Slightly OT: Multiple PID's acting on same process

OK, looking for opinions here. Let's say I am trying to maintain a constant pressure or tank level using 3 pumps with VFD's. Now instead of buying a PID controller I split the pressure reading to all 3 drives and let their internal PID's do the work. I will have some sort of setpoint controller telling the pumps when to turn off in a lead/lag/laglag fashion. Do you think it would be possible to tune the 3 drives so they will run smoothly, or would it be an erratic mess? Say pump 1 is churning along at full speed and can't keep up with demand. The setpoint controller says OK I need another pump. It kicks on lag and now the lag's PID is telling it to ramp up while the lead PID sees the change and tells it to ramp down... when pressure starts to get close to setpoint what's gonna happen? Could I just pump enough I or D into them to keep them from changing too fast (assuming dependent gains PID)? Or is this just a bad idea and we should buy a PID controller to give all pumps the same speed reference?

if the vsd have analog inputs and outputs you could use one with the internal PID and transmit the speed reference to the other two. if lead pump is 100% then lag 2 starts at 100% also, not so sure how you would load off the system. Some food for thought though.

cheers

myles

Myles said:

if the vsd have analog inputs and outputs you could use one with the internal PID and transmit the speed reference to the other two. if lead pump is 100% then lag 2 starts at 100% also, not so sure how you would load off the system. Some food for thought though.

cheers

myles

Click to expand...

The reason I don't want to do that is if your first VFD faults then the next loses it's speed ref. Also you can't alternate pumps because 1 will always have to be lead.

I take it the three pumps maintain the level in this tank by filling it? and some times one will be enough other it will need the three?
could you feed the pressure into the controller (the lead lag one) use pressure compare instructions and bring on the pumps 100% depending on how low the level went, just forgeting the PID idea?

You could depending on the controller and the amount of digital outputs send speed reference signals to each drive depending on the level, just taking step one a bit further?

I would run only one pump with a PID.

I would run the others using open loop control when necessary.

If the pressure or level feedback are not calibrated exactly the same the PIDs will 'fight' with each other. One integrator will wind up and the other will wind down.

jedft,

In this thread http://www.plctalk.net/qanda/showthread.php?t=30957&highlight=vfd+multiple , Dick DV expains in post #23 that you can run one VFD and use it's native PID loop to control three pumps (two on dol starters) with out a plc. The drive can control the starters through it's relay out contacts. This does require a ABB ACS550 drive. If you are familiar with ABB drives and their manuals, then this is a piece of cake. If you haven't worked with ABB before then my guess is you will need to at the very least have a talk with the ABB rep in your area.

Some of the other respondents to the thread I mentioned may also give you somthing to think about.

Last time I did this, I used one PID (in a PLC) to control the pump speed, with all pumps running the same speed. When the PID CV reaches, say, 98%, start another pump, and ramp up the speed in open loop until it overtakes the PID output, then lock the speeds together. When the PID CV drops below some value, ramp the speed of one pump down to stop while the others stay on PID control. The start and stop values will depend on the pump and system curves.

I seem to recall doing a job some time ago with 5 pumps, 3 fixed speed and 2 variable. The variable element needs to be larger than the fixed step to give you some control range, otherwise you may get holes in the pumping range.

I prefer to run parallel pumps at the same speed if possible. Depending on the installation, the water doesn't always divide between the pumps quite how you're hoping.

AutomaticLeigh said:

Last time I did this, I used one PID (in a PLC) to control the pump speed, with all pumps running the same speed.

Click to expand...

I've done several of these types of jobs with PLCs... this time they want it cheap and dirty without a PLC. I think what I am probably going to do is put bypass starters on all 3 pumps, which is a good idea anyway, then run the lead using it's PID and start lag and laglag full speed. This won't be the smoothest way, but it will eliminate the PID's fighting each other and the headaches of trying to tune that behaviour out of the system, which would probably never be successful. Thanks!

Originally posted by jedft:

This won't be the smoothest way,...

Click to expand...

You will get a smoother transition as you switch pumps on and off by presetting you PID integrator. Preset to 0 when you add a pump and preset to full when you subtract a pump. Assuming all the pumps are the same size this should give you a near zero net flow change at transition.

Keith

Was reading an ABB manual last night (yes I should get out more), this application is a standard one, they have done all the work for you allowing you to cascade the drives and have them working together. It might be worth checking with your VSD manufacturer to see if they have a similar standard setup.

Bryan