Gluzman - Turbine Bearing Wiped on Coastdown

8/12/2019 Gluzman - Turbine Bearing Wiped on Coastdown

1/18

Page 1 of 18

Turbine Bearing Wiped on Coastdown: Cause AnalysisVibration Institute Annual Technical Symposium 2013

By David Gluzman

Problem Description

During coastdown metal temperature of the short elliptical bearing supporting LP section of a large

steam turbine spiked from 122 deg F to 200 deg F at speed of 235 RPM. Upon disassembly the bearing

was found wiped.

Journal roughness was evaluated as being marginal and decision has been made to hone the journal in

place. Journal roundness and taper were found to be within tolerance. Original journal diameter is

20.000 inch, after honing - 19.9964 inch. A few major grooves were left in the journal in order to avoid

significant diameter reduction.

Refuted/confirmed causal factors

Bearing wipe was suspected before the turbine dissassembly based on bearing metal temperature

pattern captured during the trip [1]. This pattern is a reliable symptom of a wipe and has been well

documented in the literature. During the investigation several possible failure causal factors were

considered.

Improper Lubricating Oil temperature during coastdown (refuted).Typically, simultaneously with a turbine trip maximum cooling water flow is introduced through the oilcooler so that oil temperature 115 deg F at full speed is lowered to 90 deg F or less when the turbine

comes to turning gear operation. This condition has been met [1] and therefore the causal factor is

refuted.

Excessive operating bearing metal temperature (refuted).Reference [1] also indicates that during operation bearing metal temperature was steady at 197 deg F,

which is within the design specifications of 190-210 deg F for short elliptical bearings. This causal factor

is refuted.

Low Lubrication Oil header pressure (refuted).Turbine bearing header pressure dropped somewhat, however never reached the value at which the

emergency bearing oil pump would start. The pressure leveled out based on the design of the turning

gear oil pump, which again is typical. This causal factor was refuted.


2/18

Page 2 of 18

Improper size or plugged orifice (refuted).Both, condition and the size of the orifice have been examined and met the design requirements. This

causal factor was refuted.

Inadequate Babbitt bond (refuted).The upper half indicated normal bonding based on ultrasound test results. The lower half could not be

tested. However, based on the shaft CL position trend indicating movement toward the bottom at each

start/stop [15] this condition is refuted.

Excessive journal out-of-roundness and taper (refuted).Both parameters were examined and found to be conforming to the design specifications. This causal

factor is refuted.

Excessive bearing static loading (refuted).Although additional loading does not necessarily play a negative role at full speed as far as metal-to-

metal contact is concerned, it is a major consideration at low speed when oil wedge has lower capability

of supporting the journal due to boundary lubrication mode. Bearing static loading was analyzed based

on the orbit shape. Typically excessive loading condition is indicated by a flat orbit and relatively low

journal CL position. The vibration spectrum with excessive loading typically features relatively large

second harmonic. Higher than normal bearing metal temperature is also an indicator of excessive

preload.

In this case the orbit is indicating a lightly flattened pattern [11]. The CL plot [10] indicates that the

journal lifted up 6 mils from the bottom of the bearing at full speed, which is normal. Vibration spectra

[14] indicate relatively small second harmonic. As mentioned above, the bearing metal temperature was

197 deg F, which is also normal. Excessive loading condition is refuted.

Journal surface excessive roughness/waviness (confirmed as Cause #1)Journal surface finish condition and its effect on wiping is of great importance. This issue has been

addressed in many technical papers and prior to confirming, it will be discussed in a great detail.

The OEM addresses the issue by stating that regardless of oil condition, low speed operation can result

in bearing babbitt wipes, particularly if the journal is scored. Each time the unit is started or stopped the

condition will progressively become worse. When the turbine runs below rated speed, temperature

excursions with a characteristic spike may occur if a scored journal is present. Temperature spikes may

not be pronounced in all cases since wiping can be only momentary. The mating bearing to that journal

will be wiped also. The bottom line is that journal damage in form of scratches and grooves is caused by

dirty oil. Journals with wavy surfaces also result from particulate matter in the oil, which in time wears

away shaft material.


3/18

Page 3 of 18

When a journal becomes scored, the oil film pressure profile across the length of the bearing becomes

segmented. Consequentially, the journal rides closer to the babbitt surface. This is not necessarily a

problem at rated speed, however at coastdown or startup the oil film thickness is reduced in proportion

to the speed. As the film thickness decreases, a transition from hydrodynamic to boundary lubrication

takes place. During this transition the oil thickness becomes smaller and when already reduced by the

scored journal, the film may not provide sufficient support.

The result is metal-to-metal contact andwiping of the bearing.

