Make:

Perhaps a simpler, less invasive means would be to slip a scale underneath the keg. If you could find a reasonably priced scale capable of supporting the weight of the keg with a serial interface, you'd be in business.

Mark

Hi Mark,

I thought about that idea as well, but placing a 160lbs keg on a scale for long periods of time seems very likely it would cause the scale to loose accuracy after a few months (if not shorter). And since it's so heavy, it's not realistic to move the keg onto the scale for short periods of time.

-Aaron

What about controlling the flow rather than metering it? Maybe the beer could be dispensed by a peristaltic pump, so you would only need to record the pump's rotation.

Beer doesn't need to be pumped since the keg is pressurized via a CO2 canister (or CO2 & Nitrogen) and it flows merely by opening the tap. The keg has to be properly pressurized for the beer to have bubbles as you'd expect- otherwise it's flat which sucks.

Unfortunately, just measuring the time the tap is open wouldn't be accurate enough since the flow can vary for a number of reasons.

I do not know of a decent flow meter, but I know of a couple other ways that you could measure the volume of beer exiting the Keg:
By weight (A strain gauge or a scale would work)
By pressure (You see this a lot on commercial Coffee storage. Its a clear pipe connected to the container that will display the pressure inside)
By Capacitance (Personal favorite): You attach a wire to the keg (can be attached magnetically). The kegs capacitance with respect to ground will vary according to the amount of liquid inside of it.
Inductance: Same as above
That's all I got right now. Good luck, and let us know if you find that illusive flow meter.

Capacitance will also be affected by the kind of beer in the keg... It might be consistent enough that you can have a set of numbers for each type, or you may need to calibrate it each time you fill it.

In the case of the peristaltic pump, I don't think it would matter that the keg is pressurized... rather than actually moving the liquid, it would just be controlling the flow. The pump would basically just be pinching off the flow at regular intervals. Just run it very slowly.

Odin84gk:

Unless you have a solution to the "heavy object on scale for long periods of time causing it to loose accuracy" I don't think it's a viable long term solution. Honestly, this was my first idea, but I can't come up with a good solution for this.

Measuring pressure won't work since the pressure should be constant due to the CO2 & regulator controlling it.

Capacitance/Inductance sounds really interesting... I'll have to dust off my physics and read up on that. I wonder how much difference there is between the different beers and if we can get good enough calibration with different settings for ales, stouts, etc. But since an empty keg should have a constant value regardless of what beer it once contained just tracking the starting value & delta would be useful for providing a % full.

Thanks everyone for coming up with all these ideas!

-Aaron

This datasheet for a Rosemount 8721 magnetic flow sensor provides a great introduction to the application of Faraday's law for sensing the velocity of a conductive liquid:

http://www.emersonprocess.com/Rosemount/document/pds/4901_00n.pdf

Magnetic flow sensing is a challenge to design, but provides a great non-instrusive way to do sanitary (NSF) monitoring of flow for processing food products (milk, juice, and possibly beer in your case).

For measuring capacitance, I suspect the beer type won't have significant impact on the measured capacitance. Depends on the variation in alcohol and solids content. Beer is predominantly water which may dominate the measurement with it's dielectric constant of 80 or more. Ethyl alcohol which accounts for 5-10% of beer has a dielectric constant of 24.3 at 77F.

Maybe you could do something with ultrasonics? Perhaps you can shoot through the top surface of the keg, eliminate the primary response from the metal, and see if you can get a secondary reflection from the fluid surface.

Mark

Mark, thanks for the link. I've emailed Emerson Process to get a quote. It's prolly $500, but might as well ask right? :)

-Aaron

I would imagine it is a pricey unit.

Building a magnetic flow meter for myself is on my todo list, but unfortunately it's under a giant stack of other work.

Mark

Just got the quote... I won't say the exact price, but it's quite a bit more expensive then what I guessed. I've asked if they've got anything cheaper for a DIY project, so if I find anything reasonable, I'll let people know.

-Aaron

What is your budget and line size?

Rosemount has many competitors that offer comparable products at a lower price.

Consider also a strap on doppler meter.

http://www.dynasonics.com/products/dfx.php

Capacitance measurements use a teflon insulated semsor on conductive media. As long as minimum conductivity exceeds 200 micro siemans variations in compostition will not affect the measurement. Capacitance has been used by both Coors and Bud for beer levels.

Here is something relatively cheap

http://www.omega.com/ppt/pptsc.asp?ref=FPR301_302_303_304&Nav=gref05

So we started thinking about measuring capacitance of the keg, but we can't figure out how to make it work since the entire outside of the keg is basically one piece of metal. To measure the capacitance of the beer, we'd need two parts of the keg to be isolated from each other.

Since the kegs aren't owned by us (they're borrowed from the distributor when you buy the beer), modifying them to add an internal sensor isn't an option.

Anyways, we were thinking we should be able to build our own magnetic flow meter by wrapping part of the tube connecting the tap to the spigot in copper wire and applying a DC current. It would obviously be more work then purchasing an assembled/calibrated unit, but that's half the fun.

What about running a pair of probes in the top and modifying the tap?

Stokes: Not really feasible. Basically the top of a keg is a spring loaded large ball bearing with a collar around it. The tap locks around a collar and you depress a handle to drive a plunger to push the bearing down- thus tapping the keg. The contents of the keg are kept under pressure via a CO2 canister attached to the tap which displaces the beer as it is consumed.

Long story short, even if you were able to insert the probes down into the keg without compromising the air tight seal (which has to support up to about 65psi for beers requiring a CO2/Nitrogen mix), you'd very likely get them stuck in there.

