from the Instrument and Control department...
Isochronous Speed Control
 Posted by ZH on 22 October, 2010 - 5:31 pm
My question is about GE Gas Turbine Speedtronic Mark-v control system. According to Rung#10 of SEQ_CORE if Turbine will operating in Droop mode its formula will be
FSRN = (FSKRN2 * (TNR-TNH)) + FSKRN1

Now suppose that the turbine is running at 50% load and droop setting is 4%. Therfore, TNR will be 102%. Now at this time if operator shifts that turbine to Isoch mode then according t0 Rung#10 SEQ_CORE Isoch term will be added to Droop term.

FSRN={(FSKRN2*(TNR-TNH))+FSKRN1}+ {FSKRN3*(TNRI-TNH)+FSRNI}

but here TNRI=100% due to this error e(t)i-e (TNR-TNH)and(TNRI-TNH) in two terms (RHS of above eq) will be different and if Isoch mode is PI controller then how integral will work as e(t)is different in both terms?

 Posted by ProcessValue on 24 October, 2010 - 1:50 am
to know about the isochronous mode of operation , you must know how the FSRN block is configured as a PI controller. to my best of knowledge GE uses backward difference method for integration in the blocks. the Z^-1 symbol you see in the blocks represents the unit time delay in the signal used. i am giving a brief

writeup about the use of backward difference method for PI controller for better understanding. I hope you are familiar with mathematical analysis in Laplace and Z transform.

part 1 : explanation about sampled data integration and discrete PI controllers.

for any control system with U(s) as the output and E(s) as the input , the transfer function of a the transfer function of a PI controller is given by
`       U(s)/E(s) = Kc + Kc/Tis using the backward difference which uses the formula s = (1-z^-1)/Ts where Ts is the sampling time in the above equation we get        U(s)/E(s) = Kc + Kc*Ts/Ti(1-z^-1)converting the equation to the time domain we get U(t) = U(t-1) + Kc (E(t) - E(t-1)) + (kc*Ts/Ti)E(t) this can be rearranged and written in the form U(t) = U(t-1) + K1E(t) + K2E(t-1)`

this is called the "velocity" PI algorithm. As opposed to the fixed control reference , it uses the previous value of the control element to synthesize the new value. In essence, the control is calculated as a change, hence the term 'velocity form'.

 Posted by ProcessValue on 24 October, 2010 - 1:54 am
Part 2 : explanation about the droop control mode in GE mark Vi controller with reference to the governor chara.

Now go and check the TNR synthesis block "TNRV1". the value of the TNR is synthesized in the following manner.

TNR(t) = TNR(t)*Z^-1 + L83JD# * ( L70R - L70L) where L83JD# is the load rate array and l70r the raise command and l70l the lower command, both binary values.

converting completely to time domain as F(t)*Z^-1 = F(t-1) we get
TNR(t) = TNR(t-1) + L83JD# * ( L70R - L70L).

this is nothing but pure integration using the backward difference method. as the auto load rate is already specified , it cannot be taken to be a controller as such as the response is fixed and not dependent on the error value. the L70R and L70l can be triggered by either the manual raise command , the auto synch command , the temperature limit command or more importantly the MW control block which is what is used by the preselect load controller.

now go to the FSRN synthesis block "FSRNV4". the value of the TNR is synthesized in the following manner.

FSR(t) = (TNR - TNH(t))*FSRKN2 + FSRKN1
if we consider the (TNR - TNH(t)) as E(t) then we have
FSR(t) = FSRKN2*E(t) + FSRKN1
this term does not have previous value of the control element, thus droop mode is a proportional controller.

Physical implementation of the droop controller.

now that we have seen how the droop FSR is calculated , now let us examine what are the physical implications of such a controller. i have included a pic here ( top potion of the pic - Governor characteristics)

in the pic , you see the governor characteristic. the graph represents the droop chara of the machine. Now let us suppose that the machine is in point a. the load(MW) is l1 and the frequency f1. the machine is running independently and on droop mode. Now if i give a manual raise command. My TNR first rises. what it effectively does is that it increases the droop reference curve from "chara 1" to "chara 2". now if you will see , as the load is a constant in a independent machine (not entirely true , depends on the type of the load , but i am assuming a pure resistive load for simplicity) , the speed of the machine rises and reaches point b. Here the frequency is more , and if you give another rise command , it reaches point c with higher frequency and a shifted reference chara "chara 3". if you give the lower command the reverse happens.instead of the upward shift in the reference , the droop reference shits down.

we have now seen what happens if the manual raise and lower is given. what happens if the load changes. suppose that the machine is operating in point a and there is a sudden change in the load , and it reaches load l2. now as the machine is following droop chara 1 , the speed drops down to f2 and the machine reaches a steady state value at point d. now what will we do if we need to rivet back to the original speed. we give the manual speed increase , which in effect shits the reference to chara 2 and the machine speed rises to the previous value of f1. thus the difference is that earlier the machine was following droop chara 1 but now at steady state position e , the machine will be following droop chara 2.

the last part of the explanation is what which explains the manual frequency control. the objective of the iso mode is to keep the frequency constant irrespective of the load in the machine. the last part of the explanation is in-effect a manual iso mode , and this is what is automated in the isochronous control.

 Posted by ProcessValue on 24 October, 2010 - 2:00 am
part 3 : explanation about the isochronous mode control in GE mark VI with reference to governor chara.

