《品质管理精华理论SupplierSummit070504.ppt》由会员分享，可在线阅读，更多相关《品质管理精华理论SupplierSummit070504.ppt（112页珍藏版）》请在三一办公上搜索。
1、,Understand the difference between Inspection and Variation Reduction(SPC and Process Capability)Review New Process Capability 10-Steps and ScorecardWhat help do you need from Dell?Questions and Answers,Objectives,Process Capability Summit,Kirk ChiDell Inc.WWP BPI Business ChampionDell Black BeltASQ
2、 Certified Six Sigma Black BeltASQ Certified Quality Engineer7/05/04,Traditional Economic Model of Quality of Conformance,Total cost,Cost due to nonconformance,Cost of quality assurance,“optimal level”of quality,100%,Quality improvement based on Inspection,Modern Economic Model of Quality of Conform
3、ance,Total cost,Cost due to nonconformance,Cost of quality assurance,100%,Quality improvement based on Variation Reduction(SPC/Process Capability),Non-Value Added operations result in:Higher procurement cost of productsHigher probability of defects,Improve the Process To Reduce Non-Value-Added Opera
4、tions,Dell,VLRR/PID/CND,Hidden Costs,What is the difference between quality control based on Inspection and Variation Reduction?Inspection refers to the manufacturing operations based on Attribute Data(Pass/Fail).Current manufacturing operations are focused mainly on Pass/Fail inspections.Much of va
5、riable data is measured,but the data is converted to attribute data for Pass/Fail inspection.,If you inspect 100%,will your customer experience no failure?If you inspect 100%,and inspect again 100%,and inspect again 100%,will your customer experience no failure?Can you and Dell achieve the reduction
6、 of FIR(and VFIR)based on inspection quality control?,We will work together to understand the following points:Even if you inspect 100%,your customer will still experience failures Inspection does not detect the process and product mean shift 100%Inspection does not reduce the variation in your proc
7、ess(and product),Traditional Inspection View,Lower Spec,Upper Spec,no loss,nominal,tolerance,Traditional view:There is no failure as long as a parameter is within specification,New View(Taguchi Loss Function),nominal,tolerance,Failure rate,$,Lower Spec,Upper Spec,no loss,New View:Products still fail
8、 in time even if a parameter is within spec.The probability of failure increases as the parameter shifts away from the mean,nominal,tolerance,Failure rate,$,Lower Spec,Upper Spec,no loss,Where do we want to go?,Critical Process Parameters(x),Product Attribute(y(x),Metric(Y),VLRR,Sigma Level(Drive fo
9、r 5),PAPB,123,Yield,Cpk,Metric,What Measured,Metric(Y),Sigma Level(Drive for 5),VIFIR+90 D VFIR,Key Message:Only the reduction of process and product parameter variations will lead to the reduction of VFIR,High Cost of Quality,Lowest Cost of Quality,Approach to Process Capability,Identifying Cause a
10、nd Effect,What is Process Capability?,Process Capability is the“Voice of Process”to the“Voice of Customer”.Process Capability study provides valuable insight on how well an existing process is performing with regards to customer requirements(specifications)what needs to be done to improve performanc
11、eCapability studies enable manufacturers to improveproductivity,reduce costs,and enhance their strategicadvantage over competitors.,Process Stability Before Capability,Control of a process must be achieved first,before any attempt is made to measure capability or estimate the percentage of nonconfor
12、ming product.When a process is stable,it is repeatable,well-defined,and predictable.,You can predict the outcome only if the process is stable,time,PREDICTABLE,?,UNPREDECTIBLE,Process Stability:Statistical Process Control,Statistical Process Control(SPC),A process output is considered stable when it
13、 consists of only common-cause variation.,Sources of Variation in Production Processes,Materials,Tools,Operators,Methods,Measurement Instruments,HumanInspectionPerformance,Environment,Machines,INPUTS,PROCESS,OUTPUTS,Common Causes,Special Causes,Two Fundamental Management Mistakes,Treating as a speci
14、al cause any fault,complaint,mistake,breakdown,accident or shortage when it actually is due to common causesAttributing to common causes any fault,complaint,mistake,breakdown,accident or shortage when it actually is due to a special cause,Control Chart,Focuses attention on detecting and monitoring p
15、rocess variation over timeDistinguishes special from common causes of variationServes as a tool for on-going controlProvides a common language for discussion process performance,*,*,Commonly Used Control Charts,Variables datax-bar and R-chartsx-bar and s-chartsCharts for individuals(x-charts)Attribu
16、te dataFor“defectives”(p-chart,np-chart)For“defects”(c-chart,u-chart),Statistical Process Control(SPC),A methodology for monitoring a process to identify special causes of variation and signal the need to take corrective action when appropriate,Shift in Process Average,Identifying Potential Shifts,C