D. R. Gardner [16] and F.A. Martin [17]provideinformation oncorrelation between surface finish, peak-

to-valley surface finish, and minimum amount of oil thickness needed to prevent metal-to-metal contact

in a plain journal bearing. Data indicates that this contact occurs at predicted film thickness, which, in

turn, depends on surface finish and bearing size. A realistic failure value of the film thickness is given

by the peak-to-value surface finish Rmax of the journal as shown in Figure 1. In order to increase the

operating oil film thickness the bearing size has to be increased or a thicker lubricant used for given

conditions of load and speed. Another way to interpret this data is such that when actual roughness and

waviness increases, the existing load may exceed the value of no metal-to-metal contact operation.

Since peak-to-valley parameter is thought of as a failure value, a multiplication factor is applied to

obtain minimum allowable oil film thickness. For a 20 inch diameter journal, minimum allowable oil film

thickness is equal to 0.000,900, which is close to 0.001,000 oil wedge thickness calculated at 250 RPM.

Figure 1. Guidelines on shaft surface finish and allowable oil film thickness for a journal bearing

Surface finish

guidance

Peak to valleysurface finish

Rmax

Min allowable oil

thickness


4/18

Page 4 of 18

EPRI [18] provides instructions on performing inspection of the shaft journals for signs of abnormal

wear, such as scratches, nicks, etc.

Typical recommended tolerances and surface finish for a new journal are as follows:

Taper - 0.001 inch Out of roundness - 0.001inch

Surface finish - 0.000,016 inch

Re-machining required if the following are exceeded:

Taper - 0.0025 inch

Out of roundness - 0.0015inch

More than one circumferential score greater than 0.015 inch deep per linear inch.

( to prevent bearing from overloading at full speed).

Variation of the friction coefficient in a journal bearing as a function of Bearing Parameter [V/p], as

shown in Figure 2, where: - oil viscosity, cp; Vshaft speed, rpm; punit pressure, psi ). Note that

hydrodynamic friction slowly decreases at decreasing speed, but then increases sharply with onset of

mixed-film oil lubrication. The plot shows that dirt, deformations, misalignments, surface roughness,and reduced supply oil flow rate - all have deleterious effects on the resulting coefficient of friction

(CCW arrow). When such condition exists, the onset of metal-to-metal contact occurs at higher values of

the Bearing Parameter.Bearing wearing-in, as opposed to wearing-out, has a beneficial effect friction

(CCW arrow) since hydrodynamic full film operation is maintained to lower values of the Bearing

Parameter.

Due to the steep pattern of the curve in the mixed lubrication zone, a small change in Bearing Parameter

can result in a significant change in the Coefficient of Friction and therefore generate additional heat.

Note that in the region of boundary lubrication operation is unstable. If the temperature has increased,

viscosity will be reduced and coefficient of friction will increase causing more heat and further viscosity

reduction.

Using coefficient of friction characteristics, a curve can be constructed of bearing temperature as a

function of time during coastdown. With onset of contact starting at certain bearing parameter, the

temperature curve can be calculated (Figure 3). The form of this curve is quite similar to the field

observations in the journals that have been scored. Solid particle contaminants in the lubricant and/or

deformations and misalignments of the bearing members would also result in similar contact induced

transient temperature rise of the bearing during coastdown.


5/18

Page 5 of 18

Figure 2. Coefficient of friction curve in a journal bearing.

Wearing in

Boundary friction

Mixed

Dirt,

Misalignment,

Roughness,

Reduced oil supply

Fluid

Brg parameter, [V/P]

Operating

s eed

Friction

coeff

Figure 3. Variation of speed, temperature, and bearing wear as function

of time in a bearing experiencing metal-to-metal contact at coastdown.


6/18


7/18

Page 7 of 18

In the mixed film lubrication regime there is partial shaft-bearing contact. Inserted are thin layers of

lubricant in the remaining areas, which if pressurized add to load support ( Figure 5, a). In boundary

regime the lubricant does not provide any support ( Figure 5, b).

Figure 5. Bearing support at mixed-film and boundary lubrication regimes. Oil pressurized areas

provide additional support.

If the shaft is tapered, the stress in one bearing load zone area will be higher than in the other. In this

particular case journal dimension measurement did not indicate presence of taper and the babbitt

wiping pattern was not representative [6].

In regards to the effect of journal surface condition on wiping reference [20] concludes that

hydrodynamically lubricated bearings can operate satisfactorily with a certain amount of scoring,

however, a stage is eventually reached when it is no longer possible.

EPRI [21] is also stating that one mode of metal-to-metal contact requires simultaneous occurrence of

both, a scored journal (which in turn occurs due to hard particles contamination within the bearing oil

film or partially embedded into Babbitt surface) AND the rotor being at startup or coastdown.

