Air flow measurement in a pipe

[mcrr]

TL;DR Summary

Air flow using cta sensor in pipe

Dear friends.

It's my first post here and sorry if it's in wrong place.
I have a system with a diaphragm pump and a filter to collect dust. I need to set the flow (like 1 liter per minute) and I need to keep this flow constant during long time. The dust obstruct the filter and I need to make the compensation. I am trying to use the FS7 CTA sensor in a suction side of the pump to track the flow.
The idea is, set the desired flow, get the voltage of the CTA sensor as set point and use a PID to keep the set point. The problem is when I increase the air resistance and the flow decrease I speed up the motor until the set point CTA voltage, but when I measure the flow, it is higher than a have set. My doubt is if I should make a line pressure compensation when the air resistance increase to get the correct flow.
Thanks in advanced.

 

[erobz]

TL;DR Summary: Air flow using cta sensor in pipe

Dear friends.

It's my first post here and sorry if it's in wrong place.
I have a system with a diaphragm pump and a filter to collect dust. I need to set the flow (like 1 liter per minute) and I need to keep this flow constant during long time. The dust obstruct the filter and I need to make the compensation. I am trying to use the FS7 CTA sensor in a suction side of the pump to track the flow.
The idea is, set the desired flow, get the voltage of the CTA sensor as set point and use a PID to keep the set point. The problem is when I increase the air resistance and the flow decrease I speed up the motor until the set point CTA voltage, but when I measure the flow, it is higher than a have set. My doubt is if I should make a line pressure compensation when the air resistance increase to get the correct flow.
Thanks in advanced.

Are you using the fan curve, with fan laws to make your adjustments?

If your objective is to maintain $Q_{op}$ as the system curve transitions from $S_1 \to S_2$ via dust accumulation on the filter, you should use the manufacturers complete set of fan curves, or (less accurately) the fan laws to make your adjustment to rpm ($\omega_1 \to \omega_2$).

 

[erobz]

Is the sensor a pressure transducer, or is it actually measuring the flowrate, somehow?

 

[mcrr]

I attached my setup system. I am using flow meter hot wire sensor to control the pump.

 

[Baluncore]

CTA = Constant Temperature Anemometry ?

Maybe the pressure is falling when the diaphragm pump is turned on, because the filter restricts flow, and that cools the air at the CTA sensor, which upsets the constant temperature of the sensor.

Move the sensor from between the pump and filter, to atmospheric pressure, at the external air outlet, or to the air inlet.

A positive displacement pump before the filter should not need a mass-airflow sensor.
Diaphragm pumps can usually handle dusty air or dirty water.

 

[jrmichler]

Some questions to help us help you: The sensor should be before the air filter. Is the PID loop properly tuned? Did you try a PI controller? What is the diameter of the pipe at the point where the sensor is installed, and the air velocity past the sensor? What are you doing for flow straightening before the sensor?

 

[Baluncore]

I attached my setup system.

Where is the filter ?

 

[erobz]

I didn't think about the "positive displacement curve" characteristics...what I drew was typical for a centrifugal pump. A PD pump, should output fairly constant flow at a given RPM regardless of system curve changes. If the filter clogs either the electric motor will slip, or the pump casing will rupture.

I've never seen a PD used in a dust collector system, but that doesn't mean a whole lot.

 

[mcrr]

Some questions to help us help you: The sensor should be before the air filter. Is the PID loop properly tuned? Did you try a PI controller? What is the diameter of the pipe at the point where the sensor is installed, and the air velocity past the sensor? What are you doing for flow straightening before the

Thanks for you support jrmichier,
First of all the attached setup was create to validade my system, then the filter was changed by an air resitor, it is a valve then i can close and open to simulate filter obstruction. I can check the obstruction with the pressure gauge. I don' t have a pid, forwhile I am controlling the motor by dc power supply. The diameter of the tube is 6 mm and doesn't have any steaightening before the sensor, all the line is 6mm. I need to control the flow between 0.5 liter per minute to 5 liters to minute.

 

[mcrr]

Where is the filter ?

The filter was changed by an air resistor, then I can simulate obstruction by opening and closing the valve.
Regards

 

[jrmichler]

The problem is when I increase the air resistance and the flow decrease I speed up the motor until the set point CTA voltage, but when I measure the flow, it is higher than a have set.

I can see two problems here:

1) Your "air resistor" is apparently some sort of valve. A valve, any valve, disturbs the air flow. Hot wire air flow sensors do not work well with disturbed air flow. The disturbance will level out if there is a long enough length of pipe downstream from the valve. MUCH better is to have the sensor upstream from the valve.

2) How do you fit a 6.9 mm long sensor into a 6 mm tube? These sensors measure air velocity, and are designed to work in a smoothly flowing stream of air. If it is installed in a Tee fitting or an elbow, expect problems. Again, MUCH better is to have the sensor upstream from the valve. And in a straight section of pipe.

Please show us a detailed drawing or photo of the sensor installation.

 

[mcrr]

I can see two problems here:

1) Your "air resistor" is apparently some sort of valve. A valve, any valve, disturbs the air flow. Hot wire air flow sensors do not work well with disturbed air flow. The disturbance will level out if there is a long enough length of pipe downstream from the valve. MUCH better is to have the sensor upstream from the valve.

2) How do you fit a 6.9 mm long sensor into a 6 mm tube? These sensors measure air velocity, and are designed to work in a smoothly flowing stream of air. If it is installed in a Tee fitting or an elbow, expect problems. Again, MUCH better is to have the sensor upstream from the valve. And in a straight section of pipe.

Please show us a detailed drawing or photo of the sensor installation.

You can see my setup and sensor setup in attached files.

 

[erobz]

If this sensor measures gas flow temperature to determine flowrate, we expect the temperature to periodically rise and drop in a PD pump because we expect the pressure to rise and drop over the displacement. You have a pressure gauge there, have you plotted pressure vs time to see the variation across the stroke?

 

[mcrr]

If this sensor measures gas flow temperature to determine flowrate, we expect the temperature to periodically rise and drop in a PD pump because we expect the pressure to rise and drop over the displacement. You have a pressure gauge there, have you plotted pressure vs time to see the variation across the stroke?

Hi erobz,
No, I haven't. The cta readìng is stable, the pressure have a little oscillation. I will try to get the max and min measurement of the pressure
.

 

[mcrr]

Hi friends,

The problem here is the turbulence as jrmichler said. Now I will study about it to get the laminar flow inside the tube where sensor is. The device is small, then I don't have enough space to put the sensor far away from pump and inlet. I think a pulsation dumpner can improve the system.

Thank very much for all who have helped me.