COMPUTATIONAL WELDING MECHANICS
Discussion
Hot cracking is important in certain classes of alloys. This requires models of solidification, including segregation and texture modeling. The stress analysis near the melting point could be the most severe test constitutive models encounter. The yield strength approaches zero, texture and grain boundary effects are becoming significant. Experimentally, it is a hostile environment requiring high speed measurements. The cracking is tightly coupled to the microstructure. It could involve capillary driven flow into thin - walled cracks. Strain in the solid is thought to be a driving force. In this temperature region, viscous or creep effects are expected to be dominate. This challenge could be the most difficult in that it arises from coupling between physical phenomena with length scales ranging from atoms to meters and time scales ranging from nanoseconds to seconds. Mathematical techniques such as operator splitting and Domain Decomposition can be useful in dealing with these disparate length and time scales. To manage this complexity in software is a hard problem in software engineering, [1].
The finite element models developed by Feng et al’s study are not intended to and in fact do not reveal the microscopic deformation processes, although they could provide the necessary boundary conditions for the microscopic analysis. Rather, they are used to seek the relations between the macroscopic thermo mechanical conditions local to the trailing edge of a weld pool and the variables that are practically changeable during the welding operation (welding parameters, joint design, application of restraint and so on). The fact that the simulation results correlate well with the experimental observations provides strong support for such an approach.
Dangerously misleading results obtained by conventional Charpy F-notch (CVN) testing of narrow zones in electron beam welds in an X-70 fine grained micro-alloyed steel. In one case these results suggested a 50% fracture appearance transition temperature (FATT) of -37 °С (-38 °F) in the weld metal when a more careful investigation using a specially devised test determined the FATT was actually +90 °C(162 °F) higher at +50 °С (122 °F). It is proposed that this discrepancy arose because the yield strength of the weld metal was much higher than the yield strength of the base metal. Consequently, general yielding began in the base metal and exceeded the fracture strain of the base metal before brittle fracture initiated in the weld metal. Under these conditions, it is suggested that highly brittle weld metal can yield CVN data showing base metal toughness providing the weld metal yield strength is sufficiently high.
The strong variations of the flow stress in a welded joint, with big differences between the base materials, the weld material and the HAZ, is usually mapped out by carrying out a large number of micro-indentation tests on a cross-section of the weld. In fact, the yield stresses used for HY100 in the Tvergaard and Needleman investigation were determined experimentally in this manner. Tvergaard and Needleman studies result, that the fracture toughness of the weld is a more complex property, which not only depends on the flow stress distribution, but also on the fracture properties of each of the material components and on the weld configuration.
It should be noted that the requirements of the material model are higher if the zone near the melt for hot cracking is of particular interest. This improved material model should, at the same time, be matched by a refined spatial and temporal discretization. The modeling of material behavior at higher temperatures and in the presence of phase transformations is perhaps the most crucial ingredient in successful welding simulations. Progress in this field is dependent on the collaborative efforts from computational thermodynamics and material science. The research community active in the field of welding simulation should focus on this field. The improvement in material modeling and increasing availability of material parameters will, in combination with the computational development, increase the industrial use of welding simulations, [8].
