Hi All,

We have an intermittent problem on our mark 5 speedtronic control of a frame 6b turbine. Sometimes the field swich doesn't close and the machine sits at synchronous speed with no volts. We can manually close the field switch, volts come on and it sync's and closes the gen cb.

We looked into the wiring and software and couldn't find any fault. Changed the interposing relay between the Mark 5 output and the field swich contactor. The field flashing is the first contactor where 125vdc is put on to get it started then it changes over to a generated supply. We're not getting the field flashing. Once the field swich is closed manually, the change over from dc field flashing to the generated supply is ok.

We also changed the manual field switch open pushbutton which has a n/c contact which is in circuit for the auto field switch as there was some resistance across it. Seems to be a timing/software problem hence the intermittent nature of it. Has anyone else had this problem? It has happened on both of our machines hence why why I think it's a s/w glitch rather than a hardware/wiring fault. software only needs to see master protective healthy and correct speed. Any comments or proposals to troubleshoot are welcome.

Thanks,
David

 

davidmcc,

>We looked into the wiring and software and couldn't find any
>fault.

This is the crux of responding to your posts. You say you've troubleshot, but you won't say how and you won't say what the results were--only that you were satisfied that there is no issue with [this] or [that]. And, yet the problem still remains, no?

How can we help you if you won't tell us what you did, how you did it, and what the results were?

You say you "manually" close the field "switch" (I believe it's probably the field flashing contactor, a DC contactor). HOW do you manually close the switch? Do you force the output from the Mark V? Do you jumper some permissive?

Which "interposing" relay have you replaced? Was it the one on the TCRA card in the <QDn> core? Or one between the Mark V output and the field flashing contactor coil?

Have you used the Short Term Trend feature of the Mark Vs or one of the VIEWn tools during a start to see if the field flashing discrete output is being energized during starting when the unit reaches the speed setpoint? Does the field flashing discrete output only pick up for a brief period of time? (Most of the Mark Vs I have worked on with field flashing the field flashing discrete output picks up and stays picked up for some period of time, and the reverse bias blocking diode prevents backfeeding the battery once the field voltage builds up and until the field flashing discrete output is then dropped out.) Have you monitored the discrete output of the Mark V to be sure it's doing what it should, when it should--using trending or a voltmeter?

Tell us what you did during your testing, how you did it, and what the results were. Don't just say, we looked into it and it's all good--because, ..., well, it's not. And you're writing here for help and we're not there with you and we don't know what you did, or how you did it, or what the results were.

This is a technical forum, and this is a technical problem. You need to provide technical details if you want proper (technical) answers.

Finally, how long have the two units been running with the Mark V that there have been NO problems with the field flashing? Has this been happening since commissioning, or just recently? If it's just recently, what makes you so sure it's a software problem? Does software degrade with age at your site?

Best of luck with your problem. Write back with details when it's convenient for you and you can provide few details of how the problem was resolved.

 

Hi CSA,

Thanks for your prompt reply. I was really asking if the community knows of this as a common problem, since it happened on both of our machines. The machines were commissioned on our site in 2009. This problem fist appeared on one machine some months after commissioning, then disappeared for several months, then came back on the same machine, so in the past it was a rare occurrance on just one machine. A few years ago it happened on the other machine. Now it can happen on either machine maybe once in every 5 starts. The excitation panel has pushbuttons for excitation on and off. After excitation fails in auto, excitation can be brought on manually by pressing the close pushbutton and the sequence ie. field flashing, then change over etc. works, volts come up, sync and breaker close all work as normal. The fault seems to be on the initial command generated in the mark 5, when the machine reaches the correct speed, not getting through to the field flashing contactor.

The troubleshooting consisted of checking the field flashing contactor closing circuit. We found some resistance across the field switch off pushbutton which is in circuit even for auto, is a normally closed contact. 70 ohms was measured, but it changed every time it was pressed. Pushbutton contact changed. We checked the interposing relay in the excitation panel, a 125vdc relay with diode. Checked out ok, but the coil was a bit discoloured, changed it in case it was weak when pulling in. We monitored the programme in online mode and seen the output highlight come on when the auto worked. We never witnessed an auto fail. We could use the view2 log file to continuosly log the excitation on command and check if it was issued during an auto excitation fail. That could take months to catch the data, but it would determine if a software or hardware cause.

