Hello,

Here again an significant behaviour of a rotor.
It is from an HP/IP turbine (50Hz, frame 9). Over night the orbit is forward preccessed at flexing proces circumstances, part load). When daybreak arrives (06:00) the frame niner is going to 350Mwatt, base load)-> the 1X orbit is then backward preccessed. (including phase changing strongly). All data is measured in one same time span with ADRE3 and the known DAIU boxes. The vibration itself is low for abouth 30 microns pk-pk , direct, uncompensated.
The main question is: what could cause this behaviour in preccession. The second thing is at base load conditions the steam turbine rotor preccession is backwards! In my opinion this is causing a lot of mechanical stress. which is over long term operation not acceptable.

regards M.

MORE INFORMATION REQUIRED. ORBIT PLOT, SHAFT AVERAGE CENTERLINE PLOT, CHANGE IN OPERATING CONDITIONS IF ANY, CHANGE IN POWER . RUBBING IS THE FIRST THING FOR REVERSE PRECESSION. FLUCTUATION OF GRID FREQUENCY NEARING TO CRITICAL SPEED. A HOST OF PROBABLE REASONS.
REGARDS
IRSHAD AKHTAR

Agree with Akthar, lots of causes of backwards procession. That said at 30um Pk-Pk, I wouldn’t bother looking for specific causes as the investigation is unlikely to take you anywhere useful.

Bently Nevada courses have a tendency to put a lot of emphasis on full and partial rubs and its diagnosis using reverse procession.

There is nothing wrong with using reverse procession to diagnose rub. However, rub of a tilting pad or journal bearing will be accompanied by high vibrations and heat.

Thanks for replying.
I will set some data here tommorrow. shaft avarage centerline plots indicates a rotor position at base load in top of the bearing. i confirmed with the conditions to this investigation. the handsample was at moment of 0RPM and the radial clearance was from the OEM. it is still an indication!
Still its hard to say where the potential radial rub have been. Is it really the bearing or at the buckets nearby the bearing? Because the avarage rotor center was on top in the bearing how does this react on the bearing temperature's itself?? The velocity transducer were low.
Within a week we gonna open this bearing to do a visual inspection.

to be continued.

RUBBING MAY OCCUR IN GLAND PACKINGS, BEARING, BETWEEN, BLADES AND CASING, BLADES AND DIAPHRAGM. HOWEVER IN EACH CASE THE REASON MAY BE DIFFERENT. MINIMUM CLEARANCE IS BETWEEN SHAFT AND GLAND AND PROBABILITY OF RUBBING IS MAXIMUM . WHAT TYPE OF GLAND SEAL YOU ARE USING- SPRING LOADED, FLOATING OR ANY OTHER TYPE?
REGARDS
IRSHAD

At 30 micron, the most likely cause of backward precession is not rub. Akthar, please, fully reverse precession rubs are very rare, and generally don't last long as the machine is quickly damaged and the precesssion goes forward again.

Monitor, is it a Frame 9 Gas Turbine, or a Steam Turbine? I am confused by your intro.

Most likely it is simply that the signal to noise ratio of the runout to the vibration is very low. as a result the sum of the runout and the vibration makes it appear that there is reverse precession. This is especially the case if the orbit is flat.

Have you compensated for slow roll and checked precession? I bet you it is forward. Especially if this is a steam turbine.

The second major cause is that the probe brackets are vibrating more than the shaft, the measurement is relative and if the probes are vibrating this can also make it appear that the precession is reverse. This would not be unusual on a GE Frame series gas turbine.

The chance of this being a rub is slim to none.

Steve,

I mean the bearing #3 from a shaftline frame 9, the HP/IP steam turbine bearing coupled on the gas turbine.

The slow roll vector was very low at this bearing #3 position (1,7microns pk-pk and the pahse was not availlable (N/A). the slow roll speed was for about 150RPM (50Hz machine). So when is compensate this the orbit it still is reversed.
Meanwhile i set the gap reference when the turning gear was engaged (0RPM) and the OEM clearances are set in the ADRE3 software. this gives me a (relative) position during steady state conditions (base load) of the steamturbine. see the attachment for the direct and 1X compensated orbit.
Is this relative shaft avarage centerline position normal during normal operating conditions ??

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

Doc3.doc (78 KB, 21 downloads) shaftcenterline_orbits

quote:

the probe brackets are vibrating more than the shaft..

Steve,
Do you mean the resonance frequency of the engineered prox.probes is causing this vibrations??

No, but even on steam turbines it is possible for either bracket or the bearing strong back to be resonanant making for reverse precession.

You might also try compensation at about 500 rpm rather than 150.. Actually the best way to select a compensation speed is to look at an uncompensated bode, find the lowest speed range where the amplitude is stable and flat, and use that. If you have jacking oil that range may be above 500 rpm.

Again at 30 micron, I think this is not a machinery issue.

Also, I one time had a very strange shift in amplitude and phase on both probes of a bearing, and suddenly the vibration was reverse. Turned out someone had been out messing with the probes and hooked them up to the wrong proximitors. Might check. It is possible.

Steve,

1. When i am at the office i will take a higher RPM (500) compensate.
2. i don't think about crossed wired. the unit runs 'smooth' for long time and had no actions in meanwhile to do some actions on the probe's. also the procedure to connect the bufferd 3300 outputs cables were the same

the bearing check was already planned, i adviced to start with bearing three.

thanks for your comment.

when the data was compensated with 500rpm valid slow roll data, the 1X was stil reversed, but i had a very flat 1X compensated orbit.

This wouldn't get me too worked up as stated.

The proper speed for slowroll is best taken from the plots (data). I would probably use a lower speed than 500 on a 3000 rpm machine. 500 could be on the order of 1/2 a critical, which may be ok. Check the phase and amplitude to see when they are flat; you need both conditions to get good slowroll data.

Hello Bill,

Thank you for your reply.

The 500 slow roll data sample was the first valid data to use for slow roll compensation.
I agree it is not the most proper way to compensate but it was to try. I rather now use the uncompensated dat because the slow roll is very low.

regards Monitor