Intermittent Seismic Vibrations Increase at GE Frame 5 Turbine Compressor Casing


Thread Starter


We are facing an issue at our running unit frame 5 gas turbine driving generator w.r.t to intermittently increased seismic vibrations at BB2 (Compressor Casing Inlet) and BB1 (Bearing No. 1). The values at BB3 point (Pedestal Bearing) remain normal during these events.

Other unit is under maintenance, and subject unit is running in ISO mode. Normal values at BB1 & BB2 are 0.8 & 2.8 mm/sec, respectively.

HIGH VIBRATION EVENTS to date (from 13 Feb to 17 Feb). The vibrations at BB2 increased from normal 2.8 to 7.5~8 and returned gradually to normal.

All the event lasted for ~10 minutes, and values returned to normal at BB1 and BB2. The values reduced gradually in interval of sec. The local vibrations at the time of last event were noted and found increased from 3.1 to 3.6.

The setting on Mark-V is 12.5 mm/sec for alarm and 25 mm/sec for tripping. (Max values out of BB1, BB2, and BB3). VTs are installed at BB1, near Rigid coupling of generator and pedestal bearing. Except the erratic value of BB1, other values of VTs at BN-3300 are normal (15-35 microns). The thrust value is also normal at BB1 is 7 to 8 mils. During these events, no change was observed in any other parameter (TG load, speed, CWS temperature & pressure, Lube oil supply temperature & pressure, exhaust temperature).

Compressor side primary filters were replaced one month ago, and condition of secondary filters were also good. All the events occurred in daytime with ambient temp range from 22 to 28 deg c.

Please discuss about possible causes and checks during operation of the machine.

Thanks & Regards
The subject Unit is under parallel operation with other Unit which is under maintenance. It is mostly on ISOCH, while other is on Droop control.

You mention "VTs"--Vibration Transducers?

GE-design heavy duty gas turbines typically use one or both of two types of vibration measuring devices: velocity sensors (also called seismic sensors) and displacement sensors (also called proximity sensors).

Most every GE-design heavy duty gas turbine has seismic (velocity) pick-ups (sensors). And, some have redundant seismic pick-ups--two pick-ups mounted on the #1 bearing (at the axial compressor inlet), and for Frame 5 machines, two mounted on the #2 bearing (in the Load Compartment. For Frame 5 and -6B machines with Reduction (or Load) Gears, sometimes there are additional seismic pick-ups mounted on the Load Gear. And, then on the pedestal bearing of the generator there may be one, or two, seismic pickups mounted on the top half of the bearing pedestal, with usually only a singe seismic pick-up mounted on the collector end generator pedestal bearing (again, usually on the top half of the pedestal bearing). The philosophy of using seismic vibration sensors mounted to bearing pedestals/housing/load gear cases is that the mass of the rotating equipment is very large, and any shaft vibration will be transmitted to the bearing casing/pedestal and will be sensed by the seismic vibration pick-ups.

Proximity vibration pick-ups such as those manufactured by Bently-Nevada (a GE subsidiary), are placed in holes drilled through the bearing housing/pedestal and the proximity probe tips are very close to the shaft. Magnetic interaction of the shaft with the proximity probe tips when there is vibration produce the signals seen on the B-N monitors.

It's not clear from your post which bearing, specifically, you are referring to. In general, two-bearing GE-design heavy duty gas turbines refer to the bearing at the axial compressor inlet as bearing #1, and the bearing in the Load Compartment as bearing #2.

The bearing at the turbine (drive) end of the generator is usually called generator bearing #2, or sometimes it is called Bearing #3 in a GE turbine-generator set-up with a two-bearing turbine and a load gear and generator coupled to the exhaust end of the turbine. And, that would make the collector end (the non-drive end) of the generator the #4 bearing. So, if you're standing looking at the unit, with the turbine on the left and the generator on the right, with the load gear in between, the turbine- and generator bearings would be numbered from 1 to 4 from left to right. (Depending on the type of GE turbine control system, there is a funky numbering scheme for each of the sensors on the bearings. And for those of you reading this with GE-design F-class machines, well, the bearing numbering system is NOT the same...!)

Some GE-design heavy duty gas turbines, mostly those usually provided by licensed packagers of GE units, did NOT have redundant seismic pick-ups on the turbine and generator bearings--they only had one seismic pick-up on each bearing. The nice thing about having two seismic pick-ups--whether or not the unit also has B-N proximity probes--is that whenever there is shaft vibration BOTH seismic vibration sensors will usually register the SAME vibration levels, or nearly identical readings. It's also not clear from your post how many seismic vibration pick-ups are installed on the bearings of the unit at your site.

