1、02FTM3Application of Statistical Stability andCapability for Gear Cutting MachineAcceptance Criteriaby: T.J. Maiuri, The Gleason WorksTECHNICAL PAPERAmerican Gear Manufacturers AssociationApplication of Statistical Stability and Capability forGear Cutting Machine Acceptance CriteriaT. J. Buzz Maiuri
2、, The Gleason WorksThestatementsandopinionscontainedhereinarethoseoftheauthorandshouldnotbeconstruedasanofficialactionoropinion of the American Gear Manufacturers Association.AbstractMachine tool manufacturers supplying machines to the gearing world have been in existence for many years. Themachines
3、havechangedandsohastheacceptancecriteriaforthemachines. Beforethe1980sthecriteriaforvirtuallyallmachineacceptancewaseitherasuppliersstandardtestjoborthesupplierproducingoneortwoofthecustomerspartstowithinaprinttolerance. Itwasntuntilabout1984thatwewererequiredtoperformacapabilityanalysisforamachinea
4、cceptanceonacylindricalhobbingmachine. Capabilityrequirementsforbevelmachineacceptancedidnotoccuruntilseveral yearsafter that. Today virtually every customer requiresa capability study on at least one parameter for at leastone type of part.Sincetheintroductionofcapabilityrequirementsformachineaccept
5、ancewehaveseenthegoalpostmove. InitiallytherequirementwasforaCporCpkof1.33usinga6sigmaanalysis. Nowwehaveseenrequirementsofa1.67or2.0CporCpk with a 6, 8 or 10 sigma analysison tolerancesthathave beentightened fromthe originaltolerance! Thiscancausesomerealheadachesforthemachinetoolsupplier,andthatis
6、whyitisveryimportantforthesuppliertounderstandthetrue machine and process capability before agreeing to any capability requirement.In1991thequalityandsupplierassessmentstaffsatChrysler,FordandGeneralMotorsworkedundertheauspicesoftheAutomotiveDivisionoftheAmericanSocietyforQualityControl(ASQC)Supplie
7、rQualityRequirementsTaskForceincollaborationwiththeAutomotiveIndustryActionGroup(AIAG)toputtogethertheStatisticalProcessControl(SPC)Reference Manual. The same group developed the Measurement Systems Analysis (MSA) Reference Manual. Manycompanies, such asDelphi AutomotiveSystems, havedeveloped theiro
8、wn statisticalqualification requirementsbasedonboththeSPCandMSAReferenceManual. ThispaperwillbereferencingtheSPCandMSAManualsandtheDelphiSpecification SD-002 for much of the material presented.With this paper we hope to review some of the basics of Statistical Process Control SPC and provide a bette
9、runderstanding of its application as it relates to a machine runoff.Copyright 2002American Gear Manufacturers Association1500 King Street, Suite 201Alexandria, Virginia, 22314October, 2002ISBN: 1-55589-803-31Application of Statistical Stability and Capability for Gear Cutting MachineAcceptance Crite
10、riaT.J. Buzz Maiuri : Director - Application EngineeringThe Gleason WorksCustomer Agreement, Data Collection data collection sheets are prepared;analysis of data is laid out; and the limitations of thetest are known.Discrimination Discriminationis thelarger of theapparentandeffectiveresolutionsforsi
11、nglereadingsystems. The number of data categories is oftenreferred to as the discrimination ratio since itdescribes how many classifications can be reliablydistinguishedgiventheobserved processvariation.Distribution A way of describing the output of anatural cause system of variation, in whichindivi
12、dualvaluesarenotpredictablebutinwhichtheoutcomes as a group form a pattern that can bedescribedintermsofitslocation,spreadandshape.Location is commonly expressed by the mean oraverage, or by the median; spread is expressed interms of the standard deviation or the range of asample; shape involves man
13、y characteristics suchas symmetry and peakedness, but these are oftensummarized by using the name of a commondistribution such as the Normal, Binomial, orPoisson.Effective Resolution The size of the datacategory when the total measurement systemvariation is considered is the effective resolution.Thi
14、s size is determined by the length of theconfidence interval based on the measurementsystem variation. The number of data categoriescan be determined by dividing the size into theexpected process distribution spread. For theeffective resolution, a standard estimate of this (atthe 97% confidence leve
15、l) is 1.41 PV/Rfrequently used to refer specifically to the devicesused on the shop floor; includes go/no-go devices.Gage Repeatability thecomplete process used to obtain measurements.The actual gages or measurement devices utilizedto monitor a process.MeasurementSystemError The combinationofgagebia
16、s,repeatability,reproducibility,stabilityandlinearity.Normal Distribution A continuous, symmetrical,Gausian bell shaped frequency distribution forvariables data that underlies the control charts forvariables. When measurements have a normaldistribution, about 68.26%, 95.44%, 99.73% of allindividuals
