High speed balance for steam turbine rotor of centrifugal compressure

Dear BalanceSpecialist,
Can you advise me to decide do properly balance for steam turbine rotor of centrifugal compressor?

This rotor has 1st critical speed = 6,400 RPM, 2nd critical speed =16,000 RPM, operation speed 13,246 RPM.

Last year, the rotor had high vibration after decreased RPM to 2000 RPM and rotor blade were rubbing and shaft bow.

Vendor recommends to repair and bend the shaft and check balance by low speed.

I think it should be balanced by high speed balancing for ensure not problem after 1st speed , what do you think?
Original Post

so, the rotor in operation now is the spare rotor. i can assume the 'damage' rotor is now on its way for refurbishment. rotor straightening is one of the repair scope. depending upon the degree of damage on the blade (i guess it is on the low pressure section blade rows), blade replacement may also be required.

now, if it is now at oem shop, i would expect the vendor recommends for hsb. even for general refurbishment scope (without this rotor straightening scope) they would normally include hsb in their proposal.

nway, imo, the history of 1x vibration performance and the impact of bent & straightening repair method would be the main factors for you to assess if hsb is required. if indeed the rotor has long history of 1X vibration issue, i would go for hsb. if no concern on the 1x history and oem can guarantee the magnitude of the bent had not resulted in rotor lateral change and rotor straightening procedure will not result in any change too, then low speed balancing may be sufficient. the last thing you do not want to be in is to find out your your speed governing range is in critical speed range.

again, if the rotor is at oem shop, i would expect they will include hsb in their scope due to simple reasons; they gain more profit and assurance (they normally provide guarantee period). with your problem description, if oem is recommending low speed balancing only, i have a thought that the rotor is at 3rd party service center.

at 13K speed, it could be in the range of 20-25MW hp (100 barg steam inlet) turbine. some of the turbine in this range do have field balancing provision (weight placement). even with this provision, i will never include this in the decision making. 'oh, we dont need hsb. we can always perform field balancing should problem arose during start-up'.

if the rotor is still in operation and vibration value is normal, it indicates the rotor had been straightened during subsequent (after the problem) slow roll or at its idle speeds. so, nothing to be done.

I always thought that the tolerance is always higher than the probe landing area TIR. This is considering the procedure required to attain such surface finish at probe landing area.

With this turbine speed of 13K, the probe landing area max run-out is 6.3 microns. To find out 5.8 microns as maximum rotor radial run out is just odd. I have never use dial indicator with 1 micron precision. I would expect it would be a challenge to differentiate between surface imperfection and an outcome from deflection.

0.02mm to 0.05mm is general mech runout tolerance that i used to hear/experienced. Even for MHI turbine.

for me, it is not only the maximum, it is the shape that matters most. if you have maximum run-out located at 6 o'clock along the middle span, yes you do have sign of bow. even if 0.05mm at scattered angles along the middle spam, bow is unlikely. with 0.2-0.7mm laby clearances, 0.0058mm run-out tolerance is just too tight imo.
Valve, just to clarify, surface runout and surface finish are two separate and distict parameters and your statement "...probe landing area max run-out is 6.3 microns" (0.25 mil) makes me think you may be confusing the two or not realize the difference. At my former gear employer (high speed gear OEM) we would regularly "super finish" probe target areas with a diamond burnishing tool in an effort to minimize electrical runout. The finishing method yielded a mirror like surface finish; you could actually see your reflection in a large pristine journal. That had nothing to do with runout which was commonly held to 0.0025mm (0.1 mil) on journals up to 250 to 300mm. Such values of runout were easily attained and are absolutely needed in high speed rotors. TING states the shaft is 50mm diameter - at that size 2.5 microns (0.1 mil) runout tolerance is easily attained and should be the "norm" for any shop. Indeed I would say if a shop can't maintain that level of runout I would likely look elsewhere.

You further state "0.02mm to 0.05mm is general mech runout tolerance that i used to hear/experienced. Even for MHI turbine". I sincerely doubt that is the case, especially as applied to a 50mm journal. For those reading this that are familiar with English units, these levels of 0.02 mm (0.79 mil) and 0.05 mm (1.97 mil) are actually far in excess of most acceptance criteria for running vibration! API for instance only allows 25.4 µm direct (1.0 mil PP) or overall vibration at 12000 RPM.

For more on surface finish I suggest you review http://www.engineershandbook.c...surfaceroughness.htm or Google for other sources.

Am fully aware of the difference and that is the reason i explicitly bring up probe landing area TIR tolerance in this discussion. Please provide me the part of API where it provide us the surface runout tolerance accross a rotor. If none, I think you are wrong to use API p-p limit as surface run out 'reference' for a rotor. I have yet to obtain the general rule of thumb but i strongly believe that for the full radial surface runout tolerance of a rotor, it is the seal clearances that is referred as to set/calculate the tolerance.

