Estimates of repeatability and reproducibility are given in Table 1. The standard deviations are also shown in Figure 1. These estimates have been calculated by excluding the data for Laboratory P at Level 1. Other stragglers noted in the histograms have been included in the calculations.
In Figure 1 it can be seen that the points representing the repeatability and reproducibility standard deviations fall close to straight lines, so that it is possible to fit functional relations to these results. However, the reproducibility standard deviations appear to lie on a line that passes through (or close to) the origin, whereas the repeatability standard deviations appear to lie on a line that is nearly horizontal. These two lines will cross, implying that at low levels of MB value the repeatability is higher than the reproducibility, which is impossible.
The dye solution is added in increments of 5 mL or 2 mL, and this imposes a lower limit on the repeatability that can be achieved. An increment of 5 mL is equivalent to an MB value of:
10 × 5 ÷ 200 = 0.25 MB units
(assuming that the test portion mass is 200 g), so that adding the dye solution in increments of 5 mL is equivalent to rounding the result to the nearest 0.25 MB units, which is in turn equivalent to a repeatability standard deviation of 0.25 ÷ 12½ = 0.07 MB units. Adding the dye solution in increments of 2 mL is equivalent to rounding the result to the nearest 0.10 MB units, which is in turn equivalent to a repeatability standard deviation of 0.10 ÷ 12½ = 0.03 MB units.
In Figure 1 it can be seen that the repeatability standard deviations observed in the cross-testing experiment are consistent with an assumption that the lower limit to the repeatability standard deviation (and hence to the reproducibility standard deviation) falls between 0.03 and 0.07 MB units. It is proposed therefore to adopt 0.05 MB units as the intercept in the functional relations for both repeatability and reproducibility. This gives the lines shown in Figure 1.
It has been argued (Jorck, Sym and Powell, A study of mechanical tests of aggregates. Green Land Reclamation Report GLR 3036/03a. 1994) 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. At the present time it has not been decided how the Methylene Blue test will be called up in specifications. However, if a specification does impose a limiting value on the Methylene Blue test, then the value will be an upper limit, and the above criterion may then be applied to the test.
The results in Table 3 show that the reproducibility of the Methylene Blue test fails to meet this criterion by a wide margin. Hence if the test is to be called up in specifications it would be necessary to improve its precision. The Mandel plots show strong evidence of correlations between results for the three materials. This indicates that the cause of the variations in the measurements is a factor or factors that affects all the tests in a laboratory consistently, such as the preparation of the dye solution, or ageing of the dye solution, or some aspect of the technique employed to carry out the stain test.
It is also worth noting that the correlations between the results for the three materials demonstrate that the between-laboratory variation was not caused by variations between the samples, so that the method used for sample preparation produced samples that were sufficiently homogeneous for the purposes of this experiment.
A precision experiment has been carried out in the UK on the Methylene Blue test (Powell, Sym and Pike, 1993. Preliminary overview of European test methods for aggregates. Third UK precision trial (1993) - dye adsorption tests using red and blue dyes. Building Research Establishment Note N 135/93.) involving 14 laboratories and using two sands with adsorption values close to 2.0 MB units. This precision experiment differed from the cross-testing experiment reported here in several respects:
The coefficients of variation of reproducibility achieved in the precision experiment were close to 8 % for both sands. This shows that it is possible to improve substantially the reproducibility of the CEN version of the Methylene Blue test by simple techniques.
In Level 3, the repeatability coefficient of variation achieved in the experiment was 9.4 %. The reproducibility cannot be better than the repeatability, so whatever steps are taken to improve the reproducibility of the test, it will not be able to meet the 8 % criterion at this level. This implies that if a specification limit of around 1.0 MB units (or less) is to be applied to the results of the test, it will be necessary to improve its repeatability.
It is usually possible to improve the repeatability of a test method simply by carrying out two (or more) replications. However, as noted above, the results of the experiment suggest that there is a lower limit to the repeatability standard deviation of approximately 0.05 MB units because the dye solution is added in increments of 2 mL or 5 mL. Suppose that the correct result for a particular sand is that obtained by adding 21 mL of dye solution, then, with increments of this size, 22 mL will be needed to give a blue halo in the stain test, and the same error of about +5 % will be made however often the test is carried out. This shows that replication is not the answer for the Methylene Blue test.
The simplest way to improve the repeatability of the test would be to add the dye solution in smaller increments, and it is recommended that this should be considered if a specification limit of around 1.0 MB units (or less) is to be applied to the results of the test.
The Methylene Blue test method requires a single test portion to be tested to obtain a test result. To check their repeatability therefore, it would be necessary for a laboratory to do extra work to that required by the Standard by carrying out duplicate tests. One possibility would be for an operator to do one or two extra tests from time to time. The data generated could be used to check that a laboratory achieves similar repeatability to that given in this experiment. The ranges of pairs of test results may be compared with the repeatability limit r1. About fifteen duplicate tests are needed to allow a reliable assessment of a laboratory's repeatability.
A likely reason for poor repeatability is inexperience on the part of the operator. This is particularly likely when, as is the case with the Methylene Blue test, operators in many countries will not have used the test before. Repeatability checks should therefore be seen as an important part of the training an operator receives when he is learning to perform the Methylene Blue test.
Test results for repeatability checks should be recorded to the nearest 0.01 MB units.
The results of the cross-testing experiments reveal a number of cases where laboratories obtained erratic results. This shows that there is a need for laboratories to have some means of checking the accuracy of their results.
A very effective way that laboratories can check their accuracy is by taking part in proficiency tests, in which samples of materials are distributed regularly and the results are compared as in a cross-testing experiment. Regular comparisons will allow laboratories to identify possible faults, try out modifications to their procedures, and obtain data that show if the modifications have produced the desired improvement. It is recommended that the establishment of such proficiency testing schemes should be considered. This would be of particular value in countries where the test has not been widely used up to now.
The Methylene Blue dye is a potential cause of variation in the test because the dye can exist as either a dihydrate or trihydrate, and the solution is unstable and sensitive to light. Errors can therefore occur if insufficient care is taken during the preparation and storage of the dye solution.
A stock of kaolinite with known MB value of 9.4 g(dyea)/kg(clay), giving
V' = 28.2 mL and MBK = 0.94 mL/g,
is held by P & S Research Ltd, Century House, Telford Avenue, Crowthorne (telephone 0/44 1344 762676, fax 0/44 1344 772137). This is available for use by laboratories as a reference material to check their dye solution, for example by doing a test on the reference kaolinite at the beginning of each day on which they do Methylene Blue tests. The same material can be used in routine tests when it is considered desirable to add kaolinite to help identify the end-point of the titration.
Figures X, Y and Z suggest that there is a relation between the amount of dye adsorbed by the kaolinite and the MB result for the sand, for laboratories that added kaolinite. Operators who obtained high values on the kaolinite also obtained high values on the sands, and operators who obtained low values on the kaolinite obtained low values on the sands. This could arise because of differences of interpretation of the stain test. The use of the reference kaolinite would also show up such errors.
Also, the adsorption reported by Laboratory E (27) is very high for kaolinite. It is possible that their clay was not kaolinite but montmorillonite. The provision of a reference kaolinite would ensure that laboratories use the correct material.