Standards Column

Iso/ts 16949:2002—Quality Management Systems—particular Requirements For The Application Of Iso 9001:2000 For Automotive Production And Relevant Service Part Organizations

Earlier issues of Quality Engineering (1 2) discussed the first edition of ISO/TS 16949, which was based on ISO 9001:1994. It was also mentioned that if the project plans of ISO TC 176 were carried out as expected, a second edition of the technical specification (TS) would be developed based on ISO 9001:2000. The second edition was completed in the first quarter of 2002 and has remarkable alignment with ISO 9001:2000. As with the first edition, it includes all of the requirements of ISO 9001:2000 verbatim in boxed text, with additional automotive industry requirements or notes outside the boxes. The TS states that its goal is “…the development of a quality management system that provides for continual improvement, emphasizing defect prevention and the reduction of variation and waste in the supply chain.” The TS, along with applicable customer-specific requirements, defines the fundamental quality management system requirements for those subscribing to this document. The TS also states that it “…is intended to avoid multiple certification audits and provide a common approach to a quality management system for automotive production, and relevant service part organizations.” The TS specifies that it is intended to define the quality management system requirements for design and development, production, and, when relevant, installation and service of automotive-related products. In addition to being applicable to sites where parts are made, its applicability extends to off-site supporting functions such as corporate design centers.

A note in clause 0.3 of TS 16949:2003 states the expectation that knowledge and use of the eight quality management principles (stated in ISO 9000:2000 and ISO 9004:2000) should be “…demonstrated and cascaded … by top management.” These eight principles were reviewed in an earlier issue of Quality Engineering (3). While this expectation is only in an explanatory note, it is significant that an industry like automotive would recognize that ideas such as involvement of people, leadership, and a factual approach to decision making are important to achieving good quality.

Of course, a review of the new TS reveals additional requirements beyond the basics in ISO 9001:2000, but there is a much smaller gap between the requirements of TS 16949:2002 and ISO 9001:2000 than there was between QS 9000 and ISO 9001:1994. In fact, ISO 9001:2000 includes a number of things that were recommended by the automotive industry during its development.

TS 16949:2002 includes the definitions of 12 additional terms that do not appear in ISO 9000:2000 or ISO 9001:2000. These terms are listed in Table 1.

Table 1. Terms defined in ISO TS16949, second edition, and not found in ISO 9000:2000

Control plan	Predictive maintenance
Design responsible organization	Preventive maintenance
Error proofing	Premium freight
Laboratory	Remote location
Laboratory scope	Site
Manufacturing	Special characteristic

While a number of these terms seem to be very general in nature (like the word manufacturing), most have very specific meanings in the automotive industry and their definition can be useful in clarifying the industry's requirements.

It is also surprising that TS 16949:2002 only adds two requirements for documentation to the six specific procedures required by ISO 9001:2000. Of course, ISO 9001:2000 itself requires that extent of documentation be dependent on the size of the organization, complexity of its processes and their interactions, competency of its people, etc. Many automotive suppliers have quite complex processes that necessitate more documentation. The two situations where the TS requires documentation not called for in ISO 9001 are:

•	clause 6.2.2.2, which requires a documented procedure for identification of training needs and achieving competency, and
•	clause 7.6.3.1, which requires that the scope for internal laboratories be included in the quality management system documentation.

There are also several areas where work instruction level documents are specifically called out in the TS but not in ISO 9001:2000. The major ones are:

•	clause 7.5.1.2: Documented work instructions for all responsible for operating processes impacting product quality;
•	clause 7.5.1.4: Documented maintenance objectives; and
•	clause 8.2.3.1: Documented process capability, reliability, maintainability, and requirements for production, measurement, and test.

ISO 9001:2000 requires continual improvement of the quality management system and so does the new TS. However, TS 16949:2002 goes farther by requiring:

•	a defined process for continual improvement of the organization and
•	manufacturing process improvement focusing on control and reduction of variation in product characteristics and manufacturing process parameters.

There are a number of other additional requirements or clarifications. Some of these have been in the earlier automotive industry requirements, and some could be considered to be obvious or very minor. For example, clause 4.2.3.1 on engineering specifications is added to require the organization to “… have a process to assure the timely review, distribution and implementation of all customer engineering standards/specifications and changes based on customer-required schedule. Timely review should be as soon as possible, and shall not exceed two working weeks.” While the first part—having a process for review of requirements—appears obvious and covered by ISO 9001:2000, the requirement that it be done within 2 working weeks may not be trivial. Other additional requirements just needed to be included for clarity. For example, clause 1.2 of TS 16949 specifies that an organization can only exclude clause 7.3 on design and development if not design responsible. Organizations will need to read the document carefully, as they make the transition to the new requirements. The message is “read the details.”

