Abstract: Fire incidents in operating highway tunnels can cause sustained deterioration of the tunnel lining structure. To accurately predict the evolving risk associated with lining degradation after such hazards, a dynamic Bayesian network (DBN) model is developed to assess the long-term structural safety of highway tunnel linings. First, based on the characteristics of fire-induced damage, a safety evaluation index system is established, yielding a corresponding weight table. Then, a DBN-based evaluation model is constructed, and the risk grades and classification values of postfire lining safety are proposed. Prior probabilities are estimated from field detection results, while the transfer probabilities of each evaluation index are derived assuming an exponential distribution for tunnel service life. The model is applied to the Shiyashan tunnel. Results show extensive concrete spalling near the center of the fire zone, with a maximum depth of 40 mm and a concrete strength decay rate of 0.10 － 0.12. The safety risk value of the lining increases with service time, reaching 0.81 at 35 years. The progressive deterioration of concrete and development of cracks induced by the fire continuously impact structural safety, resulting in a rising risk value over time.

Key words: highway tunnel, fire hazard, lining long-term safety, dynamic Bayesian network, assessment model