1、Designation: C 802 96 (Reapproved 2002)Standard Practice forConducting an Interlaboratory Test Program to Determinethe Precision of Test Methods for Construction Materials1This standard is issued under the fixed designation C 802; the number immediately following the designation indicates the year o
2、foriginal adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon (e) indicates an editorial change since the last revision or reapproval.1. Scope1.1 This practice describes techniques for planning, con-duct
3、ing, and analyzing the results of an interlaboratory study ofa test method. It is designed to be used in conjunction withPractice C 670. Thus, the procedures recommended in thispractice have the limited purpose of providing reliable infor-mation on which precision statements of the type described in
4、Practice C 670 can be based. It is not appropriate for use inprograms whose purpose is to develop a test method or toassess the relative merits of two or more test methods.2. Referenced Documents2.1 ASTM Standards:C 109/C109M Test Method for Compressive Strength ofHydraulic Cement Mortars (Using 2-i
5、n. or 50-mm CubeSpecimens)2C 136 Test Method for Sieve Analysis of Fine and CoarseAggregates3C 670 Practice for Preparing Precision and Bias Statementsfor Test Methods for Construction Materials3C 1067 Practice for Conducting a Ruggedness or ScreeningProgram for Test Methods for Construction Materia
6、ls3E 105 Practice for Probability Sampling of Materials4E 177 Practice for Use of the Terms Precision and Bias inASTM Test Methods4E 178 Practice for Dealing with Outlying Observations43. Significance and Use3.1 Certain criteria need to be met before undertaking aninterlaboratory study to determine
7、the precision of a testmethod. It is not necessary that all of the following conditionsdescribed be completely fulfilled in every case; however, ifsome conditions are not met or are met incompletely, theprogram will become more complicated and require more workand expense, or may result in impaired
8、information. Therecommendations outlined in this section are intended toensure that the test method is free of technical difficulties to thegreatest extent possible before an expensive and time-consuming interlaboratory study is undertaken.3.1.1 The first requirement is the existence of a valid andw
9、ell-written test method that has been developed in onecompetent laboratory (or by cooperative work in a smallnumber of laboratories), and has been subjected to a screeningprocedure, or to ruggedness testing as described in PracticeC 1067. As a result of the screening procedure and someexperience wit
10、h the test method in the sponsoring laboratoryand one or two others, a written version of the test method hasbeen developed (but not necessarily published as a standardmethod) that describes the test procedure in terms that caneasily be followed in any properly equipped laboratory. Con-ditions that
11、affect the test results should be identified and theproper degree of control of those conditions should be specifiedin the description of the test procedure (see Note 1).NOTE 1The desired degree of control of conditions that affect testresults may not always be practically achievable, and tolerances
12、 in the testmethod should recognize this fact. Variations in test results due tovariations in such conditions contribute to the total variation whichdetermines the precision of the test method. If the resulting variation is sogreat that uncertainties in average values obtained by the test method are
13、unacceptably high, then the test method itself is at fault, and efforts shouldbe made to improve it or to replace it by a better one. An expensive andtime-consuming interlaboratory study should not be undertaken on such atest method.3.1.2 Any apparatus required for performing the test shouldbe appro
14、priately designed and available at reasonable cost.3.1.3 Personnel in participating laboratories should haveenough experience with the test method so that they arecompetent to run the test. The importance of this requirementwill vary with the complexity of the method and the degree towhich it depart
15、s from familiar procedures.3.1.4 Preliminary knowledge should exist about howchanges in materials and conditions affect the test results.There should be a reasonable degree of certainty that thewithin-laboratory variances are the same in different laborato-ries, and that troublesome interactions do
16、not exist. These1This practice is under the jurisdiction of ASTM Committee C09 on Concreteand Concrete Aggregates . This practice was developed jointly by ASTM CommitteeC01, C09, D04, and D18, and is endorsed by all four committees.Current edition approved May 10, 1996. Published July 1996. Original
17、lypublished as C 802 74 T. Last previous edition C 802 94.2Annual Book of ASTM Standards, Vol 04.01.3Annual Book of ASTM Standards, Vol 04.02.4Annual Book of ASTM Standards, Vol 14.02.1Copyright ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.c
