1、Designation: D7778 15 An American National StandardStandard Guide forConducting an Interlaboratory Study to Determine thePrecision of a Test Method1This standard is issued under the fixed designation D7778; the number immediately following the designation indicates the year oforiginal adoption or, i
2、n the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon () indicates an editorial change since the last revision or reapproval.1. Scope*1.1 This guide describes the procedures for planning andconducting an interlaboratory
3、 study (ILS) of a test method usedin Petroleum Products and Lubricants Committee D02 ofASTM for the purpose of estimating repeatability and repro-ducibility of the test method in accordance with ASTM Formand Style requirements.1.2 This guide is concerned with the management of the ILSand intended to
4、 provide guidance for the planning stage andensure the process, logistics, and tools are identified and agreedupon in advance.1.2.1 Selection of the samples and the impact of sampleselection on the final scope of the test methodboth the rangeof materials covered in the scope and precision sections a
5、nd themeasurement range covered in the precision sectionareimportant, and careful consideration needs to be given to theseaspects before an ILS is launched.1.3 This guide does not concern itself with the developmentof test methods but rather with gathering the informationneeded for a test method pre
6、cision statement.1.4 This guide is concerned with test methods which yield asingle numerical figure as the test result, although the singlefigure may be the result of a calculation from a set ofmeasurements.1.5 This guide is designed for tests of properties that arestable, such that bulk samples can
7、 be homogenized andsub-samples can be prepared that will be identical and stablefor the duration of the ILS testing period. This guide is notdesigned for unstable properties or for properties that make itdifficult to obtain stable, representative samples for an ILSconducted at multiple laboratory si
8、tes.1.6 This guide represents the suggested practices for atypical precision study. Individual subcommittees may modifythis approach as their expertise directs.1.7 This guide is voluntary, and thus, is not required for anapproved precision study.2. Referenced Documents2.1 ASTM Standards:2D130 Test M
9、ethod for Corrosiveness to Copper from Petro-leum Products by Copper Strip TestD4306 Practice for Aviation Fuel Sample Containers forTests Affected by Trace ContaminationD5854 Practice for Mixing and Handling of Liquid Samplesof Petroleum and Petroleum ProductsD6300 Practice for Determination of Pre
10、cision and BiasData for Use in Test Methods for Petroleum Products andLubricantsD6708 Practice for Statistical Assessment and Improvementof Expected Agreement Between Two Test Methods thatPurport to Measure the Same Property of a MaterialD6792 Practice for Quality System in Petroleum Productsand Lub
11、ricants Testing LaboratoriesE177 Practice for Use of the Terms Precision and Bias inASTM Test MethodsE178 Practice for Dealing With Outlying ObservationsE456 Terminology Relating to Quality and StatisticsE691 Practice for Conducting an Interlaboratory Study toDetermine the Precision of a Test Method
12、E1169 Practice for Conducting Ruggedness Tests2.2 ISO/IEC Standards:3ISO/IEC Guide 30 Terms and Definitions Used in Connec-tion with Reference MaterialsISO/IEC Guide 17025 General Requirements for the Com-petence of Testing and Calibration Laboratories3. Terminology3.1 Definitions:1This test method
13、is under the jurisdiction of ASTM Committee D02 onPetroleum Products, Liquid Fuels, and Lubricants and is the direct responsibility ofSubcommittee D02.94.01 on Task GroupA: QualityAssurance and Quality Control.Current edition approved April 1, 2015. Published May 2015. Originallyapproved in 2012. La
14、st previous edition approved in 2012 as D7778 12. DOI:10.1520/D7778-15.2For referenced ASTM standards, visit the ASTM website, www.astm.org, orcontact ASTM Customer Service at serviceastm.org. For Annual Book of ASTMStandards volume information, refer to the standards Document Summary page onthe AST
