Tolerance analysis of compliant metal plate assemblies considering welding distortion.

The ship block construction process consists mostly of cutting, welding and assembling of compliant steel plates and sections. Compared to the ship's total dimension (on the order of a hundred meters), the required accuracy between the interface of adjacent blocks (on the order of millimeters) is about one in one hundred thousand. Block joint alignment thus requires relatively very high accuracy. Misalignment and nonconformance between the building blocks being erected will adversely impact the total productivity of a shipyard by consuming one of the most critical resources in the shipyard. Distortion caused by the welding process is heavily dependent on the plate's boundary conditions such as clamping of the plates being welded and the sequence of assembly can change these boundary conditions. Also, the plate part variations can be varied through the assembly sequences due to its compliancy. The assembly sequence when designed adequately can positively affect the distortions and deformations during the assembly process without costly mitigation processes. In this dissertation, both tolerance analysis and welding distortion simulation are vigorously investigated to develop a new tolerance analysis method for ship block joining. The mechanistic variation simulation, which statistically combines the sensitivity matrix with root sum square method to determine mean and standard deviation without Monte-Carlo simulation, is used for the variation simulation and equivalent loading method based on the inherent strain for the welding distortion prediction. The method combines B-Spline finite element analysis with empirical data to determine welding distortion without computationally intensive three-dimensional nonlinear finite element analysis. A Modified Method of Influential Coefficients is developed for variation simulation of ship block assemblies considering welding distortion. For validation purposes, an example ship block which consists of two base plates, four stiffeners, and one web frame is tested by the Modified Method of Influential Coefficients and the results compare favorably with that of Monte-Carlo Simulation.PhDApplied SciencesMechanical engineeringMechanicsOcean engineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/125999/2/3224854.pd