Filling LP tanks to high pressure

[halocline]

Some tank manufacturers are producing tanks rated for low pressure now that are identical to older HP tanks. Reportedly, they're identical in material, build, and design. As far as I'm aware, the reason for the change in rating was to allow for the 3/4" neck threads instead of the smaller (weird, right?) 7/8" neck threads.


As for highway accidents, isn't gauge pressure the main concern? Like, not gauge vs rated pressure....but, just gauge pressure in the tank.

I'd be curious about the identical tanks with different working pressures. I've sort of heard that too but I'm a little skeptical. Even if it's true, the larger opening for the valve presents a potential structural weakness that wasn't there before, so the tanks aren't identical.

With highway accidents, I believe the issue is how much impact the filled tank can withstand without rupturing. If the tank is filled far beyond it's intended pressure, the presumption (backed up by structural analysis and certainly testing) is that it will rupture with a smaller impact than it would if it were filled to rated pressure, and that's true for all cylinders, LP and HP. That's pretty intuitive to me, but there are so many variables in a car accident that I couldn't imagine calculating the increased likelihood of a rupture. But it is there, anyone who denies that is simply not acknowledging the facts.

Again, it comes down to personal comfort level. There is a demonstrable greater risk based on the structural design of the tanks, but there is also a demonstrated track record of cave fills with no incident. So the greater possibility of a bang-bang is obviously low.

 

[Luis H]

As for highway accidents, isn't gauge pressure the main concern? Like, not gauge vs rated pressure....but, just gauge pressure in the tank.

If you are asking about the amount of stored energy being the same for the same pressure and volume, irrelevant of the container it is in, that is correct. If you have 100 cu ft at 3500 psi you have the same stored pneumatic energy independent of the container or pressure vessel. It could be in a HP cylinder, in a ASME 5000 psi (high safety factor cylinder), in an LP cylinder, or even in a rusty crappy cylinder. The energy stored is the same.

The big difference (as halocline pointed out) is that the higher the safety factor for the stored pressured (and corresponding stress), the more capacity the cylinder has to stay intact in the event of extraneous loads, like impact loads.


Anyone that is not familiar with this expression, please Google it: "Straw that broke the camel's back".
You can also look at this:
Straw that broke the camel's back - Wikipedia, the free encyclopedia

With highway accidents, I believe the issue is how much impact the filled tank can withstand without rupturing. If the tank is filled far beyond it's intended pressure, the presumption (backed up by structural analysis and certainly testing) is that it will rupture with a smaller impact than it would if it were filled to rated pressure, and that's true for all cylinders, LP and HP. That's pretty intuitive to me, but there are so many variables in a car accident that I couldn't imagine calculating the increased likelihood of a rupture. But it is there, anyone who denies that is simply not acknowledging the facts.

I could not have said it better.

Some tank manufacturers are producing tanks rated for low pressure now that are identical to older HP tanks. Reportedly, they're identical in material, build, and design.

I am not sure what manufacturers you could be referring to, but I am fairly certain this statement contradict the code. The code for 3AA cylinders (49cfr178.37) is fairly specific about the chrome-molybdenum steel alloys allowed for the fabrication of cylinder under this code. The alloys for HP cylinders is different.

About 8-9 years ago I worked for a company that manufactured ASME pressure cylinders. At that time I a couple of conversation with the head of the engineering department at PST. I don't recall details of the HP cylinder alloy, but I can assure you it is not AISI 4130 (or any of the other allowed 3AA alloys).

The point that I'm making is that, without actually knowing the dimensions (wall thickness, diameter, etc) of the tank and the material of the tank, we don't know what the actual design stress is as rating systems are so arbitrary. As you said, it's easy to calculate....but I just don't have the dimensions needed for it.

The actual design stress is very clearly specified for 3AA cylinders.
This is directly from 49cfr178.37

(2) For cylinders with service pressure
of 900 psig or more the minimum
wall must be such that the wall stress
at the minimum specified test pressure
may not exceed 67 percent of the minimum
tensile strength of the steel as
determined from the physical tests required
in paragraphs (k) and (l) of this
section and must be not over 70,000 psi.
(3) Calculation must be made by the
formula:
S = [P(1.3D2+0.4d2)]/(D2-d2)
Where:
S = wall stress in psi;
P = minimum test pressure prescribed for
water jacket test or 450 psig whichever is
the greater;
D = outside diameter in inches;
d = inside diameter in inches.

