purge valve caused combustor burning


Thread Starter


In our frame-5 dual fuel gas turbine we have got a high spread temp. and one TC is showing higher about 71C from the lower. There is no high temp. Spread protection configured in the Mark-V logic (which is very strange for me), the machine is old one from 1980s. we made CCI and found in one combustor direct below the nozzle is a burn in the body of the combustor and complete black part with expansion of this part like some explosion happened to it.

Specialist informed us to check exhaust thermocouples and we found some cables insulation between the thermocouples junction box and Mark-v in the J.Box it self ,this insulation has gone in some parts due to sun exposure and also high temperature then we checked the liquid purge valve PL20 and found its coil cut and it was not working.
The specialist told us that the problem of this purge valve caused this damage and burn of the combustion chamber due to oil nozzle was partially blocked and when we run on oil some of this oil was accumulated in this burned C.C. part and when we changed over to Gas this oil accumulation exploded causing the damage of the C.C.

My questions:

1- Can the failure and malfunctioning of the liquid fuel purge valve, PL-20 cause this damage to the C.C.?

2- How operator can know while turbine is running that this purge valve PL-20 has problem??

3- Is this logic not to have protection trip on high spread??

It's difficult for me to completely understand the failure here. But, let's try to reason through some things. When you're running on liquid fuel, there should be no purging of the liquid fuel lines. There are check valves (or there should be check valves) on each of the liquid fuel purge lines. Even if PL-20 was working and energized, the pressure of the liquid fuel would be much higher than the pressure of the purge air. There would have to have been a failure of the purge air check valve, but even with that because liquid fuel pressure was higher than purge air pressure the liquid fuel would flow through the failed purge air check valve and through PL-20. I don't have a Liquid Fuel Purge Piping Schematic drawing right in front of me, but I believe there is a tell-tale leak-off off port from PL-20 which would have liquid fuel flowing out of it.

This means that LESS liquid fuel would be flowing into the combustor than normal. So, it's hard to understand how any liquid fuel could accumulate in the combustor because of this problem.

I believe PL-20 is energized when running on gas fuel to purge liquid fuel from the nozzles so that it doesn't carbonize in the nozzles. There is a very small amount of liquid fuel in the nozzle that is purged when PL-20 is energized, but purge air flows through the check valves and the liquid fuel nozzle passages continuously. Even if there was no purge air flowing because of a non-working solenoid on PL-20, all that would mean is that liquid fuel wasn't being purged. There is a liquid fuel check valve at ever fuel nozzle, so there's a very small amount of liquid fuel that can be unpurged from the nozzles. And usually the worst that happens (and it's bad) is that the fuel carbonizes (turns to hard "granules") which blocks the fuel nozzles passages and makes for incomplete atomization and combustion and leads to difficulty starting and usually higher than normal spreads.

Now, here's a scenario that could have happened. Liquid fuel was not purged from the fuel nozzles and carbonized in the nozzles. When you ran on liquid fuel (you possibly transferred from gas to liquid instead of starting on liquid, I'm presuming) and most of the carbonized fuel was blown out of the nozzles--expect for one. The blockages caused poor atomization and dribbling of liquid fuel into the combustion liner, and the fuel ignited and burned. (I think "exploded" is a little bit of a stretch, but it's possible.)

Some of the 1980s-vintage Frame 5s I have worked on had liquid fuel nozzle assemblies that had internal components that were held in place by a threaded piece which had to be staked into place. It was not uncommon for the staking to be poorly done and for the internal components to loosen and cause poor atomization and dribbling of fuel into the combustor. So, there's another possibility that's not even related to the PL-20 problem which could have caused the damage you seem to be describing.

But, again, I'm not exactly clear about the damage and sequence of events.

Now, about the combustion monitor. A lot of older Frame 5s never had combustion monitors. It was incumbent on the operators to monitor exhaust temperature spreads and take appropriate action to protect the machine. Sometimes, when older turbines were retrofit with newer control systems, the new control systems were configured to most closely match the previous configuration. Frame 5s are pretty robust machines and considered to be the "workhorses of the GE fleet of installed machines. They "take a licking and keep on ticking"--and many do run without much loss of performance with some pretty extensive damage that's not discovered until a maintenance outage.

