We are operating GE Frame 6581B gas turbine on natural gas, liquid fuel as emergency fuel, TMR, simple cycle operation.
Recently our control system migrated Mark V to Mark VIe. After migration, unit was started several times without any problems.
Last startup on gas fuel unit tripped on STOP/SPEED RATIO VALVE POS ERRROR TRIP (L3GRVT), just after firing, all the flame was established.
Logic is if SRV Position Feedback FSGR is greater than 5% for 5 Sec before the warmup period unit will trip.
We observed that intermediate valve pressure FPG2 is increased to 19 Barg (261 PSIG)
TNH- Turbine speed %
FPG2- Intermediate Valve Pressure Barg
FAGR_NVR- Servo current from <R>
TNH-----FPG2-------FAGR_NVR--- FAGR_NVT--- FAGR_NVS
After several hours of investigation we found that one <R> core FPG2 transmitter not reading the pressure, it always reading zero. (We have 3 intermediate valve pressure transmitters 96FG-2A/2B/2C and one upstream pressure transmitters FGUP)
We replaced the faulty transmitter and unit started successfully.
96FG-2A/2B/2C terminated in TBQB <R> and from there it is connected to MVRA card ( in Mark V it was TCQA)
My question is if one transmitter failure unit will trip?
Then where is the redundancy? Voting?
Or something wrongly configured during the Mark V to Mark VIe migration.
Migration we used all the old terminal boards, TCQC cards.
Kindly give your valued comments.
Looking back at some of your previous posts, you have some very serious misunderstandings of redundancy. Redundancy can be accomplished in many ways, not always in the same way--and GE, to be sure, has used about all of the various ways to implement redundancy.
In the Mark IV and Mark V redundant P2 (intervalve--the area between the SRV and GCV) pressure transmitters are connected to a single terminal board on <R> and then, via ribbon cables, are individually connected to <R>, <S> and <T>. Each control processor uses what it believes to be the actual P2 (intervalve) pressure from the single transducer connected to it (96FG-2A is connected to <R>; 96FG-2B is connected to <S>; and, 96FG-2C is connected to <T>).
There are three (3) control processors, and there is the redundancy. Each control processor is making its own determination of what the actual P2 (intervalve) pressure is, and is trying to make the actual P2 (intervalve) pressure equal to the P2 (intervalve) pressure reference. Each processor gets its own speed pick-up (77NH-1 is connected to the QTBA on <R>; 77NH-2 is connected to the QTBA on <S>; and, 77NH-3 is connected to the QTBA on <T>), and each processor makes its determination of what the P2 (intervalve) pressure reference should be.
If all three speed pick-ups are giving the same indications to each control processor, then FPRG will be the same for all three processors. If one of the speed pick-ups is different from the other two, then FPRG for that processor will be different from the other two.
The same thing happens for the servo output current. If one processor sees a different P2 (intervalve) pressure than the other two processors then its servo output current will be different from the other two. BUT, in this case the voting is done at the servo valve. The amount of oil flowing through the servo valve to and from the hydraulic actuator is the result of the sum of the three servo currents in the servo valve.
That's why you see <R>'s servo current going so negative (trying to increase the P2 pressure from its P2 pressure transmitter, 96FG-2A), and the servo currents from <S> and <T> going more positive trying to limit the P2 pressure seen from their respective P2 pressure transmitters (96FG-2B and 96FG-2C). In the servo valve, the sum of the three currents produce torque which directs the spray of hydraulic oil to the ends of the spool valve which controls the flow of hydraulic oil to/from the hydraulic actuator.
Some signals are voted in Mark V/Ve software; some signals are voted in Mark V/Ve hardware (in each individual controller); and some signals are voted in the field--the servo output currents being the signals voted in the field. Redundancy can take many forms, as can voting.
A turbine control system is NOT JUST the printed circuit cards and LEDs and fuses and wiring in the turbine control panel--it includes all of the devices which are connected to the turbine control panel, and even some devices which are NOT connected to the turbine control panel (such as liquid fuel check valves, and purge air check valves, etc.). And, again, redundancy can take many forms and methods, as can voting.
As for your question about whether one failed P2 (intervalve) pressure transmitter could cause a unit to trip--no; that should not happen. UNLESS, one of the servo coils has the servo current being applied incorrectly (wrong servo polarity). Then, it's very likely that a trip WILL occur under the circumstances you describe. Or, if one coil of the servo is experiencing problems such as bad/shorted turns or wiring, etc. That's about the only way I could envision the data you provided occurring--something is amiss with the servo current in the <S> or <T> servo coil, or there's something wrong with the servo (it's sticking or failing).
Remember--it's NOT necessary to calibrate LVDTs when replacing a servo valve. (Changing a servo valve does NOT change the stroke of the device, and AutoCalibration does NOT affect anything other than LVDT calibration--which is not affected by changing a servo valve.)
