Existing shafting alignment methods ignore the inherent uncertainties of the system when modelling and inputting variables. The calculated results and the working conditions differ. Therefore, this study establishes a shafting dynamic alignment model under uncertainty using the stochastic finite element, which combines the finite element method with Latin hypercube sampling. Bearing elevations and material parameters were used as random input variables. Bearing load and cross-section bending angle were used as evaluation indicators. The shafting operational reliability was calculated for linear, rational, traditional dynamic alignment and uncertainty dynamic alignments. The results show that only the uncertainty dynamic alignment satisfies the requirements. An experiment was conducted to simulate the hull deformation by continuously changing the bearing elevation. The bearing load and axis orbit were collected and analyzed for the four alignment states. According to the linear alignment, the deviations between the experimental and theoretical results for the aft stern, front stern, and intermediate bearings loads are 0.07 %, 15.26 %, and 5.12 %, respectively. Hence, the alignment model satisfied the requirements for accuracy. Furthermore, the uncertainty dynamic alignment as the shafting's optimal state can be determined by analyzing the axis orbit in different states. Therefore, considering the uncertainty, a shafting dynamic alignment is necessary.