The isochronous mode algorithm is explained below

a. once you put the iso mode on in a independent machine , the LISOK signal goes high.
the following calculation takes place
` FSRNI(t) = FSRNI(t-1) + (TNRI -TNH(t))*FSRKN3 (as you can see this is a form of the velocity PI algorithm , andthis were the PI is actually implemented in the controller) b. After the calculation the follwing check is done.         if ( mod (FSRNI(t)) >  FSRKN6 )                   the respective l60ir ( iso setpoint raise) or thel60il (iso setppoint lower) goes high. the LISOK signal goes low thuscutting off the TNRI signal from the FSR block. The L60IR OR theL60IL increases/decreases the TNR , with the help of the L70R/L70L logic , thus raising or lowering the droop reference.        else                   the FSRNI(t) is added to the FSR to get the new FSR value.            FSR(t) =  (TNR - TNH(t))*FSRKN2 + FSRKN1 + FSRNI(t-1) + (TNRI -TNH(t))*FSRKN3in the above equation , the (TNR - TNH(t)) is no longer a error signal because , "TNR is not a setpoint any more". when the iso mode is selected , the TNR raise and lower is inhibited from any other source. only the L60IR and the L60IL have the permissive to change the TNR. as this particular code is executed only when L60IR AND L60IL is both zero , TNR is nothing but a constant in the equation. now rewriting the equation               FSR(t) = FSRNI (t-1 ) + ( (fsrkn3 * TNRI) - ( TNH(t) * (FSRKN3+FSRKN2)) ) + FSRKN1 + (TNR*FSRKN2) here now the FSR and the FSRNI now represent the same common output can be renamed as FSRI                                         FSRI(t) = FSRI (t-1 ) + ( (fsrkn3 * TNRI) - ( TNH(t) * (FSRKN3+FSRKN2)) ) + FSRKN1 + (TNR*FSRKN2)here now the equation ( (fsrkn3 * TNRI) - ( TNH(t) * (FSRKN3+FSRKN2)) ) is the error signal E(t)                                        FSRI(t) = FSRI (t-1 ) + E(t) + +FSRKN1 + (TNR*FSRKN2) `

the last equation is the final iso mode calculation equation. if you will see , it is analogous to the velocity form of the pi algorithm. thus the PI loop only contains one REDUCED E(t) signal and not two.

Physical implementation of the iso controller :

now that we have looked into how the FSR is calculated in the iso algorithm , let us see the controller inaction during actual conditions.

please refer the same diagram as before (iso chronous controller in action). which is uploaded here

a. for small load variations - the small load variations is taken care by the FSRNI block. let us suppose that the machien is running in point a where the load (MW) is l1 and the frequency f1. In this condition i am assuming that the machine is already in iso mode and reached a steady state such that the TNRI and the TNH are now equal(near equal) , the now let us suppose that there is a small load increase. the TNH value of the machine will go down as per the droop chara 1. now FSRNI is calculated and it will be found that FSRNI is less than the deadband limit. the value of FSRNI will be positive and it will be added to the FSR block. this additional FSR will help in increasing the speed of the machine to its previous value. the reverse happens if the load goes down by a small amount. the speed of the machine will rise , the FSRNI will be negative and reduce the actual FSR thus reducing the machine speed back to its previous value.

b. for large load variations - for large load variations the droop chara is changed before the TNRI comes into play. let us suppose that the machine is in the initial condition Point a. now for a sudden load increase from l1 to l2 , the frequency chages from f1 to f2. now the FSRNI is calculated and it will be found that it is more than the deadband FSRKN6. now the L60IR command is issued which inturn increases the TNR value. the rise in the TNR value will raise the droop reference and the speed of the machine as already discussed in the droop mode study. thus when the machine reaches the point b , the droop chara is raised from chara 1 to chara 2. this goes on till the TNH reaches the TNRI and the calculated FSRNI is below the deadband limit. from that point onwards the iso changes to the pi mode and FSRNI takes care of the rest. thus from the above it can be seen that , the for small load variations the FSRNI is responsible for maintaining constant frequency. in this mode , there is no droop reference change. for large variations , the FSR reference itself is changed till the speed becomes more or less equal to the setpoint value. only then does the FSRNI comes into play.

hope this clears the doubt regarding the droop and the isochronous mode of operation in GE turbine systems. i will wait for more experienced members like MIKEVI and CSA to comment so that i can know if i have left out something out or made some mistake but to the best of my knowledge this is how the controller works. I know this is a long post , but hey its a Sunday here and i have got plenty of free time. ;).

ps - now that i have explained this , you can draw a similar graph between the FSR , TNH and the TNR values to how the FSR change for a respective load change /manual change operation.

i have taken some liberties here and there make the explanation simple ,further discussion is always welcome.

 Posted by ZH on 24 October, 2010 - 4:17 pm
Thank you very much. Its helps me alot in making my concepts. Further that I have studied Velocity PI algorithm from texts. Thanks again

 Posted by ProcessValue on 25 October, 2010 - 12:18 am
> Thank you very much. <

Glad to be of help :). well , FSKRN6 is the deadband over which the iso set point tracking raise/lower command is issued. do read the part three of the explanation you will come to understand it better. and yes also look into the logic diagram of FSRN block.

 Posted by ZH on 24 October, 2010 - 4:45 pm
One more question that is what is meant by "isoch deadband FSR" (FSKRN6)? I have seen the CSP its value is "0.06CNT00"

Your use of this site is subject to the terms and conditions set forth under Legal Notices and the Privacy Policy. Please read those terms and conditions carefully. Subject to the rights expressly reserved to others under Legal Notices, the content of this site and the compilation thereof is © 1999-2013 Nerds in Control, LLC. All rights reserved.

Users of this site are benefiting from open source technologies, including PHP, MySQL and Apache. Be happy.

Fortune
"Nirvana? Thats the place where the powers that be and their friends
hang out.
-- Zonker Harris
Help keep our servers running...