A couple years ago we performed an upgrade that required some changes to our curve control constants. I made some curves to help visualize this. But when I view TTRXB, it appears that it is higher than the curve by 30F at any given CPD. It's easiest to tell when CPD is less than TTK0_C since TTK0_I = 1055 and TTRXB = 1085. What causes this difference. I see that TTKRX2 is 30F. I'm looking at 28B but I can't connect the two. What am I missing? Frame 6B/ Mk4

 

cactus_chris, eh?

Contrary to popular belief, TTRXB does not mean 'Turbine Temperature Reference-Exhaust, Base (Load). It means 'Turbine Temperature Reference-Exhaust, Biased'--where a bias can be added to the lesser of TTRXP or TTRXS--when a permissive is set to a logic "1". It's typically only used on two-shaft, mechanical drive turbines--not on single-shaft generator drive turbines.

I don't have a copy of Sh. 28B of a Mark IV Speedtronic elementary (the BEST turbine control drawing GE ever produced!), but you should be able to see the logic signal which controls a NO contact to add the bias (I think it might have been TTKRX2--but again, I don't have an elementary to look at right now) to TTRX (the lesser of TTRXP or TTRXS) to become TTRXB.

If the turbine at the site is a single-shaft generator drive unit and a bias is being added to TTRX and it was that way from "the factory" (since commissioning, in other words) it would seem unusual--very unusual, indeed. And if the unit has been running like that for any length of time--good luck trying to change it. No one gives up that kind of power without a fight.

However, since you conspicuously didn't say what the upgrade was that was performed it could well be that enabling the bias and setting the bias value to 30 deg F was part of the upgrade. You should be able to look in the original Control Specification, Sect. 03, to find out the bias was originally set and if it was originally enabled.

Lastly, I think there was, for a time, a version of this logic that added a bias to TTRXB but used a logic permissive to select TTRX or TTRXB to be the actual exhaust temperature reference value. So, unless the "selector" was enabled (it was usually a single logic driving a NO contact and a NC contact to choose which signal to pass to the algorithm). So, even if I had a Sh. 28B it might have the same version as the one at the site that you're looking at.

Hope this helps!

 

Here's my understanding from looking at 28B and now 28K,

FROM 28B

TTR MIN SEL takes the lowest value from TTRXS (FSR TEMP REF), TTRXP (CPD TEMP REF), AND TTK0_1 (ISOTHERMAL) and passes it to the TTRX through a confusing MED SEL that also gets the values TTKRXR1 and TTKRXR2 (which look to only be used initially to "ramp" up to a value through the Z-1.

TTRX is then added to a value CSRGVTXB to produce TTRXB.

FROM 28K

CSRGVTXB is a MIN SEL of CEKGVBMX (30degF) or a its Z-1 feedback value added to a value that is the MAX SEL of -1 X CSKGVBR (2degF) or IF L83GVBP IS "1" (which it is):

(CSKGVBG X (CSKGVMAX - CSRGV)) - 2

Which, using our constants is:

(3F/DGA X (86DGA - CSRGV)) - 2F

So at maximum IGV position, it looks like the TTRX is biased by CSRGVTXB, which is usually about 30degF. My elementary does not have a description for any of these 28K constants or variables. I can see by the letters they chose that it is some sort of igv/base load relationship, but my usual means of using the description to give me a better idea of what or why they did is not possible. Any ideas?

 

I wanted to add that L83GVBP is "1" because L83GVMAN is "0" and L14HS is "1". So it looks like that is true most any time, and I'm pretty sure that I am looking at an original elementary drawing for all of this.

 

cactus_chris,

I'm sort of following what you're trying to describe, and I can say I don't recall ever running across the Control Constant CEKGVBMX.

It probably goes without saying but if the actual exhaust temperature is above the calculated curve, whether it's below the corner (_C) or beyond the corner then the unit is "overfiring."

Without understanding if this was done during commissioning or during the recent modification it's really difficult to say what the effects are--except that the unit has been experiencing the benefits of over-firing (extra power). If the modification was done to the hot gas path (turbine nozzles, turbine buckets) then it's likely that they can withstand the extra firing temperature. Have you noticed any increased degradation in hot gas path parts since the modification (presuming it was the turbine nozzles/buckets, or honeycomb shrouds or brush seals or something similar done to the hot gas path section)?

