Sample Time of the PID ISA

Hi Friends;
Scanning time is 16 ms for GE 90-30 series CPU 352. I am confuse to set the sample time for the PID. If i set it 20 ms then what happen?
In the help file i noted that if scan time is 9 ms and sample time is set to 10 ms the PID execute after 18 ms why?
If i set the sample time 50 ms then after how many scan the PID execute?
Regards

Hi there,

What type of process are you trying to control? I.e. speed of a pump to maintain a level or pressure? temperature control?

I'm not that familiar with the GE 90-30 (we have one at one of the treatment plants we look after but my boss handles that one, it is more his vintage of PLC ) but i'm assuming the sample time is actually the time interval between loop updates?

Unless you have need for a very, very fast loop you probably don't need to be anywhere in the 10's of milliseconds range.

If in doubt, I like to write a bit of code to simulate a first order approximation of my process, and then attempt to control it with the PID so I can see what's likely to happen. I have an old SLC5/03 on my desk purely as a simulator!

Hi;
Process type?
Control the speed of gas turbine.
I used GE Fanuc 90-30 series by using PID ISA. The processor is CPU 352.

Can i calulate the K (Open loop process gain) without using manual command. Suppose I want PV moves to start after 50ms (Dead time) and reaches after 1.5 second to the 63.2% of the new PV value (TP).

Sir if i write whole process can u help me tune the PID?
Regards

Hi again,

It does look like you have a reasonably fast process to control after all.

i have absolutely no experience with gas turbine engine control so I'd probably be of little help to you. Most of the processes I work with are of a nature where we can "tune by feel" without running the risk of blowing anything up before I can hit a kill switch. Not sure if thats the case here!

Hopefully someone else with experience in this can help you.

Failing that, take a look at the PID tuning handbook here:

http://www.controlglobal.com/Media/MediaManager/ControlSoftInc_PID.pdf

they have a bit of info on the GE processor you're using, and some good hints for tuning by feel.

Cheers,

Steve

Take a look at the open loop tuning section of that guide, some rough calcs on getting a starting value for P and I.

I explained about the relationship between PLC scan and PID function solution frequency in your previous thread.

The PID function will never be evaluated more frequently than once per 10 mS regardless of PLC scan time. If evaluation frequency is 10 mS and PLC scan time is 9 mS, then the function gets evaluated every other scan. Think of it this way:
Scan 0 - PID gets evaluated.
Scan 1, 9 mS since last evaluation - PID does not get evaluated
Scan 3, 18 mS since last evaluation - PID gets evaluated.
Scan 4, 9 mS since last evaluation - PID does not get evaluated
etc

When the PLC gets to the rung with the PID function each scan, it first checks to see how much time has elapsed since the last evaluation. If that time is greater than the solution frequency you've set, the function gets evaluated, otherwise it waits for another scan.

Hi Steve;
Today i run the machine with sample time of 10 ms and scan time is 16 ms. It means
Scan 0= PID evaluated
Scan 1 16ms= Which is greater than sample time so PID again evaluated;
So on each scan PID evaluated. Am i right?
I observer one thing which is alarming me. I press F10 on PID and i see the detailed window.
Run the machine and carefully examine on the horizental bars of SP, PV and CO.
SP remain constant at that perticular time,
PV increaseing lineary but When PV goes as closer to the SP the CO moves very fast from 0 to 100%.
Is there is noise? IF yes how can i control it,
Is it due to sample time?
After get satbility then i increase the SP, PV moves but CO again moves very fast between zero to 100 and stable after 3 to 4 sec.
Please help me why it happens? If i increase the sample time is CV become stable suppose get PID solution after every two scan am i right?
Regards

If your CV is rapidly swinging between zero and 100%, it sounds like your proportional gain is too high.

CV = Bias + Proportional Component + Integral Component + Derivating Component

Proportional Component = Error * Proportional Gain

When you are close to the SP, error can swing between positive and negative, so the Proportional Component of the CV can change sign. If the magnitude of the Proportional Component is significantly greater than the magnitudes of the other components, the CV will react accordingly.

HI;
One thing i want to clear the rungs which have a scalling of PV and SP insert before the PID or after the PID?
If i insert before is current value at that scan manipulated by the PID?
Regards

I'm not clear what you're asking. The PID function will calculate the error each time it evaluates based on the current SP and PV values.

Hi Steve;
My point is for example
RUNG #=230 Have a scalling instructions for SP.

RUNG #=232 Have a scalling instructions for PV.

Rung # 245= Have a PID control

Suppose scan #1 have a Sp value 19000 counts and PV value 15000 Counts so
PID use these values or PID use these values on next scan.
I hope now you cleared which i want to ask.

If i use scalling after PID like
Rung # 230 have a PID
Rung # 232 Have a SP scalling instructions
Rung # 245 have a PV scalling instructions THen

PID use old values of the last scan or current scan values?

Generally, you won't see any difference, but it is better to do your scaling before the PID function.

Under steady state conditions the PID's CV output is at some value and the (scaled) PV equals the (scaled) SP (Error = 0). If your scaling is after the PID function, the first scan the PLC detects a change in the SP, it will calculate the Error based on the difference between the unscaled SP and the scaled PV. For each scan after that the Error will be calculated based on the difference between the scaled SP and the scaled PV. So for one scan, the Error will be inaccurate and the CV calculated by the function will reflect that inaccuracy. Whether that causes a problem will depend on how fast the system can react to that inaccuracy as well as whether the inaccuracy produces a larger or smaller correction than the accurate calculation of Error.

Hi Jcp,

Have you managed to tame that PID loop yet? agree with Steve Bailey that its likely one your proportional term is too large.

If your PV is likely to experience noise, then I'd normally try a zero derivative ( "D" component) for now until you can at least achieve some stability with a constant SP and slowly changing PV.

Another thing worth trying out is a much slower loop update time than you currently have (say 500ms compared to your 10ms). I've accidentally set a very fast loop update time (on a different PLC mind you), and had my previously very stable loop go haywire.

Are you testing this on a live system?

Cheers,

Steve E

Hi;
Yes i follow the steve instructions regarding Kp gain.
I decrease its value and also increase little bit KI value.
I used only PI control. Now a point come where these gains creating trouble. For example 62% speed to 84% is very fine but after 84 to 90 its create trouble (Overshoot and again back to less than SP after 5 to 6 sec then stable) But on ward like 91 to 100 is pretty fine.

How can i check the noise on PV?
I submitt some figures here please guide me how i overcome it.
SP= 0 to 11516 but intial set at 7140 62% of 11516 when PV reaches at 62% abd satbe there then user increase the set point but limited to 1 % if he press one push.
PV= 0 to 11516
CO= 0 to 32000 counts

I use delay to start PV TP= 0.5 sec
I used the time when PV reaches 63.2% of its final value is 1 sec.
I use K (Open loop asumption)=0.145

Now i set KP=7.55
KI= 0.07

Please its my humble request help me. If any body have a tool for PID tunning send me KP and KI?
If any information needed i explained more detail for the process.
Regards

HI;
Need advise Please.