GB/T 17039-1997 Standard Practice for Utilization of Test Data to Determine Conformance with Specifications

AI-translated method card — human verification pending. This is not an official translation of the standard. Unclear source values remain unverified; check the original standard and product certificate before use.

Card type: Test method

1. Method number / full name / English name / replacement relationship

• Standard number: GB/T 17039-1997
• Chinese full name: Standard Practice for Utilization of Test Data to Determine Conformance with Specifications
• English name: Standard practice for utilization of test data to determine conformance with specifications
• Replacement relationship: not provided in the available source (the foreword only states equivalent adoption of ASTM D3244-90a, and does not mention replacement of any standard)
• Adoption relationship: equivalent adoption of the American Society for Testing and Materials standard ASTM D3244-90a "Standard Practice for Utilization of Test Data to Determine Conformance with Specifications"
• Approval/implementation: approved by the State Bureau of Technical Supervision 1997-10-14, implemented on 1998-04-01
• Responsible body: Petrochemical Research Institute of China Petrochemical Corporation
• Drafting organization: Petrochemical Research Institute of China Petrochemical Corporation
• Main drafters: Yang Tingting, Yang Tianfu

2. Scope of application and non-application

• Scope of application: This standard establishes the principles for the parties involved (usually referring to the supplier and the receiver) to compare and combine their independently obtained test results when a product quality dispute arises; it specifies the technical steps for comparing test assignments with specification values.
• Applicable conditions: Applicable to test methods whose repeatability and reproducibility data conform to the definitions in this standard.
• Not applicable: no explicit non-applicable clauses are provided in the available source.

3. Method principle

• The properties of petroleum products are determined by standard test methods, and the degree of their conformity with specifications is checked.
• Due to test imprecision, the true value of a property cannot be determined exactly, and the range that may contain the true value must be inferred from the measured value.
• The acceptance limit (AL) serves as the boundary between acceptable quality and unacceptable quality; the test assignment (ATV) serves as the estimate of the true value of the product.
• If the test assignment ATV falls on the AL line or on the acceptable side of the AL line, the product is acceptable; otherwise it is rejected.
• The acceptance limit AL comprehensively considers the specification value S, the precision of the test method (reproducibility R), and the confidence level required relative to the specification value to determine the minimum acceptable quality.
• Basis of the relationship: (1) assumes that test errors follow a normal (Gaussian) distribution; (2) uses the test assignment ATV to determine conformity with the specification, and ATV is the average of acceptable results obtained by two laboratories.

4. Reagents and materials (purity / preparation)

not provided in the available source.

5. Instruments and equipment

not provided in the available source.

6. Recommended operating conditions

not provided in the available source (this standard is a practical method for statistical judgment and does not involve specific test operating conditions).

7. Sampling and sample preparation

• Sampling shall be carried out in accordance with method GB/T 4756.
• The sample amount shall be sufficient to ensure that all required determinations can be performed.
• Divide the obtained sample into three portions: the receiver's sample, the supplier's sample, and the retained sample.
• The amount of the retained sample itself shall be sufficient to ensure that it can be further divided into three portions if more tests are needed.
• Definition of test sample (3.12): a portion of product collected in accordance with regulations at the place where the product is handed over when responsibility for product quality transfers from the supplier to the receiver; in practice this sampling method is rarely used, and instead the two parties determine an appropriate sampling location through negotiation for sampling.

8. Analytical procedure

8.1 Determination of the test assignment (ATV) (Chapter 8)

• 8.1 The receiver and the supplier shall separately obtain independent test results XR and Xs.
  • Note: The supplier's result must be the test result on the test sample (see Chapter 5), not a reported value.
• 8.1.1 If the absolute value of A1=XR−Xs is less than or equal to the method reproducibility R, take the average as specified in 6.2.1, and obtain ATV according to formula (3): ATV = (XR + Xs)/2 ... (3)
• 8.1.2 If the absolute value of A1 is greater than R, discard these two results and take a sample from the retained sample for retesting to obtain X'R and X's.
• 8.1.3 If the absolute value of A2=X'R−X's is less than or equal to R, take the average as specified in 6.2.2, and obtain ATV according to formula (4): ATV = (X'R + X's)/2 ... (4)
• 8.1.4 If the absolute value of A2 is greater than R, obtain a new test result XRL from the arbitration laboratory (6.4).
• 8.1.5 If A3 is less than or equal to 1.2R, obtain ATV according to formula (5): ATV = (X'R + X's + XRL)/3 ... (5)
• 8.1.6 If A3 is greater than 1.2R, take the average of the two closest results as the test assignment ATV.
  • Note: This step does not strictly conform to statistical concepts. This approach is taken because in most cases the test sample has been consumed by this point.
• 8.2 If the bias between the supplier's and receiver's laboratories is very small or nonexistent, then in the 95% case, ATV can be obtained according to 8.1.1, and in the remaining 5% cases, when the procedure reaches 8.1.3, ATV can also be obtained with 95%.
• 8.3 If a particular supplier and receiver always have to invite an arbitration laboratory to participate in testing, both parties shall carefully examine their testing procedures; if possible, they shall check the test results with other laboratories.