P. R. Trumpler [25]defines minimum oil thickness as:

h = 0.000,200 + 0.000,040*D

For a bearing diameter D=20 inch, the minimum oil thickness is h = 0.001 inch. This is close to the wedge

thickness calculated at 235 RPM.

M.F. Spotts [26]offers the following relationship linking surface RMS roughness Rq and minimum oil

thickness hat which hydrodynamic lubrication still exists (Babbit roughness will be conservatively

assumed to be negligible).

h= 4.5*Rq

From the above, at oil film thickness of h=0.001 inch, Rq=0.000,220 inch ( 220 in), which is 13 times

higher than typical OEM journal surface finish specification of 16 in. The220 in value could be

established as journal surface roughness tolerance. Turbine manufacturers typically do not provide a

tolerance value other than one grove of certain depth per inch, which is relates only to full speed.

shaft

W

bearing

shaft

W

bearing

a b


8/18

Page 8 of 18

Cause #1 Summary:

The technical sources listed above consider journal surface roughness/waviness condition to be a critical

parameter. If excessive, it will cause a bearing wipe at low speed. The bearing in question failed

previously by being wiped during coastdown. At that time the babbitt was re-poured but the journal,

although visually possessing marginal roughness, has not been re-machined. From the previous to thecurrent wipe event there were multiple startups/coasdowns. During this period of time the journal

centerline (CL) in the bearing has gradually shifted downward by about 7 mils [15]. This number provides

the amount of wear the babbitt sustained when mixed lubrication regime was in effect. This wear has

also altered the shape and clearance of the bearinga critical parameters for creating an oil wedge

thus, negatively affecting oils ability to support thejournal, and also increased the potential of fluid

induced instability.

An accurate assessment of journal surface roughness from the photo in [7] provides an approximate

range of the grooves size being up to 0.003 inch deep. A more accurate estimate of roughness was done

based on journal diameter before and after honing when most of the scoring has been removed [8]. The

radial difference is 0.001,800 inch. It represents peak-to-valley dimension of the surface roughness

existed on the surface and is called Rmax. The RMS value ( called Rq) could be roughly estimated to be

0.001,800/1.4=0.001,200 inch or 1,200 in(for simplicity purposes a sine wave type of roughness profile

is assumed). This roughness exceeded the proposed above tolerance of 220inby factor of six.

Considering the above reasoning, it could be concluded that journal surface roughness in the journal

was excessive and therefore constitutes a confirmed cause of the wipe.

Cause #2: Large size hard particulates were being continuously generated or ingressed into

the lube oil system.

The next logical question which needs to be addressed is condition of the bearing lube oil system from

the cleanliness point of view. Oil was sampled and analyzed in detail on a monthly basis. Oil was alsosampled on a weekly basis or prior to each start and checked for cleanliness and water content. Data is

presented in [2]. The recommended contamination levels are as follows.

ISO 4406:99 16/13/10 as a normal and ISO 4406:99 20/18/15 as an alarm level.

Lab data confirmed that the turbine oil met cleanliness requirements. However manufacturers typically

warn that if great portion of particles appear in 50-250 micron size for particular substance such as

metallic, black oxide, fly ash, asbestos, their content has to be reduced.

Maintenance flush performed after the bearing wipe event revealed the elemental content of the

particles collected on the Millipore filter after several hours of flushing [3].

The size of the hard particles, such as Iron and Silicon, are in the 200-400 m (0.008-0.016 inch) range.

Although up to 0.010 inch maximum size particles are typically allowed to be collected on Millipore filter

during a high velocity/high-low temperature flush, no safe limits are known for a flush replicating

operating conditions of oil flow.


9/18

Page 9 of 18

The iron and silicon particles of 0.008-0.016 inch size circulating through the journal bearings couldve

easily machined groves into the journal while passing with the oil flow not only during coastdown but

also at full speed when oil wedge minimum thickness is 0.004-0.007 inch. OEM does specify the stop-

flushing flagif high velocity/high-low temperature flushing is used. However, flushing in this particular

case resembled normal operation oil flow and therefore no large particles should have been in existence

if oil system was truly clean. Even a small concentration of the detected sizes will eventually score thejournal after many hours of operation.

This condition invokes another question: Why sampling data from the lubricating oil tank has always

indicated a proper level of contamination in the 4, 6, and 14 m range, but never - presence of large size

particles?The sampling location is correctat the suction zone of the kidney loop pump - where

sufficient turbulence is present.

The answer is unknown at this time. It is feasible that these particles are being continuously generated

downstream of the sample collection location and therefore missed. The available contamination data

appears to be not representative.

