Process Capability Analysis ‘Sasadhar Bera, !IM RanchiIntroduction Process Capability Analysts is used in industry for Investigating how capable a process is for procucing products or services that conform within specication Timit for essential quality characteris, Process capability Index can also be usod as the Lbenchmarhing for qualty improvement activities. Companies use Process Capabity Index to demonstrate the quay oftheir products. Vendors use them to decide ‘their business relationship withthe manufacturer Specification Limits and Tolerance Limits Tolerance limits: Tolerance limits measures the variability ‘fa process and are calculated using sample observations. The sample observations are cliected from 2 process at 3 rpecified tie interval. Tolerance limits are aso known 35 onto ts onsider a process for which the quality characteristic has @ Jorma distribution with mean w and Standard beation 3 The upper and lower “tolerance ime” of the prac fll t “41 39 and u-30, respectively thats, tks W430 eye Specification Limits and Tolerance Limits Specifiation limits: Specitcation limits are determined by the needs of customer What the customer wants in Product or sence is analyzeé through market research ans Incorporated through product or service design. These limits are placed on a product characteristic by deslaners land engneers to ensure adequate functioning of the roduc. Itrapresents the acceptable bound of a product. Example: The élameter ofa part has to tan assembiy. The specication limits foe the diameter + 0.015 em. Upper specification mit (USL) «5 + 0.015 = 5.015 om ‘Lower speciation limit (St) = 5 0.015 = 4.585 cm ‘Specifiction interval = 270015 = 0.0300 Specification Limits and Tolerance Limits For 2 normal distribution, the tolerance limits Include 99.73% ofthe data points, or put another way, only 0.27% ‘ofthe process output wil fall outside the tolerance limits. “wo points shouldbe rememberee: 1.027% outside the tolerances sounds small, but this ‘corresponds to 2700 nonconforming pats per milan. | the dlstrouvion of process output © non-normal, then the percentage of output fling euside y= 30 may afer considerably trom 027% Tolerance limits reflect the variblity ofa process and have no relationship fo specifiation limits. Tolerance its are 9 known ae natural tolerance ents. .Specification Limits and Tolerance Limits. Example: The dlameter of a pat has to fit an assembly. The specification limits for the dlameter 5 + 015 cm. The ‘samples taten from the in-control process has sample mean fof 499 cm and standard deviation 1004 cm. Find the tolerance lets ofthe process What is Process Capability Index? ‘Process capability index (PCI) tas been developed to measure capability of 2 process based on various criteria, ‘which includep product consistency, process centering, process departure from a target and process loss. In ‘genera largo the value of PO, the more capable isthe process. These indies are effective tools for process ‘capability analysis and quality assurance. Iwo bose features of process copoblity study are that location f process mean in elation to taxgot value, and ‘overall process variation. The closeness to the target and smaller process vaiation are pays special attention from ‘astomer point of vew. Process Spread with Specification interval Specification interval = USL~ (St Process spread = UTL—UTL=60 + speiteaton trea $e Br a a one Process Capability Analysis Assumptions ‘he assumptions for process capability anaiyss are Process is under statistical control. in. satsally contcled process, variation occurs due to common cause only > Process data folows norma dstribution,Different Types of Process Capability Indices Different types of process capability indices are: 1) Process potential index) 2) Process capability index (C3) 4) Process capatitty inden against target (Cp. ) Process Potential index: C, —. “The valve of C, <1 indicates speciation interval is smaller than process” spread and the process is sad to be Incapable. n other words, the company is procucing junk. Tet Process dsr ry te to ’ vol Process Potential index: C, A common measure for describing the potential of 2 process 10 meet product speciation is C, Index. This Indo i dined a the ratio ofthe specification Interval 12 the process spread ust rc were USL and LSL are the upper and lower specication limits. 9 represents the process standard deviation. The main weakness of C, index is that it measures capability based on the actual process sorea (60), does not consider the process center This index does not reflect the impact ofthe shit of me process mean Gg Process Potential Index :C, ‘he vale of 6 = 1 indicates specification terval equa to the process spread. In sich stuabon, process sd t0 be tarely capable; Ht hae potent! to produce non-delectnve produce if the process mean is cemered to the spectied Toreet Tere s midpoint of speciation tts Teg Froess dsvtn st tas oy,Process Potential Index :, ‘The vale of G, > 1 indicates proces is potential capable, the process mean is centered to the specified target. ‘The goo! Isto produced within specifiations with a C, of 1.38, ts much easle to achieve, and stil easier with C, of 200. ‘aed on the emerience ofa numberof practioner, they have suggested 2 “safe” lower limit of, 2 1.33 because ‘of additonal sources of vavation are experienced in production process. Process Capability index: Cx ‘The Gy index considers both the magnitude of process sariance and location ofthe process mean ot where USL and LSL are the upper and tower specification limits. wand o represent the process mean and process standaré devation, respectively C,, and, ae the upper sided and lower sided capabilty index, respectively. Ge -sinn 8, =Minioun {c,..c,] Process Potential index: C, — (> 1 and centered process mean, pis. Peirearee omen) ae process mean et an tn ye o = Process Capability Index : Cy, — ‘Gq estimates how far the process mean is from the neorer specication fit in terms of 30 stances. ‘Gy measures how well the process mean is centered within the specification lim, and what percentage of product wl be thin specifications.How Cj, Consider Process Centering? Process Capability Index against Target: C,,, \sang ard Taguchi (1885) introduced the Indes Cg - The 2 measures process performance by taking into ‘Eount deseness to the target valve (and variably (o") fa process. c_- USL=LSL_ _ USL=LSL rs cS sere 1= JF FTN. Taeuen process oss fence Ca Is the average