Closed Loop vs. Open Loop in Flow Question

jethridge said:

My last question... We are also planning on adding this second pump in another piece of equipment, but this equipment will have a different physical configuration. This will include a holding tank. The system configuration would look like:

Pump 1 -> Flow Meter -> Pressure Transducer -> Tank -> Pressure Transducer -> Pump 2

I've proposed that this implementation would require a flow meter after pump 2 to be classified as closed-loop. Is this correct?

Click to expand...

It depends on what type of loop you are trying to create. If you are trying to control the level of the tank and the pressure transducer measures the pressure in the tank, then you could create a closed loop for tank level control. If you are trying to control the flow rate of the pump then you would need a flow meter to have a closed loop flow control.

However, you have to think of what will happen if you control the inflow to a specific flow, and the outflow to a specific flow, but they don't match, the tank will run over or run empty. You'd be better to control the level of the tank, and not control the flow. If you need understand what is critical to the process. Does the process need specific flow out of this system, or does the next system need a specific flow to it, or does it only need to have pressure available to it, and it will control the flow that goes into it.

mk42 said:

For my sanity, as someone who has had a long time since ME classes in college, and little practical experience with pumps and pumping processes:

How is flow in = flow out a conditional statement? Under what conditions could more/less water leave the pump than enter it? Am I misunderstanding the whole conversation?

I guess leakage is one instance, but I would hope that pumps don't leak enough water outside the system for that to be a measurable variable one needs to consider.

Click to expand...

I wouldn't classify the second pump as a common pump. It's 2 pistons (or 3) that operate at a pressure of ~3500 PSI and mix product as it pushes it downstream. If there isn't enough stuffing pressure to this pump, cavitation can occur inside the pistons and affect the flow out of the pump.

If you adjust the set-point of the first pump, take an snapshot of the system directly after, the flow meter just after the first pump will read a different flow than the flow meter after the 2nd pump.

jethridge said:

I wouldn't classify the second pump as a common pump. It's 2 pistons (or 3) that operate at a pressure of ~3500 PSI and mix product as it pushes it downstream. If there isn't enough stuffing pressure to this pump, cavitation can occur inside the pistons and affect the flow out of the pump.

If you adjust the set-point of the first pump, take an snapshot of the system directly after, the flow meter just after the first pump will read a different flow than the flow meter after the 2nd pump.

Click to expand...

Gotcha, interesting.

proof said:

It depends on what type of loop you are trying to create. If you are trying to control the level of the tank and the pressure transducer measures the pressure in the tank, then you could create a closed loop for tank level control. If you are trying to control the flow rate of the pump then you would need a flow meter to have a closed loop flow control.

However, you have to think of what will happen if you control the inflow to a specific flow, and the outflow to a specific flow, but they don't match, the tank will run over or run empty. You'd be better to control the level of the tank, and not control the flow. If you need understand what is critical to the process. Does the process need specific flow out of this system, or does the next system need a specific flow to it, or does it only need to have pressure available to it, and it will control the flow that goes into it.

Click to expand...

The critical parameters are the flow rate into the tank, and the flow out of the second pump. The tank's level doesn't matter, as long as it isn't empty or full. The flow into the tank must be a certain value so enough bacteria in the product is killed, and flow out of the 2nd pump must be a certain value so certain sized bottles can be filled.

My rough draft for this was a flow meter after pump 1 for feedback to control its flow, a flow meter after pump 2 for feedback to control its flow and implementing additional control around a range of measured pressures from the 2nd pressure transducer that correlate to a nearly empty or nearly full tank.

EDIT: These bottles will be filled based on a timer that controls a valve, and there won't be an instance where flow will deadhead.

[Tank problem]
Technically, you could make it closed loop of the flow, by using the flow meter 1 and the two pressure sensors to measure the flow through pump 2.

This wouldn't be good practice though, and you would be right, if you wanted better control you could use a second flow meter.

It would be open loop if you said "run the pump at 30Hz for 10L/s flow."

Also of note here, you have open loop control of your bottle level.

[Closed loop control in general]
There are three major components to a closed loop control system, as well as the links between them.
The control algorithm, the physical process, the measurement of what you are trying to control.
Your measurement must feed into your algorithm, which must feed out to your process, which is measured, etc.

If the measurement isn't very good, it is still closed loop. Do not think that poor measurement means it is not closed loop control. This simply means that it is poor closed loop control.

