《ProcessOptimization》由会员分享,可在线阅读,更多相关《ProcessOptimization(87页珍藏版)》请在金锄头文库上搜索。
1、Process OptimizationProcess Characterization61Gage R&R - Metrics from Excel SpreadsheetConsider: P/T Ratio and %GR&R Gage Repeatability and Reproducibility Analysis ResultsProcess Characterization62P/T RatioP/ T Ratio 515100%.MToleranceGreater than 30% -Not AcceptableBetween 30% and 10% - MarginalLe
2、ss than 10% -Acceptable*Less than 5% -Six Sigma*Check with customer requirements for standards of acceptabilityMeasurement System Analysis*Tolerance means process specification widthProcess Characterization63Percent Gage R&RA Six Sigma measurement system should have %GR&R of 10% or lessCurrently, %G
3、R&R is preferred over the P/T RatioMeasurement System AnalysisSigma M means measurement system variationSigma T means total system variationProcess Characterization64Classic GR&R vs DOE-based GR&R StudiesDOE-based Studies . . . are more flexible require additional skills require statistical software
4、Contact your local statistical resource for more informationMeasurement System AnalysisProcess Characterization65Gage vs. Process CapabilityRelationship Between Actual Cpk, Measured Cpk and P/T Ratio0.000.501.001.502.00Measured metric Actual Cpk = 1.7 Measured Cpk = 1.0 P/T Ratio = 70%Process Charac
5、terization66Improvement MethodsRepeatabilitymultiple readingsmodify existing gagepreventive maintenancebuy new equipmentReproducibityoperator trainingprocess methodsProcess Characterization67Measurement System SummaryAnalysis involves quantifying and understanding all of the characteristics of the m
6、easurement systemDiscrimination / resolutionAccuracycalibration intervalsbias from known standardsStabilitycontrol charts t-test, F-testLinearitycorrelation between reference values and biasPrecisionGR&RANOVAPreventive maintenance schedulesComparisons of two or more gagest-test, F-test, ANOVAProcess
7、 Characterization68Passive Data CollectionUse to evaluate stability and capabilityCollect 30 or more independent observations, make a run chartCollect as much information as possible about independent variablesNo Tweaking or changing of the processThoroughly analyze the resulting dataExamine variabl
8、e distributionsExamine correlations between Response Variables and Independent VariablesReport the resultsProcess StabilityProcess Characterization69Process StabilityIn a stable process, the distribution of the process measurements remains stable over time.Predictability of processes and equipment a
9、re crucial to the success of a manufacturing line. If a process is stable, and therefore predictable, it will be safe to proceed with process optimization, if necessary.Process StabilityProcess Characterization70Characteristics of a Stable ProcessConstant mean over timeIndividual points form a rando
10、m pattern around the meanUniform variability over timeAbsence of trends, runs, or erratic ups and downsVery few points outside control limitsMost points near center line, few points approach control limitsProcess Characterization71Time Run Charts - A Stable Process5101520253056789101112131415+3-3 Pr
11、ocess StabilityProcess Characterization72Characteristics of Unstable ProcessUpredictable changes in pattern over timeMean changes over timeIndividual points form a non-random patternTrends, runs, erratic ups and downs, sudden shiftsVariability changes over timeMany points outside of control limitsPr
12、ocess Characterization73Time Run Charts - An Unstable Process5101520253056789101112131415+3-3 Process StabilityProcess Characterization74Instability IndexFor a Stable Process with 3 sigma Control Limits . . .St = 0.27%tS =Number of Points Out-of-ControlTotal Number of Pointsx 100%This metric require
