ADVANCES IN ALUMINUM WELD SIMULATIONS APPLYING WELDSIM H. G. Fjær), O. R. Myhr(**), S. Klokkehaug(**), E. J. Holm(*) ABSTRACT This paper describes recent developments and applications of the advanced simulation model WELDSIM. This model is applicable for welding of age hardening aluminum alloys, and computes the evolution of temperatures, microstructure, residual stresses and distortions. The model is today extensively applied in seeking adequate welding procedures in the fabrication of welded automotive parts in aluminum, and it has become an attractive alternative to the traditional procedure of trial and error based optimization of the welding parameters. In the present work, results from weld simulations have been compared to corresponding measurements during welding, and a very good agreement has been obtained. The model has also been applied investigating the possibilities of minimizing the distortions in the welding of automotive parts. The simulations have shown how the fixture design, the weld sequence as well as the welding parameters, significantly affects the resulting weld distortions. INTRODUCTION Extruded aluminum profiles are to an increasing extent applied in automotive components like space frames, engine cradles and windshield frames where weight saving is essential. Welding is a key operation in the manufacturing of such parts, and robotic GMA welding is by far the most commonly applied process for high volume production. A major problem associated with welding is the thermally induced deformations caused by the intense non-homogeneous heating and cooling of the material. These deformations are unavoidable in welding, but can usually be minimized to an adequate level by proper selection of the welding parameters and the fixture design. In order to obtain weld deformations within the geometrical tolerance limits, two principally different approaches can be applied, as schematically outlined in Figure 1. The loop on the left hand side of the diagram (i.e. the "physical welding" loop) illustrates the traditional trial and error based procedure applying welding experiments and a robotic welding unit. The welding is followed by measurements of the resulting distortions and the corresponding deviations from the nominal geometry. If the distortions are outside the tolerance limits, some adjustments are done for the welding conditions, the fixture- or the geometric design as indicated in the rectangular window of the figure before another component is welded. The welding is followed by measurements of the resulting geometry, and this loop is repeated until a certain combination yields weld distortions that are acceptable. The loop on the right hand side (i.e. the "virtual welding" loop) utilizes a computer instead of a welding cell, where the simulation results provide direct information on the positions the distortions are outside the tolerance limits. It is easy to change the input data systematically in order to analyze their individual effect on the resulting local and global distortions. A post (**) Institute for Energy Technology, Box 40, N-2027 Kjeller, Norway Hydro Automotive Structures Raufoss, Box 15, 2831 Raufoss, Norway processor tool can be applied to view animations of the time evolution of the thermally induced deformations. Such illustrations of the process give valuable insight and increased understanding of the process. This provides a basis for generating rules for a better design of both the fixture and the welding sequence. A complete replacement of real welding experiments by simulations in the fabrication of welded components is not realistic in the foreseen future, but there is no doubt that the trend is irreversible moving towards a more extensive utilization of computer simulations. The impetus for this development is the substantial potential for saving both costs and time in the preproduction stage. This has been the motivation for developing the weld simulation model WELDSIM (Ref. 1-4), which provides a mathematical description of the relation between the main welding parameters and the subsequent distortions and weld properties. The model has been developed and refined to a stage where it can be applied for realistic simulations of welding of complex aluminum structures and capture important effects of the geometric design and the weld parameters, as well as material related variables like alloy composition and base metal temper condition. Moreover, the last years significant developments on computer hardware has made it possible to carry out simulations on an ordinary PC within an acceptable amount of time. The WELDSIM model is today extensively applied in the pre-production phase in seeking adequate welding procedures in fabrication of welded automotive parts in aluminum. The present paper illustrates recent developments of the model. The accuracy of the simulations is illustrated by comparisons between simulation results and measurements, and an example from a case study on welding of an automotive component is presented at the end of the paper. Figure 1: Two different approaches for satisfying geometric tolerances in fabrication of welded structures. MODEL OUTLINE WELDSIM consists of a thermal, a microstructure and a mechanical sub-model that are connected as illustrated in Figure 2, and implemented in a finite element method (FEM) software framework. The thermal field influences both the microstructure evolution as well as the evolution of the thermally induced stresses and strains. At the same time there is a link between the microstructure and the mechanical sub-model to account for the important effect of softening of the heat affected zone due to reversion of hardening particles. This is taken into account by a separate (internal state) variable fifo. f/fo is computed by the microstructure model and transmitted to the mechanical model as a field variable. There is no "back-coupling" from the mechanical model to the microstructure model. Figure 2: The connection- and transition of data between the different sub-models of WELDSIM. The Thermal Model The thermal model is a fully 3D FEM solver with an implicit time integration scheme, where the specific enthalpy is the basic unknown. In WELDSIM, the heat source is modeled as a traveling double ellipsoid heat distribution function. In order to obtain realistic temperatures, the heat distribution function has been calibrated against measurements of the weld pool geometry. Moreover, the model has a correction algorithm for compensating numerically induced deviations in the total heat input. If present, the addition of filler wire material is modeled by a continuous activation of the finite elements in the reinforcement regions when they are approached by the heat source. The dimensions of these regions must, however, be predefined. The Microstructure Model Heat-treatable aluminum alloys obtain their main strengthening contribution from the precipitation of a fine dispersion of small particles from a supersaturated solid solution. These small particles are not resistant to the high temperatures imposed by the welding, and will partly or fully dissolve at short distances from the weld. This process is usually referred to as reversion, and is the main reason for the HAZ softening taking place during welding of heat-treatable aluminum alloys. In order to quantify this effect, the model considers the evolution of the volume fraction f of the small hardening particles. The state of the precipitate structure can then be described by the single normalized state variable f/fo, where fo refers to the initial volume fraction of the particles before the welding starts. As shown in Ref. 1, f/fo can be calculated as: |