OSE. Acceptance Sampling AC sampling is the process of evaluating a portion of the poigcviaterial in a lot for the purpose of accepting or rejecting the lot as either’ conforming of not conforming to a quality Specifications, Inspection for acceptance Purpose is carried out at Many stages in anufacturing. There are generally two ways in which inspection is cartied out : (9 100% inspection. (ii) Sampling inspection, In 100% inspection all the Parts or products are Subjected to inspection, whereas in sampling inspection only a Sample is drawn from the lot and inspected. A sample may be defined as the number of items drawn from a lot, batch or population for inspection purposes, : Sampling inspection can be defined as a technique to determine the Aceplance or rejection of a lot or population on the basis of number of defective pats found in a random sample drawn from the lot. If the number of defective lems does not exceed apredefined level, the lotis accepted, otherwise it is rejected. , _ Sampling inspection is hot a new concept. In our daily life we use sampling Mspection in selecting certain consumable items, For example, while purchasing Our annual, Of monthly requirements of wheat, rice or such other food grains we Talurally take a handful of ‘grains to judge its quality for taking purchasing decision. liveare not Satisfied we take another sample and after two or three samples from tte same or different sources we take Purchasing decision. Or, let us take another Stample, Suppose we want to Purchase mangoes we nomnally take one or two iis from the lot and taste its quality, if the samples taken are found good kcide to purchase the Tequired quantity. 5 Similarly in engineering sampling inspection is preferred because it is more pote, Guick and economical as compared to 100% inspection, The main purpose agg Oe Sampling i§ to distinguish between good lots and bad lots, and to Me gy 018 according to their acceptability or non-acceptability, vantages of Sampling inspection are as follows : \ Fimpjng ,-" 2S which are subjected to destructive test must be inspected by be ™MsPection only, \ apa “ost and time required for sampling inspection is quite less as 4 9 100% inspection, | ‘ Problem Of inspection fatigue which occurs in 100% inspection is a, ‘ eat inspection staff is necessary. ugg M88 damag ig Hpecr9n $¢ to products because only few items are subjected to handling 321> STATISTICAL gy 32 ALITY © My 6. The problem of monotony and inspector error introduces 5 inspection is minimised. E. i 8. The most important advantage of sampling inspection is ta more effective pressure of anty, improvement. Since the rejection of on on the basis of sampling brings much stronger pressure on quality ane than the rejection of individual articles. . Limitations. (1) Risk of making wrong decisions : However, in sampling inspection, since only a part is inspected, its ine that the sample may not always represent the exact picture obtaining in ij. ly and hence, there will be likelihood or risk of making wrong decisions aboy jy, lot. This wrong decision can be made in two ways. Firstly, a really good lot (thy is, containing less proportion of defectives than specified) may be rejected beca the sample drawn may be bad. Secondly, a really bad lot (that is a lot containing Brealer proportion of defectives than specified) may be accepted because the sanp re 800d. In the former case, the producer has to suffer a risk oi es rR Heer and hence the associated risk (chance) is called as lle and hence the ys Be later case, the consumer nuns the risk of accepting bad lo’ sociated risk is called as consumer’s risk. (2) The sample Usually 100 per cent inspection, >, PrOY#dES Tess information about the product hs f a 0 Randomness Sampling sch ) Sam sam, cme depei vane factors ample gj Of samples nds upon the following fac! i thisheg 2 3. To deter Machine ig Machine ws for further proces!"