Correlation analysis is frequently employed to determine condition monitoring parameters, particularly for complex industrial systems like electrode furnaces where direct cause-effect parameters are not fully known.

Correlation analysis methods were recently employed by Asintech to investigate the behaviour of Søderberg electrodes in an electric arc furnace. Effective implementation of correlation analysis techniques allow industry to enhance operational efficiency and pre-emptively address potential failures.

The Problem: Predicting “Soft Breaks” Caused by “Hard Breaks”

One of the primary challenges addressed by correlation analysis is predicting soft breaks in Søderberg electrodes, which are often precipitated by hard breaks.

A soft break occurs when the baked zone of the electrode detaches from the contact shoes, leading to the loss of liquid electrode material. This can be caused by a hard break, where the solid tip of the electrode fractures due to thermal stress, impact loading, or inconsistent consumption.

Identifying the root causes and correlation between operational response of the furnace and breaks is crucial to maintain operational stability and preventing downtime.

The Importance of Correlation Analysis in Electrode Furnaces

Electrode furnaces, such as those employing Søderberg electrodes, operate under highly dynamic conditions with numerous interacting sub-systems.

Understanding the correlations between these systems and the measurable operational are challenging, due to the weak coupling between the electrode conditions and the actual vibrational of the furnace structure.

Statistical correlation analysis offers a viable technique and has been implemented successfully to identify relationships between relevant variables, such as vibration response and operating conditions. Given a large enough and representative operational dataset that includes failure related events, initial correlations may be refined end up with robust condition monitoring parameters which can employed to pre-empt future failures.

Steps in Correlation Analysis

• Data Acquisition:
  • The first step involves rigorous data collection on the entire furnace system with a range of transducers including accelerometers, microphones, thermocouples etc.
  • For the Søderberg electrode study, a suite of tri-axial accelerometers was placed on three electrode columns to capture the primary structural response due to the arcing process, while a microphone was positioned centrally to correlate noise and vibration data.
• Data Synchronization:
  • Synchronization of the collected data with the operational parameters of the furnace (current, tap position, electrode position) is crucial.
  • Event-based time synchronization was employed to align the vibration data with the operational conditions.
• Time-Spectral Analysis:
  • This step involves analysis the frequency content of the vibration signals over time by means of waterfall spectra or spectrograms.
  • Scrutiny of the highlighted the structural response “signature” while allowing the identification of important operational frequencies (such as the 50 Hz line frequency) as well as frequency content that are dependent on the stiffness of the electrode and the interaction of the electrode with the support structure.
• Correlation Tables:
  • The final step is to calculate the correlation coefficients between different vibration and noise parameters (e.g., RMS, Peak-to-Peak, Kurtosis, Crest Factor) and the operational parameters.
  • This allows identification of strong correlations that can be used for condition monitoring.

Key Findings from the Correlation Analysis

Positive Correlations: Significant positive correlations were found between RMS and Peak-to-Peak (P2P) vibration values and various operating conditions, such as electrode position and apparent power.

This indicates that as the operational parameters change, the vibration levels also vary correspondingly. Kurtosis and Crest Factor parameters exhibited variance that could potentially correlate with specific operating conditions like power fluctuations.

Practical Applications and Future Work

The findings from the correlation analysis of the Søderberg electrodes can be applied in several ways to enhance operational efficiency and prevent failures:

• Real-Time Monitoring: Implementing real-time monitoring systems that track vibration and operational data allow industries to detect anomalies early and take corrective actions before major failures occur.
• Predictive Maintenance: Understanding the correlations between operational parameters and vibration responses allows for better predictive maintenance schedules, reducing downtime and maintenance costs.
• Enhanced Design: The insights gained can inform the design of more robust electrode systems and control strategies, improving the overall efficiency and lifespan of the furnace.

Conclusion

Correlation analysis is a powerful tool in the realm of industrial condition monitoring.

The study of Søderberg electrodes in an electric arc furnace highlights its potential to significantly enhance operational efficiency and prevent failures.

By continuously refining these techniques and expanding the dataset, industries can achieve more reliable and efficient operations, ensuring long-term structural integrity and performance.