WHB Tube Rupture, possible 2-phase flow at PSV?

[Casimo5]

I have a case where the high pressure side is BFW, on the shell, and flue gas in the tubes (low pressure side). If there is a tube rupture, high pressure BFW will go to the low pressure side and flash across the rupture. Using Hysys simulation, I calculated the percent flashing that occurs. The volumetric liquid flow is about 1% of the total volumetric flow.

I understand that each system is unique and must be looked at as such, however I am looking for any references that state that the liquid flow can be discounted.

I am also performing a sanity check with the piping isometrics to estimate the total volume of the system on the low pressure side to make sure I will not fill it completely within minutes.

My initial reaction was to simply assume all the two phase flow was reaching the relief valve, however this resulted in a very large amount of area required (350 cm2). That is why I would like to check to see if there is previous experience with this scenario that would give me a precedence to assume only vapor will reach the relief valve. Vapor sizing only results in a requirement of only about 120 cm2.

Thank you in advance for your help.

 

[CJKruger]

Casimo5, without knowing the details it is difficult to know for sure, but here are some general comments:

- If the relief valve is far away from the tube rupture or with a vessel inbetween (like in a tower), we will often assume the relief valve must relief a volume of gas (at its inlet) equal to the volume of fluid displaced at the tube.

- I think your numbers look suspect. If liquid is so small a part of the fluid, the contribution to the area should not be that great. If you post all your data, some of us may be able to check it. Also post the homogeneous densities at inlet and outlet (of relief valve) if you want us to do a rough check on the omega method.

Cilliers

http://www.korf.co.uk

 

[Casimo5]

Thanks CJKruger, that's they type of confirmation that I'm looking for. Unfortunately there is no equipment between the exchanger and the PSV, but there is a significant amount of 20" pipe. The scenario is actually quite a bit more complicated, but not relevant here.

I'm comfortable with my numbers, for both single and two-phase flow from the rupture and sizing of the PSV. The only problem I'm currently having is with the assumption and whether or not it is a valid one.

But just for an example... The fluid is bubble point water. Let's just say you have 100 tons/h of BFW (at bubblepoint) at 264 deg C (about 50 bara) on HP side and you flash over to the LP side at 5 bara, your volumetric fraction of liquid left over will be very small, although the mass fraction will be large. That's why the numbers look extreme, not suspect.

If anyone knows of any references in API 520 or 521 (or any other publication for that matter) that may refer to this assumption, please let me know what section I may be able to find it. I'm currently reading through them, but am not having any luck.

Thanks again!

 

[CMA010]

CJKruger has a good point, your numbers seem to be off. It seems very odd that a 1% increase of the volumetric relief flow rate results in 200% more required area.

 

[Casimo5]

CMA010,

I'm sorry, I must not be explaining the scenario clearly enough.

The 1% volumetric flow is referring to the liquid phase of the two phase flow. We have 14,000 m3/h of flow, of that only 140 m3/h is liquid and 13,860 m3/h is vapor. Lots of vapor, very little liquid...VOLUME.

Now look at the mass flow. At the pressures I'm working with, the vapor is at 3 kg/m3, so you have a mass flow (vapor) of 42 ton/h. The liquid mass flow is 140 tons/h. (let's just say water density of 1000 kg/m3 for simplicity). So now we have lots of liquid, and much less vapor...MASS. (23% vapor, 77% liquid)

I'm using the Omega method from API 520 Annex C (Sizing for two phase liquid/vapor relief) to calculate the maximum mass flux through the relief valve.

Then, if I assume the vapor volume is so large that I can discount the liquid volume and use the total vapor volume, then (here is the mistake) back calculate the mass of vapor for that volume, then use that with my maximum mass flux to get an orifice size, I get one number. (This assumption doesn't make sense, I know).

However, if I take my total original two phase relief load mass, regardless of the volumetric fractions of vapor and liquid, and use that with the same maximum mass flux through the orifice, I get a number 4 times bigger than the one above!

That is how a 1% volumetric increase produces a 400% increase in required area. It's because that 1% increase in volume resulted in a 400% in mass.

Sorry, my initial numbers were just thrown out there. I went back and got some that were closer to my actuals. Hope this clears things up.

 

[Casimo5]

I realize now that in my original post I am comparing a vapor relief to a two phase relief, where in my response to CMA010 I am comparing one two phase relief to another.

When you compare the two-phase relief cases you get 4 times the area for 4 times the mass flow.

When you compare a single phase vapor relief to the two phase relief, you get approximately 2 times the area for 4 times the mass flow.

This is because for a fixed mass... the area required is bigger if it is all vapor. As your % liquid increases and your flow becomes two phase (still for a fixed mass), the volume significantly decreases and therefore the orifice size decreased.