1、 AEROSPACE RECOMMENDED PRACTICE Statistical Product Acceptance Requirements Using Continuous Sampling, Skip-Lot Sampling, or Methods for Special Cases SAE Technical Standards Board Rules provide that: “This report is published by SAE to advance the state of technical and engineering sciences. The us
2、e of this report is entirely voluntary, and its applicability and suitability for any particular use, including any patent infringement arising therefrom, is the sole responsibility of the user.” SAE reviews each technical report at least every five years at which time it may be reaffirmed, revised,
3、 or cancelled. SAE invites your written comments and suggestions. Copyright 2005 SAE International All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwi
4、se, without the prior written permission of SAE. TO PLACE A DOCUMENT ORDER: Tel: 877-606-7323 (inside USA and Canada) Tel: 724-776-4970 (outside USA) Fax: 724-776-0790 Email: custsvcsae.org SAE WEB ADDRESS: http:/www.sae.org Issued 2005-10 ARP9013/4 FOREWORD To assure customer satisfaction, aerospac
5、e industry organizations shall produce, and continually improve, safe, reliable products that meet or exceed customer and regulatory authority requirements. The globalization of the aerospace industry, and the resulting diversity of regional/national requirements and expectations, has complicated th
6、is objective. End-product organizations face the challenge of assuring the quality of, and integrating, product purchased from suppliers throughout the world and at all levels within the supply chain. Aerospace suppliers and processors face the challenge of delivering product to multiple customers h
7、aving varying quality expectations and requirements. The aerospace industry established the International Aerospace Quality Group (IAQG) for the purpose of achieving significant improvements in quality and safety, and reductions in cost throughout the value stream. This organization includes represe
8、ntation from aerospace companies in the Americas, Asia/Pacific, and Europe. This recommended practice has been prepared by the Americas Aerospace Quality Group (AAQG). This document standardizes, to the greatest extent possible, the requirements for product acceptance using continuous sampling, skip
9、-lot sampling, and special case statistical methods in the aerospace industry. The establishment of common requirements, for use at all levels of the supply-chain by organizations around the world, should result in improved quality and safety, and decreased costs due to the elimination or reduction
10、of organization-unique requirements and the resultant variation inherent in these multiple expectations. SAE ARP9013/4 - 2 - TABLE OF CONTENTS FOREWORD.1 1. SCOPE 3 1.1 Purpose3 1.2 Field of Application.3 2. REFERENCES.3 2.1 Applicable Documents .3 2.2 Related Publications 4 2.3 Definitions 4 3. GEN
11、ERAL REQUIREMENTS 4 4. CONTINUOUS SAMPLING .4 4.1 Quality Parameter 5 4.2 Inspection Sequence5 4.3 General Conditions 5 5. SKIP-LOT SAMPLING .5 5.1 General Conditions 5 5.2 Quality Protection Equivalency 5 5.3 Considerations for Subsequent Lots6 6. SAMPLING STRATEGIES FOR SPECIAL CASES.6 6.1 First a
12、nd Last of a Lot Produced by a Continuous Manufacturing Process .6 6.2 Predicting Quality From Correlated Variables7 6.3 Chain Sampling Inspection Plan 7 6.4 Sequential Sampling 8 6.5 Other Methods for Special Cases 8 APPENDIX A CONTINUOUS SAMPLING PLAN GUIDELINES.9 APPENDIX B EXAMPLES OF SPECIFIC S
13、AMPLING CONCEPTS 45 APPENDIX C GUIDELINES FOR SATISFYING REQUIREMENTS USING OTHER METHODS47 SAE ARP9013/4 - 3 - 1. SCOPE: 1.1 Purpose: The purpose of ARP9013/4 is to ensure conformance for each product characteristic being verified under these methods. ARP9013/4 provides the requirements for continu
14、ous sampling, skip-lot sampling, and for statistical methods for product acceptance that do not derive from industry standard tables. 1.2 Field of Application: This recommended practice is applicable when accepting product based on continuous sampling, skip-lot sampling, or special statistical metho
15、ds not documented in other industry standards. Several of these sampling strategies are not currently included in any other released standards, but are provided as valuable alternatives for efficiently meeting technical requirements. The customer may require notification or approval prior to use of
16、these special sampling techniques. 2. REFERENCES: 2.1 Applicable Documents: The following publications form a part of this document to the extent specified herein. The latest issue of SAE publications shall apply. The applicable issue of other publications shall be the issue in effect on the date of
17、 the purchase order. Nothing in this document, however, supersedes applicable laws, regulations, and contractual requirements. 2.1.1 SAE Publications: Available from SAE, 400 Commonwealth Drive, Warrendale, PA 15096-0001, Tel: 877-606-7323 (inside USA and Canada) or 724-776-4970 (outside USA), www.s