BUT, it IS necessary to ensure the polarity of the current being applied to each individual servo coil is correct. The turbine will start and run if the current being applied to one of the servo coils is incorrect--but it will likely have odd, intermittent problems (including Diagnostic Alarms) and if one of the two good coils experiences a problem with its servo current then the turbine will DEFINITELY trip.
BUT, I would suspect that <R> processor was annunciating a Diagnostic Alarm to the effect that there was something wrong with the 96FG-2A pressure transmitter input prior to the trip.... Diagnostic Alarms can be an indicator of a potential trip. Individual Diagnostic Alarms can NOT cause a unit trip, but combinations of Diagnostic Alarms can certainly indicate what caused a unit trip; it's just that most sites don't pay much, if any, attention to Diagnostic Alarms.
Lastly, I would expect that if P2 pressure increased to 261 psig during firing/warm-up that the unit would likely trip on exhaust overtemperature, and very quickly. Or, at a minimum there would have been an exhaust overtemperature alarm.
Please write back to let us know what you find!
It could also be that the polarity of the servo current from <R> being applied to its servo coil is incorrect.
Does your detailed reply apply to MARK VI TMR?
Generally what input/output signals are voted and what are not voted?
In general, the GE turbine control design philosophies of redundancy and voting are built around the idea of criticality. If a signal is deemed critical, then it is usually redundant and voted.
Beginning with Mark V, the concept of SIFT (Software-Implemented Fault Tolerance) was introduced to the Speedtronic turbine control product line for many signals, but not for all of them. Under SIFT, voting was done by each control processor using the input values of all three control processors. Note that SIFT generally refers to input signals--but, again, not all of them. Some "really" critical signals, such as turbine shaft speed and P2 (intervalve) pressure were not voted using SIFT. Each control processor used its own input value for these two signals, and voting was accomplished in different manners. In the case of P2 (intervalve) pressure (which, interestingly enough, is a function of turbine shaft speed), the voting of the signals is done at the output--the servo valve.
Shortly after the Mark VI was introduced, it was decided to handle P2 (intervalve) pressure differently than it had been handled in the past. In Mark IV and Mark V, each control processor used its own values of turbine shaft speed and P2 pressure feedback to determine what its output to the servo-valve should be. But, in the Mark VI, with faster scan rates (Mark V was generally 8 Hz, or 0.125 sec; Mark VI is generally 25 Hz, or 0.040 sec) the three P2 pressure transducer inputs (if three were used--and three were not always used for all application) were voted in software using SIFT and the three control processors use the voted valve of P2 pressure feedback in application code--not at the firmware level as was done in the Mark V--to determine what the SRV servo-valve output should be.
But, this is the only major difference that I'm aware of with respect to how redundancy is handled between Mark V, and Mark VI & Mark VIe. There have been some Mark VI and Mark VIe turbine control panels which did some unusual things with turbine shaft speed (TNH; TNH1) but they seem to have been the exception rather than the rule (at least in my experience).
The idea of criticality ("criticalness") is very subjective--that is, GE's idea of whether or not a signal or function is critical is not always the same as an owner/operator thinks when some problems occur. For example, starting means I/O is generally deemed by GE to be not critical. The presumption is that while it's necessary to get the unit started and producing torque and/or electricity, it's not necessary for normal, running operation of the unit. So, there isn't usually much redundancy for starting means I/O and it's usually "simplex" (as opposed to SIMPLEX) I/O--meaning that it is commonly connected to a single processor instead of being "fanned" out to all three processors. Plant owners/operators, however, when there's a problem with starting the unit, believe the starting means I/O is DEFINITELY critical and believe GE should change their thinking (which doesn't happen very often).
So, redundancy and voting are related to the OEM's definition of criticality, which doesn't always match everyone else's definition of criticality. Having said that, the availability and reliability rates of properly-maintained and -operated GE-design heavy duty gas turbines is generally significantly higher than that of many other manufacturer's turbines (which can be at times more complicated, or lack even minimal redundancy and/or voting).
They have done a very good job of considering what is and isn't critical, and if owners/operators properly maintain and operate their units the reliability and availability rates of their units are very high. Proper operation and maintenance includes Alarm Management (Process- and Diagnostic Alarms), as well as mechanical maintenance. Things break unexpectedly, but letting them go until they break without a plan for quick recovery, is not proper maintenance or operation.
Hope this helps!
>My question is if one transmitter failure unit will trip?
>Then where is the redundancy? Voting?
>Or something wrongly configured during the Mark V to Mark
>Migration we used all the old terminal boards, TCQC cards.
I suspect the other cores [S,T] readings were not spot on, differences in mearsurment exist (.03~.8)
And with one core down, logic interpret misoperation
There is a tuning for this, but I can't expound.