Admitting that one of causes of excessive journal surface roughness is presence of large size particulatesin the oil, it is important to reiterate that physics behind the babbitt wipe phenomenon outlined above

suggests that a wipe can occur even with perfectly clean oil if excessively rough journal surface is already

present.

Oil cleanliness data isn't representative of the oil actually reaching a bearing (Cause #3)

As stated above, oil cleanliness data collected on a regular basis in the oil lube tank did not indicate a

problem, thus creating a false sense of security.

No filters preventing contaminants from entering the bearings are present (Cause #4)

Although a kidney loop filtration system is being used to continuously circulate the oil in the tank

through the filter, there are some lube piping sections that are left out from the cleanliness perspective.

Namely, the oil piping downstream of the tank, where particulates could be generated and the dust

sucked into the tank and not completely filtered out can travel further and ultimately reach a bearing.

Multiple startups/coastdowns (Contributing factor)

During each startup/coastdown, a bearing undergoes a mixed film and boundary lubrication modea

phenomenon which occurs at low speed. The slope of the curve in the referenced above Coefficient of

Friction vs. Bearing Parameter plot is very steep. Therefore, if the contact surfaces are rough, chances of

wiping are higher during this mode even with a perfectly clean oil.If the journal and the babbitt surface

are in pristine condition, the short transition during startup/shutdown will cause minimal damage to the

babbitt surface. However, more severe roughness in journal surface increases chances of a wipe. Time

wise, wiping is accelerated by multiplicity of transitions.

By itself a large number of transitions are unlikely to cause wiping if surface roughness is low; however

under less than perfect conditions each transition accelerates chances of a wipe.


10/18

Page 10 of 18

References

1. Bearing metal temperature spiked at 235 RPM on coastdown.

2. Oil samples analyzed by an Oil Lab met recommended ISO cleanliness levels

RPM

Brg metal temp

LO cooler outlet temp

Brg drain LO temp


11/18

Page 11 of 18

3. Particles washed off of the Millipore filter at lube oil system maintenance flush.

4. Electronic Scanning Microscope (SEM) image of a sample obtained from

Millipore filter during the maintenance oil flush.


12/18

Page 12 of 18

5. EDS/SQ analysis results from Millipore sample. Hard elementsIron and Silicone and their

oxides - make up some particles.

6. The lower half of the bearing is wiped


13/18

Page 13 of 18

7. The journal as found.

8. The journal after honing. Same scale as in reference 7 for comparison.


14/18

Page 14 of 18

9. Before the failure, CL in both bearings across the coupling appear to be in the wrong quadrant

suggesting bearing alignment in horizontal direction.


15/18

Page 15 of 18

10. After repair both across the coupling bearings CL are in the lower part of the annulus with

altitude angle close to zero. Because of that, while probes for the bearing in question are

switched (see below), it did not affect the result.


16/18

Page 16 of 18

11. Before the wipe event, both bearings across the coupling operated at normal preload.

12. After bearing replacement the orbits did not change significantly ( X and Y probes on thebearing in question were accidently switched as indicated by the keyphasor ).


17/18

Page 17 of 18

13. Before and afterspectra are about the same on the across the coupling bearings

14. Before and afterspectra are about the same on the wiped bearing. Second harmonic is small.


18/18

Page 18 of 18

15. Centerline (CL ) data indicates that the bearing in question gets lower with each start/stop.

16. D. R. Gardner, Designing a plain bearing, The Glacier Metal Company Ltd, Alperton,

Wembley, UK

17. Martin, F. A., Minimum allowable oil film thickness in steadily loaded journal bearings,

Proceedings on Lubrication and wear convention, 1964; Vol. 178, Institute of Mechanical

Engineers, London, UK

18. EPRI, Guidelines for Reducing Time and Maintenance Cost of T-G overhauls, Vol.219. EPRI, Guidelines for Maintaining T-G Lubrication Systems.

20. http://www.tribology.co.uk/services/investigate/pb02-0.htm

21. EPRI Turbine Training Manual, Ch. 3, Operational Problems of Sliding Surface Bearings.

22. EPRI, T-G Auxiliary System, Vol. 1

23. EPRI, Training Manual, Ch.4

24. EPRI, CS 4555 Guideline for Maintaining Steam Turbine Lubrication Systems

25. P. R. Trumpler, Design of Film Bearings, NY, 1966, pp. 103-119

26. M.F. Spotts, Design of Machine Elements, 3rd

Ed., p. 321
http://www.tribology.co.uk/services/investigate/pb02-0.htmhttp://www.tribology.co.uk/services/investigate/pb02-0.htm

Gluzman - Turbine Bearing Wiped on Coastdown

Documents