The seismic vibration pick-ups are bolted to the bearing caps, and sometimes those bolts come loose; too often. Also, the cabling, especially in the load compartment, can get over-heated and degrade. Also, the seismic vibration sensors use screwed-on plugs to connect the cables to the sensors, and these can become loose over time, and they have also been known to corrode or oxidize, depending on environmental conditions at the site.

While not delicate instruments, they do not tolerate much abuse, and they have been known to suffer a lot of abuse during maintenance outages. And, all seismic vibration sensors--even if they look identical from the outside--are NOT the same. Some are rated for high temperature applications and some are not. And, because they do look very similar, it's common (too common) for a low temperature-rated sensor to get installed in a high temperature location.

Finally, even though they might look identical from the outside, seismic vibration sensors have different output characteristics--for example, one might have a 100mv-IPS rating, while another may have a 200mV/IPS rating. While they might appear to be identical, they are not. And, far too often, they get mixed up. (And, yes, GE does provide units with a mix of seismic pick-ups with different output ratings.)

One of the very first things to check whenever there is a disagreement with seismic pick-ups and proximity pick-ups is to check the mounting of the seismic pick-ups in question. And, the cable connection(s) and the integrity of the cable insulation. If only one of two sensors at a location is indicating high vibration and proximity probes in the same location are not indicating a high vibration, then it's likely something amiss with one of the seismic pick-ups.

If there's only one seismic vibration pick-up at a bearing and it's indicating high vibration when proximity probes in the same location are not, well, then, it would seem like something is amiss with the seismic pick-up, its mounting, or its cabling.

AND, there's always the sanity test for any vibration. When you're standing in the area that has a high vibration indication (either from a seismic pick-up or proximity probes)--do you feel (in your toes) increased vibration? One type of sensor indicating high vibration when another sensor in the same location is not indicating high vibration but a sanity check reveals there is high vibration can help detect a sensor/mounting/cabling issue--and help prevent catastrophic problems. So, it's always a good idea to go out to the unit where the high vibration is supposedly occurring and check for one's self.

Troubleshooting is often times a process of elimination. Check each component and when it's proven to be properly installed and connected and working correctly, move on to the next component.

If the vibration is occurring on the drive-end of the generator, then the problem could be related to excitation--something may be failing in the generator rotor windings, and high or low excitation current may cause the problem to get worse at certain times.

So, there's really not sufficient, understandable, actionable data to understand where the problem is, how the machine is instrumented, and what might be the problem.

And, if you need more help, PLEASE be more specific about the number--and location--of seismic (velocity) pick-ups on your machine. NOT all GE-design Frame 5 heavy duty gas turbines are instrumented identically, and subtle differences can be important.

Thanks a lot for such a detailed reply. Starting from Compressor Inlet Side, there is one seismic pickup and two proximity probes for shaft vibration and thrust monitor installed at Bearing no. 1.

Next seismic pickup is installed at Compressor inlet casing, called BB2. Next there is no instrumentation or pick up at bearing no. 2. At reduction gear the vibration is noted fort nightly thru local vibration monitoring.

Next proximity probes installed near rigid coupling and at Pedestal bearing BB3 (one seismic pick up and two proximity probes).
BB1 on Mark V - Seismic pick up at Bearing no 1
BB2 on Mark V - Seismic pickup at Compressor inlet casing
BB3 on Mark V - Seismic pick up at pedestal bearing

Now we are facing intermittent rise of vibration at BB2 daily at about 12-3 pm, with ambient as 22-28 c and L.O supply temp as 51-55 c. The vibration increases gradually from 3.2 to 8.5 and comes back to normal in 20 minutes. similarly vib at BB1 also increases at the start of event at bb2 from 0.8 to 3.2.

The load is around 14-16 MW. Can it be the reason of oil coking at either of bearing no 1 and bearing no 2, so that when that layer of oil coke breaks, the shaft vibration starts?

Also does it have any effect by temp of L.O .

Oil coking in a beating housing could probably be detected by an analysis of the L.O. as it would surely result in suspended solids in the oil. I'm presuming you are referring to a situation where the oil is overheating and resulting in the formation of solids. Can you explain how that would occur at only a particular load or specific operating condition? And what would be the source of heat in the generator bearing?

