ASTM D1898-68(1989)

NOTICE: This standard has either been superseded and replaced by a new version or discontinued.
Contact ASTM International (www.astm.org) for the latest information.
Designation: D 1898 – 68 (Reapproved 1989)
AMERICAN SOCIETY FOR TESTING AND MATERIALS
100 Barr Harbor Dr., West Conshohocken, PA 19428
Reprinted from the Annual Book of ASTM Standards. Copyright ASTM
Standard Practice for
Sampling of Plastics
This standard is issued under the fixed designation D 1898; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (e) indicates an editorial change since the last revision or reapproval.
This standard has been approved for use by agencies of the Department of Defense.
1. Scope
Specific 12 to 15
Apendixes:
1.1 This practice is primarily a statement of principles to
Glossary X1
guide purchasers of plastic materials purchased under specifi-
Inspection by Attributes X2
Inspection by Variables X3
cations to prepare sampling plans that will describe sampling
Example of Use of Previous Data for Judging the
procedures and that will enable them to determine within
Qualities of Lots and for Setting Specification
practical limits whether or not the products meet the specifi- Limits X4
cations.
1.8 The values stated in SI units are to be regarded as the
1.2 The same principles may be used to guide the prepara-
standard.
tion of specifications in quantitative terms.
1.9 This standard may involve hazardous materials, opera-
1.3 Some of the same principles may be used to determine
tions, and equipment. This standard does not purport to
the actual quality of a product, including its average and
address all of the safety problems associated with its use. It is
variation above and below that average, with respect to a
the responsibility of the user of this standard to establish
particular property.
appropriate safety and health practices and determine the
1.4 Since the design of probability sampling depends upon
applicability of regulatory limitations prior to use.
the ultimate use of the samples and the resulting data, consid-
eration must be given to the intended inspection of the samples.
2. Referenced Documents
Hence, the design of plans for examination and testing of
2.1 ASTM Standards:
samples for both their attributes and their variables is included.
E 122 Practice for Choice of Sample Size to Estimate the
1.5 This practice is intended for general guidance when 2
Average Quality of a Lot or Process
little information is available on the variability of the material
2.2 Military Standards:
and of the method of inspection. In some cases, quality control
MIL-STD-105 Sampling Procedures and Tables for Inspec-
chart methods may be substituted for the procedures herein. In
tion by Attributes
any event, a statistician should be consulted if difficulty is
MIL-STD-414 Sampling Procedures and Tables for Inspec-
encountered in applying the recommendations herein to the
tion by Variables for Percent Defective
design of a specific sampling problem.
1.6 There is no intent to guide the disposition of material
3. Significance and Use
that is found to be off-specification by sampling and inspection;
3.1 The purpose of the sample may be as follows:
disposition is a contractual matter.
3.1.1 To estimate properties of a lot or shipment, such as the
1.7 The following outline is presented to facilitate use of
percentage of some constituent, the fraction of the items that
this practice:
fail to meet a specific requirement, the average weight or
Section
property of an item, or the quality or total weight of the
shipment, or simple identity.
General Philosophy 4
Sampling of Attributes 5 3.1.2 To dispose of the lot or shipment rationally, without
Sampling for Variables 6
the intermediate step of the formation of an estimate.
Means and Standard Deviations 7
3.1.3 To define new materials in terms of their properties.
Comparison of Sampling Plans 8
Sampling Procedures:
3.1.4 To provide material for evaluation of a test method.
General 9 to 11
Random 10
4. General Philosophy
Stratified 11
4.1 Sampling is a means employed to meet the problem of
estimating the quality of a lot from the inspection of only part
This practice is under the jurisdiction of ASTM Committee D-20 on Plastics
and is the direct responsibility of Subcommittee D20.13 on Statistical Techniques. Annual Book of ASTM Standard, Vol 14.02.
Current edition approved Jan. 10, 1968. Published March 1968. Originally Available from Standardization Documents Order Desk, Bldg. 4 Section D, 700
published as D 1898 – 61 T. Last previous edition D 1898 – 61 T. Robbins Ave., Philadelphia, PA 19111-5094, Attn: NPODS.