® manutactuey ¥ing, mat ty of r section to section wil! eres, Mainining anging a Sontroy lin, ‘ tbe Quality of the produc®a item of the lot being included in the sample is the same. An i _<, placed back in the lot. "ANCE SAMPLING ee 3233 Formation Alot is a collection of items from which a samy i aE ks ple of two or m ‘i isaawn Re, to ee its acceptability, lore articles ‘The lot formation greatly i luences the outgoing quali ‘ : 4 i i quality and inspection costs. owing points F Rios § ‘The foll ; i should be taken into Consideration as guidelines for the formation of a ; 1. The products from different sources (processes, tion shifts, i uterials, etc.) should not be mixed together, unless tee baie ae Jot-o-lot variation is small enough to be ignored. 2, For lot formation the products should not be period of time. 3. The extraneous information such as process capability, inspection ec, should not be used in lot formation. er __ 4, The lot should be as large as possible consistent with the above to take advantage of low proportionate sampling costs. Sampling methods f The sampling methods can be classified as : ~ 1: Simple Random sampling 2, Stratified sampling 3. Systematic sampling 4. Cluster sampling 5. Two stage sampling. Simple Random Sampling Selecting a sample in’such a way that each item in a lot has an equal chance Of being selected, is called random sampling. Since, a judgement about the lot 1510 be made on the basis of only a sample it is very important that the sample thely represents the universe from which itis drawn. This requires that the sample Sze be large enough and the sampling procedure such as to avoid bias, Parts resting nthe bottom or in the middle of a group must be selected as well as those lying Sonveniently on top. ee According to this method, the sample of the requisite size n is drawn from Alot of si i ; ile selecting an item, the chance for Of size N, in such a manner ; that whi ig ee accumulated over extensive tratitiea Sam, pling 2 7 tit may be Thlarge lots, the difficulties of random selection my . a Pee ‘te 2 He to adopt stratified (proportional) sampling. lowing rules should be followed. «mule inspection lots should, U Draw proportional samples. According t0 this Tule inspection lots ada STATISTICAL QUALITY Conny, NTR, 34 acres ih auustind 08 ibis of the basis of homogenity (ints ceri ¢ of homogeneovs SrovPS on h sub-lots of the inspect; umber 0! sample items from all parts of suc] { Pection ly, 2. Draw je items blind (without any bias). 3, Draw a Jot into which the inspection lot is divided a Sub-samp. should ee Te size of the sub-sample should be proportional to the size ) ibe {ficient than the sim f it ay be generally more efficie! simple | This method Prag easdci sampling may not always result inthe a ae ies from such stratum of the lot, thereby affecting the ' cS representativeness of the sample drawn. p It would be advisable to ensure that a minimum of two items are selected from each sub-lot. Systematic Sampling : When the items in a lot are presented in an orderly manner, (such as piles of mild stéel sheets or stacks of cement bags) it is possible to considerably simplify the selection of a random sample of the required size. In systematic sampling on@ item is chosen at random from the lot and ceacalét the items are selected regularly at predetermined intervals. It has be Shaves method of systematic sampling is quite good approximation i ieee thei Sampling provided there is no deliberate attempt to manipulate for inspection, is ui the lot in any desired manner while the lot is presented Cluster Sampling ' ( ___ When the lot submitted fo, f ca itmay be oo even Consists of certain groups of clusters fen “amine all the items in he onal 10 Select a few clusters of items € aaa Selected clusters, ten ony fey gan Fack the ena tS PACKed in cartons and it set Islan and alt the po SElected at ranqe eee for selecting sample items 1 the Selected ga tout replacement in the fus! “8d cartons ate inspected. i the lot gupess \\ Sach consis Pabited for ine i \ ges es a of pebection Consists of 1 ages A ; a é [i eye tun 2t be