Solving High Intermittent Bearing Vibration after a Modernization and Upgrade

The Unit 1 upper guide and thrust bearings at 125 MW Tungatinah experienced intermittent high vibration after an upgrade and modernization. The author developed a novel integrated vibration analysis approach to find the root cause and propose a solution.

By Enes Zulovic

The 125 MW Tungatinah Hydro Power Station in central Tasmania has five vertical shaft generating sets commissioned from 1953 to 1956, each comprising a 26.1 MW Francis turbine and a 31.25 MVA synchronous generator. From 2010 to 2014, Units 5, 1 and 2 – in that order – underwent upgrade and modernization.

Unit 1 was returned to service (RTS) in January 2013 and almost immediately experienced unpredictable, intermittent and unacceptably high vibration. A condition monitoring system is installed on this unit, and a portable vibration system was used for commissioning. The increase was recorded at all three bearings but was particularly noticeable at the upper guide and thrust bearing assembly (see Figure 1 on page 20).1 Maximum shaft orbits increased for all three bearings and upper guide bearing housing vibration reached 3.8 mm/s (unacceptably high).

This view of Unit 1 from the turbine floor, looking up, shows the two heavy steel I-beams that replaced the removed concrete reinforced beams.
This view of Unit 1 from the turbine floor, looking up, shows the two heavy steel I-beams that replaced the removed concrete reinforced beams.

The unit was inspected in March 2013, initially focusing on the upper guide, lower guide and turbine guide bearings. Clearance on all was acceptable, with the upper guide bearing showing no damage and the oil thrower ring on the lower guide bearing found to have contacted the stationary oil pot. Testing over three days showed improvement in all vibration levels and the machine was RTS.1

Five days later, a significant increase in bearing orbits and housing vibration was recorded. The vibration trends captured show significant variation in amplitude for each machine start despite operating at consistent loads.

Novel integrated approach for vibration analysis

This article proposes a systematic, analytical approach to resolving vibration problems. Basic elements of the approach are:

  • System of logic
  • Integrated approach
  • Definition of vibration as a ratio of the dynamic forces acting on a machine to its support stiffnesses.2

System of logic

Ask: “Why did this turbine fail and not the nearly identical turbine next to it?”3

Example: Unit 1 experienced high intermittent vibration but not Unit 5.

Idea: Compare the failure case and a case as similar as possible that did not fail.

The Unit 5 concrete beams at the generator level were cut to allow for spiral casing assembly and replaced with a heavy steel welded structure that restored the foundation stiffness.

Integrated approach

Vibration analyses should include a review of the following:

Shaft alignment: Evaluate against major acceptance criteria – horizontal level <0.02 mm/m, shaft static runout at the journal of the turbine guide bearing as a percentage of bearing diametral clearance <35%,4 generator air gap stator eccentricity 3% and rotor eccentricity 1.5%.5

Balancing: ISO 1940-1 gives recommendations for maximum allowed unbalance, determined by rotating mass, rotational speed and balancing grade. Balance quality grades (in mm/sec) are designated according to the magnitude of eper, the distance of the mass center from the shaft axis. But it makes more sense to define balance quality grade in terms of eccentricity. We recommend precision dynamic runner workshop balancing and site rotor two-plane precision balancing be performed to a balance quality grade <Q1.6, or calculated eccentricity be 70% of bearing radial clearance.

Machining tolerances: Standard IEEE 810 to be applied for the shaft.3

Vibrations: According to some hydro industry experience and proven practice, ISO 7919-5 and ISO 10816-5 are not sufficiently defined as vibration standards.

