Flame Failure : Frame 6B Machine


Thread Starter


We are operating 2 Frame 6 B machines with Mark VIe control system, with DLN-1 system, synchronized with grid in floating condition which is stable. The machine is continuously running on Preix steady state / low Nox condition. On continuous basis gas turbine is running on part load (approx. 24MW). Combustion hardwares were replaced with 32K compliant. Last MI was done in Feb 2015. The HRSG steam is used for process requirement.

Since last 3weeks we are finding the issue of flame intensity in secondary scanner No 5/6 attached to combustor 2/4. Low falme alarm is prevailing. CO in the exhaust was checked ad is 3 to 4 ppm.

No increase in vibration or other deterioration in performance observed.

Machine is due for MI in 2019. Machine was running for last 3 years with out any issue (Stopped only 3 times in 3 year).

Machine tripped yesterday on flame failure. 2 scanners were replaced and machine was re started without any notable change with out forcing any parameters. In normal operation when IGV is on temperature control mode and IBH is ON machine reaches TTRF 1080 deg C at 13 MW with IGV about 47 %. If IBH is OFF, rarely we go to lan lean positive and then to premix steady state.

Now with IBH off mode also reference temperature reaches 1080 with IGV in 56 % opening at 19 MW and goes to premix steady state.

When second time machine tripped on flame failure IGV calibration was checked, GCV/SRV/GSV calibration was checked, all were found OK. We started the machine and kept on low load at 12 MW due to process requirement to meet the steam demand.

Compressor discharge temperature CTDA1 / CT DA 2 difference were observed during starting. CTDA2 was 55 CTDA1 was 110 Deg C.
While scrutinizing the trip log in both trips minor variation in P2 pressure was observed and SRV also was found varying 20 minutes before the flame intensity variation occurs.
Based on this SRV/ GCV / GSV and IGV calibration was done.
Spread normally remains below the allowable range. Spread 1 touches the limit during the load increase

Current status
Machine run on lean lean positive (Before premix transfer state) at 17 MW in IBH close mode. Flame in all scanner healthy. Vibration 3mm/sec max.

Expecting valuable feed back and experiences in this line from the esteemed team. Can share the trip logs if required.

I'm not exactly 100% clear on precisely what is happening, but here's my input. If flame is truly being lost in one or more combustors while in Premix Steady-State mode, then the exhaust temperature spread(s) will also be high (probably because the diffusion flame in the secondary combustion zone has been extinguished, so the spread won't be because of high exhaust temperatures, but rather because of low exhaust temperatures). So, if the unit is not experiencing high exhaust temperature spreads when flame is lost, then its likely flame is not really being lost.

Two things about secondary flame sensors: First, they have to be properly "aimed" to look down the tube they are screwed to and down the center body of the combustor past the secondary fuel nozzle to be able to see the flame in the secondary combustion zone. Because of the distance from the flame detector to the diffusion flame in the secondary combustion zone even a small deflection in the angle of the flame sensor can cause issues with properly detecting the flame.

Second, the flame sensors usually are cooled with small coils through which cooling water runs, and the flow through those coils is usually controlled with needle valves. If the flow of cooling water is very high that can result in the formation of condensate (from moisture in the air and combustion gases) on the lens of the flame sensor which will block the detection of flame.

Combined, these two phenomena can cause big problems. They are both pretty hard to detect or correct, especially without shutting the unit down. It should be possible to reach the needle valve(s), though, to decrease the flow of cooling water down to see if that causes the flame intensity to increase on one of the flame sensors with the problem, and if that works then do the same to the other(s).

I have even seen people INCREASE the cooling water flow to suspected problem flame sensors mistakenly believing that the problem is insufficient cooling water flow.... If it is very humid at the site, and/or the unit is using evaporative cooling and/or inlet air fogging that greatly increases the amount of water vapor in the air which is more easily condensed on the flame sensor lens.

It's difficult to prove the presence of moisture on the flame sensor lens by removing the flame sensor--because that can only safely be done when the unit is shut down, and usually by that time the moisture has evaporated. So, throttling with the cooling water flow needle valve(s) is about the only way to test for issues. (I had a Customer a couple of years ago that SWORE they weren't having moisture condensation on the flame sensor lenses, and we were able to definitively prove by throttling the cooling water needle valves they were. And, that unit was a Frame 5 GE-design heavy duty gas turbine, too, with DLN-I combustors!)

If you have a milliamp simulator you can test the Reuter-Stokes flame sensors (Flame Trakkers, I think they are called) you removed to see if they are working or not. It doesn't take very much light for those detectors to indicate flame. You can use most good LED torches, or just point it out an open door to sunlight, to get a reading.