nen Packages (pr, Cled 5 BES (8 in onc: ao te required umber Units) are ye Stages, Ty we of cluster samplind, Sampling EMS ate chosen SEM at ran Tn ae Lust stage a desi (i) Sampling Bee MY be carne OM the gece Second sae ables, 1 Selected primary units ‘aMpling py attribu'esANCE SAMPLING 533.02 25 more ‘sti i i istics. The decision eee perptrermp tree n into account, sampling by attributes can be oe bias — ble a Go and no-go gauges: can be ets pection Se le characteristics it is not necessary. to register the ements palo, gersicaly the number of defective items found in the sample. n sampling are it is necessary to take measurements on the tics inspected. A statistical treatment will then be i or rejection of the lot. wen Is Used in Relation to Sampling Acceptance »,,the items, ' att ce are Classified as non-defectives or 1g : N= number of pieces in a given lot (Lot size) n= number of pieces in a sample (Sample size) M = number of defective pieces in a given lot of size Ne m= number of defective pieces in a given sample of size ne ‘= acceptance number, the maximum allowable number of defective pieces in a sample of size n. P= fraction defective. In a given si sample p = nv/n.'~ : p’ = true process average fractit inspection. = average fraction defective in observed samples. : probability of acceptance. : Pc = consumer’s risk, the probability of ‘accepting product : . quality. The probability of rejecting product of some stated quality © ig referred to as producers risk, (1 — Pa). Pos Po.s0» = fraction defective having a probability of acceptance of 095, 0. 0.15 etc. under any given ‘acceptance criteria. ating Characteristic (OC) Curve ating characteristic curve for an attri efective in a lot against the probability of acoeplane’. OC curye sho ubmitted lot p = MN ina given ion defective of a product submitted forSTATISTICAL QUALITY CONTROL) N=100 ling plan is | = 5 If the sampling C=2 BI) differ. To construct an oc curve, WE s ‘lity of accepting lots with varying per cen “etl the table of Poisson’s distribution (given Producer's risk Indifferent Quality region risk 0 AQL UPD ——— Proportion of detectives (P ) ——> ; i Fig. 10.1..0C curve. Producer’s Risk and Consumer’s Risk q ‘There ae always two partes to an submits ‘acceptance procedure the party is thep for aceeptance and the party for whom the decision is made ree OC curve for pcan a Moducer and consumer), Fig. 10.2 shows a” ® rate then Oe ubet® it i desired to accept all los of 2% defective lots greater than 2% defectives have a probability of acceptance of zero. Act)” however, no sampling Plan exists that can discriminate perfectly, ther alway remains some risk that a good lt wll be rejected or that a bad lot will be cee Producer’s risk. Since, ideal samplin, 4 th Une fj : 18 Plan which will satisfy b© consumer and producer is not possible, x je an they have (o tolerate certain risk,” COMPFOMIse has to be mad e_ ie P coor sens 4c A TiIAUD Ads if he quality is good still from eee eee pas to suffer. The producer's ) ility of rejecting a good fot which oDerWwiSe would have been — oopiedsSo the producer should be HOF g.-against the rejection of rae The producer can decrease his risk by producing provi at a better quality level than specified AQL (explained latter) depending on ° other economical considerations. Saying a producers risk « = 0.05 | tems that in the Tong run about 1 Tot in 20 wi ss controlled at AQL. quality level. Jots are coming from a proce: Consumer’s risk. If the quality is ill suffer. Consumer's risk Jots are to be accepted the consumer W? of defective lots being accept Po,10 = 2.5% means the {more than 2.5% defectives and : 25% defectives. At first impression, it appears completely opposite viewpoints tow: | pierce Benne relationship betw f 800d products in the effort to Ba of consumer, The consumer a In the long run the costs pues on by the producer # Tejects is not ayailable for Aol gi: » UAcceptable Quality Level es which the consumer o the maximum per cent n AOL is an acceptable