Stiffness: Vibration is a ratio of the dynamic forces acting on the machine to its stiffnesses. Viewed this way, the focus becomes, “What has changed in the machine, the forces acting on it or its stiffnesses?”2 Vibration is a result of other root causes occurring. To prove the relationship between vibration and stiffness, the following actions were conducted:

  1. Concrete foundation vibration testing using seismic accelerometers: The foundation responded with vibration amplitudes of 30%, exceeding the recommend values of 25% of the bearing housing vibrations. Unit 4 is very sensitive to machine load and the concrete foundation vibration increased up to 2.5 mm/s root mean square (rms) when load exceeded 24 MW;
  2. Turbine shaft bump test for natural frequency: Shaft natural frequency is less than calculated;
  3. Stator frame bump test for natural frequency: Result not repeatable;
  4. Trial by compactor of excitation of the concrete foundation to determine its natural frequency: It was not possible to excite the concrete foundation at the generator level with the small compactor when speed control was not variable;
  5. Exciter shaft balancing: Exciter shaft resonant frequency is about 36 Hz, but due to the low stiffness of the exciter support, it was subject to high vibrations in dynamic operating conditions;
  6. Adding two metal stiffeners to the concrete foundations: This increased the foundation stiffness, reduced overall machine vibration and moved the shaft resonance far from the operating speed;
  7. Adding eight radial stiffeners to the thrust bracket: Vibration testing provided confidence that the additional radial stiffeners were providing a stabilizing effect on the thrust bracket assembly; and
  8. Trim rotor balancing: This provides additional margin on the shaft critical speed by reducing dynamic unbalanced forces.

Shaft critical speed

In traditional design processes, the only needed check is calculating the shaft first natural frequency and assuring this is at least 1.25 higher than the highest runaway rotational speed. This frequency was calculated under the assumption that the bearings are a rigid support model, so that the calculated natural frequency was always much larger than the frequency experienced in operation.

To calculate the shaft system natural frequencies and machine steady state response, the computational model for a hydro rotor should combine a finite element model (FEM) with varying guide bearing stiffnesses, foundation stiffnesses, the gyroscopic effect, unbalanced magnetic pull (generator air gap), and hydraulic and mechanical forces (shaft misalignment and runner and rotor unbalance).

The turbine generator shaft rotating system natural frequency of lateral vibrations varies with bearing stiffness and can be significantly less than calculated if this stiffness is insufficient.

Detailed investigation

Hydro Tasmania technical personnel conducted root cause analyses from April to June 2013 to help identify the cause of the unusual machine condition. Tests of changing conditions, loads and sequences on Unit 1 and the other machines in service were performed. Results showed the vibration was not speed (mechanical balance) related but that increased vibrations of the stator frame and stator core when excitation (electrical force) was applied were evident.

Comprehensive inspection and partial strip down of Unit 1 was performed, and a thorough investigation of the surrounding structure.

When the unit operates at its resonant frequency, it responds more strongly, which is observed as higher vibration levels. If the unit is operating close to its resonant frequency, a small change in force or perturbation may cause the resonance to be excited. Analyzing a polar plot of the shaft displacement after a Unit 1 overspeed test suggests a resonant frequency exists close to synchronous speed. Resonance is indicated by a 180 degree phase shift in vibration and is displayed in polar coordinates as a circle formed by changing machine speed, with actual resonant frequency being the furthest point on the circle.

Evidence was seen of a resonant frequency at the upper guide bearing around 608 rpm, very close to the operating frequency of 600 rpm. With a resonant frequency so close to the synchronous speed, the sudden step changes in vibration may be explained by only a small perturbation to the machine causing the resonance to be excited. Likewise, a similar perturbation could de-excite the resonance and cause a step down in vibration.

Figure 1 shows that once the unit is steady at 0 MW, it oscillates between a resonant and non-resonant condition, evident by the dramatic variation in vibration levels at the upper guide bearing.

Foundation structure

During work on Unit 1, sections of flooring concrete were removed to install a new spiral casing down to the turbine floor. Enlargement of the existing openings included removing a concrete floor beam and slab in the operating and generator floor. ‘As left’ alignment was good, with a shaft plumb of 0.005 mm/m and the runner centralized in the wearing rings. However, ‘as found’ shaft plumb was 0.029 mm/m (acceptance <0.020 mm/m) and the runner was about 0.500 mm offset in the seals.

