Urban tunnel construction often uses pressurized front Tunnel Boring Machines (TBM). It is difficult to predict the displacements induced by the pressurized digging of tunnels in proximity to existing piles, particularly when the distance is smaller than the diameter of the tunnel. Limiting the impact of TBM on piles is crucial for the Grand Paris Express project to reduce the risk of damage to neighbouring structures. The TULIP research project (TUnnel boring machines and Limitation of Impacts on Piles), established by Société des Grands Projets, CETU, ENTPE, and UGE, conducted a full-scale experiment on line 16 lot 1. Eiffage Génie Civil, HEIA-FR and GeoMod have developed a probabilistic approach coupled with the observational method. This article presents the application of this probabilistic approach to better understand the mechanism of tunnel/soil/piles interaction, based on the published results of experimental tests. A 3D model is first produced to simulate the tunnel construction, then the sensitivity analyses are carried out to determine the probabilistic input variables that are the most important on the quantities of interest. Using the monitoring measurements in the pile zone, the Bayesian inference allows updating the calculation hypotheses to refine the evaluation of the probability of exceeding acceptable thresholds for both the surface settlement and the normal force in the pile (from a priori to a posteriori), thus optimizing the design if the value of the probability of exceeding these thresholds decreases a posteriori.