Prediction of fire spalling behaviour of fiber reinforced concrete

Fire spalling prediction of fiber reinforced concrete containing polypropylene (PP) fiber and steel fiber at elevated temperature is a challenging problem. The conventional methods such as FEM and DEM are difficult to deal with the problem as a result of complicate coupling mechanism of polypropylene (PP) fiber and steel fiber in concrete. To this end, two artificial neural network (ANN) models, one (ANN1) is on the basis of concrete mix study and the other one (ANN2) is based on compressive strength study, are introduced in current study to assess the resistance of concrete to explosive spalling. A number of 321 and 318 test data gathered from literature are utilized to train the two proposed ANN models. Twenty-four concrete mixes (96 groups), i.e., seven plain concrete (PC) mixes, four high performance concrete (HPC) mixes reinforced with PP fiber, three ultra-high-performance concrete (UHPC) with reinforced PP fiber and ten ultra-high-performance concrete (UHPC) mixes reinforced with PP and steel hybrid fiber are designed and tested to validate the accuracy of the two models. It demonstrates that ANN1 and ANN2 can achieve a predictive accuracy of 89.6% and 84.4% for the explosive spalling respectively, which indicates the feasibility of proposed ANN models for predicting explosive spalling threat of the hybrid fiber reinforced concrete.

Automated summary

This study introduces two artificial neural network (ANN) models to assess the resistance of fiber reinforced concrete containing polypropylene (PP) fiber and steel fiber to fire spalling at elevated temperature. The models achieve a predictive accuracy of 89.6% and 84.4% respectively, indicating the feasibility of using ANN for predicting the explosive spalling threat of hybrid fiber reinforced concrete.