[We have a dual fuel unit: HSD/Gas, Non-DLN, Generator Drive ~125MW, Mark-VI controlled, equipped with inlet evaporative coolers, operating in combined cycle configuration (HRSG with no GT bypass stack), commissioned a couple of months ago, has been operating smoothly]

Following a trip due to some electrical fault, we tried to restart our GT today. Things go pretty much as usual till about 2000rpm, which is the onset of the critical speed of the machine. The machine is accelerating through its critical speed, FSR is at elevated levels. The TNH curve is "accelerating" (TNH curving upwards). However past 2000rpm, the TNH continues to rise but curves downwards instead of maintaining that profile. During this time, mark-VI tries to increase the acceleration by increasing FSR, so much so that it gets clamped "starting FSR clamped" is annunciated in the alarm display.

So, during critical speed, there is a noticeable decrease in acceleration when compared to past startups. We tried the start up again, once with and once without the evaporative cooler, but to no avail.

No major maintenance activity has been done on the unit, it is a relatively new unit.

The trend and the alarms for the tripping can be downloaded from the following link (will need GE control system Toolbox for the .trn files)

http://rapidshare.com/files/403899021/P809_Startup
_Acceleration_too_low_ALARM.zip.html

For the purpose of comparison, I am also uploading the trend of a startup only 2 days ago. Please use the link below.

http://rapidshare.com/files/403899699/2_days_ago.zip.html

[Alternatively, I can email you the files as well, drop me a message]
[email protected]

What are the things that we need to check/look for? Has anyone else faced this problem on their unit? We have checked the IGV (it stays at 24 degress), TTXM (its 750 F at the time of tripping), Fuel pressures and flows, CTIM (29C, quite cool).

 

Why can't you tell us exactly what Process Alarm is annunciated at the instant that L4T picks up and L4 drops out? How hard is that? That should be the first thing to be investigated--the Process Alarm that indicates the precise reason why the unit tripped.

I no longer have a copy of Toolbox nor a dongle to use it, so I can't review the trends.

But, one thing that jumps out is that the evaporative coolers should not be in operation until the unit is on line (generator breaker closed). There's no use in having them running if the unit isn't generating power. The air flow is so low that there is likely very little, if any, evaporation with the IGVs closed.

You didn't say what fuel you are trying to start the unit on, HSD or gas fuel.

I love this next statement:

>However past 2000rpm, the TNH continues
>to rise but curves downwards instead of >maintaining that profile.

Speed can't be rising and curving downwards at the same time.

I hope that was a typo on the IGV angle; 24 is pretty low! 34 is more like it.

Doesn't sound like you're hitting exhaust temperature control if the TTXM is only at about 750 deg F.

On a Frame 9E the torque converter is usually de-pressurized at 60% speed, which is 1800 RPM, which is pretty close to 2000 RPM. It just seems like there isn't enough fuel to maintain acceleration once the torque converter is dropped out.

And I've only seen one or two GE-design heavy duty gas turbines where the critical speeds were very noticeable at all, and they were very old machines. GE doesn't publish the criticals and the units are usually very well balanced so that they are barely noticeable if at all pronounced. One usually has to be watching very closely to see a pronounced rise, or using some more sophisticated vibration sensors than the velocity pick-ups usually provided on GE-design heavy duty gas turbines.

If you have a Mark VI, then the acceleration is likely a function of TNHAR (TNH Acceleration Reference) and the feedback is THNA (TNH Acceleration).

Are you certain you have sufficient fuel flow? If the unit is starting on gas fuel, is the P2 pressure equal to the P2 pressure reference? Is the SRV wide open or only partially open?

If the unit is starting on HSD, what is the position of the Liquid Fuel Bypass Valve (presuming there are LVDTs on the LFBV)? Is it nearly closed or what is it doing? What is the liquid fuel supply pressure upstream of the Liquid Fuel Stop Valve? Are the forwarding pumps working correctly? Are the strainers upstream of the forwarding pumps clean and unrestricted? What are the differential pressures on the Liquid Fuel Filters?

