SPC基本概念介绍.pptx

《SPC基本概念介绍.pptx》由会员分享，可在线阅读，更多相关《SPC基本概念介绍.pptx（57页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、Sinderella LiuSPC ConceptsSinderella Liu- Fundamental SPC Concept- Shewhart Control Chart - Introduction- Average and Range Control Chart- Average and Moving Range Control Chart-Western Rules - Cp/Cpk Introduction - Real AppliedSinderella LiuTIME 精神 產業型態 1920 - QI 檢驗出來 -檢驗員 大多停留在手工製造 1940 QC 製造出來 -統

2、計的品管 開始有生產線之形成 1950 - QA 設計出來 -品質保證 大戰-需高可靠度武器 1970 - TQC 管理出來 -全面品質管制 1980 - TQA 習慣出來 -全面品質保證 企業開始強調整體形象 Sinderella LiuSTATISTICALPROCESSCONTROLA structured mathematical method of analyzing data or numbersA set of conditions or causes which work together to produce a given resultTo make a process b

3、ehave the way we want it to behaveSinderella LiuVarious Definitions of SPC: Controlling Processes using Statistical Techniques or Methods It is the use of Statistical Control Charts to monitor and control manufacturing operations in real time It is the use of Control Charts & Statistical Techniques

4、to prevent a process from going out of control which affects the product yield as well as quality/reliability levels It is the elimination of unnatural variance to improve the process Special Cause Sinderella LiuSPC is a method of using statistics to identify patterns and trends in order to control

5、the process DataCollectionProcessStatisticalMethodAbnormalStableProcessAnalysisActionsContinuous ImprovementSinderella Liu Improve Quality/Yields Increase Product Life and Reliability Greater Customer Satisfaction Government & Customer Requirement To be More ProfitableSinderella Liu Improve Quality

6、by removing the causes of problems in the system . inevitably . leads to improved productivity The person doing the job is most knowledgeable about the job People want to be involved and do their jobs well Every person wants to feel like a valued contributorSinderella Liu A structured problem solvin

7、g process using statistical methods and graphical techniques procedures better solutions than in an unstructured process Graphical problem solving techniques let you know: the relative importance of problems to be solved whether the changes made have had the desired impact Adversarial relationship b

8、etween labor & management is counterproductive & outmoded Every Organization has undiscovered “gems” waiting to be developedSinderella LiuProduct DesignProcessCustomer needsQuality ResponseNominal valueConceptual Framework for SPCThe origins and natural of variabilitySinderella LiuConceptual Framewo

9、rk for SPCThe origins and natural of variabilityManufacturing Process Raw Material Tool ConditionQuality ResponseSinderella LiuConceptual Framework for SPCThe origins and natural of variabilityProduct use in the field Temperature PowerQuality ResponseSinderella LiuConceptual Framework for SPCThe ori

10、gins and natural of variabilityProduct use Over Time Wear AgingQuality ResponseSinderella LiuProcess Evaluation Over TimeAccuracy & PrecisionAccuracy is how well numbers center on your target.Precision is how alike the numbers are.AccuratePreciseYesNoYesNo*Sinderella LiuProcess Evaluation Over TimeA

11、ccuracy & PrecisionTargetTargetHigh Accuracy & high PrecisionLow Accuracy & high PrecisionHigh Accuracy & low PrecisionLow Accuracy & low PrecisionSinderella LiuProcess Evaluation Over TimeTimeProcess in statistical controlone unique causal systemSinderella LiuProcess Evaluation Over TimeTimeProcess

12、 in transitionin search of a causal systemSinderella LiuProcess Evaluation Over TimeTimeProcess not in statistical controlpopulation changes over timeSinderella LiuOutput DataOutput DataOutput DataOutput DataOutput Datat1t3t4t5t2Process MeanUpper Control LimitLower Control Limit 將製程之品質特性以圖形之方式表示將製程之

13、品質特性以圖形之方式表示，把量測所得之資料點繪製把量測所得之資料點繪製於以時間為橫軸之圖形上於以時間為橫軸之圖形上。管制圖必須包括中心線管制圖必須包括中心線(CenterLine) 、管管制界限制界限(Control Limit) ，並應用管制規則並應用管制規則(Control Rules)判斷之判斷之。Sinderella LiuLCLUCLCLSinderella LiuUSLLSLUCLLCLUSLLSLUCLLCLUSLLSLUCLLCLUCLLCLIn SpecificationOut of SpecificationIn controlOut of controlUSLLSLSin

