Regulating pump output with load-cells

Hello,
we are a group of students doing a school project, who stumbled upon a problem with regulation.



We have to automatize the part of the system from the picture above. The input is delivered into the funnel that leads to the pump. The funnel's weight is measured by load cells below.

The idea is to regulate the pump's frequency output according to the funnel setpoint weight with PI controller.



So basically when level of liquid (calculated from weight) is lower than setpoint the pump will slow down and on the same principle it will speed up when liquid level starts to exceed the setpoint value.

Can this kind of regulation be applied to the system?
Is it a good practice in real life?

If not, can we regulate this system according to the PT (pressure sensor) that we have after the pump?

We will be extremely happy to get your feedback.

Kind regards,
Hubert

Dependent on what you are pumping , you are on the right track. But you maybe better of using volume as your method. You would then have some form of transducer measuring the volume by looking down from the top, then regulating the speed via the 4-20ma coming out of the transducer to ramp speed up/down of the VSD on the pump.

The pressure transducer is sensing the output pressure and depending on what the output is doing(whether keeping pressure constant is important for the process) it would speed up/down pump via the VSD. But this does not watch/control what is coming into the hopper, so it could overflow or run dry. The weight sensor of course is a good way to go, but not maybe a normal practical approach, as the cost is so much higher than just a transducer looking in from the top. But this is all conjecture dependent on the process.

Thanks for your answer Duckman, we have also considered the level sensor on the top, however this is how we recieved the project from the company that made the whole plant.
I think we will send them an email if level sensor can be applied instead.

More than likely, load cells would be the preferred method of measurement here for many reasons...

First, assuming that the material is the same and therefore has a known specific gravity, volume and weight are fixed proportionally. So, you know the true volume at all times from weight.

Second, top down sensor in most cases gives you linear volume unless you build a strapping table to account for the irregular shape of the vessel. May or may not be important in the application, but one reason NOT to use a top down sensor.

Third, top down sensors are either non-contact (laser, ultrasonic) or contact (radar, capacitance), both of which have their inherent problems. Non-contact sensors are subject to environmental AND tank wall effects. Dust, smoke, steam, etc. can effect the readings of top down sensors. Small diameter tanks are historically poor applications for sonic style sensors. Contact sensors have their own issues with placement in the tank (agitators or mixers in the way) or product buildup on the cable/rod over time.

So, unless there's a reason NOT to use load cells, in most applications like this load cell is the preferred method as it is non-contact and not subject to environmental or tank configuration issues. The only caveats with load cells are that the material weight to volume ratio doesn't change so much that there's danger of overflow (which is why a secondary high level switch is required whether it's load cells or top down sensor) and that you have to isolate the tank from surrounding support members (piping for example) to get accurate readings.

1)Is TT, a temperature sensor on the substance? and if it is why?
2) Is the pump centrifugal or displacement?
3) Which leads to this question, what is it you are trying to pump, a liquid or more of a thick slurry?
4) Looks like a valve of some sort on the output, why?
Their would be lots of questions that would needed to be answered to continue on to a satisfactory control of your process.
Regards

Hi Robertmee, l am aware or the issues you raise, but without knowing the process, l don't believe the first choice would be weight, given the costs involved, compared to the easy of top down measurement, Scaling the transducer is not to complicated and in there diagram isn't required.

ask the company if they want to control the pressure in the discharge pipe, or the level in the tank
if level in tank you rproposal is good,otherwise use the pressure sensor
Think how you connect the inputs to your controller.

duckman said:

1)Is TT, a temperature sensor on the substance? and if it is why?
2) Is the pump centrifugal or displacement?
3) Which leads to this question, what is it you are trying to pump, a liquid or more of a thick slurry?
4) Looks like a valve of some sort on the output, why?
Their would be lots of questions that would needed to be answered to continue on to a satisfactory control of your process.
Regards

Click to expand...

1) It is to monitor pump's temp.
2) Positive displacement mono pump
3) It is a part of waste water treatment plant where solids are filtered out from water claryfier then taken through centrifuge which filters out the remaining water from our input.
4) No idea.

Are you measuring sludge or water going into the hopper?

Sludge

Hi hubert, it is now your turn to ask some questions, after l ask.
1) Is it still required to operate this via a load cell?
2) Is the "sludge" going to be agitated in the hopper (to keep it in suspension), as it may settle if it sits, which l believe would be an issue with a load cell.
3) What was the reason for the pressure switch, are they putting this thru a cyclone (centrifugal separator) to remove the heavies, which seems the most likely reason, then the pressure switch needs to be involved with the control.
Just reread your answer to my previous and now see l was correct and hence the control needs to incorporate the pressure transducer.
You may now have enough info to start your programming, if this is your next stage.
Regards

A pump controlling the out-flow is the simple case of tank level/weight/volume control. A simple proportional band will work if the level can vary within a range. PI control will control the level exactly in a steady state situation.

duckman said:

Hi hubert, it is now your turn to ask some questions, after l ask.
1) Is it still required to operate this via a load cell?
2) Is the "sludge" going to be agitated in the hopper (to keep it in suspension), as it may settle if it sits, which l believe would be an issue with a load cell.
3) What was the reason for the pressure switch, are they putting this thru a cyclone (centrifugal separator) to remove the heavies, which seems the most likely reason, then the pressure switch needs to be involved with the control.
Just reread your answer to my previous and now see l was correct and hence the control needs to incorporate the pressure transducer.
You may now have enough info to start your programming, if this is your next stage.
Regards

Click to expand...

1) Still waiting for answer
2) I belive it will be constantly agitated because it is a continous process.
3) Sorry but I was wrong with the centrifuge, it is a drum thickener instead. This is the exact device that produces the input: http://www.dwe.dk/product/KD30/KD%C2%A030%C2%A0Drum%C2%A0Thickener

This layout is very common and is called a Loss-in-Weight controller. There are commercial versions available, or you can program your own LIW controller using the PID instruction in your PLC software. The LIW control function is based on calculating the rate of change of the tank weight. That gives you the output rate of the pump, expressed in weight change per second, or converted to volume change/time.

The change of the tank weight is calculated by storing the previous weight, then subtracting the current weight. (Weight Old - Weight New)/Time = Rate of Weight Change). Select a time period to match the accuracy needed for the project. If the weight changes slowly (small pump volume), then use a longer time period. If the weight changes rapidly, use a shorter time period.

The refill of your "funnel" from the Input will cause a process upset. You have to stop the Loss of Weight calculation while the refill is being done, then reset the Old Weight = Tank Full Weight.

I have several examples of a LIW controller using Allen Bradley RSLogix 500, if that would help.