Downloaded from https://www.everyspec.com Certain situations may arise where high asaur- PROBABILITY Or ACCEPTANCE ances of achieving specific reliabilities are de- o.N sired. In such cases, the utilization of some level, other than the AQL, would be necessary -_ to achieve such desired protection. Thus if 95 percent assurance of achiev , a reliability of - 99 pecent were desired, it wuld be similar to a probability of 5 percent of accepting lots I .... percent or more detective. This point would - _ _ then be the basis for inspection. -- "" 5-16. fesubmiuslon and Retest. The disttnc- M - -.. tion between resubmission and retest is fre- quently lost in applying sampling procedures.
0/ It is true that both resubmission and retest re- 1 " . quire additional inspection, but the similarity / endi there, A retest is essentially a double- - (or multiple-) sampling procedure. It acknow- ledges the fact that the first sample is not large - - enough to distinguish adequately between ac- . / ceptable and nonacceptable materiel in border- line cases. When materiel is subjected to a Figure 5-9. Affect of acceptance number retest, it should not have been altered in any with fixed sample size way prior to retest. The materiel subjected to retest has not yet been rejected or accepted.
inspection by inspecting a large number of con- The risks involved in the case of retest are secutive units of the product as produced, and those of a specific plan and are, therefore, then the sampling of the product that is quail- clearly defined, tied. Resubmission, on the other hand, is a proce- dure for reinspscting materiel that has pre- One of the major virtues of many of these plans dureforreinpecing ateiel hathas re- is that products which have passed the inspec- viously been rejected. It implies (and the pro- cedures for resubmission should require) some tlon station are accepted, and are free of re- procedure for removing the defective portions inspection.
This type of plan attempts to as- thceue lotor reworking the lofet.iTo poreventsh of esure that the average outgoing quality will not F resubmission of nonscreened or nonreworked exceed a specified limit where the process is materiel, the acceptance criteria for a resub- in control. mitted lot should be more stringent than those applied when the lot was originally submitted. It is recognized that in some continuous-sam- 5-17. Continuous-Sampling Plans. It has be- pling plans, questions arise as to the quality come evident that where production is contin- of the product between the time that a defect is uous, as in conveyor lines, the formation of in- detected and the last preceeding sample.
How- oever one of toneyo assumtio, thf mormtio of thes spection lots for lot-by-lot acceptance issome- ever, one of the assumptions of many of these plans is that in the long run, the quality will be what artificial, and may he impracticable or un- no worse than quality specified by the plan. duly costly. Some other plan of inspection, therefore, is necessary for this type of produc- tion. Such inspection schemes are advantageous to the manufacturer because he does not risk hay- Continuous-sampling has been developed to fill ing large quantities rejected, does not need this need.
Continuous-sampling dispenses with storage facilities for static lots or rejected ma- the notion of the lot as a static entity, and con- teriel, and does not have to tie up his production siders instead a continuous flow of the product. lines for reinspection a .d reprocessing. Such The essence of most continuous-sampling plans plans also work to the consumer's advantage, is the qualifying of the product for aampling because they result in lower cost. 5-10