Here is an old post of mine at VintageDoubleHose.com (Sat May 08, 2010 1:02 pm) showing some of the dimensions and calculations I have done with some of my personal cylinders.

I spent the day yesterday doing hydros on seven of my tanks. My LDS has a hydro station that I use… I still pay for the hydros, but I do the work and control the duration of the pre-test, etc. I like doing my own hydros.

I also got wall thickness readings on 9 of my steel 72. A co-worker took the readings using a precision ultra-sound NDT equipment. I drew a pattern of dots on my tanks and took reading every 3 inches to get a precise average.

This procedure meets (and exceed) CGA C-5 in determining an average REE for this class of tanks. I own enough Norris and PST steel 72 tanks that I can calculate an average REE number for the type of tank (per the CGA procedures). Whether your local hydro station accept this REE number to plus “+” stamp your tanks or not, I don’t know. You are always welcome to bring them up to Maine I can recommend a nice dive shop that would gladly hydro them with the + stamp for you (assuming it meets the criteria, which probably does).

All but a few of my 3AA tanks are already + stamped. By the after the next hydro I expect all my tanks to be plus stamped, including my 45s and 40s double tank sets.


Below is some of the data on my tanks. There is still a lot of empty blanks, but I should have at least one data set for all of my tanks by the end of the year or early next year.

In the first two tables each row is for the data associated to a tank. The third table is in columns; with each column has the calculations for the tank identified on the top. On the far right of table three shows the equations used in each column.

Tank dimensions and wall thickness:

Tank hydro data and volume data:

Tank REE and hydro calculations:

more later

I have 14 steel 72 tanks that I have been taking some precise measurements.


Wall thickness measurements were taken using a precision ultrasound measuring device. The ultrasound probe was calibrated before each set of readings using a cut off (condemned) steel 72 tank. The cut off tank wall thickness could be confirm with a caliper and provided a sample of the exact same material to calibrate the ultra sound equipment. Measurements are good to about 2/1000 of an inch.


I measured the actual internal volume by using water at room temperature and precisely measuring the weight of the water. I took into account the reduced volume from the valve threads.


I am only showing some of the data below to save space. The “Tank #” is only the number I assign and painted on all my tanks to keep track of them. I have full records (in a spread sheet) with all the tank information, including serial number and all the tank markings (including all the hydro test markings).


I also have records of all the hydro test data, etc.


I am only including seven of my tanks at this point (plus a friends Walter Kidde tank). I still need to take precise volume data on the other tanks. I do have wall thickness for all of them. I need wall thickness and volume data to calculate REE numbers for that specific tank.


At this point I have complete data for four Norris tanks, three PST tanks, and a friends Walter Kidde tank (I don’t own any Walter Kidde).


The data is shown as follows:
Tank# / manufacturer >> average wall thickness / tank empty weight (with boot and valve) / tank actual volume at 2475 psi / actual REE nunber


1 / Norris >> 0.177 inches / 30.26 Lbs / 72.2 cu ft of air (@ STP) / REE = 61.3
2 / PST >>>> 0.171 inches / 28.88 Lbs / 70.5 cu ft of air (@ STP) / REE = 58.6
4 / PST >>>> 0.181 inches / 30.11 Lbs / 70.8 cu ft of air (@ STP) / REE = 61.5
5 / Norris >> 0.179 inches / 29.73 Lbs / 71.4 cu ft of air (@ STP) / REE = 60.7
6 / PST >>>> 0.173 inches / 30.43 Lbs / 70.9 cu ft of air (@ STP) / REE = 58.7
7 / Norris >> 0.178 inches / 31.01 Lbs / 70.3 cu ft of air (@ STP) / REE = 58.4
8 / Norris >> 0.175 inches / 29.73 Lbs / 71.4 cu ft of air (@ STP) / REE = 61.2
No# / WK >> 0.183 inches / 29.63 Lbs / 71.0 cu ft of air (@ STP) / REE = 63.6




The variation in volumes is not very much considering the manufacturing process used for these tanks. As a mater of fact I consider the variation in wall thickness to be fairly impressive. The wall thickness standard deviation for the data collected for each tank is also relatively small.




Tanks with a + stamp (DOT 3AA code) are always advertised as having the capacity with the extra 10 % overfill.


