Using data mining algorithms to predict the bond strength of NSM FRP systems in concrete

- Formulations for predicting the bond strength in NSM FRP systems are analyzed.
- The prediction's accuracy of ACI 4402R-08 and HB 305-2008 guidelines are tested.
- Alternative formulations based on data mining algorithms are suggested.
- The results showed that the data mining algorithms can be a good alternative.
- A website is presented to ease the use of the data mining models developed.

This paper presents the effectiveness of soft computing algorithms in analyzing the bond behavior of fiber reinforced polymer (FRP) systems inserted in the cover of concrete elements, commonly known as the near-surface mounted (NSM) technique. It focuses on the use of Data Mining (DM) algorithms as an alternative to the existing guidelines' models to predict the bond strength of NSM FRP systems. To ease and spread the use of DM algorithms, a web-based tool is presented. This tool was developed to allow an easy use of the DM prediction models presented in this work, where the user simply provides the values of the input variables, the same as those used by the guidelines, in order to get the predictions. The results presented herein show that the DM based models are robust and more accurate than the guidelines' models and can be considered as a relevant alternative to those analytical methods.