Multimodal optimization of concrete mix design for sustainable load bearing wall panels: Mean-mix − Artificial Intelligence − experimentation fusion

Abstract

This work used abundant database to intelligently select a mix-proportion of conventional OPC and fly ash (FA) based concretes. First, data was sorted and bifurcated according to the grades of concrete, and a mean mix-design (MMD) approach was proposed for selecting constituents. Afterward, both data types were trained using an artificial intelligence (AI) tool, and strength was predicted for trained sets and testing sets of data. It was further compared with the MMD approach for considering a 30 MPa strength. Moreover, experiments were designed for 30 MPa targeted strength by ACI and MMD methods. It was observed that MMD, AI, and experimentation approaches have close correspondence for selected targeted strength. However, ACI-based mix-design overestimated the strength, resulting in higher costs and CO2 emissions than their counterparts. It has also been observed that FA-based concrete has a similar cost to lower-grade conventional concrete but has lesser CO2 emissions. Furthermore, a multi-objective optimization model was proposed for intelligent mix design by integrating strength, cost, and CO2 emissions. It was revealed that the proposed MMD proportions of constituents are among the top-ranked intelligent mixes, thus verifying the proposed approach. Finally, the wall panel was developed of size 600 × 300 × 125 mm using conventional concrete and FA-based concrete, and it fulfills the performance requirement of masonry defined by standards. The cost of masonry with wall panels is 35 % lower than conventional brick masonry, and CO2 emission was 68 % lower, respectively.

Keywords

• Mix-Design
• Database and Experimentation
• Artificial Intelligence
• Wall Panel
• Sustainability