It can be anticipated that over time most of the automotive-specific documents based on ISO 9001 and developed in various countries (such as VDA 6.1 in Germany and QS 9000 from Ford, GM, and Chrysler) will be replaced as the basis for quality management system registration by ISO TS 16949.

Iso/iwa 2, International Workshop Agreement 2—Quality Management Systems—guidelines For The Application Of Iso 9001:2000 On Education

This document is ISO's second IWA. An earlier issue of Quality Engineering (4) discussed the development of ISO's first IWA, i.e., IWA 1 on quality management systems—Guidelines for process improvement in healthcare. The basic idea behind an IWA was covered earlier. It is important to remember that IWAs are not international standards.

The proposal for this IWA came from DGN (Direccion General de Normas), the Mexican member body of ISO and was approved by the ISO Technical Management Board. The workshop was hosted by DGN and INLAC (the Latin American Institute for Quality) October 18–20, 2002, in Mexico. The workshop was attended by about 400 people from 14 countries.

The draft resulting from the workshop states that its purpose is to provide guidelines for organizations to implement an effective quality management system that meets ISO 9001:2000. It includes the full text of IOS 9001:2000, Quality management systems—Requirements, and guidance for implementing those requirements in educational institutions. The draft also includes text from ISO 9004:2000, Quality management systems—Guidelines to performance improvement, for organizations that might wish to go beyond the requirements of ISO 9001 with a focus on performance improvement. The scope statement in the draft makes it clear that the intent is to provide guidance and not to change, modify, or add to the requirements of ISO 9001:2000.

The process for creating IWA 2 was similar to the one used for IWA 1, but some of the lessons learned in the IWA 1 process made this one appear to go more smoothly. Dr. Craig Johnson, who participated in this work from start to finish, described the process as follows: “A first draft of IWA-2 was circulated at the end of September 2002 among potential delegates to the workshop.

A second draft, incorporating comments from more than 15 countries on the first draft, was produced and circulated as a working document for use during the Acapulco workshop.

The working document, including comments from ISO Technical Committee (TC) 176 members, was given to four workshop task groups. Each task group was assigned one or two clauses of the working draft and was asked to produce a revised draft that addressed international needs. The working document, including comments from ISO Technical Committee (TC) 176 members, was given to four workshop task groups. Each task group was assigned one or two clauses of the working draft and was asked to produce a revised draft that addressed international needs. A broad consensus led to approval of a final draft of IWA-2 on October 20, 2002.

An Editing Group was established with representatives from France, Germany, Kenya, Mexico, Russia, the United Kingdom and United States and provided a revised version.”

This development work was followed by a ballot of the workshop participants based on the edited draft. It is expected that the document will be issued in early 2003.

Iso/fdis 10576-1—Statistical Methods—guidelines For The Evaluation Of Conformity With Specified Requirements—part 1: General Principles

An earlier issue of Quality Engineering (5) mentioned a new work item proposal (NWIP) that was to expand this project to include a part 2 on uncertainty intervals and probabilities of correct and incorrect declarations of conformity. That expansion has not yet happened, and the fate of the proposed part 2 remains in question. Part 1 has reached the final draft international standard (FDIS) stage. Part 1 is a guide for taking into account the uncertainty of the entire measurement process in making decisions about conformity or nonconformity. It does not provide the methods for calculating measurement uncertainty; that is to be accomplished using the rules of the Guide to the Expression of Uncertainty in Measurement (GUM). It also does not deal with determining the sources of uncertainty in the overall measurement system. As an aid in identification of components of uncertainty, ISO 5725 may be used. Rather, ISO 10576 deals with how specification tolerance limits should be specified and provides rules for using the calculated measurement uncertainty in the determination of conformity or nonconformity. It does not deal with acceptance sampling, but when sampling is used, it indicates that uncertainties due to the sampling plan are to be included as one of the sources of uncertainty.

