1、MIL-HDBK-HL07 QC W 7977970 OOqOL53 9- Quality Control and Reliability HANDBOOK (Interim) 4 STATISTICAL PROCEDURES FOR DETERMINING VALIDITY OF SUPPLIERS ATTRIBUTES INSPECTION _- 6 MAY 1960 OFFICE OF THE ASSISTANT SECRETARY OF DEFENSE - c (Supply and logistics) WASHINGTON 25, D.C. P Provided by IHSNot
2、 for ResaleNo reproduction or networking permitted without license from IHS-,-,-MIL-HDBK-HlO7 QC m 7777770 OOLiOL54 O OFFICE OF THE ASSISTANT SECRETARY OF DEFENSE WASHINGTON 25, D. C. SUPPLY AND LOGISTICS Statistical Procedures for Determining Validity of Suppliers Attributes Inspection H-109 This (
3、interim) handbook was prepared by the Chemical Corps Materiel Conmiand, Department of the Arrgy, on behalf of the Office of the Assistant Secretary of Defense (Supply and logistics). developed statistical techniques, together with appropriate critical values, to test the effectiveness of a suppliers
4、 system of sampling inspection in assuring conformance of submitted supplies with technical requirements. It provides specially The statistical methods presented are useful in procurement, storage and maintenance inspection operations when an independent check is desired of the fractions defective r
5、eported. This publication is issued to make procedures available to those in the Department of Defense and in industry who are concerned with resolving inspection and quality control problems to whicn such techniques might apply. Suggestions for improvement of this handbook should be addressed to th
6、e Office of the Assistant Secretary of Defense (Supply and Logistics), Washington 25, D. 6. F * J Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-. . . . - - MIL-HDBK-HL07 QC H 7777770 0040355 2 ?aragraph TABLE OF CONTENTS Title Page 2. VALIDATION OF
7、 SUPPLI.ERS INSPECTION 2.1 - Independent Sampling and Inspection- 3 2.2 Test for Homogeneity- 3 2.3 Statistical Criteria for Paired Attribute Samplings- 5 2.4 Combination of Tests for Homogeneity- 7 2.4.1 Examples of Use of Check Ratings- 9 2.3.1 Example of Use of Table I- 6 3. 3.1 3.2 3.2.1 4. 4.1
8、4.1.1 4.2 4.3 4.4 4.4.1 5. 5.1 5.2 CORRECTIVE ACTION Statistical Criteria Indicate “Warning“- Statistical Criteria Indicate “Action“- - 11 12 14 15 16 16 17 17 18 19 19 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-MIL-HDBK-HL07 QC 7799770 0040L56
9、4 TABLE OF CONTENTS (Cont) APPENDIX Table I Table II Table III Critical Limits for Cumulative Check Ratings . Table IV Limits for Determining Discrepancies Between Suppliers and Check Ratings for Paired Attribute Sampling Inspections Probability of Acceptance of Homogeneity in Each of K Veri- ficati
10、on Trials for a Specified Acceptance Probability in a Single Trial Probability of Acceptance of 4iomogeneity of Paired Attribute Consumers Paired Attributes Sampling Inspections Table V . I Samples Proportional in Size 1. . -e Operating Characteristic Curves of Two-Sample Test for Homogeneity I Prov
11、ided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-_ - MIL-HDBK-HL09 I1C I 7797770 0040357 b I- 1. INTRODUCTION 1.1 statistical tests and tables of critical values to Department of Defense procurement inspection and quality control activities for use in det
12、ermin- PURPOSE. The purpose of this handbook is to provide appropriate ing the validity of suppliers inspection records when sampling inspection by attributes is specified. only supplies conforming with technical requirements are being of fered or Such records serve to assure the consumer that - _ -
13、 - . - . 1 I . acceptance. 1.2 BACKGROUND. DoD Instruction 4155.6, “Department of Defense -_ Quality Assurance Concept and Policy,t dated 14 April 1954, established a- uniform concept and policy on quality assurance as related to procurement inspection. This policy requires that, after recorded data
14、 generated by. - examination and testing of manufactured product are ascertained to be reliable, optimum use be made of these data obtained by suppliers in deter- mining the acceptability of supplies. DoD Instruction 4155.8, “Department of Defense Procurement Inspection Policies and Procedures for I
15、tems Covered by Military and Federal Specifications,“ dated 7 May 1957, implements the broad concept and policy published in 4155.6 by prescribing uniform policies and procedures for relatively noncomplex items of supply which are techni- cally specified in the military and federal series of specifi
16、cations. Dol) Instruction 4155.8 outlines the suppliers responsibilities for perfonn- - ing the prescribed examinations and tests itemized in Section 4 (Quality Assurance Provisions) of specifications and requires him to maintain records of his inspection results. DoD Instruction.4155.8 also prescri
