Annealing CRGO laminations: when you need it and how it changes properties

1. What annealing really fixes in a lamination stack

Very short version:

• Cutting, punching, bending → plastic strain at edges and corners → dislocations, residual stress, disturbed Goss texture. That pushes up core loss and drags down permeability.
• Stress-relief annealing (SRA) at ~700–850 °C in controlled atmosphere lets recovery and partial recrystallization clean up that damage, releasing stored energy and restoring a big chunk of the original magnetic behavior.
• Big ANSI/IEC-style datasheet values from producers like POSCO and JFE Steel are typically guaranteed after a defined SRA cycle, not in the “as-cut” condition.

So if you buy laminations based only on catalog numbers but skip annealing, you’re quietly running a different material than you think.

2. When you actually need annealing on CRGO laminations

You don’t anneal “because everybody does.” You anneal because something in your process has pushed the steel away from what the mill delivered.

Let’s break it down by situation.

2.1 Cases where annealing is basically non-negotiable

a) Semi-processed or “customer-annealed” grades

Some electrical steels are sold with the expectation that the final core loss and B-H behavior will only be reached after the customer runs a stress-relief cycle. JFE’s JNA / similar semi-processed lines and multiple “N-CORE” / “PNA-CORE” style products from producers fall in this camp.

If the datasheet explicitly says:

• “Properties specified after stress-relief annealing by customer”
• or gives a standard “750–840 °C, 1–2 h, non-oxidizing atmosphere” test condition

…then running these grades without SRA is asking for:

• Higher than catalog core loss at 1.5–1.7 T
• Lower effective permeability
• More unit-to-unit scatter

b) Wound cores from CRGO strip

Wound cores see:

• Tight bending radii
• Continuous plastic strain along the strip
• Clamping and banding during winding

Without annealing, those stresses don’t magically relax. Vacuum or protective-gas annealing of wound CRGO cores is standard precisely to drop iron loss and magnetizing current back to mill-level values.

If you are buying wound cores without SRA, treat them as a different loss class than the raw strip.

c) Heavy punching, slotting, or multiple hole patterns

The more metal you displace, the more you cold-work the edges. Shear-cut regions can be magnetically “damaged” up to a few hundred microns in from the cut line, with visible microstructural changes and harder edge zones.

If your lamination design has:

• Multiple holes for clamping, leads or sensors
• Long punched windows
• Small radii around yoke/leg transitions

…then a post-stack anneal is usually the only realistic way to claw back the lost permeability and reduce the local loss hotspots.

2.2 Cases where annealing is optional but usually pays back

You can run many stacked CRGO cores “as-cut” and still meet nameplate. But you may be giving away efficiency and margin.

Typical literature numbers for moderately deformed silicon steels show:

• 10–20 % reduction in total core loss at ~1.5 T and 50/60 Hz after a suitable stress-relief anneal, relative to as-cut samples with edge strain.
• Noticeable improvement in relative permeability and reduction in magnetizing current.

That kind of delta, on a 10+ MVA unit, is not small.

So annealing usually makes sense when:

• You’re chasing high efficiency tiers or utility tender limits with tight no-load loss caps.
• The lamination thickness is already low (0.23–0.27 mm) and you want to exploit the full material capability.
• Your design margin on flux density is narrow, e.g. you’re sitting near 1.7–1.8 T at nominal.

2.3 Cases where annealing is a bad idea or needs extra care

Not everything likes a trip to 800 °C.

a) Domain-refined / laser-scribed grades not meant for SRA

Domain-refined (laser scribed) GOES grades rely on micro-strain patterns on the surface to subdivide domains and cut loss by ~10–15 % compared with non-refined steel.

A classic trap: some domain-refined CRGO is explicitly marked “not suitable for stress-relief annealing”. If you anneal it anyway, you:

• Relax the laser-induced strain
• Wipe out the domain refinement effect
• End up with higher core loss than if you had never bought the expensive DR grade in the first place

So: check datasheets carefully. If it says “for applications without SRA”, believe it.

b) Assemblies with non-metallic fixtures not rated for high temperature

Any of these in the core stack?

• Epoxy, adhesives, tapes
• Organic spacers, 3D-printed polymer parts
• Low-temp gaskets or seals

They won’t survive 800 °C in nitrogen. Obvious, but this gets overlooked in repair/rebuild scenarios. If the lamination stack is already integrated into a larger assembly, SRA might be off the table unless you redesign the mechanical concept.

c) Coatings not designed for re-anneal

Insulation coatings fall into three rough buckets:

• Mill anneal / SRA-stable inorganic (ceramic-like)
• Thin organic hybrids
• Pure organic varnish-type

Only the first group is comfortable with repeated high-temperature cycles. Coating datasheets from makers like Aperam and others explicitly state which coatings will maintain insulation after SRA and which are “as-cut only.”

If the insulation bakes, cracks, or bonds laminations together, you may lose lamination factor and introduce extra mechanical stress.

3. How annealing changes CRGO lamination properties

Let’s treat your lamination stack as a small ecosystem. You change the heat history once; multiple properties shift at the same time.

