Shear behaviour of tunnels subjected to fire

Abstract

Recently experiments were conducted at the Technical University of Delft on the size effect of concrete. The size effect is a term used for the relative decrease in shear capacity with an increase in height of the structural member. The beams observed in the experiment failed much sooner than was predicted. These test results have implications for the Heinenoordtunnel, the roof of which shares many of the characteristics of the beams that were used in the experiments. The question is posed what happens to the Heinenoordtunnel in case of a fire, when also considering the recent tests on the size effect of concrete. The Heinenoordtunnel is analysed with a numerical model. First however, in order to account for the observed size effect, the beams from the experiment are recreated. A study is performed on the effect of various parameters on the numerically obtained failure load, cracking load, crack pattern and deflection in order to find a set of parameters to approximate the observed size effect. It was found that a significant reduction in tensile strength and fracture energy is necessary to obtain a better approximation of the experimental results. However, despite these changes the numerical model still overestimates the shear capacity. This information is used to create a model of the Heinenoordtunnel. A situation without a fire load is analysed and validated. The model is compared with the analytical IBBC-TNO method. Consequently, the model is subjected to a fire load. The fire is modelled using temperature dependent properties and by determining the temperature ingress for a 2 hour RWS fire. A significant shear crack is found present due the fire load, the location and shape of the crack suggesting onset of shear compression failure. The model however is still considered to be in equilibrium and so failure has not actually occurred in the model. A comparison with an analytical model suggests that a shift in bending moments from the increase in temperature results in a shift of shear capacity in the roof. It is concluded that, while the numerical model does not fail, some caution is advised for the translation of these results to practical application. The change of material parameters found in modelling the size effect tests still leads to an overestimation of the shear capacity. On the other hand, the situation that was modelled was an extremity. In the model of the Heinenoordtunnel the absolute physical maximum water load was assumed in conjunction with an extreme fire. It is recommended to check the fire load with computational fluid dynamics modelling, to see if the fire load could possibly be less severe than assumed.