17、bes the Governments responsibility for inspection, including verification of the suppliers compliance with:Eechnical requirements of the contract, s Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-MIL-HDBK-H107 dC m 7977970 0040158 m 1.3 . TH6 GOVERN
18、MENTS-BXOHT To REJECT. For Government procurement ,applications, the right to rej.ect all nonconforming produces offered for acceptance is reserved to the consumer regardless of the findings of the supplier. At -its discretion, the consumer may discontinue at any time the use of the suppliers inspec
19、tion results or determining acceptability of submitted supplies until all discrepancies in the records and inspec- tion procedure6 have been rectified. Failure of -the supplier to correct deficiencies in his inspection system responsible for the discrepancies noted will prompt the Government to take
20、 action to protect and conserve its interests. 1.4 REINSPECTION OF SUPPLIERS SAMPLE -UNITS. Where the specifica- tion prescribes sampling inspection, reinspection of the supplier 6 samp.lze units at the start of production may prove beneficial. adopteil by the consumer are made manihst by his examha
21、tian of rejeczed and borderline units, and the dhecking of inspection aids can be accm- plished. However, cmplete verification of the inspection operatsons The standards -performed by the suppliers inspectors cannot consistently be carried aut. The requirement that the units of the sample be selecd
22、at random without regard to their quality cannot be verified by reinspecting the items drawn by the supplier. Only an independent sample selected by the consumer can authenticate the over-all effect of the suppliers inspection in assuTing conformance of supplies with technical requirements and evalu
23、ate the true , s qualfty of the supplies offered to the consumer for acceptance. 2 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-MIL-HDBK-HLO7 dC a 7777970 0040357 T a .- 2. VALIDATION OF SUPPLIERS INSPECTION 2.1 INDEPENDENT SAMPLING AND INSPECTION
24、. Selection of an independent sample from the lot and performance of product verification inspection pro- vides the consumer with an objective basis for validating a suppliers inspection. The two samples of product from the same lot can be examined for fractions defective and compared for agreement,
25、 making allowance solely for random sampling (chance) variation. A large difference between the proportions of nonconformance in the two samples may lead to a decision to investigate factors suspected of bringing about the disparity in sampling results. Some of the key factors may be failure to sele
26、ct random samples, improper use of inspection aids, and misinterpretation of inspection standards. This handbook provides quantitative criteria based on statistical principles to guide inspectors in conducting verification inspection of independently selected sample units from the lot submitted for
27、consumer inspection and acceptance. 2.2 TEST FOR HOMOGENEITY. When two independent samples are drawn at random from a lot and each item in the samples is classified as con- forming and nonconforming, the fractions defective in the samples may be compared. The comparison is considered a test of homog
28、eneity of the two samples since the concern is whether the percentages of defectives observed would be such as would only occur by chance selection of the sample units, inspection being uniformly accomplished. In other words, the question to be resolved is whether the differences in sampling inspect
29、ion results between the supplier and consumer can be attributed to the “luck of the draw“ in selecting sample units at random from the lot or whether the differences can be assigned to real differences in inspection practice. 3 -. F -. -_ .- - - - - I Provided by IHSNot for ResaleNo reproduction or
30、networking permitted without license from IHS-,-,-.- MIL-HDBK-HL07 OC 9797970 0040160 h m In this handbook, the inspection performed by the consumer will be regarded the standard against which the performance of the supplier will be judged. accordance with the sampling standard. A second sample, pro
31、portional in size to the prescribed sample taken by the supplier will be drawn from the same lot by the consumer. sizes, the consumer ascertains an “action“ number associated with the number of defectives observed and recorded by the supplier and compares the number of defectives found by the consum
32、er with the “action“ number. If the latter equals or exceeds the“action“ number, dc(A), the consumers inspector adopts a course of action on the premise that a discrepancy actually exists in the suppliers inspection system. Each individual lot will be inspected by the supplier in From Table I, for a
