1、1By Strive ZhouCapability Analysis能力分析2By Strive ZhouProcess Improvement Methodology 制程改善方法论制程改善方法论 3By Strive ZhouWhat is Capability 制程能力是制程能力是 LSLUSLq When process is under control,capability is decided by customer demand and process performance(product or service quality shifting degree),The more
2、 the process meets the customer need,the better the capability will be.制程在受控状态下时，客户要求与制程表现（产品品质或服务的品质变动程度）的比值，如果制程表现越能满足客户要求，则制程能力越充分，反之则不足。4By Strive ZhouWhat is Capability 制程能力是制程能力是 LSLUSLWhat can we conclude form the capability of these process?这些制程的制程能力这些制程的制程能力如何？如何？ALSLUSLCLSLUSLBLSLUSLD5By S
3、trive ZhouHow Capable Is It?制程能力如何？制程能力如何？LSLUSLCapable Process制程能力充分制程能力充分6 LSLUSLA VeryCapable Process制程能力很充分制程能力很充分6 LSLUSLA Very,VeryCapable Process制程能力非常充分制程能力非常充分6 6By Strive ZhouHow Capable Is It?制程能力如何？制程能力如何？Capable Process制程能力充分制程能力充分A VeryCapable Process制程能力很充分制程能力很充分A Very,VeryCapable Pr
4、ocess制程能力非常充分制程能力非常充分LSLUSL3 3 LSLUSL4 4 LSLUSL6 6 =0.08Standard Deviation 标准差Sigma Level Sigma 水平3 Process=0.064 Process=0.046 Process7By Strive ZhouCustomer Requirement and capability能力与客户要求能力与客户要求 8By Strive ZhouWhat brings defects?什么导致缺陷？什么导致缺陷？Unstable incoming material 不稳定的原材料不稳定的原材料Poor proce
5、ssCapability 不充分的制程能力不充分的制程能力Inappropriate specification design 不适当的规格设计不适当的规格设计9By Strive ZhouCapability Analysis制程能力分析制程能力分析 10By Strive ZhouProcess capability analysis using different Types of data 使用不同类别数据进行制程能力分析使用不同类别数据进行制程能力分析 Variable 计量型Attribute记数型11By Strive ZhouProcess Sampling short ter
6、m过程抽样短期过程抽样短期 12By Strive ZhouProcess Sampling short term过程抽样短期过程抽样短期 13By Strive ZhouProcess Sampling Short Term/Long Term过程抽样短期过程抽样短期/长期长期 14By Strive ZhouProcess Sampling Long Term过程抽样过程抽样/长期长期 15By Strive ZhouProcess Sampling Long Term过程抽样过程抽样/长期长期 v From many shifts of production 从许多生产班次而来v Obs
7、erving many machines 使用许多机器v Analyzing the performance of many operators 许多操作者v Using many lots of incoming parts 使用多批原材料16By Strive ZhouProcess Sampling Long Term过程抽样过程抽样/长期长期 17By Strive ZhouLong-term and Short term Capability长期长期/短期制程能力短期制程能力 LSL 规格下限规格下限USL 规格上限规格上限Short-termLong term18By Strive
8、 ZhouLong-term and Short term Capability长期长期/短期制程能力短期制程能力 19By Strive ZhouLong-term and short-term capability(Z-shift)长期和短期能力（长期和短期能力（Z偏差）偏差）Normally,the difference between long-term and short-term Sigma Level is the 1.5 shift.通常，长期与短期的Sigma 水平1.5 20By Strive ZhouLong-term and short-term capability(
9、Z-shift)长期和短期能力（长期和短期能力（Z偏差）偏差）21By Strive ZhouLong-term and short-term capability(Z-shift)长期和短期能力（长期和短期能力（Z偏差）偏差）Sigma Level Sigma 水平水平Short-term DPMO 短期的短期的DPMOLong-Term DPMO 长期的长期的DPMO1158655.3691462.5222750.1308537.53135066807.2431.76209.750.3232.760.03.422By Strive ZhouTypes of Capability Analy
10、sis 能力分析的类型能力分析的类型23By Strive ZhouTypes of Capability Analysis 能力分析的类型能力分析的类型24By Strive ZhouCapability Issues能力相关的问题能力相关的问题25By Strive ZhouProcess Capability Variable Data制程能力计量型数据制程能力计量型数据Frequency26By Strive ZhouVariable Output:The Method计量输出：方法计量输出：方法27By Strive ZhouStep 1:Check Specification步骤步
11、骤1：规格确认：规格确认 Are the specifications correct for the parameter(Y)(output variable)?参数Y的规格正确吗？（过程输出变量）Is the parameter(Y)centerbetween the lower and upper limits?参数Y是否在规格上下限之间的中心点？To ascertain true capability,we must use a true specification for calculation.要确定真实的能力，真实的规格是必要的28By Strive ZhouExample 举例