14、derella LiuAll processes are subject to two fundamental types of faults or problems. They are:類別機遇原因(一般原因)Chance Cause(Common Cause)可歸屬原因(特殊原因)Assignable Cause(Special Cause)特性 1.包括許多隨機存在之個別原因。 2.任何單一機遇原因僅導致微量變異(但若許多機遇原因彙總在一起，可能產生頗大之影響)。 1.包含一個或少數偶爾發生之原因。 2. 任何單一可歸屬原因均可造成大量之製程變異。可能原因機器之微震原料之略微差異作業員之

15、錯誤機器不正確之設定不合格原料 解說1. 消除製程中之機遇變異不符合經濟原則。2. 當僅有機遇變異出現時，製程處於可接受水準；倘若仍有不合格品產生，則必需進行基本製程改變或修定規格，以減少不合格品。3. 當觀測值在管制界限內時，表示製程不應調整。4.當僅有機遇變異時，製程相同穩定，可以用抽樣程序預測產品品質。 1. 可歸屬變異可以被偵測出；消除該變異通常均合乎經濟原則。2. 當有可歸屬原因出現時，製程未在可接受水準。3. 當觀測值超出管制界限時，通常表示製程應予以調整和矯正。4.當有可歸屬原因出現時，製程不夠穩定，不宜以抽樣程序預測產品品質。 Sinderella Liu DEFINITION

16、:Its a snapshot of your process over time providing further insight as to the consistency and quality of that process. PURPOSE:A Control Chart is a tool which allows you to determine if the “Process” has changed.The upper and lower control limits(which are set statistically) further allow you to det

17、ermine, with a degree of certainty, that the change is a real change; i.e., statistically significant change.Sinderella Liu Shewhart Control Chart Thinking:In the context of the Shewhart control chart method what we do is collect an amount of data (say, k samples of size n each) that we think is suf

18、ficient to allow us to treat the average of all these data as if it were the actual process mean. Control Charts is introduced to Industry by Walter Shewhart in the 1920sSinderella Liu 計量計量值值 (Variable Data / Measurable Data) 種類適用狀況 平均值與全距管制圖可用以管制分組之計量數據，即每次同時取得幾組數據之工程，如長度、濃度、成份等是掌握工程狀態最有效之管制圖 平均值與標

19、準差管制圖與 相同，用於 subgroup size10時 中位數與全距管制圖與 相同，但檢出力較差 個別值與移動全距管制圖適用於找出不同subgroup間之變化時 ChartXRChartXSChartXRChartRX ChartXRChartmRX Sinderella Liu種類適用狀況 P-Chart 不良率管制圖 其數據有兩種類別，如合格/不合格、通過/不通過等 NP-Chart 不良數管制圖與 P-Chart 相同，每組樣本需相同 C-Chart 缺點數管制圖每單位數內缺點發生之次數 U-Chart 單位缺點數管制圖與 C-Chart 相同，但樣本大小不同 計數計數值值 (Att

20、ribute Data / Countable Data) 管制圖之選擇方式管制圖之選擇方式： 請參閱請參閱 QS9000 SPC Manual, Appendix CQS9000 SPC Manual, Appendix C Sinderella LiuData CollectionRational samples (unbiased and random)Groups measurementVariation is attributable only to a constant system of common cause. Minimize the occurrence of speci

21、al causes within sample Calculation Necessary ItemsControl limit (2025 subgroups of data are required)Central lineEstablish Control ChartProcess monitorAlarm systemDetect special cause Sinderella LiuThere are two process characteristics that are general interest in the statistical process control: T

22、he mean level of the processThe amount of variation in the processAverage ChartRange ChartTo interpret the patterns in the control chart Always Begin with R Chart X - R ChartSinderella LiuSample Size: nTotal : k Subgroups Run A Run BRun CRun DRun ERun FRun G Run HRun IRun J Run K Run L2.772.123.681.

