Artificial Neural Networks in Prediction of Concrete Strength Reduction Due to High Temperature


  • The International Concrete Abstracts Portal is an ACI led collaboration with leading technical organizations from within the international concrete industry and offers the most comprehensive collection of published concrete abstracts.

International Concrete Abstracts Portal


Title: Artificial Neural Networks in Prediction of Concrete Strength Reduction Due to High Temperature

Author(s): Chih-Hung Chiang and Chung-Chia Yang

Publication: Materials Journal

Volume: 102

Issue: 2

Appears on pages(s): 93-102

Keywords: exposure; pulse velocity; strength

Date: 3/1/2005

The effect of temperature on the compressive strength of concrete has been thoroughly explored. The effect of exposure time, however, still needs systematic exploration. This paper demonstrates that artificial neural networks can be used to predict the residual strength of heated concrete effectively. Experimental investigation reveals that loss of strength becomes significant as exposure times prolonged at exposure temperatures of 400 °C and higher. Regression analysis of residual pulse velocity and residual strength shows that a certain linear relationship exists for exposure time is between 30 and 120 min. The dependence of residual strength on exposure time and temperature is highly nonlinear. Due to the absence of a theoretical relationship, neural network analysis is applied to identify a possible general relationship between residual strength and variables including exposure temperature, exposure time, water-cement ratio (w/c), and residual pulse velocity. Three neural networks are designed and trained by data from the experiments and data collected from the literature. Good linear relationships are found when comparing the residual strength predicted by networks with correspondent target values. Such results are promising and can be extended to be part of assessment for fire-damaged concrete prior to repair of structures.