PSV sizing for contactor tower in TEG dehydration 1

[hskk]

Hi,I have some doubt regarding the sizing of PSV.

Here is the case:

the contactor will be operating at 100F, 855 psig. The PSV sizing for the contactor shall be based on fire case. The contactor shall be packed column.

The latent heat of glycol solution at 855 psig (corresponding boiling temp is about 800F)is very small (0.0126 Btu/lb).

According to API 520/521, the vapour to be relieved due to vaporization and the expansion of gas needs to be estimated when sizing the PSV. However, the vapour to be relieved is going to be a large number because of the small value of latent heat.

My doubts are as follow:
1) the latent heat of glycol at 855 psig (0.0126 Btu/lb)is correct? (I read the value from HYSYS) if not, what should be the value? where can i check?

2) if the latent heat of glycol is so small (however, i still think that it is possible for the temp of contactor to increase to 800 F in a very short period of time when exposed to fire), can I consider only the expansion of gas as vapour to be relieved(heat input through unwetted surface)?

 

[rbcoulter]

Your T reduced is very small so you are close to the critical point where the heat of vaporization would be expected to be very small. At the critical point it is zero.

 

[chemjock]

It would be conservative to size for some mimimum latent heat value. API 521 mentions 50 btu/lb. Only use vapor expansion if well above the critical point.

 

[MortenA]

the 50 btu/lb isnt that for simgle compoennt paraffin hyrocarbon liquids? Maybe a bit far from TEG?

How about gas breakthrough from the contactpr? Is that a case or is the piping sufficiently small so that this is not a problem?

Best regards

Morten

 

[phorth]

hskk,

I would not expect the standard HYSYS equation of state to be capable of calculating accurately the latent heat of TEG . You need to get a data book on TEG, where enthalpy and physical properties are provided. I believe such books are published by Union Carbide and by Dow.
The latent heat should be taken at the partial pressure of TEG in the column (very low) not the total pressure, so I would think that the latent heat at atmospheric pressure would be fine.
Also, the TEG in the contactor bottom will contain water. When the column is heated by a fire it is water which will be driven into the vapour much more than TEG. However, I believe that the latent heat of TEG will be lower than that of water, so using the TEG latent heat will give more vapour and a conservative sizing for the relief valve.

 

[Zoobie]

While not totally on topic, I thought that I would point out (in response to phorth's post) is that the latest version of HYSYS has a TEG property package. I believe it resolves a lot of the issues that HYSYS had in the past with glycol properties (though I cannot specifically comment on the latent heat calculation).

 

[tenac]

I think that, according to API 521, physical properties for PSV sizing should be evaluated at relief pressure, not at PSV set pressure.

So, I would search for latent heat (and relief temperature) @ 1.21 x 855 psig = 1034 psig (21% overpressure for fire case), which may be well above critical point. If that's the case, I would probably size PSV using the equations corresponding to unwetted heat absorption. Is the complete vapourization of TEG in bottoms enough to pressurize contactor from operating to PSV set pressure?

I don't agree with phorth - in case of wetted fire calculations, I would consider the properties of the liquid phase initially contained in the vessel evaluated at relief conditions.

 

[EGT01]

For a mixture that could change composition over the course of a relief event, the proper way to select the controlling basis is to actually evaluate what happens and check the sizing at various points during the event. Have a look at this thread...
thread124-65447

In particular, look at the case study at the ioMosaic website....

http://www.iomosaic.com/iomosaic/pdf/bench1.pdf

If you are only talking about a two component mixture (TEG and water), this should not be to difficult to evaluate.

 

[phorth]

I agree with Tenac - I was wrong to say that the TEG will boil at the partial pressure of TEG in the column. Over here in Britain, every schoolboy knows that you can't make a decent cup of tea on Mount Everest, because the water boils at a lower temperature. That's why not many of us go there. This demonstrates that the boiling point is determined by the total pressure, not the partial pressure as I posted earlier. The vapour in contact with the liquid is 100% boiling vapour with a partial pressure equal to the total pressure. It will be the same in the TEG column.

 

[hskk]

Hi guy, thanks for your valuable comments.

here I got another doubt on the relieving temp for fire case relief.

