Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Avoidance of false negative rotor false diagnoses in industrial induction motors

Trachalakis Nikolaos

Full record

---

Summary

One of the most common problems that electric machines exhibit, which can be diagnosed before causing serious damage due to their slow progression, are rotor electrical faults. Thanks to their nature, these faults provide diagnostic science with the opportunity to study them and take the necessary actions to avoid unnecessary operational interruptions as well as the catastrophic effects on the machine and other connected systems, thereby avoiding significant maintenance costs.This thesis focuses on squirrel-cage induction motors, which often experience breaks or cracks in the bars and end rings of the rotor. In many cases, traditional diagnostic methods fail to detect such faults until they cause severe impacts on the machine’s functionality. This phenomenon is even more pronounced in cases of non-adjacent broken bars, where the field asymmetries caused by each broken bar cancel each other out, leading to false negative diagnoses even though a fault exists in the machine.One solution provided in the literature for such cases involves measuring current and magnetic flux during transient operation, such as during startup, where the harmonic content caused by the broken bars stands out among the machine’s normal harmonics as the speed changes. However, in many industrial applications, this solution is costly and impractical, as the number of starts for certain machines is directly tied to their lifespan or may result in production loss for a significant period.This thesis proposes a solution to this problem, based on the analysis of zero-sequence magnetic flux during steady-state operation. For the execution of this research, we modeled an induction motor 400V in finite element analysis software (Simcenter Magnet), which simulated both the startup and steady-state operation of the motor in healthy conditions as well as in cases with broken bars. The data from these cases were processed using Matlab software, and the reliability of this diagnostic method was examined in comparison to previous methods, both during startup and steady-state operation of the machine.

Available Files

Services