Page 431 - Practical Design Ships and Floating Structures

406

to the difference in tack weld positions. In the simulation it was found that the positions of the first few
tacks were decisive for the general deformation mode as the influence of the succeeding tack welds is
insufficient to cause large alterations once the deformation mode has been initiated.
The computation time used for each model is 120 hrs on a HP 55000 workstation, including the
thermal and the mechanical simulation for both fillets. Holding the capability of the workstation
against the standards of today, a considerable reduction in computation time is realistic.

4 THE LOCAL/GLOBAL TEMPLATE

The efficiency obtained by graded elements and moving mesh algorithms is inadequate when the
objective is simulation of welding in large structures such as ship sections. The essential problem of
dynamic mesh refinement is that the structure should be discretised at least to the level where
geometry, stresses and deformation can be represented with adequate accuracy at ambient temperature.
More precisely, unwelded regions should be discretised to represent geometry and restraints properly.
The welded regions should also hold the stresses and deformations accumulated in welding and, hence,
further discretisation is required. As a result the number of DOFs necessary to represent a given
structure at the coarsest refinement level increases very quickly beyond what is computationally
attainable when the size of the structure grows.

In the present context a template is defined by a bundle of techniques for welding simulation in large
structures. Generally, templates are based on hrther approximations compared to the full simulation.
The essence is how these approximations are chosen and how they affect the accuracy of the prediction.
To predict the distortions of an assembly rather extensive requirements are imposed on the template to
be developed. In addition to the factors essential to the local modelling, it is necessary further to
account for the stiffness of the surrounding structure and to inched the stress and deformation caused
by preceding welding on the assembly. These factors are central to a template capable of predicting the
effect of different welding sequences. A template denoted the LocaVGlobal template or just the UG
template has been developed for the purpose in Andersen (2000,2001b).
The subassembly chosen for analysis is a part of a transverse web frame in a container vessel and is
particularly interesting as it has caused significant problems in production due to welding deformations.
The global model of the subassembly is shown in Figure 4(a) and is composed of linear elastic shell
elements.

As the global shell model is not suited for welding simulation, each weld line is represented by a local
solid model, equivalent to the dynamically meshed models shown previously. The idea behind the L/G
template is to link the local models to the global and thus enable a proper representation of both the
boundary conditions for the local models and the accumulated distortions in the global model.
Moreover, it is exploited that at large part of the structure can be modelled by substructures due to its
linear elastic response at low temperatures. The steps in the L/G template may be described briefly as
follows: The global model is substructured and included as boundary conditions for the first local
model as illustrated by Figure 4@). The welding response is calculated and the deformations are used
to update the distortions of the global shell model as shown in Figure 4(c). The deformed global model
is used to establish the boundary conditions in the modelling of the next weld line so that the
interaction between weld lines is included in the modelling and the effect of welding sequence can
therefore be accounted for. The specific calculation steps of the template and issues such as the
updating of geometry, the combination of shells and solids, the extraction of boundary conditions and
substructuring are addresses in detail in Andersen (2000,2001 b).