You don't say what the gas is, but from the rest of your description I'll assume it is flamable.

What you've described is similar to what happens in a diesel engine.  The high-pressure gas rushing into a closed and (apparently) depressurized section of pipe can be moving at sonic velocity.  A sonic stream has a very high density and will not mix with a static stream.  In effect it will act as a piston to compress the gas in front of it.  The heat of compression can raise the temperature of the compressed gas to above the autoignition temperature of a flamable gas.  If there is an explosive mixture in the hose then it will ignite--explosively.  My guess is that when the hose bursts and the metal stays cold you have had an explosion with inadequate oxygen to continue the fire immediately after the pressure is released.

After a man in my old company died from a similar activity I started researching this type of accident and found that in the U.S. the Oil & Gas industry has averaged about 1 fatality per month between 1920 and 2000 from similar accidents.  What you've described is incredibly dangerous.

David Simpson, PE
MuleShoe Engineering
www.muleshoe-eng.com
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

The harder I work, the luckier I seem

zdas04 (Mechanical HAS GIVEN A GOOD EXPLANATION.
You can also look at simpler problems such as putting air into your tire.  The tire will heat up.

Also, you problem might be modelled as a rupture disc bursting with flow into a dead ended pipe. With no mixing of the upstream and downstream gases and adiabatic conditions, the dead ended gas will initially be compressed under isentropic conditions.  If the upstream and downstream pressure differences are high enough, supersonic flow, followed by shock waves will occur.

For clarification: at a lower gas pressure, the adiabatic heating always takes place at the end of the hose where there is a metal jacket section (several 100 mm long) in direct contact with the compressed gas. This is evident because the jacket gets hot (and the hose stays cold).

However, at increased pressure (but still well within the safe maximum pressure rating of the hose) the adiabatic compression seems to take place not at the end in the jacketed section, but just before the start of the jacket and sometimes the hose bursts there and always gets hot even when it does not burst. It's almost as if the shock wave does not want to enter the metal jacket, which has a slightly smaller diameter.

I believe this effect has nothing to do with the gas being combustible or not.

What is going on?

Some more info: the hose has an internal lining of PTFE (Teflon) and the gas is oxygen (!). I've just seen a high speed video and there's signs of combustion during the explosion.

bigbuzz (Chemical)The hose even though feeling cold may not be heating up as fast as the metal pipe, because of heat transfer properties.  Obviously, the process is not adiabatic, since the metal is rapidly heating up.  However, the heatup effect may be small compared to the general compression process.
What are the upstream and downstream gasses, their inital temperatures, pressures and how rapid (time)is the opening and what size (orifice) is the rapid opening?

Hi sailoday28, valve opening time 15 ms, high pressure section is 360 bar, 60 degrees centigrade, low pressure region (in hose) is 1 bar, temperature 20 degrees centigrade. Hose internal diameter 6.35 mm, metal section internal diameter is 5.00 mm. Interestingly for a valve opeining time of 22 ms all goes well (i.e. hose stays cold, metal section gets hot).

Although PTFE has a high limiting oxygen index, LOI, of 95%, under pure oxygen the polymer may burn after ignition, depending on the compression-developed temperatures.

Sudden or quick opening valve time relating to hammer effects is related to length upstream (or downstream)and sound speed.
 For example consider water, with sound speed relativly high and a long length of pipe. For an acoustic wave to travel both ends upstream of a valve that is opened, the time is  t= 2L/a.  Where a is sound speed, and L the length.  Valves fully opening, in less that t are considered fast opening.
For highly compressible gases the formula for the time is less than 2L/a.
 What seems to be a slow opening valve in a long length of pipe, may really be fast opening.
I don't have the sound speed for the gas or the piping lengths. It would be interesting to see how your opening times compare with the above formula.

The approximation above neglects friction, which is generally small compared to the initial transient effects.
Also the valve throat or orifice equivalent are not known.

The velocity of sound will change wrt the density of the gas medium also.

I think Davids observation, explanation and warning is very important.

Normally you would assume that the cold gas rushing in (beeing cooled by the JT effect) would mix with the gas being "compressed" and thus not cause a major increase in pressure. But this phenomena occurs and causus deaths!

Best regards Morten

MortenA (Petroleum)"Normally you would assume that the cold gas rushing in (beeing cooled by the JT effect)"
JT can be applied to STEADY FLOW problems, where upstream and downstream KE and potential energies are negligible.

Since the problem is a transient, even if velocities are neglected (KE effect),and the system well insulated, JT, by itself has no meaning.
Consider
Filling air temp, same as downstream temp

For example, the filling of an evacuated volume that was originally filled with air or other gas.
First evacuate the volume.
Let the inteernal air come to temperature equilibrium with the surrounding environment of air.
Let it now be well insulated. (Some how on the inside)
Put a small opening in the vessel and let outside air flow in.
The vessel air will heat up.  Solution is obtained from conservation of energy. (No relation to JT)

Change of internal energy of the air in the volume=
stagnation enthalpy (or static enthalpy of outside air, since there is no KE outside) times mass added to vessel.

For simplication, let initial mass in tank be negligible compared to mass added.  
Then   m*u =ho*m    where m is final mass in vessel , u internal energy of air in vessel, ho stag enthalpy of outside air.
or    u=ho
For perfect gas, constant spec heat  CvT=CpTo  T=kTo
where T is container air temp , To outside air temp and k ratio of specific heats  (Cp/Cv).   Note temp is absolute.

My above example, based on a lumped mode analysis does not relate to hammer effects.
 For hammer effects, such as the original thread problem indicates, I would suggest a method of characteristics solution. This would require pipe lengths, valve opening time and valve characteristics, etc. Combustion or detonation effects could be included, if need be.
 

Well i think that JT still exists in transient modes (valve do believe so) but you may be rigth with regards to using caution when aplying these calculations in special transient phenomena - just my point BTW. So we agree except that JT does not apply to trancient calculations. As i understand it JT exists - but you have a situation where the two gasses does not mix and you thus get a more complex problem that cannot be solved using the "normal" 1D methodes ofen used in flow simulation (both transient and SS).

There are many types of problem scenarios and in the problem described by David it would be dangerous to assume that a 1D solution is adequete while in other it would be OK.

If my point is not clear i will try one more twist: Not all transient problems are related to filling empty tanks or closed pipe sections.

Best regards

Morten