ABSTRACT
In tunnel structures, the risk of safety hazards significantly increases when exposed to fire, emphasizing the importance of investigating the corresponding structural responses. However, existing models for concrete under elevated temperatures inadequately capture the complex stress-strain relationship and material degradation under fire conditions, which may lead to deviations in calculations of structural behavior. This paper proposes an improved stress-strain relationship for concrete based on neural network methodologies, which leverages the predictive power of neural networks to more accurately reflect the behavior of concrete exposed to elevated temperature. The proposed model is then applied to simulate the structural behavior of a certain tunnel structure in fire scenario. Further analysis and discussion about stress redistribution are provided based on FEM simulation result.