In addition, ISO 10576 explains the concept of setting limiting values for a characteristic of an item that will be measured in some way to determine conformity. One-sided and double specification limits are described and examples given. Two methods are given (two-stage and one-stage) for making conformity or nonconformity decisions using the actual measured value and the calculated measurement uncertainty.

In the two-stage process, a first test is run and measurement uncertainty applied to make an initial determination of conformity or nonconformity. Conformity to the requirements may be assured if, and only if, the uncertainty interval of the measurement result is inside the established region of permissible values. If this is the case, stage two is not conducted. Stage two is conducted if, and only if, the uncertainty interval calculated after the first stage includes a specification limit. In this case, a second test is run and the results averaged with stage one. Conformity may be assured after stage two if, and only if, the uncertainty interval of the final measurement result is inside the region of permissible values. Thus, the two-stage process favors not making an erroneous determination of nonconformity. Note that ISO 10576 indicates that the two-stage method is preferred. Obviously, this has the effect of increasing consumer's risk. If this increase in consumer's risk is unacceptable, then a one-stage process should be used.

In the one-stage process, the test is conducted in a manner similar to stage one above, but the criteria for determining conformity or nonconformity are slightly different and include the possibility that the result would be inconclusive.

The document does have appeal, at least from an academic point of view. On the other hand, its usefulness in day-to-day decision making may be questionable. The need to refer to other documents such as the GUM and ISO 6725 makes it cumbersome to use.

Standards Related to Sampling

In this issue we continue our review of standards related to sampling. This time we will look at two standards related to acceptance sampling based on allocation of priorities.

Iso/cd 13448-1—Acceptance Sampling Procedures Based On The Allocation Of Priorities Principles (App)—part 1: Guidelines To The App Approach, And Iso/cd 13448-2—Acceptance Sampling Procedures Based On The Allocation Of Priorities Principles (App)—part 2: Coordinated Single Sampling Plans For Acceptance Sampling By Attributes

An earlier issue of Quality Engineering (6) discussed the project to develop these standards based on earlier Russian documents. That column pointed out several difficulties with the proposal, such as:

•	use of terminology that differs from other ISO standards on acceptance sampling;
•	their inclusion of only sampling plans for relatively poor quality levels (0.15% being the highest level included);
•	the ability to expend effort on such a document at a time when work was needed on integrating existing sampling standards and address acceptance at ppm levels.

The project has continued at a slow pace and does not seem to have interfered with other work that could be considered to be of higher priority. On the other hand, most of the objections remain in the latest draft.

Parts 1 and 2 of ISO/CD 13448 cover sampling using a priori information on the supplier's capabilities to meet quality requirements as an input to defining initial data for choosing sampling plans. Part 1 gives the general guidance for use of the APP approach, while Part 2 gives the actual APP attributes sampling plans and instructions for their use. The concept is based on the observation that there is often a great deal of a priori information available that may give indications of the quality of lots. Such information may include the supplier's quality history, the existence of data on statistical process capability and control, evidence that the supplier has an effective quality management system, inspection data from product or process development, the application criticality of the items to be inspected, etc. Using the information can provide a means to allocate inspection resources where they are needed most. A premise of the document is that ignoring a priori information can both increase cost and actually fail to provide the needed level of consumer protection. Note that ISO/CD 13448-1 states that the ISO/CD 13448 sampling system may assist in situations where:

1.	“ inspection is first conducted by the supplier on final inspection and then for the same lot by the consumer on incoming inspection (occasionally by a third party);
2.	there is a long-term relationship between the producer and the consumer;
3.	a priori information about the supplier's capabilities to meet specified requirements is, or is not, available;
4.	the supplier's responsibility for a quality guarantee involving a sampling inspection was agreed upon in the contract;
5.	both parties are interested in reducing the cost of inspection.”

The establishment and agreement between the customer and supplier of lot quality requirements is a prerequisite to using the standard, and these requirements are to be specified in terms of the required normative quality level (NQL). The NQL is defined as “limiting value of the lot quality level specified for the purpose of acceptance as a guaranteed lot quality level.” In concept, a limiting quality (LQ) can also be thought of as a guaranteed lot quality, but it is assured only by the sampling plan having a low probability of acceptance when the lot is of the LQ. A specified NQL, on the other hand, is considered as a guarantee of lot quality not only due to the sampling plan but also through other evidence based on the a priori information. The ISO/CD 13448-1 states that a sampling plan for LQ is used when there is an a priori distrust of lot quality, while NQL plans can take into account a priori information that increases confidence in lot quality. Also, ISO/CD 13448-1 clearly states that the assignment of an NQL is not license to ship nonconforming items. The supplier must be fully responsible for any nonconformities found, even if a lot with one or more nonconformities is accepted.

