Deformations in welded panels
Typ
Projektarbete, avancerad nivå
Program
Publicerad
2020
Författare
Selvaraj, Srikannan
Pamfil, Bogdan
Hård, Daniel
Gurram, Shivaprasad
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Welded panels are commonly used for ships hull manufacturing, and in order to
lighten the hull the stiffened panel thickness tends to be reduced, thus increasing
the distortion effects due to welding. This project will be part of Chalmers
contribution to the benchmark exercise in a specialist committee, Material and Fabrication
Technology, for the International Ship and Offshore Structures Congress
(ISSC) 2021 conference. The project’s purpose is to compute the deformation for
the ISSC2021v.3 benchmark [1] geometry of a stiffened welded panel which comprises
of a stiffener 1000 x 100 x 5 mm DH36 plate which is initially tack welded on
one side to a 1000 x 400 x 5 mm DH36. The welding procedure consists of a first
weld fillet being deposited on one side of the stiffener, followed by a cooling time,
and then another weld fillet being deposited on the other side and followed by a
second cooling time before the welded panel is finally unclamped. To achieve that,
both thermal and mechanical analyses are carried out by relying on the nonlinear
Finite Element Analysis (FEA). The commercial FEA program ABAQUS is used
for analyses in this project. Those simulations are however performed for a shorter
plate and stiffener of length 200mm.
The material properties were implemented in ABAQUS input file form, for both the
weld and the plate and stiffener. Nonlinear thermal and mechanical properties were
taken into account. The specific heat and thermal conductivity were implemented
for the DH36 steel, however other properties such as thermal expansion coefficient,
Poisson’s ratio etc., were implemented for a similar steel, S355. The weld material
and the plate and stiffener are altogether modelled as being the same material except
regarding the hardening modulus and the yield stress. When compared to the
plate and stiffener material, the weld material itself has been modelled as having a
higher yield stress and higher hardening modulus at room temperature.
Two thermal models are considered with each having 1000 C as the heat source
temperature of the weld. First, the fillet weld is activated simultaneously and in the
second, the fillet weld material is activated sequentially by dividing the weld into
segments. After each weld pass the model is allowed to cool down for 200 seconds
so that the welded panel has cooled to around 50 C at the end of the simulation.
The thermal simulations are coupled to mechanical simulations. In mechanical simulations,
different clamping sequences are carried out. One with fully fixed and
two with sliding clamps. In total six configurations are considered, since there are
two thermal models and three clamping conditions for the mechanical simulations.
Clamping is taken care of by taking the top and bottom nodes of the plate at designated
locations.
The same mesh model was used for both the thermal and mechanical simulations,
and it consists of linear 8-node hexahedral and linear 4-node tetrahedral solid elements.
The linear hexahedral elements are used to model the plate and stiffener
and the welding joint is modelled by linear tetrahedral elements. Prior to choosing
vi
a final mesh a mesh convergence study was conducted by monitoring the temperature
variation at a location close to the weld. The resulting mesh had 31200 nodes
and elements with an aspect ratio lower than 10 to avoid complications with the
mechanical simulations.
The results of this project showed that the type of heating source and the clamping
conditions have definite effects on the behaviour of the model. The displacements
results for the mechanical simulations indicated that the sequential heating results
were closer to the intended ones than for the simultaneous heating case. Also, the
extent of clamping conditions, i.e., either fully clamped or sliding clamps, along
with the methods used to prevent rigid body motions for a sliding clamp are all
factors which have an effect on the behaviour of the model. And these differences
in the thermal and mechanical loading configurations will be magnified when they
are applied to the full length of 1000 mm for the plate and stiffener as per the
ISSC2021v.3 benchmark.
Beskrivning
Ämne/nyckelord
ABAQUS , FEA , DH36 , Inertia relief , Simultaneous , Sequential