Hydrostatic Test Failure
Hydrostatic Test Failure
(OP)
I came across a YouTube Video yesterday that has completely changed my thinking on hydrostatic testing vs. pneumatic testing. First, after a lot of digging, I found some of the details on the line:

You can see from the picture that the failure started next to the weld, not in the weld.
I watched the video and asked myself "where did the energy come from to tear out a flap of steel and bend it up 90° against the curve of the pipe?" That was a huge force.
That is when I realized that the bulk modulus (i.e., the amount of pressure that would decrease the volume by 1%) of water is 319,000 psia, so to reach 1870 psig you would have to add 0.006% of the system volume. I looked at the enthalpy of the water at rest and the enthalpy of the water at test pressure and found a 5.3 BTU/lbm change. For a test that was more than a few joints long, W=m*ΔH+Δm*h turns out to be a really really big number. Archimedes Principle says that a force applied (or removed) from a closed volume will be transmitted everywhere within the closed volume, so this huge energy acts like a coiled spring that releases its entire energy at the failure point.
I've talked on this site many times about the errors in the NASA Glenn Research Methodology used by many to calculate the energy of a pneumatic test. Primarily my point has been that there is no way for distant mass to "know" about a failure and the mass that participates in a in a failure is limited to a few joints worth of gas.
The change in my position is that I've always been a bit defensive about the use of pneumatic tests, but I was wrong to be defensive. The energy available to strike nearby workers is far higher in a pipeline hydrostatic test than a pipeline pneumatic test. If we assume that the failure in the video was 5 miles of 30-inch pipe, then I calculate that the energy release in a pneumatic test would be 0.2% of the energy that could be released in a hydrostatic test. Yep, pneumatic tests are irresponsibly dangerous. NOT.
- ANSI 600 line
- Test was a fully degassed hydrostatic test (water for the test had been loaded 2 days before the test and allowed to degas)
- Tested to 130% of MAWP (1870 psig test pressure)
- All fittings and pipe were U.S. made
- All welds around the failure passed x-ray
- The line was owned by one of the largest pipeline companies in the U.S.
- The construction contractor had done hundreds of hydrostatic tests for that particular pipeline company and others
- The test was done following a procedure that had been reviewed both by the engineering contractor and the pipeline company

You can see from the picture that the failure started next to the weld, not in the weld.
I watched the video and asked myself "where did the energy come from to tear out a flap of steel and bend it up 90° against the curve of the pipe?" That was a huge force.
That is when I realized that the bulk modulus (i.e., the amount of pressure that would decrease the volume by 1%) of water is 319,000 psia, so to reach 1870 psig you would have to add 0.006% of the system volume. I looked at the enthalpy of the water at rest and the enthalpy of the water at test pressure and found a 5.3 BTU/lbm change. For a test that was more than a few joints long, W=m*ΔH+Δm*h turns out to be a really really big number. Archimedes Principle says that a force applied (or removed) from a closed volume will be transmitted everywhere within the closed volume, so this huge energy acts like a coiled spring that releases its entire energy at the failure point.
I've talked on this site many times about the errors in the NASA Glenn Research Methodology used by many to calculate the energy of a pneumatic test. Primarily my point has been that there is no way for distant mass to "know" about a failure and the mass that participates in a in a failure is limited to a few joints worth of gas.
The change in my position is that I've always been a bit defensive about the use of pneumatic tests, but I was wrong to be defensive. The energy available to strike nearby workers is far higher in a pipeline hydrostatic test than a pipeline pneumatic test. If we assume that the failure in the video was 5 miles of 30-inch pipe, then I calculate that the energy release in a pneumatic test would be 0.2% of the energy that could be released in a hydrostatic test. Yep, pneumatic tests are irresponsibly dangerous. NOT.
David Simpson, PE
MuleShoe Engineering
In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist





RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
What would be the velocity differences between liquid and gas at rupture at the immediate rupture scene? Might there be a kinetic component to the evolving rupture that would promote 'peeling' results.
Interesting point David.
Keith Cress
kcress - http://www.flaminsystems.com
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
I personally saw a buried 30" blow under hydrotest a long time ago. Under an asphalt parking lot. It was a mess. There was dirt and asphalt chunks hitting cars 100ft away. If it had been a pneumatic test, I'm sure nothing would have happened. NOT!