With reference to API 520/521,
relieving temp = operating temp/operating pressure x relieving pressure

however, my client told me that PV = nRT is valid only for adiabatic and not for the case of vessel exposed to fire. Is it true??

In my opinion,
when there is a fire, the vessel will get isolated (there are shut down valves between the vessel) and hence the V = constant; n = constant. the heat will transfer to the gas in the vessel causing the gas to expand (temp increase causing pressure in the vessel to increase). Hence i think the equation proposed in API is valid (regardless of whether it is adiabatic).

 

[MortenA]

If the TEG contains water then in the fire case the water will boil off BEFORE the TEG! This means that the relieving temperature will be equal to the boiling point of water at the relieving pressure if water is present.

The ideal gas law is only valid near atmospheric pressure and fairly low temeparture - but the methode can be used. However since the gas "developed" comes mainly from the boiling liquid its not that important. Of course there will be some relief that comes from expanding gasses - but the important step is to find the boiling point of (TEG - or water if there is water in the TEG) at the relieving pressure. It should be possible to find these values in databooks. Rembeber that API has a special section for steam relieving

Best regards

Morten

 

[Montemayor]

hskk:

The properties of Triethylene Glycol (TEG) are as follows:
Boiling Point @ 760 mmHg = 550 ^oF
Initial Decomposition Temperature = 404 ^oF
Latent Heat of Vaporization (@ 760 mmHg) = 174 Btu/lb

Additionally, we know from experience that TEG is regenerated at 400 oF to a level of 99.1 – 99.9% purity before recirculating it to the TEG absorber, where it picks up water from the natural gas being dried (I’m assuming you are drying natural gas, since you haven’t said so). The water picked up in the absorber only dilutes the TEG somewhat. The solution in the Absorber is approximately 97-98% TEG, overall. For practical purposes, you are dealing with a TEG liquid system in the absorber.

Although you don’t state your direct, specific application and we don’t have any P&ID information – among other things - I believe you are using the wrong pressure relief philosophy when you apply API 520 without first reviewing the essential, basic criteria in RP API 521. Specifically you should be focusing on:

a) 3.15 External Fire; and
b) 3.19 Vapor Depressuring

Not until your second post do you mention that you are blocking in the subject absorber during the fire – as is recommended in API 521. This is an important action that cannot go undiscussed during a relief scenario agenda. You do not have a steady state condition as might be inferred in your first post. As specifically defined in API 521, you have created a potential blocked-in “boiler” effect and the practical fact is that you also have a very limited amount of liquid (essentially TEG) at the sump of the absorber to absorb the fire heat and convert it into saturated TEG vapor. Bear in mind that when you automatically block in the absorber, you have a very limited amount of TEG liquid in the vessel that represents the cooling of the vessel. I don’t think I’m wrong when I estimate that your TEG liquid level in the absorber will be far below the nominal 25-30 ft of fire height. If that is the case, then you will have a relatively large exterior surface area of the absorber directly exposed to the pool fire flames and the wall temperature will be exorbitant at those points above the liquid level – unless you have the absorber insulated and/or water sprayed (as mentioned in RP API 521). Once your “heat sink” (the liquid TEG inside the absorber sump) is vaporized or gone, the result will immediately be one of potential absorber steel wall stress failure and ultimate collapse if the fire persists. Also bear in mind that you can't practically depend on natural gas convective currents inside the absorber to keep your vessel walls cool. Gas heat transfer coefficients are notoriously horrible and that's why the API has had to revise their thinking on protecting vessels during the fire case scenario. You can only rely on internal liquids to maintain a reasonable wall temperature during a fire case. Vessels filled with gas should be depressured.