qu ted which otherwise would have consumer does not wanl he would at the most accep! that the producer ards the selection intersects are conflicting). But, more criti een producer and consumer ; subs! exclude bad is interested in quality, incident to the rejection of g to the consumer, ‘econdly, any good product his immediate use. _sallty Indices for ‘Acceptance Sampling Plans (AQL). It represents finds definitely accepl ie defect Can be considered satisfactory Acceptance line (0,0) Cua ecaeeea e 100p- ——> Fig, 10.2. Ideal OC curve. i ll be rejected provided that the bad still from the sampling plan some is the probability been rejected. Saying ta worse quality containing t 10% of lots containing ‘and consumer should have of sampling plans (ie. the cal consideration will show that, there products are not necessarily in the he is also interested ood products tend the maximum proportion table. AQL can also be tives that for the purpose of sampling ‘as a process average. It is the fraction yy serious effect upon further processing ility of acceptance for AQL ality level, the probal cer’s safe point is termed as » MA Gi ens: as QU should be high. In fact the Pr IHSTATISTICAL QUALITY CONTROL. OL \ ». Rejectable Quality Level (ROL). It is also called as Lot Tolerance Pep cent Defective (LTPD). This isa definition of unsatisfactory quality. Tt represents tion of defectives which the consumer finds definitely unacceptabig. ‘AsROL isan unacceptable quality level, the probability of acceptance for an ROL Jot should be low. The probability of accepting a lot at RQL level represents consumer’ risk, Z 4. Indifference Quality Level (IQL). This is a quality level somewhere petween the AQL and RQL. It is frequently defined as the quality level having: a probability of acceptance of 0.50 for a given sampling plan. 4, Average Outgoing Quality (AOQ). It represents the average % defecti re ~ in the outgoing products after inspection, including all accepted and all rejected: ~ jots which have been 100% inspected and defectives replaced by non-defectives: So, for a given fraction defectives, the lot accepted as a result of first sampling inspection will have a fraction defective p’, the rejected lots are subjected to 100% inspection and rectification (defective articles are either replaced or corrected) the AQQ will therefore, be less than p’: iad Let, n= sample size, ,.. : umber of lots submitted for acceptance, P fraction efective} #95" ViieN probability’ of acceptance (1. - P= Probability of rejection. Poot rie Bese wit ty sn of the lots K(1— Pg) which are subjected to 1004 inspection and rectification. He an ee originally Present in k lots = K.N.P’. 4 Of defectives in the outgoing lots a =P Pak(N-n) ‘otal number of defectives =D’ Pak(N ~ n) + k(n)(O f ) ) AQQ = Pap 0) + Kd O) TaOE 329 UIT hs an Y given n and c there. Between these extremes is the point at which the percent of defectives going: product will reach its maximum, This point is known as the average g limit (AOQL). Therefore, AOQL is the maximuim possible value of the Need defectives in the outgoing products after inspection and. “acceplacne/rejection scheme the OC curve is used but for rectification the curve of-AOQ plotted against p’ is used. : Any acceptance/rectification plan guarantees that regardless of the incoming ubmitted, the outgoing quality in the long run will not be worse than the t igh svras (> i P Bai sto esulae hovig, ina od PNA 1, Shows a common’ ‘type of approximate calculation to determine the AOQL. This table refers to the plan, n ='75 and C= 1 when N is large in son with n. The right hand column gives average outgoing quality (AOQ) assumed per cent defective in submitted lots, The maximum value of 12% occuring when submitted lots are 2.2% defective. This maximum AOL: hy ig 75, ¢ = 1 for Acceptance/Rectification plan. yo = ce | A0Q = 100p’Pa 0.385 SF TAI ; 0 409 from n Gant | Per ‘cent defectives'in |, 5. __ submitted lots ,STATISTICAL QUALIT ¥ conn, OL ‘The above computed figures are plotted