18、onditions are investigated in the analysis of the data of aninterlaboratory study, and are discussed further in 8.2.2, 8.2.3,and Appendix X1.3.1.5 Facilities and procedures for procurement, prepara-tion, and distribution of samples must be available and shouldbe as simple and free of difficulties as
19、 practicable.3.1.6 Selection of samples must be done by a randomizationprocess, and one person who is familiar with randomizationprocedures should be responsible for seeing that the procedureis carried out. Refer to Recommended Practice E 105.3.1.7 Adequate numbers of participating laboratories, op-
20、erators, and materials must be available. Requirements in theseareas are specified in Sections 4 and 5.3.1.8 The entire interlaboratory test program should bedeveloped from the beginning with the help and advice ofpersons familiar with statistical procedures and with thematerials involved (see Note
21、2). The same persons who designthe experiment should also carry out, or at least have controlover, the process of analysis of the data.NOTE 2It may not always be possible to obtain people who arefamiliar with the materials involved who have a sufficient knowledge ofthe proper statistical techniques
22、and their proper use. In this case, thecommittee should obtain the services of a competent statistician who hasexperience in practical work with data from materials testing, and providehim with an opportunity for learning something about the particularmaterials and test method involved. Planners of
23、an interlaboratory studyshould also be warned to avoid the pitfall of assuming that the use of alarge computer necessarily results in special expertise in the handling ofdata or the interpretation of results.3.2 It is important to bear in mind that estimates of theprecision of a test method are alwa
24、ys based on a particular setof data obtained at a particular time and they need to be keptup-to-date. As materials, apparatus, and conditions change, andoperators change or gain more experience, the characteristicprecision of the results obtained may change, especially if thetest method is new. In s
25、ome cases, it may even be desirable toconduct more tests at a later date (though not necessarily arepetition of the complete interlaboratory study) in order toprovide a check on estimates previously obtained and eitherverify them or introduce revisions.4. Laboratories4.1 The problem of obtaining com
26、petent participating labo-ratories for an interlaboratory study is often one of the mostdifficult ones connected with the process. The number oflaboratories available is seldom as extensive as one would like,and if the test method is new, complicated, or expensive andtime-consuming to run, the probl
27、em is further complicated.The problem usually becomes one of finding and obtaining thecooperation of enough qualified laboratories to obtain mean-ingful estimates of precision, rather than that of selectionamong a group of available laboratories. If there is greatdifficulty in obtaining a sufficient
28、 number of competent andwilling laboratories, then the possibility exists that the testmethod should not be subjected to a formal interlaboratorystudy.4.2 For the purposes of programs using this recommendedpractice, it is recommended that at least ten participatinglaboratories be included (1, 2).5In
29、 cases where it is impossibleto obtain ten laboratories, the effect of an increased numbermay be obtained by repeating the program with the same groupof laboratories six months later. Usually, results obtained fromthe same laboratory after a time lapse of approximately sixmonths display most of the
30、characteristics of results from adifferent laboratory, especially if a different operator andapparatus can be used. If this procedure is followed, it isnecessary to be sure that the same materials are used, and thattheir characteristics have not changed in the interim.4.3 In general, it is recommend
31、ed that any laboratory that isconsidered qualified to run the test in routine testing situationsshould be permitted and encouraged to participate. “Qualified”implies proper laboratory facilities and testing equipment,competent operators familiar with the test method, a reputationfor reliable testing
32、 work, and sufficient time and interest to doa good job. It does not mean, however, that only a select groupof laboratories that are considered to be those best qualified forthe interlaboratory study should be picked. Precision estimatesfor inclusion in a test method must be obtained under condi-tio
33、ns and through the efforts of laboratories and personnel thatare representative of the situations in which the test methodwill be used in practice (3). If a laboratory has all the otherrequirements, but its personnel has had insufficient experiencewith the method, the operators in that laboratory sh