17、ycles,Trend,Quantitative Comparison ofTraditional Inspection vs.Control Chart,Review of Variable-Data Control Charts,+3,-3,+1,+2,-1,-2,+/-1=68%+/-2=95%+/-3=99.73%,Individual measurement distribution,SPC Control Limits,x-bar,x-bar+3X-bar,x-bar=/,UCL(Upper Control Limit)=X(double bar)+3X-barLCL(Lower
18、Control Limit)=X(double bar)-3X-bar,Sample mean distribution,Distribution of X vs.Distribution of Sample Means(X-bar),+3,-3,x-bar,x-bar+3X-bar,x-bar=/,Individual measurement distribution,Sample mean distribution,There is no relationship.Spec limits are determined by engineering based on customer req
19、uirements Control limits are determined by the common-cause variation in the process Spec limits are typically wider than control limits,Spec Limit vs.Control Limit,+3,-3,Control Chart:Example:Control chart is monitored by UCL and LCL.(Note:USL and UCL do not have any relation.),USL,LSL,x-bar,x-bar+
20、3X-bar,UCL,LCL,*If=2 for Inspection of individual parts,of subgroup size of 4 samples=1.,Why Control Charts Are Better Than Inspection?,+3,-3,Inspection:Example:One part is checked each hour.One part is checked every 20.Part passes the inspection if the part measurement is within spec.,USL,LSL,*Assu
21、mption:spec limit lis+/-3 sigma.,+3,-3,What happens to Inspection if process mean shifts 3 sigma?,USL,LSL,Anything above USL are rejected.,50%,Because half of the pieces are still within spec,there is 50%chance of selecting such a part and mistakenly deciding to continue running this modified proces
22、s.,What happens to Control Chart if process mean shifts 3sigma?,x-bar,UCL,LCL,*If=2 for Inspection of individual parts,of subgroup size of 4 samples=1.,99.865%,Only 0.135%of the subgroup averages would fall within the control limits.Conversely,99.865%will be above UCL.There is almost 100%chance this
23、 shift in the process mean will be detected.,+3 shift,Power of X-bar Chart to Detect Process Changes,Producers Risk(Alpha error):Rejecting a good partAlpha error measure the probability of rejecting good parts in the factory.Consumers Risk(Beta error):Shipping bad partsBeta error measures the probab
24、ility of shipping bad parts to customers,(1-),Probability of detecting a process shift is higher with larger n,i.e.Beta error(consumers risk)decreases as n increases.,Process Capability,Process Capability is defined as the ability of a process to satisfy customer expectations.Because the specificati
25、on limits are assumed to reflect customer desires,capability measures are said to relate the“Voice of the Process”to the“Voice of the Customer”,Process Capability,Process Capability Index,Cp=,USL-LSL 6s,Cpl,Cpu,USL-m 3s,Cpl=,m-LSL 3s,Cpk=min,Cpu=,Process Capability,Process Capability Index,CpCp 1.33
26、,CapableCp=1.00 1.33,Capable with tight controlCp 1.00,Incapable,Process Variation,Specifications,Process Capability,Cp=1 and Cpk=1,LSL Target USL,LSL Target USL,Cp=2 and Cpk=1,LSL Target USL,Cp=2 and Cpk=0,New Process Capability 10-Steps,1.Characterize the entire manufacturing process.Process Map t
27、he entire manufacturing operations from the beginningto the end.Later when the critical parameters are identified,locatewhich processes are aligned with the critical parameters.,Initiate an attribute control chart(P-chart)on Rolling-Through-Put Yield.If a supplier is sensitive to sharing the yield d
28、ata,use the coded data.The goal is to monitor the stability of RTY.The goal of control chart is to identify any special-causes.Identify the root causes and take corrective actions immediately for any out-of-control points.Does the supplier have the documented process on P-chart,business owner,and ac
29、tion steps for correcting out-of-control points?After CPKs of critical characteristics are improved,check if there is any impact to RTY.,3.Identify the critical parameters.This is the most important step.Critical parameters can be identified in many different ways.Critical parameters should be tied
30、to customer experience.Critical parameters can be both Process and Product characteristics.One way to identify the critical parameters is to analyze the nonconforming defectives(fall outs)from supplier manufacturing line,customer manufacturing line,and customer fields.Pareto,FMEA and Cause-and-Effec
31、t diagram are common tools to use for the analysis.,3.Identify the critical parameters(continued)Another way of identifying critical parameters is to get Engineering(Design and New Product Introduction)engaged to identify key critical Product characteristics that impact customer experience.Identifyi
32、ng both Process and Product critical parameters is desirable.Identifying the critical parameters from both defective analysis and engineering involvement is desirable.,Review the specifications of the critical characteristics.Specifications should be defined based on customer requirements.Does the s
33、upplier have the process to define specs based on engineering and tolerance analyses?Or does the supplier define some specs arbitrarily?Does the supplier have the process to tighten up specs or do they maintain the same spec always?The value of process capability will be dependent on the specificati
