PID for proportional air valve

Hello everyone.

I have a proportional air valve FESTO MPPE-1/4-420-B. It controls pressure for a spray gun. The operational pressure is 800 mBar. The gun consume about 2 l/sec. During the spaying the pressure drops sometime down to 600 mBar, sometime even to 400 mBar. I tried to use a PID. That valve has internal feedback. My PID doesn’t work. Too much oscillation. Am I missing something in PID configuration?

Thanks.

Do you have a way of knowing that the gun will be on/off?
If so, feedforwards can help.

Are you sure you have a good supply? If not can you add a surge tank? This could help with the pressure drops and help with the PID. Anytime I think of oscilations I think of too much P gain. Feedforwards are added into the PID loop. We have had many many post here about PID loops and feedforward gains. Do a search and then if you still have questions post them.

What is controlling the PID loop?

<<What is controlling the PID loop?>>

It is a good question. I direct the PID control variable to the analog output of proportional valve and it is not what I would like to have. For example at start point: SP = 800, PV = 0, SE = 800, Kc = 1.0, Ti = 0, Dt = 0 and Control Output CV = 100%. It switches the pressure to the MAX = 6000. Just right after, it switches the pressure back to 0.

I did several PID to temperature control that works fine, but I am missing something with pressure control.

IMO, you are working with such large numbers you need to decrease your gain. Your gain is a direct representation of your error. You need a gain of about 0.01 and maybe go from there, and of course you can start playing with a little Integral to overcome any offset. There are alot of articles on here to tune PID loops. Also you can look at delaying the PID activation for a set time, during that time open the valve to 50% first. Lots of things you can try, high pass filter etc etc.

There have been a lot of good suggestions so far.

Use feed forwards. When the spray gun opens you say that it releases 2 liters / second. The valve must supply the 2 litters per second. How far must you open the Festo valve to supply 2 liters per second? When the spray gun opens, open the festo valve the supply the same amount.

A feed forward will be off a little but that shouldn't be a big deal because the system should reach an equilibrium. If the pressure goes to low the flow through the gun should drop to and the Festo valve should be able to catch up until the in flow equals the out flow. If the pressure goes too high the flow through the gun will be more than 2 liters per second and the pressure will drop again. You need to adjust the feed forward output until the equilibrium is reached around 800 mbar.

If the equilibrium can't be maintained then you have a supply pressure problem. Getting a PID to correct for this will be tricky because pneumatic systems are non-linear and respond slowly. Are you sure you can supply 2 liters per second? The open loop test will tell you.

If there is a supply pressure drop then you will need some sort of supply tank so the pressure will not drop too much but it will still drop. If the supply to the tank still isn't able to keep up then you will need to reduce the flow through the spray gun.

I would get the feed forward working by itself without the PID. You should be able to get close. Then use just a little proportional gain.

<< You need a gain of about 0.01>>

Well I started from Kc = 0.01. SP = 800, PV = 0, SE = 800, Kc = 0.01, Ti = 0, Dt = 0 and Control Output CV = 1%. It set the valve about 8 mBar and doesn’t go any further. The valve has own pneumatic feedback and it need only set point to maintain the pressure in few sec.

<< Feedforwards are added into the PID loop. We have had many many post here about PID loops and feedforward gains. Do a search and then if you still have questions post them.>>

How to add the feedforward to PID? I couldn’t find any room in PID configuration. Is it separate ladder instruction? As far as I understand, in my case the feedforward or bias should be equal my set point (800) and PID should react for sudden drop when gun opens. Is it correct?

<<Use feed forwards. When the spray gun opens you say that it releases 2 liters / second. The valve must supply the 2 litters per second. How far must you open the Festo valve to supply 2 liters per second? When the spray gun opens, open the festo valve the supply the same amount.>>

From FESTO valve specification I found it can handle 2000 L/min. I need only 120 L/min. It self controlling valve, it supposes to maintain the set point pressure and machine came without any feedback control. Our production is not happy, because of drooping pressure during the spraying. Spraying cycle is 4 sec. It was my idea to add a PID. It looks like a wrong way. Peter, the valve works like, if I increase the output from 4 to 20 mA it increases the pressure from 0 to 6000 mBar. What do you mean: open the FESTO valve?