There is a third issue--but it would also be combined with high exhaust temperature spreads--and that is that the secondary fuel nozzle orifices in those two combustors (and likely others, too) are plugged or plugging due to contaminants in the gas fuel.

Also, the CTDAn readings are very suspect. Typical axial compressor discharge temperatures while on load are around or above approximately 325 deg C, depending on load and IGV angle (and to a certain extent IBH flow). CTDA is an important input to the combustion firing temperature reference calculation.

Finally, you have not mentioned if you have any Diagnostic Alarms on the Mark VIe. They can be indicators of Mark VIe hardware problems (such as with the mA input channels/cards the flame sensors are connected to).

Hope this helps! Please write back to let us know what you find.
We are currently running at 16 MW with IBH off mode in Lean Lean positive mode.

After 12 hour operation the flame intensity in secondary scanner 7/8 is now showing 20 % approx. But machine stable, no major CO upset or vibration issues. We are operating in middle east / Oman. Last few days the humidity is higher and atmosphere is dustier than normal. Filters are OK and pulse cleaning is in put on continuous mode as a safety measure. (Machine is only gas fired)

What has been mentioned we will check in the next opportunity which requires machine shutdown.

We will update regarding the latest developments. Any additional monitoring please guide. Thanx.

You should know that operating in Lean-Lean Positive (or Extended Lean-Lean) for prolonged periods of time is NOT recommended by GE.

Because there are large diffusion flames in both the primary- and secondary combustion zones, this causes high thermal stresses on the combustion liner, the venturi in the combustion liner, and the secondary nozzle "tube" in the combustion liner.

GER-3620 has recommendations for prolonged operation of DLN-I combustors--and how it affects maintenance outage planning (it reduces the time between outages due to the need to inspect the combustion liners because of the high thermal stresses).

DLN-I combustion systems were designed to be operating in Premix Steady State combustion mode. Primary and Lean-Lean combustion modes are just meant to be transitory modes on the way to Premix Steady State during start-up and loading, and during shutdown.

Extended Lean-Lean is just a mode that prevents tripping when the unit "drops out" of Premix Steady State, and the operators should take action quickly to return to Premix Steady State (presuming conditions will allow it; high exhaust temperature spreads, for one, will keep dropping the unit from Premix Steady State back to Extended Lean-Lean and likely requires a shutdown for investigation (borescope of combustors/nozzles, usually)).

IBH is a means of allowing Premix Steady State operation at loads lower than would be possible without IBH (it protects the compressor when the IGV angles are below approximately 57 DGA, the typical minimum rated speed operating angle for most GE-design Frame 6B heavy duty gas turbine axial compressors).

You previously mentioned fluctuations SRV fluctuations and P2 pressure variations. These kinds of fluctuations can cause flame instability in Premix Steady State, particularly at low loads. It would be good if you understood what was causing them. P2 pressure is a function of turbine speed/generator frequency. Was the turbine speed/generator frequency stable when the P2 pressure fluctuations occurred?

Sometimes P2 pressure can be affected by upstream pressure (gas fuel supply) pressure fluctuations. Was the gas fuel supply pressure stable when the P2/SRV fluctuations occurred?

Most GE-design heavy duty gas turbines use compressor discharge extraction air for pulsing the self-cleaning inlet air filters. This means that a momentary flow of air will be periodically taken off the axial compressor discharge to supply the self-cleaning inlet air filter system. At low loads and with low IGV angles, the sudden periodic flows of air can cause flame instability in DLN-I combustors, which can cause primary zone re-ignitions when operating in Premix Steady State (dropping back to Extended Lean-Lean). If your unit uses separate electric motor-driven air compressors or plant air (that is, not axial compressor discharge air) to supply the self-cleaning inlet filter system you can disregard this paragraph.

You can check the flame sensors you removed with a mA simulator; they just require 24 VDC in two-wire simulation mode from the mA simulator. Flame intensity is reflected in the mA flowing in the circuit. When you are pointing the flame sensor at a strong light (LED torch; sunlight) the mA value will be closer to 20 mA. If you cover the end of the flame sensor with your hand, the mA value will be closer to 4 mA --if they are working correctly. I suspect they are fine and did not need to be replaced--and this test could prove that.

If they are good, I would suspect there is too much cooling water flow to the coils around the sensor bodies causing condensate on the lens. You mentioned the humidity is high now; this could be contributing to the problem. Have a look at the Cooling Water P&ID supplied with the turbines to see if they included needle valves to throttle the water flow to the cooling coils around the flame sensor bodies. If so, you should be able to start reducing the cooling water flows of the suspect flame sensors, a turn or so at a time, while monitoring the flame intensity to see if it increases. If the exhaust temperature spreads are good, it's likely the flame sensors just can't see the flame because of moisture condensing on the lens.

