Mixing by Weight?

[Akimbo]

Is anyone mixing gases by weight? This was the realm of commercial gas plants when I started in the 1970s because the scales were so expensive the technique was of little value offshore.

However, you can buy a 50 Kg/110 Lb scale that is +/- 20 grams for $50 now. I may have made a quick math error but with Oxygen at 40.465882512 Grams/Ft³ it seems that would be accurate enough for home-brew mixing — onshore anyway. This strikes me as less expensive and more accurate than PP mixing with a precision pressure gauge — and is less temperature sensitive.

Sorry if the subject was covered already, but couldn't find it on a search.

 

[rjack321]

How accurate do you need to be? I can get plus-minus 1% of target with a non-precision analog gauge.

 

[dannobee]

How accurate do you need to be? I can get plus-minus 1% of target with a non-precision analog gauge.

It has to be pretty frickin' accurate if the bottles are small, but it is doable. 1% doesn't cut it.

I did some rough calculations before when I was going to do it (using refrigerant scales-accurate to within an ounce) and it wasn't worth the hassle with small cylinders. Big cylinders (T-bottles and the like) would obviously be easier.

One of the problems that I had was that the scale (digital) wouldn't repeat. There was some hysteresis.

 

[cool_hardware52]

Part of the problem getting the scales to repeat is the forces the fill whip can exert on the scale is variable.

I built some fuel injection systems ~30 years ago that used propane as the fuel. I wanted to get accurate mass flow numbers for the injectors at various pressures, orifice size, injection duration and frequency.

That involved a lot of timed test run runs and weighing the propane cylinder.

To get accurate repeatable readings it was always necessary to disconnect the tank from the associated hoses.

If you have just the tank on scale and it's connected via some sort of flexible hose to any apparatus that's not on the scale it's very hard to have isolate the forces the hose can impose on the scale.

Unless the cylinders being filled are huge and the gas weights are very large compared to the "hose force error" blending by weight seems like it would still be a struggle.

It more a matter of learning the idiosyncrasies of a given fill station / gauges and adjust from there.

Tobin

 

[Thalassamania]

Some of the new hose materials, Kevlar (I've had Kevlar filling hoses made up for shipboard use at a local hydraulic supply hose) or Miflex might help a lot.

 

[cool_hardware52]

Some of the new hose materials, Kevlar (I've had Kevlar filling hoses made up for shipboard use at a local hydraulic supply hose) or Miflex might help a lot.

Miflex at 4000psi + seems like a bad idea....

But I would agree that super flexy hose would help.

Tobin

 

[Thalassamania]

I thought there was a new Miflex that was > 10,000 PSI (which is what the Kevar is, and it's pretty damn light too).

 

[Akimbo]

Part of the problem getting the scales to repeat is the forces the fill whip can exert on the scale is variable.

I built some fuel injection systems ~30 years ago that used propane as the fuel. I wanted to get accurate mass flow numbers for the injectors at various pressures, orifice size, injection duration and frequency.

That involved a lot of timed test run runs and weighing the propane cylinder.

To get accurate repeatable readings it was always necessary to disconnect the tank from the associated hoses.

If you have just the tank on scale and it's connected via some sort of flexible hose to any apparatus that's not on the scale it's very hard to have isolate the forces the hose can impose on the scale.

Unless the cylinders being filled are huge and the gas weights are very large compared to the "hose force error" blending by weight seems like it would still be a struggle.

It more a matter of learning the idiosyncrasies of a given fill station / gauges and adjust from there.

Tobin

I don’t think hose whip variables matter in this application. We are only measuring the added weight. Total weight repeatability isn't a factor either. Just for round numbers, let's say you calculate that your cylinders at 500 PSI need another 500 Grams of O2. The tank and whip weight would be stable before you add the 500 Grams. Total weight does not matter unless it exceeds the scale capacity.

The next step would be to hang the cylinder on the scale with the compressor/bank whip. Again the weight would be stable and you would add X grams of air. I have not bothered with a tolerance analysis, but I suspect the scale's precision would have to be much higher than +/- 20 grams for Helium mixes. The good part is temperature is virtually taken out of the equation when the top 60% of the fill is rapid/hot.

I have only seen high precision cal gases mixed three times from commercial suppliers, but they were done on precision balance scales — 25 years ago anyway. They delivered certified 1% HeO2 +/- .02%. They didn’t care about how much weight was used to balance the cylinder(s) and whip. The precision counter-weights came out when the pointer was on zero and the calculated gas weight was ready to add. I have no idea how precise the scales were. The cylinders could weigh several tons though because they used the same rig for six packs delivered to non-diving customers.

 

[cool_hardware52]

I don’t think hose whip variables matter in this application.

Trust me, I've tried and it matters, up and until the variables introduced by the hose become a small part of your error budget.

Go find any good high resolution digital scale and invert a bottle of beer (or your favorite beverage) cap side down on the scale, try to simply balance the bottle with your hand, no load, just keep the bottle from falling over.

If the scale is worth a damn it won't register a weight because it's constantly changing just slightly due to your hand touching the bottle.

Tobin

 

[Akimbo]

Trust me, I've tried and it matters, up and until the variables introduced by the hose become a small part of your error budget.

Go find any good high resolution digital scale and invert a bottle of beer (or your favorite beverage) cap side down on the scale, try to simply balance the bottle with your hand, no load, just keep the bottle from falling over.

If the scale is worth a damn it won't register a weight because it's constantly changing just slightly due to your hand touching the bottle.

Tobin

Apparently it does not matter to people who certify precision cal gases for a living, sometimes with six-packs and sometimes with 1 Liter sample bombs. Perhaps the key is to make sure you do not transfer forces through the whip as you manipulate bottles.