Determination of the effectiveness of various mineral additives against sodium and magnesium sulfate attack in concrete by Taguchi method

In this study, pumice (P), waste glass powder (GP), ground granulated blast furnace slag (GGBFS) and colemanite waste (CW) were used in combination as mineral additives in concrete production. By employing the L16 design of the Taguchi method, four different mineral additives were used at four different levels (0%, 5%, 10% and 20% for P, GP and GGBFS, 0%, 1%, 3% and 5% for CW). The produced concretes were exposed to magnesium and sodium sulfate for 360 days. At the end of these periods, the compressive strength and ultrasonic pulse velocity (UPV) values, as well as the compressive strength loss and mass loss values of the concrete samples before and after the sulfates’ effect were determined. The Taguchi analysis was utilized to assess the efficiency levels and optimum mixing ratios of the mineral additives employed in each experiment. At the same time, the contribution level of each parameter was calculated using ANOVA (analysis of variance). In addition, confirmation experiments were carried out to determine optimum conditions and to ensure the accuracy of the analyses. The analysis results indicated that the ideal mixing ratio for the compressive strength after the effect of sulfates can be prepared with using 5% P, 10% GP, 10% GGBFS and 1% CW. It has been determined that 5% P, 5% GP, 10% GGBFS and 1% CW should be used in order to minimize mass loss due to the sulfate effect. In addition, scanning electron microscope (SEM) images of the most successful and unsuccessful samples were taken and their microstructures were examined. •Pumice, waste glass powder, GGBFS and colemanite wastes were used in combination as mineral additives in concrete.•Effects of mineral additives on properties of concretes were investigated after 360-day MgSO4 and Na2SO4 exposure.•The Taguchi analysis was utilized to assess the efficiency levels and optimum ratios of the mineral additives employed.•The contribution level of each parameter was calculated using ANOVA (analysis of variance).