Controlling a valve with backlash

Thanks for all the help!

I think I'm going to write a positioning routine that keeps the valve on the open side of the backlash. It will take a requested position and compare it with the current commanded position. If it's less, it will set the commanded position directly to the requested position and start a timer that will give it time to finish moving. If it's greater it will set the commanded position to the requested position plus 10% of full range, start a timer to give it time to complete the move, and when that timer finishes it can set the commanded position to the requested position and set another timer to give it time to finish.

That takes care of the backlash, so I can either use the built-in PID loop or roll my own proportional loop to determine the desired position from the current position and the difference between the setpoint and the measured water flow. It can run fairly slowly, but in any event it won't be allowed to run when any of the valve movement timers are timing since it's not in an equilibrium state yet.

If that doesn't get it, then I'll probably have to resize the valve. There don't seem to be a whole lot of choices that are bigger than 1/4" tubing but smaller than 1/2" pipe, but there's got to be something suitable.

Water flow is changed based on the consistency of the extrusions, but it takes 8-10 minutes from when the water flow is changed to when the consistency of the extrusions change, since it changes the water in the extruder mixing chamber and that material takes a while to make it through the augers and get pushed out through the grids. The hardest part is training the operators not to overcontrol the water - they want to use water to control the extruder motor amps (which is not what they're supposed to control with the water in the first place), and they want to make a change, wait a few seconds, and if it doesn't affect the process enough they want to change it more. If the water valve control runs slowly, it's OK - the big challenge is going to be training the operators to give it time to work.

aley said:

Water flow is changed based on the consistency of the extrusions, but it takes 8-10 minutes from when the water flow is changed to when the consistency of the extrusions change, since it changes the water in the extruder mixing chamber and that material takes a while to make it through the augers and get pushed out through the grids. The hardest part is training the operators not to overcontrol the water - they want to use water to control the extruder motor amps (which is not what they're supposed to control with the water in the first place), and they want to make a change, wait a few seconds, and if it doesn't affect the process enough they want to change it more. If the water valve control runs slowly, it's OK - the big challenge is going to be training the operators to give it time to work.

Click to expand...

Heh, good luck with that. I was wondering about the amps as a measured/controlled value. If you *did* try to control on amps, the deadtime, and D of the PID, might just be so big that the backlash would look like slop, and get lost, in the deadtime. This would be a cool process to model.



That said, I would guess that the amps ramp up very quickly at some point with decreasing water fraction, and that would be very difficult to tune well: once the amps get to that (inverted) knee of the curve it's too late.


Does the process and/or production rate ever change? You could perhaps add feedforward on the auger speed.

drbitboy said:

Heh, good luck with that. I was wondering about the amps as a measured/controlled value. If you *did* try to control on amps, the deadtime, and D of the PID, might just be so big that the backlash would look like slop, and get lost, in the deadtime. This would be a cool process to model.

That said, I would guess that the amps ramp up very quickly at some point with decreasing water fraction, and that would be very difficult to tune well: once the amps get to that (inverted) knee of the curve it's too late.

Does the process and/or production rate ever change? You could perhaps add feedforward on the auger speed.

Click to expand...

The amps on the extruder motor depend on a bunch of factors - the delivery rate from the box feeder, the amount of recycled dust, the moisture content of the incoming clay, and of course the amount of water being added.

The biggest factor is the rate from the box feeder, which depends on the consistency of the mined material - we're discussing adding a crusher to get better homogeneity of our incoming material, but that's an expensive addition that doesn't fit well in the existing plant.

I've daydreamed about controlling auger speed to even out the flow of material into the rotary drum dryer downstream of the extruder. We have a system installed on the conveyor that feeds the extruder that includes a load cell under one of the trough rollers, but it's not been operational in a number of years, and I have doubts about whether it can accurately weigh the few pounds of material on a very heavy moving belt. The signal would be buried in a whole lot of noise.

I've also looked at systems to measure the moisture content of the clay before and after the extruder to try to calculate the correct amount of water to add, but it comes back to the problem of not really knowing the weight of the material going into the extruder. Non-contact systems also run north of $15k, and since our incoming moisture content varies we'd really need two, one before and one after the extruder. When I took the $30k price tag to the plant manager, he shot it down, but I may be able to get it into the capital budget one of these years.

I put in a little code yesterday to keep the valve on the same side of the backlash, and it seems to be functioning pretty well. I haven't had time to put in a control loop yet; hopefully by the end of the week. The more I want to work on the dryer controls, the more I end up having to do troubleshooting and repairs on the transporter systems - Murphy has had fun with me lately.

aley said:

I put in a little code yesterday to keep the valve on the same side of the backlash, and it seems to be functioning pretty well.

Click to expand...

Oooh, do (show and) tell ...


Please.