PDA

View Full Version : Exploded CO2 bottle



Peter.
04-02-2009, 03:27 PM
Here's a story I remembered reading from last year. It's a drag-racing guy who had a bottle of CO2 for his air shifter stored in his bike trailer. Quite sobering seeing how close to the friends eyes the damage was done.



Hi All,
We have a little accident here last week and I thought I post a few pics. Everyone is OK. It seems the blow off valve did not do its job on my spare co2 shifter bottle. Bottle was on the oil shelf in the race trailer and blew up just as we were hooking the trailer to the truck.

Here is the hole it made
http://www.wildbros.com/images/P50400502.JPG

Here is my friend Wayne that was down from Mn for his birthday weekend. Happy B-Day Wayne :lol:
This pic is after he pulled the 5" spike out of this neck.
http://www.wildbros.com/images/P50400442.JPG

Close up of his shoulder. These are deep in there :grn:
http://www.wildbros.com/images/P50400452.JPG

Here is one of me and the inside of the trailer. There is 2 1/2 qts of oil not in bottles. The CO2 bottle was on the oil shelf right above the 30gal drum of C16.

Peter.
04-02-2009, 03:28 PM
More pics:


http://www.wildbros.com/images/P50400492.JPG

Some pics of what is left of the bottle :bmb:
http://www.wildbros.com/images/P50400462.JPG

http://www.wildbros.com/images/P50400472.JPG

http://www.wildbros.com/images/P50400512.JPG

Liger Zero
04-02-2009, 03:40 PM
Yikes. Any idea what set it off?

Fellow I worked with at Harbec got blasted with hot plastic in the face and upper body, THANK GOD he was wearing goggles :cool:

Peter.
04-02-2009, 03:44 PM
They never said why - I guess no-one knows. One important thing is that like so many bottles mounted to bikes that one appears to have been polished. They are generally quite rough on the casting so perhaps they thinned the walls when cleaning it up? Does cast ally work-harden? Perhaps that is why?

Liger Zero
04-02-2009, 03:47 PM
Maybe... Bottle could have been damaged too from use.

Or maybe sunlight heated the trailer too much and it caused the bottle to pop.

TGTool
04-02-2009, 03:53 PM
A place I once worked sold Cushman industrial vehicles among other things. Sometimes we had requests to supply them converted to propane which we did with a conversion kit specifically for that vehicle and engine. One day we had a visit from an investigator verifying what we had supplied and done, etc. We were in the clear, but the story turned out to be that the customer had replaced the bottle connector we supplied with a quick disconnect. Unfortunately, the new one did not have the relief valve of the original. By itself, that was bad, but compounding it was the fact that they refilled their own containers and bottles are never supposed to be filled more than 90 percent. In this case, it had been filled 100 percent, had sat in the sun, and we were told that it had expanded the bottle to almost double. :eek: Catastrophe waiting to happen.

Evan
04-02-2009, 04:04 PM
It's an aluminum bottle. Aluminum shouldn't be used for pressure vessels. The critical pressure for CO2 is 800 psi. It is unlikely that pressure change with temperature was the direct cause of the failure. The root cause is certainly the fact that aluminum undergoes metal fatigue any time that stress is applied. It doesn't matter how little, it still results in a proportionate amount of fatigue. That especially applies to the number of pressurization/ depressurization cycles it went through as those are full range stress events.

Eventually an aluminum part will fail. The number of cycles of stress it takes depends on the alloy, the amount of stress, the repetition rate and the alloy's starting condition. Pure aluminum shows very little fatigue even after many cycles. High strength alloys such as 7075 can fatigue so quickly if stressed repeatedly near the plastic limit that they will fail below that limit after maybe only a few hundred cycles.

There was a serious problem some years ago with aluminum SCUBA tanks exploding for the same reasons.

Quetico Bob
04-02-2009, 04:15 PM
Yikes, makes me wonder. Have 4, 80 cu/ft scuba tanks sitting at 3,000psi.
But they incorporate a burst disk thats supposed to let go for that very reason (sitting in a hot trunk) or what ever.. Maybe that failed or the wrong one was inserted on annual inspection.
Cheers, Bob

Doc Nickel
04-02-2009, 04:17 PM
Yikes. Any idea what set it off?

-Probably overfilling. It's very easy, probably too easy, to refill small CO2 tanks from larger ones. The paintball marketplace started supplying inexpensive "fill stations", and that's carried over to racers (who use the tanks to power air shifters and the like) and combat robots (who use onboard tanks to power pneumatic spikes, hammers and flippers) and even fratboys (who tinker together their own "kegerator" CO2 sources.)