8.2 Application of precision data in test methods (Chapter 6)

• 6.1.1 Acceptance of results (repeatability): When two results are obtained under the conditions specified for repeatability and the difference is equal to or less than the method repeatability, the operator may report the average of these two results as the result for the test sample.
• 6.1.2 Rejection of results (repeatability): When the difference between two results is greater than the method repeatability, these two results shall be discarded, and another two results shall be obtained under the conditions specified for repeatability; if the difference is equal to or less than the method repeatability, report the average; if it still exceeds, reject these two results and investigate the use of the method.
• 6.2.1 Acceptance of results (reproducibility): When two results are obtained in different laboratories and the difference is equal to or less than the method reproducibility, these two results shall be considered acceptable, and the result for the test sample is the average of these two results.
• 6.2.2 Rejection of results (reproducibility): If the difference between the results obtained by two laboratories is greater than the method reproducibility, discard these two results, and each laboratory shall retest the retained sample; if the difference between the two results obtained by retesting is equal to or less than the reproducibility, report their average; if it is still greater than the reproducibility, reject these two results and investigate the application of the method in each laboratory.
• 6.3 Multiple tests: If multiple results are obtained in one or each of the two laboratories, the allowable difference R' between the average results obtained by the two laboratories is calculated according to formula (1): R'=√{R² − r²[1 − 1/(2n₁) − 1/(2n₂)]} ... (1) Where: R— reproducibility of the test method; r— repeatability of the test method; n₁—number of results obtained by the first laboratory; n₂—number of results obtained by the second laboratory. (Note: In the source, formula (1) source-text extraction is shown as "R'=VR2-P[1-1/(2n,) — 1/(22)]"; the formula form requires confirmation against the source standard (source-text extraction).)
• 6.4 Arbitration laboratory: When the retest results from the two laboratories in 6.2.2 are still rejected, a third laboratory or arbitration laboratory may be invited to participate, and one retained sample from 6.2.2 shall be used for testing to obtain a third result. At this time, multiply the reproducibility R by 1.2 (converting from two laboratories to three laboratories), and compare the difference between the two most dispersed results obtained with this value to check the acceptability of the results. If the results are acceptable, the result for the test sample is the average of these three results.

