Seismic vulnerability of a soil-steel composite tunnel – Norway Tolpinrud Railway Tunnel Case Study

•The longitudinal stiffening of RC beams significantly affected CSP tunnel response;•Modelling the CSP tunnel shell with orthotropic characteristics leads to underestimating tunnel response compared to corrugated shell;•The method of modelling a steel shell affects the time of the appearance of maximum displacements, velocities and accelerations;•The duration of the seismic excitation affects the response of the soil-steel tunnel;•The results from Canadian Highway Bridge Design Code obtained are quite far from those obtained from FEM; Transport infrastructure, such as tunnels and underground structures, are usually built using reinforced concrete solutions. Since the 1970s, many transport structures (culverts, bridges, tunnels and animal crossings) have been made from corrugated steel plates and surrounding backfill (also known as soil-steel structures). These structures are a good alternative to reinforced concrete solutions because they can be built rapidly with substantial 40–50% cost savings. The behaviour of soil-steel structures under typical static and dynamic loads is reasonably well recognised. However, the seismic analysis of these structures is usually conducted in a limited range, mainly for small culverts. Therefore, it is important to know how soil-steel composite tunnels behave under seismic excitations. The analysis in this work was made for the Tolpinrud railway tunnel in Norway. The paper presents the numerical analysis under seismic excitations using the DIANA FEA program based on the finite element method. Non-linear models with the El Centro and Ancona records were used to analyse the problem. Four different numerical models of the tunnel were considered. The paper analyses two methods of modelling the steel shell of the tunnel and the impact of using reinforced concrete (RC) beams to strengthen the shell. In models I and II, the shell was modelled as corrugated, and in models III and IV as flat. Additionally, it was found that the influence of additional RC beams (models II and IV) is negligible. Moreover, significant frequencies caused by given seismic excitations were determined. It was also found that the use of destructive earthquakes of similar intensity and different duration caused a low level of stresses in the soil-steel tunnel models. As a result of the conducted research, it was recommended that model I is the most suitable for this type of analysis. The obtained results can help recognise the behaviour of such tunnels loca