Valve control: Fast Close, Slow Open

I don't do enough PID loops to ever get good at them. I've been pondering over this one for a couple days:

The level of a waste tank is to be controlled by a gas valve. Water and waste come in with the gas which collects in the tank. The gas will flow up and back out of the tank. Sepeartely, the tank processes/dumps water out of the bottom through an on/off valve.

Right now, this gas valve is being bypassed. Gas is flowing freely at all times. The tank has hi and low level switches. The hi level swith stops the process completely and requires the operator to manually drain the tank. The low level switch closes the valve on the bottom of the tank to keep it from emptying completely.

This new gas valve is supposed to regulate gas flow so that the water level never reaches these hi/low switches under normal conditions. They want the valve to begin to pinch off and close when the tank gets to 50% and be completely closed at 80%. They want this to happen quickly. But as the tank drains, they want the valve to open back up extremely slowly.

I am having two problems while trying to implement this: the system was set up with a PID, but I'm not sure how to make a PID behave slowly on close and fast on open. Is this even possible?

The second problem is that they want the system to operate over this 50-80% range, but my PID loop seems to want to operate over the entire range of the tank level. I could do this without a PID loop, but they want to be able to set a setpoint (presumably between 50-80%), so purely proportional control doesn't seem to be an option.

Maybe someone here can nudge me in the right direction. Also, I'm sure there are things that need clarification in my description - let me know what other info is needed.

Sounds like an air lift pump application. I have a couple of suggestions.

First, many PID controllers and PID loops in PLCs have a bias setting to make opening and closing operate differently.

Second, maybe a canned PID isn't really what you need. Try a proportional floating control. If the level is above 50% modulate the valve to close, eihter with the change in position proportional to error or in steps, continuing to step closed until the level is OK. You can do the same on low level, with the steps smaller.

This is trivial. I agree with Tom that a PID probably isn't what you want. It doesn't look like you have a set point anyway. I would use a SCP instruction to map 50 to 80 percent level to 100% to 0% open. The output of the SCP is then compared to the current valve position. If the SCP output is higher than the valve position then add a small number to the valve position. This will open the valve slowly and approach the desired output indicated by the output of the SCP instruction slowly. If the output of the SCP instruction is lower than the valve position then subtract a bigger number so the valve will close more quickly.

Hopefully your PLC has some sort of scaling instruction.

Peter Nachtwey said:

This is trivial.

Click to expand...

Only if you know what you're doing.

Peter, your method was exactly what I thought to do first, but these guys really seem to like their setpoints and PID's. I'm just going to try and insist on this; thanks guys for your input.

J E said:

Peter, your method was exactly what I thought to do first,

Click to expand...

Then you had a pretty good idea of what you were doing.

but these guys really seem to like their setpoints and PID's.

Click to expand...

Did "these guys" tell you what the set point should and or how to make the PID control in the shut direction faster than the open direction? If not then they have no idea what they want and just want to dump it on you.

I'm just going to try and insist on this; thanks guys for your input.

Click to expand...

If they insist the make them come up with all the answers.

I know how to make the valve close faster than open. It takes two set of PID gains. You would need to use the incremental or velocity form of a PID. If your PLC doesn't have that then it would need to be written from scratch.

Peter Nachtwey said:

I know how to make the valve close faster than open. It takes two set of PID gains. You would need to use the incremental or velocity form of a PID. If your PLC doesn't have that then it would need to be written from scratch.

Click to expand...

Yes, the PID (Logix PIDE) seems to use the velocity algorithm, which now explains why it wasn't behaving as I expected when testing it.

Just for my future knowledge: would I just have to have some logic monitoring the valve movement external to the PID and adjust the gains on the fly based on this?

You better have some sort of feed back on the valve and the waste level.

You would need to compare the SP and PV and change the gains depending on which one was greater.

Another trick is to calculate the error=SP-PV if the error is positive multiply the error by 0.1 or something small so the valve will not open up as fast. This will mess with your proportional band calculations.

This is Logix5000, so the analog modules have a Ramp parameter you can change on-the-fly.

You will need to know if you are opening or closing (compare SP to PV for direction), and you will need to use a "Module Reconfiguration" message each time you want to change the ramp rate.

Of course, limiting the analog out will upset the PID, it will integrate a larger error when you slow the response of the valve, so you may have to have two sets of PID terms as well

Did "these guys" tell you what the set point should be?

Click to expand...

Of course, limiting the analog out will upset the PID, it will integrate a larger error when you slow the response of the valve, so you may have to have two sets of PID terms as well

Click to expand...

Yes, as Tom, Peter, me, and all think, forget the PID and K.I.S.S. You don't need it, you have no Sepoint to control, you will have to do a bunch of fancy stuff just to be able to use the PID, so why make your life difficult for no noticable gain?

Agree with daba. Ramp generators will do this in a simple way