What happens when the Generator excitation is varied over a wide range while the load is maintained at a steady value?

The temperature of the oil entering the bearing should not be less than approximately 60 deg F or it can result in high vibrations, but it's difficult to understand how that would occur only at a specific load.

When a generator is synchronized to a grid and the frequency is stable the speed of the generator and turbine is also stable. So, aren't you looking for something that is changing in a specific range of load?

I think you are attempting to relate unrelated conditions.
The most suspected bearing to have oil coking is the bearing that is in the exhaust area (high temperature environment).

The last time we had an oil coking, the vibration was cyclic. The vibration was detected by the proximity sensor. The problem was not detected by the seismic sensor to my knowledge.

Oil coking will decrease the clearance between the shaft and the labyrinth seal of the oil deflector. This decreased clearance will cause the shaft to rub against the burned (coked) oil.

I believe you have to plot what is called orbit graph to tell whether the shaft is rubbing or not.


We have done oil analysis of the oil and found it normal.

All we are trying to relate is the occurrence of the seismic vibration peak at peak temperature time of the day.

Can it be due to any choking in seals or vapor formation due to temperature (No mist eliminator installed, only vent pipe)? We have done online wet washing also. the peak has not appeared since last two days, the only change is wet washing two days back and ambient lesser than previous week.

Yes, we too are suspecting the same bearing in exhaust (Bearing no. 2, which is initiating the vibration scenario. Unfortunately there is no instrumentation available at Bearing no 2 (neither seismic nor proximity).

The proximity probes at bearing no 1 (are giving faulty reading). While the ones at Reduction gear and pedestal bearing in generator end have true readings. We connected ADRE for the data collection at reduction gear and Generator, but there is no such thing at that end.
The vibration pick up is shown at Mark VIE at seismic f inlet compressor casing and then seismic at Bearing no 1 (Acc. Coupling side).

A rub due to oil coke is a good candidate to this description. Unfortunately, with the existing data it is difficult to confirm.

If you have an Adre, have you tried to get phase referenced data of the event? Plot the 1X vibration in a polar plot and see how it is (assuming that the vibration increases due to the 1X component). You need the correct instrumentation in order to do this ... do you have a keyphasor in your machine? Which Mark do you have? Are the seismic sensors connected to the Mark through a VVIB or similar card (with BNC connectors)?
Thanks Stecav,

We have ADRE, but as told earlier, the vibration of proximity probes at bearing no 1 is not accurate, and there is no proximity probe at bearing no 2 Exhaust hood. we can take plot of ADRE only on reduction gear proximity probes and pedestal bearing proximity probes. ( but the vibrations are normal at both gear box and pedestal bearing)

Yes, We have a keyphasor, but it is connected to Bentley Nevada vibration monitoring system which is separate to Mark V system. Probe has a jig with BNC connector, but in Mark V it is terminated at terminal board of R Core with I lugs.
If the keyphasor goes to a Bently monitor you can probably hook-up your Adre Kph input to the BNC connector in the Bently. At the same time you would need to measure vibration on the bearing 1.

If I understand it correctly, the proximity probes give bad readings and the seismic sensor is wired directly to the Mark V so it is not possible to take that reading in paralell to the Adre.

Then I think that the only option would be to use a spare sensor with a magnetic base and placing it next to the seismic sensor in bearing one (assuming that you can access that bearing while the turbine is running). Then you would need to run a wire from that seismic sensor to the Adre (assuming that this doable, I don't know the distance from the turbine to the Bently monitor) or if the proximity probes are not working you can take the wire just until the JB of the proximity probes and use the cables of the proximity probes until the Bently (assuming that your plant policies allow you to do all this while the turbine is running).

Then you could plot with the Adre a polar plot and see how it looks like. That could give additional hints to sustain the coke rub hypothesis.

On the other hand, I've seen turbines suffering of a coke rub running without problems for months and they were not cleaned until a planned shutdown several months afterwards.

The Gingerbeer

I am aware that I am late into this conversation. however, there may be advantage in surveying horizontal vertical and axial magnitudes and their frequencies to help isolate the issue and remedy. You mentioned that a water wash has been done and the vibration has not recurred. The compressor may have been dirty and that could have been the issue. Also the vibration occurred in the hottest part of the day. If the unit operates with modulating IGV then the specific ambient conditions coupled with a dirty compressor and specific IGV angle may have been the cause. Keep monitoring and if you can on line wash then repeat as often as you can. Some sites do this daily to maintain compressor performance.