Estimates of repeatability and reproducibility are given in Table 1. The standard deviations are also shown in Figure 1.
Note that these standard deviations (and the values given for r and R) are estimates that can be influenced by individual results. Thus there are some stragglers in the between-determination ranges for Laboratories A, B and M, and the data for Laboratory A give a straggler in the laboratory averages for Level 3. If these stragglers were not included in the calculations, the repeatability standard deviation for Level 1 would be reduced from 0.021 to 0.014, and that for Level 3 would be reduced from 0.047 to 0.030, and the reproducibility standard deviation for Level 3 would be reduced from 0.102 to 0.081.
The functional relationships given in Table 2, and shown as the lines on Figure 1, were calculated using the standard deviations for Levels 1, 2 and 3. However, data for only three materials are not really sufficient to establish these relationships with a high degree of confidence. It is possible that the precision of the test method may depend, to some extent, on the material under test. Partner 3 noted that the slag used for Level 2 was difficult to homogenise, and this may explain why the reproducibility is worse for this material than for the ground blast furnace slag used for Level 3.
If the functional relationships are quoted in the European Standard test method then the precision results for the individual levels must also be given (so that the reader can see how much information was used to derive the relationships).
The proposed draft European Standard specification for aggregates for concrete (Proposed draft for aggregates for concrete including those for use in roads and pavements. CEN/TC 154 committee paper N 232. January, 1994.) contains the following information about limits for the total sulfur content of aggregates:
"The total sulfur content of the aggregate shall not exceed:
2% S for slags
1% S for aggregates other than slags"
(Presumable these limits should be interpreted to mean 2.0% and 1.0%.)
The materials used in the cross-testing experiment gave overall average sulfur contents of 0.5% in Level 1, 1.1% in Level 2, and 1.5% in Level 3. They thus cover the range of sulfur contents that will be of interest in practical situations when the European Standard comes into effect.
It has been argued (J»rck, Sym and Powell, 1994 A study of mechanical tests of aggregates. Green Land Reclamation Ltd Report GLR 3036/03a.) that the reproducibility standard deviation of a mechanical test, when expressed as a coefficient of variation, should be no more than about 8%, if the test method is to be used to assess the compliance of aggregates with specifications. The argument used to justify this value applies equally-well to non-mechanical tests when their results are to be compared with upper limits in specifications, so it can be applied to the determination of total sulfur contents.
The reproducibility coefficients of variation (see Table 3) for Levels 1 and 2 are greater than 8%, and that for Level 3 is only marginally smaller than 8%. It should also be remembered that these reproducibility coefficients of variation were obtained in an experiment in which the participants were provided with samples that had already been reduced to pass a 0.125mm sieve. In practice, laboratories have to carry out the reduction process themselves, as part of the test procedure. The reproducibility of the whole test procedure (reduction and analysis) cannot be better than that of one stage (analysis), and could be substantially worse. These results show, therefore, that the reproducibility of the determination of total sulfate is unsatisfactory when assessed by the above criterion.
This conclusion, together with the adverse comments made about the method by a number of the participating laboratories, suggests that the method for determining total sulfur used in the cross-testing experiment is not a good choice for a reference method for the European Standard.
If the method is to be retained as the reference method, then there is a need to improve its reproducibility. The correlation between the averages for Levels 1 and 3 suggests that it may be possible to achieve some improvement in the reproducibility by tightening up the procedure. It is also possible that the reproducibility of the method could be improved, without introducing a substantial bias, by omitting the second step in the procedure (the step that is intended to remove iron and other constituents): simplifying the method will eliminate some sources of variation. EN 196-2 (BS EN 196-2 Methods of testing cement; Part 2 Chemical analysis of cement. British Standards Institution.) describes a method for determining the sulfate content of cement which, apart from the bromine oxidation step, is similar to the method used in the cross-testing experiment reported here. The method in EN 196-2 does not require the separation of dissolved iron, aluminium and other Group 111 elements even though the amounts in solution are likely to be much higher than for the aggregate method. BS EN 196-2 contains a National Annex that describes a method of determining the total sulfur content of cement that includes a bromine oxidation step: again the method does not require the separation of iron, aluminium and other Group 111 elements.
There are a number of general methods for improving reproducibility that should also be considered:
The determination of total sulfur consists of two main stages: sample reduction to produce material that passes 0.125mm, and the analytical procedure to determine the sulfur content. Both stages are potential sources of error, so work to try to improve reproducibility should not be concentrated just on the analytical stage.
A factor that may well have a significant influence on the reproducibility of the test is the degree of familiarity that operators have with the procedure. The nature of the method is such that inexperienced operators will be prone to error. If the test is carried out infrequently, then operators cannot acquire experience and cannot be expected to produce reliable results, and it would be better to use an automated method in which checks on the accuracy can be built in to the procedure.
The British Standard version of the European Standard that describes methods for carrying out chemical analyses of cement, BS EN 196-2, contains a National Annex giving a method for determining the major oxide contents of Portland cement using X-ray fluorescence spectrometry. It is suggested that consideration should be given to adopting this method either as the reference method for determining the total sulfur content of aggregates, or as an alternative method.
The National Annex in BS EN 196-2 contains information about the precision of the XRF procedure. The information was derived from an experiment involving twelve British laboratories and seven cements. The materials were circulated to the participants as ground samples, so the results are directly comparable with those given in this report for the precision of the determination of total sulfur in aggregates. Figures 1 and 2 have been drawn to the same scale so that the two may be compared directly. The functional relationships shown on Figure 2 are derived from those given in BS EN 196-2, by converting from SO3 contents to S contents, and the points for the seven materials that appear in Figure 2 were obtained from the author's own records of the experiment.
If the method based on X-ray fluorescence were to be applied to aggregates, it would be necessary to check that no interferences from other elements could occur that might bias the determination of sulfur. It would also be desirable to produce some reference materials (derived from aggregates not cements) so that laboratories could check the accuracy of their results.
The draft European Standard (Draft prEN 1744-1 Tests for chemical properties of aggregates. Part 1 Chemical analysis. November, 1994.) contains a general requirement that the results of chemical tests should be reported to the nearest 0.01%. The repeatability standard deviations given in Table 1 fall between 0.02% and 0.05%, so that to be able to carry out repeatability checks it is necessary to report results to 0.01%. However sulfur contents around 1% or 2% will be of interest in practice, so for the purpose of checking compliance with the specification it would be more appropriate to report sulfur contents to the nearest 0.1%.