Replacing only the damaged phase can avoid disturbing two serviceable windings. It can also leave a rebuilt transformer with one new mechanical and insulation history alongside two old ones. The question is whether that mixed assembly meets its intended duty, not whether every phase has the same date of manufacture. [1]
Start by confirming the fault’s extent. A failure concentrated in one coil can still contaminate neighboring insulation, move shared clamps or affect the core. The proposed repair boundary should be based on observed damage and accessible evidence rather than on the phase that tripped first.
Define which similarities are required
Some phase differences belong to the original design. Lead lengths, core-limb position and internal routes can differ. Other differences arise from the replacement winding. Compare each result with the expected construction and the measured conditions; an unexplained difference deserves investigation, while an intentional one deserves documentation.
| Comparison | Conditions to align | What an unexplained difference suggests |
|---|---|---|
| Turns and tap schedule | Same connection interpretation and tap identity | Section count or connection issue |
| Winding resistance | Defined temperature and comparable paths | Conductor, joint or path difference |
| Active dimensions | Same compressed/manufactured state | Geometry or support mismatch |
| Leakage behavior | Same test basis and tap | Changed winding distribution or position |
| Insulation assessment | Appropriate phase boundaries and condition | Retained-phase damage or new-phase defect |
| Mechanical restraint | Shared assembly condition | Unequal compression or load path |
The comparison should include the retained phases as found and after assembly. Removing a common clamping member to install the new coil can change their condition. A pre-dismantling result cannot describe a later assembly movement that was never checked.
Assess the shared interfaces
The new phase must fit the existing core, support structure, leads and barriers. Matching turn count cannot settle the condition of a shared end-insulation piece. A replacement coil with different springback can alter how the common clamp distributes load. These are interface questions, separate from the quality of the new winding itself.
Keep the thermal comparison equally specific. The new phase may use a revised conductor arrangement or drying history, while the retained phases have different moisture and ageing evidence. Do not derive their condition from the new coil’s manufacturing record. Review the complete unit’s cooling paths and intended operating limits.
At equal temperature, a change in conductor or joint resistance changes the winding’s I²R loss. Compare the repaired phase with its own expected path resistance and lead geometry, rather than treating three unequal phases as an identical-resistance population. [2]
Compare the new phase on matched conditions
Assumptions: equal-temperature phase resistances after repair are 0.500, 0.505 and 0.530 Ω, with the last phase rewound. The design’s corresponding path predictions are 0.500, 0.505 and 0.510 Ω. The new phase is 0.020 Ω, or 3.92%, above its own prediction. Against an assumed ±1% agreement gate, its allowable difference is 0.00510 Ω and the observed excess over that allowance is 0.01490 Ω. Hold release and reconcile conductor length/area and joints in the replaced phase. Its 3.58% deviation from the measured three-phase mean is a different statistic and would partly absorb the same fault into its reference. Check retained-phase dielectric and shared-clamp evidence independently. [1][2]
References
- U.S. Bureau of Reclamation, FIST Volume 3-31, Transformer Diagnostics, June 2003.
- U.S. Department of Energy, DOE-HDBK-1011/4-92, Electrical Science, Volume 4, 1992.