Hi Experts?

I am very grateful to this site who helps me on each steps but now i am stuck to tune up the PID setting of the gas turbine fuel.
I hope many experts are there who helps me out. I attached an image while PID Enable.

http://i41.tinypic.com/66ym47.jpg

<b>moderator's note:</b> <i>You cannot imbed images or video in forum posts. JCP had this link plus one for video that I could not bring up.</i>

First of all i mark two lines on the image. I can not understand that is it a PV minimum (00000)or maximum (11522)?

At that time SP increase or decrease by the user the following logic execute. For example user press the push button a 1 is add in the previous setpoint and then this setpoint is converted into frequency. Suppose 62% is equal to 7144 of the 11522 Hz.

Now PV is directly address by the %AI (Frequency).

But the problem is that there is always a error show within 1 second so the speed sometime goes too 61% or some time goes to 63% if the setpoint is 62%.

I want to use percentage instead of %AI and as well as SP is also in percentage. But when i do it the engine is continuously ramp up and shutdown.

I also change the 11522 (which i mark) to 100 when using percentage with same gains.

CV is not manipulated when using %age at SP and PV it shows always 20000(Lower clamp) and 0%.

How can i minimize the speed variation?
Is KP increase or decrease? Is KI increase or decrease?

My scan time while engine is running is 15ms. I set the sample time to 10 ms is it correct?

i want that when both error is zero then speed not changed until user increase or decrease the setpoint.

whenever user increase the setpoint suppose 62% to 65% the PV goes to 66% then back to 63% and after that it stays on 65% but after sometime 64 to 65 fluctuation noted.

I check the noise but there is nothing found.
I hope guys you will help me.
Thanking you i will remain.
Regards

 

hi
In my opinion, your P parameter maybe is set too big,and I parameter maybe is too small.

and your cycle time is too fast, maybe.

 

The marked numbers are the minimum and maximum values of the PV and SP,or their range.If your PV is overshooting on correction, add derivative component and try decreasing the gain a bit.

 

> add derivative component and

Seriously? You're going to advise a total stranger to turn on derivative on a process you are not responsible for and not sitting in front of ???

In answer to the original poster, Don't do this unless you know EXACTLY what to expect. Reducing gain settings and / or making the Integral Time longer will make your PID controller slower and ~<b>probably</b>~ reduce the overshoot.

Turning on Derivative action when you don't understand how it works or have not set it up properly could significantly overshoot the process or cause the control loop to become unstable - I would hope that the worst effect would only be to trip the turbine.

As ever, I'm going to repeat my standard advice at this time, which is that you should invest in either a proper PID tuning tool, or hire someone who has one.

Rob

 

Hi;

I check with changes in integral and proportional gains. over shoot is now less but still there. I can not understand why it goes back that is if setpoint is 80% then why it goes to 78 or 79. How i control it is KP or KI?

I want also if error is zero then how can i calculate that what is the CV value? I it must be between 16000 to 20000 counts?

Regards

 

http://expertune.com/ArticlesFull.aspx#PIDOpt

Kp = Gain = the response of the controller to a new error - ie a setpoint change or a process change.

If Kp is Too Big, the controller output change will be too large and the controller will overshoot. If Kp is Too Small the controller will not reach the setpoint. Since getting Kp right is hard, we use Integral to correct the remaining error.

Integral Time = Ti = the time to repeat the Kp Control action.

If there is still an error after the Kp correction is made, repeat some of the control action. If Ti is Too Short, the controller repeats too quickly and the process overshoots. If Ti is Too Long the controller does not repeat quickly enough and the controller takes longer to reach setpoint.

The definitions of "Too Big", "Too Small", "Too Short" and "Too Long" are completely dependent upon your process and your requirements and the subject of much argument amongst engineers. To confuse engineers even more Kp and Ti are often expressed in different units and forms.

Further explanation of PID tuning will take more space than available in this post and generate more arguments.

Rob
www[.]lymac.co.nz

 

Hi,

I understand a lot of things from expertune thanks for the link.
But my problem is fix little bit. Overshoot is due to KP high and KI oversome this am i right?

But i can not understand that why PV went goes below after overshoot. For example if i set the setpoint 74% the PV goes 75% but why it goes 72to 73 % and then stable at 74%.

Which parameter control it?

And i want to know if i use reset wind in GE fanuc 90-30 CPU352 series what effects on the PID output? Please reset wind up clear by an example.

Thanking you i will remain
Regards

 

Ah, yes, you need the critical "don't oscillate" parameter enabled. Not sure if the 90-30 PID has that one.

Just kidding, what causes the PV to swing back to the other side of the SP is the same as what caused the overshoot in the first place -- the proportional or integral get too wound out in THAT direction before the process has time to respond.

Forget the history of the deviation for a moment and pretend you're approaching the setpoint for the first time after a disturbance that produced error in that opposite direction. The reason the "overshoot" is smaller than the first one that happened in real life is that you're starting from closer to SP than you did the first time, so P and I don't get wound up as bad.

 

Call a vendor, kidding

JCP could you tell me which controlling machine you are using look like Bentley Nevada company, I have useful website about how to understand and calculate the KP and KI. look at it http://www.engin.umich.edu/group/ctm/PID/PID.html

 

Hi Ahmed;

First i try to control the gas turbine speed. Gas turbine model Saturn 1100. As i stated before the speed is in control but the correct values of gains is still needed.

The overshhot is minimize but i can not understand why it goes below to the setpoint? which parameters i increase or decrease?

Regards

 

By now you've figured out that tuning by hand takes experience and educated guess work. Often you cant get optimal tuning and have to live with what you've got.

Overshoot and oscillation is a sign of controller tuning that is overly aggressive for your process. Likely you have some significant deadtime (time delay) in your closed loop control, which is very difficult to tune for by hand.

And all of this assumes that your problem is in the PID controller settings. It is just as likely that it is a worn valve or linkage somewhere in your plant that is the real cause.

Nobody here on this forum can tell you the "right" tuning numbers for your process - We can only suggest some changes that you might try. Seriously, if the current controller performance isn't good enough for you then you have 2 choices ...

Keep playing with the numbers until its better - or call someone with the tools, skills and experience to do it better.

Rob
www[.]lymac.co.nz

 

HI;

Yes i fix the number very well now. But a one point occur where PV overshoot this is the 2nd critical speed of the turbine that is 88.5 to 90.5

In this region the speed overshoot if i give the setpoint from 85 to 92 then works fine. If i increase 0.5% then it overshoot. I check the PV and CV on oscilloscope but no noise found. I think some resonance disturb the PV

Is there any method to minimize it?

Regards

 

Hi;
Please help me in this issue?
Regards

> In this region the speed overshoot if i give the setpoint from 85 to 92 then works fine. If i increase 0.5% then it
> overshoot. I check the PV and CV on oscilloscope but no noise found. I think some resonance disturb the PV

> Is there any method to minimize it?

 

Listen to Ahmed. Study the CTM PID examples.

However, the CTM PID examples assume you have some idea of plant model!!!

The system identification part of tuning is the part that no one wants to, or seems to be able to do, but after you have system identification licked then the tuning parameters are easy to compute.

When I look at a PLC screen it means nothing to me since I don't know or care to know GE PLCs. If you provide a transfer function for the generator speed then I am sure many of us can help you compute gains that will work or be close.

So far I have not seen a transfer function or even a log of time, control output and speed that can be used to calculate a transfer function.

BTW, a PLC is not really a good controller for a turbine.