PID Control in SLC 5/04

T

Thread Starter

Ting

I am using a SLC 5/04 to control a process. All went fine during our commissioning except for 1 of our PID control. Our client wants the process to have at least +/- 0.4 error (in a scale of 10). There are two PIDs in the tank. One for the level and one for the pressure. This is a closed tank and the medium is steam. My level control is quite okay now except for the pressure control that goes beyond or below 0.4 of the SP(setpoint is 3.7 bar and my PV will sometimes at 4.1 to 4.5 or below 3.1 or 2.9). The controller output goes to a steam valve. Anyone who can help me fine tune the process? My current P=1.7 and I=0.1. If I increase my I, it can hardly control my output and it overshoots. I am not using a D at the moment.

Please help.
Ting
 
D

David McDaniel

It appears that you have a slow response to the change in pressure. I.E. the pressure chages rapidly when the valve is opened or closed but response to that change is slow. I would be tempted to use just a little D. Though I am sure other may disagree.

Dave
 
I
Unfortunately, there is too little information about the application. The below reasoning is valid only if there is a tank with a controlled steam flow into and uncontrolled flow out of the tank though some properties of the process are valid for other situations too.

One possible scenario of the situation developed (not guaranteed that is necessarily the case) is as follows. You are controlling the pressure in the tank - the process, which has integrating properties (pressure can be estimated as an integral of the difference of the steam flow into the tank minus the steam flow out of the tank).

As a result, you have to apply very little I-action of the PI-control. A lot of I-action would result in instability or low stability margins. On the other hand, you want to eliminate the static error, which can be done via application of sufficient I-action. The static error in your system is caused by the steam flow out of the tank that can be considered an external disturbance. To compensate for this disturbance while retaining integral properties of the control I would suggest using a feedforward signal proportional to the steam flow out of the tank. Ideally, you would use two cascaded loops for the pressure control: the main loop produces a signal equal to the flow demand, which serves as a setpoint for the second loop that controls the steam flow into the tank. The feedforward signal is added to the output signal of the first controller. In the main loop, you would apply a little of I-action, in the flow control loop it can be larger. Of course, it is more complicated as requires two flow transmitters and one more PID-controller.

Even in the control utilizing only one loop, application of the feedforward signal proportional to the output steam flow would be beneficial for your application.

Regards,
Dr. Igor Boiko
Consulting in Control is available
(including modeling, simulations and control design)
[email protected]
Tel: 1-403-294-2745
 
M

Mark Blunier

> I am using a SLC 5/04 to control a process. All went fine during our
> commissioning except for 1 of our PID control. Our client wants the
> process to have at least +/- 0.4 error (in a scale of 10).=20

You are not being clear as to what you want.
If it is +/- 0.4 bar, then it doesn't matter what the scale is.
If it is +/- 0.4 of the displayed units, change the units to meters of mercury and the 0.4 error will become .29 error.

> There are two
> PIDs in the tank. One for the level and one for the pressure.
> This is a
> closed tank and the medium is steam. My level control is
> quite okay now
> except for the pressure control that goes beyond or below 0.4 of the
> SP(setpoint is 3.7 bar and my PV will sometimes at 4.1 to 4.5 or below
> 3.1 or 2.9). The controller output goes to a steam valve.
> Anyone who can > help me fine tune the process? My current P=1.7 and I=0.1. If
> I increase
> my I, it can hardly control my output and it overshoots. I am
> not using a
> D at the moment. Please help.

Why is the system fluctuating? Is the process changing rates? Are there disturbances to the system? Is the temperature self-regulating? What is the response time of the system? Does your control valve have a linear response in the range that you are operating?

Mark Blunier
Any opinions expressed in this message are not necessarily those of the company.
 
Ting,

Your problem is a PID tuning problem. I can offer free PID tuning service for one or two loops. Please send me about 1000 samples of the PV and CO (also called OP or CV) and tell me the sampling period, the dead-time of your pressure process (i.e., the time period from the time your CO has a change to the time your PV begins to respond to this CO change). Than I can email you the three PID constants. For more info, take a look at

"http://bestune.50megs.com":http://bestune.50megs.com (Section K)

I have offered this service to some people and the results were quite good.

Regards,

Dan::

#108, 10615-111 Street
Edmonton, Alberta
Canada T5H 3G1
Phone: +1-780-424-5074
WWW: http://bestune.50megs.com
Email: [email protected]
 
M

Michel A. Levesque, eng.

I've had basically the same problem (including P,I and D parameter values) as you, so here are my tips for tuning this loop:
1. Start by making sure you have a fast loop update time (100ms or so).
2. To be safe cut down P to 1.2 or so and set I=0
(effectively disabling it)
3. Start adding D. This will crack the steam valve open quickly then close back down with time. Steam valves are usually slow acting, you will probably end up with 0.4 or so depending on the valve action.
4. Next start adding I. This might have to go higher than you had before (maybe I=0.6)
5. Finish off by tuning P for a "stiffer" response.

Overall, I found that I=P/2 and D=P/4 is a good ballpark figure for steam valves using a SLC-5/04 with 100ms loops. Again, this depends on the valve action and the volume of the vessel.

NOTE: If you have a large process lag...all bets are off. Large process lags, especially with steam valves almost always needs advanced process controllers (Smith predictors).
 
L

layton baker

Maybe I'm reading the problem wrong...but it seems like the system was mis-designed to begin with. Install an adjustable pressure regulator on the steam source...and control liquid level via a PID controlled drain valve. The regulator will "self PID" the tank pressure based on simple physics and the PID control for your level will be about as simple a loop as you could ask for.

I'm curious as to the actual description of the system and it's intended use. Is this an SIP process of some sort or perhap a steam let-down/flash drum?

What causes the pressure to fluctuate? What are the flow requirements from the tank, if any? It may also be your line/valve sizes are wrong, in which case your in trouble because your P and I numbers will only be valid depending upon the time constant of your system. In which case, your better off using a fuzzy logic enabled controller from Yokogawa or somesuch rather than the rather crude PID control provided in in SLC's.

Just my two cents...
 
Top