 3. A B-10 life of 11 hours wasobtained at the higher torque levels. Once again, completingsteps 4 and 5 in Table 3 resulted in reducing shot peenprocess costs with a lowering of the energy transfer to a

SHOT PEENING, SURFACE TRAITEMENTS GRENAILLAGE... 17

level between the 3. 8L production peen process and theoptimized level shown in Fig. 4.

99

90

50

20

10

5 -

Weibull DoloSlope B 10 Life.32.2966 11.088..1

c

(

>>

10 0 = failure 100Life "> = suspension

Figure 4. Weibull probability plot of gear lifepotential at 3.2L SHO loads using shotpeen.

The resultant dataand B-10 life of9 . 6 hours is shownin Fig. 5.

Benefit/costcomparison

The following willanalyze thejustification ofsuch an expensiveendeavor based uponcomparing theres u 1 tan tbenefit/cost ratio(B/C) to the otheroptions available.

50 -

20

10

Weibull DoloSlope B 10 Life11.566 9.58907

10

LifeFigure 5. Weibull probability plot of gear life at

3.2L SHO loads wi th lowest cost shotpeen.

100

For dLicussionpurposes, theauthors def ine B/ Cas tb~ percentageinc):-ease inhorsepower ratingdivided by the costof obtaining theincrease. Thebaseline forcomparison is a 63%increase inh 0 r s e powerapplication at acost of less than$1.00 pertransmission. Thisproduces a B/C ofgreater than 63.

other options available for increasing horsepower ratinginclude designing a larger transmission and using a higherstrength steel. The cost associated with designing,validating, and tooling for production of a new gearset isconservatively estimated to be in the mUltiple millions ofdollars. The resultant B/C' is less than 0.0001. Notaccounted for in this figure is the corresponding performanceand fuel economy penalties resulting from the weight increaseassociated with a larger transmission, as well as the time­to-manufacture penalty measured in years.

Upgrading to higher fatigue strength steels usually resultsin higher machining, heat treat, and raw material costs.

18 MAT-TEC-93

only a limited amount of fatigue strength improvement can beobtained, such as the change from 5130 to 8620 or 4615M,without the cost increases reaching significant proportions.Changing to either of these materials still requires theapplication of shot peening to produce the required fatiguestrength.

To obtain a comparable fatigue strength increase without shotpeening would require materials in the class of 9310 orAermet 100, both used in aerospace applications. Thesematerials, besides being more expensive, must be subjectedto additional heat treatments to achieve a machineablestructure, plus additional operations to eliminate austeniticmicrostructure. If we assume a $2.00 - $3.00 cost penaltyfor using this material, the resultant BjC ratio is 25.2.Not included in this figure is the time-to-manufacturepenalty of 2 - 3 years for validation and tooling.

Another option to consider is conventional controlled shotpeening. It has been demonstrated, and now is commonlyaccepted, that statistical process control increases qualitywhile reducing cost (23). Therefore although unquantifiedin this example, conventional controlled shot peening, asdefined in this paper, will lack the level of fatiguestrength benefit obtainable with statistically capable shotpeening. If it were possible to obtain a comparable benefit,the costs associated with such a hypothetical scenerio wouldbe higher for conventional controlled shot peening than forstatistically capable shot peening. This is due to the factthat the statistical indices necessary to assure complianceto a minimum fatigue strength requirement are unknown.

The time-to-manufacture required for engineering astatistically capable shot peen process ranges from 3 to 14months, depending upon fatigue test scheduling and existingshot peen capacity availability. This characteristic of theshot peen process becomes a benefit when comparing to time­to-manufacture for transmission redesign and for materialupgrades.

Summary

Table 5 provides a comparison of the cost effectivenessassociated with alternatives for increasing the strength ofthe AXOD final drive planetary gearset. The requirement wasa 63% increase in horsepower capacity.

statistically capable shot peening I as described by theauthors, has the highest benefit/cost ratio (BjC) and theshortest time-to-manufacture of all the strengthening optionsavailable to Ford. Conventional controlled shot peening isthe worst choice due to its lack of statistical capability.

SHOT PEENING, SURFACE TRAITEMENTS GRENAILLAGE...

Table 5

Cost effectiveness comparisons of options for increasing AXOD transmission strength.

