Dear All;

The plant where I am, there are A, B, C and D turbo-compressors (GE MS5002 driving BCL CC). The D machine is experiencing Combustion Troubles. I already posted it on this forum and could be reached through this link: http://control.com/thread/1417205798 (Exhaust Temperature Spread MS5002D Gas Turbine).

My concern this time is not about the D machine but it's about the others machines. When investigating the root causes of the combustion trouble on this D machine, I observed that TNH of the others machines are fluctuating and in the same time their TTXSPL jump away from the initials values.

On machine B; TNH is changing value from 99,97% to 100,01% and back to 99,97%. If the time between the changes is greater than 120 seconds than the TTXSPL reading is about 57°C but when this laps is less than the 120 seconds (and it's the case mostly) TTXSPL jump and stay at 168°C.

As said most of time, the fluctuation in TNH occurs at below the 120 seconds and therefore TTXPL is almost equal to 168, 1 °C. <b>My concern is because the TTXSPL is too higher we'll NOT get Combustion Trouble Alarm even if a real trouble occurs.</b>

My questions are: Does 0,04% TNH fluctuation affect the Allowable Spread (TTXSPL)? If yes, what could be the root causes of this fluctuation and how to avoid it? How does MARK V logic works with same fluctuation?

One more detail: When TTXSPL changes value all the exhaust temperature thresholds will change also. See below.<pre>
TNH ------------- 100,01% --- 99,97%
TTXM------------- 521 °C --- 522
TTXSPL----------- 57,3 °C --- 168,1
TTHT------------- 547 °C --- 598
TTHA------------- 544 --- 589
TTLT-------------- 485 --- 416
TTLA-------------- 498 --- 455</pre>

Many thanks In Advance.
[email protected]

 

bkarim55,

TTXSPL is biased by changes in TN<b>R</b>--not TN<b>H</b>, and on a two-shaft machine TNR controls LP turbine shaft speed, not HP turbine shaft speed. So, when whatever is controlling the compressor loading is changing the LP Turbine Speed Reference, the Speedtronic then biases TTXSPL during load changes since fuel flow-rate is changing. There's probably two discrete inputs (contact inputs) or an analog input that is driving TNR up and down periodically which is applying the bias to TTXSPL.

All of this can be observed using Dynamic Rung Display (for a Mark V) and looking at the block that generates TTXSPL.

>One more detail: When TTXSPL changes value all the exhaust
>temperature thresholds will change also. See below.<pre>
>TNH ------------- 100,01% --- 99,97%
>TTXM------------- 521 °C --- 522
>TTXSPL----------- 57,3 °C --- 168,1
>TTHT------------- 547 °C --- 598
>TTHA------------- 544 --- 589
>TTLT-------------- 485 --- 416
>TTLA-------------- 498 --- 455</pre>
I don't understand this question at all. It would be necessary to see the logic/sequencing/application code running in the Speedtronic to understand what's driving the TTHx and TTLx values.

 

Dear CSA,

At first; thanks for the quick response and many thanks for all what you are doing at this forum. Usually when I've to conduct any trouble investigation I begin to Google it first. Now, I have 2 pages all time opened Google and control.com. My way to say thank you very much Control.com.

CSA, When reading your comments and responses through the different topics, the following writing caught my attention: <i>"the Mechanical Department did nothing wrong--because the Speedtronic turbine control panel is the cause of all problems! Just ask any mechanic or Plant Manager--the Speedtroinc is either the cause of every problem, or the solution to every problem. And, just in case you were wondering--the Speedtronic is RARELY the cause of this type of problem."</i> May be because I'm mechanical:-( I have not too bad experience on rotating machine <pre><i>but my knowledge on MarkV/Speedtronic is really limited.</pre></i>

As a head of a mechanicals team I work very closely with control/instrument engineers specially when investigating root causes of any troubleshoot and I never ever had any doubt about the control or monitoring systems used (MarkV, DCS, PLC). The only time where I have been seen that the control system was the root cause of the trouble was on GE EDR Unit and when the system rebooted trouble gone away!

Yes, I'm all time after the control guys to check, re-check and double check again because there are so many instrumented parameters (input, outpout, wirings, connections, etc.) to take in consideration when troubleshooting.

Regarding my experience, I could say 80% of the troubleshoots on rotating machine happened on the instrumentation side (bad signal, instrument faulty, etc.) and more than 80% of the outages (trip, emergency shutdown) are due to a small thing like a looseness of junction box or a faulty reading or instrument.

From the mechanic point of view; Gas turbine or any combustion engine could be assimilated to a rotating piece inside of a statoric piece with lot of piping around these two pieces and this configuration is the same over 50 years ago.

But nowadays, gas turbine or rotating machines are more powerful, delivered in smaller enclosure with huge quantities of instrumentation around them. Even, old units have been upgraded many times.

