Flow meter 3

[luciansm]

Hi evryone.
My name is Lucian, I'm a Computer Engineer.
I'm working in a project of a company to develop a flow meter using a microcontroller and Infrared sensors.
I'm stuck in the mechanic fluid part.
We're trying to develop this product basing in the mechanic model of the attached picture.
The liquid will enter in the top and pairs of IR leds and Phototransistors and the A/D converter of the microcontroller will read the voltage oscilation caused by the difference of the density of the liquid.
The liquid will pass with pulsating vacuum.
The goal of the 3 channels is to segment the liquid and increase the precision of the read.
I'm having sobre problems with the voltage read and I'm starting to think that this problems are related with fluid mechanics like turbulent flow.
We're developing this mechanics prototypes using a 3-D Printer.
My goal is to develop a mechanic prototype that will segment the liquid with no turbulent flow (or less tuebulent possible).

I don't know much about fluid mechanic, but I studied a little bit about Reynold Number and turbulent flow.

If someone could help I would be grateful.

Thanks and sorry for the bad english.

 

[BigInch]

A lot of wall surface very near to the flow will tend to create many near-wall laminar and turbulent boundaries. There looks to be a significantly reduced flow area as the fluid enters the device. That may tend to cause increased velocities with reduced pressures similar to a "venturi". A lot of potential acceleration vectors are possible entering the flow channels. Square and bluntly pointed corners don't help.

 

[luciansm]

Hi BigInch, thanks for the answer.
I'll try to increase the distance between the walls of each channel, with the goal to decrease the speed of the liquid. Do you think that if I increase the length of the channel, the liquid would be less turbulent?
I had read a little bit about development flow that is related with the RE number and the length of the pipe.

 

[Latexman]

What fluid(s) or fluid type(s) will be the target market? What are the typical physical properties and operating conditions? Density? Viscosity? Normal boiling point? Temperature? Pressure?

Good luck,
Latexman

Technically, the glass is always full - 1/2 air and 1/2 water.

 

[luciansm]

Hi Latexman.
Ok here We go.
The fluid will be milk from a milking machine. The density of the milk that I founded was 1.034 g/cm^3. The dynamic viscosity is 0,00149 Pa.s . The temperature of the cow milk is about 35ºC. The milking machine uses a pump to create a negative pressure to suck the milk through the pipes. I have not found information about the pump yet, but I'll find.

 

[Latexman]

It's been 40 years since I've been to my cousin's milk farm. Is the flow pulsating or steady?

Good luck,
Latexman

Technically, the glass is always full - 1/2 air and 1/2 water.

 

[Latexman]

1.5 cP. It may not be easy to minimize turbulence.

Good luck,
Latexman

Technically, the glass is always full - 1/2 air and 1/2 water.

 

[luciansm]

It's pulsating.

 

[luciansm]

I'm posting a picture with the plot of the values of the middle channel and the right channel.
I don't know if my assumptions are right, but I think that something like that.

 

[bimr]

The problems that you are encountering is why the output of milk is weighed, not measured with a flow meter.

http://www.delaval.com/en/-/Product-Information1/Milking/Products/Milking-point/Milk-recording/MM15/

 

[luciansm]

Hi bimr, thanksfor the answer.
I already knew about this product of delaval. I did a research about milk meters
There is a lot of milk meters that use a free flow technology.
I know that's ways more difficult to implement than weigth the milk, but I think that is possible.
I think that will envolve a lot of fluid mechanics theory.

 

[dbill74]

First observation I have is your device looks square. I have never seen a square pipe.

Second if you want more laminar flow, then your arrangement is, well problematic. The widening at the inlet, then the vanes dividing the flow are going to create turbulent flow in those areas. You would then need a much longer straight path for the flow to 'even' out. If you look at other manufacturer's meters installation requirements, they require 10x pipe diameter straight pipe upstream of the meter to get the best possible flow profile through their meters. They also require 5x dia of straight pipe downstream to minimize back pressure issues.
Also, flow velocities in pipe are generally faster than near the edges (due to friction), I foresee momentum causing faster/more flow through your center channel.

Have you considered using a tubular device with your IR sensors looking at the fluid from different angles? Or would such an arrangement cause the sensors to interfere with each other? This would also mean all your sensors are measuring the same flow stream.
In a typical water meter the flow is forced through an area smaller than the pipe it is connected to. This is done to increase the velocity through the meter and reduce the chance of getting air in the meter.

 

[BigInch]

Can we take one, or two, steps back. Why IR? Do we know IR works in this app? Why not multipath ultrasonic? That's the standard metering technology today.

 

[luciansm]

Hi dbill74, thanks for the answer.
About the square pipe, yes, I already thougth about that, but our 3-D printer is incapable to print a circular pipe. I'll try to find something to implement a circular pipe.
We were using a patent of a company called SCR as a reference.
I'm posting the image of the flow meter of SCR. This meter is really small, about 11x11cm of area.
The IR sensors are not interfering each other because the maximum angle of the beam is about 30º.

 

[luciansm]

Hi BigInch.
The goal of the IR sensors it's because it's really cheap and easy to implement.
And the size of the device would be really bigger if we use ultrasonic components.
With IR sensors we are capable to use a algorithm called Cross correlation to compare the 2 signals of the sensors (top and bottom).
With this algorithm we can find the speed of the liquid (in theory). The problem will be to find the height of the liquid in each path to use the equation:

Q = K.v.A
Where K is the calibration of the height (0 to 1).

I had research about ultrasonic flow meter using time transition, and it's really interesting, but we have to consider a size restriction of the device.

 

[btrueblood]

The patent you are referencing is describing a meter that measures the milk content in a two-phase vacuum conveyor system. The water (milk) droplets/slugs impair the IR beam transmission, so that the size of the drop passing by the sensor can be estimated, and the velocity must be inferred by some other measurement, or by correlation between multiple sensors in the same channel path. The flow from the (typically) multiple milking machines is again two-phased, I.e. water drops or slugs mixed with enough air to keep the milk moving along the pipe to a collection vessel. Because the milk will arrive at the meter in randomly sized packets, I think you are worrying about the turbulence un-necessarily. I.e. the flow will always have a highly random nature with some channels running full and others seeing a more "frothy" mix, and the only way this meter can work is to test it: the algorithm that integrates the signal vs. time must be correlated to actual fluid collected in the downstream "bucket" across a wide range of delivery conditions.

 

[luciansm]

Hi btrueblood, thanks for the answer.
You're trying to say that these patent that I showed is something like this (I posted a picture in paint brush D)?

Thanks.

 

[BigInch]

OK, I didn't know ir worked like that. Good luck. Thanks

 

[luciansm]

Thanks BigInch!
And thanks for the hints!

 

[btrueblood]

I'm familiar with a meter like the one you showed that is mounted inline on a single pipe, not on a branch like you showed. Yes, I have spent some time in dairies. Being able to track milk production from individual cows

At some point, you will need to think about how to make the meter clean-able, preferably it should be capable of being cleaned in place. The best solutions for in situ cleaning milk from a line is hot lye solution followed by hot acid, rinse and repeat several times. And no porous material allowed, no crevices for bugs to grow in. So, while a 3d-printed part sounds good from the crevice-free standpoint, it will suffer with the porosity requirement, and good luck finding materials that can withstand the cleaning chemicals.