in Fig. 10.3. The curve plons i known 25 A0Q curve. Curve without rectification 7 7 \ Max AQ fe iQ =AOQL=1-12 % Rivcmommsi 20m 25 030 35 = +0 : = — Percent Detective (100p') —~ {/Fle. 103, Average Outgoing Quality for Acceptance’ Rectification plan. eee erate omsens) what would happen if there wee 1 cease ane the AOQ curve falls below the AOQ P -ASp) increases, the proportion of rectified lots increases and hence ‘the AOQ euye falls below the line AOO = p'. Steps in the Design of ps in the Design of an‘ Acceptance’ Plan e tely ‘The protecnen ar wiiaa G8 described by the OG rant 10 the consumer and the producer is comple! it calls for designin, Suirve Of the plan {When the acceptance plan is designe! Points agreed Rear Whose OC curve will. pass through two SUPUl* acceptable quality level (AQL) ee and producer) The first of these 'S Which the consumer finds depetresenting the maximum proportion of defectives Poroent Defective (LTPD) llely aéceptable, The second, is the Lot TO : consumer finds definit, 5 whic! Which represer ive finitely presents the proportion of defective: ae Accra. 0C canon a Fe eecrved easier. we ; De divided into three regions, (zones) 3 shove 1, Acceptable Quatity esti Region, . | 2, Indifferent Quatity Beja) 3, Objectionable Quality Ree ion,” represented by AQL or less represen by LTPD or mai a leet all tots whi , ‘AQL and LTPD are conce es in the thes Sa ys a fraction efeeet said to be “‘indifferent’’ to them. But nad contro} Bees ane const net 8 available by acceptance sampling, the Advanta; seg mae consumer ust Re a sume ee ives ~ely acceptable os ened sae equals or is less than the s are agreed and explicit wantitative terms. For example, th producer may agree w acne fire rape the lots with a fraction defect fective of say 0.01 (AQL) will € consumer may agree to accept the risk isk that 5 per cent of the defective 25 (TPD) wil i I be accepted, discuss the hypothe! case we have Presented light of Fig. 104, consumer defined the AQL as being 0.01 and the LTPD as being cer is willing to accepta 10 per cent risk that lots with a proportion 0.01 would be Tejected and the consumer is willing to accept a ots with a proportion of defectives of 0.25 would be nae The symbols commonly used to race them are Pi = AQL = 001 i Po=LTPD= 025 oth “= producer's risk = 0.10 Saka a ual the consumet’s and producer 0-025 (P2) ‘ ' Proportion of detectives (p) “104, 0C curve (= 010 B= 005).STATISTICAL QUALITY i 332 CONTRG for this set of conditions would have the general form shoy The OC curve in Fig. 104. a Bd ain rect, what the consumer and producer ag is a set of two poi for ac oC. ane ; these points are (py, 1 — a) and (2, B). In other words, they agree that any acceptance plan whose OC curve passes through these pre-seleg points will be satisfactory. A tial and error approach must be employed in thi design. To illustrate this, the question may be ; given a lot size of 100, what sh be the sample size and acceptance number ? The only way in which this can determined is by arbitrary selecting a sample size and an acceptance number calculating the respective probabilities of acceptance for the AQL and LTPD, an hoping that they coincide with (1 — a) and B: Chances are that they will n Another combination of sample size and acceptance number must then be selec and the procedure repeatd until they coincide. ' 0-9 OB 07 0-6 0-5 Probability of acceptance (Pa) dan ad. = 0 2 4 6 8 10 12 14 Percent defective (100 p’) Fig. 10.5, Comparison of OC curve,a sas 333 The problem of design is further compli js licated oad B remain fixed, a different Pian ou gd the fact that even if Pr a Tequired for €ach lot si: lity Computar; lot size characteristics of OC Curve Patations, 1, The Operating Characteristic ce plan shows the ability of the plan to di stinguish bes Of an Acceptance Saraplin, : ween 2. Sampling acceptance Plans with same per pignen = bad lots. i t quality protection. For e: 7 iples gives very cena “ample, the curves (Fig. 10.5) shows that lor 1-0 -———_—_ 