Getting more power and high Efficiency under all the Law Restrictions we have today make that the work area of a HDGT is very close to the alarms.

Yes! All these instrumentations and controls are very helpful when conducting trouble investigations. But they could be themselves the source of trouble especially when machine or installation is getting older.

I think for theses reasons; Mechanical Department (like me) and management are getting after control & instrument people (like you). And I know, time to time, we are pushing you a little bit hard.

**Back to the actual topic**
Using MarkV Forcing Logic Display; I put all the parameters below on the same screen: All TTXD1x, TTXM, TNL, TNH, TTXSPL, TTHT, TTHA, TTLT, TTLA, CTDA1, CTDA2 and CPD.

Facing the monitor screen; The only change I observed is %TNH (from 99,97% to 100,01% and back to 99,97%). If the time between the changes is greater than 120 seconds nothing will happened and the TTXSPL is about 57°C. But if the time is less than 120 TTXSPL become 168°C and most of time TTXSPL is equal to 168°C.

Next week I'll be on site and I will make another test and at that time I will add TNR on MarkV Logic Display. And with Control engineers I will check what is driving TNR up and down.

On GE training courses I've got on my hands (I'm Home); The spread Protection (2nd type) could be resumed as following:

MarkV Algorithm calculates the allowable TTXSPL and compare it to the actual calculated TTXSPx than the Combustion Monitor Enable (L83SPM) L83SPM = L14HS • L4 • ~L94X • L60SPME under these conditions 1) HP speed above minimum operative (L14HS)
2)Master protective signal (L4)
3)NOT Shutdown in progress (~ L94X)
4)Actual spread more than allowable (L60SPME)

<b>Moderator's note:</b> my editing UI doesn't like some special codes for characters. so not sure what • L4 • ~L94X • actually is.

I can't see THR on this Algorithm. Could you please help?

The thresholds Temperatures are: TTHx = TTXM + (TTXSPL*TTKSPx) and TTLx = TTXM - (TTXSPL*TTKSPx)

Let's take the case where there are 2 adjacent TCs below TTLA; MarkV will generate L30SPA - Combustion trouble Alarm-
My concern, when TTXPL is biased does the algorithm calculate TTLA by the same formula? If it's the case! TC4 and TC5 (see below) could be higher than the new calculated TTLA and <b>NO alarm will be generated??</b>

When TNR is driven up and down and TTXSPL is biased; How the operator will be aware that he lost the Spread Protection?
Temperatures<pre>
|
|
|__________________________ TTHA
| x x x x
|__________________________ TTXM
| x x x xxx
|__________________________ TTLA
| x <b>TC5</b>
| x <b>TC4</b>
|__________________________ TTLA when TTXPL is biased
|
|
|_____________________________ TCs</pre>

Thanks in advance & Kind regards

 

bkarim55,<pre>
| |
Speedtronic | | Turbine,
^ | Inputs/ | Generator
--------|--------| Outputs | and
v | | Auxiliaries
HMI | |
| |</pre>The sketch above shows four distinct areas of the unit and its control system. In the snippet you quoted I was trying to emphasize that the <i>Speedtronic</i> was rarely the cause of an inability to enter into or remain in Premix combustion mode. Most every site that experiences DLN troubles immediately after a maintenance outage always blames the Speedtronic and it's I/O (Inuts/Outputs) as the cause of the inability to transfer into or remain in Premix combustion mode.

The Speedtronic, is in fact, rarely the cause of most turbine operational problems; it's more the inputs/outputs or those parts of the unit control system not physically connected to the Speedtronic (for example, the Liquid Fuel Forwarding Pressure Regulator, or the Liquid Fuel Check Valves, or the Liquid Fuel Purge Air Check Valves, etc.).

Good, knowledgeable and conscientious instrumentation & controls technicians are going to frequently check and tighten terminal board screws that are exposed to temperature changes (caused by starting/stopping of the turbine) and vibration. They know and recognize the importance of taking this small step to preventing nuisance alarms and even trips. They are also going to oversee and review device calibration data sheets to ensure that devices are calibrated and working properly. They are going to supervise and/or review the re-installation of RTDs and T/Cs to ensure they are properly installed and capable of producing accurate outputs. This includes ensuring all manual hand-valves in sensing lines to field devices and instruments are in the proper position before the unit is re-started.

They are involved and engaged in ensuring the "little things" as well as the "big things" are looked after and maintained so that they are not asked at every step and turn to prove the problem is or is NOT the control system. Because once the smoking finger of suspicion/blame is pointed at the "control system" it becomes the I&C technician's responsibility to prove the problem is, or is not, the control system--because, again, it's never the work of the mechanical department that's causing the problem. And a thinking I&C tech is going to be actively working to ensure during every outage, planned or forced, that the little things as well as the big things are looked after.

