Title: Neural Network Modeling of Concrete Carbonation
Author(s): Ali Akbar Ramezanianpour and Amir Tarighat
Publication: Symposium Paper
Appears on pages(s): 899-916
Keywords: advanced concrete technology; complex phenomenon;
concrete carbonation depth; input-output relationships; modeling;
neural networks; nonlinearity
Corrosion is one of the dominating causes of deterioration of reinforced concrete structures. Carbonation of concrete can initiate the corrosion of reinforcements. Many parameters are affecting the concrete carbonation process. Due to the combination of these parameters, phenomenon of concrete carbonation is very complex. It is therefore necessary to implement numerous experimental works to find the relationship between input and output parameters. These tests are slow and time-consuming. On the other hand t h e great number __ of __ r eq uir ed tests makes the investigations costly. Thus it is worth to use numerical methods as new tools to find the relationships between input and output parameters. Neural Networks are capable of showing the relationships of inputs and outputs even in complex nonlinearity. They can be used even in the cases of little background of the theoretical rules, which govern the phenomenon. Due two these advantages of neural networks, a new model of concrete carbonation (NNCC) have been developed to show the appropriateness of the neural networks in civil engineering fields especially in advanced concrete technology, together with its usage as a new prediction model instead of conventional fitted type models.