13、s a very large sample size (n2000) Process StabilityGOAL:Process Characterization75Stability of Upstream and Downstream ProcessesUpstream ProcessProcessbeing CharacterizedDownstream ProcessIndependent VariablesConsider the stability of upstream processes and factorsProcess StabilityProcess Character
14、ization76Process / Equipment EvaluationProcess Characterization77Process/Machine EvaluationProcess Characterization78Process/Machine EvaluationProcess Characterization79Stability EvaluationTwo types of dataIndividual readings taken in consecutive orderrun chartsuse 3 std dev limitsControl chart samp
15、lesx-bar, range or x-bar, sigma chartsuse control chart limits Process stability vs. process capabilitystability is the ability of the process to produce similar results over timecapability is the ability of the process to fit within specification limitsAs process capability becomes worse, process s
16、tability becomes more critical to producing good parts.Process Characterization80Long Term StabilityMake certain to observe the stability of a process over an extended period of time (30-60 Days)Look for signs of long term instability, often due to equipment wear, buildup of residues or dirt, or gra
17、dual breakdown of standard methodologies Process StabilityLong term study is most effective in step 4 to evaluatethe effectiveness of process optimization and processcontrols over time.Process Characterization81Identify & Eliminate Assignable CausesIf the process is found to be unstable, look for ch
18、anges in the independent variables that might be associated with changes in the response variables. Look for switches turned on and off, worn parts, dirty parts . . .Consider the possibility that an important, uncontrolled independent variable has yet to be identified.Consider bad lots, unannounced/
19、unnoticed supply changes, changes in water/air compression, uncalibrated gages . . .Process StabilityProcess Characterization82Identify & Eliminate Assignable Causesbrainstormingcause and effect analysisparetosFMEApassive data collectioncorrelationsmulti-vary-analysisB vs. CProcess Characterization8
20、3Process CapabilityNormal DistributionsCapability MetricsCpCpkOne-Sided MetricsNormal Probability PlotsCapability Metrics for Non-Normal DistributionsTransformationNon-conforming PercentagePearson tablesLong Term CapabilityProcess CapabilityProcess Characterization84Normal Distributions-4-3-2-112340
21、-4-3-2-101234-6-556Most Capability Indices Assume Process Data to be Normally DistributedProcess CapabilityProcess Characterization85Capability Metrics - CpCp = USL - LSL6-4-3-2-101234-6-556LSLUSLLSLUSL-4-3-2-101234-6-556Cp = 2.0Cp = 1.33The Cp metric does not account for off-target processes 6 6 Pr
22、ocess CapabilityProcess Characterization86Capability Metrics - Cp vs Sigma3.01.003.51.174.01.334.51.505.01.675.51.836.02.00SigmaCpPoor CapabilityMarginal CapabilityGood CapabilitySix Sigma CapabilityAssuming the process is on target . . . What if the process is not on target?Process CapabilityProces
23、s Characterization87Capability Metrics - CpkLSLUSLMeanCpk = 2.0LSLUSLMean3 Cpk = 1.5 Cpk = Min ,USL - MeanMean - LSL3 3 3 -4-3-2-101234-6-556 -4-3-2-101234-6-556-7-8 Process Capability6 Distribution On-Center6 Distribution 1.5 Off-Center Process Characterization88Capability Metrics - Cpk - Propertie
24、sCpk = 1.5LSLUSLMeanLSLUSLMeanCpk = 1.5Cpk = Min ,USL - MeanMean - LSL3 3 3 3 -4-3-2-101234-6-556-7-8 0-4 -3 -2 -1 1 2 3 4 -6 -5 5 6 Process CapabilityNarrow Distribution Off-CenterWide Distribution On-CenterProcess Characterization89Capability Metrics - Cpk vs Sigma3.00.503.50.674.00.834.51.005.01.
