We have a Frame 5 DLN machine running on fuel gas with MarkV control and have a problem that all four flame intensity signals fluctuate in all modes. It is higher when in ELL or premix modes but the fluctuations are relatively constant and anywhere from 10 to 50 counts across the detectors. These are new detectors fitted after an overhaul (looking back historically the old detectors did this also). All wiring and earthing has been checked. All fuel valves are stable, fuel gas inlet temperature is ok and all other turbine parameters are ok with no indication of any combustion problems. It has been running like this for three months but due to the fluctuations we keep dropping out of premix mode. The fluctuations are causing the flame intensity signal to catch the thresholds that were set during the DLN tuning. They could be raised but our turbine man is reluctant to change them as the same threshold is used for flame on during primary mode. Any suggestions?
> We have a Frame 5 DLN machine running on fuel gas with MarkV control
There must be adequate safety margins to accurately measure the flame temperature. Many sources of fluctuation can exist which can be eliminated;
1) aperture size from Field of View (FOV) , distance,
2) electrical noise when digitized from AC power on sensor DC current current.
3) electrical integration time or averaging circuit.
These apply to all flame sensors.
Without a close examination of the issues, it is hard to pin-point your issue. Can you answer the questions above? Do you know FOV, distance of sensor? Sensor uA and noise level? length of sensor cable? shielded? ferrite sleeve? There are many other improvements that can be made. Good margin on thresholds is important for reliability and longer periods between maintenance.
Blowing by the exhaust spread question, the FOV and reflectivity issues are pretty well known in the DLN-I community. Alignment of flame detectors and the length of sight tubes is also an issue. Using non-OEM nozzles and liners can also cause issues.
Fluctuating gas fuel supply pressure and fluctuating frequency (that is, trying to use a DLN-I unit in Isoch mode or on a small island with fluctuating loads) can contribute to problems.
We don't have enough information to be of much help.
Thanks for the replies. To address your points
1)aperture size from Field of View (FOV), distance - I can't find a dimensional drawing so have asked GENP rep to provide.
2) electrical noise when digitized from AC power on sensor DC current current - Our model is 330 Vdc. Electrical noise where my first thoughts, the fluctuations are constant and regular. I plan to connect a flame detector directly to one channel in the Mark V panel to see how that looks.
3) electrical integration time or averaging circuit - I'm not clear what you mean here? It is the original Mark V configuration counting pulses from the detector. Detection threshold values are flame detector pulses during a 1/16 second sample period. There is no integration time or averaging circuit?
CSA, when you say FOV and reflectivity issues are pretty well known in the DLN-I community are you indicating that these two issues could be a cause of the fluctuations? Is UV reflection off metallic surfaces a common issue that could cause the fluctuations?
It is all OEM equipment fitted by GENP during a recent shutdown. We are awaiting a further response from GENP.
My thoughts were that in order to accommodate both the requirements of flame out during start up and flashback in the primary combustion chamber during premix mode, the threshold settings needs to be between the highest flame count seen in premix mode and the lowest flame count seen in primary mode during start up. Do you see any issues in doing this? GENP say they determined the thresholds during DLN tuning by recording the output of each flame detector with a fast scan rate application, in all the operating modes from firing up to premix. The thresholds were then calculated, based on a statistical study of the values shown in the recordings. Does this method give average values for flame intensity over time or would that method account for the peaks of the fluctuations?
Once again thanks for taking the time to answer my post.
Yes, reflectivity can definitely cause problems like the one you are describing.
I have never heard of "tuning" the flame detection parameters; that seems like asking for trouble when Geiger-Mueller (Honeywell) style flame detectors are used on a Mark V on a unit with DLN combustors.
Please write back to let us know what GENP replies/advises.
Another thought came to mind. Some Mark V panels used a low-level setting for primary mode and a high-level setting for Premix mode. This was done differently depending on the vintage of Mark V panel. Sometimes it was done in the CSP and other times (later, newer panels) it was done in firmware by PROM and I/O Configurator settings.
How old is the Mark V at your site? Do you know the panel "type"? Sometimes when you run CARD_ID.EXE it will identify the type ("A" or "B",etc.).
Can you tell from the CSP and/or the I/O Configurator if the panel has the low/high level capability?
You may not have it now, but perhaps GENP could provide it?
Also, the TCEB card and the PTBA card are both common to the flame detector circuit. It's not really clear this "fluctuation" stuff. You could try replacing those cards one at a time to see if it had any positive effect. If that didn't I would suggest trying to replace the TCEA in Loc. 1. If that didn't improve the fluctuation you could swap the new card to Loc. 3, and then to Loc. 5. You would then have replaced all of the cards in the flame detector circuit in an effort to see if any one of them was causing the problem or contributing to the problem. A shutdown would be required for every card change/swap....
Write back to let us know what you find!
Thanks again for the info. Card changes are the next step but that will not be for some time as a shutdown now is not an option. Still waiting on further feed back from GENP.
An update and further question.
OEM has proposed new flame thresholds based on current operating data.
Thresholds were originally determined when ambient temperatures were around 40 Deg C. Now it is winter and we are -10 to -20 Deg C. I can't see any connection but could this difference in ambient temperature have any effect?