Again, this takes patience to determine because it can only be done when the unit is running (which means someone has to enter the Turbine Compartment when the unit is running, which is not permitted at some sites, and is under certain conditions at others).

Finally, inlet filter houses have to be assembled at site during construction. And that means a LOT of welding has to be done. Sometimes, stitch welds are used which means there is the possibility over time that the spaces between the stitch welds can open and allow dirty air to enter the inlet ductwork--and the axial compressor.

Also, the inlet ductwork has to be assembled at site, and many times stitch welds are also used when assembling the ductwork, also. It's best to go inside the filter house (downstream of the filters) during the daytime and close the door behind you and look at all the seams/joints. If you see daylight anywhere, that means that dirty air can enter through that area. (A lot of people are surprised when they make this inspection. And, high ambient temperatures can make these areas open even more as things expand during the day, and contract during the night. It can lead to twisting and breaking of stitch welds, also. So, periodic inspections are always good; not just one.

If you would be so kind as to share the Process- and Diagnostic Alarms which are currently annunciated on the HMI we could be analyzing those to see if there are other actions we might recommend.

Hope this helps, and keep us informed how you fare in resolving this problem.
Dear Sir,
Thank you for sharing your experience.

Based on your observation we were looking for an opportunity to verify the issues. Incidentally today the issue of loss of flame reoccurred while the machine was transferring to premix steady state. We checked the scanners and the scanners were having drops of condensate. In Scanner 8 which was most troublesome, 3-4 drops of water is observed. We are now replacing the scanner, inspecting the IBH for any short circuit of air. Throttling cooling water up to 60% we tried, but it didn't give any result.

Any other proposal you have to address this issue? Waiting for your reply.

Once again thank you for the guidance provided .

So, were talking about the secondary flame detectors. I know some sites have drilled small holes in the secondary flame detector tube (the part that protrudes out of the back of the secondary fuel nozzle block) and piped a very small amount of compressor discharge air into the pipe. Compressor discharge air is hot, and because there is little to no air flow in that pipe in its normal configuration this hot air, which will flow into the combustor, helps to keep the pipe warm and push air, and moisture back into the combustor, helping to keep the lens of the flame sensor warmer, and hopefully drier. (NOTE: Compressor discharge air is moist, itself, so if the cooling water flow to the flame sensor is still too high, this can add to the moisture in the pipe and it can condense on the flame sensor lens.

It doesn't take much air. The problem, in my personal opinion, is the additional piping/tubing and fittings required to pipe the compressor discharge air to the flame sensor tube. There's already enough pipes and tubes in most turbine compartments....

If you try this, I would suggest trying it on the most troublesome flame sensor (you said it was #8, I believe) first (just a single sensor). You can always cap or plug the hole if this doesn't work.

But, I'm also curious how you determined that the flow had been cut by 60%. I would try cutting it back more and watching the results. Not all valves have a linear flow-rate through them that is a function of the opening.

Moisture on the lens of a flame sensor does not make it unusable; not at all. Just dry the lens and it can be reused, no problem. Again, they are VERY easy to test (if they are SiC type of flame sensors; the UV flame sensors (usually made by Honeywell) require a "stronger" light source, but some high-intensity LED torches seem to do the job).

I would also suggest you have an analysis of the fuel gas currently being burned--with particular attention to the dewpoint of the fuel gas. If it's too low, then fuel gas liquids can condense in the unit; there are pressure drops at every control valve, as well as across each nozzle/orifice which can cause condensation if the fuel gas dewpoint temperature is too low--and that can cause flame stability problems, too. GE recommends the fuel gas dewpoint temperature be at least 50 deg F above the fuel gas temperature.

Hope this helps!!!
Dear Sir,
We had taken following actions and now GT flame is healthy for last 2 days with 23 MW load . We are running the trend also, it is showing healthy reading.

The flame scanners were replaced. After opening the flame scanners, we had started the machine on cranking for about 15 minutes. IBH local drain and false start local drain were kept open till loading utilizing the available option in the unit. These we did to ensure unexpected condensate accumulation.

The scanner cooling water was throttled to about 20% (ball valve) during start up till the machine was on load. After this, slowly the flame intensity was getting normalized. We increased the load and machine was taken to premix steady state. We were monitoring CO through online monitor during this operation to ensure that combustion was proper. After reaching premix we normalized the cooling water in scanner.

Now the machine is running on premix steady state with all 4 secondary scanners.