Visualizing the Composition of Aluminum Alloys

by Colin Fletcher

Hundreds of aluminum alloys have been developed for a wide variety of applications since the first industrially significant aluminum alloy, Duralumin, was created over 100 years ago. Many alloys of commercial importance were registered with bodies such as the Aluminum Association (AA), who documented the alloy designation and its composition limits, as part of industry efforts to standardize alloy designations.

The Aluminum Association alloy registration documents currently contain over 900 alloys, and this composition data can be used to visualize the composition-space currently filled by commercial aluminum alloys. With so many alloys covering a wide range of compositions, it can be difficult to determine how similar or different they may be from one another. Since this seemed like a good application for visualization, I used this dataset to map out the aluminum alloy composition space.

For the following visualizations, I use wrought and cast alloys listed in the AA “Teal Sheets” and “Pink Sheets”, respectively. Alloys are classified into families based on their primary alloying elements as described in ANSI H35.1; examples include Al-Cu (2XXX) and Al-Mg-Si (6XXX). Each alloy was assigned a color to represent its family classification.

From each alloy’s record, a nominal value was calculated from the minimum and maximum composition limits. This value, which is in the center of the allowed composition range for each alloying element, was used for plotting. Thirty alloying elements are also present throughout the sheets, which is highly dimensional for composition data. By using Principal Component Analysis (PCA), the data can be reduced to 2 or 3 dimensions, which is easily visualized.

A direct reduction to two dimensions using PCA obscures the alloy families (as does standardizing the composition data), but a suitable 2D projection of a 3D reduction shows clear differences between alloy families in the composition-space, as expected. The families overlap in the center near the 1XXX series, which contains various commercially pure aluminum alloys. As additions of alloying elements increase, alloys move further from this location. Some of the more significant alloying elements are indicated with axes on the plot, showing where certain additions move alloys in this space.

The central region is also quite dense and contains many alloys with small alloying additions and composition differences which are small in magnitude. A heatmap of the 2D visualization shows just how dense that central region is. Approximately one third of the entire dataset is located in the three brightest hexagons in the center!

Of course, the compositions are more easily visualized in 3D, and I have created an interactive 3D plot for just that purpose. Feel free to explore the aluminum alloy space for yourself!