Magnesium alloys are becoming a popular choice for making vehicles and airplanes.

These metals are strong, lightweight, and easy to shape, which makes them ideal for reducing energy use and emissions.

However, predicting how long parts made of magnesium alloys will last has always been a challenge. Over time, these materials can develop tiny cracks from repeated stress, leading to potential failure.

Until now, engineers relied on traditional methods to predict when these cracks would form.

But these methods require constant adjustments for different conditions, making them less practical for industries where parts are exposed to varying loads and directions.

To solve this issue, a research team led by Professor Taekyung Lee from Pusan National University in South Korea has developed a new prediction model.

The team combined machine learning with physical modeling to create a tool that can better predict the “fatigue life” of magnesium alloys.

Their findings were published in the Journal of Magnesium Alloys.

The model uses a neural network to analyze stress patterns in the metal during repeated use.

It also includes a physics-based system that helps the machine learning component stay grounded in the rules of material science. Together, these tools predict how and when cracks are likely to form.

The researchers tested their model on a magnesium alloy called AZ31. They collected data on how the metal behaved during repeated stress cycles, called hysteresis loops, where the metal stretches, bends, and returns to shape.

“The neural network learns from these stress cycles and identifies patterns in how the material behaves,” explains Prof. Lee. “Then, the physics-based model converts this data into a prediction of how many cycles the metal can endure before it cracks.”

This new approach is more accurate and adaptable than traditional methods. Because the machine learning system improves as it analyzes more data, it doesn’t need manual adjustments for different conditions.

This makes it useful for predicting metal wear in real-world environments where loads and stress directions constantly change.

This breakthrough could lead to safer, lighter, and longer-lasting designs in cars, planes, and other critical applications.

By improving predictions for magnesium alloys, manufacturers can create more reliable and cost-effective components, reducing risks and extending the life of important parts.

Source: KSR.