But, forget about the evap coolers until the generator breaker closes. The Mark VI is supposed to bias fuel flow based on compressor inlet temp so that if the inlet temp is low there will be slightly more fuel (because the air is more dense) and if the inlet temp is high there will be slightly less fuel (because the air is less dense). It's done by TNHCOR.

And, have a look at the Alarms and work on the condition that caused the trip.

 

Hello CSA!

Our unit can only start on HSD, but there are no LVDTs on VC-3. Since flows from FQLM1 are good (our reference being trends from previous startups), we don't suspect a problem with VC-3. The valve moves freely and follows fal. Good FQLM1 values more or less preclude the possibility of fuel supply problems (it is clear from the trends that FQLM1 is merely following FSR1, and that too in the same fashion as it does in past startups). Filter delta P, FDL1 pressure at the HSD forward skid (5.2bar), and flow divider outlet pressure (when on warmup L2W- ~8bar) are all OK. Delta pressure on the forwarding skid filters are also OK. We have also drained air from all the possible points (filters, bypass, pump, even flow divider).

Yes, of course the alarms and events were the first things we look at. The first alarm is "start-up FSR clamped", and the next one, after a few seconds, is "startup Acceleration too low", and that's what causes L4T to pick up and L4 to drop out. "startup Acceleration too low" picks up if the acceleration stays under 0.08%/s for .1 s.

>However past 2000rpm, the TNH continues
>to rise but curves downwards instead of >maintaining that profile.

What I meant to say was: TNH continues to rise, but the rate of increase (slope) decreases. I should have used the word slope instead of curve. Sorry for the confusion!

Yes, that was a typo. Its 34 degress. And yes, the machine does not go to temp control while starting up, TTXM is normal (not much higher than it should during startup).

>On a Frame 9E the torque converter is usually de->pressurized at 60% speed, which is 1800 RPM, >which is pretty close to 2000 RPM. It just seems >like there isn't enough fuel to maintain >acceleration once the torque converter is dropped >out.

The fuel flows (from FQLM1)from past startups is almost identical right uptil about 2000rpm, when it increases - following FSRACC. at 2000 rpm, FSRACC rises until it reaches the startup clamp limit since it is unable to achieve the desired acceleration rate. So there is actually MORE fuel past 2000rpm, as opposed to being less (when compared to past start ups).

You can see this behavior clearly in the two trends in the zip file at the following link. It has PDF prints from the toolbox trender. You will notice the sharp (almost sudden) drop in TNHA.I have also included the process alarm screenshot.

https://www.yousendit.com/download/K0JRblRnUzhBNkZFQlE9PQ

The only reason we turned evap on was that it was thought that maybe the cooler air (our purge timer is 17 mins, so it does give some time to the evap to cool the air a bit) would give us better CPD which would allow better acceleration- kind of a long a shot but worth a try I guess. Didn't work.

Any ideas?

 

Well, it's very difficult to precisely correlate the two graphs because the scaling is different and the data points are not the same on the two graphs.

The 'Start-Up Conditions Abnormal - Trip' is an alarm I've never seen before. What does the application code that drives this trip do?

Also, the L83FL event is interesting; what drives L83FL in the application code?

It's also interesting that the maximum vibration more than doubles when the acceleration starts to decrease.

I don't think this is a controls issue. Rather, I think something happened during the trip or the coastdown. Did the unit go on Cooldown after the trip? You said there was an electrical fault. Did the plant go black?

It seems that something is actually causing the shaft to slow down and vibration is increasing. This is borne out by the fact that the Mark VI is increasing fuel to try to maintain acceleration but acceleration is actually decreasing.

As for "increasing CPD" by turning on the evap coolers increasing the CPD is going to require MORE TORQUE which is going to require MORE FUEL.

Another interesting thing about the most recent start-up is that TTXM levels out for a while and so does FSR, which suggests that TTXM may be equal to TTRX, or, that the exhaust temperature has hit the exhaust temperature reference while accelerating.

