Title:
Artificial Neural Networks to Predict Mulit-Fold Performance of Advanced Cementitious Composities: Can We Tranform Information into Knowledge?
Author(s):
Liberato Ferrara
Publication:
Web Session
Volume:
Issue:
Appears on pages(s):
Keywords:
DOI:
Date:
3/24/2024
Abstract:
The increasing demand of concrete structures raised the issue of raw material consumption and scarcity. Meanwhile, the service life of several buildings is coming to an end, resulting in substantial formation of construction and demolition waste (CDW). The crushing of old concrete structures to form recycled aggregates is one of the most common routes to both reduce the consumption of raw materials and avoid landfilling. Numerous studies addressed the properties of recycled aggregate concrete (RAC), where the coarse aggregates are replaced by the crushed material. Owing to the wide availability of data, it was possible to create a database including input and output properties of RAC, respectively related to origins of the waste material and mix design and slump, shrinkage, and strength. The rate of water absorption was also included as a durability parameter of the resulting concrete. The correlation between input and output parameters for concrete production is a crucial aspect in mix design and the significant variability of the parameters involved requires powerful tools to be addressed. For this purpose, artificial neural networks (ANN) are a suitable artificial intelligence (AI) application to bridge the knowledge gap currently present in RAC production. This work developed a neural network to correlate various input parameters with the expected performance of the RAC concrete both at fresh and hardened state.