18、ae.org. ARP9013 Statistical Product Acceptance Requirements ARP9013/1 Statistical Product Acceptance Requirements Using Isolated Lot Sampling Methods ARP9013/3 Statistical Product Acceptance Requirements Using Process Control Methods 2.1.2 ANSI/ASQC Publications: Available from the American Society
19、for Quality (ASQ), 600 North Plankinton Avenue, Milwaukee, WI 53203, Tel: 800-248-1946 or 414- 272-8575, www.asq.org. ANSI/ASQC S1 An Attribute Skip-Lot Sampling Program SAE ARP9013/4 - 4 - 2.1.3 Applicable References: Available from the Document Automation and Production Service (DAPS), Building 4/
20、D, 700 Robbins Avenue, Philadelphia, PA 19111-5094, Tel: 215-697-6257, http:/assist.daps.dla.mil/quicksearch/ MIL-STD-1235 Single and Multi-Level Continuous Sampling Procedures and Table for Inspection by Attributes NOTE: MIL-STD-1235 has been cancelled but may be used as a guide for derivation and
21、application of continuous sampling plans. Appendix A herein contains the essential text and tables of MIL-STD-1235C with some editing, reformatting, and paragraph renumbering; however, all the functional curves have been purposely omitted to conserve the length of this recommended practice. 2.2 Rela
22、ted Publications: H. F. Dodge, “Chain Sampling Inspection Plan,“ Industrial Quality Control, 11, No. 4 (January 1955), pages 10-13 Abraham Wald, “Sequential Analysis” (New York: John Wiley (2) the required objective evidence shall be reestablished when a lot is rejected and there is no documented ev
23、idence of a special cause with corrective action; and (3) any skipped lots shall be retrievable at the point in time when the results of the next inspected lot become known. See Appendix B for guidelines. 5.2 Quality Protection Equivalency: Each skip-lot sampling plan shall show its AOQL. If the qua
24、lity requirement is stated as an AQL, then Table 1 above provides the equivalent AOQL requirement. SAE ARP9013/4 - 6 - 5.3 Considerations for Subsequent Lots: As with other continuous sampling plans, if quality results dictate an increase in inspection, then the units on which that inspection would
25、be performed shall be inspectable. For that reason, if there is a time delay for an inspection process, all subsequent lots presented for inspection during that time shall be held where they are available for the increased inspection if that becomes necessary. NOTE: This stipulation may place additi
26、onal storage and inventory information requirements on the skip-lot inspection process. 6. SAMPLING STRATEGIES FOR SPECIAL CASES: The sampling strategies in this section are provided as valuable alternatives for efficiently meeting technical requirements. The customer may require notification or app
27、roval prior to use of these special sampling techniques. See ARP9013 paragraph 4.8. In each case, if the organization chooses to use sampling strategies for special cases, the organization shall demonstrate the statistical validity of these approaches before they are used. This section addresses wha
28、t is needed to establish that statistical validity. 6.1 First and Last of a Lot Produced by a Continuous Manufacturing Process: Statistical evidence for using autoregressive or auto-correlated statistical analysis, if used, shall compare the prediction interval for the last measurement in a lot to s
29、pecifications when the first measurement is given. The time series model used for analysis shall be documented together with parameter estimates, and prediction intervals. See Appendix B for guidance. 6.1.1 Showing that any “first and last” model applies shall require showing the exact sequence of u
30、nit quality in lots that are of mixed quality. Attribute data from lots that are 100% conforming or 100% nonconforming does not demonstrate that the first and last units in the lot are more likely to be nonconforming than the others. 6.1.2 When doing this on an attribute basis, the number of mixed l
31、ots in which the sequence of conforming and nonconforming units is shown depends on both the quality requirement and the degree to which the actual data matches the model. The model shall show evidence that among lots in which the fraction nonconforming equals the RQL, the fraction of those lots whi
32、ch would pass this sample (that is, that have both the first and last units good) shall be less than or equal to 5%. SAE ARP9013/4 - 7 - 6.2 Predicting Quality From Correlated Variables: The quality of a product feature may be accepted on the basis of measurements taken from something else if the me
33、asurements are shown to be predictive. Under these conditions, there are two sets of characteristics, and both shall be measurable, or countable with large enough counts to be treated as measurable (for example, particle counts). One of the sets of characteristics is directly subject to engineering