8.3 Application of precision data in specifications (Chapter 7)

• 7.1.1 A specification value is the limit specified for the true value of a particular property.
• 7.1.2 If the probability distribution function of the test method is known (such as the normal distribution curve related to the method reproducibility), the probability of dispersion of any test result relative to the true value can be calculated.
• 7.2 Critical specification: Due to the characteristics of the product or the needs of the final use of the product, some specifications require a high degree of assurance for the receiver, even if the product reaches or exceeds the quality level represented by the specification value. Non-critical specification: If the specification only requires ensuring that the product quality is not substantially lower than the quality level represented by the specification value.
• 7.3.2 The value located between the acceptable region and the unacceptable region of the product test results is the acceptance limit. The acceptance limit may or may not be consistent with the specification value that specifies the product quality or grade.
• 7.3.3 The acceptance limit must be determined through negotiation between the supplier and the receiver. If the true value of the product is equal to the acceptance limit, the quality level represented by the acceptance limit can make the probability of accepting or rejecting the tested product each 50%.
• 7.3.4 In the absence of an agreement, the specification will be considered a non-critical specification. In this case, if the true quality of the product is the quality represented by the specification value, the probability that the product can be accepted is guaranteed to be 95%, and the acceptance limit can be determined as described in 7.3.6 using the acceptance probability P=0.95.
• 7.3.5 When the true value of the product is equal to the specification value, the probability P of accepting the product is a function of D=(AL−S)/0.255R (as shown in Table 1 and Figure 1), and D is a direct measure of the difference between the acceptance limit AL and the specification value S.
• 7.3.6 In addition to directly determining the acceptance limit through negotiation, when the true value of the product is equal to the specification value, the receiver may also choose a specific product acceptance probability P, look up the corresponding D value from Table 1 or Figure 1, and calculate the acceptance limit AL according to formula (2): AL = S + 0.255RD ... (2) If N is not two laboratories, then multiply 0.255 in the formula by √(2/N). The derivation of the constants in the formula is given in Appendix B.
• 7.3.6.1 For specifications that have both an upper limit and a lower limit, the steps of 7.3.6 shall be performed twice to obtain two acceptance limits, one high and one low, and there must exist some permissible interval between these two acceptance limits.
• 7.3.7 When only one test result can be obtained, the case of N=1 shall be used with formula (2) in 7.3.6. Obviously, with only one test result, reproducibility cannot be tested, and this single result becomes the test assignment of the test sample.
• 7.3.8 The determination of the acceptance limit for a critical specification (for the upper specification value) is such that when the true value is less than or equal to the specification value, the product has only a small probability of acceptance (determined by the choice of P).
• 7.3.9 The determination of the acceptance limit for a non-critical specification (for the upper specification value) is such that when the true value is equal to or greater than the specification value, the product has a large probability of acceptance (determined by the choice of P).
• 7.3.10 The relationship between the acceptance limits for critical specifications and non-critical specifications is shown in Figure 2 (for the lower specification value).

8.4 Product Quality Conformity (Chapter 9)

• 9.1 If the test assignment of every property of the product meets the requirements of the acceptance limits, the product is considered to conform to the specification requirements.
• 9.2 The supplier shall deliver the product only when it ensures that every property of the product conforms to the specification requirements.
• 9.3 When the receiver obtains a single result, if this result does not meet the requirements of the acceptance limits obtained according to B3 in Appendix B, the product quality is considered questionable.
• 9.4 As long as the result obtained by the receiver does not meet the requirements of the acceptance limits, a dispute will arise between the supplier and the receiver.
• 9.5 The method for resolving the dispute is to determine the test assignment ATV according to Chapter 8 as the estimated value of the true value of the product, and to determine the acceptance limit AL according to 7.3, and compare the test assignment with the acceptance limit.

8.5 Acceptance and Rejection of Products (Chapter 10)

• 10.1 If the test assignment ATV of the product is equal to or better than the acceptance limit AL, the product is considered to conform to the specification requirements and is accepted.
• 10.2 If the test assignment ATV of the product is worse than the acceptance limit AL, the product will be rejected for not meeting the specification requirements.
• 10.3.1 If the two results fall to the left of the (XR+Xs)/2=ATV=AL line and at the same time fall between the two lines of XR−Xs=±R, the test sample is considered acceptable.
• 10.3.2 If the two results fall to the right of the (XR+Xs)/2=ATV=AL line, the test sample is considered unacceptable.
• When the initial test result falls in the area marked "resample", the test needs to be repeated. If the test result after the second sampling still falls in the "resample" area, an arbitration laboratory should be invited to participate in the test.
• The actual consequences caused by the product being rejected for not conforming to the specification requirements shall be agreed upon in advance through negotiation between the parties concerned.