There's a debate about pneumatic vs hydro testing safety? I thought the debate was about if pneumatic testing could ever be made as safe as hydro (relatively speaking), given that at times pneumatic testing might be more convenient and is permitted under some codes, if no practical alternatives are available, but nobody has ever said to my knowledge that it is actually safer to do pneumatic testing ... until now??? Really? I'd like to stop the conjecture. Where's the proof? Has NASA admitted they are wrong? One of our members has new calculations to show us? There is new evidence somewhere? Or is it YouTube, Urban Mythbusters ...? Where?
B31.4 prohibits pneumatic tests.
B31.8 allows, but with condition of the impracticality of hydrotesting, which I take to be that there is no water available, but which others feel that it means water, or hydrotesting itself, is too expensive to be practical (ie. That's what a good lawyer can do)
RE: Hydrostatic Test Failure
Key thing for me is why on earth were they testing what is clearly supposed to be a buried system without burial?? The extra stress caused by hydro which is not being restrained or supported by the soil is enough to cause failure, not withstanding what appears to be alack of fusion on the weld which is my guess as to the root cause.
Hydro vs pneumatic has been done to death here and by zdas04s engineering guild paper.
I must admit I have changed my mind from "no way" to "has merits in certain locations and conditions". Any pressurised system of a decent volume has the potential for damage at a local level whether filled with air or water.
I didn't follow the equation in the OP I'm afraid, but even looking at the video you can see there is a very short term effect which would be greater with pneumatic, so safer?? Don't think so - could be made acceptably safe - Yes.
B 31.4 allows pneumatic testing for system operating below 20% of SMYS as a leak test so maybe doesn't qualify directly as a strength test.
Remember - More details = better answers
Also: If you get a response it's polite to respond to it.
RE: Hydrostatic Test Failure
Thrust a practical man, LI.
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
I reran the math with 8-inch and got the same results. BigInch, you didn't address my point at all. You sound like the global warming zealots when you say Where did the energy to bend that pipe come from? If it wasn't from decompressing the water then where did it come from? I am really disappointed in this discussion.
Btrueblood,
No, I'm saying that in a liquid system the energy can be transmitted without significant mass flow much like a spring uncoiling. In a gas the energy transfer requires mass flow which is limited to sonic velocity.
CompositePro,
Popping an air-filled balloon is not more violent than popping a water balloon. The mass flow simply expresses differently, instead of the majority of the energy expressing as sound it expresses as fluid velocity. You can see the difference and the similarities in high speed video (which is available on YouTube if you want to look for it, it is pretty interesting)
LittleInch,
What does it matter? Had it been underground when it blew the forces would have been applied differently, but what does that have to do with anything?
If this subject were really "done to death" then I don't think I would be being attacked like I am in this thread. I think all of the locked in "NASA said it, it must be right" mindset is blocking rational discussion.
rconnor,
What? In my OP I said And then you say "we don't even know if the air was removed from this line".
All,
I have been working with some researchers in another country (the details are under an NDA) that did a series of experiments that found the force exerted by a failed pneumatic test was independent of pipe length after a (very short) minimum, less than 2 joints of pipe) for a given size pipe. The data showed vividly that only the gas that could get there at 1.0 Mach in about 0.02 seconds participated in the explosion. I wish I could share the details, but I can't so please reject this out of hand because it doesn't fit with your preconceived ideas.
I give the hell up. If anyone wants to think for themselves, please ponder the idea that hydrostatic testes are dangerous. Pneumatic tests are dangerous. The mechanism of the energy transfer allows more energy to participate in a large-volume hydrostatic test than in the same volume pneumatic test.
David Simpson, PE
MuleShoe Engineering
In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist
RE: Hydrostatic Test Failure
I know my pipe sizes, thank you.
Remember - More details = better answers
Also: If you get a response it's polite to respond to it.
RE: Hydrostatic Test Failure
Continued expansion of a gas behind travelling particles continuously propels the particles until the expanding gas has reached ambient pressure.
LI thanks for reminding us about the B31.4 test option for <20% SMYS. I don't usually mention it, since we know "pipelines" aren't practical to build to <20% SMYS and the safety factor during testing, hydro or pneumatic, would be so enormous even I wouldn't have any objection whatsoever to pneumatic testing such a pipe.
OK 24". That's the biggest controversy I see here, including climate change. I get National Geographic cable TV.
