Sampling inspection permits the estimation of the overall quality of a group of product articles through the inspection of a relatively small number of product articles drawn from the group.

The specification of a sampling plan provides purchasers and sellers a means of identifying the minimum quality level that is considered to be satisfactory.

Because sampling plans yield estimates of the quality of a product, the results of the inspection are subject to error. Through the selection of a sampling plan, the potential error is known and controlled.

Sampling inspection is used when a decision must be made about what to do with a quantity of articles. This quantity may be a shipment from a supplier, articles that are ready for a subsequent manufacturing operation, or articles ready for shipment to a customer.

In sampling inspection, a relatively small number of articles (the sample) is selected randomly from a larger number of articles (the inspection lot); the sample is inspected for conformance to the requirements placed on the articles. Based on the results, a decision is made whether or not the lot conforms to the requirements.

Since only a portion of a production lot is inspected, the quality of the uninspected articles is not known. The possibility exists that some of the uninspected articles are nonconforming. Therefore, basic to any sampling inspection plan is the willingness of the buyer to accept lots that contain some nonconforming articles. The number of nonconforming articles in accepted lots is controlled by the size of the sample and the criteria of acceptance that are placed on the sample.

Acceptance sampling plans are used for the following reasons:

When the cost of inspection is high and the consequences of accepting a nonconforming article are not serious.

When 100 % inspection is fatiguing and boring and, therefore, likely to result in errors.

When inspection requires a destructive test, sampling inspection must be used.

In acceptance sampling by variables, the coating characteristic of each article in the sample is measured. Using the arithmetic mean of these values, the standard deviation of the process, and the factor k that is found in the Tables, a number is calculated (see 9.3). If this number equals or exceeds the specified minimum, the inspection lot conforms to the requirements. If it is less, the lot does not conform. If the standard deviation of the process is not known, the standard deviation of the sample is calculated and used.

The use of a sampling plan involves the balancing of the costs of inspection against the consequences of accepting an undesirable number of nonconforming articles. There is always a risk that a random sample will not describe correctly the characteristics of the lot from which it is drawn, and that an unacceptable lot will be accepted or an acceptable lot will be rejected. The larger the sample, the smaller this risk but the larger the cost of inspection.

To understand the risks, consider that if every article in an inspection lot conforms to its requirements, every article in the sample will conform also. Such lots will be accepted (Note 1). If only a few articles in an inspection lot are nonconforming, the sample probably will indicate that the lot is acceptable; but there is a small probability that the sample will indicate that the lot is unacceptable. The larger the proportion of nonconforming articles in an inspection lot, the more likely it will be that the sample will indicate that the lot is unacceptable. If every article in an inspection lot is nonconforming, a sample will always indicate that the lot is unacceptable.

Note 1—Throughout this method, it is assumed that no mistakes are made in sampling, measurement, and calculation.

The probability of accepting an inspection lot that contains nonconforming items is often described in terms of the Acceptable Quality Level (AQL) and the Limiting Quality Level (LQL). The AQL is the quality level that is considered to be acceptable. The LQL is a quality level that is considered to be barely tolerable. A sampling plan is selected that has a high probability of accepting lots of AQL quality and of rejecting lots of LQL quality. In this method, the AQL given for a sampling plan is the quality level of lots (expressed as the percentage of nonconforming articles) that have a 95 % probability of being accepted. The LQL is the quality level of lots that have a 10 % probability of being accepted or, in other words, a 90 % probability of being rejected. The tables in this method give the AQL and LQL of each plan. They also give the 50/50 point, the quality level of a lot that is just as likely to be accepted as rejected.

The disposition of nonconforming inspection lots is beyond the scope of this method because, depending on the circumstances, lots may be returned to the supplier, kept and used, put to a different use, scrapped, reworked, or dealt with in some other way. An alternative is rectifying inspection in which rejected lots are screened and used.

In rectifying inspection, when an inspection lot is rejected, all of the articles in the lot are inspected and nonconforming ones are removed. They may be replaced with conforming articles. The now 100 % conforming lot is accepted. With this practice, the average quality level for a series of lots taken as a whole will be better because of the addition of the 100 % conforming lots. When the incoming lots are of a good quality level, the average quality level of a series of lots will be even better when the rejected lots are screened and resubmitted. When incoming lots are of a poor quality level, the average quality of a series of accepted lots will again be good because many of the incoming lots will be rejected and upgraded. At intermediate quality levels of incoming lots, the average quality level of a series of accepted lots will again be improved, but it will not be improved as much as in either of the above cases; and there will be an intermediate quality level where the degree of improvement is the least. This improved quality level is called the Average Outgoing Quality Limit (AOQL). It is the worst condition that can occur under rectifying inspection. The tables give the AOQL for each plan. There is no AOQL for the plans used with destructive tests because destructive tests cannot be used to screen rejected lots.

Note 2—The AOQLs given in the tables are strictly correct only when the sample is small with respect to the lot. If this is not the case, the correct AOQL will be smaller than the tabulated value. The correct values are obtained by multiplying the tabulated values by the following equation:

1.1 This test method provides sampling plans that are intended for use in the inspection of metallic and inorganic coatings on products for the purpose of deciding whether submitted lots of coated products comply with the specifications applicable to the coating.

1.2 The sampling plans are variables plans. In plans of this type, several articles of product are drawn from a production lot. A characteristic of the coating on the drawn articles is measured. The values obtained are used to estimate the number of articles in the lot that do not conform to a numerical limit, for example a minimum thickness. The number is compared to a maximum allowable.

1.3 Variables plans can only be used when the characteristic of interest is measurable, the test method gives a numerical measure of the characteristic, and the specification places a numerical limit on the measured value. It is also necessary that the variation of the characteristic from article to article in a production lot be normally distributed (see Appendix X2). Each article must be tested in the same way (for example, coating thickness must be measured at the same location, see X2.7) so that the values from article to article are comparable. If one or more of these conditions are not met, a variables plan cannot be used. Instead, an attributes plan must be used. These are given in Test Method B602 and Guide B697.

1.4 This standard does not purport to address all of the safety concerns, if any, associated with its use. It is the responsibility of the user of this standard to establish appropriate safety and health practices and determine the applicability of regulatory limitations prior to use.