Hydrotesting of custom pressure vessel

[MrComputer]

I am working on a project where we have a custom pressure vessel that is made with 1in DOM tube with some end caps brazed onto the ends. The intention is to pressurize this up to 9ksi with nitrogen gas for a test we are doing. I had no involvement with the design of the vessel; however, I am not trusting of the brazed end caps for such high pressure. I had two of these vessels hydro tested today and both of them lost pressure and began leaking at 8ksi. Further inspection does not reveal any leaks or visible cracks in the vessel. Is it possible this leak/loss of pressure was solely due to the threads? Unfortunately, it is an NPT thread. After the test failed at 8ksi, we were able to retest it up to 8ksi before it started leaking again. Being new to hydro-testing, is this a common result? Is the ability to re-pressurize to a high pressure a sign that there is a bad connection or would this be an indicator of a crack in the material or end cap braze?

 

[Tuckabag]

Where was the leak coming from? That's your first port of call.
I assume that the brazed end caps then have NPT threads machined in them?
If you can re-pressurize to around the same pressure again before leaks occur it's a fair bet that the threaded connections are leaking.

 

[LittleInch]

I'm lost here?

Where are these threaded components? What is their rating and size?
Why is NPT a bad thing?
Can you draw out and describe each item (valves, pressure source, gauges etc)

Can you test with soap solution like this

http://www.wintonproducts.com/id21.htm

If not why not?

It might be a failure eof the item or maybe a failure eof your test equipment. Can you try pressure testing something you know is good for the pressure to see?

I think you're right to be nervous of brazed end caps - they could detach and become projectiles - so I hope this has been considered in your test environment. 9000 psi sounds like a lot to me for such items.

why can't this be tested using water or oil / hydraulic fluid?


Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[MrComputer]

The leak was not visible as the vessel was tested in a closed water jacket cylinder hydro testing apparatus. The machine is designed to test DOT rated cylinders. It was the closest testing facility I could find. The vessel is literally a 1in tube, say 1ft long with two end caps brazed to the ends of the tube to enclose. On one end the end cap was drilled and tapped to NPT. This was tested with water at which lost pressure at 8ksi. We intend to hit them with gas to 9ksi so the hydro test is mainly to check the brazed caps. However, I found it odd that 2 separate vessels were able to maintain pressure up to 8ksi multiple occasions then leak. All components are rated to at least 10ksi. Could it be not using enough teflon on the npt fittings.

I say unfortunately NPT because of the designed to leak situation that is going on there.

Thanks for all the advice.

 

[btrueblood]

Why do you have a hydro test article submerged? Generally, something that size is relatively safe to stand next to, even at 9ksi, IF you have done your setup right and purged all the air out. At the very least, put it on the other side of a wall, and watch it via a webcam, and see where the leak is coming from.

Brazing doesn't mean much either way, I've seen braze joints at higher pressures work well - depends on the design of the joint.

"Could it be not using enough teflon on the npt fittings. " Could be. Could also be too much. Did you use any pipe sealant (Loctite or dope) along with the ptfe tape?

Could also be that your pump can't make it to 9 ksi, and starts leaking back at 8k. Did you test the pump with a dead-headed gauge first? When was the gauge last calibrated?

 

[JGard1985]

Good point problem with the tools, not the product!

Jeff
Pipe Stress Analysis Engineer

http://www.xceed-eng.com

 

[MrComputer]

The only reason it was submerged was that the company/location I work at does not have any hydro testing equipment. Had to take it to a local shop that tests scuba tanks and the such. Right before the test we were able to test one of the testing locations cylinders up to 10ksi. I did not use any loctite or similar sealant just the ptfe tape. I appreciate the help. Would it be best to throw some sealant on there with the ptfe?

 

[btrueblood]

Submerging a cylinder being tested with air or nitrogen makes sense to me. Submerging a cylinder being pressurized by water makes less sense. But, I get that this is not YOUR test facility.

You can buy hand pumps that will pump to 10-12 ksi with hydraulic oil, see McMaster-Carr. For less than $500 your company would suddenly have hydro testing equipment.

Yes, use sealant.

 

[MrComputer]

Thanks for the advice. I checked McMaster and could only find 6ksi hand pumps. But I do have a hydraulic jack system I might try to tap into. I believe they submerge these things for scuba tanks because its more of a consumer type test that the DOT requires, so it is found in marine shops locally and may be required for additional safety since they likely wont have room for a test range for larger cylinders. That is just pure speculation. Is it common to use hydraulic oil to hydrotest something? If I cant get the jack unit to pressurize it to what I need I will take it to a shop again but try the sealant option. I will report back my findings. Thanks again.

 

[itsmoked]

btrue; The tests consist of putting the DUTs in a water vessel that is 100% full of water. Tested by bleeding out a top-most petcock. A gauge is set then the DUT is pressurized with the gauge showing the water displaced by the expansion of the DUT under pressure. A marker is set then the pressure is released from the DUT which causes it to contract back towards its unpressurized size. The difference on the gauge is the amount the DUT did not recover. DUTs that fail the test are the ones that have lost their elasticity and hence don't recover more than X% of their original size.