Station foundation vibrations

To understand the relationship between vibration and foundation stiffness, tests were conducted using seismic accelerometers from PCB Piezotronics installed at the generator level on all five machines.6 Vibration analysis revealed the power station structure is susceptible to vibrations produced particularly by Unit 4. With Unit 1 shut down, vibration up to 0.4 mm/s was recorded outside the Unit 1 alternator enclosure (see Figure 2 on page 22).

High sensitivity accelerometers from PCB Piezotronics were installed outside the Unit 1 stator. Further analysis conducted during RTS testing revealed that as Unit 4 load fluctuates above 24 MW, it would occasionally initiate structural vibration up to 1.4 mm/s in the station floors. Similarly, high vibration was observed in the Unit 1 exciter housing. The low stiffness of the exciter housing combined with its shape as a cantilever beam extending from the operating floor caused it to vibrate up to 12 mm/s rms on occasion as the foundation structural vibration initiated from Unit 4 “tuned in” to the exciter housing at 36 Hz.

Stator design

The stator is assembled in two halves. The key bars are welded into cut-outs in the frame and shimmed dovetail strips are bolted to the inner surface of the bars for stacking of the lamination plates. The stator frame/core consists of 32 key bars welded to the frame diametrically around the inner bore. A dovetail key is spaced from the key bars using small sections of plate and fastened by 11 countersunk 0.5-inch screws.

Based on stator core and frame vibrations increasing when excitation is applied, it appeared Unit 1 was experiencing a loss of core to frame connection. Specifically, the loss of the dovetail to keybar connection was identified as the likely possible root cause due to inherent flaws in the dovetail design arrangement using screws.1,7

The core assembly shows the presence of a split core design, which is inherent to occurrence of vibrations of 100 Hz. In the split design, the joints between the stator core sections should be tight to reduce the core vibrations amplitude due to the pole passing magnetic forces. If the joints become loose, high amplitude core vibrations at 100 Hz frequency could develop.7,8,9

Stator frame shims

The Unit 1 stator frame was shimmed using 10 mm-thick plates to correct axial alignment of the rotor and stator, which was found to be incorrect during the upgrade and modernization alignment. This contributed to a decrease in stiffness of the stator supporting assembly.

Other issues

Other minor abnormalities discovered and tests performed during the investigation are:

  • Stator frame weld cracks. During upgrade and modernization of Unit 5, non-destructive testing (NDT) of the stator frame revealed many weld cracks. Inspections of the other four units revealed similar cracking. The investigation revealed the cracks were welding defects from manufacture of the frames and there is a low risk of failure even with multiple failed joints.
  • Alternator concrete foundation. Samples of the concrete inside the alternator enclosure revealed it to be of poor quality and soaked with oil after years of leaks from the lower guide bearing.
  • Thrust bracket bolts. The thrust bracket hold-down bolts had been over-torqued during installation.
  • Bump tests. A bump test of the shaft was conducted with the unit free on the thrust bearing. The shaft at the turbine pit recorded a distinct natural frequency of 3.9 Hz.

Discussion

Combining the vibration definition (dynamic force/stiffness) with results of the detailed analysis and inspections, the following was discussed to identify the root cause of the Unit 1 high intermittent vibration and provide a corrective action plan.

  • Unit experiences a resonant condition at the upper guide bearing support assembly with the natural frequency too close to the operating frequency;
  • Removing the concrete beams reduced unit stiffness and shifted the natural frequency at the upper guide bearing support assembly closer to the operating frequency;
  • As the upgrade and modernization included the installation and concreting of a new metal spiral case, the changed shaft lean and shift in position is most likely due to settling of concrete components. This corresponding change in position at the turbine increased the perturbation on the machine, helping excite the resonance during operation;
  • Station concrete structure is not rigid and is noticeably affected by the operation of Unit 4 when above 24 MW;
  • There is no simple method for determining the vibration root cause on hydro generating units. The integrated approach described in this article provided distinctly better results;
  • Results presented here indicate the value of site testing for quantifying vibration, natural frequencies and stiffness,
  • Additional stiffening of the supporting system reduces vibration below the acceptance by a big margin; and
  • A challenge is to use new technologies, such as the exciter/shaker testing, to obtain the natural frequency of the components, stiffness calculations, and site verification tests by using a vibration diagnostic software such as Veski (stiffness, shaft critical speed).8,10,11

It was decided to: increase the stiffness of the upper guide bearing and shift the resonant frequency away from the operating frequency by re-installing the floor beams and adding stiffening beams, realign the unit to the “settled” position of the turbine components and perform a trim balance of the rotor.

