Magnetostriction Analysis
Theory: strain caused by magnetisation
Overview
Is a transformer's hum the wires vibrating?
The main cause is core magnetostriction. Core steel stretches and shrinks very slightly each time it magnetises. Since magnetostriction stretches the same way whether magnetisation is positive or negative, it cycles twice per AC period, so the fundamental tone is twice the supply frequency. Strain is tiny, around 1 ppm, but in large cores the vibration amplitude reaches micrometres and travels through oil and tank as sound. Conversely, materials with large magnetostriction make precise actuators and powerful underwater sound sources. Magnetostriction analysis couples the magnetic field with structural deformation.
Stroke, force and field
λ is magnetostrictive strain, L rod length, E^H Young's modulus at constant field, A cross-section, N turns and I current. Magnetostriction doesn't depend on flux direction, so AC excitation vibrates at twice supply frequency.
Joule's discovery and giant magnetostriction
Joule discovered magnetostriction in 1842 by measuring the change in length of an iron rod when magnetised. Iron and nickel show little magnetostriction, yet nickel-based sonar was built around the Second World War. In the 1970s the US Naval Ordnance Laboratory found that alloys of iron with rare earths terbium and dysprosium showed magnetostriction tens of times nickel's, naming it Terfenol-D after its elements (Ter for terbium, Fe for iron) and the laboratory (NOL).
Worked examples
Example 1: saturation magnetostriction and 100 mm rod stroke
| Material | Saturation strain | 100 mm rod stroke |
|---|---|---|
| Iron | about 10 ppm | 1.0 μm |
| Nickel | −33 ppm | 3.3 μm |
| Galfenol (Fe-Ga) | 250 ppm | 25 μm |
| Terfenol-D | 1,600 ppm | 160 μm |
Example 2: Terfenol-D actuator (100 mm long, 10 mm diameter, 1,000 ppm)
| Item | Value |
|---|---|
| Stroke | 100 μm |
| Blocked force (E = 30 GPa) | 2,356 N |
| Stiffness | 2.36×10⁷ N/m |
| Required field | 80 kA/m |
| MMF (current with 500 turns) | 8,000 AT (16 A) |
| Work density (½Eλ²) | 15 kJ/m³ |
By the same measure, the earlier piezo example (PZT stack, 890 ppm at 1.5 kV/mm, E = 48 GPa) has a work density of 19 kJ/m³, comparable in size.
It gives force similar to piezo, but needs as much as 16 A.
Magnetostrictive actuators are driven by magnetic field, so coils need large current. Piezo is voltage-driven and needs almost no current when static, whereas magnetostrictive coils heat continuously through resistance, raising temperature and changing properties. In return they drive at low voltage, are rugged, and handle large forces and large-amplitude vibration, suiting powerful underwater sound sources and high-force vibrators. Magnetostrictive materials also stretch more under compressive prestress, which aligns domains, so they're normally held by springs. Analysis couples magnetostrictive nonlinearity, prestress, eddy currents and heating.
Analysis workflow
- Compute field distribution in the material from coil current by magnetic analysis.
- Enter field- and stress-dependent (nonlinear) magnetostriction data.
- Pass magnetostrictive strain to structural analysis for displacement and stress, feeding back if needed.
- For AC drive include eddy currents and heating, with temperature-dependent properties.
- For transformer noise, analyse from core vibration through tank vibration to sound radiation.
“A transformer that should have been quiet”
At a substation, a transformer designed as low-noise was louder than expected once installed, with a particularly strong 200 Hz component. Magnetostrictive core vibration contains 100 Hz and its multiples, and 200 Hz nearly matched a tank-wall natural frequency, causing resonance. Core-only analysis showed small vibration, but analysis coupled through to the tank reproduced the cause. Adding stiffeners to shift the tank frequency brought noise down to target.
Common mistakes
Mistakes and fixes
| Mistake | Effect | Fix |
|---|---|---|
| Evaluating at supply frequency | Wrong frequency | Use twice the frequency |
| Treating magnetostriction as linear | Wrong stroke | Use nonlinear data |
| Ignoring prestress | Strain underestimated | Include prestress dependence |
| Ignoring coil heating | Properties drift with heat | Couple heat |
| Core-only noise evaluation | Tank resonance missed | Include the tank |
I'd like to learn related topics.
Related pages include piezoelectric analysis, magnetic bearing analysis, Lorentz force, EMAT simulation and electroactive polymers.
Related Topics
Experience the theory firsthand with the interactive simulator for this field
All Simulators