25、175.51.336.01.50SigmaCpkPoor CapabilityMarginal CapabilityGood CapabilitySix Sigma CapabilityAssuming the process is shifted 1.5 from target . . . If process is on target, the Sigma Level will be even higher! Process CapabilityProcess Characterization90Capability Metrics - One-sided SpecOne-sided Cp
26、k = Mean - Spec. Limit30-4-3-2-11234-6-556USLMeanA one-sided Cp does not exist, but . . . 3 Process CapabilityProcess Characterization91Normal Probability Plots0306090120150 0.1 1 5 20 50 80 95 9999.9cumulative percent048121620frequencyX25354555657585Y 0.1 1 5 20 50 80 95 9999.9cumulative percent024
27、681012frequencyUse a normal probability plot to determine if the data are normally-distributedUse statistical software to obtain normal probability plotsProcess CapabilityProcess Characterization92Capability Metrics for Non-Normal DistributionsTransformationTry transforming the data using Log, Ln, o
28、r powersCheck normality using normal probability plotsUse transformed data to calculate capability metricsNon-Conforming PercentageRequires very large sample sizes (n1000)Cp = USL - LSLP0.9987- P0.0013Cpk = Min ,USL - P0.5P0.5- LSLP0.9987- P0.5P0.5- P0.0013The 99.87th PercentileThe MedianProcess Cap
29、abilityProcess Characterization93Capability MetricsGive both Cp and CpkCp - the best it can be.Cpk - what it actually is.Process CapabilityProcess Characterization94Stability vs Capability67891011121314USLLSLUSLLSL5101520253067891011121314USLLSL51015202530USLLSLCapableNot CapableStableNot StableProc
30、ess stability and capability are two distinct propertiesProcess CapabilityProcess Characterization95If the Process is Not Capable . . .Check to see if the process is significantly off target? If so, the process must be optimized in such a way as to move the process mean to the target.If the process
31、is essentially on target, it will be necessary to reduce process variation by identifying the independent variables that are potential sources of that variation.The optimization will identify levels of these independent variables that result in improved process performance.Use the Cross Reference Ma
32、trix to identify Independent Variables which may be major sources of variabilityProcess OptimizationProcess Characterization96Capability ImprovementsCpk = 1.0LSLUSLMeanLSLUSLMeanCpk = 1.03 3 -4-3-2-101234-6-556-7-8 0-4 -3 -2 -1 1 2 3 4 -6 -5 5 6 Process CapabilityDistribution Off-CenterDistribution
33、On-CenterCp = 2.0Cp = 1.0Process Characterization97Step 2: Process CapabilityProcess Characterization98Process OptimizationIdentify/prioritize independent variables which may have an effect on the response variable characteristicList controlled independent variables (experimental factors)List uncont
34、rolled independent variables (noise factors)Determine levels of controlled independent variablesIdentify significant Independent VariablesScreening ExperimentsOptimize the processResponse Surface ExperimentsConfirm the optimization modelConfirmation RunsProcess OptimizationOptimization involves an i
35、terative approach. Dont try to optimize a process in one experiment.it is unlikely to locate the true optimum levels. Process Characterization99The Goal of OptimizationOptimize a process both in terms of the mean output and the variation around the meanDOE is used to systematically investigate the r
36、elationship between the inputs and the outputsWell designed experiments provide the most efficient means for understanding a process with the highest probability of successPoorly designed experiments can waste time and resources; little or no useful informationProcess Characterization100Identify/Pri
37、oritize Independent VariablesMost of the influential independent variables have been identified while describing the total process.Some new independent variables may have been added as a result of the passive data collection analysis.Select those that have the highest probability of influencing the
38、response variable.List both the controlled independent variables and uncontrolled independent variables (noise factors).Process OptimizationProcess Characterization101Identify Significant Independent VariablesUse a screening experiment to determine which of the independent variables have a significa
39、nt impact on the response variable.The screening experiment design should allow estimation of each independent variables main effects and estimation of the two-way interactions between each pair of independent variables.Main EffectsABCDTwo-Way InteractionsABACADBCBDCDProcess OptimizationProcess Char
