This report gives the result of a cross-test on the determination of the Shape Index. 22 laboratories from 11 countries took part in the experiment. Three materials with different levels of poorly shaped particles were tested.
The results obtained on two of these materials were such that the precision data derived from this cross-test cannot be used in practice. Several factors influencing the results are discussed and recommendations given to improve the test procedure.
Considering the number of difficulties encountered, it is a remarkably good result that the reproducibility standard deviation obtained for Level 1 is slightly smaller than the corresponding one calculated according to the equation from DIN 52114.
This is not the case for Levels 2 and 3, where very big reproducibility standard deviations were obtained. The reasons for these big variations in the results are not completely clear. Though the factors discussed above were certainly of influence, they cannot be the only reason for the variations found.
Given the fact that the extreme values come from laboratories which are quite possibly not familiar with the test method, one might be tempted to eliminate the results of laboratories with no previous experience with this test. This, however, would be unfair to those laboratories, who had no experience, but arrived at results close to the average.
Also, this procedure would not eliminate traditional differences between countries which are familiar with the test: in Austria, for example, the over- and undersize is not tested, contrary to the practice in Germany.
It should be possible to improve the test procedure. The following recommendations therefore refer to prEN 933-4.
It should be checked in TG 6 of CEN/TC 154, whether the test portion masses of prEN 933-1 and prEN 933-4 can be made to agree. This would allow the Shape Index test to be carried out on the same test portion as the grading test, would also allow the results of the grading test to be used as weights in the calculation of the weighted Shape Index when sub-samples of size fractions are examined in the Shape Index test.
The term "appropriate sieves" should be better defined and include upper and lower limiting sieves.
If, for the aggregate to be tested, upper and lower limiting test sieves are not known, the grading must be established first. For this purpose preferably a series of sieves should be used, where Di/di <= 2. The resulting percentages will be used for the calculation of the weighted Shape Index. After having thus established the limiting upper and lower sieves and the amount of over- and undersize, it should be clear whether the oversize and the undersize > 4 mm must be tested or not. It is not clear whether this is a decision of the test procedure or of the requirement specifications.
In case the mass of a size fraction exceeds by far 300 particles, it should be allowed in the test procedure to reduce the mass of this size fraction by a sample reduction procedure.
The minimum number of particles cannot be easily converted exactly into minimum masses for each size fraction, because a size fraction with D/d <= 2 may comprise particles of very different mass and shape. In the present cross-test, for example, the 8/16 mm fraction of the Level 3 material was very flaky and contained a concentration of small particles, and the approximate mass of 300 particles of the 8/16 mm fraction ranged from 250 g for Level 3, through 400 g for Level 2, to 1200 g for Level 1.
It should also be made clear in the test procedure what is to be done if a size fraction does not reach 300 particles.
Last not least, it would make the test procedure more efficient if a reference instrument would be indicated for the decision whether a particle is cubical or non-cubical.