Title:
From Data-Driven Models to Material Characterization: A New Approach to Improve Durability and Mechanical Performance of High-Early Strength Concretes
Author(s):
Tavares
Publication:
Web Session
Volume:
ws_S23_Tavares.PDF
Issue:
Appears on pages(s):
Keywords:
DOI:
Date:
4/2/2023
Abstract:
The properties of high-early-strength concretes (HESC) are desired for applications such as rapid repair engineering, precast concrete, cold weather concreting and rapid construction. Nevertheless, the fast formation of ettringite associated with calcium sulfoaluminate and calcium aluminate-based mineral admixtures leads to rapid set and low workability, raising problems related to early-cracking and long-term durability. This makes their use in 3D printing and rapid pavement repair more challenging. The increasing need for tools capable of intuitively demonstrating the effect of concrete mixture composition on material performance has motivated this work. The type and proportion of ingredients required to produce HESC synergistically dictate material performance, which presents a great opportunity for artificial intelligence (AI) techniques. Yet, the effectiveness of AI models is highly dependent on the size, distribution, and quality of the data, which has led researchers to use datasets from multiple sources in lieu of performing extensive experiments, which can become time- and resource-intensive. This study couples orthogonal arrays with AI models to optimize HESC mixtures while performing reduced experimental runs, allowing designers to quickly generate efficient models using their own data and materials. Mixtures performing within an optimum level for multi-objectives are further characterized to understand the generated microstructures.