Generalized Trial Mix Design Approach for Concrete with Various Cementitious Constituents

Abstract

Concrete's versatility and structural importance make optimizing its mix design crucial to meeting modern construction standards for strength and durability. Existing models for predicting concrete strength, however, are often constrained by small data sets and specific additive compositions, highlighting the need for a more generalized empirical approach. This study addresses this gap by developing a comprehensive compressive strength prediction model applicable across a broad spectrum of concrete compositions and strength levels. Utilizing a comprehensive database of over 3,500 concrete mix designs with compressive strengths ranging from 2 to 220 MPa, this research integrates a wide variety of cementitious additives, including superplasticizers, silica fume, fly ash, blast furnace slag, nanosilica, limestone powder, quartz powder, and fibers, with ages spanning from 1 to 365 days. A parametric approach was employed to split the database into 80% training and 20% testing subsets, leading to a nonlinear regression model for strength prediction. The model demonstrated high accuracy, achieving an R2 of approximately 0.94 and an root mean square error (RMSE) of 11 MPa across both subsets. Furthermore, the derived equation was supported by a step-by-step compressive strength-based trial mix design procedure which was validated on 34 experimental mixes with varied strength targets, ages, and additive types. This approach maintained predictive accuracy within a 10% error margin when compared to actual results, highlighting its robustness. This study provides a practical empirical tool for practitioners, enabling effective compressive strength prediction across diverse mix compositions and concrete strength ranges, facilitating more informed mix design decisions in the field.