WDJ3.2 Finite Element Modeling of Self-Reacting Friction Stir Welding for Designing Laser Peening Patterns

Tuesday, June 22, 2010: 2:00 PM
409 (Meydenbauer Center)
Mr. Anupam S. Badamikar , National Institute for Aviation Research, Wichita State University, Wichita, KS
Ms. Ester Sanseviero , Metall-Kunststoff-Engineering, Immenstaad, Germany
Dr. Gerardo Olivares , National Institute for Aviation Research, Wichita State University, Wichita, KS
Dr. Dwight Burford , National Institute for Aviation Research, Wichita State University, Wichita, KS
A 3D finite element model for self-reacting friction stir welding of a flat plate is developed using LS-DYNA. It is designed to help develop Laser Peening patterns for forming and / or straightening friction stir welded structure. This new thermo-mechanical model is capable of predicting the transient temperature distribution and stress field in the component being welded or processed.

 

 The modeling was carried out for three different cases with increasing order of complexity of applied loads. In the first case, only the moving heat source load was applied to take into account effect of weld heat generation. The second and third case added clamping loads and tool load respectively to the case one model. Additional cases for heat flux distribution in the weld zone are under study to identify effect over residual stresses.

 

 The model setup includes the plate and welding fixture. Tool load during the process includes the pinch force imposed by drawing the top and bottom shoulders together using forge load (top) and the draw pin load (bottom); model includes effective pinch force. The simulation sequence begins with plate clamping, followed by welding, subsequent cooling, and unclamping. For initial model validation testing, the data is collected using an ARAMIS optical 3D deformation analysis system, strain gages, and thermocouples. ARAMIS displacement and strain contour data shows plate longitudinal and lateral shrinkage and the patterns compares well with the predicted data. Comparison of the measured peak temperature data with the model prediction shows maximum 9% deviation. For the far field strain gage unclamp state data, longitudinal strains show average 10% variation; whereas lateral strains show 8% variation.

 

 Modeling results show maximum predicted longitudinal residual stress of 309MPa. For the cases modeled, the mechanical clamping loads found to have a minimal influence on the residual stresses developed in the weld zone as compared to the tool load.