There's a picture of a beer keg here which might help:
http://en.wikipedia.org/wiki/Image:Keg_geschnitten.jpg

How about a strip of piezo transducers mounted down the side of the keg? Using a microcontroller to ping them, you could probably do the equivalent of knocking on the side of the keg with your knuckle and hearing the level of the liquid. I'm guessing the ring-out of the impulse will vary depending on the amount of liquid on the other side of the metal wall from the sensor.


Edit: Apparently similar things already exist...
news.thomasnet.com/fullstory/466277

mechanical.pnl.gov/docs/ultrasonic_lld_cost.pdf

Hmmm ... how to describe this without a diagram ....

Picture a pneumatic (or hydraulic) cylinder with the rod (ram) pointing up.
A platform is attached to the the ram, and the keg is sitting on the platform.
The port of the cylinder is connected directly to a pressure transducer.
As the beer is consumed the pressure appearing at the transducer will drop in a predictable manner.

Obviously the mechanics are a little bit more complicated than I've given above.
Also, depending on the medium that's in the cylinder (air/oil/water/?), there may have to be an ambient temperature offset to the pressure reading.

By the way, if you're not looking for super-accuracy, pressure transducers can be pretty darned cheap ....

Ask the auto-zone guy for the oil pressure sensor for a 1972 Dodge pick-up, 225 engine, pressure gauge in the dash (as opposed to an idiot light). <grin>

John

bluenoise:

That's an interesting idea... I'll have to call around and see what the cost would be. FYI, I think the second link should be:

mechanical.pnl.gov/docs/ultrasonic_lld_cost.pdf

jschuch:
I think I understand what you're saying. Seems possible. Would be bit of a challenge fitting everything in the refrigerator, but worth investigating.

You're correct about that second link, but it won't change when I edit it. The edit and the preview show the correct URL, but it sticks my username into it after I'm done editing. Weird.

Edit: I think I figured it out...The URL starts with '(slash)me,' which must prompt the forum software to substitute my profile name.

bluenoise

I still think the scale is the cheapest and most feasible solution.

Build a small frame and platform to support the keg with a way to raise the scale up underneath the keg to take a measurement and then drop back down to relieve the load. It would only need to raise up a few fractions of an inch, and the whole platform could be made barely larger than the thickness of the scale you are using.

Alternatively make the frame the part that drops down to load the scale. You could use casters that could cam up to position to become a mobile base and then cam out of the way to drop the keg down onto the scale.

Just trying to inject a little practicality.

For the magnetic sensor, what you're really doing is making a variable core transformer. As liquid flows through at different speeds, you get different inductances. What you need to do, then, is wind two coaxial (intertwined) coils about a tube that has the fluid flowing through it. The next thing you need is a signal source and a signal sensor. One of the coils is your primary, the other is your secondary. Put an AC signal through your primary (just take a look around for any audio circuits to generate a nice sine wave), and, with nothing flowing through, look at the voltage (Vrms) across your secondary. Now use some known flow (like timing the emptying of a gallon of water/alcohol 5-15% blend through the hose) to calibrate different flow rates and the voltages that they create. If you want, you can calculate a flow constant in V/(L/min), and then feed the secondary into an analog input on an Arduino or somesuch to have it automatically calculate the flow rate from your flow constant.

Now for the practical engineering side: you need...
1 roll magnet wire
1 piece of low-µ tubing (thin walled glass or plastic would work best, maybe a drinking straw if it can handle the pressure)
1 AF sine wave generator, probably in the 100Hz to 1kHz range for best Vrms sampling (Vrms should be between 1 and 5 VAC)
1 AC voltmeter (True-RMS like those from Fluke are better for this application, but any AC voltmeter should work)
1 piece of sandpaper (for the magnet wire)
1 liter of beer (or two pints... or three pints? At lunchtime? Arthur, it's muscle relaxant!)
1 funnel
1 stopwatch
1 length food grade tubing, cut into halves
*1 Arduino and an LCD display (optional)

1. Cut two 1-meter lengths of magnet wire and label the ends of each with tape saying "Primary" and "Secondary" (so as to not confuse them later.
2. Tape the wires together about 20 cm from each end, and once more in the middle.
3. Consult this video (http://blog.makezine.com/archive/2007/09/make_a_foxhole_radio_week.html) to hear about the frustration of winding a 120 turn coil. We're going for 500 turns, so start by taping the middle of the wires to the middle of the straw, pick a direction, and start making your coils. Make certain that the entire winding of both wires is either right-handed (counterclockwise) or left-handed (clockwise), and not both.
4. Use the sandpaper (although a soldering iron would work, too) to bare the ends of the magnet wire.
5. Hook up the primary to the AF oscillator and the secondary to the voltmeter.
6. Hook up the plumbing, i.e. funnel to tubing to sensor tube to rest of the tubing, and put something under the end of the tube to collect the beer.
7. Time how long it takes for a liter of beer to traverse the tube, and record your voltage reading every 5 seconds.
8. Average your voltage readings, and now divide this average by the amount of time (in seconds) that it took your liter of beer to traverse the system. You now have your flow constant in V/(L/sec).
*9. Hook up the secondary to the ground and analog inputs on an Arduino, and write a program to display the flow rate on the LCD (a whole 'nother project in and of itself).

For those of you wondering what I mean by:
Vrms:
http://en.wikipedia.org/wiki/Root_mean_square
low-µ:
http://en.wikipedia.org/wiki/Permeability_(electromagnetism)

Hope this helps!
-Pete

Have a look at our flow meters drinkbrook.co.uk
low cost and easy to use