40、acterization102Determine Levels of Controlled VariablesUse best judgment to select low and high levels Sensitivity studies, center pointsLevels too close may hide a significant effectProcess variation may mask effect of the experimentLevels too extreme may result in bad outputSome cells may be impos
41、sible to performProcess OptimizationProcess Characterization103Screening Experiment DesignsFull Factorial DesignsFractional Factorial DesignsProcess OptimizationScreening Designs are two-level designs used to determine which factors have a significant effect on the response variableUsed to “screen”
42、out unimportant factorsAlso used to “hunt” for an optimumProcess Characterization104Full Factorial Experimental DesignRun every possible combination of High and Low levelsRun12345678ALowLowLowLowHighHighHighHighBLowLowHighHighLowLowHighHighCLowHighLowHighLowHighLowHighTempPressPowerIndependent Varia
43、blesResponse2.64.14.47.83.91.78.03.2Process OptimizationProcess Characterization105Full Factorial Design - Main EffectsTo evaluate the main effect of A (Temperature), compare the averages of the two levelsRun12345678ALowLowLowLowHighHighHighHighBLowLowHighHighLowLowHighHighCLowHighLowHighLowHighLowH
44、ighTempPressPowerIndependent VariablesResponse2.64.44.17.83.98.01.73.24.734.20Significant Difference?The analysis requires a measure of experimental variability Process OptimizationProcess Characterization106Replication in a Factorial DesignEach run of the experiments design must be replicated in or
45、der to estimate experimental variation and use statistical tests of significance.Run12345678ALowLowLowLowHighHighHighHighBLowLowHighHighLowLowHighHighCLowHighLowHighLowHighLowHighTempPressPowerIndependent Variables2.64.44.17.83.98.01.73.2Response2.44.24.38.13.87.91.73.02.74.54.07.73.98.11.53.3RANDOM
46、IZE!Process OptimizationProcess Characterization107Full Factorial Design - Two-way InteractionsRun12345678ALowLowLowLowHighHighHighHighBLowLowHighHighLowLowHighHighCLowHighLowHighLowHighLowHighTempPressPowerIndependent VariablesResponse2.64.44.17.83.98.01.73.23.505.955.954.90Consider the AB interact
47、ion.Obtain the averages for each combinationProcess OptimizationProcess Characterization108Interaction PlotA = LowA = High3.03.54.04.55.05.56.06.5B = LowB = HighResponseThe difference in slopes indicates there is an AB interaction Process OptimizationProcess Characterization109Larger Screening Desig
48、nsSuppose we are considering 8 independent variablesFULL FACTORIAL 256 runsOverall Mean1Main Effects8A,B,C,D,E,F,G,HTwo-Way Interactions 28AB,AC,BC,BD,EF,EG, . . .Higher Order Interactions 219FRACTIONAL FACTORIAL 64 runsOverall Mean1Main Effects8A,B,C,D,E,F,G,HTwo-Way Interactions 28AB,AC,BC,BD,EF,E
49、G, . . .Higher Order Interactions 27MoreEfficientFractional Factorial designs are more efficient Resolution V DesignProcess OptimizationProcess Characterization110Resolution V Fractional Factorial Screening DesignsResolution V designs are a class of fractional factorial designs that provide estimate
50、s of the main effects and the two-way interactions, and minimize the number of runs required.Designs with lower resolution (III and IV) cannot provide clean estimates of the main effects and two-way interactions. Such designs are sometimes used to screen through a very large number of independent va
51、riables, and are usually followed by a resolution V Interaction experiment.Contact your local statistical resource before using resolution III or IV fractional factorial designs Process OptimizationProcess Characterization111Resolution V Fractional Factorial Screening DesignsLarger fractional factor
52、ial designs are even more efficient Process OptimizationProcess Characterization112Analysis of Factorial Screening ExperimentsCheck data for outliers and/or collection errors. Remove improper data values.If there are no Significant Effects: Consider increasing differences between factor levels. Take
53、 steps to decrease response measurement variation. Consider additional Factors.Experiment should have been replicated. Consult a Specialist.Use Normal Probability Plot of Effects to determine importance of EffectsUse Interaction Plot to determine effect of interacting factors. Factors involved in si
54、gnificant interactions cannot be analyzed using Main Effects plots. Use confidence intervals if possible.Use ANOVA table to determine importance of EffectsCalculate the natural log of the standard deviation for each treatment combination. Repeat the analysis using this response variable to understan
