I am trying to use PID to control the temperature in a insulated pressure vessel.

At present due to the slow acting nature of the process the valve tends to ramp to 100% open then not begin closing until after the temperature setpoint is reached.

After the setpoint is reached the valve then completely closes, during which the temperature continues to rise.

What is the best method od preventing overshooting in this case?

Thanks in advance.

Need more information.
Are you using PID? or PI? For slow temp. loops I usually don't use the D.
What are your current tuning parameters?

Greetings moorecontrol,

let's try doing this the easy (easiest?) way ... can you post your .RSS program file? ... can you post a trend showing the response of your process? ...

and welcome to the forum ...

Thanks for having me in the forum.

My current parameters are

Gain: 50
Integral: 0.2

No derivative

My PID call is the only thing in the rung and is in a ladder file that is called constantly

Thanks again

[QUOTE=moorecontrol,Jul 6 2005, 06:17 PM]I am trying to use PID to control the temperature in a insulated pressure vessel.

At present due to the slow acting nature of the process the valve tends to ramp to 100% open then not begin closing until after the temperature setpoint is reached.

After the setpoint is reached the valve then completely closes, during which the temperature continues to rise.
[/quote]

That happens because you aren't using enough differentiator relative to the I and P gains and/or you have a significant dead time.

[quote]
What is the best method od preventing overshooting in this case?
[/quote]
Your desired response is going to be either overdamped or critically damped. You can start by reducing your gains and you may achieve an overdamped response. An over damped response will not achieve the set point as quickly as a critically damped response and may delay production.

I suggest that you post the trend file as Ron suggested. It would be best if the trend showed the response to a big step change in the set point, without that we/you can only guess and use trial and error.
[/QUOTE]
[quote]I am trying to use PID to control the temperature in a insulated pressure vessel.

At present due to the slow acting nature of the process the valve tends to ramp to 100% open then not begin closing until after the temperature setpoint is reached.

After the setpoint is reached the valve then completely closes, during which the temperature continues to rise.
[/quote]

That happens because you aren't using enough differentiator relative to the I and P gains and/or you have a significant dead time.

[quote]
What is the best method od preventing overshooting in this case?
[/quote]
Your desired response is going to be either overdamped or critically damped. You can start by reducing your gains and you may achieve an overdamped response. An over damped response will not achieve the set point as quickly as a critically damped response and may delay production.

I suggest that you post the trend file as Ron suggested. It would be best if the trend showed the response to a big step change in the set point, without that we/you can only guess and use trial and error.

You can try programming a smith predicter into the plc. Ive never done it before, but i know a smith predicter is used on loops with an excessive period of oscillation

Thanks to all for their help.

My final solution was to manually ramp open the valve until the setpoint was reached, then hand the control over to the PID loop in automatic.

Not ideal I know but it works well.

Again thanks to all