An interesting side note. I also measured two PST HP 80s. They are supposed to have 80 cu ft of air at 3442 psi… they actually have 85.3 cu ft and 85.0 cu ft.


Note: as expected, the actual REE numbers are all higher (or equal) to the published number by PST. As a mater of fact I believe the REE number on tank 7 came out low due to lack of precision on the hydro test data. I believe I had a tiny leak during hydro, but not bad enough to do it over. I will confirm the data in 5 years.

---------- Post added January 14th, 2014 at 08:13 PM ----------

The last portion of this post is just general LP steel cylinder information.

 

[victorzamora]

Luis, your first couple of posts I totally get. With the gauge pressure question, I was more confused because I was understanding something different from the conversation that had been going on. I was just clarifying.

As for the stress and materials issue, thanks for the details. From what I understood, they simply changed the neck size and changed the certification. As for the bigger threads being a failure point, they're only marginally bigger...I don't think there's a HUGE difference.

 

[pocky21]

Thanks.

The hydro test pressure is intended to be at the lowest limit of yield strength......

......The elastic expansion is normally around 55 cc and
The permanent residual (plastic) expansion often runs less than 1 cc, but the worst case was 4 cc.
For a few of my cylinders the permanent expansion was 0%. Their expansion was 100% elastic.
So you can see that the hydro test pressure is the very low end of the yield stress level, but it is right at the starting region.

Thank you. This is exactly the information that I was looking for, but I do have a few follow up questions. Suppose a tank has been overfilled to the point of plastic expansion enough to produce a volume of, say, 50cc more than it's baseline capacity. The tank then goes in for hydro which adds, say, another 5cc (but still passes) and it's service is continued.

As far as I can tell, there is no baseline established for a particular cylinder's volume, and the hydrotest will measure only the initial, loaded, and rebound volumes of a particular tank. In theory, an overfilled LP tank could see any number of plastic expansions before it's deformation is measured, and even then it's an isolated measurement. In this case, the tank that left the hydro facility with 55cc of expansion from it's born date might return next round with 100cc of expansion, and nobody would know it.

Obviously very few keep records like you have, but I am curious...how much expansion is too much when looking at the life of a cylinder? It would seem to me that, for a cylinder operating continuously at close to or above yield strength, a simple hydrostatic test may be limited in it's usefulness. Is this a valid point to consider?

 

[Jared0425]

What's going to be the difference between a tank exploding at 2600 compared to 3500? When a tank goes she is gonna make on heck of a mess no matter what the pressure is. As I stated in my first post, I was questioning the material the LP tanks were made out of, not talking about a catastrophic failure or some freak accident that would cause any tank to blow. As I said before, people in the "cave" country encouraged me to over fill the tank 4000, which I said was not my intention but 3400-3500 is good enough for me. They claim that they have dove the tanks for years and years and have never suffered any problem what so ever. My curiosity was if there is a material difference in the manufacturing of the tank that causes it to be rated to low pressure. I was also told that LP tanks exist because some dive shops or compressors cannot pump past 2500.

 

[flots am]

What's going to be the difference between a tank exploding at 2600 compared to 3500?

The difference is that the 2600 tank might not expode in situations where the 3500 tank will.

The "overfillers" are using the the additional pressure for additional air, instead of as a safety margin as designed.

 

[Luis H]

Luis, your first couple of posts I totally get. With the gauge pressure question, I was more confused because I was understanding something different from the conversation that had been going on. I was just clarifying.

As for the stress and materials issue, thanks for the details. From what I understood, they simply changed the neck size and changed the certification. As for the bigger threads being a failure point, they're only marginally bigger...I don't think there's a HUGE difference.

Oh, now I know what you are you are talking about.

Yes they change the tank pressure rating from 3500 psi down to 3442 psi.

I don't know all the details, but my understanding is that the 3500 psi is the threshold of what CGA consider the limit for a yoke connection. Therefore 3500 or higher can not use the 200 Bar/ yoke convertible valve.

According to CGA those tanks could not accept the that valve and therefore the manufacturer (I think they were mostly PST) made them with the narrower neck.

Structurally either neck is fine. I have seen 4500 psi cylinders with the 3/4 -14 NPSM threads.

Going from 3500 psi to 3442 psi was just a trivial change to meet a code.