12、举例Pen.mtv29By Strive ZhouStep 1:Check Specification步骤步骤1：规格确认：规格确认30By Strive ZhouStep 2:Sample Data-Rational Grouping步骤步骤2：采集数据合理分组：采集数据合理分组 Data collected should be able to reflect term and long term performance(if possible)数据采集应能捕捉过程的短期绩效和长期绩效（若有可能）Data can be collected within a short period of t
13、ime and arranged in a time sequence.可通过快速采集一系列数据来完成，这些数据按照时间序列排列；q Data collected from Rational Grouping Samples.快速凭票采集的数据应从合理分组中抽样0123456789O oooo oO oo o o oO o oo ooO o o oo oO oo oo oO oo o o oO o o oO o o oo o31By Strive ZhouStep 2:Sample Data-Rational Grouping步骤步骤2：采集数据合理分组：采集数据合理分组32By Strive
14、 ZhouStep 2:Sampling步骤步骤2：取样：取样 Eg.The engineer takes 5 data a day for consecutively 7 days.例：工程师每天采集5次数据,连续7天进行以上数据采集工作 Collected data entered into file PEN.mtw 采集的数据放在PEN.Mtw 文件中；Pen.mtv33By Strive ZhouAnalysis short-term and long-term variation分析短期和长期波动分析短期和长期波动34By Strive ZhouShort-term or Long-
15、term短期还是长期？短期还是长期？Rule:If 80%of the inputs vary within their natural limit,then the output data can be considered long term.准则：如果80%的输入是在它们自然的范围中起伏，则输出的数据可被认为是长期的；35By Strive Zhou Another may is though observation:另外一个方法可用来观察Short-term or Long-term短期还是长期？短期还是长期？21121211yYyYnyYijmikjikjjijmikj36By St
16、rive ZhouShort-term or Long-term短期还是长期？短期还是长期？Opportunity 机会机会37By Strive ZhouShort-term or Long-term短期还是长期？短期还是长期？38By Strive ZhouShort-term or Long-term短期还是长期？短期还是长期？39By Strive ZhouShort-term or Long-term短期还是长期？短期还是长期？40By Strive ZhouExample-Short Term Shift短期波动的例子短期波动的例子v Calculate short term sh
17、ift(with Minitab)使用Minitab来计算短期波动v StatQuality ToolsCapability analysis(Normal)41By Strive ZhouExample-Short Term Shift短期波动的例子短期波动的例子42By Strive ZhouExample-Short Term Shift短期波动的例子短期波动的例子v Calculate short term shift(with Minitab)使用Minitab来计算短期波动v StatQuality ToolsCapability analysis(Normal)43By Stri
18、ve ZhouStep3:Calculate Zu and Z1步骤步骤3：计算：计算Zu和和Z127.249.061.1245.12386.349.061.1245.126termshortlsltermshortuslsxlslzsxuslz44By Strive ZhouStep4:Calculate Cp 步骤步骤4：计算：计算Cp制程的容差（散布）产品公差制程的表现客户的要求ToleranceocessToleranceoductCLSLUSLCePerformancocessquiremetnsCustomerCpppPrPr6PrRe45By Strive ZhouStep4:C
19、alculate Cp 步骤步骤4：计算：计算Cp Potential Capability(Cp)过程潜在的能力(Cp):46By Strive ZhouStep4:Cp and Entitlement步骤步骤4：Cp与潜在最佳值（与潜在最佳值（entitlement)47By Strive ZhouStep4:Pen Length:Cp 步骤步骤4：钢笔长度的：钢笔长度的Cp LSL=123.5 USL=126.5=124.61 =0.49stpLSLUSLToleranceocessToleranceoductC6PrPr02.194.2349.065.1235.126stX48By S
20、trive ZhouStep4:Minitab Result 计算结果计算结果49By Strive ZhouStep4:A Question About CP步骤步骤4：一个关于：一个关于CP的问题的问题v Which process is better?哪个过程更好？v What is the difference between these two Cps 这两个过程中Cp有什么差异v What needs to be done in order to use Cp as a more effective indicator for process capability?要使得Cp作更加
21、有效地作为过程能力的统计指标，还需要做些什么？50By Strive ZhouStep4:Process Performance 步骤步骤4：认识：认识 Cpk过程绩效过程绩效51By Strive ZhouStep4:Pen Length:Cpk 步骤步骤4：钢笔长度的：钢笔长度的Cpk02.149.0*65.1235.126*6termshortpSLSLUSLC755.049.0*36.1245.126,49.0*35.1236.124min*3,*3minststpkSXUSLSLSLXC52By Strive ZhouStep4:Minitab Output 步骤步骤4：Minita
22、b 计算结果计算结果53By Strive ZhouStep4:Calculating Sigma Level 步骤步骤4：计算：计算Sigma 水平（水平（Zst）54By Strive ZhouStep4:Calculating Sigma Level 步骤步骤4：计算：计算Sigma 水平（水平（Zst）Estimate short-term defect rate 估计短期缺陷率Defects 缺陷缺陷Defects 缺陷缺陷55By Strive ZhouStep4:Calculating Probability Using Excel 步骤步骤4：用：用Excel 来计算概率来计算
