Solving over-run in a liquid batch system

I need some help in how to solve over-run/under-run correction in a liquid batch system. Thanks for any help.

Please more precisely define your system and your problem. This will relieve us from a long series of guesses based on your one sentence description.

The best way to implement overrun or underrun correction is to get it right the first time so you don't need to correct.

Steve Bailey said:

The best way to implement overrun or underrun correction is to get it right the first time so you don't need to correct.

Click to expand...

Though this is not always possible. But with liquids it helps if you have 2 valves. One normal one, and one with a orifice in it.


If its a weight dosing system (load cells).
If the deviation of the dosing amount and actual amount is consistent its fairly easy to make a correction for it. At the end of the dosing calculate the difference. At the next dosing you would add or subtract this from the dosing amount. I would limited the amount that is corrected per dosing for example only 30% of the difference, see it as a rough PID control. Also i would set a max amount that can be corrected.

Sorry I have micrologix 1000 (1761-L1BBB) that uses a HSC function that counts pulse's from a flow meter that is rated for 200 pulse's per gallon. Then I have a Redlion graphite HMI as the operator and calibration interface. This is a new program that I am working on and I am having a problem solving the logic for over-run and under-run conditions, especially when I am working with three different liquids to be scaled into a mixer.I'm thinking I need to take my rate/time and then compare it with a fixed value in regards of what it should be and then change timing on the output.

When you get close to your desired volume, slow down the pump. If you can't change the speed of the pump, switch to a smaller valve, preferably one that includes an adjustable flow restriction.

You need a control system with feedback.

It's easy. You just adjust the setpoint by taking a percentage of the difference between you want and what you got (the error).

Next time you weigh something you take the setpoint you want and subtract the adjustment. After you weigh something next time you get a certain error, you take a percentage of that, and on and on. Finally the system will have corrected itself. Problem solved.

Have a different correction for each media type or valve.

You should also think about doing pulses when getting close to the setpoint - unless you have a proportional control valve. So open for a short bit wait, close and then check the weight and repeat until weight is right.

If you use proportional control you need a minimum output otherwise you might get zero or very low little flow and never reach the setpoint.

You can combine all these methods.


.

There is only one solenoid valve on the water system to be scaled
and on the corn syrup and soybean oil to be scaled uses a pump and a solenoid on those systems.
These are an existing systems that use an SJ batch controller but these units are going obsolete and I want to get them converted over before this happens. I know there are other batch controllers out there but what I have seen doesn't work for the operators because either they are to small or not visible from a distant.
On the sj controller it uses multipliers and dividers for accuracy and a zero floating point to correct over-run problems

picture of homework

One word, PREACT. Make an adjustable preact that you subtract from your setpoint. This is the amount of ingredient that flows while the valve is closing. Then have a timer that starts after the valve is de-energized. This timer should be slightly longer than the time it takes the valve to close. Keep totalizing until the timer is done. With practice you'll be spot on in no time.

JaxGTO said:

One word, PREACT. Make an adjustable preact that you subtract from your setpoint. This is the amount of ingredient that flows while the valve is closing. Then have a timer that starts after the valve is de-energized. This timer should be slightly longer than the time it takes the valve to close. Keep totalizing until the timer is done. With practice you'll be spot on in no time.

Click to expand...

+1

And, if you want to get slick, you can calculate the error after a dispense cycle is complete and automatically correct your pre-act value. It may be useful to do this on a limited basis or only if the error exceeds a limit for multiple consecutive cycles.