D 1898
of the lot. When 100 % inspection (“screening”) of a lot is part by the frequency of occurrence of their attributes, that is,
performed, sampling does not apply. properties or conditions which may or may not be present in a
4.2 The inspection of the sample or portion according to a given sample. Examples include visual defects, contamination,
sampling plan results in an estimation of properties, hence, an and success or failure in meeting dimensional or test specifi-
uncertainty which can be expressed for a given sampling plan cations. Probability sampling should be employed to assess the
in terms of an operating characteristic curve. Each such curve, actual frequency, provided that a small number of undesirable
that is, for a given acceptable quality level, can be used to attributes can be tolerated.
obtain two numerical quantities: a, the seller’s risk, and b, the
NOTE 2—If the presence of a given attribute is unacceptable, all the lot
purchaser’s risk. An example is given in Appendix X2.
must be inspected, and statistical techniques do not apply.
4.3 The acceptable quality level (AQL) provides assurance
5.2 The quality of a lot with respect to attributes is evaluated
to the producer that good material will seldom be rejected.
by inspection. The sampling plans for such an evaluation are
There is a large probability that quality that is close to and
conveniently expressed in tabular and graphic forms (Appen-
slightly better than the AQL will be accepted (X2.2.1, Note
dix X2). The selection of AQL’s for a given material is
X2.1). Consequently the seller is well protected against rejec-
traditionally a contractual matter. Factors in determining the
tion of submitted lots of material from a process that is at the
number of samples to be taken from a lot include the size of the
AQL or slightly above it. The operating characteristic curve is
lot and the degree and level of inspection.
a graph of the risks that are taken in sampling inspection
5.3 Attributes may be evaluated by examination or by
plotted as a function of incoming lot quality. Any of several
testing.
units such as average, median, or coefficient of variation, can
5.4 Sampling for Examination presents fewer problems.
be used to establish the seller’s and purchaser’s risks. The unit
Examination generally is nondestructive and hence less costly.
most frequently used for this purpose is the arithmetic average.
Contamination of molding powder, bubbles in sheets, fisheyes
4.4 An immediate question which arises in approaching the
in film, warp and twist in extruded shapes, and dimensions by
sampling of a given lot is that of selecting the number of units
go and no-go gages fall in this category. Because of their
to be sampled, that is, defining the sampling plan. A sample of
subjective nature, it is difficult to put firm numbers on visual
relatively small size will generally be adequate for a large lot.
defects and contamination for a plastics material specification.
For a small lot, however, a relatively large sample is required
Once the specification has been established, however, a suffi-
for reliable estimation of properties. Such a sample is often
cient number of samples for a good statistical evaluation of a
uneconomical, and a compromise must then be found between
lot can generally be examined at reasonable cost.
the cost of an adequate sample and its inspection, and the risk
5.5 Sampling for Physical and Chemical Tests must be
entailed in the less certain information from a smaller sample.
planned while keeping in mind that the tests are often time-
Actually, such a balance should always be sought in statistical
consuming and destructive of the sample. The cost tends to
practice so that the value of the information to be obtained
limit the number of tests that will be performed. Tests include
from a given program is commensurate with the cost of
the determination of particle size and of moisture of molding
obtaining the information. This practice includes in Appendix
powder, measurement of properties such as tensile strength,
X2 and Appendix X3 some specific directions for choosing the
hardness, and deflection temperature, and measurements of
number of units to be sampled. This has proved useful in
color and transparency. For many of these tests, specimens
extensive use in certain seller-purchaser relationships.
must be cut or molded. The manipulations and calculations
NOTE 1—Some detailed procedures for setting up the numerical fea-
may require considerable time for each sample.
tures of a sampling plan for a particular characteristic of a material will
5.6 The definition of the sampling unit is important, since it
also be found in Practice E 122.
affects the size of the sample. The conception of the sampling
4.5 In the early stages of experience with a material, the
unit should also take into account the intended use of the
required sample quantity for an isolated lot is generally
material. For instance, a single bubble may seriously impair the
calculated while considering the lot as an entity. Considerable
usefulness of a large sheet of glazing material that is to be used
advantage is often gained in evaluating a series of shipments of
in its entirety in military aircraft. If the sheet is to be cut into
similar material by regarding their manufacture as a whole.
