I'm sure you've all read the warnings on the side of old propane bottles:
Federal Law forbids transportation if refilled - penalty up to $500,000 fine and five years imprisonment (49 U.S.C. 5124)
But have you ever wondered why they can't be refilled? Is it a vast gas-cylinder-manufacturer conspiracy to sell more gas cylinders? Or is there really a reason? What is the reason? Crap metal? Thin metal? Why?
A corollary question is: at what pressure is a refilled non-refillable tank safe? Half of the rated pressure? A quarter?
How many people are killed or injured because of refilled non-refillable tanks? Is it really so dangerous that it warrants a half-million dollar fine?
Well, I did a little research and are some of the answers I came up with.
The standard for non-refillable cylinders in the U.S. is DOT 39. You can identify a DOT 39 cylinder from the labeling on it, which is as follows:
- The service pressure
- The test pressure
- The registration number (M****) of the manufacturer
- The lot number
- The date of manufacture if the lot number does not establish the date of manufacture
- With one of the following statements:
- For cylinders manufactured prior to October 1, 1996: "Federal law forbids transportation if refilled-penalty up to $25,000 fine and 5 years imprisonment (49 U.S.C. 1809)" or "Federal law forbids transportation if refilled-penalty up to $500,000 fine and 5 years imprisonment (49 U.S.C. 5124)."
- For cylinders manufactured on or after October 1, 1996: "Federal law forbids transportation if refilled-penalty up to $500,000 fine and 5 years imprisonment (49 U.S.C. 5124)."
The markings required by ... this section must be in numbers and letters at least 1/8 inch high and displayed sequentially. For example: DOT-39 NRC 250/500 M1001.
On the other hand, there are several standards for refillable cylinders, depending on their use, size, how they're welded, etc. One of the more common is DOT 4BA, "a cylinder, either spherical or cylindrical in shape, with a water capacity of 1,000 pounds or less and a service pressure of at least 225 and not over 500 psig."
All the DOT "Specifications For Packagings" can be found here.
Physically, the refillable and non-refillable tanks are pretty much the same. The specifications for the steel is the same:
(The DOT spec for steel for refillable cylinders also specifies various other elements, while the DOT 39 spec doesn't specify them.) (I am NOT addressing aluminum DOT 39 tanks - only steel tanks.)
How about the thickness of the metal? DOT 4BA specifies a minimum wall thickness of 0.078 inches (14 gauge), and the lower of a maximum stress of 35,000 psi or half the emprically determined strength of the material. DOT 39 only specifies that "the wall stress at test pressure does not exceed the yield strength of the material", so it's hard to say how these compare.
Both standards specify that wall stress shall be calculated using this formula:
S = [ P ( 1.3D + 0.4d ) ] / (D - d)
Where S = Wall stress, in psi; P = Test pressure in psig; D = Outside diameter, in inches; d = Inside diameter, in inches. Obviously, wall stress in tanks of identical proportions will be identical.
As it appears metal of both refillable and non-refillable tanks is the same, then
it seems reasonable to assume that both have a yield strength of 35,000 psi. [Wrong! see below.] If so, the wall thickness of a 9" (ID) DOT 39 tank at 325 psi works out to a minimum of 0.072 inches, only 8% thinner than the 0.078 inches required by DOT 4BA. Moreover, 0.072 is an odd size, between 14 guage and 15 gauge. I think it's unlikely that DOT 39 tanks would actually use 0.072 instead of 0.078 inch metal, though that is just a hunch. There are way too many assumptions made in this paragraph to come to any conclusions. I've got a 9" DOT 39 tank here - when I get a chance I'll cut it open and measure it.
The few other distinctions (like labeling and testing procedures) are pretty minor and don't really effect the nature of the tanks themselves, just how they're handled.
First, DOT 39 tanks are not heat treated.
DOT 4BA reads:
Heat treatment. Cylinders must be heat treated in accordance with the following requirements:
- Each cylinder must be uniformly and properly heat treated prior to test by the applicable method shown in table 1 of appendix A to this part. Heat treatment must be accomplished after all forming and welding operations, except that when brazed joints are used, heat treatment must follow any forming and welding operations, but may be done before, during or after the brazing operations.
- Heat treatment is not required after the welding or brazing of weldable low carbon parts to attachments of similar material which have been previously welded or brazed to the top or bottom of cylinders and properly heat treated, provided such subsequent welding or brazing does not produce a temperature in excess of 400°F in any part of the top or bottom material.
"Table 1 of Appendix A" specifies "any suitable heat treatment in excess of above 1100°F, except that liquid quenching is not permitted." (Liquid quenching would harden the metal.)
As a result of the manufacturing process, the metal that tanks are made from is work-hardened and brittle. Heat treatment softens the metal and restores it's elasticity. Without this elasticity, the metal won't "give" and distribute loads evenly, instead concentrating them around flaws in the metal, dents, scratches, crappy welds, rust spots, or sharp bends. These high load concentrations fatigue the metal greatly - and more important, unpredictably - making reuse of a DOT 39 tank a gamble.
Second, DOT 39 tanks are stessed to the limit of their capacity. I had missed this point at first, but Wesley Scott (who works for a company that wholesales industrial gases, but he asked that I not name it) wrote to me and pointed out that:
The wall thickness [of the DOT 4BA cylinder] must be increased to the point where the wall stress will be below the lower of 35,000 psi or 1/2 the empirically determined strength of the material.
The DOT 39 cylinder is limited to not exceeding the yield strength.
To compare these specifications consider a steel with a yield strength of 80,000 psi. The stress in the the [DOT] 4BA cylinder is limited to 35,000 psi since 35,000 is less than 1/2*80,000 = 40,000. The stress in the [DOT] 39 cylinder is limited to 79,999 psi. This means the 39 cylinder can be made of thinnner steel than the 4BA cylinder.
The heat treatment improving the characteristics of the metal is also a valid concept. The DOT 39 cylinders are frequently brazed and the brazing material is specified as "must have a melting point of not lower than 1,000 F". If you were heat treating at 1100F, you would probably have the brazing material melting resulting in unusable cylinders.
Refilling non-refillable gas cylinders is dangerous because the metal is stressed to it's maximum, and because the inelasticity of the metal concentrates stress and fatigues the metal.
But HOW dangerous is it? Where are the statistics on how many people blew themselves up refilling non-refillable tanks? Empirical observation presents a problem: some people, including friends of mine, refill non-refillable propane cylinders, apparently without any problems at all beyond leaks. Presumably, this is just some fluke. There may well be data out there somewhere, but I haven't been able to find it. One would hope that our benign, beneficent government wouldn't impose a $500,000 fine on any activity that wasn't truly dangerous, but until I see some numbers, I'm skeptical about how dangerous is really is.
But, that said, I'm going to Sears, Lowes, or Ace Hardware and buying a refillable tank instead.
© 2003 W. E. Johns