Can an I&C tech predict the failure of a T/C? No. The failure of a solenoid, or a pressure transmitter? No. But, they can--and do--look after the things that are in their control: manual hand-valve positions; terminal screw tightness; proper installation/re-installation; replacement of devices that don't meet specifications or are intermittent; etc.

So, I will stand on my statement that the <i>Speedtronic</i> is rarely the source of any problem--but you yourself suggested because it has so many wires and devices connected to it it simply and logically (in your mind, anyway) is the source of the majority of problems. Young, or inexperienced I&C tech's may not yet recognize the importance of the "little things" and the "big things"--but they will learn or be constantly responding to calls at all hours of the day, night, weekends, and holidays to the incessant blame that the "control system" is the root of all problems.

I can assure you, that other combustion turbine OEM control systems and control system philosophies are not nearly as well thought out and tested/proven as the GE Speedtroinc turbine control system and GE's heavy duty gas turbine control philosophy and supporting systems and auxiliaries.

Now, to your original question, have a look at the TTXSPVn algorithm ("n" is usually the digit '4'). In particular, look at the section of the block that outputs TTXSPL. It is affected by L83SPMB (the longname description is almost ALWAYS wrong.... it should be 'Exhaust Temperature Spread Monitor Bias Enable'), which is affected the L70R and L70L--which are the signals that drive TNR up and down.

The machine at your site is a two-shaft machine, so there may be some slight differences (such as 120 seconds for the bias to be removed, instead of 60 for most single-shaft machines), and a different bias (200 deg F for most single-shaft machines), but I'm confident in saying that once you return to site and monitor the block/algorithm <i>using Toolbox</i> and observe the L83SPMB input and the TTXSPL output you will see the correlation. It's not TNH--remember, the machine at your site is a two-shaft machine; LP shaft speed is the more critical speed, the one that's affected by commands to increase or decrease output. HP shaft speed is controlled by the variable second-stage nozzle position to hold a particular speed based on machine load. From the sounds of it, the machine at your site is operating at or near rated LP shaft speed/power output which would mean that the HP shaft speed would normally be at or near 100.00%, with some slight variations, as you have noted (0.04%--which ain't much).

I'm still trying to understand the whole TTHx and TTLx purpose. But, if as I have said, TTXSPL is biased out of the way when TNR is changed--again, this is done because the designers recognized that there may be some spread abnormalities during load (fuel flow-rate) changes and so biased TTXSPL out of the way during load (fuel flow-rate) changes)--then even the "usual" spread is "masked" during load changes, and that's by design and acceptable. Once the bias is removed (it's usually ramped down from the initial high value to the actual calculated TTXSPL), then the spreads are again "fully normal."

And, if TTHx and TTLx are calculated using TTXSPL, then, yes, they are going to be "masked" during the time the bias is active and greater than the actual calculated TTXSPL. And, again--that's accepted practice for the "usual" spreads, so one would think it's acceptable for these "auxiliary" spread calculations.

Hope this helps!

If you're worried that real, harmful spreads are being masked by the continual biasing of TTXSPL as a result of the RAISE- and LOWER SPEED/LOAD signals, then it's incumbent on you to advise the Operations Department of the dangers and develop an operating procedure to follow in the event a real, harmful spread is being "masked" by some external signal's fluctuation, intended or unintended.

I have been to several "clever" sites that "discovered" this TTXSPL bias that's applied when RAISE/LOWER signals are toggled and actually had a operating procedure to have the operators periodically click on RAISE then LOWER then RAISE, and so on, all day long, to keep units running that exhaust temperature spreads high enough they would trip the machine. Of course, some of these units crashed when turbine buckets or combustion liners or transition pieces let go when the unit was running, or they found quite a lot of damage when they shut down for a planned maintenance outage. And, why do you think I was there? Because, <b><i>of course,</b></i> the hardware damage was caused by the "Speedtronic"--even though the Speedtronic was being manually "manipulated" to ignore a real, harmful spread. And, once it was pointed out to site management that the operators had been instructed to manually bias the allowable exhaust spread by their supervisors (because, of course, they weren't aware of this when the unit was running), they were furious that the Speedtronic could be manipulated, and they wanted the OEM to pay for the damage their "clever" personnel had exploited because the Speedtroinc was capable of being manipulated.... True stor<b>ies</b>--because it happened more than thrice!

Isn't this fun?

 

Dear CSA

> Isn't this fun?

I discovered how to bias TTXSPL so I'm the "<b>clever personnel</b>" (just a joke!).

When onsite, first thing to do is to inform the Ops Department about what's happening on their machines and I will share with them the true story of the "clever" personnel.

And I'll turn back to you with the findings.

Thanks again & See you soon
Karim