Btw, I have seen many technical professionals have commented that diamond burnihsing is used to improve electrical run out. I am yet to able to digest it since what i understand, the objective of diamond burnishing is majorly on attaining the mirror like surface finish (MECHANICAL). As for electrical run out, degaussing or demagnetizing is the procedure involved. If this understanding is true on burnishing (mainly mechanical), so, does this mean that the radial surface for the surface runout measurement accross rotor shall be burnished or honned in order to attain 5.8 um tolerance???

I guess those that have strong experience in machienry maintenance and inspection should come forward and comment.


I suggest you have a look on this article. Been trying to find extablished paper that suits this discussion. The turbolab paper is the closest i could get. The operating speed is 8676 rpm.

Excerpted from the article; "except for the instance when the runout at the centre of rotor is performed, 5 mils of runout were measured, which was acceptable for this rotor".

what do you think of this?

The article you site is talking about mid-span rotor bow not allowable tolerances at bearing journals and/or probe target areas. Having said that excessive rotor bow will also likely evidence itself at the journals as excessive journal runout.

The removal of mechanical runout is simply handled by proper tolerancing and the selection of the machine process, i. e., a lathe can't do as good a job as a journal grinder. As far as electrical runout degaussing is the first step but often doesn't get the rotor in spec.

In a proximity probe the magnitude of current supplied and the depth of the eddy current flow into the shaft are a function of the conductivity and permeability of the target material. Inclusions, cracks and carbon segrations cause discontinuities in the eddy current flow, which changes the eddy current response. Keep in mind that whenever there is a change in the electrical properties of the material there is likely a change in the mechanical properties as well. This is often seen as a change in hardness around the periphery of the journal. Inclusions below the surface of the steel may be too deep or localized to easily identify by mechanical means. A point to make here is the eddy probe "sees" below the surface, depending on the material about 300 to 380 µm (12 to 15 mils) in depth. When a shaft is diamond burnished the surface is cold worked and becomes significantly more uniform in surface hardness. This adds significant compressive stress to the outer skin of the shaft that can be seen by the eddy current magnetic field. BN has long stated "...is not fully understood why this reduces runout, but the fact that burnishing generally reduces runout implies that stress variations in the surface of the rotor have a significant impact on electrical runout." The diamond burnishing tool doesn't really remove material. Burnishing is a relatively inexpensive process as it can be done in a lathe. The tool is run into the surface and is loaded against the surface to about 60 kg (125#) by a compressed spring in the tool. It is rotated at low speed in the presence of significant lubricant and results in a mirror finish. Burnishing after grinding will reduce runout by about 60% to 80% but those numbers are highly dependent on the type of steel. High carbon steel (AISI 4140 or 4340) responds nicely while something like 9310 (very high in mangenese) has minimal response to the burnishing tool.

Go to http://www.ge-mcs.com/download...05/3q2005_runout.pdf for an Orbit article (Vol. 25, No. 3, 2005, pp. 5-17) titled "Understanding and Mitigating Shaft Runout” by N. Littrell. It will provide some good insight into mechanical and electrical runout and their reduction. I'd attach it here but it is likely too large.
The article you site is talking about mid-span rotor bow not allowable tolerances at bearing journals and/or probe target areas.

Again, i want to stress that i am fully aware of probe landing area runout tolerance. The MAIN REASON i brought up probe landing area runout tolerance is just for compartive figure to the discussed subject here i.e the rotor radial surface runout tolerance. It is brought up since the tolerance the op has (5.8 um) is lower than its probe landing area tolerance of 6.3 um. Is this a valid tolerance? I am still with my initial comment i.e it is too tight. As clearly pointed in the turbolab article, 5 mils runout is acceptable for their rotor.

As said earlier, those with strong machinery maintenance should come forward and share their exp.

John, btw, thanks for the insight on how burnishing can improve electrical runout.
To minimize the runout error, API 541 and API 546 define limits for the maximum probe track runout. In simple terms, these runout limits are 25% of the unfiltered vibration limits, or 0.45 mils (thousandths of an inch) for most induction machines and 0.5 mils for most synchronous machines. These correspond to 11.4 and 12.7 μm respectively.

Taken from: http://www.ge-mcs.com/download...t_v27n207_runout.pdf
If the area on which one is measuring is a good machined surface, true with the centerline, then 5 mils seems large to me. I thought the article talked about the runout being measured near the shaft end.

I know that John has seen much larger GT's (much much larger) whose measured runouts near the center would be tiny compared to this figure. Now, shaft sag on the larger rotor is much more than this, but you don't measure it as runout.

A temporary bow of 5 mils is very believable. Roll out the bow before measuring.

Low speed balance of a steam turbine with the criticals as described may be taking a chance.

Move the mass 5 mils (or 2.5 if mearing pp, tir) and use a good rotordynamics model to see what can happen. What is the damping at the first critical? Use a simple approximation of amplification factor times mass eccentricity for a first guess (The vibration would be measured pp.).
Many customers specify the amount of allowable runout
for new or refurbished rotors as part of their purchasing
documentation to their vendors. American Petroleum
Institute (API) Standard 612 is one such frequently cited
specification. It pertains specifically to mechanical drive
steam turbines and requires the TIR to be 0.25 mil pp or
25% of allowable vibration, whichever is greater. API 617
has identical runout requirements and pertains to
process centrifugal and axial compressors as well as
turbo-expanders. Similar API standards exist for other
machine types.