33、 specified ratio of sample The critical values for indicating a discrepancy in paired attribute sampling are not to be confused with the “rejection“ numbers of sampling plans in the standard prescribed for determining acceptance of supplies offered to the consumer. arriving at a decision as to the r
34、eliability of a suppliers inspection data from an individual lot submitted for acceptance. provides criteria for arriving at a decision as to the acceptability of the lot offered to the consumer. Accordingly, the decision as to reliability of inspection results is distinct from the decision to accep
35、tor reject a lot although the latter decision may be contingent upon the former. having no reason to impugn the validity of the suppliers inspection records, the consumer assumes that the suppliers inspector is doing Table I provides criteria or The sampling Standard Thus, 4 Provided by IHSNot for R
36、esaleNo reproduction or networking permitted without license from IHS-,-,-MIL-HDBK-HLOS QC 7979770 OOYOZbL B everything that the consumers inspector would do. the consumer can rely upon the suppliers inspection records to support Under such conditions, acceptance of supplies offered to him. performe
37、d by the consumer can thus be reduced to the minimum requited to assure a flow of material meeting acceptance quality standards. 2.3 STATISTICAL CRITERIA FOR PAIRED ATTRIBUTE SAMPLINGS. For con- venience in referring in general terms to the quantities required for the - homogeneity test, the results
38、 of paired attribute samplings are represented symbolically in Table A: The amountof sampling inspection TABLE A Notation for Two-Sample Test for Homogenei-y Defective Effective Sample Size Suppliers Sample dS - d, “S “C n - de C Consumers Sample These data are recorded to decide whether the results
39、 of inspecting two samples, one the suppliers of size ns and the other, the consumers of size nc, which are found to contain d, and dc defectives (or defects), respectively, are significantly different. conveniently tested for statistical significance at approximately the The recorded data can be 5
40、per cent level (one chance in 20 that observed differences might be due to the “luck of the draw“) by means of Table I, which provides the cxical values for various ratios of n8/nc : r. The critical limits of Table I are the Itaction“ nders for de when associated with an observed d, for specified ra
41、tios of ns/nc equal to 1, 2, 3, 5, and 8, respectively. 5 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-2.3.1- lot sampling fnspection in accordance with Military Standard 105 at inspection level II for AQL (Major) = 1.0 per cent defective and AQL
42、(Minor) = 2.5 per cent defectpre. If the supplier produced lots of 1,000 units each and used single sampling, the appropriate sampling plan for raajor defects would be sample size 110 and acceptance number = 3. The supplier performs sampling inspection and records that he found 2 defec- EXAMPLE OF U
43、SE OF TABLE I. Suppose a specification prescribes tivas in the sample of 110 units inspected for majors. inspection, the supplier submits the lot with his inspection record to the consumers inspector. Since this is the first lot, the inspector might appropriately choose to perform product verificati
44、on inspection using the same sample size as the supplier. tion and finds 5 defectives n his sample of 110 units inspected. The sampling results are displayed as follows: After completing his The inspector performs sampling inspec- Defect ive Effect ive Sample Size Suppliers Sample 2 a08 110 Consumer
45、s Sample 5 105 110 Referring to Table I; for r = 1 (since the ratio of the suppliers sample size to the consumers sample size equals i), the inspector finds that, corresponding to the 2 defectives observed by the supplier, the “action“ limit for the number of defectives observed in the consumers sam
46、ple is 7. Since the consumer noted only 5 defectives in his verifica- tion sample, he has no evidence at this time to doubt the validity of the suppliers inspection result for the lot in question. 6 Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-MIL
47、-HDBK-HI07 6lC = 7777770 OOLlOIb3 L = After the consumers inspector has performed verification inspec- tion for the first few lots and has found no significant discrepancies between the suppliers results and his he may reduce his sample size to one-half that of the supplier. (More definite guidance
48、for reducing verification sampling is available from the operating characteristics of the test as explained in the next section of this handbook.) Now, suppose the contractors record for the lot verified showed 3 defectives in a sample of 110 units, but the consumers inspector found 3 defectives in a sample of 55 units from the same lot. suppliers sample size to consumers sample = 2) with the entry ds.= 3 Referring to Table I for r = 2 (ratio of we not