23、概率P(x126.5)=1-0.99994=0.00006 Total defects=0.0117+0.00006 =0.0118该例中：该例中：Mean=124.61，Sigma=0.49，LSL=123.5，USL=126.51.用NORMDIST函数计算比例不一致的P值56By Strive ZhouStep4:Calculating Sigma Level Using Excel 步骤步骤4：用：用Excel 来计算来计算Sigma 水平水平2.用NORMSINV函数计算比例不一致的Z值Zshort-term=2.26Defects(0.0118)Z=2.2657By Strive
24、ZhouStep4:Sigma Level-Example(Using Minitab)步骤步骤4：Sigma 水平水平-使用使用Minitab 的实例的实例 使用Minitab:Calc-Probability Distributions-NormalTotal=0.0117+0.00006=0.0118=11,800DPMM58By Strive ZhouStep4:Sigma Level-Example(Using Minitab)步骤步骤4：Sigma 水平水平-使用使用Minitab 的实例的实例 计算 Sigma 值(Zst)59By Strive ZhouStep4:Minita
25、b Output步骤步骤4:Minitab 的计算结果的计算结果60By Strive ZhouLong Term长期Step4:Calculating Zit步骤步骤4:计算计算ZitShort Term 短期Short Term 短期Short Term 短期Short Term 短期61By Strive ZhouStep4:1.5shift-from ZST to ZLT步骤步骤4:1.5 漂移漂移-从从ZST到到ZLT62By Strive ZhouStep4:Estimate Long-term Capability步骤步骤4:预测长期能力预测长期能力PpkItItpkststpk
26、LSLUSLPLSLUSLC3,3min3,3min Replace short-term shift with long-term capability 用长期能力代替短期波动63By Strive ZhouStep4:Estimate Long-term Capability步骤步骤4:预测长期能力预测长期能力Ppk25.0376.3*376.05.126.25.1LTpkLTLTLTstLTZPPpkionSpecificatZZZZ64By Strive ZhouStep4:Minitab Output步骤步骤4:Minitab 的计算结果的计算结果Minitab 人为长期波动可通过组
27、间的波动来预测人为长期波动可通过组间的波动来预测65By Strive ZhouFormulae for Capability Calculation能力计算公式能力计算公式66By Strive ZhouExercise 1:Computing Process Capability过程能力计算练习过程能力计算练习1 A technician inherits the sterilization process.Because sterility is critical,each run is evaluated by tracking chamber exposure temperature
28、.Components from failed runs are reprocessed.某工程师欲分析杀菌过程。分析方法是监控每批产品杀菌室温度，所有不合格产品会重新返工。According to yield data,115%of the runs have been scrapped during the previous nine months 良率数据显示过去9个月有11.5%批的产品67By Strive ZhouCalculating PPM and Z-Score计算计算PPM和和Z值值PPM=Fraction Defective(缺陷比率）X 1,000,000 =0.115
29、 X 1,000,000=115,000Z-Score:For worst case,assume all defects are in one tail.So,P(z?)=0.115 Z值：假定所有缺陷发生在左或右的一尾。值：假定所有缺陷发生在左或右的一尾。What is the z that P(z?)=1-0.115?P(z?)=1-0.115,那么那么Z为？为？P(z1.2)=0.885Confirm using Minitab 用用Minitab计算计算Zh=1.2 查查Z值表得出的结果是否一致值表得出的结果是否一致?68By Strive ZhouCalculating Ppk,Z
30、st and Cpk计算计算Ppk,Zst 和和Cpk40.03/2.13/ITpkZP7.25.12.15.1termLongtermshortZZ90.03/7.23/STpkZC69By Strive ZhouAttribute Process Capability计数型制程能务计数型制程能务Variable data is difficult to find!寻找计量型数据太难了！寻找计量型数据太难了！What to do with attribute data?遇上记数型数据该怎么遇上记数型数据该怎么办？办？70By Strive ZhouCp,Cpk,Pp,Ppk(using Mi
31、nitab)Cp,Cpk,Pp,Ppk (使用使用 Minitab)Minitab71By Strive ZhouQualitative Y(attribute data)Y的定性分析的定性分析(计数型数据）计数型数据）Capability Determination for an attribute output:对计数型输出的能力确定：72By Strive ZhouQualitative Y(attribute data)Y的定性分析的定性分析(计数型数据）计数型数据）73By Strive ZhouQualitative Y(attribute data)Y的定性分析的定性分析(计数型
32、数据）计数型数据）74By Strive Zhou75By Strive ZhouStep1:Tally Defect Data步骤步骤1:缺陷数据计数缺陷数据计数76By Strive ZhouSteps3 Calculate Z-score步骤步骤3：计算：计算Z值值Step 2(步骤步骤2):77By Strive ZhouSteps3 Calculate Z-score步骤步骤3：计算：计算Z值值Step 3(步骤步骤3):78By Strive ZhouSteps3 Calculate Z-score步骤步骤3：计算：计算Z值值79By Strive ZhouSteps4:Calculate Cpk,Ppk步骤步骤4：计算：计算Cpk、Ppk 值值Step 4(步骤步骤4):80By Strive ZhouCalculating Defect Levels计算缺陷水平计算缺陷水平81By Strive Zhou