RE: Hydrostatic Test Failure
I really didn't get this bit - can you explain further?
I looked at the enthalpy of the water at rest and the enthalpy of the water at test pressure and found a 5.3 BTU/lbm change[heat energy?]. For a test that was more than a few joints long, W=m*ΔH+Δm*h [units please and an explanation] turns out to be a really really big number. Archimedes Principle says that a force applied (or removed) from a closed volume will be transmitted everywhere within the closed volume, so this huge energy acts like a coiled spring that releases its entire energy at the failure point.
Also surely the point made before about pneumatic testing is valid here, i.e. the energy involved in an incident / burst is really quite local to the incident, so how are you now saying "....releases its entire energy at the failure point"?? The mechanism ofr energy / pressure pulse must run at speed of sound in liquid as well surely. - similar principle to pressure surge from sudden valve closure.
I've never come across this mysterious NASA paper - maybe you attached it or included it in previous discussions - I can't recall - so I'm not blinded by something I've never seen.
I don't think anyone has said any high pressure test isn't potentially dangerous, just one type is generally less risky than another type.
Remember - More details = better answers
Also: If you get a response it's polite to respond to it.
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
Specific enthalpy of water at atmospheric pressure and 60F is 28.137 BTU/lbm. Specific entropy of water at 1887 psia and 60F is 33.409 BTU/lbm. If I must add 300 lbm of water to get 23 million lbm of water up to test pressure then:
Wstored = 23,000,300 lbm * 5.272 BTU/lbm = 121 million BTU
Then I need to account for the mass that has to leave to get back to atmospheric pressure
Wejected=300 lbm * 28.137 BTU/lbm = 0.844 million BTU
An incompressible fluid in a closed container acts as a rigid body (like a spring). That means that if the rigid body is allowed to move, the entire 121.8 MMBTU of energy is applied at the point of movement as fast as the point of movement can accept it (very very fast). Once that energy is expended, the water will leak out if it can or it will simply stand in the pipe if the failure was physically higher than connected water volumes. I think of this like the push-rod and rocker arm on an internal combustion engine--if the rod doesn't bend then all of the energy of the spring is applied to moving the valve with nothing lost in shifting mass in the intervening linkages.
A compressed gas does not act as a rigid body and transferring its stored energy requires mass transfer which is limited to the speed of sound. Once the explosive decompression event is completed (between 20 and 50 mS according to experiments done at the University of Nebraska and repeated at other facilities that I can't talk about), the rest of the stored energy is dissipated as a "blow down". We've all blown gas pipelines down and it is noisy and blows stuff around, but it isn't an "explosion".
The new conclusion that I reached when I watched the linked video is that the rate of energy transfer makes hydrostatic testing MORE dangerous than pneumatic testing and we have once again demonized the wrong thing. Not that I think anything should be demonized in this discussion (except maybe NASA and regulators/standards-writers who set regulations and standards based on an analysis that happened to have NASA's logo on it without looking for the gross flaws in the analysis). Pneumatic testing should be a tool in our tool box and very possibly should be the preferred tool for pipelines (not so much for vessels since in a vessel all of the energy is close enough to a failure to make pneumatic testing riskier than hydrostatic testing). Instead we have people getting stars for pointing out that B31.4 doesn't allow pneumatic testing or saying that the appropriate closest point of approach to a pneumatic test is the next county. That kind of fear and superstition is harmful to a rational evaluation.
The NASA Glenn Research Methodology paper seems to have been totally eradicated from the Internet, something that I didn't think was possible but whoever wanted it gone seems to be very good. At least 6 people that I know of have searched for it for years without success. I never saved a copy when it was available, mostly because it irritated me so much. Laws are based on it, and the links in the regulations get a "404 error--page not found" message.
David Simpson, PE
MuleShoe Engineering
In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist
RE: Hydrostatic Test Failure
However the immediate flaw to me is "An incompressible fluid in a closed container acts as a rigid body (like a spring)". True, however,
Water is not 100% incompressible.
Neither is a pipeline a classic "rigid body"
Water compresses under pressure ( Bulk modulus)
Pipe expands under pressure (Hoop stress / strain)
I am actually on your side here, but I don't think you can make the leap you're talking about based on the hypotheses above. Anyway I'll come back to this later.
LI
Remember - More details = better answers
Also: If you get a response it's polite to respond to it.