19

B/C

Time-lo-manufacture

Statistically CapableShot Peening

63

3 - 14 months

0.0001

3 - 5 years

MaterialUpgrade

25.2

2 - 3 years

ConventionalShot Peening

lower

1- 12 months

A statistically capable shot peen process is based uponstatistically relating critical peening process parametersto actual workpiece strength. A conventional controlled shotpeen process is based upon the statistically incapable almenstrip measurement system. A statistically capable shot peenprocess is one of the most cost effective technologies forincreasing automotive transmission capacity.

REFERENCES

1. Horger, o. J., "Mechanical and Metallurgical Advantagesof Shot Peening", The Iron Age, March 29, and April5, 1945.

2. Fuchs, H.O., and Bickel, P.E., "Shot Peening ofSprings", springs, Vol. 2, No.1, pp. 16 - 20, May,1963.

3. Daly, J. J., "Boosting Gear Life Through Shot Peening",Machine Design, May 12, 1977.

4. Schreiber, R., Wohlfahrt, H., and Macherauch, E., "TheEffect of Shot Peening on The Bending Fatigue strengthof the Case Hardened Steel 16Mm Cr 5", Arch,Eisenhuttenwes, Vol 49, No.1, January 1978.

5. Hatano, A., Namiki, K., "Application of Hard ShotPeening to Automotive Transmission Gears", Daido SteelCo., Ltd., Steel Res. section.

6 . Gar ibay, R. P., and Chang, N. S., "Improved FatigueLife of a Carburized Gear by Shot Peening ParameterOptimization", ASM International Conference onCarburizing, Processing, and Performance, July, 1989.

7. Simpson, R.S., and McCagg, D.R., "Effects of Peeningon Fatigue Life of Wrought SAE 5454 AluminumAutomotive Road Wheels", Second InternationalConference on Shot Peening, May, 1984.

8. Simpson, R. S ., and Chiasson, G. L., "Quantification ofthe Effects of Various Levels of Several Critical ShotPeening Process variables on Workpiece SurfaceIntegrity and the Resultant Effect on WorkpieceFatigue Life Behavior", Flight Dynamics Laboratory AirForce Wright Aeronautical Laboratories, October, 1988.

9. Almen, J., "Shot Blasting to Increase FatigueResistance", SAE Transactions, Vol. 51, No 7.

20 MAT-TEC-93

10. W.P. Koster, L.R. Gatto, and J.T. Cammett, "TheInfluence of Shot Peening on Surface Integrity of SomeMachined Aerospace Materials", First InternationalConference on Shot Peening, 1981.

11. Miwa, Y., et aI, "carbonitriding and Hard Shot Peeningfor High Strength Gears", SAE Technical Paper Series880666.

12. "Material Characterization of Aermet 100", WLjMLSE,Wright Patterson AFB, Ohio 45433-6533.

13. Simpson, R.S., and Chiasson, G.L., "A New concept forDefining optimum Levels of a critical Shot PeeningProcess Variable", Second International Conference onImpact Treatment Processes, September, 1986.

14. Fuchs, H., "Defects and virtues of the Almen IntensityScale", Second International Conference on ShotPeening, May, 1984.

15. simpson, R.S., and Clark, D.O., "The Effects ofInherent Tolerances in the Almen Test strips on ShotPeening Process Reliability", Third InternationalConference on Shot Peening, October, 1987.

16. Simpson, R.S., "Acceptable Product Variability and theShot Peen Process: Performance Oriented Processing",Engineering Materials Advisory Services Ltd., 1993.

17. Bonde, carl, letter to the editor as printed in theShot Peener, Vol. 6. Issue I, pp.9

18. Chang, N.S., and Garibay, R.P., "Maximizing theFatigue Strength of a Carburized Helical Gear ViaOptimizing Shot Peening Process Parameters Part IAlmen Intensity", ASME 5th International PowerTransmission and Gearing conference, April 1989.

19. Garibay, R. P., and Terry, S. L., "High Technology ShotPeening. for the Automotive Industry", IITTInternational Conference on Shot Peening Theory andApplication, May 1990.

20. Garibay, R. P., "Optimized High Production Shot Peeningfor the Automotive Industry", IITT InternationalConference on Shot Peening Theory and Application,September 1992.

21. Simpson, R.S., "An Analysis of the Internalstatistical Capability of Several Common ControlledShot Peening Specifications", Fifth InternationalConference on Shot Peening, September 1993.

22. Wieland, R. C., "Performance Specifications - A ProductEngineering Process Specification Alternative", FifthInternational Conference on Shot Peening, September1993.

23. Campanella, J., "Principles of Quality Costs, SecondEdition, Principles, Implementation, and Use", ASQCQuality Press, American Society for Quality control,1990.