If this was done at the time of the modification, then it was likely the simplest way to increase firing temperature--but adding a bias. The issue is that the isothermal (_I) constant is chosen not to protect turbine nozzles and -buckets, it's chosen to protect the exhaust diffuser and exhaust components, sometimes including a HRSG (Heat Recovery Steam Generator). So, raising the exhaust temperature by 30 deg F when operating below the corner is putting extra heat on the exhaust diffuser and exhaust components. And, while the effects of that may take a while to become apparent, it's probably worth a look at the earliest convenient time to see if there are any adverse effects on the exhaust diffuser and -components.

Lots of people don't know how to shift curves (they're just y=mx+b, with a negative slope and a maximum limit in this case). If there are multiple curves for a particular fuel, then the intersections also have to be re-calculated to ensure they are continuous with no large offsets. That may be why someone chose to just add a bias in this case, to avoid having to calculate new offsets and intersections to shift the curve (beyond the isothermal) to get more power. And, didn't realize--or took a gamble--that the exhaust diffuser and -components would withstand the extra temperature below the corner (when operating on the isothermal).

Hope this helps!

 

I have a lot of respect for you for your giving so much to this community, and taking the time to understand and help people like me with understanding our turbines and their problems. Thank you.

CEKGVBMX should be CSKGVBMX.

I had a hard time finding these in the control reference because I was scanning for the constants, and they are spelled differently in the book than the MK4 or elementary.

CSRGVXXXX is spelled CSKCGXXXX in the control spec.

From 06.11.09, IGV CONTROL REFERENCE, MK4_CSRGVV2_00:

CSKGVBMX IS 30DdegF and described as the MAX FUEL TEMP CONTROL BIAS.

The MK4_CSRGVV2_00 algorithm description states:

"This VIGV temperature control reference uses fuel temperature limit. The reference is biased up by CSRGCTXB as a function of the difference between CSKGVMAZ and CSRGV to allow transient response. It can only decrease at rate CSK GVBR to avoid overtemperature alarms. At full open IGVs, the temperature bias is zero. (but its 30 usually)

If sudden load pickup is required, extra capability is present because of the bias, and the IGVs open first on proportional action to ~87% power, then on integral action to full power within ~20 sec.

When IGV temperature control is selected, a suppressed reference is imposed at initial loading (on MS6001, MS7001, and MS9001 units). This keeps IGVs closed to avoid combustor pulsation at low load.
..."<pre>
_ _____ CSRTVGXB
(CSKGVBMX)-------------------------------------------------------->|MIN |------->28B, TTRX
_ |SEL | |
_ -1-------- ------>|____| |
_ | _____ | |
(CSKGVBR)-------------------------- X------|MAX| | |
_ |SEL |-----> + |
_ ---->|____| + |
L83GVBP---------------------- | | |
_ 28Q55 | | | |
_ -2 | | | |
(CSKGVBG)--------- X ------| |---- + - -------------------------------------------
_ |
_ |
_ |
_ MAX >---------- + - -------< CSRGV
_(OPEN)</pre>
L83GVBP is "1" now, and as long as 14HS IS "1" and IGVs are not in Manual. There is a logic point, OPTVTNH (TEMP CONTROL REFERENCE, (100% TNH-TNH)*TTKRXN), in 28B that if it were 1, would replace this CSRTVGXB with a simple TNH related output, but this has been a 0, as far back as I can see.


MK4_CSRGVV2_00 does appear to be a transient response bias that ramps back down to 0 or near 0 as the IGvs steady out. I do not see how this would remain at or about 30degF indefinitely. I'm looking into whether this would have been changed by our PEAK modification (it was added after commissioning). I know it wasn't part of the shroud seal mod. the curves were modified the right way for that, slopes and all. I posted a thread on that a while back when it happened under a different similar user name, before that account expired.

Is there a way to see if that algorithm changed?

I downloaded C_TI, C_TL and Q_TI, Q_TL files from the MK4 for an aftermarket HMI vendor a while back, would it be possible to check those for a difference in installed vs. elementary algorithms?

Thanks again for your help so far! I hope my ASCII art didn't turn out too bad, It didn't transfer to HTML well, so I did the best I could, using preview.