Corrective work

Work was carried out to increase overall stiffness of the unit foundation and at the upper guide bearing support and to reduce the dynamic forces: alignment and rotor unbalance to control response of the unit during operation, by shifting the resonant frequency from rated speed and reducing the subsequent vibrations.

Stiffening metal beams: The concrete floor beams were replaced with steel I-beams. To provide additional stiffness to the upper guide bearing assembly, eight more steel beams were installed at each upper guide bearing bracket leg extending outwards to the alternator enclosure wall attached to the reinforced top concrete beam. The new beams were mounted with only light contact to allow for thermal expansion once the unit is online.

Alignment: The unit was realigned to the settled turbine components and with the runner centralized to the wearing rings. The upper guide bearing was offset to achieve a shaft plumb of 0.011 mm/m. Besides shifting the resonant frequency of the upper guide bearing by improving the stiffness, the improved alignment will help minimize the machine perturbation exciting the resonant frequency.

Balancing: During RTS testing of Unit 1, the balance grade of G1.0 achieved during initial commissioning was again recorded. Further minimizing the unit perturbation, the rotor was successfully trim balanced to a grade of G0.8 according to ISO 1940.12

Exciter shaft: The new exciter shaft installed during the upgrade and modernization was out of balance. The shaft was removed and dynamically balanced in April to June 2013.

Unit 4 restriction: Structural vibration originating from Unit 4 operating above 24 MW was recorded to similar levels during the Unit 1 RTS. To prevent long term damage to the concrete foundation and further minimize any disturbance exciting the resonance on Unit 1, a 23 MW load restriction was placed on Unit 4.

Unit 1 RTS

RTS testing was performed in June 2013. The unit was subject to machine and station conditions previously shown to induce the resonant vibration, as well as other conditions common during normal operation. The unit was subject to multiple starts and stops as well as long term operation, and the previous resonant condition could not be excited.

Overall, the shaft orbits and bearing housing vibration were much lower than during commissioning in January to February 2013 and dramatically lower than the previous high vibration recorded in March 2013.

Further evidence suggests the stiffening beams have shifted the resonant frequency away from the operating frequency (see Figure 3). The machine is again decreasing in speed after an overspeed test, but there is no indication of a resonant frequency close to the operating frequency of 10 Hz (600 rpm) in the polar plot.

Since the Unit 1 RTS, vibration levels have remained well below the previous levels and the unit has continued unrestricted operation.

Following completion of the Unit 2 upgrade and modernization project and installation of new metal support beams, which increased the overall stiffness of the foundation at adjacent Unit 2, vibration levels at Unit 1 unit have been stabilized and reduced.

Recommendations

Experience on this project highlights the importance of the following recommendations:

A hydro-generator rotor model shall combine a FEM with varying guide bearing stiffnesses, foundation stiffness, the gyroscopic effect, unbalanced magnetic pull (generator air gap), hydraulic force and mechanical forces (shaft misalignment and runner and rotor unbalance) to calculate natural frequencies and machine steady state vibration response.

Technical specifications shall include all the factors above and vibration levels determined in conjunction with machining tolerances, runner workshop balance specified as eccentricity (not residual weight), site shaft alignment tolerances, rotor site balancing, sufficient stiffness of the support and concrete foundation to keep the shaft critical speed 15% above overspeed and vibration levels within acceptance.

An integrated approach to vibration analyses should be applied that considers machining tolerances, installation tolerances, runner workshop dynamic balancing, shaft alignment, rotor dynamic balancing, stiffness, stresses, dynamic forces.

Install and monitor vibration equipment on each unit. Vibration information is crucial to timely identification and diagnosis of machine problems. Use vibration software – such as that from Veski – that can produce such data as shaft orbits, shaft deflection, slow roll vector, guide bearing alignment, air gap, calculation of shaft critical speed and stiffness of the supporting structure.

