Title: Towards a Service Life Prediction System of Concrete Structures Based on a Neural-Computing Approach
Author(s): Bakhta Boukhatem, Arezki Tagnit-Hamou, Mohamed Chekired and Mohamed Ghrici
Publication: Symposium Paper
Appears on pages(s): 41.1-41.12
Keywords: carbonation, concrete structure, corrosion, durability, fly ash, neural network, service life, sulfate attack.
The cost of repairing and rehabilitating damaged reinforced-concrete structures in Canada and elsewhere continues to rise. Predicting the service life and life-cycle cost of these structures can help identify the most cost-effective solution. Many companies have joined with research partners on projects to develop reliable tools to predict the service life of concrete structures. Given the problem’s complexity, most of these projects are based on different modeling approaches producing widely different values, greatly limiting their application.
Therefore, our project consisted in applying a connectionist approach, including artificial neural networks (ANNs) models and a database, to create an intelligent system. In addition, each ANN model better grasps the complex mechanisms of concrete degradation (carbonation, sulfate expansion, chloride-induced corrosion, etc.). The proposed system will yield a powerful solution for predicting the service life of concrete structures and be useful in designing new structures. It will significantly improve codes by contributing more realistic