RE: Hydrostatic Test Failure
But I think your argument holds up less well when testing more mundane and compact things, where the longest distance to a wall/pressure boundary is on the order of meters or fractions of a meter, not kilometers. Dunno, will do some thinking and some quicky math when I get some time.
That said, we routinely pressure leak test some of our products with air, in suitably restrained fixtures, under water, and I have worked on many devices (e.g. rocket engines) where introducing water or other fluids for pressure testing could cause complications more difficult to deal with than explosive decompression. I guess when you are working on stuff that can go kablooie in normal operation, you have more bunkers and blast walls at your disposal.
RE: Hydrostatic Test Failure
David Simpson, PE
MuleShoe Engineering
In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist
RE: Hydrostatic Test Failure
The vessel expansion/contraction is the same regardless of the test fluid, right? It is part of the energy release, though, and that adds yet another transient with its own characteristic time/length/speed contribution. I worked on a hypersonic cannon project as a grad student, and the math for that pipeline got hairy fast. https://www.aa.washington.edu/research/ramaccel
But do you get my point about the vessel shape and dimension (i.e. 20km pipeline vs. a 35 gallon barrel) playing into the discussion?
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
https://www.youtube.com/watch?v=GIqCE3yLJ4U
RE: Hydrostatic Test Failure
I've done those vessel/pipe expansion calcs before and it didn't matter if the pressure came from a gas or a liquid. The numbers were really small.
I've always agreed that any test where the farthest fluid was less than about 20 m from the failure would have more energy applied to the failure in a pneumatic test than in a hydrostatic test. Nothing about this change in my thinking changes that position. For small volumes the amount of liquid you have to add to raise pressure in a hydrostatic test is tiny and the compressive energy is also pretty small, so when those tests fail you get a "pop" and water on the floor. The physics of this kind of test was very well explained in 1988 in:
“A Review of Energy Release Processes from the Failure of Pneumatic Pressure Vessels”, M. Coleman, M. Cain, R. Danna, C. Harley, D. Sharp, General Physics Corporation Cape Canaveral Air Force Station, Florida.
The thinking fell off the rails when NASA has a summer intern "scale up" that analysis to pipeline scale in the 1990s and he failed to take into account that energy transfer in gas required mass transfer (which is limited by the speed of sound), while in hydraulic systems the energy transfer happens with minimal mass transfer.
Weldstan,
Exactly right, I have had a lot of engineers imply that hydrostatic tests are zero risk and pneumatic tests are literally "playing with dynamite". The change in my thinking is that hydrostatic tests are playing with a lot more dynamite than pneumatic test are on pipeline tests.
David Simpson, PE
MuleShoe Engineering
In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist
RE: Hydrostatic Test Failure
"The hazard posed by a pressure test is a combination of the energy stored in the pressurized fluid in the equipment and the degree of ignorance about the suitability of the equipment to contain pressure..."
When I later took time to look at the OP linked "video", I have a tendency to wonder along with LI on this one. At one thousand eight hundred and seventy psi on this system (and who knows what else e.g. thermal etc?), I believe there would have had to be a substantial transverse resultant thrust on the outside of the 90 ell several feet from the tee. With no backfill around the piping (i.e. nothing but air around same and behind the ell, that of course has little resistance), there would appear to be relatively very little to restrict at least Bourdon movements of that piping in at least the ell area under pressure. Even slight deformation/movement of the 90 ell and adjacent branch piping would of course beam load the connection at the tee branch, introducing some additive localized stress within same (i.e. in "pully-bone" fashion, as they say down South, or "wish-bone" fashion to Northeners). While I'm sure different forensic folks would probably put their finger on what they wish after this event and I would not venture cause without seeing a stress analysis in this condition, I wouldn't be surprised if quite high local stresses were imposed on this one (and one wonders if this layout of testing and the as-built tee strecture was analyzed for such condition?) While I guess the perception may be that modern pipers know a whole lot more and have a whole lot more control over all things that might or can go on in the cradle to grave lifecycle of a pipeline job, maybe one could add just a few words to the HSE statement,
The hazard posed by a pressure test is a combination of the energy stored in the pressurized fluid in the equipment and the degree of ignorance about the suitability of the equipment to contain pressure in the tested condition.