3.1 Magnetic properties

Key shifts you’ll actually notice in test data:

• Core loss (W/kg)
  • Edge strain, bending, and punching raise both hysteresis and classical loss components.
  • Recovery and partial recrystallization during anneal push losses back down.
  • For typical CRGO grades, a well-chosen SRA cycle can bring losses close to “virgin” mill-test values, but extreme deformation damage may not be fully reversible.
• Permeability / magnetizing current
  • Residual stress pins domain walls; you see flatter magnetization curves and higher current for the same induction.
  • Annealing frees a lot of those pinned walls, giving higher µ and lower no-load current at your design induction.
• Anisotropy and directionality
  • Some studies show that annealing around 800–825 °C for 2–4 h provides a sweet spot: good loss, high permeability, and manageable anisotropy between rolling and transverse directions.
  • Too low a temperature: incomplete stress relief → you keep the damage.
  • Too high or too long: grain growth and texture changes erode the benefit.
• Magnetostriction / noise
  • Magnetic annealing or annealing under mechanical stress can tweak magnetostriction. Done right, it can smooth out vibration behavior; done wrong, it can hurt the steel’s magnetic properties.

3.2 Mechanical properties

The mill has already done the big recrystallization steps. Your SRA is mostly “clean-up.”

• Hardness drops slightly, especially at highly strained edges.
• Ductility improves, which can reduce micro-cracking risk in any later handling.
• Residual stress dies down, which also means less tendency for stacks to “spring” or distort when unclamped.

None of this turns CRGO into butter, but it does make the lamination stack less fragile in service and during final core assembly.

3.3 Coating and dimensional behavior

• Coating stability
  • Many inorganic coatings are specifically qualified to go through an SRA cycle at 750–840 °C without losing insulation resistance or flaking.
  • Organic-heavy coatings may discolor, shrink, or partially carbonize; this might still be acceptable for insulation but can change friction and stackability.
• Dimensional drift
  • You will see some dimensional relaxation if the steel has significant residual stress.
  • For tight-tolerance cores, it’s usually better to anneal before final machining of frames, shims, or fit-critical interfaces.

4. Indicative “before vs after” table for SRA on CRGO laminations

The numbers below are not a spec. They’re typical ranges seen in literature and practice when you take a reasonably deformed stack of fully processed CRGO, then run a good SRA cycle (around 800 °C in non-oxidizing atmosphere for a few hours). Actual results depend heavily on grade, deformation, and furnace discipline.

Again, treat those as orders of magnitude, not promises.

5. Process parameters that matter to lamination buyers

You may not run the furnace yourself, but your supplier’s choices will show up in your test bay.

5.1 Temperature window

In practice, industrial SRA cycles for GOES sit around:

• 700–900 °C, often tighter at 800–825 °C, depending on grade and coating.

Too low: incomplete recovery. Too high or too long: grain coarsening, coating issues, or unwanted changes in texture.

5.2 Time at temperature

• Common dwells are 1–4 hours at peak temperature.
• Going from 0.5 h to ~2 h often brings a noticeable drop in core loss; extending to 4 h may give diminishing returns or only shift anisotropy.

For purchasing, you don’t need the exact curve – you just need to know that your supplier’s recipe isn’t a 20-minute flash bake.

5.3 Atmosphere and pressure

Datasheets and process notes repeatedly emphasize:

• Neutral or slightly oxidizing atmospheres (e.g. dry nitrogen, sometimes specific DX gas mixes) to avoid heavy scale and to maintain coating.
• For wound cores, vacuum annealing is common to improve temperature uniformity and protect coatings.

Ask your supplier:

• What atmosphere do you use?
• Is it continuously monitored?
• Any oxygen or dew point limits?

5.4 Heating and cooling profile

Large cores do not like thermal shock.

• Fast ramps can introduce new thermal stresses, especially in thick wound cores with large cross-sections.
• Controlled heating and cooling (multi-step ramps, holding zones) avoid internal cracking and distortion.

If you are buying large power-transformer cores, it’s reasonable to expect your lamination supplier to have:

• Logged temperature profiles
• Max gradient limits (e.g. °C/min) for big stacks

6. A short design + purchasing checklist

If you’re specifying lamination stacks or cores, here’s a quick practical list. You can treat it almost like a talk track with your supplier.

1. Grade and condition
   • Is the steel fully processed or semi-processed / customer-annealed?
   • Does the datasheet quote losses “as-cut” or “after SRA”?
2. Domain refinement
   • Is the grade laser-scribed or otherwise domain-refined?
   • If yes, is it compatible with stress-relief annealing?
3. Coating
   • Coating type and thickness.
   • Confirm it is qualified for one full SRA cycle at your target conditions.
4. Geometry and processing
   • Cutting method: shear, punch, laser, waterjet.
   • Any heavy bending, multiple holes, or complex windows?
   • Wound core vs stacked EI/step-lap laminations.
5. Annealing recipe
   • Peak temperature and dwell time.
   • Atmosphere (gas type, vacuum, dew point limits).
   • Heating/cooling rate limits, especially for large cores.
6. Verification
   • Are loss and magnetizing current measured on cores after the actual production SRA?
   • What sampling plan is used?
   • Can you get a typical test report with W/kg and magnetizing current vs catalog?

This is the stuff that separates a “CRGO core” that meets the drawing from one that quietly saves you watts and copper every hour of its life.

7. FAQ: Annealing CRGO laminations for purchasing and engineers