Hiya! I didn't want to write back until I had something intelligent to say. Sorry for the delay, folks!
And may I also add, Dr AJ is a boon to this forum--everyone should read the sticky on water for brewing. Very beneficial, generous, and spot on.
So it is with much respect and humility that I dare suggest that in this one case--specifically, the risk analysis assessment of cylinder failure due to repeated dry-ice refills--that AJ may be somewhat inaccurate in suggesting acid erosion as being more important than temp-cycle embrittlement of metal. And for my part, I'll try and be civil & non-condescending (non-condensing?
) & stuff, as if we were at the table sharing a brew.
I do reserve the right to be silly, because, well, life is too short not to be silly. :fro:
More to the point: if you do a bit of a search on the subject you will find many references to the fact that carbonic acid solutions are corrosive.
Never contending that CO2 in water wasn't acidic, just wondering how much water could be in the cylinder with the liquid CO2.
The paper is an interesting read, though I often found it confusing keeping track of when a figure was "CO2 in water" vs "water in CO2", especially in Figure 9, perhaps* the most pertinent to this item, they call the former "Water-rich phase" and the later "CO2-rich". Holy rusted phraseology, Batman! But the discussion in Section 4 helps.
*Figure 9 describes the erosion rates of carbon steel under different conditions. The black line says that water saturated in liquid CO2 erodes at about 0.2 millimeters per year at almost any pressure. This is a projection after 24 hours of exposure at 50C. (
Maybe it's that hot in a shed in Australia right this moment. And although error bars in measuring erosion thickness tend to matter when you multiply by 365, I'm too lazy to dig in to estimate how big they might be in this case. ... ) So 0.2 mm/yr could be worrisome for a thin walled tank, or after 5-10 years (or less) of a normal tank, but that was never my point.
As AJ mentioned, water saturation isn't very high. Previously, I was suggesting--based on a guess--that any adhered water would be even less than saturation. Per Figure 2, saturation for our case is about mole_fraction=.003 at 950psi=65.5 bar. (Lower at room temp, but let's use the highest arguable value.) 5 lbs = 2.3 kgs liquid CO2, molecular weight 44.1 g/mol leads to 52 mols. .003 of that is .15 moles, 18 g/mol = 2.8g. I roughly estimated 0.5 mL = 0.5 g under Houston mouth breather conditions, let's say it's more like 1 mL. That's still a third of max solubility. At 25C it's lower. And I rounded up a lot here.
& Hey:
before our cheap, I mean thrifty, thread readers sigh in relief that they have 15-30 years before the corrosion explodes their beloved mammals, please note that a) I still think you have much less time than that due to the rapid cryo/room temp cycling and b) that's a 15 year error window. 15 years waiting for your bomb-in-waiting to explode is really long & frankly f-ed up. I mean, you might as well start smoking crack. And besides, you've missed point c), which has something to do with assuming that the published proposed self-limiting mechanism of passivating FeCO3 crystals still holds at lower concentrations. It might not, and AJ might be right.
So there's that.
Either way, don't do it.