The problem is, tanks need to be filled by weight, not like any other container where you fill it 'til it can't take any more. There has to be a "vapor bubble" in the tank, to allow for the expansion and contraction of the CO2- which is dramatic when it's in it's liquid, pressurized form.

That looks like it was about a 6 to 9-ounce tank. But it could theoretically hold about two ounces more if it were overfilled, in which case, a mere 10-degree rise in temperature could better than double the pressure in the tank.

There's also lots of additional factors; the tank could easily have had some physical damage. In that photo of just the base disc, a big notch is seen right by the tip of his ring finger. It may be merely shadowed, but it appears to be 'dirty'- IE, an old ding, not freshly-caused by the blast. A much less significant ding only a little further up the body could have been the "weak point" where the wall failed.

The tank could have been repeatedly overfilled, which over time stressed and weakened it, to the point where even a normal fill could have caused it to fail.

That, after all, is exactly the reason pressure tanks are hydrotested on a regular basis. And yes, even small ones like that have to be periodically tested, if they're larger than 2" in diameter.

And that, the weaken-over-time part, could explain why the burst disc failed to act. On that bottle, the burst disc should be rated at 2200 psi, as the tank is rated for 1800, and CO2 has a "room temperature" working pressure around 800 psi.

A severely weakened tank- either from a ding/notch/scar or from regular overstress- could have easily blown at a pressure well below the burst disc rating.

Or, as has also happened, the burst disc could have popped previously, and the owner could have easily replaced it with a higher rated one, either accidentally, which is possible, or intentionally, which I've seen done.

Paintball tanks are also available in HPA as well as CO2. The compressed air tanks are rated at either 3,000 psi or 4,500, with appropriately-scaled burst discs. It's entirely possible, especially if that's an older valve, that someone put in a burst disc rated for a 3K psi air tank, rather than a 1.8K psi rated CO2 tank.

I've seen that done intentionally, because the owner kept overfilling, because he didn't know the proper procedure, and therefore kept popping burst discs.

A little forensic work could narrow it down. Who last filled the tank? How old is the tank? What's the disc rated at?

But assuming no maliciousness or stupidity on the part of the owner/user, if I had to guess, I'd say probably mechanical damage. It's a race bike, chances are the tank to a nick or ding sometime in it's working life, and that caused a weak spot where the failure began.

Doc.

camdigger
04-02-2009, 04:35 PM
Owwww

Looking up CO2 liquid/gas ratios in an industrial handbook, 1 unit of liquid CO2 yields 543.7 units of gaseous CO2 or 1 gal CO2 liquid yields 72.9 cu ft gaseous CO2:eek:

That ratio is why we pump hundreds of barrels of liquid CO2 in well stimulations.....

Doc Nickel
04-02-2009, 04:36 PM
IThe critical pressure for CO2 is 800 psi. It is unlikely that pressure change with temperature was the direct cause of the failure.

-Actually, as I said above, that easily could be.

A 7-ounce tank, for example, actually has a 'liquid capacity' closer to 10.3 ounces. The extra room is for the "vapor bubble" which gives the liquid CO2 room to expand and contract with temperature changes.

Unfortunately, as I said, it's very easy for an inexperienced or undereducated user to overfill a small tank.

Even still, the burst disc should have prevented a blowout- which means there was either an issue with the disc itself, or with the tank.

The fatigue you and I both mentioned is entirely all too possible- the more it's overfilled, the more the tank is stressed, and the weaker it gets. Eventually the tank gets weaker than the burst disc, and it becomes a grenade waiting to go off.

Unfortunately, the burst disc issue is also possible- I work with these tanks virtually daily, and the number of them I've seen where the disc has essentially been disabled "because it kept blowing out on me!" is sometimes frightening. In the days before the "unified" burst discs, people would literally stack two discs in instead of one. I've seen unified discs replaced with higher rated ones, and once or twice, the entire bolt replaced with a setscrew and lots of Loctite.

So without further data, there could be at least three possible causes: Overfilling and a malfunctioning burst disc, age or repeated-overfill fatigue, or mechanical damage to the wall of the tank. Or, it could have been a combination of any of the three.

