Abstract:
This thesis presents an analysis of data based on the results of a previous study into the
path independence of friction stir welding (FSW). The original study was conducted in two
phases using AA2024 and AA2198 aluminum alloy material. In Phase I, welds were made with
six different tool designs, and tensile data was entered into Statgraphics® software as part of a
design of experiments (DOE) approach for the purpose of optimizing the weld parameter process
windows for each tool design. Phase II included a round robin study where welds were produced
at four sites to evaluate site-to-site variability. In the present study, testing of the welds included
additional tensile testing, full-field microhardness testing, and conductivity testing of the welds
produced in the prior two phases of the program. The welds were inspected for defects, and the
method of failure on the tensile specimens is discussed. Tensile data was evaluated statistically
using Statgraphics® software. Previously, as part of Phase I, that data was evaluated using the
techniques of chapter nine of the Metallic Materials Properties Development and Standardization
(MMPDS), and design allowables were calculated at that time. In the present study, Phase II
data was compared using the design allowables from Phase I.
Variability was not found to be significant when evaluated according to location of the
tensile coupon along the weld joint line or as a function of weld parameters in the process
window. Although site-to-site variability was significant, it was low, with the highest variation
for each material being 2 to 5.5 ksi. No significant outliers were identified.
There is evidence that the friction stir welding process is path-independent and that a
defect-free weld is uniform in tensile strength from beginning to end. There is also evidence that
a DOE approach can be used to optimize the weld parameter process window for any tool in
order to identify a range of weld parameters where a defect-free weld can be produced.