Also, just because you are not directly measuring something, doesn't mean you are not measuring it. Let me give you two examples.
Ex1. You measure the flow out of a tank by measuring the flow in, the pressure in the top of the tank and the pressure in the bottom of the tank. The calculation is then flow out = flow in - d(P2-P1)/dt / (density * 9.81) * vessel cross sectional area.
Ex2. You measure the flow out of your tank with a flow meter downstream of your pump. The calculation is then flow out = the value in the 32bit float = the result of scaling the raw 12bit value to 0-100L/s. The 12 bit value is made up of the result of the DAC chip measuring the voltage across a 150 Ohm resistor which will be proportional to the 4-20mA signal and any noise and cable impedance effecting it's rate of change. That 4-20mA signal is the output of some chip which is measuring the resistance of a variable resistor which is touching the diaphragm of a differential pressure sensor. The position of the diaphragm is effected by the pressure on its two surfaces, which is effected by the pressure across an orifice plate, which your liquid is flowing through. Your flow meter manufacturer has worked out the relationship between flow rate of a particular fluid and pressure across his orifice, and has used that value to write the flow rate to mA conversion factor in the manual.

mk42 said:

For my sanity, as someone who has had a long time since ME classes in college, and little practical experience with pumps and pumping processes:

How is flow in = flow out a conditional statement? Under what conditions could more/less water leave the pump than enter it? Am I misunderstanding the whole conversation?

I guess leakage is one instance, but I would hope that pumps don't leak enough water outside the system for that to be a measurable variable one needs to consider.

Click to expand...

This is absolutely the key question. The flow rate for both pumps, if they are in series as shown, will be equal. Refer to the Law of Conservation of Mass. Unless the diagram is in error this is a questionable system, and closed loop versus open loop is besides the point. Are you sure the second pump isn't intended to boost pressure?

Not sure where to begin here....

Let's start with your initial process description. I understand you have 2 pumps in series. The first one is a progressive cavity pump, which is a positive displacement type pump. It is pumping a low viscosity, water-like liquid.
The 2nd pump is fed by the first and is a 'piston type' pump; piston pumps that I'm familiar with are all positive displacement (PD) type pumps.

Having 2 PD pumps operate in series is simply a bad idea (unless there's something peculiar to your process that I don't know). I would think the first pump should be a (far cheaper) centrifugal-type pump as this would be used to control the suction pressure of the second pump. (As I see it, this is more a process design problem than a control problem).

As for your second configuration that has a tank, you will need to concern yourself with controlling the level as most level processes are not self-regulating (if you don't mind it, it will simply empty or overflow).

Tom Jenkins said:

This is absolutely the key question. The flow rate for both pumps, if they are in series as shown, will be equal. Refer to the Law of Conservation of Mass. Unless the diagram is in error this is a questionable system, and closed loop versus open loop is besides the point. Are you sure the second pump isn't intended to boost pressure?

Click to expand...

Your post got me thinking...

I tested it multiple times in our test bay last week and I kept seeing the same phenomena. I'd have the system running at a constant speed and temperature and only increase pump 1's speed. When I did, flow meter 1 (directly after pump 1) would see an increase and flow meter 2 (directly after pump 2) would not. Momentarily, flow in was not equalling flow out and I didn't observe any leaks or malfunctions on the system.

But, I think I can explain why now.. As I stated earlier in the thread, pump 2 (the 2 piston PD pump) operates at ~3500 PSI and that pressure can be manually adjusted. When I only increased pump 1's speed, the pressure in pump 2 would bump up slightly. Once this pressure stabilized, flow in did indeed equal flow out.

So I thought of the PD pump like a tank. It should always have a constant value for maximum volume. If you fill a tank, flow in is always going to equal flow out only if the pressure (and volume) in the tank is constant. But, being a PD pump, water within the pump is trapped and not always subjected to the pressure of the water coming into the pump as water within a tank would be. More water comes into a tank, it pushes more water out. More water comes into a PD pump, it takes time for the pressure within the pump to stabilize since water in can't simply push water out. Only when this pressure has stabilized does flow in = flow out.

Does this make sense? Or am I totally off base... I'm just wondering before I go into a meeting Tuesday afternoon about this. When I wrote our functional design spec for this feature, I included that flow in = flow out was conditional (and sited the "snapshot" of the system after a set point change on pump 1 as proof) I instantly met some internal restriction. It now seems that it is conditional upon the "black box" between flow in and flow out having a constant pressure. I'd never really given it much thought since a couple of classes in school, but that thought process applied to PD pumps seems logical. If you increase the set point of pump 1, it takes time for the first "pocket" within the 2nd PD pump to reach the discharge tube. At some point in time you're going to have multiple pockets within the 2nd pump at different pressures causing flow in not to equal flow out.

If I'm totally wrong, then it seems the flaw is in our equipment.

jethridge said:

Does this make sense?

Click to expand...

I think it makes sense. Conservation of mass says flow in and flow out must be equal over the long term, but if there are short term disturbances causing the amount of water inside the pump to change, that would cause a delay in the change in flow.