Doc.

steverice
04-02-2009, 06:02 PM
bet if you look you will probably find a plug where the blow off goes.

wment
04-02-2009, 06:13 PM
I remember a automotive a/c course I took years ago. The instructor said the relation between temp and psi was pretty linear. He then asked what the pressure was in the 1lb can of R12 at 70 degs F. One guy said 70psi, the instructor said Correct. What about at 80 degs F, another guy said 80 psi, Correct again. What about 110 degs, someone said 110psi, Wrong.... It would be 0 psi.... The can explodes at about 100degs/psi so dont leave a can of R12 in the trunk of your car...

Evan
04-02-2009, 06:53 PM
-Actually, as I said above, that easily could be.

A 7-ounce tank, for example, actually has a 'liquid capacity' closer to 10.3 ounces. The extra room is for the "vapor bubble" which gives the liquid CO2 room to expand and contract with temperature changes.

Unfortunately, as I said, it's very easy for an inexperienced or undereducated user to overfill a small tank.


CO2 is a special case. You can't really overfill a CO2 container. Once it is above it's critical temperature (31C) and it's critical pressure (73 bar [I misstated the critical pressure]) it becomes a supercritical fluid. It has the density of a liquid but the compressibility of a gas. That means it won't expand like liquid water does to burst the pressure vessel. In fact, the pressure does not increase linearly with temperature at that point. When CO2 reaches the critical point and becomes a supercritical fluid it takes on entirely different properties. It has infinite heat capacity which means an increase in temperature increases density instead of pressure.

kvom
04-02-2009, 07:51 PM
I was wondering about the overfilling explanation as well.

You do need the "vapor bubble" in order to get a sufficient gas flow for your intended use. If it's too small, the gas releases quickly and it takes time for additional gas to sublimate from the liquid to refill it.

Evan
04-02-2009, 08:23 PM
The vapor bubble only exists if the CO2 is below the critical point. Any tank made to hold CO2 must be able to withstand the critical pressure of 73 bar. Including a reasonable safety factor it would normally be able to take at least twice that. As soon as the pressure drops below 73 bar the supercritical phase disappears and the CO2 separates into gas and liquid phases. In that case the pressure is always less than 73 bar.

GKman
04-02-2009, 08:27 PM
The root cause is certainly the fact that aluminum undergoes metal fatigue any time that stress is applied. It doesn't matter how little, it still results in a proportionate amount of fatigue. That especially applies to the number of pressurization/ depressurization cycles it went through as those are full range stress events.

Eventually an aluminum part will fail. The number of cycles of stress it takes depends on the alloy, the amount of stress, the repetition rate and the alloy's starting condition. Pure aluminum shows very little fatigue even after many cycles. High strength alloys such as 7075 can fatigue so quickly if stressed repeatedly near the plastic limit that they will fail below that limit after maybe only a few hundred cycles. .

Seems like I recall that fatigue only occurs when material is cycled through tension - compression cycles and tension is sometimes intensionally increased in a design so the material never undergoes compression. A bandsaw blade is an example.

Evan
04-02-2009, 08:37 PM
Aluminum and steel don't share the same reaction to repeated stress. With steel if the stress is below a certain value then the part does not fatigue at all. With aluminum there is no lower limit and all stress results in some fatigue. With aluminum any type of stress in any direction causes fatigue.

boslab
04-02-2009, 09:44 PM
Aluminum and steel don't share the same reaction to repeated stress. With steel if the stress is below a certain value then the part does not fatigue at all. With aluminum there is no lower limit and all stress results in some fatigue. With aluminum any type of stress in any direction causes fatigue.
Evans right folks, thats why ali uses 'proof stress' not UTS, or rather 0.2 % proof as a strength indicator, trouble with fatigue is its invisible, you cant tell a parts going to fail by looking at it, you can however come up with some elegant mathematical models to predict when its going to happen if you do a lot of destructive testing, after manufacture of some alloys you have to refrigerate the stuff before fabrication to stop the metal hardening itself, age hardening, temp plays a big part in fatigue, the copper bearing alloys were notorius for age hardening [ they were called 'Duralamin' but that was a long time ago, the result of work done by Rolls-Royce and the High Duty Aluminium company, brand named Hiderminium or RRxx from memory, Reynolds had thier own version], i dont get the polishing may weaken bit yet as it would seem that the removal of stress raisers should strengthen it provided you dont remove a pile of metal while buffing?
is there a requirement to heat treat these cylynders after manufacture, eg percipitation treat or whatever?
regards
mark

Doc Nickel
04-02-2009, 10:24 PM
CO2 is a special case. You can't really overfill a CO2 container.

-Um, yes, you can. Easily.