34、requirements, and shall be accepted as conforming before the product having those characteristics can be used to produce aerospace deliverables. This is the “dependent” set of characteristics. The other set, the “independent” characteristics, may or may not be directly subject to engineering require
35、ments, but is related so closely to the first set that their values can be used to predict the acceptability of the first set. The conditional probability of the dependent characteristics being acceptable, given the measure or count of the independent characteristics, shall satisfy the required qual
36、ity parameter. Studies showing this conditional probability shall be sufficient to understand the ongoing relationship of such parameters and their resulting influence on the deliverable property or characteristic. An adequate relationship shall show that the probability is high (relative to the qua
37、lity parameter) that the inspected feature will predict when the uninspected feature is nonconforming. Records of these correlation studies or other evidence of validity shall be maintained for as long as this method is used in the acceptance of product. See Appendix B for an example of this concept
38、. 6.3 Chain Sampling Inspection Plan1: This is a special type of lot-by-lot acceptance sampling plan for attributes when destructive or costly tests are required to determine quality characteristics. The sample size will be small out of necessity. Chain sampling plans use cumulative results of sever
39、al preceding samples. When using a chain sampling technique the following conditions shall be met: 6.3.1 This applies when the lots of a continuing series of products are sampled in substantially the order of production. 6.3.2 Customers can normally expect the lots to be essentially the same quality
40、. Provision for an occasional nonconformity is satisfactory for major or minor classifications, but not for safety/critical classifications. 1H.F. Dodge, “Chain Sampling Inspection Plan,“ Industrial Quality Control, 11, No. 4 (January 1955), pages 10-13 SAE ARP9013/4 - 8 - 6.4 Sequential Sampling: S
41、equential Sampling2is used for destructive or costly tests, with a subgroup size of 1, thereby making it an item-by-item plan. Item-by-item sequential sampling is based on the concept of the sequential probability ratio test (SPRT). 6.5 Other Methods for Special Cases: When a sampling strategy for a
42、 special case is developed that is not described in ARP9013/4, it shall satisfy the requirements of ARP9013. Its mathematical basis shall be documented and available for review. See Appendix C for examples and guidance. PREPARED UNDER THE JURISDICTION OF SAE COMMITTEE G-14, AMERICAS AEROSPACE QUALIT
43、Y GROUP (AAQG) 2Abraham Wald, “Sequential Analysis” (New York: John Wiley or (2) the assembly of small lot sizes greatly increases the amount of inspection and thus results in increased inspection costs. The customer may specify other retrievability requirements than shown in 4.3, depending on their
44、 usage of the product. SAE ARP9013/4 - 10 - A.4 CONTINUOUS SAMPLING PLANS: This Appendix section contains the essential text and tables of MIL-STD-1235C with some editing, reformatting, and paragraph renumbering; however, all the functional curves have been purposely omitted to conserve the length o
45、f this recommended practice. In doing so, this section provides the tables and procedures for applying five different types of continuous sampling plans for inspection by attributes: CSP-1, a single-level continuous sampling procedure, which provides for alternating between sequences of 100% inspect
46、ion and sampling inspection. CSP-F, a variation of the CSP-1 plans in that CSP-F plans are applied to a relatively short run of product, thereby permitting smaller clearance numbers to be used. CSP-2, a modification of CSP-1 in that 100% inspection resumes only after a prescribed number of defect-fr
47、ee units separate any two nonconforming sample units. CSP-T, a multi-level continuous sampling procedure that provides for reducing the sampling frequency upon demonstration of superior product quality. CSP-V, a single-level continuous sampling procedure that is an alternative to CSP-T in that these
48、 plans provide for reducing the clearance number in good quality situations where reduction of sampling frequency has no economic merit. A.4.1 Scope: A.4.1.1 Purpose: This Appendix establishes continuous sampling plans and procedures for inspection by attributes. A.4.1.2 Applicability: The condition
49、s that are recommended to exist before the sampling plans can be used are: a. Moving product (see ARP9013 paragraph 2.3). b. Ample space, equipment and manpower at or near the site of inspection to permit rapid 100% inspection when required. c. Relatively easy and quick inspection. d. A process that is producing, or is capable of producing, material whose quality is stable. See A.4.7.2.b.i, A.4.8.2.b.i, A.4.9.2.b.i, A.4.10.2.b.i, and A.4.11.2.b.i. e. The inspection is non-destructive. SAE ARP9013/4 - 11 - A.4.1.2 (Continued): The sampling plans designated herein
