Due to increased traffic density, there is a tendency to construct larger immersed tunnels with larger tunnel elements both in transversal and longitudinal direction resulting in a more critical design of the joints. Especially for those tunnels with relatively soft foundation and high siltation, there is no mature experience for reference. This not only requires a more detailed analyze of the joint itself, but also of the total behavior of the tunnel and its interaction with the subsoil.
Conventional deterministic designing procedure is based on average soil characteristics with sensitivity analyses including upper and lower boundaries of soil stiffness. The mentioned above trend of larger structures requires a more sophisticated probabilistic approach with deeper understanding of uncertainties and insight in probability of exceedance of requirements in serviceability limit state.
In this paper, a probabilistic analysis approach has been implemented to understand the shear behavior of tunnel joints. The biggest contributor for tunnel joints shear behavior is differential settlement between adjacent parts of the tunnel structure. Two probabilistic methods, Point Estimate Method and Monte Carlo Simulation are performed on settlement determination. The former is more efficient and with a considerable accuracy while, the latter is extremely accurate with high computational costs. For simple cases, Monte Carlo Simulations are commonly used to solve soil structure interaction problems due to its straightforward process and easily understandable theory. While for large and complex soil-structure interaction problem, it is computationally intensive to complete even a single run. Such practical disadvantage can be solved by modifying the algorithm or by performing computationally efficient probabilistic methods.
The soil-structure interaction analysis is researched by models set up in PLAXIS 2D. Python scripts are programmed in cooperation with PLAXIS models to perform Monte-Carlo Simulation in determination of shear forces in tunnel joints. The function of Python scripts is helping PLAXIS to select properties automatically and to storing the output after every single run.
The reliability analysis is done for different conditions in 3 models. Comparison shows that reliability significantly influenced by the distribution and correlation length of soil parameters.