Oxygen Explosion - YouTube
---------- Post added ----------
This reply intriqued me and I wanted to understand why this could be. I decided to play around with some numbers to try and understand and/or explain if everything I said earlier was correct or what others are saying is correct. The following is my attempt to explain what's going on during a fill.
When a cylinder is Hydrostatically tested, it is filled to 5/3'rd (or 166%) of its service pressure. This means that AL 80's rated at 3000 PSI would be filled to 5000 PSI (3000 PSI / 3 = 1000 PSI then * 5 = 5000 PSI) at the hydrostatic testing facility. The expansion of the cylinder is measured and recorded before and after the test and if it remains expanded beyond certain tolerances then the cylinder is condemned otherwise it passes hydrostatic testing.
Cylinder valves have a burst disc to help prevent cylinders failing from over pressurization (commonly in-case a fill station operator leaves the fill whip connected and walks away for too long). A properly matched burst disc generally tolerates a 40% pressure expansion beyond the cylinders service pressure before it fails. An AL 80 cylinder rated at 3000 PSI with a properly matched burst disc should fail at 4200 PSI (Service Pressure * 40% = 3000 PSI * 1.4 = 4200 PSI).
NOTE: If an aluminum cylinder filled to its rated service pressure is heated high enough to cause the burst disc to fail, then if the same cylinder were filled with to half or less pressure, it may weaken from structural fatique before reaching burst disc pressure. Aluminum cylinders begin to fatique after about 300* F.
Cylinder volumes/pressures are considered to be at "room temperature" because they spend the majority of their time sitting in a room (not in a fire, not in the trunk of a car in the middle of the Saudi Arabian desert, and not in a Hydostatic testing facility). Room temperature is defined as 72* F. Manurfacturers use this value when establishing a tank's volume/service pressure rating.
There are a few different methods used to fill a cylinder (my definitions may be a bit off)
*) Hot-fill. When a cylinder is rapidly filled to reduce the fill time. Generally this fill is beyond roughly 600 PSI/minute.
*) Slow-fill. When a cylinder is filled at roughly 500 PSI/minute or less
*) Wet-fill. When a cylinder is filled inside a water bin -- usually slow'ish. Wet-fills convect heat away from the cylinder faster than if it was sitting dry at room-temperature.
The very act of filling a cylinder generates heat. This is because molecules generate heat as they are squeezed into a confined space. The more rapid the squeezing, the higher the temperatures. This squeezing is called "compression". Compression is a hot-topic! (bada-boom, bada-bing)
Charles' Law states that (Pressure * Volume)/Tempurature is constant. This means that P1/T1 = P2/T2. When computing pressure changes related to temperature, we must first adjust F to absolute zero which brings us to the Rankine scale -- which is -460* F. This means that zero degrees F = 460* Rankine and 72* F = 532* Rankine (460+72).
I could not easily find an actual measurement or statmeent of the temperature of a cylinder during hot-fill. Let us assume a cylinder reaches 120*F during a hot-fill in a room that is 72* F then we could compute:
P1/T1 = P2/T2
=3000/(460+72) = P2/(460+120)
=3000/532 = P2/580
P2=(3000/532) * 580 = 3384 PSI
What does this mean? This means that if we hot-fill an AL80 cylinder with a service pressure of 3000 PSI to 3384 PSI (with an internal cylinder tempurature of 120* F during fill), we are still well beneath the burst disc pressure of 4200 PSI and signicantly below the Hydrostatic pressure test of 5000 PSI.
This means that it should be safe to over fill a currently-in-hydro AL80 cylinder to 3384 PSI without fear of structural fatique *IF* that cylinder has not previously been exposed to heat in exceess of 300* F at any point in the past and *IF* it is in good condition (no nicks, scratches, dings, bulges, bows or corrosion beyond prescribed tolerances) and *IF* the burst disc itself is not faulty.
Just for fun, I wanted to see what the pressure would be of a fully filed AL80 cylinder if exposed to a 300* F fire:
P1/T1 = P2/T2
=3000/(460+72) = P2/(460+300)
=3000/532 * 760 = 4285 PSI
Recall earlier that the burst disc of this cyclinder should tolerate up to 4200 PSI? This burst disc should fail just before reaching 300* F if it was filled to 3000 PSI at room temperature before being exposed to the heat source. But what if the same cylinder filled to 1200 PSI was exposed to the same fire?
1200/532 * 760 = 1714 PSI
This means that the cylinder itself would likely fail from heat featique long before the burst disc fails from over pressurization.
Next time a fill operator refuses to fill beyond service pressure alone perhaps you can explain this to them but remember: "their shop, their rules". You are always free to go somewhere else for a fill.
Hope this helps!
EDIT: The DOT does allow a cylinder to be over-filled to a pressure that will cool to its service pressure at room temperature. I was not able to dig in an find an exact reference online to link to but hopefully someone here can (J1Scuba has a Hydro facility perhaps he can point us in the right direction ??... J1... J1?)
---------- Post added ----------
Now that I've explained how pressures and temperatures work with cylinder fills, I wish to complete my investigation of the wisdom of people who do such overfills.
Low-pressure tanks are generally 2400/2640 PSI. 3000 PSI is generally considered medium-pressure and 3442+ PSI is considered high-pressure. Because I do not know what cylinders CBRICH observed, I will assume a use an LP95 as an example. An LP95 is a low-pressure steel cylinder rated to 2640 PSI when "+" rated. They are popular in cave-country because they can be overfilled and have characteristics that make them more preferable to use than HP 100's or 130's. For purposes of this investigation, I will assume the cylinder is "+" rated.
The burst disc of such a cylinder should tolerate 3696 PSI before failing to over-pressurization. Let us assume the tanks filled at 3500 PSI heat to 130* while driving through a desert in the summer. The pressure should be:
P1/T1 = P2/T2
=3500/(460+72) = P2/(460+130)
=3500/532 = P2/590
P2 = (3500/532) * 590 = 3881 PSI
What does this mean? Well, first it means that the tanks will overheat causing the PSI to skyrocket beyond the 3696 burst disc tolerance to a whopping 3881 PSI. In order to allow this, a higher rating burst disc must be applied to the tanks. Even a burst disc for a 3000 PSI rated cylinder would not be enough. One would have to use a 3500 PSI service pressure rated burst disc.
What is the hydrostatic testing pressure of the same tank? (2640 / 3 * 5 = 4400 PSI). Doing this would bring the pressure of the LP95, filled at 3500 PSI at a tempurature of 130* F to within 600 PSI of hydrostatic testing limits.
This means that the cylinder could potentially fail and explode before the burst disc.
But what if the cylinder was an AL80? The same formula would apply above to reach 3381 PSI but the burst disc pressure would be 4200 PSI bringing the cylinder pressure to within 400 PSI of burst disc pressure.
I'll leave it up to you to decide if that is a risk worth taking.