We are facing an issue with exhaust TC in GE Frame 9E Gas Turbine unit with Mark 6e control system as follows:

When unit is running, TTXD-6 reading fluctuation from normal value to minus value and accordingly Spread is increasing. Once we go for shutdown, TC value is showing steady value but relatively less than other TC (around 15 to 20 C less ). I understand that it looks like loose connection issue. But the confusion is happening due to the following :

During shutdown we tried to investigate the issue, we opened the TC cable from Field Junction box, measured its oh-mic value and mv and found normal. (Please remember that it was showing 20 C less on Mark 6e display). Once we connect the cables back in the Junction Box, it is reading normally and health like other thermocouples. The same phenomena happened 2 times in the same unit.
Still we didn't get the opportunity to go inside the load tunnel and inspect the TC physically as it requires zero speed to access it. I just wanted to know if someone faced the same phenomena earlier or have similar experience to advise.

Thanks in advance

 

This is one of the common T/C failure modes; they sometimes intermittently go open circuit, especially when heat is available at the T/C tip.

If you can't shut the machine down and wait for the shaft to go to zero speed, there's not a lot which can be done. Some sites try jumpering thermocouples (it should be from the T/C on one side or the other of the intermittent/failed T/C!!!) and that works until the T/C can be withdrawn and replaced with a new one. (And I'm speaking of the T/C which is physically next to the intermittent/failed T/C--not the one next to it on the T/C terminal board which may or may not be the physically adjacent T/C.

When did this problem start? After a maintenance outage? Which interval was it? Were the T/Cs removed and replaced? Was there work done in the Load Compartment during the maintenance outage?

 

This is one of the common T/C failure modes; they sometimes intermittently go open circuit, especially when heat is available at the T/C tip.

If you can't shut the machine down and wait for the shaft to go to zero speed, there's not a lot which can be done. Some sites try jumpering thermocouples (it should be from the T/C on one side or the other of the intermittent/failed T/C!!!) and that works until the T/C can be withdrawn and replaced with a new one. (And I'm speaking of the T/C which is physically next to the intermittent/failed T/C--not the one next to it on the T/C terminal board which may or may not be the physically adjacent T/C.

When did this problem start? After a maintenance outage? Which interval was it? Were the T/Cs removed and replaced? Was there work done in the Load Compartment during the maintenance outage?

We will get a shutdown soon with standstill condition. We will have the access to replace the TC.

The unit was on outage before one month but is was started several times during the last period without problems. The TC was not replaced and there is no work done on load tunnel>

So, you recommend the TC with a new one. agreed.

But what is the explanation of this phenomena: TC showing 20 C less at shutdown condition, When you open it from JB and then fix it back, it is reading correctly. I mean electrically or theoretically, how it can be explained.

Thanks for your support.

 

1. When a thermocouple reads low, it can be due to cold junction compensation not being enabled or being measured for that circuit/input/T/C.

If the cold junction measurement is made locally in a remote terminal junction box, where copper wire is then used for the final connection to the thermocouple inputs, then it could be proposed that mechanical activity in the junction box, like jiggling wires where there is a loose connection, could have re-enabled a faulted, open circuit cold junction measurement.

But is this the case? Loss of cold junction measurement (a resistance measurement, usually a thermistor where the loss of a closed path for the excitation current is easily detected) is usually an alarmed event. Also, most installations only use one cold junction measurement for multiple remote terminal thermocouples, not for just a single thermocouple (nor multiple CJ measurements - one for each thermocouple), so a faulted cold junction would affect the readings of other temperature points, which does not seem to be the case here.

I have seen single loop controllers with the cold junction sensor externally mounted on the input terminals (a very effective CJ measurement) where a broken CJ wire, a bad internal CJ wire connection on the CJ sensor could conceivably fault a CJ measurement, but again, I doubt that describes the remote terminal junction box situation above.

2. Another reason for a thermocouple to read low is hot junction drift due to chemical pollution of the thermocouple elements. The thermocouple still generates an EMF, but the EMF bears no relationship to the standarized tables because the chemistry of the thermocouple elements is not the same as whatever type thermocouple is in use.

Physical inspection might or might not reveal a chemically polluted, drifting thermocouple (Type K run in a reducing atmosphere (800-1260 Deg C) show 'green rot' - green is oxidized chromium in the NiAl leg.

Measuring the resistance and comparing that to the resistance of a new thermocouple of the same type, gauge and lead wire length will show an increase in resistance of a drifted thermocouple. Several manufacturers use that fact to detect thermocouple drift.

There is NO compensation for thermocouple drift because there are no standardized tables for polluted chemical alloys. Drifted thermocouples need to be replaced.