23、62.363.332.813.33.253.575.171.982.424.172.661.631.93.493.843.073.542.394.672.112.812.532.551.842.223.522.413.542.222.23.492.332.832.442.742.121.873.133.292.742.152.223.853.443.031.852.563.033.184.6723.162.441.783.821.812.691.73.12.883.982.582.743.132.591.893.381.932.371.92.362.032.22.812.753.022.973

24、.163.161.814.561.733.191.893.252.944.462.972.752.123.841.82Sinderella LiuX Chart : Step 1: Calculate Mean and Range of each Subgroup Where: Mean is the average value of each subgroupRange is the difference between Max and Minvalue in each subgroupMeanRangeRun A Run B Run C Run D Run E Run F Run G Ru

25、n HRun IRun J Run K Run L2.942.312.862.132.623.313.043.122.742.423.922.152.192.471.321.432.112.092.460.801.391.792.011.63Sinderella LiuX Chart : Step 2:Calculate Centerline 、 Upper and Lower control limit as follows:Centerline:kxxkii1Where: k is the number of subgroups is the mean value of ith subgr

26、oup ixxis the arithmetic average of all available sample average Sinderella LiuX Chart : Upper and Lower control limitUpper Control Limit:RAxUCLx2Lower Control Limit:RAxLCLx2Where:R is the average of the sample rangesA is a constant which varying sample size n and tabulated2Sinderella LiuR Chart : k

27、RRkii1Where: k is the number of subgroups is the Range of ith subgroup iRRis the arithmetic average of all available sample range Calculate Centerline 、 Upper and Lower control limit as follows:Centerline:Sinderella LiuR Chart : Upper and Lower control limitUpper Control Limit:Lower Control Limit:Wh

28、ere:RDUCLR4RDLCLR3R is the average of the sample rangesD and D is a constant which varying sample size n and tabulated 3 4Sinderella Liu 23.7601.8800.0003.2671.1280.0003.2672.6601.88032.3941.0230.0002.5681.6930.0002.5751.7721.18741.8800.7290.0002.2662.0590.0002.2821.4570.79651.5960.5770.0002.0892.32

29、60.0002.1151.2900.691 61.4100.4830.0301.9702.5340.0002.0041.1840.54971.2770.4190.1181.8822.7040.0761.9241.1090.50981.1750.3730.1851.8152.8470.1361.8641.0540.43291.0940.3370.2391.7612.9700.1841.8161.0100.412101.0280.3080.2841.7163.0780.2231.7770.9750.3631A2A3B4B2D3D4D2E23AMnThe following table are th

30、e constants for determining from R the 3 Sigma Control Limits for Control Charts and for estimating the process Standard Deviation from RSinderella LiuCalculation :MeanRangeRun A Run B Run C Run D Run E Run F Run G Run HRun IRun J Run K Run L2.942.312.862.132.623.313.043.122.742.423.922.152.192.471.

31、321.432.112.092.460.801.391.792.011.63Centerline:kxxkii1= 2.795Centerline:kRRkii1Average Chart= 1.808Range Chart=1215. 231. 294. 21263. 147. 219. 2=Sinderella LiuCalculation :X Chart : Upper Control Limit:RAxUCLx2Lower Control Limit:RAxLCLx2R Chart : Upper Control Limit:RDUCLR4Lower Control Limit:RD

32、LCLR3469. 3808. 1373. 0795. 2120. 2808. 1373. 0795. 2369. 3808. 1864. 1246. 0808. 1136. 0Sinderella LiuX-bar Chart0.001.002.003.004.005.00Run ARun BRun CRun DRun ERun FRun GRun HRun IRun JRun KRun LRun #LPDUCLCLLCL R Chart0.001.002.003.004.005.00Run ARun BRun CRun DRun ERun FRun GRun HRun IRun JRun

33、KRun LRun #LPDUCLCLLCL X - R Chart RepresentationSinderella LiuOBJECTIVE Reveal nature of a process Reveal significant change in process Early detection of processing problemsSinderella LiuCalculation :MeanMoving RangeRun A Run B Run C Run D Run E Run F Run G Run HRun IRun J Run K Run L2.942.312.862

34、.132.623.313.043.122.742.423.922.150.630.550.730.490.690.270.080.380.321.511.77Centerline:kxxkii1= 2.795Centerline:Average Chart= 0.674MR Chart=1215.231.294.2=121277.155.063.0Moving Range ( ) = )(1XXniiabs111nkRnkimmiRiRmSinderella LiuCalculation :X Chart : Upper Control Limit:Lower Control Limit:MR

35、 Chart : Upper Control Limit:Lower Control Limit:RmxExUCL2587. 4674. 066. 2795. 2RmxExLCL2002. 1674. 066. 2795. 2RmDRUCLm4202. 2674. 0267. 3RmDRLCLm3000. 0674. 0000. 0Sinderella LiuX - MR Chart RepresentationX-bar Chart0.001.002.003.004.005.00Run ARun BRun CRun DRun ERun FRun GRun HRun IRun JRun KRu

