Electrician Talk banner
1 - 17 of 17 Posts

·
Electron Factory.Worker
Joined
·
355 Posts
The only diiference is on a PI no derivative term is used in the calculation of the output. The derivative term is the most misunderstood because it often works against the proportional and integral term.

The derivative looks at (depending on the controller). How quickly the error or process variable is changing. The faster the rate of change the more the derivative term will work against the output to restrain it. It knows that the output can't continue in any one direction for too long because it will overshoot the set point.
 

·
Registered
Scada Supervisor
Joined
·
4,512 Posts
I was taught that the D was an anticipator, as in electric heating elements where it might take time for the heat to reach the reading device, and for heat that was still being sent from the element after it was off.
 

·
Registered
Joined
·
5,632 Posts
PID is all about the response to an "error" (deviation) in a process that you are trying to control.

P = Proportional to the SIZE of the error, so it is the Past error because it has already happened.

I = Integral to the RATE OF CHANGE (change in value INTEGRATED over time) of the error, so it is the Present state of the actions of the error right now

D = Derivative of the NUMBER OF CHANGES across time, so it is the anticipation of Future errors.

Most things like pumps and fans only really use PI control, the D is there in case someone decides later that they need it, but mostly the D function is used in the processing industry where you have a continuous process of making something where one or more of the components varies in somewhat predictable ways..
 

·
Registered
Joined
·
1,612 Posts
To add to what JRaef said, D is often left out in many processes where the process has excessive noise as a phenomenon called derivative kick could occur.

In a PID controller, Error = SP - PV. Since any change in the setpoint (SP) causes an instantaneous change in error, the derivative with respect to time (dt) is infinity. This tendency towards infinity causes a kick or a spike at the output which can throw the control loop into an unstable state - IE it can saturate the control variable which might be impossible to recover from.

There are many techniques used to mitigate this in actual practise (such as introducing a lag filter), but sometimes it's just easier to forget about the D term and use a PI controller - especially in faster systems.

In solutions that I've personally implemented, I think I've only used derivative twice where PI controllers are used 90% of the time.
 

·
Registered
Joined
·
5,632 Posts
To add to what JRaef said, D is often left out in many processes where the process has excessive noise as a phenomenon called derivative kick could occur.

In a PID controller, Error = SP - PV. Since any change in the setpoint (SP) causes an instantaneous change in error, the derivative with respect to time (dt) is infinity. ...
You know something, I never thought of that. It not only makes perfect sense, but it explains something that happened to me once!

Live and learn... thanks.

I recently put on a series of VFD classes in which I teach people who do not get exposure to it what PI(D) control means. When I ask people if they have ever used one, they mostly say no. Then I point out to them that the Cruise Control on your car is a PI controller!
 

·
Electron Factory.Worker
Joined
·
355 Posts
You know something, I never thought of that. It not only makes perfect sense, but it explains something that happened to me once!

Live and learn... thanks.

I recently put on a series of VFD classes in which I teach people who do not get exposure to it what PI(D) control means. When I ask people if they have ever used one, they mostly say no. Then I point out to them that the Cruise Control on your car is a PI controller!
Many modern PID controllers allow you to control the derivative based on the pv only. This way set point changes don't give you the derivative kick. A good example of a process that requires derivative is the segway. It requires quick Derivative adjustments to keep it from falling over from small changes in its position.
 

·
Premium Member
Joined
·
20,981 Posts
In practice, heating and cooling are (mainly) the only time you'd normally have occasion to use D. Applications involving flow, level, or pressure are typically only PI.
 

·
Premium Member
Joined
·
20,981 Posts
And that's to prevent short cycling. :nerd:
Not so much. I prevents overshoot, more than anything. It's sometimes used in pumping when the pump needs to follow another pump to keep a fluid mixture (in the same pipe or vessel) in spec.
 

·
Banned
Joined
·
16,967 Posts
Not so much. I prevents overshoot, more than anything. It's sometimes used in pumping when the pump needs to follow another pump to keep a fluid mixture (in the same pipe or vessel) in spec.
Now I'm confused. The heat anticipator in old fashioned thermostats served the "I" purpose, did it not? So what does the "D" accomplish? I though it was to prevent short cycling. 'Splain please.
 

·
Premium Member
Joined
·
20,981 Posts
All D does is shut the controlled item off early for times when the measured PV is known to be laggy, like in temperature applications. Nothing about PID will prevent short cycling. It maintains a setpoint at all costs, assuming a moron didn't tune it. If anything, it creates short cycling.
 
  • Like
Reactions: MTW

·
Banned
Joined
·
16,967 Posts
All D does is shut the controlled item off early for times when the measured PV is known to be laggy, like in temperature applications. Nothing about PID will prevent short cycling. It maintains a setpoint at all costs, assuming a moron didn't tune it. If anything, it creates short cycling.
Right, now I get it. :thumbsup:
 

·
Monkey
Joined
·
185 Posts
In practice, heating and cooling are (mainly) the only time you'd normally have occasion to use D. Applications involving flow, level, or pressure are typically only PI.
I have been taught the same thing. Usually used for slow changing process variable like temp.

You don't even really need it on temp unless you are concerned about an overshoot.
 
1 - 17 of 17 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top