 

cactus_chris,

Thanks for the kind words. I'm just trying to 'pay it back' for the knowledge and information people shared with me when I was just getting started. Without the patience and kindness of a couple of people in particular, I would not be able to do what I can today--and I'm forever grateful for my good fortune. So, I just try to repay it as best I can.

Some of the early Mark IV Control Specifications were a 'work in progress,' meaning that they were developing the CSPEC and the Mark IV simultaneously--and it wasn't always a coordinated effort, things were changing so quickly at the time.

From the description you copied from the CSPEC, I would say this was never intended to be a 'permanent' bias, but, rather, one that would become effective during certain transient conditions. So, it would seem the logic driving the bias selection/enable permissive is not--and has not been--correct for a very long time.

Again, I would say that if you were able to force L83GVBP to zero without affecting any other rungs (I suspect it's only used for this purpose--but I may be wrong) you would find that the power output of the unit would drop, and Management and Bean-Counters ain't gonna be happy about that! NO AT ALL!

I have tried re-arranging the ASCII art you sent, but I can't make heads or tails of it (sorry!). I did look at some Mark IV Speedtronic Elementaries I have and I still didn't see any case of where the bias was enabled--the logic driving the bias enable/selection logic was always false (logic "0").

And, as for looking at the "logic dumps"--that was one of the major shortcomings of the Mark IV: they couldn't be downloaded and compared to the elementary depictions. Only the logic (sequencing) could be downloaded and examined/compared to the elementary.

Wish I had better news for you. I think you're stuck with the bias, 'cause like I said if you disable it the power output is gonna go down and ain't no one whose salary is based on power generation gonna like--or accept--that.

Again, I'm presuming that TTRX is actually 30 deg F above the lesser of TTRXP or TTRXS (meaning that the bias is, indeed, being added to the exhaust temperature reference/limit calculation from the control curve values). I would really need to see the blocks from the elementary at your site to make a better comment. (Have you considered scanning the blocks/logic and posting them to a web hosting site and then posting the URL here where we could download them and examine them? tinypic is one site that's been used here before.)

 

CSA,
>From the description you copied from the CSPEC, I would say
>this was never intended to be a 'permanent' bias, but,
>rather, one that would become effective during certain
>transient conditions. So, it would seem the logic driving
>the bias selection/enable permissive is not--and has not
>been--correct for a very long time.

After watching this and building a graph for the operators on the DCS to show them how these curves and limits come together, I think I understand it a lot better now. It is certainly not intuitive.

Scenario:

Initial Conditions:
TTXM is at some operating value, CPD is > TTK0_C. TTXM is below the slope of TTK0_S. TTRXB is 30 greater than the curve defined by TTK0_C. DW goes up Because IGVs are only at 72DGA.

Dynamic condition:
TTXM goes up. IGVs are in Temp control, they open. TTRXB lowers closer to the curve as CSRGVTXB lowers. If TTXM continues to go up IGVs continue to close to cause TTRXB = curve defined by TTK0_C. TTXM may be higher than TTRXB only for a few moments during this period.

This is not how I envisioned "temporarily going over a limit" to work, but that seems to be the way it works here. It does seem like maybe every one isn't in as hot of water as I thought after all, and now we can see it a little easier. (See GIF file in link below)

>I have tried re-arranging the ASCII art you sent, but I
>can't make heads or tails of it (sorry!). I did look at some
>Mark IV Speedtronic Elementaries I have and I still didn't
>see any case of where the bias was enabled--the logic
>driving the bias enable/selection logic was always false
>(logic "0").

Here's the related info in non-ascii art form:

https://www.dropbox.com/sh/vogcn44l4ptje5a/AABFwO7vALru1YyJfjmheldVa?dl=0

Thanks again for the help!

 

Sorry, I had to make a couple small corrections

Scenario:

Initial Conditions:
TTXM is at some operating value, CPD is > TTK0_C. TTXM is
below the slope of TTK0_S. TTRXB is 30 greater than the
curve defined by TTK0_C, because IGVs are only at
72DGA.

Dynamic condition:
TTXM goes up. IGVs are in Temp control, they open. TTRXB
lowers closer to the curve as CSRGVTXB lowers. If TTXM
continues to go up, IGVs continue to open to cause TTRXB =
curve defined by TTK0_C once IGVs are full open. TTXM may be
higher than TTRXB only for a few moments during this period.