In this case study the benefits of performing condition monitoring on a new machine were revealed, as Randy Wallman and Réjean Beaudoin* explain
The expansion Unit 7 at Montana Power’s Thompson Falls station in the US was commissioned in December 1995. This 50MW run-of-river unit is powered by a four-blade Kaplan turbine and equipped with a 76-pole rotor spinning at 94.7rpm with a 13mm (0.512in) nominal air gap. The machine is designed with two guide bearings at the ends, ie one above the generator and one at the turbine.
A ZOOM® machine condition monitoring system (MCMS) from vibrosystm was installed prior to commissioning of the plant to measure a variety of vital dynamic and status parameters. Measured parameters include:
•Eight air gap inputs (four at the stator splits and four equally spaced in between).
•Radial and axial relative shaft vibration at the guide and thrust bearings.
•Twenty-three temperature inputs (bearing water, oil and metal, stator RTDs, turbine seals).
•Gate opening, headwater and tailwater levels.
•MW and MVAR.
Studying the machine through its full range of operating and transient conditions, from standstill to overspeed, allows a comprehensive assessment to be made of the machine condition and behaviour by correlating various dynamic and status parameters through rotor pole reference.
As part of the system implementation, VibroSystM conducted its results interpretation services (RIS). The RIS consists of a series of data interpretation reports from test measurements recorded under different operating conditions. These include static (slow-roll),
SNL, overspeed, field flash, synchronised to grid, various loads, overexcited, underexcited, and full load hot. The RIS reports prepared by independent experts present observations, data analysis results, and specific recommendations.
The first RIS tests were conducted concurrently with the initial commissioning to verify the machine tolerances and behaviour. In September 1997 a second set of tests was conducted to re-evaluate the mechanical behaviour and condition, in order to compare results and enforce warranty terms if necessary. Finally, a third RIS set was conducted in January 1998 to investigate a vibration problem discovered at the generator guide bearing.
Comparing the findings
The first set of tests was conducted between October 1995 and January 1996. It concluded that the unit maintained a good mechanical stability in all operating conditions, including overspeed.
The static measurements taken under slow-roll conditions matched the tolerances specified at the initial erection of this unit. However, air gap measurements at full load slightly exceeded erection tolerances for a new machine. The rotor roundness and concentricity slightly exceeded erection tolerances. However, as the load increases the rotor roundness improves, so it was not considered a problem.
The angle and the range of the vibration at the generator guide bearing revealed a slight magnetic shifting caused by the rotor eccentricity. During the field flash test, the peak–peak vibration levels at the generator guide bearing increased, where it should normally decrease. This situation was noted and addressed in the follow-up tests.
In September 1997 a new series of RIS tests was conducted. Although the machine was considered reliable, there were some changes in the form of a significant deterioration of the stator roundness, a slight increase of the rotor roundness, and an improvement in both rotor and stator concentricities, as well as high shaft vibration at the generator guide bearing.
These tests have shed light on a premature aging of the stator geometry. A polar view of the generator shows that it has adopted an elliptic shape at full load. Its roundness has deteriorated from 6.9% to 11.6% of nominal air gap in less than two years. This rate is considered abnormal and exceeds the expected tolerances for generators with over 15 years of operation.
It is suspected that this premature ageing originates from design or assembly problems with the radial key system designed to facilitate thermal expansion of the stator. In the long term, an increase in the stator oval shape could bring about the distortion of the rotor rim due to cyclic strain at each revolution. While the machine was still under warranty, it was recommended to initiate discussions with the generator manufacturer to study the problem. It was also recommended to add stator frame displacement sensors radially with air gap sensors to study the anomaly further.
The study of shaft vibration showed that the relative vibration increased with load. Vibration levels at the generator guide bearing were still high at full load, reaching almost 0.254µm (10mils). Axial vibration exceeded 75% of radial vibration at the guide bearing, which is also considered abnormal.
A final set of tests was conducted in January 1998. The goal was to resolve the high vibration problem at the generator guide bearing. For these new tests, the ZOOM system gathered direct DC displacement versus time data. This mode allowed for analysis of the direction of vibration correlated with rotor shape and centre offset.
Diagnosing the problem
The diagram on p50 shows vibration behaviour versus various operating conditions for both the generator and turbine guide bearings, revealing that peak to peak vibration increased exponentially with the rotational speed of the machine and stabilised upon field application. The spectral analysis in the diagram indicates a predominant frequency which matched the unit rotational frequency, while vibration at 2x and 4x harmonics (turbine frequency) were low. Both analyses indicated a mechanical imbalance at the generator level.
The study of shaft behaviour versus operating conditions in relation to Pole 70 (pole corresponding with rotor geometric offset, or eccentricity) showed that the rotor eccentricity created a magnetic pull sufficient to slightly modify the shaft rotation axis, producing a noticeable mechanical imbalance which induced excessive vibration.
High vibration levels
The analysis then looked at the rotor shape stability over various operating conditions, since any noticeable modification in the roundness would also affect the eccentricity, and thus the vibration behaviour. The rotor shape varied very little once under load and the offset angle maintained itself. Therefore at the moment, the rotor shape is not responsible for the higher vibration levels as the load increases. However, the rotor roundness should be monitored continuously, and trended to detect any change which could result in an increase in eccentricity.
The analysis of the vibratory behaviour has confirmed that mechanical unbalance due to the rotor eccentricity is primarily responsible for the high vibration of the generator shaft at full load. The design of this generator demands a very precise assembly to eliminate geometrical errors. Since the rotor eccentricity remains relatively small under various operating conditions, it was determined that balancing the rotor should be sufficient to correct the vibration problem.
Based on the ZOOM software data, recommendations were made to install a trial weight of 9kg (20lb) in the opposite angle of the offset (135°), then to repeat the measurement at SNL to determine the final weight and angle needed to reduce the vibration to 75µm (3mils) or less.
At the conclusion of the tests, Montana Power was undergoing the sale of its generating assets. The remedial actions as a result are still under consideration and implementation will be scheduled in the near future.
This case demonstrates the importance of conducting comprehensive tests and analysis upon commissioning of a new or refurbished machine, and of repeating the process on a regular basis to establish valuable trends. It also shows the significance of having an overall machine monitoring system coupled with powerful analysis software, making comprehensive analysis and diagnosis possible.