15、M website.3Available from Available from Aerospace Industries Association of America,Inc. (AIA), 1000 Wilson Blvd., Suite 1700, Arlington, VA 22209-3928, http:/www.aia-aerospace.org.*A Summary of Changes section appears at the end of this standardCopyright ASTM International, 100 Barr Harbor Drive,
16、PO Box C700, West Conshohocken, PA 19428-2959. United States13.1.1 accuracy, ncloseness of agreement between an in-dividual test result and an accepted reference value. E4563.1.2 bias, ndifference between the population mean ofthe test results and an accepted reference value. E4563.1.3 outlier, nobs
17、ervation that appears to deviate mark-edly from the other observations of the sample (also referred toas extreme result, outlying or doubtful observation, or aberrantvalue. E1783.1.4 precision, ncloseness of agreement among test re-sults obtained under prescribed conditions. E4563.1.5 reference mate
18、rial, nmaterial or substance, one ormore of whose property values are sufficiently homogenousand well established to be used for the calibration of anapparatus, the assessment of a measurement method, or forassigning values to other materials. E1773.1.6 repeatability, nquantitative expression of the
19、 ran-dom error associated with a single operator in a givenlaboratory obtaining repetitive results by applying the same testmethod with the same apparatus under constant operatingconditions on identical test material within a short interval oftime on the same day. It is defined as the difference bet
20、weentwo such results at the 95% confidence level. D6792, E1773.1.7 reproducibility, nquantitative expression of the ran-dom error associated with different operators from differentlaboratories, using different apparatus, each obtaining a singleresult by applying the same method on an identical test
21、sample.It is defined as the 95% confidence limit for the differencebetween two such single and independent results. D6792, E1773.1.8 trueness, ncloseness of agreement between thepopulation mean of the measurements or the test results and anaccepted reference value. E4564. Summary of Guide4.1 The pro
22、cedure presented in this guide consists of fivebasic steps:(1) Planning an interlaboratory study (including objectivesand expected outcome).(2) Disclosure of plan details (participants, equipment,samples, logistics).(3) Approval by membership.(4) Guiding the testing phase of the study and collecting
23、test data results.(5) Post validation that the ILS was executed in accordanceto the agreed plan.4.1.1 This guide does not address the data analysis tech-niques. For details of such analysis techniques refer to PracticeD6300 and E691, or other technically equivalent documents.4.1.2 The relevant subco
24、mmittee is the sole arbiter oftechnical equivalence.5. Significance and Use5.1 All ASTM standard test methods are required to includestatements on precision and bias. To obtain such data it isnecessary to conduct an interlaboratory study. This document isdesigned to provide a brief overview of the s
25、teps necessary inan ILS and to suggest an appropriate sequence in carrying outthese steps.5.1.1 Qualitative tests such as Test Method D130 coppercorrosion may not require an ILS.6. Planning an Interlaboratory Study6.1 Suggested steps in an ILS are given in Table 1.6.2 Support for conducting the ILS
26、is available from ASTMthrough its Interlaboratory Studies Office (ILSastm.org).Individuals involved in conducting an ILS are listed below.6.2.1 “Support” does not include financial support for newor alternate equipment or procedure beyond sampling andshipping materials to the participants.6.2.2 A Ta
27、sk Force should be formed consisting at aminimum of the method developers, users and ASTM statisti-cian. Based on the objectives of the ILS, this task force shouldhave the overall responsibility for the design, disclosure, andexecution of the ILS. The Task Force is charged with thedevelopment of an
28、ILS plan that includes scope and purpose,funding, staffing, participants (equipment structure), sampletypes, tests required, special instructions, reporting form, andsample distribution, participant protocol, and a questionnaireand statistical tools to be used, and criteria for acceptance ofresults.
