Pneumatic valve with pid

Hello Everyone

I'm trying to control pneumatic valve with pid but I have a problem about valve sensitivity. I used FB41 and also set parameters like below.

Cycle time : 200 ms
Gain : 2
Integration time : 10 sn
P_SEL : True
I_SEL : True
COM_RST : DB....
MAN_ON : DB....
SP_INT : MD....
PV_IN : DB....
MAN : DB....
TI : 10 sn
LMN_HLM : 100
LMN_LLM : 0
ENO : DB....

First of all the value was calculated for sp_int and compared with PV_IN. During process, output of FB41 was changed to % 0 and % 100. There was no value between that. Rarely I noticed different values. Because of that, pneumatic valve was unstable and always valve's setpoint was changed(%0 and %100). How can I change that to make the valve work more stable ?
In addition to that am I supposed to use another block for control valve?

Any comments are welcome.

This is a very difficult problem to answer without knowing more details of the process, the sensor(s) and actuator.

If we were sure that the only source of problem lay in the PLC PID block,
then the symptoms you describe (0%<-->100%) are quite typical of excessive gain (unfortunately they could also be accounted for by incorrect sampling interval and other factors)
else ...

First try the valve with the manual control bit set true.
Can you accurately control it throughout the range of movement by directly writing a value to the output address (QW...)?
Can you do the same by writing a value to the LMN of the PID block?
If either of these fail to give smooth continuous control, you need to attend to the output scaling, or ranging of the output module, or other issues before tackling the PID.
Still in manual, does the PV accurately reflect what you expect to see when you alter the LMN? Are you getting direct feedback of the valve position, or are you relying on another variable (flow, pressure ...)?

The block FB41 is widely used and well proven. The problem won't be with the block itself. It may be with the parameters you're feeding it, but it could also be with these 'external' issues. You need to eliminate these first and work inwards towards the block. Once you've proven everything else, switching the block in to automatic and tuning it correctly should be the final part of the problem, not the starting point.

Ken

Pneumatic Valve with pid

Dear Sir

I've just checked and results are given below.

Sensor type : Electromagnetic Flowmeter ( Endress )
Actuator type : Pneumatic Valve ( Foxboro )

I use Pid_output as an input of Unscale Block(FC106) and there is no problem. it was changed normally.
Range of output module : 0-27648
Pv_in : Direct
Sp_int : Variable
I can change LMN_HLM or LMN_LLM without any problem.

Any comments are welcome.

I use Pid_output as an input of Unscale Block(FC106) and there is no problem. it was changed normally.

Click to expand...

Does this mean you have solved your problem?
Did you then pass the unscaled value from FC106 to the valve and everything worked OK? What values did you use as parameters for FC106?
I'm not clear about 'Pid_output' - this doesn't actually define a specific parameter of the FB41 block. Did you use FB41.LMN as the input to FC106? If that works OK, you should be able to use FB41.LMN_PER to pass the unscaled peripheral value directly to the valve without FC106.

There are lots of other questions could be asked here. Are you calling FB41 from a cyclic interrupt block set at 200mS? Applying a value of 200mS to FB41 won't make it run at that interval. That simply informs the block what time interval it is actually being executed at.

Ken

007NN said:

Pv_in : Direct

Click to expand...

I've no idea of the PLC or syntax used here, but I'm wondering if the loop needs the opposite 'action', for instance, reverse action, rather than direct action.

The symptoms correspond. The wrong 'action' will drive to an output limit and stay there.

Dan

Thanks for the suggestion, Dan. That was one of the issues I was saving for later. I know that FB41 is a PID control block in a Siemens S7 PLC. Unfortunately, although the block does have a formal input parameter called PV_IN it expects to receive a REAL value (i.e. the process variable). I have no idea what 'direct' means in this context. One thing of which I am sure is that there is absolutely no way that 007NN could have applied it to a parameter named PV_IN. It is so far off the mark I couldn't even guess what to ask.

Ken

Well direct or reverse action is determined by the sign of controller's gain. I believe that in this context "Direct" means that PV is going directly to summation block without scaling blocks like shown in the attached picture, but one cannot really know for sure....
http://www.plctalk.net/qanda/uploads/Direct_PID.jpg

Back in Texas Instrument days (late 1980's), TI defined reverse/direct action the opposite of ISA's definition. Drove me nuts until I figured it out. I've always been on the alert for inverted implementations, since.Dan

I think it is good to define some kind of convenction and stick to it. A year ago I made one post about similar issue. You can find it here

OK guys we're all agreed that using direct action where it should be reverse action and vice versa is a bad thing. But how does that account for the symptoms in the original post? I agree with DanW that I'd expect incorrect selection of action to push the valve to one limit or another and then stick there. It would be an incredibly stable system. Wrong, but stable. The problem described by the OP was instability. I take that to mean he is seeing bang-bang control 0%-100%-0% with no real proportioning taking place. I don't understand how incorrect action would produce that symptom.

Ken

Ken, you're right that the wrong action won't create a symptom of 0 to 100 to 0 to 100 . . . .

Your first suggestion, excessive gain is the most likely. It's likely a fast flow loop where a gain of 2 is producing quasi-on/off response.

Dan

Pneumatic Valve

I am so sorry for late reply. Now it works properly. I've just changed FB 41's parameters as below:

OLD VALUE NOW

CYCLE 100MS 50MS
GAIN 3 0,001
TI 10SN 1SN

All suggests are welcome, thanks.