36、n LRun #LPDUCLCLLCL MR Chart0.001.002.003.004.005.00Run ARun BRun CRun DRun ERun FRun GRun HRun IRun JRun KRun LRun #LPDUCLCLLCL Sinderella LiuSpecial Notes Many times we have processes where the major variation is not WITHIN the normal subgroup but is BETWEEN the normal subgroups. This is especiall

37、y true in batch operations such as Epi, Diffusion, Solder Dip, etc. In these cases, X & R or X & s, which are based on WITHIN subgroup variation, are not adequate and we must use Moving Range Charts (MR Chart).Sinderella LiuTest For InstabilityTest For InstabilityProcess MeanUCL (Upper Control Limit

38、)LCL (Lower Control Limit)AABBCCZone A 1/40 of points will be here +1+1X+2+2-1-1-2-2-3-3+3+3Zone A 1/40 of points will be here Zone B 1/7 of points will be here Zone B 1/7 of points will be here Zone C 1/3 of points will be here Point OutsidePoint OutsideZone C 1/3 of points will be here In Applying

39、 Western Rules consider only 1/2 of control chart at a time i.e., upper half or lower half Each zone is one sigma ()Sinderella LiuTest 1: Extreme Points / Out of Control Applies both for X and R control chart Process is OUT OF CONTROL if one point falls outside the control limit(s) SPECIAL CAUSES ar

40、e PRESENT at THIS POINTv IMMEDIATE ANALYSIS of the operation is requiredv Mark Out of Control Point(s) with and the Rule Number which has been violated.Process MeanUpper Control LimitLower Control LimitAABBCC1Sinderella LiuTest 2: Two out of Three Points in Zone A or Beyond Applies for X chart only

41、Process may be OUT OF CONTROL SPECIAL CAUSES may be PresentProcess MeanUpper Control LimitLower Control LimitAABBCC1,22Sinderella LiuTest 3: Four out of Five Points in Zone B or Beyond Applies for X chart only Process may be OUT OF CONTROLProcess MeanUpper Control LimitLower Control LimitAABBCC3331S

42、inderella LiuTest 4: Successive 8 Points Above or Below the Centerline Applies both for X and R control chart Process may be OUT OF CONTROLProcess MeanUpper Control LimitLower Control LimitAABBCC4Sinderella LiuTest 5: Linear Trend Identification Applies both for X and R control chart Process may be

43、OUT OF CONTROL - if seven successive points show a continuing increase or decrease, a systematic trend in the process is signaled.Process MeanUpper Control LimitLower Control LimitAABBCC55Sinderella LiuTest 6: Oscillatory Trend Identification Applies both for X and R control chart Process may be OUT

44、 OF CONTROL - if 14 successive points oscillate up and down, a systematic trend in the process is signaled.Process MeanUpper Control LimitLower Control LimitAABBCC6Sinderella LiuTest 7: Avoidance of Zone C Test Applies for X chart only Process may be OUT OF CONTROL - if 8 successive points occurring

45、 on either side of the centerline avoid zone C Process MeanUpper Control LimitLower Control LimitABCABC7Sinderella LiuTest 8: Run in Zone C Test Applies for X chart only Process may be OUT OF CONTROL - if 15 successive points fall in zone C only, to either side of the centerlineProcess MeanUpper Con

46、trol LimitLower Control LimitABCABC8Sinderella LiuPp (Process Performance): Pp is the performance index which is defined as the tolerance width divided by the process variation. S 6LSLUSLPpninXiXS112)(, whereAnd here “n” refers to the total number of all of the individual values sampled. Sinderella

47、LiuPpk (Process Performance Index): Ppk is the performance index which accounts for process centering and is defined as the minimum of ninXiXS112)( (Note: Both Pp and Ppk should be used only to compare to or with Cp and Cpk and to measure and prioritize improvement over time.), whereS3USL XorS3LSLXS

48、inderella LiuCp (Capability of Precision): Cp is the capability index which is defined as the tolerance width divided by the process variation, irrespective of process centering. , where2d/R 6LSLUSLCp2d/Ris the estimatedstandard deviation by , here is the average of the moving ranges and d2 is a con

49、stant varying by the sample size n , used in grouping the moving ranges as shown in the partial table below : 2d/RRn2345678910d21.131.692.062.332.532.702.852.973.08Sinderella LiuCpk (Process Capability Index): Cpk is the capability index which accounts for process centering and is defined as the minimum of 2d/Ris defined as previous page.It relates the scaled distance between the process mean and the closest specification limit to half the total process spread. , whereor2d/XR3USL2d/XR3LSLSinderella Liu The End