GE frame 9e speed fluctuation during crank mode (gas turbine purge)

Good afternoon, at the moment we are commissioning a GE frame 9e gas turbine, and during the crank mode (purge of GT), the turbine speed fluctuates from 475 to 541 rpm. Contractors insist that this is normal, we tell them the speed is not stable during gt purge (crank mode)(475-541rmp) and it is not ok. Could you please help us if this is normal or not and how to prove it to the contractor. Also, when the speed reaches 541 rpm, the 20tu valve energizes and relieves pressure. when the speed reaches 475 rpm the valve closes and the speed increase begins, and so the speed of the turbine floats from 475-541. Are there any documents that describe this in detail. URGENT!!!

Good afternoon,

Do you have a clear picture of what should going on ? What should be " purge speed " ? 20%of rated speed ?

I have a diagram explaining well startup curves ..
You need to ask the GE-design Frame 9E experts at GE Belfort, or the packager/supplier of the turbine-generator. We don’t know what kind of starting means the unit employs or the logic that’s required to make it work as designed. GE Belfort have ultimate responsibility for this Frame-size equipment, but licensed packagers (e.g., BHEL) have some leeway in auxiliaries like starting means. As long as the auxiliaries meet the guidelines for machine operation and protection.

GE Belfort have used some different methods of controlling purge speed and monitoring/limiting starting motor current draw. This doesn’t seem to be the case and if the unit uses a motor-operated torque adjustor on a Voith variable output torque converter what you describe seems to be abnormal operation, which should be smooth acceleration to purge speed (which wasn’t usually a set speed, per se, but the speed that corresponded to approximately 150-169% of rated starting motor current), some time above a minimum purge speed (in RPM) and a smooth deceleration to less than 9.5% speed and a slow acceleration up through 10% speed at which point the firing sequence would begin.

It would be necessary to see all of the application code and the P&ID for the starting means system and the torque adjustor drive motor schematic to say for sure how the machine was intended to operate—but what you are describing doesn’t seem normal based on past practices. It might’ve also be misadjustment of the torque adjustor limit switches (though GE Belfort seems to be trying to eliminate them) or a problem with the application code that controls the torque adjustor. But we would need a lot of information to say for sure.

GE hasn’t really been properly training field service/commissioning personnel for a few years now. GE Belfort, in their defense, usually does a reasonably good job of producing good Operation & Maintenance Manuals and the Starting Means section should have a good description of how the sequence should progress. But, they are not always consistent.

We wish you much luck!
Without being able to see the Starting Means P&ID for this machine we can’t tell if 20TU-1 should be energized or de-energized during breakaway, acceleration to purge speed, during and after successful firing and during acceleration. My belief is 20TU-1 should be energized only when torque is to be transmitted from the starting means to the turbine-generator shaft (as described above) and de-energized at other times, mostly as a fail-safe measure to prevent applying torque to the shaft in the event of failure of the solenoid or the control system.

But—that’s just an educated guess without being able to see the Starting Means P&ID.

If you can’t convince the commissioning personnel that the data in the graph is indicative of a problem then you should probably ask for a meeting with the turbine-generator packager and the commissioning personnel manager to discuss how to move forward with resolving the issue of how to achieve and maintain firing speed during purge time—as well as after firing so the machine will accelerate properly. ‘Cause it don’t look promising to even fire and if it does fire then how is it going to accelerate to self-sustaining speed and eventually to FSNL (Full Speed-No Load)?

We don’t often get feedback about issues like this; some B.S. explanation is made and the application code gets corrected and people are just so happy to continue with commissioning that we just never hear what really happened. So—PLEASE write back to let us know how the problem was resolved.

[If this is truly an application code problem the question in my mind is: Why wasn’t this identified and corrected during software tests after the modification was made to prove it was correct? If the starting means is typical of what GE provided for decades, complete with proven application code, why the HELL was the application code changed??!?!! And why wasn’t it tested before it was sent to the field?]