3. I have seen incorrect thermocouple readings due to corrosion of the iron leg in Type J thermocouple. Cleaning the leg to bare wire would fix the situation temporarily, but moisture corrodes iron, does it not? Are these Type J thermocouples where re-connecting to screw terminals would clean the corrosion or rust off the iron leg?

 

there was an interesting case with exhaust thermocouples, the problem in the mismatch of readings was due to a draft of cold air through a slot in the bottom of the cabinet with Mark6 boards

 

Cold junction compensation is usually done at the I/O Pack for T/Cs in Mark* VIe turbine control systems. My guess would be the device being used to measure the T/C either didn’t have CJ compensation or hadn’t been allowed to acclimate to the temperature in the area where the measurement was made.

MANY people mistakenly believe the room or compartment where the Mark* is located should be very cool, even cold. That’s entirely wrong—the primary purpose of air conditioning in the location where the Mark* is located is humidity control, followed by temperature control. Dust and humidity are as bad or worse for electronics than intense heat.

The heat produced by individual Mark* VIe I/O Packs can be pretty high. Many TMR I/O Packs are located vertically with very little space between them on I/O terminal boards. AND the I/O terminal boards are often located in neat vertical columns—looks great, makes wiring easier, but the temperature at the top of the column can be much higher than in the middle or at the bottom. Most Mark* VIe panels are cooled by convection which draws air from outside the panel through vents/louvers and the air rises as it is heated cooling the I/O Packs. If the air temperature outside the panel is very cool or cold then the I/O Packs at the bottom of the columns will get more cooling than those at the top which will result in different CJ compensation temperatures in the panel from bottom to top.

Finally, if you’re looking at the Exhaust T/C display to readT/C temperatures when testing like thisyou should remember: Exhaust T/Cs are biased from their “raw” (CJ compensated) value before being used in the temp control and protection algorithms, so the values you see on the display don’t match the incoming, CJ compensated values. You should be using the input value reported by the I/O Pack, not the one on the Exhaust T/C display when comparing T/C as you have been doing.

Finally, if you’re using meter leads made of copper connected to tin alligator clips to measure T/C output temperatures you could be introducing unwanted cold junctions in the measurement. Also, a good multimeter is always better for this kind of troubleshooting than an inexpensive one; the “acceptable” range/tolerance of some digital meters these days is pretty outrageous.

Replace that T/C. Ensure the chromel wires are connected with chromel wires all along the circuit from the T/C to the Mark* VIe I/O terminal board (meaning the alumel wires are also connected to each other all along the circuit—T/C wiring must be connected properly to prevent unwanted cold junctions from affecting readings).

And write back to let us know what you find.

 

> when testing like this you should remember: Exhaust T/Cs are biased from their “raw” (CJ compensated) value before being used in the temp control and protection algorithms, so the values you see on the display don’t match the incoming, CJ compensated values (post #6)

Sounds like a major consideration in checking the reported temperatures.

 

Thanks to all for their feedback.

Just wanted to update you, we had a shutdown and inspected the thermocouple. we found thermocouple cable is partially damaged at one point. we replaced it, started the unit and it is ok now.

Again many thanks for your support.

 

Thank you for the feedback!

This is a very good example of checking the sensors and wiring first when troubleshooting a problem with inputs. While many sensors are very robust, the wiring is not always so robust. Many older GE-design heavy duty gas turbines used thermocouples with very long stainless steel sheaths which had to be "coiled" in order to prevent them from becoming entangled in the Load Compartment or from rubbing against other metal in the compartment which would usually lead to problems like the one in this thread. Many technicians intensely dislike having to remove the clips holding the existing sheaths to stationary brackets, and in the process drop the pieces of the clips on the floor (which usually has oil on it, and has a lot of really difficult to access locations, and so they don't go "bilge diving" to recover them and don't have spares for them. As a result, the sheaths are not properly secured during installation and are free to wave around and rub against other metal components, wearing a hole in the sheath and eventually through the conductor insulation underneath. And, that's when the intermittent problems start.

And for thermocouples, if EITHER conductor (chromel OR alumel in the case of Type K--which is the usual T/C type for exhaust T/Cs on GE-design heavy duty gas turbines) touches ground anywhere along it's length that can introduce a single cold junction which can cause the Mark* to misinterpret the voltage and display inaccurate readings. And, this unintended cold junction can introduce varying biases depending on the temperature in the area where the ground has occurred (and the temperature in the Load Compartment can change A LOT from start-up to Base Load and back to zero speed!).

The Mark* is a fairly robust control system--it has been designed specifically for turbine control applications and the types of environments the sensors and outputs can be exposed to and located in. The Mark* is only as good as its inputs--garbage in equals garbage out, as they say for control systems, all control systems.

Anyway, thanks again for the feedback!