152-mm (6-in.) portholes, however, more defects could pre-
This is generally done after there is some assurance that the
sumably be tolerated because they can be marked and then
manufacturing process is under control, as by the application of
avoided in the cutting.
control chart methods.
4.6 If the sampling plan is to be designed to help establish
6. Sampling for Variables
whether a lot meets the requirements of a material specifica-
6.1 The ultimate object of sampling for variables is to
tion, it is imperative that the specification be formulated
estimate the average value of a property of the material that can
carefully.
be measured on a continuous scale and the variability of the
5. Sampling for Attributes property, with a desired degree of confidence. Such a procedure
can be used to obtain data for establishing a specification, as
5.1 The quality of plastic materials or products is judged in
well as for comparing a new lot with a specification.
6.2 In working against a specification for a variable, the
See the ASTM Manual on Quality Control of Materials, STP 15-C, Am. Soc.
concept of AQL is useful because it reduces to tabular and
Testing Mats. (1951).
See Form and Style for ASTM Standards, available from ASTM Headquarters. graphic forms the choice of sample size (that is, the number of
D 1898
units from the lot to be drawn into the sample) that is necessary 7.2 Even in the most careful measurements, a certain
in order to minimize the risk of accepting substandard material. amount of variability or dispersion will be encountered. This
6.3 In both inspection for attributes by means of tests, and in must also be reported in order to allow meaningful compari-
inspection for variables, values measured in tests are the sons of the values. For the purpose of this practice, the most
ultimate bases for judgment of quality, that is, both types of useful numerical expression of this variability is an estimate of
inspection are intended to ensure that the percent defective is the standard deviation, s, of the individual values, calculated as
less than a certain value. The difference lies in that, in the first follows:
instance, the individual test results, or perhaps an average for
¯
s 5 ~(~X 2 X! /~n 2 1! (1)
=
i
each sampling unit, are used in deciding acceptance or rejec-
tion. In inspection for variables, however, the range (spread) of
where:
the group of measurements is also important. The estimated
s 5 estimated standard deviation,
standard deviation, derived from the measurements, is multi-
X 5 value of a single observation,
i
¯
plied by a factor from tables, and the resulting product is then X 5 arithmetic means of a set of observations, and
n 5 number of observations in the set.
subtracted from the average. The corrected average is used for
comparison with a minimum specification value. (When the The reliability of the mean can now be estimated.
7.3 The condition of the average and the standard deviation
specification is a maximum limit, the product is added to the
average.) This means, as one result, that a lot of product having can also lead to an estimation of the “confidence limits” of the
average, provided that only random errors influence the mea-
a relatively poor average value, but a narrow range of values
for a certain property, may be accepted as meeting a specifi- surements. In order to achieve narrow confidence limits, a large
cation, while another lot having a better average but greater number of determinations must be made. This is seldom
variation may be unacceptable. This puts a premium on realized in practice because of economic considerations of the
consistent performance, which is probably more economical in cost of repeated measurements. Hence, it becomes necessary to
the long run for everyone concerned. report also the numbers of measurements made in arriving at
6.4 In setting up a plan for sampling for variables, it must be the stated average. Then, confidence limits can be calculated
remembered that it is generally easier to attain a desired degree from tables of Student’s “t” values.
of assurance for the average value than for its standard 7.4 Sampling may be designed simply to obtain such
deviation. This can easily be demonstrated by an experiment information as the above on a new material or test method. This
which involves drawing numbers truly at random from a bowl. information is then applied to make decisions on the quality of
In general, it takes a very few trials in order to establish a subsequent lots of the material.
reasonable value for the average number. It takes many more
8. Comparison of Sampling Plans
samples to arrive at a good estimate of the standard deviation
8.1 Table 1 shows the sample sizes (number of units of
of the average. This woul
...