The more I look at the two graphs, I see that the fuel flow in the earlier start-up at 79% TNH is approximately 3.84 lbm/s for an FSR of approximately 19.967. In the most recent start-up attempt, at 71% TNH the fuel flow is 3.83 lbm/s and FSR is 22.179%. So, it would seem that something is causing the unit to require more fuel for a lower speed.

I don't see an alarm "Start-up FSR Clamped".

That's about all I can add from the information provided. I've been wrong before and I'll be wrong again; but I don't think this is a controls-related issue.

I have seen cases where the shaft was hot when the unit tripped, and the unit couldn't be put on cooldown and the compressor shaft "took a set" (slightly warped) and during acceleration the vibration was very high, sometimes high enough to actually cause a trip.

What's usually done in these cases is to put the unit in FIRE mode, initiate a START, let the unit purge and fire and establish flame, and then let the unit warm up in FIRE mode for as much as 20 minutes to help the compressor rotor return to "straight" as it warms. Usually, then just selecting AUTO mode will allow the unit to start accelerating to FSNL. Sometimes the vibration will increase, but usually not high enough to trip the turbine.

We don't know the circumstances of the trip and what happened afterward. But, it sure seems like something is "holding" the unit from accelerating and vibration is increasing and more fuel is required for a lower speed than during a previous successful start.

Let us know what happens.

 

Good news first: GT fired!

but first I would like answer your queries.
> The 'Start-Up Conditions Abnormal - Trip' is an alarm I've never seen before. What does the application code that drives this trip do? <

Startup Conditions Abnormal - Trip is driven by LSTART_FLT_T, which is driven by any of:

A: START-UP FSR CLAMPED AT MAX VALUE (LFSRUHZ_ALM)
B: START-UP FSR CLAMPED AND IGV OPEN (LCSGV_O_T)
C: START_UP ACCELERATION TOO LOW (LTNHALZ_ALM)
(see "LSTART_FLT_T.jpg in the link provided below)

> Also, the L83FL event is interesting; what drives L83FL in the application code? <

L83FL is "Increase Liquid Fuel", driven true if the

[[{(fuel pump clutch is engaged) AND (L14HSX is picked up)} OR L2TVZ firing permissive present] AND fuel tranfer to gas not initiated. ]
(see "L83FL.jpg in the link provided below)

> I don't see an alarm "Start-up FSR Clamped". <

It was reset by the operator before I could do the screen capture. It was there though, can be confirmed from the alarms and event log.

> It's also interesting that the maximum vibration more than doubles when the acceleration starts to decrease. <

While you are correct that GE does not "publish" the critical speed of the machines, but the "range" of critical rpm is generally known, and falls in the the range of TNH when THE GT is in acceleration mode (FSRACC has control). It is precisely in this range that the unit trips. Normally the GT accelerates through this region, but in our case, right in the middle (or so we think), the GT gives up (trips on low acceleration). Then it has nothing but bearing friction and compressor torque to slow it down- it takes a long time to come out of the critical zone. It sits there, slowing down right in its criticxal zone. Vibrations are bound to go up (there is a physically noticeable effect too- stand close to our GT when this was happening and there would be no doubting the fact that the vibrations are high (but still below trip levels anyway- which is a moo point, since GT is tripped anyway)

> I don't think this is a controls issue. Rather, I think something happened during the trip or the coastdown. Did the unit go on Cooldown after the trip? You said there was an electrical fault. Did the plant go black? <

The plant did not go blank and did arrive on Turning gear, so there are no "rotor bow" issues. For the sake of completeness of information, before this shutdown, the unit had operated for only 2 days. Prior to that the unit did shutdown, there was blackout, and rotor stayed standstill for too long, rotor bow was suspected, was left to cool for 36 hours, turned manually, found free, no evidence of rubbing, so put in turning gear, and then fire and left there for about half an hour, and then put on auto- exactly how you recommended.

