Iron Core Saturation Analysis
Iron Core Saturation: Theoretical Foundations
Iron Core Saturation
Professor, what exactly is the phenomenon of iron core saturation?
A phenomenon where the magnetic flux density $B$ hardly increases even when the magnetic field $H$ is increased. When all the magnetic moments (magnetic domains) of the iron are aligned, further magnetization becomes impossible.
B-H Curve
The nonlinear $B$-$H$ relationship:
| Region | $\mu_r$ | Characteristics |
|---|---|---|
| Initial Region | 100–500 | Low magnetic field. Reversible. |
| Steep Rise Region | 1000–10000 | Magnetic wall movement dominates. |
| Near Saturation (Knee) | 10–100 | Magnetic domain rotation. |
| Saturation Region | ≈1 | $B \approx \mu_0 H + B_{sat}$ |
When saturated, $\mu_r \approx 1$... is it the same as air?
Exactly. A saturated iron core stops conducting magnetic flux effectively. This causes flux leakage, leading to reduced efficiency, increased losses, and heat generation.
Typical Saturation Flux Densities
| Material | $B_{sat}$ [T] | Applications |
|---|---|---|
| Electrical Steel Sheet (Si Steel) | 1.8–2.0 | Motors, Transformers |
| Permendur (FeCo) | 2.3–2.4 | Aerospace, High Density |
| Ferrite | 0.3–0.5 | High-Frequency Transformers |
| Amorphous Alloy | 1.5–1.6 | High-Efficiency Transformers |
| Nanocrystalline Alloy | 1.2–1.3 | High-Frequency Reactors |
Summary
- Saturation → $\mu_r ≈ 1$ — The iron core stops conducting magnetic flux.
- B-H Curve is Nonlinear — Requires Newton-Raphson iteration in FEM.
- Electrical Steel Sheet: $B_{sat} ≈ 2.0$ T — The upper limit for motor design.
The Physics of Iron Core Saturation—How Magnetic Domain Structure Creates the BH Curve's "Knee"
The "Knee Point" that appears on the magnetization curve of electrical steel sheets marks the transition stage where magnetic wall movement is complete and only magnetization rotation remains. The Bsat (saturation flux density) of soft steel is about 2.0–2.1 T, while ferrite is significantly different at 0.3–0.5 T. In transformer and motor core design, it is fundamental to keep the operating point below the knee point, but cases are increasing where B is set near the knee point due to demands for weight reduction and higher output. Accurately incorporating the nonlinear BH curve in the saturation region into FEM is crucial for the accuracy of iron loss and flux distribution, making the choice between Lagrange elements vs. Nedelec elements important.
Computational Methods for Iron Core Saturation
Nonlinear Magnetic Field FEM
$\nu = \nu(|\mathbf{B}|)$ ($\nu = 1/\mu$) depends on magnetic flux density. Solved using Newton-Raphson iteration:
1. Initial estimate (linear solution with e.g., $\mu_r = 1000$)
2. Update $\nu$ from B-H curve based on $|\mathbf{B}|$
3. Form Jacobian (tangent stiffness) using tangent permeability $d\nu/d(B^2)$
4. Iterate until residual reaches zero
It's the same nonlinear solver as for structural elastoplasticity.
Exactly. The B-H curve corresponds to the stress-strain curve.
B-H Curve Input
Input the steel sheet manufacturer's datasheet (magnetization curve) as a table. Notes:
- Extrapolation in Saturation Region — If data only goes up to $B = 1.8$ T, extend with slope of $\mu_0$.
- Interpolation Method — Spline or linear. Sharp bends negatively affect convergence.
Summary
- Newton-Raphson Iteration — Same as for structural elastoplasticity.
- B-H Curve Table Input — Pay attention to extrapolation in the saturation region.
Iterative Methods for Nonlinear Magnetic Field Analysis—Newton-Raphson Method and Handling BH Curves
Magnetic field analysis considering iron core saturation is a nonlinear problem, and the Newton-Raphson (NR) method is the standard iterative solver. The NR method has quadratic convergence (error decreases with the square) and is fast, but has the drawback of diverging with poor initial values. In the "steep gradient part (near the knee)" of the BH curve, the change in permeability is large, and the calculation accuracy of the Jacobian matrix directly affects convergence. In practice, two-stage methods providing a linear initial solution before switching to NR, or techniques combining it with Picard iteration (fixed-point method) to enhance stability, are adopted even in commercial tools like ANSYS Maxwell.