When conducting major work on a unit, consider its impact on unit stiffness, which directly affects vibration.

During the rebuild of a unit, ensure the alignment is within the design specification, and allow time for unit balancing during RTS.

Consider the acceptable levels of vibration within the power station concrete foundation structure. There are no specific hydro machine standards addressing this, but other standards referencing rotating machinery concrete foundation vibration state acceptable vibration is up to 25% to 30% of the bearing housing vibration.

Due to the complexity of vibrations of vertical-shaft units, it is important to periodically present advances in analytical and diagnostic tools and methods and standards and any new knowledge and experience (Unit 1 is a good example), as well as to review prior knowledge and concepts.

Notes

1. Zulovic, Enes, and J. Vandervelde, “The Tungatinah Vibration Station; Multiple Vibration Issues Requiring Unique Solutions,” Proceedings of 16th HPEE, Hobart, Tasmania, Australia, 2014.

2. Sabin, S., “Understanding and Using Dynamic Stiffness – A Tutorial,” Bently Nevada, 2000.

3. Keith, G.G., P. Loustau and M. Melin, “Lloyd’s Register: Failure Analysis of Rotating Equipment using Root Cause Analysis Methods,” Proceedings of POWER-GEN International 2011, PennWell, Tulsa, Okla., USA, 2011.

4. Hydroelectric Turbine-Generator Units Guide for Erection Tolerances and Shaft System Alignment; Part V – Maintenance of Vertical Shaft Units (All Types of Turbines or Pump-Turbines) Limits for Key Parameters, Canadian Electricity Association.

5. Nasselgvist, M., R. Gustavsson, and J.-O. Aidanpaa, “A Methodology for Protective Monitoring of Hydropower Units Based on the Mechanical Properties,” ASME Journal of Dynamic Measurements, March 2011.

6. Machinery Protection Systems, American Petroleum Institute 670, 2003.

7. Vladislavlev, L.A., “Vibration of Hydro Units in Hydroelectric Power Plants,” (translated from Russian) Amering Publishing, New Delhi, India, 1979.

8. Oreskovic, O., “Implementation of Total Machine Condition Monitoring System on Pump Storage Generator in PSPP Avce (Slovenia),” Proceedings of HydroVision International 2013, PennWell Corp., Tulsa, Okla., USA, 2013.

9. Instructions on Measuring and Interpreting the Vibrations on Machines, VDI 3839.

10. Internal documentation prepared during testing of Tungatinah unit number 1, Hydro Tasmania, Hobart, Tasmania, Australia, 2013.

11. Oreskovic, G., Computerised Diagnostic System, CoDis, Veski, Melbourne, Victoria, Australia.

12. API Standard Paragraphs Roto Dynamic Tutorial: Lateral Critical Speeds, Unbalance Response, Stability, Train Torsional, and Rotor Balancing, American Petroleum Institute 684, 2005.

Reference

. Zulovic, Enes, “A Novel Approach Solving High Intermittent Vibrations Following Modernisation and Upgrade Project,” Proceedings of HydroVision International 2015, PennWell Corp., Tulsa, Okla., 2015.

Enes Zulovic, MIEAust, CPEng, is specialist hydro mechanical engineer with Acutel Consulting in Australia.

Novel integrated vibration analysis system

INTVibraâ„¢ starts by stating a definition: Vibration we measure is a ratio between the dynamic forces and stiffness. Key elements ( 1 to 17) of integrated vibration analyses known now under INTVibra are: vibration definition; system of logic; vibration standards; precision dynamic balancing; shaft alignment; shaft critical speed; stiffness; mechanical machining tolerances; precision installation tolerances; natural frequencies; exciting frequencies; rotor dynamics; vibration software plots; people, plant, process, training; failure modes, history, root cause analysis, reliability centered maintenance; technical specification, design review; vibration condition monitoring and diagnostic systems.

Previous articleInternational Guide to Hydro Innovations and New Technology
Next articleViewpoints from the Board Room: International Finance Corporation

No posts to display