55、d and control variation.Use Main Effects plots to determine the effects of Levels for Significant Factors. Use confidence intervals if possible.Design is not balanced. Contact a specialist for analysis Are there extremely large or small data values?Are there more than two factors?Is every treatment
56、combination represented by more than one independent data value?Is every treatment combination represented by at least one data value?Consider the Interaction Effects first. Are Interactions Significant?Are Interaction Free Main Effects Significant?YesNoNoNoYesYesNoYesYesNoYesNoStartWrite ReportProc
57、ess OptimizationProcess Characterization113Analysis of Screening DesignsAre there Any Outliers?.Unusual data points?indication of measurement problems, incorrect logging of dataBest to watch for those while running an experiment!Is the experiment balanced?missing treatment combinations, missing repl
58、icatesWhich interactions are significant?always look for interactions before main effectsHow do the interactions influence the response variable?Which interaction-free main effects are significant?How do the interaction-free main effects influence the response variable?What is next?Process Character
59、ization114Optimize the Process - Response Surface ExperimentsOnce the list of independent variables has been trimmed to include only those that have a significant impact on the response variable . . . A response surface experiment can be used to find the levels of the controlled independent variable
60、s that provide an optimal response.Process OptimizationResponse Surface designs require more runs due to multiple levels.it is always best to screen factors carefully before running RS designs.Process Characterization115Response Surface DesignsThree-Level Factorial DesignsCentral Composite DesignsPr
61、ocess OptimizationProcess Characterization116Three-Level Factorial DesignsEach controlled independent variable is represented by three levelsProcess OptimizationProcess Characterization117Response Surface Experiment Design Efficiency3-LEVEL FULL FACTORIAL 81 runsOverall Mean1Main Effects4Two-Way Int
62、eractions 6Quadratic Effects 4Other 66CENTRAL COMPOSITE 26 runsOverall Mean1Main Effects4Two-Way Interactions 6Quadratic Effects 4Other 11 MoreEfficientCentral Composite designs are more efficient Process OptimizationProcess Characterization118Central Composite Designs3 Factor Central Composite Desi
63、gn8 Corner Points6 Star Points5 Center Points Use statistical software to design & analyze central composite designsProcess OptimizationProcess Characterization119Analysis of Response Surface ExperimentsUse regression analysis to estimate the coefficients for main, interaction, and quadratic effects
64、 as shown in the above equation.Use this mathematical model to find levels of the independent variables that result in an optimum response.This is typically done using three-dimensional plots.Process OptimizationY = A + B + C + AB + AC + BC + A + B + C222ccbbaabcacabcba Process Characterization1203-
65、D Response Surface PlotProcess Optimization280380480580680Power0306090120150Force-4.1-2.1-0.11.93.95.9Process Characterization121Confirm the Optimization ModelReserve enough resources (time and materials) to perform a confirmation run.The confirmation run will determine if the “optimal” levels of th
66、e independent variables actually improve the performance of the response variable.Test the impact of changes on the customer using environmental testing of samples from the confirmation run.Process Characterization122ReplicationReplication has occurred when each treatment combination is applied to m
67、ore than one independent experimental unit.An experimental unit is the smallest, most fundamental object to which the treatment combinations can be applied in a random order.4 Experimental Units24 measures on each16 Experimental Units6 measures on eachTreatment: Conformal CoatingBreak CouponsProcess
68、 OptimizationProcess Characterization123RandomizationRandomize Randomize RandomizeRandomization removes biasProcess OptimizationProcess Characterization124Process Analysis and ImprovementProcess Characterization125Process ControlPotential Failure Mode and Effects Analysis (FMEA)Control PlanPositrol
69、PlanControl ChartsOut-of-Control Action Plan (OCAP)Set-Up Check ListPreventative MaintenanceProceduresCheck ListScheduleLogReferences / InformationProcess ControlProcess Characterization126Potential Failure Mode and Effects AnalysisPotential Failure ModesPareto ChartsPotential Effects of the Failure