This NQL system is based on the notion that sampling conducted by the supplier is done for the purpose of demonstrating to the customer (or a third party) the validity of information about the adequacy of lot quality. Therefore, sampling by the customer can be viewed as a means to demonstrate the inadequacy of the supplier's information. But if this information is false, it is often more important to get the supplier to fix the quality system than it is to do sampling inspection. The ISO/CD 13448-1 appears to recognize this by stating: “Normally when using the sampling system of ISO 13448 a consumer inspection is not required. It is far more effective to audit a supplier's acceptance sampling procedure and quality system.”

There are of course constraints. There are constraints on consumers' risks for supplier inspection and constraints on the supplier's risks on consumer inspection. Protection of the opposite party's interests is provided by constraints on the sampling plans and on constraints on the schemes by which decisions on lot conformity are made.

Two types of constraint systems are provided—type I and type II. For type I constraints, sampling plans are selected in a conventional manner based on agreed criteria and acceptance decisions made based on rules that are in the permissible inspection plans given in ISO/CD 13448-2.

•

Type I constraints applied to supplier inspection-—For type I constraints, the supplier's permissible sampling plans are defined by the agreed NQL, trust risk level, and thus a consumer's risk on supplier inspection (β 0), and the lot size. For supplier inspection, sampling plans can be selected based on levels of a priori “trust risk level” in the supplier's capabilities, and ISO/CD 13448-1 contains tables giving 7 and 10 levels of trust with associated consumer risk and confidence level. The supplier uses only his own optimality criteria when selecting among permissible sampling plans given in the tables of Annex A to ISO 13488-2. For sampling a continuing series of lots, there are provisions for using switching rules.

•

Type I constraints applied to consumer inspection—For consumer or third-party inspection, the supplier's risk on consumer's inspection should always be specified in the contract and is normally 0.01, 0.05, or 0.1. If the supplier's risk on consumer's inspection is not specified in the contract, ISO/CD 13448-1 requires that the value 0.05 be used. The consumer uses both his own objectives and optimality criteria as well as the specified supplier's risk on consumer inspection.

For type II constraints, each party selects his own rules, which may be different. Here, the permissible sampling plans are defined as constraints on confidence levels when decision rules involve confidence limits (intervals or regions) for the lot quality level. In this case, supplier and consumer select sampling plans with regard to their own objectives, abilities, and optimality criteria. Decisions on lot acceptance are made using decision rules based on confidence intervals, and different decision rules are applicable for the supplier and the consumer. It is common with AQL acceptance plans for the consumer and supplier to agree on a full sampling plan in attempts to avoid getting different results. With ISO/CD 13448, customer and supplier can actually select their own detailed sampling scheme, and they can be different. In fact, they have flexibility to select sample sizes that suit their own situations and resources.

While the concept behind the document seems to have merit, the latest draft does not resolve the issues identified nearly 3 years ago. It is also fairly complex and difficult to use. A comment by India sums it up: “This document was highly mathematical and would be difficult for the common user to understand and implement. Further, existing ISO 2859-1 might be used with zero acceptance as a substitute for this document.” If quality is consistent and switching rules are applied, more conventional approaches like those in ISO 2859 can be effective and economical (even without going to zero acceptance numbers).

There is an even more important reason that the CD may prove problematic. Collection of truly appropriate a priori information and using it to establish sampling plans is a difficult and time-consuming activity. In the explanation and examples, one of the most prominent pieces of a priori information mentioned is the supplier's achieving ISO 9000 certification or registration. While this may be a useful piece of information, there is inference that it could be the only information considered. It is not the overall quality management system registration that is important but rather how that system is applied to the specific items being sampled. It is important to consider a number of factors, including the production processes, measurement and monitoring of those processes, capability of those processes, maintenance to ensure capability is maintained, etc. This could lead users to change their sampling based only on knowledge that a supplier holds ISO 9001 registration (a piece of information that is easy to get) without even thinking about other, more detailed information that actually should be considered in making such a decision. While, ISO 13448-1 and 13448-2 may yet become published standards, their usefulness is questionable.

Acknowledgments