The "supercritical" point comes at over 1,000 PSI and about 84 degrees. But when filling the tanks, there is invaribly a cryogenic effect- I wear gloves when filling, because the tank and fill apparatus virtually always form frost.

This is, actually, something of a requirement in order to fill- if both tanks are the same pressure, no gas gets transferred. But blowing some of the remnant contents out of the tank to be filled cools it, and thus greatly reduces the pressure. At that point, the standard pressure in the warm bulk tank can then feed the smaller tank easily.

It is not uncommon for an inexperienced user to "freeze" the small tank, leaving a chunk of dry ice inside. It's then possible to fill liquid on top of that.

But even without any dry ice, it's very possible- all too easy, in fact- to fill a small tank completely full- or very close to it- of very cold, very dense liquid CO2 at around minus-100F.

In this case, the liquid cannot enter the supercritical phase- at least, not before the hydraulic pressure of the expanding liquid exceeds the burst pressure of the tank. The liquid could still be below zero F and still overpressure the tank. It's happened, and it's entirely to easy to do. That's why the filling regulations expressly state the tanks must be filled by weight. Pressure and volume are mallable with temperature, but weight is a constant.

Ten ounces of supercritical fluid weighs the same as ten ounces of dry ice- but the pressure is considerably different.


It has infinite heat capacity which means an increase in temperature increases density instead of pressure.

-Yeah, and theoretically, under the right conditions, it can form a glasslike solid. But under anything close to normal conditions (http://www.warpig.com/paintball/technical/gasses/co2pv.gif), the pressure in the tank rises along with the temperature.

As noted, a 7-ounce tank (once an "industry standard" for early paintball guns) can technically hold 10.3 ounces of liquid at 60 degrees. That's right at a 150% overfill, and you can see on the chart, that would reach the burst-disc relief pressure at barely over 30 degrees F.


Any tank made to hold CO2 must be able to withstand the critical pressure of 73 bar. Including a reasonable safety factor it would normally be able to take at least twice that.

-Aluminum CO2 tanks are rated at 1,800 psi working pressure, and are typically fitted with a 2200 psi burst disc. 73 bar is a little over 1,000 psi, which gives us that 2X safety factor.


Aluminum and steel don't share the same reaction to repeated stress.

-Which is why most paintball manufacturers, or rather, the tank makers supplying the manufacturers, have been slowly shifting over to chrome-moly steel tanks, and away from the spun-aluminum tanks.

Doc.

Doc Nickel
04-02-2009, 10:38 PM
is there a requirement to heat treat these cylynders after manufacture, eg percipitation treat or whatever?

-Oh yes. I don't know exactly what kind of heat treating is done, but I recall one spec noting that some 40% of the tank's strength comes from the heat treat. Or to put it another way, the heat-treated tank is 40% stronger than the same tank, annealed.

But then, the same is true for steel tanks as well- maybe not the same ratio, but the material is still definitely heat-treated.

DOT regulations, in fact, require the destruction of any tank that's been exposed to more than X temperature for a given time period, though I can't recall if that spec was 150 or 200 degrees.

Many years ago, a few people were killed trying to fill some SCUBA tanks that the owners had had powder coated. The heat of the baking process annealed the tanks, which then exploded when filled. As noted earlier, the tank wall had become weaker than the burst disc, so the relief couldn't save them.

Personally, I had one customer seriously pissed off with me when I "destroyed"- I drilled a hole- in a tank he'd had powder coated to match his paintball gun. I don't know who did the actual work, but he sent it to me (along with the gun) for assembly and tuning. There was no way I was going to return the tank, knowing, as I did, it was horribly unsafe to use. If I refused to assemble it, he'd have just found some other doofus to do it- or done it himself- and wound up dead or injured for it.

I replaced the tank, but I doubt the customer really knew what I saved him from, and probably never forgave me. :D

Doc.

Rich Carlstedt
04-03-2009, 01:06 AM
Hey doc
Aluminum age hardens by natural occurance.
If those tanks blew after powder coasting, its because they were used immediately.
They should have set them aside for a month, or post treated them at 260 degrees for 24 hours
6061 for example goes from T1 to T 6 after 30 days at room temperature.

Looking at the failure says the bottle was faulty in my opinion.
The straight line up the side, like a zipper suggest a thin spot along the axis of the bottle. It could also have been subjected to a vise like clamp that marked the side , but the failure is not a normal hoop strength loss ?

