Filler of three fraction grain size composition has been studied in order to prepare concrete with good physicotechnical properties. This composition is most technologically efficient in manufacturing refractory concrete objects. Fractionation of corundum grade K object scrap is accomplished through a standard set of screens with cell sizes of 5.0, 1.25, 0.63, and 0.14 mm. Residues on each of the screens (partial residues) specifying filler grain size composition were labelled with letters. Residue on a screen with cell size 1.25 mm C, 0.63 mm B, and 0.14 mm A.

Provision of minimum intergranular space and correspondingly maximum average filler mix density was achieved by mixing grains of different size in prescribed ratios. Results of experiments are given in Table 1 and shown in a diagram Fig. 1.

Table 1 Filler Grain Size Composition and Average Bulk Density
Full size table
Fig. 1
figure 1

Filler composition— bulk density diagram.

Full size image

It follows from the data obtained that the greatest filler bulk density applies to compositions in the range 50 – 65% A, 10 – 30% B and 10 – 40% C. In order to select the optimum ratio between filler, high-alumina cement, and mixing water in refractory concrete studies were carried out by a completely factored experiment plan (1–3) of eight tests.

For reliability of the results obtained average values of parameters in each experiment were calculated from the results of measuring 12 specimens.

The following factors were varied: X 1 is amount of filler, kg; X 2 is amount of high-alumina cement, kg;, X 3 is water-cement ratio (W/C).

Values of variation factor levels and step are provided in Table 2, and a planned experiment matrix is given in Table 3. As a result if implementing the experimental plan values of parameters provided in Table 4 were obtained.

Table 2 Variable factors and Levels of Component Variation
Full size table
Table 3 Experiment Planning Matrix
Full size table
Table 4 Optimized Concrete Composition and Properties
Full size table

Statistical treatment of results was carried out. Numerical average line-by-line dispersion and verification for uniformity showed that results are reproducible. Coefficients of regression equations were calculated for the average values of parameters. Finally equations were reduced to the form:

$$ \begin{array}{l}y={\upalpha}_0={\upalpha}_1{x}_1+{\upalpha}_2{x}_2+{\upalpha}_3{x}_3+{\upalpha}_{12}{x}_1{x}_2+{\upalpha}_{13}{x}_1{x}_3+{\upalpha}_{23}{x}_2{x}_3+{\upalpha}_{123}{x}_1{x}_2{x}_3,\hfill \\ {}y=19.9+0.18{x}_1+1.09{x}_2+0.09{x}_3+0.68{x}_1{x}_2+1.52{x}_1{x}_3+0.49{x}_1{x}_3+0.51{x}_1{x}_2{x}_3.\hfill \end{array} $$

According to the equations obtained the optimum concrete mix composition is as follows, %: coarse filler fraction 5 – 1.25 mm 62, fine fraction 1.25 – 0.14 mm 20; high-alumina cement 18, concrete water-cement ration 0.43.