Numerical modeling of seismic performance of shallow steel tunnel

Advanced numerical models can be used to complement experimental results and further elucidate their findings. This is particularly true when certain scaled model "adjustments" are incorporated in the testing scheme to reduce the testing cost or schedule. This study conducts comprehensive finite element modeling to simulate large-scale shaking table tests conducted by Kim et al. (2023) [1] to evaluate the seismic performance of steel tunnels. The numerical model was calibrated by comparing its predictions with the experimental observations in the initial tests. The calibrated model was then validated by comparing its predictions with the results obtained from other test models and subjected to different input ground motions. The validated model was employed to comprehensively investigate the effects of the testing scheme parameters including the input motion time scaling, tunnel burial depth, and soil relative density. It was found that reducing the time scale factor during the tests led to decreased seismic responses. It was also found that the presence of overburden soil on top of the tunnel resulted in higher seismic bending moment values, and that higher overburden height increased both lateral and vertical soil pressures on the tunnel side walls and top slab. In addition, lateral soil distortion and tunnel deformations were found to be directly correlated with the overburden soil height. Furthermore, in the absence of overburden soil, there was a significant amplification in horizontal soil acceleration. The results demonstrated that lower-density sand bed experienced greater acceleration amplification and larger displacement, and the tunnel experienced larger lateral soil pressure and higher racking. Additionally, the results revealed that subjecting the same soil bed to several shaking tests affected its stiffness, and hence affected the test observations in the subsequent tests. This should be considered when planning shaking table tests. Finally, the strain values and racking ratios obtained from the numerical simulations and the shaking tests were in agreement with the values obtained from the FHWA procedure.