These are realistic and practical points that will surface during the Hazop you hold while trying to document a Management of Change (MOC) or, if this is a new design-construction, during the classical Hazop required prior to approving the P&ID. In my opinion, the relevant engineering guidelines and recommended practices should be those spelled out in RP API 521 and should focus on depressuring the vessel safely and also applying either insulation or water spray. A vessel structural failure should be of major concern not because of the capital cost involved, but because of the potential catastrophic results it can impose on your facilities and on human life. Unless you can accurately predict the duration of the vessel’s exposure to the fire flames (like for example, using positive external liquid drainage) you have little to predict whether any liquid will remain in the absorber – and bear in mind that as the liquid is differentially being boiled off, the wall temperatures are differentially increasing while subjected to external flames. Of course the absorber should carry the mandatory ASME pressure relief device. But from what you describe, I firmly would recommend you apply vessel protection in the form of insulation/water spray and depressuring as described in RP API 521.

I hope this experience helps.

 

[phorth]

I agree with everything in the last post (Montemayor). The relief valve will not prevent the vessel from rupturing in a fire, it should be automatically depressured at a rate that ensures that this does not happen. Traditionally this rate is to depressure to < 100 psi in 15 minutes, though I believe that API 521 now recognises that this may not be rapid enough for thin walled vessels.
Nevertheless, the fire relief valve still needs to be sized, and the relief sizing rate should be the total of vapour boiloff plus gas expansion. This is not an easy calculation and needs to be done after understanding the physical processes that are taking place, not by simply applying an approximate rule which could be lurking in the pages of API.
Referring to an earlier post, PV=nRT applies to all changes, not only adiabatic. Fire that client! But it is true for ideal gases only, which is definitely not the case for this TEG absorber.

 

[MortenA]

Art:

Just for my own benefit: Can boil-off water be disregarded in a fire case for a TEG contactor?

Best regards Morten

 

[Montemayor]

Morten:

I consider this a valid and pertinent point because it addresses the actual conditions in a realistic manner. I would not disregard the amount of water (& its rate) of boil-off. I believe an effort should be made to estimate the water amount trapped within the absorber with the blocked-in TEG and also make a conservative estimate as to how long the resultant steam will be relieving. I believe this rate will control the conservative sizing of the required PSV on the absorber. However, let's not overlook nor forget that we are fighting time in trying to preserve the mechanical integrity of the vessel due to the amount of liquid inventory in the absorber sump (which should also be estimated, taking into consideration the normal operating level and the amount of TEG wetting the packing prior to block-in). As I stated, the estimated total blocked-in quantity of TEG-water in the sump may easily not be very high and allow for flame impingement on relative dry vessel walls resulting in the early debilitation of the vessel's strength. I believe that such standards as Norsok take this situation to be so serious as to consider that the vessel is practically all gas and therefore should be immediately subjected to thermal protection and prescribed de-pressurization.

Here, I am addressing the subject in a general manner since we don't have specific basic data on the application. With all the specific data in hand I believe we can arrive at a better defined and engineered safe solution to the scenario.

 

[MortenA]

I see your point - this must be valid for all colums (more or less?)

Best regards

Morten

 

[hskk]

Hi guy,
Regarding PSV sizing for fire case,

my customer is very concern about the relieving temp during a fire.

how to calculate or estimate the relieving temp for isothermal approach?

the guy made a sentence as follow,"un-wetted & wetted conditions, the relieving temp may be different"

Is it true that 'isothermal' means the temp of the gas in the vessel (contactor tower in TEG dehydration unit) remains constant during safety relief when there is a fire?


In my opinion, isothermal is used only for unwetted case. this is because the gas temp in vessel will drop but the heat input causes the temp to remain constant.

For a vessel contained liquid, the temp of the gas in a vessel will rise very quickly during a fire and hence isothermal is not applicable. am i correct?

 

[MortenA]

I think this is exactly Art point.

Since the fire case PSV is only about preventing the "catastrophy" to spread you should consider the vessel to be un-wetted (and check with wetted).

As far as i understand the fire case is teh situation where deluge of the vessels fail. The vessel is more or less assumed "lost" anyway and the purpose is to prevent rupture so that escalation is avoided.

Since high wall temperature may mean that the steel "softens" this should be considered.

As far as i understand isothermal do mean that the temperature remains constant - but this is seen as opposed to a situation where the depressuring will cause the gas temp to decrease (adiabatic). The isothermal case assumes that energy is transferred to the gas (from the vessel walls) as the temperature drops.

Best regards

Morten