Alex Pel said:

How to add the feedforward to PID? I couldn’t find any room in PID configuration. Is it separate ladder instruction? As far as I understand, in my case the feedforward or bias should be equal my set point (800) and PID should react for sudden drop when gun opens. Is it correct?

Click to expand...

Yes. If the Festo valve can supply 16 times the desired flow then the valve is TOO BIG!. I can see why it would be difficult to control. It is like just touching the gas pedal and the car goes 100kph! The Festo valve controls the rate of change in pressure. You need to be able to more finely control the rate of change.

Keep in mind that Alex Pel is using an MPPE series valve, which is a pressure regulating valve. If he sets a pressure command the MPPE should do everything in it's power to reach that pressure. If it can't reach that pressure either the supply isn't there to do it, the flow of the MPPE has been exceeded or the pressure element in the MPPE is actually measuring the right pressure but the pressure has dropped by the time you see it on a guage.

I think that the MPYE series valves are more similar to the valves Peter deals with in hydraulics. The MPYE series is a dynamic flow control valve.

Since you can get this thing to oscillate the issue probably isn't supply pressure or flow. It is most likely that the pressure element in the E/P is measuring the correct pressure and you are getting line losses at higher flows. 800mBar isn't much pressure. It is pretty easy for flow effects to look like this pressure level.

I think you may need a hybrid system. Command the baseline pressure as you were doing originally. The add the output of the lower gain PID to this baseline value.

Any flow based feed forward term needs to be the projected pressure loss in the components between the MPPE and the final element. This pressure increase at the E/P should compensate for the losses you are seeing.

Keith


NOTE: I just looked at the MPPE specs and it looks like it will accept an external pressure feedback signal. This may be a way for you to allow the valve to compensate for losses on it's own.

kamenges said:

Keep in mind that Alex Pel is using an MPPE series valve, which is a pressure regulating valve. If he sets a pressure command the MPPE should do everything in it's power to reach that pressure.

Click to expand...

Alex shouldn't even need a PID or FF then. The valve is still oversized.

kamenges said:

If it can't reach that pressure either the supply isn't there to do it, the flow of the MPPE has been exceeded or the pressure element in the MPPE is actually measuring the right pressure but the pressure has dropped by the time you see it on a guage.

Click to expand...

You would think Alex would have told us if the supply pressure droops. In this case a tank on the outlet of the Festo valve would help minimize the pressure changes and also reduce the sensitivity of the too big Festo valve since it will take more flow now to change the pressure at the gun and tank.

I agree there is no excuse for not maintaining pressure unless the supply can't keep up. So far we have not been told there is a supply problem.

kamenges said:

NOTE: I just looked at the MPPE specs and it looks like it will accept an external pressure feedback signal. This may be a way for you to allow the valve to compensate for losses on it's own.

Click to expand...

To do this right you need at least 10 pipe diameters from the sampling point to any elbows or pipeline diameter changes. 20 pipe diameters would be better.

Ok. Now I removed PID and add some logic: When gun is open I add 200 mBar, I call it feedforward (correct me if I still misunderstanding this term) and also add the error like SP-PV. It works much better: within 700 – 900. Before it drooped down to 300. I think the production guys will be happy.

Thank everybody.

Originally posted by Alex Pel:

Click to expand...

When gun is open I add 200 mBar, I call it feedforward (correct me if I still misunderstanding this term) ...

Click to expand...

I think technically this is disturbance rejection as opposed to feed forward. Feed forward would be based on the setpoint value alone. But the end result is the same regardless of name. You accounted for a known loss before it had a chance to create an error, which is the ultimate goal of both feed forward and disturbance rejection.

Ideally you would be able to quantify what is causing your loss in the first place. This will help you if something in the process changes in the future and the problem returns.

Keith