The tank wall thickness should be miked !

rich

airsmith282
04-03-2009, 07:22 AM
thats a big mess , co2 is pretty dangerous stuff ,i been shooting and working on co2 pellet guns for a long time now and co2 runs about 1000 to 1500 psi on average usually closer to 1500 psi , tanks are suppoosed to be hydro tested but perfered to be replaced every 5 years your also supposed to pre chill co2 bolltles before filling them, and keep them away from heat sources, i let a gun sit on the table for about an hour one day it was about 25 C out side and all the seals in my gun just up and blew and that was only a gun using a 12gram bottle , i have to agree as well aluim should not be used for co2 but 12gram carts are steel and just as dangerous , but the aluim is pretty thick walled , if your going to strip down a bottle for reapainting id use something like poly stipper, taking off any kind of ruff surfacing id just go buy a new bottle the surface is supposed to be smooth on thoes bottles, aluim is a bad choice in that any out side heating of the bottle past reccomened will make them swell and blow up like bomb if the safty valve failes and sometimes the valve will fail if its not replaced from time to time as well, the HPA stuff i realy dont like, it runs 3000+ psi and the bottles are no where near the quility , i hav e seen them made from aluim to carbon fiber for paint ball guns and i dont trust the stuff at all expeciially the carbon fiber ,, so iam guessing that this bottle explosion had to be a heat issue and a failer of the safty valve..
anyhow i hope your friend is alright...

pressurerelief
04-03-2009, 07:50 AM
Rich, the failure of the cylinder is typical of a pressurized vessel failure. When a railcar goes like that we call it a dance floor. Lays it out nice and flat alot of times. I have a steel cylinder sitting on my bench that has the same type of result straight up the sidewall. The failure was from disassociation of the liquified gas in the cylinder and built up pressure from the hydrogen. Quite common from Hydrogen Fluoride and Hydrogen Bromide. I happened to be standing about 20 deet from the cylinder when it "went" and the only negative was ringing in the ears for hours. The gent closest to it might have had to clean out his drawers.

Airsmith, CO2 has a vapor pressure of approximately 835 psi. It is not and will not be over 1000 psi in your house. Refer to this link.

http://encyclopedia.airliquide.com/Encyclopedia.asp?GasID=26

Carbon Fibre cylinders are safe and have a well documented service history. They can withstand thermal cycling to some degree and still pass a hydrotest. If you doubt this look on the back of most firefighters, this is what they wear.

As far as CO2 cylinders Doc has it right and you can overfill a cylinder. I would like at an improper burst disk first and mechanical damage second as a root cause. Aluminum cylinders are prevalent in the market and have a low failure rate. Most of your calibration gas cylinders are aluminum. The DOT regulates these and keeps stats on failures. If they were not safe they would not be approved by engineers far more versed on this subject than I.


We had this same keyboard disagreement about overfilling a propane cylinder on this board. Alot of conflicting information here.

P/R

Jpfalt
04-03-2009, 11:52 AM
Doc and Rich,

Precipitation hardened (or age hardened) aluminum has to be solution annealed before it can be rehardened. Solution annealing happens at about 950 deg F. and is immediately quenched in water or such. At this point the preciptating material is dissolved in the aluminum like sugar dissolves in water. After solutiuon annealing, the aluminum is baked at about 250 deg F for a set time (time and temp depend on the alloy). The most common precipitating addition to aluminum is silicon or copper. The highest strength for the aluminum is just before the silicon or copper actually starts to separate out of the aluminum in particles.

Precipitation hardened aluminum that is heated over time becomes weaker and more brittle with time and temperature as the particles of silicon or copper get larger and less silicon or copper remains dissolved in the aluminum. This condition is called over-aging and the microstructure of the material would show a lot of precipitated inclusions in the aluminum.

Powder coating is usually baked around 300 to 350 deg F which is way below the solution annealing temperature, but well above the temperature needed to get the silicon or copper to come out of solution.

The only way to fix over-aged material is to re-solution anneal the material by heating the material to somewhere around 1000 deg F, quench and then re-age the material at some time and temperature around 250 deg. F.

philbur
04-03-2009, 03:27 PM
How does a fixed mass of gas in a tank with a fixed volume increase in density by increasing the temperature.

Phil


It has infinite heat capacity which means an increase in temperature increases density instead of pressure.

Rich Carlstedt
04-03-2009, 09:15 PM
Thank you PR and Jpfalt !
Good concise explanations
My experience is with high pressure liquids (15 K PSI) and Hoop failure. and decidedly different results.
Most likely because of the non-compressable nature of liquids (?)
Rich

(non-compressable as compared to gases.)