34、ons.Quite often,very unreasonably high CPKs are reported because of very wide spec ranges.If the spec is too wide,products may pass through suppliers manufacturing process,but products will not conform to customers expectations.Define the required CPKs for the critical characteristics.,5.Conduct Mea
35、surement System Analysis.The goal of running Gage R&R is to check if the data collection is notcontaminated by the inaccuracy and non-repeatability of themeasuring devices and operators.,Implement X-bar and R variable control charts on the critical parameters.Determine the subgroup sample size and t
36、he frequency of sampling.Preferable subgroup size is n=10,in order to have 95%confidence indetecting 1.5 sigma process shift.However,if the circumstances donot allow n=10,use smaller subgroup size with more frequentsampling.Use the Beta-error curves to figure out the probability ofdetecting the proc
37、ess shift.The goal is not to collect the data.The goal is to identify specialcauses using the control charts.Once a special cause is identified,root cause analysis and corrective action need to follow upimmediately.,6.ContinuedHave passion for identifying special and common causes and driveactions f
38、or the continuous improvement.The process can be declared to be Stable if 25 points(with subgroupsample size)are plotted consecutively within the control limits.Does the supplier have the documented process on X-bar and R charts,business owner,and action steps for correcting out-of-control points?,A
39、fter the process is stable,calculate CP and CPK.Use the same X-bar and R chart template to calculate CP and CPK.Check the normality of the data first.What is CP?What is CPK?Does calculated CPK meet the required value?,If Cp is not same as CPK,plan the actions to make CP=CPK.If CPK is below 1.33,plan
40、 the actions to raise CPK to 1.33.,Using the attached control plan is optional.Modify the format to fit your needs.,Conduct statistical correlation between supplier critical parameters and Dell quality metrics.Do you find any correlations?Conduct the correlation analysis between critical parameters
41、and supplier RTY.Are there any correlations?If there is no correlation,are we missing the right critical parameter that impact customer experience?Work on the action plans on identifying the critical parameters that impact customer experience.,Implement Process Capability with sub-tier suppliers.Mov
42、e upstream.Identify the core suppliers that impact customer experience.Implement the same Process Capability program with the coresuppliers.,New Scoring Criteria and Scorecard,New Scorecard Criteria,Six-Sigma Quality,Ensuring that process variation is half the design tolerance(Cp=2.0)while allowing
43、the mean to shift as much as 1.5 standard deviations.,Motorola Definition of Six Sigma,Defect rate goal is less than 3.4 DPPM(with 1.5 sigma shift)CP goal=2CPK goal=1.5,Six Sigma Metrics with 1.5 Sigma Dynamic Shift,Six Sigma Metrics with 1.5 Sigma Dynamic Process Shift,(1-),Use this plot to determi
44、ne what subgroup size to use for detecting the process shift,Calculation of System Level Yield,Calculate based on lower limit CPKs.If your factory has:10 processes with CPK=1.33(31 dppm)-20 processes with CPK=1(1350 dppm)30 processes with CPK=0.83(6200 dppm)Predicted system level yield=(1-.000031)10
45、*(1-.00135)20*(1-.0062)30=.99969*.97334*.8297=.807,What if suppliers CPK is high and their manufacturing failure rate is low(i.e.high yield),but Dell line and field failure rates are high?This shows that suppliers specifications are set too wide.Therefore,nonconforming products(per suppliers require
46、ments)are passing through suppliers final tests.,How is Process Capability analysis going to help Lead Free transition?Can compare the process capability of before and after the transition Predict the failure rate based on spec change and the existing variation,Back Up Slides,FPY,RTY,and NAY,FPY(YFP
47、),First Pass Yield or FTY(YFT),First Time YieldGives the probability of going through only one step in the process with zero defectsNumber of good units produced the first time through the process divided by the total number of unitsRTY(YRT)or Rolled Throughput Yield Probability of a unit passing th
48、rough each step of a process defect-freeRTY=FPY1*FPY2*FPY3NAY(YNA),Normalized Average YieldUsed to characterize a process using the z normal tablesUsed to compare two different processesNAY=RTY 1/number of opportunities or stepsUse NAY to compare processes with different numbers of steps or opportun
49、ities,P Chart Plot First Pass Yield Fractions(Attribute Data),Pareto Analysis to Identify Major Defect Categories,Work On 80%Issues,Forget,Use 80/20 rule to focus on major issues,Cause and Effect Diagram,Enables a team to focus on the content of a problem,not on the history of the problem or differi
50、ng personal interests of team membersCreates a snapshot of collective knowledge and consensus of a team;builds support for solutionsFocuses the team on causes,not symptoms,Effect,Cause,FMEA What is it?,Failure mode effects analysis(FMEA):A technique to identify,define,and eliminate known and/or pote