70、sPotential Causes of the FailuresCurrent Controls for detecting or reducing the failures Process ControlProcess Characterization127Potential Failure Mode and Effects AnalysisProcess ControlExisting ConditionsResultsProcess Description / PurposePotential Failure ModePotential Effect(s) of FailurePote
71、ntial Cause(s) of FailureCurrent ControlsSeverityOccurrenceDetectionR.P.N.Recommended Action(s) and StatusResponsible Area/Individual for Corrective ActionActions TakenSeverityOccurrenceDetectionRPNDeposit a uniform layer of phosphorus glass of desired thickness & compositionFilm thickness below spe
72、c limitEarly life failure of device due to moisture or contamination ingressWrong deposition time setTest wafer in each run82232 Control software enhancementNovellus Corp. 10/31/90Software installed81216Wrong recipe enteredTest water in each run2232 Control software enhancementNovellus Corp. 10/31/9
73、0Software installed81216Equipment malfunctionIf malfunction occurs, every wafer is checked2116 No action recommendedPhosphorus Concentration above spec limitPad corrosion in high humidity conditionsWrong phosphate flow rateCheck equipment at start-up & test wafer before each shift8118 No action reco
74、mmendedWrong recipe enteredFTIR test wafer in each process run2116 Control software enhancementNovellus Corp. 10/31/90Software installed8118Equipment malfuncitonFTIR test wafer in each process run2116 No action recommendedhigh hydrogen content in the film FTIREarly life failureImproper system pressu
75、re at depFTIR test wafer in each process run8118 No action recommendedImproper gas flowsFTIR test wafer in each process run118 No action recommendedImproper system temperature at depositionFTIR test wafer in each process run2116 No action recommendedProcess Characterization128Control PlanProcessEqui
76、pment used in the processSignificant Characteristics to be controlledCritical Independent VariablesResponse VariablesMeasurement Methods for collecting dataSample Sizes and Sampling FrequencyAnalysis Methods, Tools, and TechniquesIdentify . . .Process ControlProcess Characterization129Positrol PlanC
77、reate a matrix . . .Rows:What - The process variable to be controlledHow - How to perform the required controlWho - The person responsible for the specified actionsWhen - The frequency of monitoringType of Control - The type of control technique usedColumns:Set-up Requirements - Procedures to set-up
78、 a machine or processIndependent Variables - All important independent variables to be controlledResponse Variables - All important response variables to be controlledPreventive Maintenance (PM) - All PM requirementsProcess ControlProcess Characterization130Positrol PlanProcess ControlProcess: Scree
79、n PrintEquipment:SetupControlled Independent VariablesResponse VariablePreventive MaintenanceWhatSetup Check ListStencil LifeSqueegee PressureAir PressureNumber of StrokesAmbient TemperatureStencil ThicknessSolder HeightPMHowRefer to Setup Check ListCheck Cycle CounterPressure GagePressure GageSet C
80、ontrolsThermometer MicrometerVisionPM Check List and ProcedureWhoOperator Technician EngineerOperatorOperatorMaintenanceTechnicianMaintenanceOperatorOperatorTechnicianWhenPer Setup Check ListEvery HourEvery HourThree Times per ShiftStart of ShiftThree Times per ShiftEvery Hour Every BoardPer PM Sche
81、duleType of ControlSetup Check ListChartChartMaintenance Check LogScreen Print Check LogMaintenance Check LogChartChartPM Check ListProcess Characterization131Control ChartsControl Charts are trend charts that are used to provide real-time monitoring of processes. Decision rules are used to determin
82、e whether the process is currently in-control or out-of-control.The specific type of control chart used depends upon the kind of measurement data being recorded. A single reference should contain: ChartsProceduresDecision rulesOut-of-control action plansCorrective action logsProcess ControlProcess C
83、haracterization132Control Chart123456789 10 11 12 13 14 15 16 17 18 19 2078910111213UCLLCLProcess ControlProcess Characterization133Types of Control ChartsVariables MeasurementsGroups of Measurements: X-bar, Range, or Standard Deviation ChartsIndividual Measurements: Individual or Moving Range Chart
84、sAttribute MeasurementsFraction Defective: p-chartsNumber of Defective: np-chartsNumber of Defects per Subgroup: c-chartsAverage Number of Defects per Unit: u-chartsProcess ControlProcess Characterization134Control ChartUpper Control LimitLower Control Limit + 3 - 3 Process ControlProcess Characteri
