Calculation and Analysis of the Strands Short-Circuit in Stator Transposition Bar for Large Generators

Chenguang Wang, Yanping Liang, Lei Ni, Dongmei Wang, Xu BianIEEE Access 2019 · DOI

Abstract

Reliability in large-scale power generation hinges on the integrity of stator bar insulation. This research introduces a high-precision 3D Finite Element Method (FEM) coupled with a circuit network to model strand-level short circuits in the complex transposition (Roebel) bars of large generators. The study identifies a distinct "current jump" in specific strands as a reliable signature for insulation failure. Validated against experimental data from a physical test bench, the model achieves high accuracy, providing a robust framework for diagnosing internal stator bar health and preventing catastrophic failures.

Introduction: The Challenge of Stator Bar Insulation

Large generators are critical components of the modern power grid. To minimize energy losses from eddy currents, their stator bars are not solid conductors but are composed of numerous thin, insulated copper strands woven in a complex 3D pattern known as transposition. While effective for efficiency, this design introduces a vulnerability: the insulation between strands is extremely thin (0.1mm–0.2mm).

Subjected to intense electromagnetic vibration and thermal stress, this insulation can fail, leading to internal short circuits. While a single short might not trigger an immediate shutdown, it induces circulating currents that cause localized overheating and eventual breakdown.

Comparison of (a) a healthy stator bar winding and (b) a winding damaged by a short-circuit fault.

Historically, calculating the exact impact of these faults has been difficult due to the intricate 3D geometry of the twisted strands. Traditional models often struggle to balance computational efficiency with the geometric fidelity required to track individual strand currents.

Modeling Methodology

To address these complexities, the researchers developed a hybrid 3D Piecewise FEM and Field-Circuit Coupling approach.

3D Piecewise FEM

Rather than meshing the entire twisted bar as a single complex object—which would be computationally prohibitive—the team divided the stator bar into multiple piecewise segments along its axis. Within each segment, the strands are treated as straight (non-transposed). This simplification allows for high-resolution local modeling without sacrificing global accuracy.

Field-Circuit Coupling

The actual twisting of the strands is accounted for by linking these piecewise segments through a parallel circuit network. This allows the model to simulate the 3D electromagnetic field while maintaining the electrical continuity of the transposed strands. The electromagnetic field equations are solved using the $A, \phi - A$ method (vector magnetic potential and scalar voltage potential).

Schematic diagram of a stator transposition bar, illustrating the complex geometry of top/bottom layers and the transposition pitch within the slot and end regions.

Key Insights: The "Current Jump" Signature

The most significant finding of this research is the identification of a "current jump" signature. When a short circuit occurs, the affected strands exhibit a sharp increase in current magnitude and a noticeable shift in phase angle. This signature is unique to fault conditions and can be used to distinguish insulation failure from normal operational fluctuations.

Several factors influence the severity of this signature:

  • Radial Position: Shorts occurring in the "top layer" (closer to the air gap) produce larger current jumps than those in the "bottom layer." This is because the slot leakage magnetic field is significantly stronger at the top of the slot.
  • Axial Symmetry: Faults located symmetrically relative to the transposition bend exhibit nearly identical current characteristics, suggesting that the physical path of the strand relative to the magnetic field is the primary driver of the fault current.
  • Transposition Path: The current in a shorted strand is heavily influenced by its path before and after the fault point, as the different leakage fields linked by the parallel branches change the balance of the circuit.

Experimental Validation and Results

The model was validated using a physical laboratory rig featuring a $308^\circ$ stator transposition bar. The rig was equipped with current transformers and a multi-channel A/D converter to measure the current in every single strand.

The results demonstrated exceptional accuracy:

  • Healthy State: The maximum error in strand current magnitude was only 4.83%.
  • Faulted State: Under simulated short-circuit conditions, the model maintained high precision, with a maximum magnitude error of 7.85% and a phase angle error of 5.35%.

These results prove that strand current monitoring is a viable and accurate method for diagnosing the internal health of stator bars.

Limitations and Future Outlook

While the current model provides a powerful diagnostic tool, it operates under certain simplified assumptions, such as ignoring rotor magnetic fields and stator core saturation. In a real-world generator, these factors would introduce additional noise.

Future research will focus on:

  1. Multi-physics Coupling: Integrating the electromagnetic model with a thermal model to predict the exact temperature rise caused by specific short-circuit faults.
  2. Circulating Loss Analysis: Quantifying the impact of faults on total circulating current loss to understand long-term efficiency degradation.
  3. Real-world Noise Reduction: Refining the "current jump" detection algorithms to account for the complex electromagnetic environment of an operating power plant.

By quantifying how specific fault locations change current distribution, this work paves the way for advanced online monitoring systems that can detect insulation degradation long before it leads to catastrophic failure.

Summary updated

BibTeX
@article{wang2019calculation,
  author = {Chenguang Wang and Yanping Liang and Lei Ni and Dongmei Wang and Xu Bian},
  title = {{Calculation and Analysis of the Strands Short-Circuit in Stator Transposition Bar for Large Generators}},
  journal = {IEEE Access},
  year = {2019},
  doi = {10.1109/access.2019.2905362},
}
Figure