it is 16MW condensing turbine. coast up data is perfect. we were doing solo run for overspeed test. overspeed setpoint is 5800 rpm. critical speed is 2700 rpm. coast down period is more or less 8 minutes. about 15 seconds after overspeed trip at about 5200rpm, all 4 radial probes start to pick up vibration. it reaches its max range of 150 um at ~2500 and stay overrange fopr close to 4 mintues. it is genuine high 1X vibration where the surrondings was vibrating and even the radial probes at uncoupled compressor registered significant vibration. comp probes register 20 um max at around 1180rpm leaving me the fact the peak indeed occurs at that speed (shifted 1st critical?). vibration starts to drop down from ~1000rpm until 0 speed. the vib was still significant before reaching zero speed.

most data i presented here is from DCS following unavailability of complete data by System1. only few transient samples were captured during the coast down which is enough to conclude that it is purely 1X and no other component exist in the spectrum except 2X (3% of 1X). no rubbing.

now, we re-ran the turbine until overspeed again. perfect (critically damped rotor) coast up data like before and surprisingly, the abnormal vibration did not recur this time. we even had another ran after that i.e we ran at max gov speed and trip it but again no recurrence. most of our part is glad but to some extent it is worrying as it might recur in future. what do you think that could have excited the 1X vibration in the first place (5200 rpm)?

IS IT OIL WHIP TRACKING NATURAL FREQUENCY?
IRSHAD

oil whip 'tracking' NF? first time i heard tracking. i never thought that oil whip can be dynamic. anyway, there is no subsync component in the spectrum. it is purely 1X.

If the first run was after some overhaul, shaft may have been less straight and bend out more at crit speed passage or minor touch helping it to bend, reason for it not doing it anymore could be some longer operation time that keep it more straight or slow roll for some time. Why not at run up then? Maybe the passing of the speed range was faster so it didn´t get so far. Tried for once a positive evaluation, just one of many possibilities and may have some flaw but then someone will surely find it. So the best way to avoid it, don´t trip . OLI

Oli, it is still in pre-comm stage. 2 overspeed protection test were done. the coast up data with same ramping up rate for both runs were perfect i.e negligible change (~2 um) as it passes critical speed. overall is below 10 um. this strange high 1x vib ONLY occur during 1st coast down after 1st OST. during coast down after 2nd OST, it did not recurs. strange isn't?

The turbine experienced a light rub which induced a rotor bend, giving high 1X - most likely touching off on a new seal - this will have opened up the clearance for subsequent problem free runs.

Vibration was high all the way down to zero speed because the rub induced bend was still present- if you can, compare orbits at low speed (~150rpm)from the 'bad' shutdown with a 'good' shutdown (after the rotor had cooled and straightened out)

Terry, thanks. It sounds technically sound. Too bad that we do not have the waveform data the time it starts to pick up vibration at 5200rpm to confirm the possibility of momentarily rubbing that causes it to bend.. we do have waveform data at somewhere ~4000 and it is clean 1X with no sign of rubbing in the orbit timebase. anyway, i will compare slow roll data between the 2 event, if the data is there (the S1 server was and still having communication issue with vib rack)

Could it be possible that the rotor sat for a period of time and became slightly bowed? Then once it heated the centrifugal force straightened it out?
Dave

Dave, it occurs during coast down. It was on slow roll for 2 hours and another 4 continuous hours at 2 idles during the stepped start-up.

Sounds like rub is the leading theory and I can't argue with that.

Just to throw out a few more remote ideas. Did the duration of coastdown change for some reason ? (longer/slower coastdown would result in higher and broader resonant peak). Was anything else different during the high-vib coastdown compared to later coastdowns (condensor pressure, load configuration, temperatures ->thermal growth) ?

pete, the only process influence involved during coast down are sealing steam, vacuum, and oil. no signifcant change of these parameters noted between good and bad shutdown or at the time the vibration started picking up vibration @ 5200rpm of which way outside the critical band. thermal growth of casing im afraid we do not have the data. the period of 8 minutes coast down is normal and more or less consistent for 3 coast downs we had so far.

With very little data very little can be said.

You seem to have some set ideas about rubs, not explicitly stated. 1X rubs have been investigated many years ago (1920's). These type of rubs are indeed 1X. Yes, especially with large vibration, harmonics msy be present.

The onset symptoms are consistent with a rub. The coastdown with changed and elevated vibration in your case all the way to zero speed indicates a bowed rotor, likely caused by a rub -- Oh for the lack of data. The fact that the subsequent run did not have this pattern is a sign that the bow was gone during that run, also consistent with a rub/thermal induced temporary bow of the rotor.

As far as repeating, one would like to know the cause. Some have suggested a bow from the outage; this is a possibility. Unfortunately, you don't have the data.

This System 1 is a very capable system. Was it not set up properly to capture startups and coastdowns?

indeed s1 is very capable system. think transient data collection configuration is ok. it did collect good transient data for 1st start up. somehow right after the the overspeed trip, it experience intermittent loss of data (all probes - vib & process parameters). we do have some sample during the coast down. very few. all of them feature 1X dominant and negligible 2X component ONLY. i will see details of of 1st start-up transient once the opportunity is there (i have limited access at present). yes, momentary rubbing seems to be the closest hypothesis as yet but can't really understand why would it occurs during coast down just after ost and yet rubs likely did not occur during the 1st start up (i presume this since the overall throughout the start-up is max 10 um)until ost.

This message has been edited. Last edited by: Valve,

I think it is a rub as well, there are a various reasons why it might rub on shutdown and not start up. Really depends on where and how it is rubbing.

I think on System 1 you can select plot invalid data. If so you may be able to look at the data during the overrange event. Since it is overrange, you cant read much into it, but if it remains 1X and big fairly round orbits then that is pretty consistent with a rub induced bow.

Attached is a paper I did for the Vib Inst National meeting on rubs, with several examples of this sort of thing. Enjoy.

diagnosing_rubs.pdf (1,055 KB, 54 downloads)

Steve, invalid data is plottable but not sure in this case since the gap between sampled/available coast down data is big i.e 4200rpm...2000rpm...~1000 rpm..few hundreds rpm.. anyhow, i will try to plot it and see how it goes. Btw, i am looking/focusing on rub at present. It is a big round orbit shape. Am studying the laby seal of the turbine in specific. And thanks for your write up. It helped me to understand better.

This message has been edited. Last edited by: Valve,

Gentlemen,

We have come to a conclusion what had went wrong. Firstly, i'd like to wiithdraw my earlier statement saying that there were no rubbing signature in the orbit. I revisited the S1 2 days ago and it turns out that there are few good data for me to conclude that it is rubbing induced rotor bow. The orbit i saw before was a normalised (no more rubbing) orbit. The rubs phenomenon occured immediately after the OST. The seal is of retractable type. The fact the seal is of retractable explains why rubs took place only during 1st coast down and not during operation and the its next 2 coast downs. Thanks all for your response.

Steve, what we had here is identical to your Case History #1.

Cool, I hope to see you next week. I will be on site April 10. I am doing a readiness review for the whole facility.

Great meeting you. Have fun and let us know how it starts up!

Yep! Im looking forward for the start-up. We should be able to conclude it right from begining if we have data on the steam consumptions during the run.