As to relevance to air vs hydrotesting, I don't know what exactly would have happened here had this been a gas test, but I kind of suspect the aftermath may not have looked any prettier (nor the folks standing around's drawers being any drier when it blew!) All have a good weekend.
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
http://www.whatispiping.com/pressure-tests-of-pipi...
RE: Hydrostatic Test Failure
Speaking of "complete ignorance" please take a glance at a mirror.
The link you provided started with the assumption that hydrostatic tests are inherently safe and that pneumatic tests are inherently hazardous and built from those assumptions. The video that I started this thread with shows that there is nothing inherently safe about hydrostatic tests. The basic assumption is that water is incompressible. That assumption is inherently wrong. If you add 600 gallons of water to a full system, then you have obviously compressed it.
The difference between the results of the failure is "energy distribution" and "energy transport". A gas can only transfer energy at the speed of sound. A liquid will transport its energy in a single event (obvious when you track pressure miles away from a failure and see that pressure drops to zero in micro-seconds in a hydrostatic test while it may take a couple of hours for the pressure at the same place to drop to zero. That energy had to go somewhere. In a pneumatic test you are limited to about 50 ft of pipe that can get to a failure during the "explosive decompression" period. In a hydrostatic test the entire volume participates in the explosive decompression.
David Simpson, PE
MuleShoe Engineering
In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
I understand this is pipeline related but I'm going to make some very broad assumptions and generalizations for the sake of getting some numbers, not accurate or directly related to this system;
Code - ASME B31.3
Material - API 5L X52
Pipe Size - 24"
Long Weld Factor - 1.00 Seamless - Why not, for the sake of easy math?
This would put minimum thickness around 3/4". API 5L X52 follows curve A, which for 3/4" gives a MDMT of 53 °F.
Material and welding may have been based on normal operation of 70+ °F process but hydrotesting in winter when it's 40 °F may have not been properly accounted for. It was also mentioned that they left the line to vent entrained air for two days, could have provided time to cool down to ambient temperature. Just curious on the possibility that temperature wasn't properly considered when the hydrotesting was performed and stress risers at the weld caused overstress when the material went brittle.
Thanks,
Ehzin
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
https://www.youtube.com/watch?v=AelS89Az3YA
RE: Hydrostatic Test Failure
While there is certainly not enough firm info defined at least in the OP of this thread to know whether or not any of this is applicable to the present case, see that the results of the ORNL research revealed "maximum stress ratios" (it appears basically stress multipliers they defined as the ratio of the actual determined maximum principal stress divided by the nominal hoop stress of a similar dimensioned cylinder) of from 9.0 to 13.3, due to internal pressure alone on the model tees.
It would thus be at least interesting to look at the "design" of this actual connection for withstanding the internal pressure and any other loads imposed, given enough firm information.
I also happen to have a hard-copy only of another paper entitled, "Approximate Analysis of Intersecting Equal Diameter Cylindrical Shells Under Internal Pressure" written by professors Kalnins and Updike of Lehigh I think not long after the ORNL work (obtained when same could be had for $1.50 from ASME!) that also elaborated on "radial displacement"/shape changes of intersecting cylinders. It explained that due to the fact the "hoops" of one cylinder, which intersect those of another. are in effect "broken" (if there is to be unimpeded flow passage), the load in addition to the stress concentrations also tends to push the top meridian of the branch outward, and thereby ovalling same outward in the process. In looking at the one picture provided, it thus appears that once the fracture cut loose that flap of metal went basically where it was wont to go, as the specific meridian the Lehigh profs referred to would appear to basically bisect that pointy flap of metal looking at one in the pic. In conclusion, one has to wonder about the localized stresses that actually came to bear on this connection, and accept that the flap of metal was bent basically in the direction it was loaded. I'm not sure what else can be concluded now, but the new video added here may well be helpful.
All have a good weekend.
RE: Hydrostatic Test Failure
The video compared a 20 gallon vessel hydrostatic test to a pipeline test of undisclosed (but very large) volume. Yep, a thousand pounds of TNT makes a bigger hole than a firecracker would, even though both have about the same blast-propagation speed. So what?
The audio in the link was garbled so I couldn't tell what they were saying, but the dismembered body parts make it pretty clear that they weren't being favorable towards pneumatic testing.
David Simpson, PE
MuleShoe Engineering
In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
RE: Hydrostatic Test Failure
David Simpson, PE
MuleShoe Engineering
In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist