A unified rate-dependent anisotropic hysteresis model for grain-oriented, non-oriented, and directional electrical steels

Imagine you are trying to predict how a crowd of people moves through a giant, complex maze. In the world of electricity, this "crowd" is made of tiny magnetic arrows inside metal sheets, and the "maze" is the invisible energy landscape they have to navigate. This is the heart of magnetic hysteresis: the study of how materials remember their past magnetic states. When you run electricity through a motor or a transformer, these magnetic arrows spin and flip, but they don't just snap instantly; they lag, they get stuck, and they create heat. Engineers need to predict exactly how this happens to build efficient devices. For decades, the tools used to predict this were like trying to describe a 3D dance by only looking at a shadow on the wall. They worked okay for simple, uniform metals, but when it came to special "grain-oriented" steels—where the atoms are lined up in a specific, crystal-clear direction like soldiers in a parade—those old tools broke down. They couldn't explain why the metal behaved one way when you pushed it forward and a completely different way when you pushed it sideways, unless you used two totally different sets of rules for each direction.

This paper introduces a new, unified way to model that magnetic dance. The authors, Mario Bendra and Florian Poltschak, built a sophisticated mathematical engine based on a concept called the Fokker–Planck equation. Think of this as a high-definition weather forecast for the magnetic arrows. Instead of guessing where they go, this model tracks the probability of every single arrow pointing in every possible direction on a sphere. It combines three different "forces" that guide the arrows: the pull of the external magnet (Zeeman), the preference for a specific crystal direction (uniaxial), and the complex grid-like structure of the metal itself (cubic). The result is a single, shared set of rules that works for both the "forward" and "sideways" directions of high-performance steel, without needing to adjust the fundamental properties of the material for each test.

The Problem: The "Two-Faced" Steel

Imagine you have a piece of special steel used in electric motors. If you run a magnetic field along the grain (the "Rolling Direction"), the steel snaps into magnetism like a light switch—sharp, fast, and efficient. But if you run that same field sideways (the "Transverse Direction"), it behaves like a sluggish sponge, taking longer to react and creating a much fatter, rounder loop.

For years, engineers tried to model this with existing software. To make the math work, they had to adjust. They would tell the computer, "Okay, for the forward direction, the steel's maximum magnetism is this number. But for the sideways direction, it's actually 35% less." This was a mathematical adjustment to force the model to fit the data, but it didn't make physical sense. Magnetism is a property of the material itself; it shouldn't change just because you turned the sheet of metal. It was like saying a person is 35% shorter when they walk sideways. The old models were essentially adjusting to get the answer right.

The New Solution: A 3D Map for Magnetic Arrows

The authors decided to stop adjusting and start mapping the terrain properly. They built a model that treats the magnetic arrows as a crowd of people on a sphere, moving under the influence of thermal noise (jittering from heat) and energy landscapes.

Here is the magic of their approach:

1. One Set of Rules: They used a single, shared set of physical parameters (like the material's saturation magnetization) for both the forward and sideways directions. They didn't need to change the "height" of the steel's magnetism depending on which way they pushed it.
2. The "Skin-Depth" Insight: They noticed that as the frequency of the magnetic field increased (from 50 Hz up to 200 Hz), the behavior changed. They introduced a clever, frequency-dependent tweak to the "cubic" part of the energy landscape. This tweak aligns perfectly with the physics of how electricity penetrates the metal (the "skin depth"). It's as if the metal gets slightly "stiffer" to magnetic changes at higher speeds, and their model captures this naturally.
3. The "Two-Knee" Mystery: When pushing the steel sideways, the magnetic curve has a weird "two-knee" shape (it bends twice before flattening out). The basic rotation model couldn't see this. So, the authors added a "hysteron overlay"—a small, extra layer of math that acts like a set of tiny, discrete switches. This layer snaps into place exactly where the real-world measurements show the steel is switching its internal magnetic walls.

What They Found

When they tested their new model against real-world data from a high-grade steel called Hi-B 27M-OH, the results were impressive.

• Accuracy: They achieved an average error of just 0.069 T (Tesla) across eight different tests (forward and sideways at four different frequencies).
• The Fix: They successfully removed the "35% direction dependence" artifact. Their model found a single saturation magnetization value of 1.41×10⁶ A·m⁻¹ that worked for both directions.
• The Trade-off: To get this perfect unification, they had to accept a slightly lower value for the "cubic anisotropy" (a measure of how much the crystal structure resists turning) than what is seen in a perfect single crystal. The paper explains this isn't a mistake; it's because the steel is made of many tiny grains, and the specific texture of the steel "hides" some of the crystal directions, effectively reducing the overall resistance the model sees.

They also tested the model on non-oriented steels (randomly arranged grains) and found it reduced perfectly to the standard, simple behavior expected for those materials, proving it doesn't invent fake complexity where none exists.

What It's Not (and What's Next)

The authors are very clear about what their model doesn't do yet.

• It's not a magic bullet for every angle: The model works perfectly for the main forward and sideways directions, but predicting the behavior at a 45-degree angle is still an open challenge.
• It's not a full physics simulation of the whole sheet: At very high frequencies (above 150 Hz), the magnetic field doesn't penetrate the metal evenly (the skin effect). The model handles this with a clever mathematical shortcut, but the authors admit that for the highest precision, you would eventually need to couple this with a full electromagnetic solver to see the field inside the metal slice-by-slice.
• It's not perfect for every tiny detail: While it handles the main "major loops" of magnetism beautifully, it doesn't yet fully capture the complex memory effects of tiny "minor loops" (small back-and-forth movements) or the effects of DC bias (a constant magnetic offset).

The Bottom Line

This paper presents a unified, physics-based model that finally treats grain-oriented and non-oriented electrical steels with a single, consistent set of rules. It stops the "adjusting" of changing material properties based on direction and replaces it with a sophisticated understanding of how magnetic arrows rotate and switch in a 3D world. While it has limits at extreme frequencies and angles, it provides a robust, open-source foundation (called hysterpy) that engineers can use to simulate motors and transformers more accurately than ever before. It turns a messy, two-faced problem into a single, coherent story.