85、zation135Out-of-Control Action PlansDecision Trees - Flow charts or other diagrams showing diagnostic methodology, containment procedures, and corrective actionsDescription of diagnostic methodology, containment procedures, and corrective actionsInteractive computer softwareProcess ControlProcess Ch
86、aracterization136Interactive Computer SoftwareSoftware should sound an alarm and then lead the operator through diagnostics, containment, and corrective actionsSoftware Requirements:Must use standard methods for calculating centerline, control limits, and capability indicesMust signal an alarm for c
87、orrective action, and verify and document corrective action before process is resumedMust have decision rules which conform to local requirementsMust record and recall corrective actionsDoes not automatically revise control limitsHas security to prevent unauthorized modification of data or control l
88、imitsProcess ControlProcess Characterization137Preventive MaintenanceProceduresChecklistScheduleLogsA PM Log should be kept with critical equipment Process ControlProcess Characterization138Process ControlProcess Characterization139DocumentationIdentify/Prioritize opportunities for improvementSelect
89、 the teamDefine objectives and goalsDescribe the total processAssess measurement system capabilityImprove measurement systemEvaluate process stabilityRemove assignable causesContinue to MonitorImplement/Enhance process control systemIdentify Significant Independent VariablesEvaluate Process Capabili
90、tyOptimize the ProcessConfirm the Optimization ModelRemove assignable causesIs measurement system capable?Is process stable?Do Cp & Cpk meet goal?Is process stable?Do Cp & Cpk meet next goal?YESNOYESNOYESNOYESNOYESNOSTART123456789121110131514DocumentationProcess Characterization140DocumentationMacro
91、 Maps - flow diagrams and listsPareto Charts, Quality Reports, Customer Reports, FMEA ReportsProcess Characterization ChampionProcess Champion / Team LeaderTeam SponsorList of Team MembersList of Consultants & ExpertsResponsibility AssignmentsDefinition of Objectives & GoalsSchedule of Activities 12
92、3DocumentationProcess Characterization141DocumentationMicro Maps - flow diagrams and listsFunctional CharacteristicCause & Effect DiagramsCross-Reference MatrixList of Independent Variables - descriptionsList of Response Variables - descriptionsCriticality Ranking%GR&Rs and P/T Ratios for all critic
93、al independent variables and all response variablesDocumentation of Measurement System Improvements 456DocumentationProcess Characterization142DocumentationTrend Charts w/ Specification LimitsInstability Index ValuesPassive Data Analysis ReportLong Term Stability Trend Charts - 30-60 daysDocumentati
94、on of Steps Taken to Find & Remove Assignable CausesCapability Metrics - Cp, CpkPercent of Response Variables w/ Metrics789DocumentationProcess Characterization143DocumentationDescription of Independent and Response Variables for Screening Screening Experiment Design / MethodologyScreening Design An
95、alysisANOVAInteraction PlotsMain Effects PlotsList of Critical Independent VariablesDescriptions of Independent and Response Variables for OptimizationOptimization Experiment Design / MethodologyOptimization Design AnalysisANOVA & Regression CoefficientsOptimization Model with Estimated Coefficients
96、3-D Surface Plots and/or Contour PlotsList of Optimal Levels of Independent Variables and Predicted Response Variable Levels1011DocumentationProcess Characterization144DocumentationDescription of Confirmation Run MethodologyComparison between Optimization Model Predictions and Confirmation Run Obser
97、ved ResultsFMEA Tables & DocumentationControl Plan DocumentationPositrol Plan Table & DocumentationControl ChartsProcedures, Check Lists, Decision RulesOut-of-Control Action PlansDecision Trees, Procedures for Containment and Corrective Actions, Corrective Action LogsPreventive MaintenanceProcedures
98、, Check Lists, Schedules, Logs1213DocumentationProcess Characterization145DocumentationDocumentation of Steps Taken to Find & Remove Assignable CausesSummary of Process Characterization ResultsReferencesOther Relevant InformationOngoing Log Books1415Number of Key Processes% of Key Characterized Processes % of Optimized Processes% of Key Processes under Variation ControlMonthlyDocumentationNumber of Key Variables% of Key Variables with Cpks 1.5Process Characterization146Documentation