29、6.2.3 An ILS Coordinator should be appointed as an overallperson responsible for the distribution of materials and proto-cols to the laboratories, and receive the test results from thelaboratories. Eventually this person should be responsible forall aspects of conducting the ILS, and writing the res
30、earchreport.6.2.3.1 A running activity record may be useful to bemaintained by the ILS Coordinator to overview the progress ofthe ILS activity.6.2.4 StatisticianThe Task Force should obtain the ser-vices of a statistician who is familiar with the protocols used inASTM for precision calculations duri
31、ng the planning stage forinput towards the design of the ILS, material selection,establishment of the statistical tools to be used, acceptancecriteria, and testing protocol. This person may be a member ofthe Task Force. This person should also assist the task group ininterpreting the results of the
32、data analysis.TABLE 1 Sequence of an Interlaboratory StudySequence Procedure Section Number in thisGuide1 Select ILS membership 6.2.22 Prepare Basic Design 6.33 Specify Test Method(s) 6.44 List ParticipatingLaboratories6.55 List Materials 6.66 Number of Test Resultsper Test Material perLab6.77 Provi
33、de Protocol 6.88 Ruggedness Study(Pilot Run)7 and 89 Full Scale Run 910 Data Handling 1011 Data Presentation 1112 Prepare Study Report 12D7778 1526.2.4.1 The statistical software available from D02 Com-mittee can help in these calculations, but it is highly recom-mended that this work be done by, or
34、 under the oversight of thestatistician engaged in 6.2.4.6.3 Basic DesignThe design should be kept as simple aspossible in order to obtain estimates of within- and between-laboratory variability that are free of secondary effects. Thebasic design is represented by a two-way classification table inwh
35、ich the rows represent the laboratories, the columns repre-sent the materials, and each cell (that is, the intersection of arow with a column) contains the test results obtained by aparticular laboratory on a particular material. See Table 2 forillustration where x and y are the test results obtaine
36、d on thesame materials by the same operator from each laboratory.6.3.1 Samples to be tested shall be presented as a singleseries of test samples, requiring only a single analysis persample. Both the material identification and the correspondingduplicate shall be blinded such that the tester does not
37、 know thematerial identification of the test sample, nor its correspondingduplicate run. An example of a randomized series for the ILSdesign in Table 2 is presented in Table 3. If this singlerandomized series approach is not possible, then blind repeatsof the same material(s) should be embedded amon
38、g thesamples provided. The blind samples should represent thematerials used in the scope of the test method and concentra-tion range for which the method is applicable. At the least, theblind duplicates should include a range of easy and difficultsamples and include the full sample matrix in the sam
39、ple set.By including only easy samples in blind testing or the full ILSfor that matter, the resulting precision may demonstrate betterrepeatability than otherwise would be found in practice.6.3.2 The randomness of testing samples is so importantthat this has to be a requirement, not just a good prac
40、tice. Theactual duplicates run for each test material shall be randomizedwithin the series.6.4 Test MethodTo establish initial precision statementsfor a new test method, a draft of the test method is required thathas been developed in one or more competent laboratories andstudied in a pilot ILS (see
41、 Section 8). This draft method shoulddescribe the test procedure in terms that the steps can beunequivocally followed in a typical laboratory by competentpersonnel with the knowledge of the materials and the propertyto be tested. The test conditions that affect the test resultsappreciably should be
42、identified and the proper degree ofcontrol of sample handling and the test conditions be specifiedin the description of the test procedure. The test method shouldspecify calibration procedure and its frequency and the formatfor reporting the test results.6.4.1 Reevaluation of Existing Precision Stat
43、ementsToreevaluate existing precision statements for test methods, allaspects of the existing test method that are not intended to berevised shall be followed. Intended revisions to the test methodshall be highlighted in the test method and explicitly stated tothe participants.6.4.1.1 ILS Execution
44、ProtocolThe procedures and specialinstructions provided in the ILS protocol for participation shallnot introduce restrictions or controls in the execution of themethod that are not in the normal execution of the methodbeing evaluated.6.4.1.2 VariablesAll variables in the test method that areintended
45、 to be variables shall be left as variables in the ILS. Toremove, control, or set limits on factors that are not intended tobe controlled in the normal usage of the test method in theexecution of the ILS will result in the precision being overlyoptimistic, and as such precision statements thus gener
46、ated willlikely be unattainable by users in the normal usage of themethod.6.5 LaboratoriesAn ILS should ideally include enoughlaboratories to produce a reproducibility estimate with at least30 degrees of freedom. See Practice D6300 for further guid-ance. It is important that a reasonable cross secti
47、on of thepopulation of qualified laboratories be represented in the ILS.Special attention is required to ensure that at least 75% of theparticipants represent product manufacturer laboratories orthird party commercial independent testing laboratories, orboth. Take care to ensure that the population
48、is random and notbiased towards special interest participants such as equipmentvendors. Vendor participation shall be managed carefully so asnot to bias the data population.6.5.1 The final statement of precision of a test method shallnot be based on fewer than six laboratories with acceptable testre
49、sults for each material. More than 7 or 8 laboratories shouldparticipate, preferably a minimum of 10 laboratories in casesome data is unacceptable and data from one or more labora-tories has to be excluded from the data evaluation.6.5.2 Any laboratory considered qualified to run the testroutinely should be encouraged to participate in the ILS. Inorder to obtain an adequate number of participatinglaboratories, the ILS study should be publicized through trademagazines (if feasible), meetings, broadcast through pertinentASTM committees, etc.TABLE 2 Basic Design of an
