Abstract

Reliable mechanical strength values of refractory products are relevant parameters with respect to the design of durable and safe installations and accurate tensile strength values are in this context of paramount importance. Whereas CCS method enables to correctly evaluate the compressive strength, bending strength values (i.e. MOR values) are unfortunately not sufficiently representative of the actual tensile strength of the products. Accordingly, a simple, standardized method able to reliably approach the true tensile strength of the products appears still needed.

The splitting tensile-strength test could be such a method. It consists in loading diametrally a right-circular cylinder till the fracture strength of the material is reached and the specimen splits along its vertical diameter. The splitting tensile strength is subsequently calculated from the measured maximum load sustained and the dimensions of the specimen.

The main advantages of this test stems from its ease of implementation. It allows indeed for the use of simple specimen geometry compared to what would be needed for pure tensile test, readily obtained by drilling, and compatible with other, already standardized test methods for refractory products (CCS, density/porosity). Thereupon, the stress state in the specimen during the test is well documented in the literature: the diametral loading induces a uniform tensile stress normal to the loading axis across a rather large part of the anticipated fracture plane whereas relatively high compressive stresses develop around the loading points. This method has for long been applied for the indirect determination of the tensile strength of concrete and has already been standardized in the civil engineering sector.

In the current study, the parameters susceptible to influence the test have been thoroughly screened through factorial designs of experiments and subsequent ANOVA analysis. Practically, the effects of both specimen size and testing conditions (preload level, loading rate, use of load bearing strips and characteristics thereof) have been investigated experimentally on three representative dense shaped commercial products : two high alumina grades with respectively 76 and 81 % alumina, and a magnesia-carbon bonded grade with a carbon content higher than 9 wt.%. A total of more than 250 splitting tests have been performed. By far the largest effect observed is the one related to the use of bearing strips, the role of which consists in distributing the load applied along the length of the cylinder and preventing compressive-stress failure near the loading points.

All by all, sets of optimum conditions have been identified with regard to repeatability. These will be further evaluated with respect to reproductibility through interlaboratory round robin tests and, if proven satisfactory, drafted as potential EN standard.

Funding support of FPS Economy is acknowledged.