17、 lie within plus and minus one, two, andthree standard deviation from the mean,respectively. These percentages are the basis forcontrol limits and control charts analysis.Out-of-Control Condition describing a processfrom which all special causes of variation have notbeen eliminated. This condition i
18、s evident on acontrol chart by the presence of point beyond thecontrol limits or by patterns that are not randomwithin the control limits.Ppk The performance index for a stable process,typically defined as the minimum of PpkU or PpkL.PR Performance RatioProbability Set of conditions or causes workin
19、gtogether to produce an outcome.Process The combination of people, machinesand equipment, raw materials, methods, andenvironment that produces a given product orservice.Process Capability The total range of a stableprocesss inherent variation (6Rd2)ProcessPerformanceThetotalrangeofastableprocesss to
20、tal variation (6s).ProcessRoutings The documents that describethe processes required to produce a product.Range (R) The difference between the highestand lowest values in a subgroup.Reference Value - A value that services as an agreed upon refer-ence for comparison. It may be:- a theoretical or esta
21、blished value based onscientific principles;24- an assigned value based on some nationalor international organization;- a consensus value based on collaborativeexperimental work under the auspices of ascientific or engineering group; or- for a specific application, an agreed uponvalue obtained using
22、 an accepted refer-ence method.- A value attributed to a specific quantity and ac-cepted, sometimes by convention, as appropri-ate for a given purpose.- Avalueconsistentwiththedefinitionofaspecif-ic quantity and accepted, sometimes by con-vention, as appropriate for a given purpose.Regression Analys
23、is A calculation to define themathematical relationship between two or morevariables.Repeatability Variation in measurementsobtainedwithonegagewhenusedseveral timesbyone appraiser while measuring a characteristic onone part.Reproducibility Variation in the average of themeasurements made by differen
24、t appraisers usingthe same gage when measuring a characteristic onone part.Resolution The capability of the measurementsystem to detect and faithfully indicate even smallchanges of the measured characteristic. (See alsodiscrimination.)Scatter Diagram A plot of two variables, oneagainst the other, to
25、 display trends.Sigma () The measure of variability, ordispersionthatindicateshowdataspreadsoutfromthemean. Itgivesinformation aboutthe variationina process. Standard DeviationStabilityTheconditiondescribingaprocessfromwhich all special causes of variation have beeneliminated, and only common causes
26、 remain;evidenced by the absence of points beyond thecontrol limits and by the absence of non-randompatterns or trends within the control limits.Standard Deviation (see Sigma)StatisticalProcessControlTheuseofstatisticalmethods and techniques such as control charts toanalyze a process or its output s
27、o as to takeappropriate actions to achieve and maintain a stateof statistical control and continue improvement ofprocess variability.SubgroupOneormoreeventsormeasuresusedto analyze the performance of a process. Rationalsubgroups are chosen so that the variationrepresented within each subgroup is as
28、small asfeasible.Tolerance Allowable deviation from standard.That is, the permitted range of variation about anominal value. The permitted tolerance is thedifference between the upper and lowerspecification limits. Specification limits should notbe confused with control limits.Unilateral Specificati
29、on Unilateral tolerancesarethosewhichhaveonlyasinglelimite.g.mustnotexceed 1000 lb., hardness to be 60 Rockwell orless.Variation The inevitable differences amongindividual outputs of a process; the source ofvariation can be grouped into two major classes:Common Causes and Assignable (Special)Causes.
30、X and R Chart - (see control chart)Zero Based Dimensions These dimensionshave a value of zero as their inherent target value(e.g. roundness, concentricity, and surface finish).They usually generate distributions that have avisible amount of skewness or non-normality.ReferencesChrysler Corporation, F
31、ord Motor Company, andGeneral Motors Corporation, Statistical ProcessControl SPC - Reference Manual, Issued 1992Chrysler Corporation, Ford Motor Company, andGeneral Motors Corporation, MeasurementSystems Analysis MSA - Reference Manual,Issued 1992DelphiSaginawSteeringManufacturingEquipmentStatistica
32、l Qualification Requirements SD-002,Saginaw Steering Systems, Issued March 1, 1993:Revised January 7, 2002Quality Planning and Analysis, J.M. Juran andFrank M Gryna, Jr., McGraw Hill PublishingCompany, Copyright 1980Quality Control Handbook, J.M. Juran, ThirdEdition, McGraw Hill Book Company, Copyright1979Phoenixis a registeredtrade markof theGleasonWorksMicrosoft is a registered trademark of MicrosoftCorporation
