A new customer program utilized various Cast and extruded Al components. Early indications that porosity was a problem were inconclusive for identifying the specific materials to be further analyzed. An investigation used to rate the severity of porosity on gas metal arc welds (GMAW) of several extruded and cast aluminum component combinations was completed. The goal was to obtain an unbiased assessment of porosity formation during GMAW on representative components. The approach was to provide weld process development assistance by performing a design of experiments (DoE) study. Besides examining the 15 component combinations, the effect of baking and abrading the components prior to welding was assessed.
Experimental Procedure and Sample Preparation
This DoE approach was based on a 32-run Hadamard Matrix design. A portion of the DoE design is shown in Table 1. While this traditional experimental matrix is non-standard as used by today’s common statistical programs, this matrix has a long history of being used for designing DoE’s. Before modern computers, the Hadamard matrices (developed prior to 1893) was one of the original approaches for generating DoE designs. However, computer computations favor using the Yates matrix which can produce DOE’s with greater ease. Still, there are a number of advantages for selecting the Hadamard matrix; chiefly, it regularly produces smaller DoE designs, these designs can be subdivided into additional orthogonal and balanced DoE units. This latter feature allows it to more efficiently categorize the 15 material combinations examined. The experimental levels for the material combinations were generated using a standard method of specifying multiple DoE factor levels into a robust DoE design. This design could therefore distinguish between the material combinations to pick out the materials causing the issue.
GMAW was performed using appropriate welding equipment and settings for Aluminum. The same set of welding parameters were used across all material combinations. The welding equipment was validated to not contribute to the porosity obtained. Each material combination was tested twice in the DoE. One material combination was used in all four quadrants of the DoE (Baking and Abrading combinations).
In the preparation for welding, all of the test samples were first cleaned (wiped) with methanol to remove any cutting or other fluids from the surface. As specified in the DoE, baking occurred first and followed by abrading (where appropriate). Some samples received neither treatment.
Table 1 – A portion of the DoE specified with Practical Units and Listed in the Order of Testing
The test measurements included both the fillet weld break test and various measures of porosity using metallographic examinations. The fillet weld break test evaluates the weld along the weakest plane through the throat of the weld. If excessive porosity is aligned in the cross section, it will fail along that plane. In contrast, the weld cross section only determines porosity at a specific point transverse to the weld.
Fillet Break Results
The fillet breaks were performed on 25mm (1-in) long portions of the weld specimen. In order to develop a visual rating system, five samples were selected to represent the primary numerical values creating a 1-5 rating scale. The fillet break DoE data was statistically analyzed and a reduced regression model was developed. The material combination, baking, and abrading were all identified as significant variables by the DoE. Although the published results were much more thorough, an example of one of the results is shown in Figure 1. The box plot shows the wide variation of porosity observed in this study.
Figure 1 – Box Plot of Fillet Break Rating vs Material Combination ID
Metallographic sections were also examined. Because of the point-specific characteristic of metallographic evaluations, the results can be dramatically different from the fillet break test evaluation. DoE reduced model results were again obtained and extensively analyzed. In this case, the material combination was a much more prevalent factor in the metallographic measures of porosity. An example of one result is shown in Figure 2. The boxplot shows the effect of material combination on percentage of weld area porosity.
Figure 2 – Box Plot of Percent Porosity of the Fused Area vs Material Combination ID
This study detected specific materials in the material combinations that were responsible for GMAW porosity unrelated to weld equipment issues. The remedial measures of baking and/or abrading were identified as an alternative solution. It is uncertain that these remedial measures will be utilized, but a knowledge of pin-pointing which materials to be further investigated was invaluable in identifying the problem for further investigation.