Uneven air gap in hydro generators can cause high vibrations, due to magnetic unbalance. To alleviate the vibration levels, the air gap needs to be adjusted, which requires a major forced outage. Depending on the severity of the magnetic unbalance vector, balancing the unit to operate safely until the next scheduled outage is possible. This article presents a case study.
By Bernard F Boueri
The turbine-generator unit being discussed in this article is part of a three-unit powerhouse with a total capacity of 210 MW, owned by Ontario Power Generation in Canada. The units are propeller types rated at a speed of 100 rpm. This unit is 60 years old and has a magnetic unbalance due to uneven expansion of the stator, causing a non-uniform air gap. The uneven expansion of the stator seems to change over time and had been occurring for three years. This change causes an increase in the generator overall shaft relative displacement levels. This increase caused the unit to trip several times, resulting in an unreliable unit.
The unit was not scheduled for a full stator rewind and refurbishment until a few years down the road. Thus, the company’s Machine Dynamics group decided to attempt balancing the unit to make it available and safe for operation until the stator is fully refurbished.
The unit protection system consists of two proximity transducers installed 90degrees apart to measure shaft relative displacement levels on the generator and turbine guide bearings. Figure 1 shows the proximity transducers on the turbine guide bearing. A keyphasor®, supplied by Bently Nevada, is installed in-line with one of the proximity transducers to measure speed and phase. A small piece of reflective tape was installed in-line with pole No. 1 to provide one pulse per revolution. All proximity transducers on the unit are Bently Nevada 8 mm (200 mV/mil) probes. All signals were input into a Bently Nevada ADRE 408 data acquisition system.
Table 1 provides a summary of the unit frequencies of interest.
Vibration assessment and balancing
Figures 2 and 3 are the shaft relative displacement levels at the generator and turbine guide bearings, respectively. Unit operation was restricted at speed no load (SNL) field off and all conditions below 70% wicket gate (WG) opening due to excessive shaft relative displacement levels at the generator guide bearing. The unit was operated briefly at SNL field off to assess the mechanical unbalance and obtain the vector response at this condition.
Tables 2 and 3 summarize the shaft relative displacement levels at all tested conditions at the generator and turbine guide bearings, respectively. The data show a change in shaft relative displacement level and a phase shift of about 60degrees in the y direction when the field is applied. This indicates a possible magnetic unbalance due to an uneven air gap.
Figure 4 shows the generator shaft centerline movement at all tested conditions. Once the field is applied, the generator shaft shifts 20 mils. This phenomenon is consistent with an uneven air gap causing a magnetic unbalance once the field is applied on the unit. The data show that the phase is relatively constant once the field is applied. Generator shaft relative displacement levels are above the acceptable limit of 10.2 mils peak-to-peak (p-p) and above the recorded generator diametrical bearing clearance.
Figure 5 is the orbit and time waveform plots at 100% WG opening at the generator and turbine guide bearings. The data show a dominant vibration component at 1x rpm. A possible preload on the generator guide bearing is also noted, most likely due to the large overall shaft relative displacement levels, coupled with the excessive movement of the generator shaft centerline.
Figures 6 and 7 are the spectra at the generator and turbine guide bearings at 100% WG opening, respectively. The spectra confirm the dominant vibration component at 1x rpm. Spectra also show a significant vibration component at 2x rpm. This component is usually associated with alignment/verticality issues in the unit and most likely brought about by the magnetic unbalance. Harmonics are also noted in the spectra. Harmonics are associated with nonlinear behavior of the rotor system, which could be caused by a preloaded bearing.
Because the 1x rpm vibration component is the dominant vibration component with a level of 33 mils p-p at the generator guide bearing, balancing the unit would reduce the 1x rpm vibration component and thus the overall generator shaft relative displacement levels. Based on previous experience with this unit, the sensitivity was calculated to be about 20 lb/mil with a phase lag between the high spot and heavy spot of about 25 degrees. Based on the unbalance vector of 32 mils âˆ 282 degrees once the field is applied, a balance weight of 400 lb was installed at 75 degrees opposite rotation. It was not possible to manufacture one lump mass of the large weight required, so nine smaller weights were installed totaling 400 lb (see Figure 8).
Figures 9 and 10 show the post-balancing shaft relative displacement level trends at the generator and turbine guide bearings, respectively. The overall shaft displacement levels on both the generator and turbine guide bearings improved once weights were installed.
Tables 4 and 5 summarize the overall 1x rpm and 1x phase shaft relative displacement values at the generator and turbine guide bearings, respectively. The SNL field off condition did not improve and that was expected. However, a reduction of about 20 mils in the 1x rpm vibration component is noted once the field is applied. Even though the overall generator shaft relative displacement levels are still above the acceptable limit of 10.2 mils p-p, they are well within the generator guide bearing diametrical clearance.
The final response of the unit at 100% WG opening was 12.3 mils âˆ 347 degrees, a shift of about 65 degrees from the original vector. This indicates that the magnetic unbalance force is quite significant. However, the use of additional weights to try and improve the unbalance response further is not recommended. A larger weight might cause even higher vibrations if the magnetic vector shifts, which could damage the bearings/seals. The author recommended that unit operation be restricted to SNL field on and to 70% WG opening and above.
Figure 11 is the orbit and time waveform plots at 100% WG opening at the generator and turbine guide bearing post balancing.
Figures 12 and 13 show the post-balancing generator and turbine shaft relative spectra at 100% WG, respectively. The generator spectra show a reduction in the 1x rpm amplitude due to the balancing weight. The figure 8 orbit at the generator guide (see Figure 11) is due to the fact that the 1x rpm vibration component and 2x rpm vibration component are at about the same amplitudes. The generator x spectrum also shows that the 2x rpm vibration component amplitude was reduced from about 8 mils p-p pre-balancing to about 4 mils p-p post balancing. This reduction could lead to a reduction of the preload identified at the generator guide bearing pre-balancing (see Figure 5). The turbine spectra harmonics of running speed are most likely due to the scratch/protrusion on the turbine shaft identified in the time waveform of Figure 11.
This article presented a case study illustrating that balancing hydro machines with a magnetic unbalance is possible. However, such balancing should be done on a case-by-case basis. In this particular case, the unbalance vector when the field was applied was relatively constant and a decision was made to balance the unit when it was energized. This would result in significant large shaft relative displacement levels when the field is not applied on the unit. However, the unit only operates without field during run-up and shutdown. It was recommended that the field be applied on the unit as soon as it reaches its rated speed. Clearly, in this case, balancing the unit required a large amount of weight to overcome the magnetic unbalance. The unit was able to operate for three years, until a full rewind was completed to alleviate the stator core uneven expansion.
Note that such balancing should not be considered a permanent solution and the unit should be constantly monitored to make sure that the unbalance vector has not shifted. The author highly recommends that such balancing be done by qualified individuals.
Bernard F. Boueri is a vibration specialist with 25 years of experience in the field of vibrations and rotordynamics and their application in the diagnostics of rotating equipment. He works in the Machine Dynamics Group at Ontario Power Generation and has been responsible for vibration diagnostics across the OPG fleet, including steam turbines, hydro units and all rotating auxiliary equipment. He is a Category IV certified vibration analyst by the Vibration Institute.