A prototype shunt trial should measure local hot-spot improvement and the relevant total-loss change on one controlled basis. Combine spatial field and temperature evidence with a matched power comparison. [1]
Register the local measurement plan
Place observations at predicted shunt peaks, ends, nearby tank regions and plausible hot-spot migration locations. Retain sensor coordinates relative to the assembly. A sensor left at the original hot spot may miss the new limiting location after the shunt is installed. [2]
Specify what each field measurement senses: component, orientation, bandwidth and sampled region. For temperature, record the surface or internal location, attachment and any disturbance introduced by the sensor. Thermal imaging also needs an appropriate surface and viewing basis; colour scales cannot be compared fairly if they change between trials.
| Measurement group | Purpose | Control to retain |
|---|---|---|
| Tank and shunt field | Check collection and bypass distribution | Excitation, orientation and position |
| Local temperature | Locate the governing thermal condition | Initial state, cooling and elapsed time |
| Aggregate power | Measure the relevant watt difference | Current, waveform and temperature basis |
| Assembly condition | Account for construction changes | Mounting, contacts and oil access |
Make the before-and-after comparison credible
Keep winding current, waveform, relevant temperature and cooling condition matched. Record any changes needed to install the shunt. If supports or flow paths also change, the trial evaluates that combined arrangement rather than magnetic shielding alone.
Use the laboratory’s established total-loss and component-loss accounting. The winding-asymmetry paper itself notes that separate tank losses are not directly measured through the aggregate transformer loss test. Subtracting estimated winding contributions introduces additional uncertainty and assumptions. [3] [2]
Check whether the saving is resolved
Define the watt difference on the chosen measurement basis and evaluate its uncertainty. Shared calibration or repeated-test errors may be correlated, so the uncertainty of the difference should follow the actual measurement relationship. A small saving obtained by subtracting large powers may not be distinguishable.
Use repeat reference conditions to check drift. Compare local observations with model predictions at the same locations, not just with the most favourable part of a field plot. The multi-step shield study benchmarks a field calculation against 3D simulation; prototype measurements provide a different level of evidence. [1]
Assume paired total-loss measurements of 12,000 W without a shunt and 11,850 W with it, each with independent standard uncertainty 100 W. The estimated saving is 150 W, while standard uncertainty of the difference is √(100² + 100²) = 141.42 W. A stated ±2u interval is −132.84 to +432.84 W and includes zero. An assumed local hot-spot reduction from 105 °C to 95 °C therefore supplies a different result from a statistically resolved total-loss saving. Retain the temperature improvement and improve the paired electrical measurement before pricing a guaranteed 150 W reduction.
Report local temperature change and resolved aggregate loss change separately. Identify the measurement coverage and uncertainty behind each finding.
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.
- I. A. Hernandez-Robles et al., Analysing and Computing the Impact of Geometric Asymmetric Coils on Transformer Stray Losses, Applied System Innovation 7, 26, 2024. DOI: 10.3390/asi7020026.
- B. N. Taylor and C. E. Kuyatt, NIST Technical Note 1297, Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, 1994, sections 5 and 6.