It would be necessary to see the retrofit contract and see what was supposed to be provided to see if something was missing.

As for how does the operator know if PL-20 is working or not, well, that's a problem. Some sites have added a limit switch (usually a non-contact proximity switch) to monitor valve actuator position.

Lastly, with respect to exhaust temperature spreads, most spreads are the result of cold spots--not hot spots. How is this so? For there to be a "hot spot" that means there must be an excess of fuel flowing into a combustor and that's pretty difficult to occur, though it does happen but very infrequently (for Frame 5s the usual cause is a broken liquid fuel flow divider shaft). There is a LOT of excess air flowing in a GE-design heavy duty gas turbine--that's air in excess of that required for combustion. If there is a hole in a liner (such as happened at your site) or a crack in a transition piece or a broken hula skirt deal, those allow more air to flow into the combustor/hot gas path than normal which results in a cold spot. Blockages of fuel nozzle passages or orifices also reduce the amount of fuel flowing into a combustor which reduces the temperature of the gases leaving the combustor. So, most spreads are the result of cold spots--not hot spots.

[NOTE: Those of you reading this post that operate F-class machines with DLN combustors, exhaust temperature spread hot spots occur as often as cold spots, because of the design of the combustor/nozzles, so the combustion monitor is configured a little differently to take this into account. Remember: DLN combustors are run in a very lean fuel-air mixture to reduce combustion gas temperature thereby reducing NOx formation, and when there is a problem that results in diffusion flame being present where is it should not be present then combustion gas temperatures are higher than they should be.]
Thanks for your kind reply and only for confirmation of the expected scenario that happened to us that expected one nozzle is partially blocked due to non-purging of it due to defect PL-20. then we change from Gas to Oil, and due to poor atomizing some dribbling of liquid fuel collected into combustion liner causing its burn and stretch. How to avoid this above scenario from happening in future?? Is there any precautions we can take?

Also for PL-20, it is actually a solenoid valve so we can not put limit switch or proximity switch for feedback indication. Is there any other recommended solution?

What are the consequence of high spread if not noticed by the operator for long time in this machine as there is no alarm at all configured in Mark-V?

Thanks again for your time and help.
Thank you for the confirmation of the method of damage.

If there was dribbling in the combustor due to carbonization of liquid fuel in the nozzle it probably manifested itself as a cold spot spread and could have been pinpointed using the manual selector valve/pressure gauge at the Liquid Fuel Flow Divider. The general rule is that the fuel pressures to all the combustors should be within approximately +/-10% of each other, and any pressure outside that range means there is some problem with fuel flow to that combustor. A high pressure would likely mean some kind of blockage; a low pressure would likely mean some problem with purge- or liquid fuel check valves. Often overlooked, the manual selector valve/pressure gauge is an invaluable tool for troubleshooting liquid fuel exhaust temperature spreads, even difficulties with liquid fuel firing/starting. I have seen Customers install one pressure transmitter for each fuel line and connected the transmitters to the Speedtronic panel--they found the information so useful to be able to see it on the operator interface (so their operators wouldn't have to get out of their chairs to go to the Accessory Compartment).

If 20PL (I couldn't put my finger on what was wrong with the PL-20 description; it's typically called 20PL!) is just a simple solenoid valve that ports air to another air-operated valve, can you put a limit switch on the air-operated valve? Other than that, you might try sourcing a suitable solenoid valve that has a limit switch that is closed when energized. Maybe not so easy since you would likely want an explosion-proof assembly, but it's worth a try.

The consequences of high spreads left unattended are catastrophic damage to turbine buckets. You could purchase the services of some Mark V-knowledgeable firm to modify the CSP to add the combustion monitor function, either both the BBL and associated logic to alarm and trip the turbine, or just the BBL to generate alarm(s) on high spreads.