Choose the shunt validation scale that preserves the mechanism governing the decision. Coupon, panel and full-unit tests each provide distinct evidence about material behavior, local interfaces and the complete source boundary. [1]
Match the test level to the question
| Validation level | Useful evidence | Principal transfer limit |
|---|---|---|
| Local coupon or packet | Directional magnetic response, coating or contact behaviour | Does not reproduce the transformer leakage source or whole mounting circuit |
| Representative tank panel | Local shunt, wall, aperture and mounting interaction | Imposed field and cooling must represent the relevant region |
| Full transformer | Actual coupled source, structure and aggregate performance | Measurements still need uncertainty and condition control |
Preserve the source-field character
A uniform laboratory field can test a material response but may miss a steep spatial gradient, changing direction or edge concentration. Record the field components, spectrum and spatial variation imposed on the sample. A panel test needs enough surrounding geometry to reproduce the local return paths and fringing.
The bushing-region paper considers one axial conductor and constant tank permeability. Its analytical-to-FEM agreement does not establish a three-phase lead-exit result. Likewise, the multi-step shunt study compares its field predictions with 3D simulation under stated source assumptions; it is not a prototype loss or temperature test. [2] [1]
Check what scaling changes
For conducting parts, geometry relative to skin depth matters. Reducing thickness while retaining material and frequency changes that ratio. A geometrically scaled copper plate therefore does not automatically preserve current distribution or loss density.
For magnetic material, the operating B–H range and direction must remain representative. Geometric similarity does not preserve nonlinear loading unless the source and material state are also addressed. Electrical contacts, coating gaps and fastener dimensions may not scale in the same way as the main structure.
Keep thermal similarity separate
A small panel’s cooling and thermal mass can differ greatly from an installed tank region. Record flow access, conduction paths and the time scale of the test. A measured panel temperature cannot be transferred to the full transformer simply by matching magnetic field magnitude.
Use each level to reduce a specific uncertainty. Coupon evidence can qualify the material or interface; a panel can validate the local field and loss model; the full unit can assess coupled performance. State the remaining extrapolation after each stage.
Assume a copper panel has thickness 3.0 mm and an electromagnetic skin depth of 9.35 mm at 50 Hz for the stated material condition. Its thickness/skin-depth ratio is 0.321. A half-scale 1.5 mm panel at unchanged material and frequency has ratio 0.160, so geometric scaling has changed the current-distribution parameter. To restore that ratio with the thinner panel in the ideal linear conductor model, frequency must increase fourfold to 200 Hz. That change can invalidate other magnetic and thermal similarities. Use the coupon for the quantities it actually preserves and retain a representative panel or full-unit check for the coupled decision.
List the measured and simulated quantities, preserved mechanisms and remaining scale differences. Match the design decision to that validation coverage.
References
- Z. Jurković, B. Jurišić and T. Župan, Multi-Step Approach for Fast Calculation of Magnetic Field in Transformer Tank Shields, Energies 17, 1378, 2024. DOI: 10.3390/en17061378.
- S. Khan et al., Computation of Stray Losses in Transformer Bushing Regions Considering Harmonics in the Load Current, Applied Sciences 10, 3527, 2020. DOI: 10.3390/app10103527.