> The more I look at the two graphs, I see that the fuel flow in the earlier start-up at 79% TNH is approximately 3.84 lbm/s for an FSR of approximately 19.967. In the most recent start-up attempt, at 71% TNH the fuel flow is 3.83 lbm/s and FSR is 22.179%. So, it would seem that something is causing the unit to require more fuel for a lower speed. <

--- snip ---
> We don't know the circumstances of the trip and what happened afterward. But, it sure seems like something is "holding" the unit from accelerating and vibration is increasing and more fuel is required for a lower speed than during a previous successful start. <

You are right! When you highlighted this point, it led us to suspect that the issue is right there from the start, only it becomes more critical during the acceleration phase.

For firing, you need air (cpd and igv's are fine), fuel (there is enough FQLM1). But what happens if the fuel is not being properly atomized? Not all of it would burn, not enough energy will be released to accelerate the GT. the check valve NV101 across the motor driven atomizing air blower was suspected stuck. We opened it, but found it OK. In total, we checked the following

1- Check IGV angles - physical inspection of IGV angles from inside the inlet plenum- found OK.

2- Check VA18-1 (atomizing air re-circulation valve ;connects outlet of shaft driven pump back to the inlet of the atomizing air system) - checked for close position in shutdown state, operated with instrument air, found OK

3- Check NV101 - removed valve from line and checked if it was stuck, found OK

4- VA19-1 & VA33-1- (liquid fuel purge and water purge air header valves) checked for close position in shutdown state, operated with instrument air, found OK

After taking these steps, we had a successful GT fire! I suspect a problem with NV101 - when mechanical team removed it, it probably got reset on its own. NVs are rarely too stuck too badly - a slight tap is usually enough to dislodge them if stuck. I can only think of this is the most probable cause. I guess we will never know - or at least until the next shutdown! At least for now the unit is up, and we can get some well deserved sleep.

I am including the pdf of the latest successful fire trend in the link below. The scaling is the same on the y axis, and the x axis has been fixed to show the same TNH start and end value. Note that in the successful start, the GT reaches the end point (71%) took 5 seconds lesser to reach.

What do you make of the two trends? Do you think it really could have been a problem with NV101 - or even the atomizing system for that matter?

One more question (sort of unrelated though): When the GT first engages the fuel pump clutch, flow divider motor 88Fm turns on for 5 seconds, and then shuts down. This is shown by the initial 5 second value shown by FQLM1 in the first 5 seconds in the trends. It then goes to zero and stays there for a few seconds. In this time, VC-3 is closing. when its fully closed, flow is seen on FQLM1 once again, and this time its fuel (not just 88FM). My question is, is this normal? what is the purpose of 88FM? Flow divider manual says that the flow divider must be run atleast once a month for some time otherwise it might become stuck. Is this motor there to "free" the flow divider if it is stuck? If not, what purpose does it serve, since it comes back to zero rpm before it is freewheeled by fuel a few seconds later. Or should the VC-3 valve close more quickly so that before the flow divider has time to come back to 0 rpm, fuel should already be pushing its way along and keeping it rotating even if the motor turns of?

Here's the link with all the trends and screenshots referred to in the discussion above:

https://www.yousendit.com/download/K0JRYlJYcVhtNEt4dnc9PQ

[If anyone else refers to this post later, and is unable to download from the link (it expires after some time), please feel free to email me at freelancingengineer [at] gmail.com , I will email you the trends]

 

So, a lot of "new" information....

If a unit has incomplete combustion as you suggest then there will be LOTS of white smoke coming from the exhaust and there will likely be high exhaust temperature spreads and flickering flame detectors.

I have never been able to find anyone who could explain when a motor is required for a liquid fuel flow divider or when it's not required. They seemed to be pretty common on many older liquid-fuel fired machines, and especially if the units used heavy fuel oil. But, there must be some kind of criteria for when one is required and when it's not. I've seen them used on units in tropical areas where the ambient temperature (and fuel temperature) never gets below 25 deg C. And, I've not seen them on units in arctic zones where the ambient temperature gets below -40 deg C and fuel temperatures can be less than 0 deg C and fuel heating is required.

I have never been able to understand when they are required and when they aren't. I don't know if it's related to the type of fuel, or the expected viscosity or what criteria is used to deem a motor necessary or unnecessary.

So, I can't answer the question about 88FM.