0-9 0-8 3 07 a 8 Cc 0-6 8 a ¢ 8 N=1000 2 05 n= 20 ® c=0 > 30-4 N=200 a - 84 x 0,025=21 From Table G, for np? = 2.1: S c=1), Probability of acceptance : Py = 0,380. Therefore, total number of lots accepted, when ¢ = 1 ; = 0,641 x 1000 + 0.380 x 1000 = 1,021. oe Problem 8, Design a sequential sampling *cifcations : plan for the following o = 0.05, Pi = By B= 0.20, P, = 0.3 _Abso compute : Dy Oe , @ Average Outgoing quality ad hee sci be i ra Ast ‘ection of the lot. t Minimum number _ Ppecuves for ce : toe th loti Pr imum number of ¢ : Desi giben tage number of item!STATISTICAL QUALITY CONTR, 350 [ Sol. For sequential sampling plan Acceptance line, dj = Sp - hy Rejection line, dz = S, + ha 82 Sr ate nie (ob hy ~ 8 + 82 a nate te where, i 81 = 108 eae 0.30 i = tog 239 = 0.4771 = 108 019 = ° 1-pi 82 = log 1-py eB log o> = 0.6767 Substituting the values, we have. di = 0,186 n — 1,154 ana 4, = 0.186 n + 2.054 (a) When P'=py A AOQ = Py.p’ / =Q-a@p' = 0,95 x 0.10 = 0,095ANCE SAMPLING. | pec3St (0) Minimum number of items Inspected for acceptance i ett asa — 57. S 0.186 Of defectives for Tejecting the lot hn = 2.054 Say 3, si (d) Average number of items inspectea when (© Minimum number the lot quality is py =_%+b)-p PGi + 82) — p . oy 0.05(1.8808) — 0.6767 Ee 0.1(0.5862) - 0.1091 Problem 9, Design’ a Single-sampling inspection will meet or nearly meet the following Tequirements + @=0.05 p, =0.008' B=0.10 \p, =004 - Sol. Producer's tisk’ > - @=0.05 he. for quality of. Pi = 0.008 ed : ~ be 08% of defectives, the Expected probability of acceptance is } (1- a) = 095, Trying with acceptance number % ; wey From Table G, corresponding to Pee : c=0 and P, = 0.95 a np’. = 0.05 (by interpolation) ss oe 005 15 y p’ 0,008 Nike. Now, Plan by attribute which it should satisfy the consumer's risk also for which e relia see G, but actually it should have “sponding to ¢ = 0 ; P, = 0.756 ,7 SHO TICAL QUALITY . 35zet . CONTR Trying with other acceptance numbers, results can be tabulated ay folly, mp1 Pado.to corresponding Probability of to Pa= 9.95 acceptance foy From Table G consumer's risk g 0.10 from Table G % From the above table c = 3 seems to be giving very close results and satisfying both the producer's and consumer's tisk nearly. . n=171,, i j and c=3 is the required plan. d Problem 10. A single sampling plan is given as N = 10,000, n = 100 and “c= 2. i i (@) Compute the Se, probabiily on RCS of lots with 1% defective (use Poisson). - (b) Determine the A0Q value for the above lots. (©) What will be the average i inspection in per cent ? (Assume acceptance, rectification plan), Sol. 1 =100 001 2 1. ; From Table G, Probability of 2. or less defective = 0.920 i Bet +. Probability of acceptance of lot. = 92%, (0) AQQ = P,p’ = 0.92 x 0, 01 = 0. .0092, Of say 100 lots = 100 articles cach in 92 tots + 10,000 articles in m 8 lols (© Total inspection 7 ™ 9200 + 80,000 = 89,209AMPLING 211 os oo Ne s - ,, Total average percentage inspection 89,200 x 100 = 8.9%. 400 x 10,000 \ limitations of acceptance and explain the advantages and 1. @ State sampling over 400% inspection. (6) Compare Random Sampling and Stratified Sampling. 2, Explain the OC curve with reference to sampling inspection and the meming of the terms : \ 1. AQL \ 2, LTPD 1 3. IQL 4, Producer’s risk 5, Consumer's risk. : : 3. (® Explain the various types of sampling plans which are in practice | inindustry with their res! tive tance criteria. ‘ _ @ Differentiate Be single Sampling Plan and Double Sampling Plan. of the consumer and the producer in the selection * 4.@ Discuss the interest aa... its differen' () Re ormcc re ee 5. (a) What is an item by item sequential sampling plan ? Explain stains ls application. (© Explain the characteristics of OC eUFVE: ay an @ Whar is ATI? How will compute ; ible sampling plans ig; (b) Differentiate betwee? AQL and aod. . 7. Write short notes 09 29 jy ehree of the followin * fh ay 1. Single sampling plan. 4, 2. Double sampling PI? "3, Systematic samplins- 353° t zones (regions). ee