Pressurizing the DUT with a gas in one of these test fixtures would be a very bad idea in my opinion.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[LittleInch]

This is a piece of 1" pipe 300mm long with an end cap with a screwed connection in the end cap.

"On one end the end cap was drilled and tapped to NPT. "

What thickness are we talking about for the screwed part of the connection? Can't be more than 5mm?

I think the fact it has held 8000 psi without blowing out is a near miracle.

Have you got a drawing / sketch / photo?

I think you really need to re-think your connection into this minute piece of pipe.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[saplanti]

I have designed similar vessel almost 35 years ago for filed gas samples for the diameter of 2 1/2 in and 10ksi internal pressure. I used an o-ring seal on each side in addition to the NPT threaded connection of heads. It was hydraulic and leak tested before use. Seal design is an important part for sealing against the leak under gas pressure.

Please remember that there are other connections for valves and pressure gauge in the application. In case they are not leak free you will loose pressure anyhow.

Good luck.

 

[MrComputer]

That's a little sketch. The end caps are much thicker than the tube. There is about 0.5in of NPT thread in the end cap. I recognize the poor design in this situation; unfortunately for me, my only responsibility is to test these vessels and have them fabricated. I appreciate all the advice, good stuff. The actual pressure system I have to pressurize these vessels was designed to use coned and threaded connections and is rated at least 2.5 times greater than the pressure I will be using. The main purpose of testing these vessels was originally to ensure that those brazed end caps can hold that pressure. The numbers I have crunched for the vessel itself suggests that we will be close to yield but about 10ksi under. I am going to try some additional sealant tomorrow and retest this sucker. I will update the results.

 

[mk3223]

Since it's your responsibility, you need to find a proper hydro testing apparatus for your equipment, so you may identify where is the leak. Otherwise, we may only guess every possibilities, which may not help you to solve the issue.
Good luck.

 

[btrueblood]

MrC,

Search "hydraulic hand pumps" in McMaster. Sounds like you may already have one.

 

[MrComputer]

Well, I had the units tested again today and had some interesting results. The use of the water jacket during the test definitely was the cause of the bizarre readings I was getting and what left the impression of the leak. I am curious why this was the case but when it was in the water jacket, we took the pressure up to 6ksi and the water level in the water jacket began to rise, suggesting a leak. The pressure was also fluctuating quite a bit. When this occurred, we maintained the pressure and removed it from the water. As soon as the the vessel was removed from the water jacket, the pressure stabilized and no longer bounced around. We put the unit back under water and brought the pressure up to around 9ksi. The gauge began fluctuating quite a bit and then the vessel failed. The gauge was around 9ksi when this burst occurred.

Does this look like a typical failure?

So to investigate this water situation, and since the water in the chamber was released when the previous vessel burst, we tested the second unit without the water jacket. In this instance, there were no pressure fluctuations and we were able to keep it at 9.5ksi for about 10 minutes without pressure loss. I am not familiar with the water jacket testing so I have no idea what the issue was or if the burst pressure was legitimate since the vessel itself should be good up to 11ksi before yield. I can only wager that there was some sort of issue with the test apparatus and/or due to the use of such a small vessel in a large cylinder test chamber.

I also used 2 different thread sealant methods, one being liquid sealant and the other being mil-spec ptfe tape. Both sealed the vessel and there were no visible leaks around this connection when observed. The main purpose was to ensure the brazed end caps would hold and they did (quite incredibly). So its been an interesting situation, thanks for all the help. If I had access to a better testing facility, this likely would have been avoided completely; however, gotta make due with what's available.

 

[prex]

You really mean 0.065" thick and 1" OD ? You are about at 1/3 the required thickness for 9 ksi, so no surprise it failed with that very classical pattern.
You better go back to the 'designers' and beat them on their heads, telling them they should not put you into such a danger...

prex
[URL unfurl="true"]http://www.xcalcs.com[/url] : Online engineering calculations
[URL unfurl="true"]https://www.megamag.it[/url] : Magnetic brakes and launchers for fun rides
[URL unfurl="true"]https://www.levitans.com[/url] : Air bearing pads

 

[LittleInch]

We don't know the strength of the tubing. If smys》70000 psi you're ok, but are at the mercy of any small thinning or damage.

What the pressure fluctuations are I can't work out.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[MrComputer]

Yeah, I checked the certs for the tube and the yield was 70ksi and ultimate was 80ksi. Given the numbers it should have destroyed the tube at 11ksi. That one that burst was actually a little less than the calculated pressure it could hold. I think it burst at 9ksi(if that was the true value given the weird fluctuations) due to some rusting, since these were the same that I tested originally and the water had taken its course over the next couple days before the retest. Or it was because of some uneven thickness in the tube or just some shit tube from a different lot than the certs I received.