BTW
My wife has a couple of HP 80 (3442 psi) and she really likes them. I have been looking to by her a third one, but I will not buy the ones with the narrow neck. We don't normally use DIN fitting, but far more important is that I hate to service cylinders with narrow necks.

 

[halocline]

Thank you. This is exactly the information that I was looking for, but I do have a few follow up questions. Suppose a tank has been overfilled to the point of plastic expansion enough to produce a volume of, say, 50cc more than it's baseline capacity. The tank then goes in for hydro which adds, say, another 5cc (but still passes) and it's service is continued.

As far as I can tell, there is no baseline established for a particular cylinder's volume, and the hydrotest will measure only the initial, loaded, and rebound volumes of a particular tank. In theory, an overfilled LP tank could see any number of plastic expansions before it's deformation is measured, and even then it's an isolated measurement. In this case, the tank that left the hydro facility with 55cc of expansion from it's born date might return next round with 100cc of expansion, and nobody would know it.

Obviously very few keep records like you have, but I am curious...how much expansion is too much when looking at the life of a cylinder? It would seem to me that, for a cylinder operating continuously at close to or above yield strength, a simple hydrostatic test may be limited in it's usefulness. Is this a valid point to consider?

I bet Luis will reply to this, but in the meantime, if I understand your question, you are thinking that repeated overfills to near test pressure would cause some permanent expansion of the tank, and that the hydrotest does not measure this, it only measures the percentage of return to the pre-test volume. I think you're correct. You're supposing that since the hydrotest doesn't measure previous amounts of permanent expansion, it can't measure the true condition of the tank?

I think the answer is that if the tank undergoes permanent expansion before the test (i.e that the test won't measure because it was done through prior fills at near test pressure) then the elastic characteristic of the steel also changes, and that results in the tank being more likely to fail hydrotest, even though the total expansion during the test could be less due to the prior permanent expansion. That's a guess on my part, it would be interesting to hear if it's more-or-less accurate.

 

[NAM001]

Jarad

This thing about lp tanks are made for shops that cant pump to 2500 i think is a stetch. If you go back a few decades you will find that (pardon me for saying this) all the tanks were 1800-2250 tanks (pre al days). It is my opinion that as the sport caught on the demand for higher volumn tanks came to be. You either had to increase the size of the tank or increase teh presure of the tank. I think that at one time the al tanks as they entered the market were like 2200-2500 tanks and then after a while they became what we have grown to love as the al 80's at 3k. They may have been the same size as the curent al80s but at a lower presure, i dont remember. Even the hp tanks are relatively new to the scene. Many compressor systems were not made to handle the (todays) hp preasure areas we now use. Now it is rare to find a compressor that can not pump to 3000. It is true that many shops can not fill hp tanks completely. Note the line between hp and lp is at around 3000-3200 now as opposed to <2k so even a al80 is now a lp tank. You see this referenced to when using din vs yolk regs yolk are for lp tanks and din is for hp tanks. So in that regard yes many shops can/will only fill lp tanks. In street terminology a 3k tank is not considered a lp tank. Lp functually refers to 2400 psi tanks and below. If you were to ask most people what lp tanks are mad from they would say steel. There was once a time that people believed that al was stonger because steel tanks were lp 2000-2500 psi tanks adn al's were hp 3000 # tanks.

 

[pocky21]

I bet Luis will reply to this, but in the meantime, if I understand your question, you are thinking that repeated overfills to near test pressure would cause some permanent expansion of the tank, and that the hydrotest does not measure this, it only measures the percentage of return to the pre-test volume. I think you're correct. You're supposing that since the hydrotest doesn't measure previous amounts of permanent expansion, it can't measure the true condition of the tank?

I think the answer is that if the tank undergoes permanent expansion before the test (i.e that the test won't measure because it was done through prior fills at near test pressure) then the elastic characteristic of the steel also changes, and that results in the tank being more likely to fail hydrotest, even though the total expansion during the test could be less due to the prior permanent expansion. That's a guess on my part, it would be interesting to hear if it's more-or-less accurate.

That would be the question I am presenting, yes.

The elastic characteristics changing makes perfect sense, and is to be expected, but what's important is the phrase "more likely to fail". I guess what would be important to think about is whether expansion per cycle, modeled as a function of total lifetime expansion, is more exponential or linear in nature. If it's linear, a hydrotest may be sufficient at predicting an impending failure. If it's exponential, maybe not so much.