process loss relative to the sanutactring tolerance Effect of Reducing Variability in Process Process Capability Index against Target: Cus Pea et al (2902) process capability index Gon - Wis constructed by combining both G, index and C,indox = Minimum | USL . = When the process mean (k) departs from the target value (1, the reduction of the value of Cy is more sensitive than the others two indices C, , Gy and aso being sensitive tothe changes process variation. The target value is taken to be the midpoint of the specication lint: T= m, alas stated otherwise.Process Capability, Z Distribution and PPM When the process isi statistical control and normally stributed, calculated Z values can be used to estimate the ‘Proportion of output beyond any speciation limit. Lat doting 2 Minimums) ‘expression of Gy In terms of 2 at b> PrdlahD dE ante 2 [Link](t) pres 2 Laman saneles) Process Performance indices the recess performance indo are ng term eapabity of 2 proceso meet the prod specication. ‘he equation and relatonstip of process performance Indicos (P,P) are same as process capability indices. ‘he only itforence is hat process standard deviation (0) ‘estimated by using all ofthe individual reading ts 2 long ‘erm estimate elses ferme sng MS Excel PPM» 2°(-NORIMS DIST, TRUE) ‘is totenotd that n PPM caleutin ty proces sft by 1.50 [St coer yarn Lage? 13" Zaneee 22S sf ve hams 2 erlbinale ansa = sonm-a a2 57 (0,70) Difference between Process Capability Indices and Process Performance Indices Process capability analysis and process performance ‘anoyss are diflerentiated by standard deviation estimate, Process capability considers short-term variably & long term variality Is considered for process. performance analysis which address directly how well the process performing relative tothe needs ofthe customer. Process capability indices C, and Cu typically assess the potential “short-term” capabiity by using a “short-term” Standard deviation estimate, whe P, and P, estimate “longterm” standard deviation by using all ofthe individual reading.Estimating Process Standard Deviation There are two types of process standard deviation (0) estimates: ‘Short term o estimate: 1) Using R-bar chart, 2) Using bar chart 4) Using Pooted-standard deviation Long term © estimate: 1) Using individu data point Short term estimate of o — Jing Sar chart: Callao the standard deviation value (5) ‘or each sample, Next calculate 5 forall samples. ‘hort term astimate of proces standard deviation (0) Short term estimate of Using far chart: Calculate the Range value (R) for each ‘ample, ext ealculate for al samples. ‘Short term estimate of process standard deviation (0)Long term estimate of Long term 0: Using individual data point Sample sterdard devation: $= Y= ——— Total ebeenatoni) = Mumba of ample sample se >rocess Capability Index for Non-Normal Data Incase of non normal data, C and G, values caeulsted using ‘conventional methods often leading, to erroneous resut because they are defined under” the astymptions of romaty, ‘One approach to desing with non-normal data is to transform the data to make i normally dsinbuted. Typical ‘ansformations include tating the teciprocal, reciprocal seqare root, natural fog, oF square reat ofthe raw data. The corresponting speciation lint must be silly transformed. An akerative approach so identify probability istration by using goodness of test. Determine the parameters) of the distbutlon. Determine the process ‘anabilty and arepertion of noncanlerming Process Capability Index for Non-Normal Data ‘There are several characteristics that do not ‘ellow the rrormal ditrbution such as roundness, mold dimensions, surface finish, eccentricity, angularty, straightness, Squareness, wold strength, casting hardness, insertion depth, shrinkage, inductance, dynamic imbalance etc. wer Ps tale’ Process Capability Analysis for Attribute Data ‘An output item may be nonconforming. unit (called defective, or may have number of noncorformities oF defecs, n case of attribute data, the process performance |S measured by parts per milion (PPM) defect rate oF Aefacts per millon eppertuitios(2PM0} When 2 fraction of nonconforming is the measure of performance, it i typical to use the parts per milion {Pt defect rate as a measure of process capably. PPM measures how many pats ae defectives out of every ‘millon of items preduced. PPM makes easer for anyone te visualae and understand the quality level of 3 roduction process.>rocess Capability Analysis for Attribute Data —. ‘hen deating with nonconformities or defects, defects per anit (QPL) i fst calculated, To Incude compleaty of the tem produced, the number of potential chances fora defect 12 occur is Incorporated in the caeulation. These potential chonees are known as opportunities. spy __ Totalnumiber of defects ‘Toualnuenber of untsproduced Defects per milion opportunities (DPMO) = bru “Tal numberof opportunities *'”” rocess Capability Indices for Multivariate Data \¥ we assume that the process quality characteristics flows rmolvaiate normal stibtion, then liptical contour [two dimensions) or lipsoid gher dmonsions) define the process region, and the tolerance region is rectangle [two dimensions) or hypercube la higher dimension, The shape and size of elipsoid Is influenced by correlation structure (vaiance-covariance mati) of the qualty tharactersties. Therefore, we cannot easily extend the existing definition of univariate process capability indices Into muitvarite setup. Process Capability Indices for Multivariate Data Capability indices Cy Gas Go + and Guy hive been proposed to evaluate process capably burt restricted to ‘eases wath single quolty characters ‘The process capability of » product with multiple quay characteristics called multivariate capability index Inthe univariate ease, the location and dispersion of the rormal distibution curve | used to measure process Interval length. Specification limits prove the defined Imerval length ofthe product tolerance. Therefore, we can compare the interval length in unhariate domain Process Capability Indices for Multivariate Data ‘There are three types of Multwariate Capability indices |. Multivariate capability index (MC,): Used in Multivariate ‘poral distribution data, i, Multivariate capability inde (MC): Used in Maktivarite ‘normal dstibutlon data 1, Muttnarate process capabiity index using. princpal component analysis (MG nc I Generally used in ‘multvarite non-normal dtibution date