Taken from p4: http://www.ge-mcs.com/download...05/3q2005_runout.pdf
Josh, we were debating on runout for full rotor radial runout assessment.

Probe landing area TIR tolerance can be obtained from API doc. But as for rotor/shaft run out tolerance (NOT THE PROBE LANDING AREA), it is no where to be seen in API. So, question is, is 5.8 microns (lower than probe track tolerance) shaft run out tolerance valid?
By the way, when i was in maintenance 5 years ago, i used to jointly assess full runout (as part of as recieved inspection) check (radial and axial runout) for refurbished or spare rotor on vee blocks. I remember doing the activity for a centrifugal compressor spare rotor for change out. We have 0.03-0.05mm as tolerance. maximum radial runout obtained was 0.07mm (on one of the impeller OD). no sign of bow though. the rotor was accepted and it is running well after change out done

Cause and Effect of Rotor Bowing

A large steam turbine rotor must run with
low vibration levels during operation and
when going through critical speeds. Rotors
must be straight and true within a tight
tolerance, typically, 0.0005” (half a mil)
per foot length of span. For most HP and
HP/IP rotors, the Total Indicated Runout
(TIR) should not exceed 0.007” (7 mils).

Josh, I must say your googling skill is superior than mine. Thanks for the 2 references.

In addition to what you have shared, this is another turbolab paper i just found. The writer states the obtained runout readings (figure 1 : Shaft max runout is 1 mil and impeller eye max runout is 3 mils) are WELL WITHIN OEM tolerance.


I guess these 3 references should be adequate to conclude that probe landing area TIR tolerance is not to be mixed up with rotor full runout assessment.
As Josh has pointed out most API specifications and in particular API 670, the instrumentation spec, state "These probe areas shall be properly demagnetized or otherwise treated so that the combined total electrical and mechanical runout does not exceed 25% of the maximum allowed peak-to-peak vibration amplitude or 6 micrometers (0.25 mil), whichever is greater."

Later editions of API specifications typically have a statement for allowable vibration based on a formula using rotor speed as a parameter; i. e., 25.4[(12000/RPM)]^½ in µm or [(12000/RPM)]^½ in mils. The key part of the acceptance criteria is it also states the vibration is limited to the value of the formula "...or 25 µm (1 mil) whichever is less.." So a 3000 RPM turbine is allowed 25 µm (1 mil) based on the portion "whichever is less", not 50 µm or 2 mils.

Taken together the two statements mean that 6 µm or 0.25 mil should be the maximum observed combined electrical and mechanical runout and in addition the running vibration will not exceed 25 µm or 1 mil pp. This holds for most compressors and turbines where an API specification is invoked.

The reference I cited by URL and which josh attached (and quoted from) applies to motors. I don't have the API motor spec but I can tell you that motor shafting is typically very poor steel as compared to that used in turbines, compressors or high speed gearboxes. A motor is commonly cold or hot rolled bar stock, often AISI 1040 steel. Turbines, compressors, high speed gearboxes, etc. are commonly vacumm degassed high quality forgings. The point here is I would anticipate a motor rotor to exhibit higher mechanical and electrical runout as compared to turbines, compressors, and high speed gearboxes.
Must be an echo...

John, thats the point John. Thats the point.

are you sure 0.0058mm is the run-out tolerance? it is very tight imo

maximum readout is one thing, how about the shape of the rotor based on the full (along the shaft) max runout/angle readings?
maximum readout is one thing, how about the shape of the rotor based on the full (along the shaft) max runout/angle readings? ...

I was trying to prove 5.8um is likely a very tight tolerance for TING's turbine. Even with the 3 references, you are still giving me the details on probe landing area TIR tolerance. Probe track area tolerance was brought up just to show people that for the size of turbine TING has, it is odd to see rotor full run out tolerance is lower than probe landing area tol. As of now, i am yet to see a standard that precisely provide us the way to calculate shaft/rotor runout tolerance. Now with the turbolab paper i last attached the link, even if there is a standardized max run out for a specific size/rotor/eqpt, it is still the shape of the rotor that matters most.
Given two different OEM's, even in physically very similar rotors, the OEM tolerances may vary significantly. The author, John East, was on the staff of Hickham Industries, an excellent repair facility of all types of turbo machinery when the paper was published. It is not considered good practice in a technical paper to publish the name of the OEM nor their engineering tolerances without the permission of the OEM. Very often it is best to keep details "generic", especially these days when lawyers tend to get involved with technical literature.

Also, keep in mind the most important part of the paper, the Conclusion which reads in part
"Though balancing is probably the simplest aspect of turbomachinery maintenance and reliability, many problems have occured due to inexperience, oversights, and lack of sound judgement when rotors are assembled and shop balanced."


Files (1)
Sorry, I was reluctant to accept the statement that the as found runout values are "well within the OEM specification" because the OEM values were not stated in there. Besides the lawyers, are these values also regarded as trade secret?

Add Reply

Likes (0)