34、ould begiven an opportunity to familiarize themselves with the methodand to practice its application before the interlaboratory studystarts.5. Materials5.1 The number and type of materials to be included in aninterlaboratory study will depend on the following:5.1.1 The range of the values of the pro
35、perty being mea-sured on a given material and how the precision varies overthat range,5.1.2 The number of different materials to which the testmethod is to be applied,5.1.3 The difficulty and expense involved in obtaining,processing, and distributing samples,5.1.4 The difficulty of, length of time r
36、equired for, andexpense of performing the tests, and5.1.5 The uncertainty of prior information on any of thesepoints. For example, if it is already known that the precision isrelatively constant or proportional to the average level over therange of values of interest, a smaller number of materials w
37、illbe needed than if it is known that the precision changeserratically at different levels. A preliminary pilot or screeningprogram may help to settle some of these questions, and mayoften result in the saving of considerable time and expense inthe full interlaboratory study (4).5.2 In general, a mi
38、nimum of three materials should beconsidered acceptable.6. Estimates of Precision6.1 In accordance with Recommended Practice C 670, theprocedure described in this practice is designed to provide two5The boldface numbers in parentheses refer to the list of references at the end ofthis practice.C 802
39、96 (2002)2basic estimates of the precision of a test method: (a) single-operator precision, and (b) multilaboratory precision. If otherestimates of precision are desired, other references should beconsulted (see Practice E 177) (5).6.2 Single-operator precision provides an estimate of thedifference
40、that may be expected between duplicate measure-ments made on the same material in the same laboratory by thesame operator using the same apparatus within a time span ofa few days.6.3 Multilaboratory precision provides an estimate of thedifference that may be expected between measurements madeon the
41、same material in two different laboratories.7. Collection of Data7.1 In order to minimize the problems concerned withanalysis of data, a definite form and instructions for obtainingand recording the data should be developed and distributed toall participating laboratories.7.2 Directions to Laborator
42、iesThe directions to the labo-ratories should deal mainly with reporting of data. No specialinstructions for performing the tests that differ from thosegiven in the description of the test method should be included.The laboratories should be cautioned to conduct tests andreport results exactly as sp
43、ecified in the test method, with theone exception as noted in 7.2.2.7.2.1 Averaging Test ResultsLaboratories should particu-larly be cautioned against practices such as running a numberof tests and selecting the “best” ones or reporting the averageof several determinations, except as such averaging
44、is specifiedin the test method. For example, Test Method C 109/C 109Mspecifies three or more test specimens, and requires that thestrength of all acceptable test specimens made from the samesample and tested at the same period shall be averaged andreported. In this case, the directions for the inter
45、laboratory testshould specify the number of determinations to be obtained andreported. Whenever a test result is defined, either in the testmethod or in the instructions to laboratories participating in aninterlaboratory test program, as the average of a particularnumber of determinations, the indiv
46、idual determinationsshould be reported, in addition to the averages. When two ormore measurements are averaged to obtain a test result, thedata from the interlaboratory test program may be used todevelop an estimate of the precision of these individualmeasurements. See 3.3.3 of Practice C 670.7.2.2
47、Rounding of Data:7.2.2.1 Generally, laboratories should be required to reportall figures obtained in making the measurements, rather thanrounding the results before recording them. In some cases, thismay result in recording of more digits than is customary oreven more than the test method calls for
48、in the section onReporting (see X1.3.1). This is necessary because the variationfrom which information about the precision of the test methodcomes is contained in the least significant digits, which areoften discarded in reporting the results of routine testing (6).For example, Method C 136 calls fo
49、r reporting of percentagesto the nearest whole number. This is adequate for the usualreporting purposes, but for purposes of determining precision,at least one decimal place is needed. It is better to require thereporting of too many decimal places than too few, since adecision about rounding all data can be made when the analysisis done. If too few places are reported, however, valuableinformation may be irretrievably lost, and the result might wellbe the impairment of the entire program.7.2.2.2 In cases where a test result is the result of acalculation based on two or more meas
