Actuator with delay

[lyraKenza]

Hello,
I have to design a PID controller for a second order system of transfer function of follwing form:
G(s)=K/s(1+Ts)
my command is an angle (on degree) After doing several tests with my actuator, I found that , it takes 10 s to respond (give me the required angle of command)
The PID I designed before is no long working. I'm thinking about model my system as a delay system, it's possible ? if so how can I add the delay of my actuator ?
Thank you

 

[controlnovice]

School problems are not permitted in this forum.

______________________________________________________________________________
This is normally the space where people post something insightful.

 

[roydm]

controlnovice (Electrical) 13 Mar 17 16:37

School problems are not permitted in this forum.

Just as well, I have no idea what he's talking about LOL

 

[PNachtwey]

School problems are not permitted in this forum.

True but they should be with a huge but......

Just as well, I have no idea what he's talking about LOL

This is why student questions should be allowed. I bet most that view this forum can't solve the problem and they could learn something.

What I think the OP wants are the PID gains so that the set point or position will be reached in 10 seconds.
The OP's wording is awful. He misuses the word delay but I doubt English is his first language.
I didn't answer because I thought the thread would be deleted by now.
I could have immediately posted this solution within minutes

http://deltamotion.com/peter/Mathcad/Mathcad - t1p1 pid pi-d i-pd.pdf

The OP would just need to change where the closed loop poles are to yield a 10 second response instead of a 0.3 second response.

Now for the but.....
Obviously this is homework and we SHOULD NOT provide answers within 1 week so the home work will be past due before we give the student a the answer. They should come here for the education but not for the answers.
Meanwhile we should quiz the student about his problem to make him think.
I would ask:
Where did that transfer function come from? This is one of my pet peeves about college control classes and books.
What kind of system is it for?
What are the units for K?
What does the first s in the denominator do? It integrates but it integrates what?
Can you write the differential equation from the Laplace Transform.
Why use a PID instead of a PI controller?
Now assume this is for a simple motor in position control.
If the current position is far from the set point position, what will be the velocity? Can it move that fast. When the control signal is at 100%
When control signal is the saturated at 100% the system is doesn't care about the feed back. It is essentially open loop. That is not good. How do you avoid that?

I think we could have some fun with the students and it would be education to non-students.








Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

http://forum.deltamotion.com/

 

[roydm]

After all this time as an instrument tech (40+ years) I like to think I have a fairly good idea of process control, and how to tune a loop, its the modern terminology I don't understand then I was never taught that way, I sometimes think these big words are just designed to confuse.
My old boss had a favorite saying "BS baffles brains"

 

[PNachtwey]

After all this time as an instrument tech (40+ years) I like to think I have a fairly good idea of process control,

The OP's transfer function is for a motor or something similar.

and how to tune a loop,

There are so many different types. The way don't tune the same.

I was never taught that way,

Me neither.

I sometimes think these big words are just designed to confuse.

No, they only confuse the confused.

My old boss had a favorite saying "BS baffles brains"

One should be able to tell what is BS and what is not and not be baffled but this is not BS.
When you understand you can write auto tuning programs.
No tweaking gains and drinking coffee waiting to see what happens.
This a system similar to what the OP is asking about.

https://www.youtube.com/watch?v=Ba1yJhvOmvc&t=3s

Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

http://forum.deltamotion.com/

 

[roydm]

Peter, no doubt you are right, I certainly couldn't solve it, I have no idea what a transfer function is but ask me to tune a loop, I can do that without big words and a computer simulator with little wheels wiring around.

Why does he want to delay the actuator?

Cheers
Roy

 

[xnuke]

Why does he want to delay the actuator?

The OP wants to model the delay of his actuator mathematically so he can find a better controller algorithm, one that works with the delay. Delay is often a significant destabilizing factor in closed-loop systems that requires more than just PID tuning to overcome.

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[roydm]

Ok, I take it the actuator is electric, a pneumatic actuator usually has very little delay.
Either that or a faster actuator with process delay where you make a change in the feed rate but it doesn't show up in the PV until a few seconds later.

A long time ago (about 40 years) I was using Ramsey controllers to control the power draw of cone crushers, a change in the feed rate would take about 20 seconds to show up as power draw in the crusher because that's how long it took the rock to arrive. 400 HP motors would be run at 398-400 HP 24/7/365, something like 120,000 tons per day for 8 crushers.

These controllers would make a change then wait for the process variable to reflect the change (20 second delay) before making another proportional change. They worked far better than a PID controller in this situation, the PID Integral action would wind up the output.
I'm sure it would be real easy to do in any PLC.

 

[PNachtwey]

I think the OP just has trouble explaining what he wants to do. There is no reason for an electric actuator to have a delay or dead time of 10 seconds. I interpreted this as wanting a 10 second response from starting angle to the set point angle.

This should be easy to do

but ask me to tune a loop, I can do that without big words and a computer simulator with little wheels wiring around

The test system with little wheels has little friction and therefore relatively long time constants. If you think you can start by adjusting the proportional gain you would be wrong.
I can think of systems that are very difficult to tune without knowledge of the transfer function and how to place poles and zeros.

For systems with relatively long dead times there is the Smith Predictor. Smith Predictors don't wait and the integrator doesn't wind up. The key is the predictor part and it only needs to be close. It doesn't need to be perfect.

xnuke, don't you moderate this forum? Can't you change the rules so that students can post as long as they are willing to get grilled for a week.










Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

http://forum.deltamotion.com/

 

[xnuke]

There's no official moderators, and the subject of student posting has been brought up to site management before many times, and it has been decided to keep to a strict no-student policy. I believe the last thing I heard was that site management was investigating the creation of student forums at engineering.com as an alternative so that Eng-Tips could remain for professionals only. I don't know if that's gone anywhere yet.

Personally, between textbooks, TAs, other students, professors, and the rest of the internet apart from Eng-Tips, I feel students have enough resources elsewhere, and I'd rather see them use those resources. Most forum posters don't know how to educate students while trying to help them. Students typically want the solution that gets them the right answer for the least amount of work, but only through work is learning truly possible. I still have student groans echoing in my ears when I'd answer a question with a question, but it was the right way to get them to learn. Forum posters usually don't do things that way because they want to be helpful, and they think by showing the full solution to a student they are helping. Often, they aren't, because the student simply copies it down for their assignment, then forgets it. The students don't learn how to think when that happens.

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[PNachtwey]

xnuke, I agree with most of what you say but honestly, most teachers and books teach what has been taught in the past because that has always been acceptable. They show little real experience.

I still have student groans echoing in my ears when I'd answer a question with a question, but it was the right way to get them to learn.

Yes, that is what I was proposing above.

Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

http://forum.deltamotion.com/