9. Qualitative and Quantitative and Calculation Formulas

• Test assignment ATV (two results, formula 3): ATV = (XR + Xs)/2
• Test assignment ATV (two results of retest, formula 4): ATV = (X'R + X's)/2
• Test assignment ATV (three results, formula 5): ATV = (X'R + X's + XRL)/3
• Acceptance limit (formula 2): AL = S + 0.255RD
  • If N is not two laboratories, then multiply 0.255 in the formula by √(2/N)
• Permissible difference for multiple tests (formula 1): R'=√{R² − r²[1 − 1/(2n₁) − 1/(2n₂)]} (formula form requires confirmation against the source standard (source-text extraction))
• Appendix B Derivation of constants:
  • AL = S + (aD/√N) …(B1)
  • R = a(1.96)√2 = 2.77a …(B3)
  • a = R/2.77 = 0.361R …(B4)
  • When N=2: AL = S + (0.361RD)/√2 = S + 0.255RD …(B5)
  • Upper specification value, P=0.95 (D=+1.645): AL = S + 0.419R
  • Lower specification value, P=0.95 (D=−1.645): AL = S − 0.419R
  • When N=1: AL = S + 0.361RD …(B8)
  • N=1, upper specification value: AL = S + 0.594R …(B9)
  • N=1, lower specification value: AL = S − 0.594R …(B10)
• Table 1: Deviation of acceptance limit from specification value at a specific product acceptance probability (D=(AL−S)/0.255R)
  • Note: The test assignment ATV is obtained using N=2, that is, results obtained by two different laboratories.
  • Acceptance probability P and D value (upper specification value / lower specification value):
    • 0.001: −3.090 / 3.090
    • 0.005: −2.576 / 2.576
    • 0.010: −2.326 / 2.326
    • 0.025: −1.960 / 1.960
    • 0.050: −1.645 / 1.645
    • 0.100: −1.282 / 1.282
    • 0.150: −1.036 / 1.036
    • 0.200: −0.842 / 0.842
    • 0.300: −0.524 / 0.524
    • 0.500: 0.000 / 0.000
    • 0.700: 0.524 / −0.524
    • 0.800: 0.842 / −0.842
    • 0.850: 1.036 / −1.036
    • 0.900: 1.282 / −1.282
    • 0.950: 1.645 / −1.645
    • 0.975: 1.960 / −1.960
    • 0.990: 2.326 / −2.326
    • 0.995: 2.576 / −2.576
    • 0.999: 3.090 / −3.090
  • (Note: In Table 1 source-text extraction, the correspondence between the "critical specification/non-critical specification" column and the numerical values is misaligned; the above table is organized according to the two columns "upper specification value/lower specification value", and the specific correspondence requires confirmation against the source standard (source-text extraction).)

10. Precision

• Repeatability (r) definition (3.8): the quantitative expression of the random error of repeated results obtained by the same operator in a given laboratory using the same instrument under constant operating conditions on the same test material within a short period of time. It is defined as the difference between two single results under normal and correct operating conditions. In long-term operating practice, this difference can be exceeded only once in 20 times (that is, 95% confidence level).
  • Note 1: Repeatability and reproducibility data shall be determined according to the method of GB/T 6683.
  • Note 2: This difference is related to the standard deviations of repeatability and reproducibility, but is not the standard deviation itself.
  • Note 3: Since the definitions of repeatability and reproducibility in various test methods are not necessarily expressed in exactly the same wording, their definitions should be checked before comparing precision data.
• Reproducibility (R) definition (3.9): the quantitative expression of the random error of single results obtained by operators in different laboratories using the same method on the same test material. It is defined as the difference between two single and independent results under normal and correct operating conditions. In long-term operating practice, this difference can be exceeded only once in 20 times.
• Precision is specified by the repeatability and reproducibility of the test method.
• Specific precision values: not provided in the available source (this standard is a general method and does not give specific values).

11. Report

• 6.1.1: The operator may report the average of two results as the result of the test sample.
• 6.1.2: If the difference between the two results of the retest is equal to or less than the repeatability of the method, the operator shall report the average of these two results as the result of the test sample.
• 6.2.1: The result of the test sample shall be the average of these two results.
• 6.2.2: If the difference between the two results obtained from the retest is equal to or less than the reproducibility, report their average.
• 6.4: If the results are acceptable, the result of the test sample is the average of these three results.
• 8.1.6: Take the average of the two closest results as the test assignment ATV.
• Rounding of results: All results shall be rounded according to the provisions of GB/T 8170 (3.10).
• Report format: not provided in the available source.

12. Referenced Documents

• GB/T 1250 Representation and determination methods for limiting values
• GB/T 4756 Sampling method for petroleum and liquid petroleum products (manual method)
• GB/T 6683 Determination method for precision data of petroleum product test methods
• GB/T 8170 Rules for rounding off numerical values
• Note: The above referenced standards shall all be currently valid standards (unless otherwise explicitly specified in the standard).

13. Source and verification

This page is an AI translation of the available Chinese method card. Human verification is pending. It is not an official translation of the standard. Check the source standard and product certificate before use.

14. Quick Reference for Questions and Answers