Tube Rupture - High pressure gas into water 2

[Justice100]

Hello all,

I have a shell and tube HE with dense phase HC fluid on tube side at 350 barg and hot water on the shell side at 15 barg and 150 DegC with a design pressure of 24 barg. The shell will be protected by 2 bursting discs connected to the main flare header via tail pipes and a sub header for the tube rupture scenario. On tube rupture the dense phase fluid flashed to the lower pressure is almost all gas. I've worked out the steady state relief rate and we have a consultant doing the transient analysis on the scenario.

The one thing I am not sure about is when the tube rupture occurs and the bursting disc bursts then hot water will be pushed out of the exchanger but at this initial point the water in the tail pipe will immediately vaporize to steam since the flare sits at 0 barg. As this flow develops down the tail pipe/header a back pressure will develop eventually so the water won't vaporize. The shell volume is very small compared to the flare header.

I am trying to size the burst disc tail pipe and assess the flare header, which is fine for the steady state case but I am not sure what to consider for this transient case. Any ideas? Has anyone come across this before? The transient consultant was not much help...

 

[georgeverghese]

Agreed, the initial displacement through the RD would be single phase water and 2phase V-L downstream in the tail pipe. Presume you have the tail pipe sloping down into the main flare collection header. Built up backpressure would be minimal in this tail pipe given the HX shell volume is small, I would guess. Only the dynamic simulation engineer can tell you what this backpressure would be. An approximation of this backpressure may be necessary. Are you sure the max operating pressure coincident with this relief is only 0 barg ? What happens when there is coincident continuous flaring when a compressor trips on some mechanical failure or similar ?
Conventional RDs' are a pain and a hazard to replace while the plant is running, so instead of 2x100 % RDs', why not 1x100% RD and a small PSV to take care of minor leaks on the tubes or the tube-tubesheet joints? The small PSV can be set at a lower setpoint than the RD. Typical max normal operating pressure should be no more than 80% of RD setpoint. See if a buckling pin activated safety valve is a better alternative.

 

[Snickster]

I have never run into this situation in designing relief systems. However, seems like the water will flash to about a 20% quality water steam mixture at 50C based on water/steam pressure-enthalpy diagram assuming not much more heat is picked up in the relief system, so basically will still be mostly water. You could do a pressure drop calculation to see what the back pressure is on the rupture disc when the mixture flow reaches the connection to the main flare header (assuming the flare header is large enough as to not contribute much to the pressure drop). This will give you the maximum back pressure at the rupture disc outlet with all two-phase water/steam flow in the outlet and sub-headers up to the main flare header, from which you can check if you get the flow through the rupture disc required based on flashing flow through the rupture disc.

The volumetric flowrate to use in the calculation of the two phase flow pressure drop would be the volumetric flowrate of the single phase water upstream of the rupture disc that must flow to equal the volumetric flowrate into the shell of the high density fluid , at resulting P/T in the shell, so that the pressure increase in the shell does not increase above the design pressure of the shell.

 

[Justice100]

Agreed, the initial displacement through the RD would be single phase water and 2phase V-L downstream in the tail pipe. Presume you have the tail pipe sloping down into the main flare collection header. Built up backpressure would be minimal in this tail pipe given the HX shell volume is small, I would guess. Only the dynamic simulation engineer can tell you what this backpressure would be. An approximation of this backpressure may be necessary. Are you sure the max operating pressure coincident with this relief is only 0 barg ? What happens when there is coincident continuous flaring when a compressor trips on some mechanical failure or similar ?
Conventional RDs' are a pain and a hazard to replace while the plant is running, so instead of 2x100 % RDs', why not 1x100% RD and a small PSV to take care of minor leaks on the tubes or the tube-tubesheet joints? The small PSV can be set at a lower setpoint than the RD. Typical max normal operating pressure should be no more than 80% of RD setpoint. See if a buckling pin activated safety valve is a better alternative.

I need to have a look into tail pipe slope. My initial thought was that it would as per an RV, located such that it slopes down into the flare header but from a transient perspective it needs to be as close to the shell as possible. I'll speak to the guy doing the transient and see if he can accommodate it away from the HE such that we can slope down to the flare. If BDs need to go directly on the exchanger then could potentially justify that as it goes into a cold flare header purged by fuel gas... not sure how dry the fuel gas is.

Good point regards other coincident operating flare cases, I will investigate.

Yes, plan to put a "D" RV on there if accepted by client in addition to the two BDs. Then wait and see if the transient analysis requires one or two BDs.

What advantage would a safety pin device have over a bursting disc in this scenario?

 

[Justice100]

I have never run into this situation in designing relief systems. However, seems like the water will flash to about a 20% quality water steam mixture at 50C based on water/steam pressure-enthalpy diagram assuming not much more heat is picked up in the relief system, so basically will still be mostly water. You could do a pressure drop calculation to see what the back pressure is on the rupture disc when the mixture flow reaches the connection to the main flare header (assuming the flare header is large enough as to not contribute much to the pressure drop). This will give you the maximum back pressure at the rupture disc outlet with all two-phase water/steam flow in the outlet and sub-headers up to the main flare header, from which you can check if you get the flow through the rupture disc required based on flashing flow through the rupture disc.

The volumetric flowrate to use in the calculation of the two phase flow pressure drop would be the volumetric flowrate of the single phase water upstream of the rupture disc that must flow to equal the volumetric flowrate into the shell of the high density fluid , at resulting P/T in the shell, so that the pressure increase in the shell does not increase above the design pressure of the shell.

Thanks for responding.

I have run it on Aspen Flare System Analyser with hot water at 145 DegC and a volume flow equivalent to the gas flow into the shell at shell design pressure. The case won't completely converge, the water flashes at the end of the header/entrance to the KO Drum but the results seem pretty close to what I'd expect, I just need to get it to converge. On that basis, I have no back pressure or rhov2 issues so I think that's all I can do.

 

[georgeverghese]

Buckling pin is much easier and safer to replace. Talk to the supplier to see if a buckling pin will allow you to operate at pressures higher than 80% of set pressure, assuming operators could use the additional headroom on pressure - I dont have this info offhand. There may be some other advantages also. A BD would be my last preference, if you ask me.

Note the PSV setpoint would have to be no more than 80% of BD set pressure for obvious reasons. This is another reason to check out buckling pin option.

 

[goutam_freelance]

Sizing and velocity calculations for safety valve discharges in steam service, including wet steam are given in ASME B31,1 Non Mandatory Appendix II.

I believe a preliminary estimate of back pressure and velocity in the flare headers can be obtained from this.

Referring to clause II.2.2.2, initially, during the transient phase, there is a possibility of shock waves in the header due to which the design pressure of the header should be more than 2 times the steady state pressure.

 

[KevinNZ]

Two phase safety valve and disk sizing is covered by ISO4126-10, but this come back using two phase flow calculations for the pressure upstream and downstream of the disk/valve