When I said I suspected a software problem, I know the software doesn't degrade over time, but I thought a software flaw could mean success or failure depending on the state of other variables. Maybe I'm predisposed to suspect this as we found a previous flaw in the software where the machine failed to sync when the ac lube oil pump happened to be running for heating. It should have cut out when the machine was at speed, but the s/w didn't include for that. It always sync'd when re-tried because the oil was up to temp, and the ac pump was off. Hence it was an intermittent/random fault: if the ac lube pump was in an off cycle it sync'd or if in an on cycle sync failed. The pump off period is longer than the on period so it usually worked. This excitation fault is similar in frequency of occurrance. The software was corrected and sync never failed again. This excitation fault may have a similar cause.

I think we'll set the view2 file to log the excitation on command and wait for a fail and take it from there.

Thanks for you input.

Regards,
David

 

davidmcc,

If the Mark V software CSP (Control Sequence Program) is not energizing the field flashing discrete output, then the VIEWn tools or Short Term Trending would catch it. All you need to do is start the program every time the unit is started--not too difficult. Oh, and remember to stop it after synchronization. Also, not too difficult.

That little "bug" you think you found in the CSP--that's not a bug. That's a failure to understand basic GE heavy duty gas turbine controls philosophy AND to respond to alarms. Because there was an alarm telling you the Aux. L.O. Pump was running when it should not have been. And, anyone looking at the CSP would have realized the unit did not achieve 'Complete Sequence' (L3 equal to logic "1"). GE heavy duty gas turbine control philosophy is that all Auxiliary Pumps (and the Emer. L.O. Pump) must all be off when reaching FSNL before an Auto Synchronization can be initiated and completed. If the L.O. Tank Heater was running then the L.O. temperature was already below approximately 80 deg F, and when that happens the vibration during a start and acceleration to FSNL is usually higher than normal.

So, there's likely a LOT of things you haven't told about that particular situation, some of which you might not even have been aware of. But, a "bug" or fault you did not find. Think about it this way: If the unit is started and synchronized with the Aux. L.O. Pump running and then Main L.O. Pump is not developing pressure, and then the Aux. L.O. Pump pressure decays, the unit is going to trip and come down on the Emer. L.O. Pump. Does that sound like something which should be allowed to happen?

People are ALWAYS complaining about GE "logic" without understanding or even trying to understand why things are done by GE the way they are done. GE heavy duty gas turbines are the most reliable in the world--and they didn't get that way with "bugs" and bad sequencing. That bit of code has been around for decades--literally. And, yes, it does catch some people unaware. But, then, documentation about GE heavy duty gas turbine operation and control philosophy--and that is a serious failing of GE, and of its licensed packagers, too.

So, just about whenever I encounter intermittent problems like this, the first thing I always do is check wiring all along the circuit, from the Mark V, to the end, and back. Sometimes terminations can be tight, but the crimp-on connectors are loose. As can the terminations on the Mark V high-density terminal boards. Recently, I was at a Mark V site where they were complaining of intermittent problems, and it was discovered that someone had tried to put the wires on the wrong side of the sliding clips by turning the screws in the opposite direction to hold the wires in place. We found about 70+ wires which were loose and incorrectly terminated. And, a LOT of the problems which had been complained about for 20 years suddenly stopped.

Everyone wants to blame the Mark*--it MUST be that blasted Mark*!!! It has too many wires and flashing lights and it just has to be the Mark*!. I can tell you from decades of personal experience--it's much easier to check everything outside the Mark* than it is to prove an intermittent problem is caused by the Mark*.

 

CSA,

Yes, there certainly is a lot that could be learned about GE philosophies at our business. But given that there is no dedicated resource for these machines, only what is left over after our main plant is attended to, it is a testament to the design that they remain fairly reliable. The problem with starting/stopping the logging is that they are dispatched based on system demand (peaking), usually out of hours so we would have to leave the batch file running for long periods to catch it. We could delete the 'empty' text files every day to stop then disc filling up. The ac l.o. pump running /fail to sync issue was resolved by a software mod by the gas turbine specialist who commissioned the machines; so I'm intrigued to check what mod they did, maybe they overlooked the root cause i.e. why the ac l.o. pump was still running at fsnl...

The reality is that these machines are run on minimal resources. There is very little time available to resolve issues because issues on our main plant, of which there are many, take priority.

Thanks for your guidance. I'll update when I have some data on excitation failing i.e. if the Mark 5 issued the command or not when it failed.

Regards,
David