API 2000, PVRVs, and crude oil

[KernOily]

Hi guys. Looking for opinions/thoughts/nasty comments/rude gestures.

I have a 23,000 bbl tank with 10' of water and 5' of 13 deg API heavy crude oil. I am trying to size the PVRV. The fire sizing equation in API 2000 wants a molecular weight and a latent heat.

It is reasonable to assume that a fire outside the tank will cause the lowest NBP cuts to boil off first and thus the PVRV should be sized using the M and Hfg for those cuts. Problem is, I do not have a crude assay so I have no idea what the first cuts are. All I have are pseudocomponents generated by my simulator using a black oil model. The lowest cut from the simulator is NBP513. Not exactly a light cut, that...

So my take is to used water as the fluid exiting the PVRV, so use M and Hfg for water. I don't think I can defend using anything else. I could use the relieving rate so generated and add an arbitratry mutiplier to it to account for the light cuts which will go off before the water starts to go off. Or, just go up one size on the PVRV, say from an 18" to a 20". What say ye? Thanks ! ! ! Pete

 

[pleckner]

API suggests the use of 50 Btu/lb for a latent heat if you don't have anything else. But years ago I did a relief system for a Lub oil plant and our simulators gave me a latent heat of around 15 Btu/lb. For molecular weight use the lightest component.

 

[MJCronin]

Kern,

Would you get better responses if you were to post on the "Safety Relief valve Engineering forum" ?

regards

-MJC

 

[KernOily]

Hi guys. Thanks for the replies.

pleckner - I do not have a crude analysis, therefore I don't know what the lightest component is. Can you please post the section from API2000 where it recommends using 50 Btu/lb for Hfg? I can't find it in there.

MJC - I dodn't post it over there because a tank PVRV is not a PSV or SRV, per se. That's why I posted it here. If I don't get any more replies, I 'll post over there next.

Thanks! Pete

 

[pleckner]

It's not API 2000, its API Standard 521, Section 5.15.3.2. This really deals with what to do near the critical point but I would be comfortable to extend that to anytime you don't have accurate information.

 

[KernOily]

pleacker - thanks. As is typical of API STDs and RPs, there is no shortage of confusion and conflict. I am not sure that 521 is applicable to storage tanks. 521 itself references 2000. But 521 also calls out information for tanks. In Appendix A, the discussion talks about 'tanks', yes the Table column heading is 'Vessels'. The disctinction is usually clear between a storage tank and a vessel in regard to the API docs most of the time.

In short, I'm not sure I can stand up in court with a straight face and defend the use of 521 for a storage tank PVRV, especially when that tank is an API 650 tank. 650 does not reference 521 but does reference 2000. In section 3.10.8.3, 650 calls out specifically that "Emergency venting requirements are satisfied if... the tank is
equipped with pressure relief devices adequate to prevent
internal pressure from exceeding the tank design pressure and meeting the requirements specified in API Standard 2000 for emergency venting." Looks to me like I'm stuck with 650... Thanks! Pete

 

[pleckner]

I totallly agree. For low pressure tanks, API 520/521 doesn't directly apply. Take a look at NFPA 30 as well.

 

[JoeWong88]

It is reasonable to assume that a fire outside the tank will cause the lowest NBP cuts to boil off first and thus the PVRV should be sized using the M and Hfg for those cuts.

For multicomponent heating, you are right that the light end (low MW component) will leave the liquid mixture when it reaches the initial boiling point (IBP). BUT, in my previous experiences, lowest NBP cuts with low molecular weight (MW) and Heat of vaporisation (Hfg) may or may not results largest Pressure Relieving Area. I have made some simple studies on step-wise heating with multicomponent flash, the maximum relieving area (peak) may (or may NOT) occur at the initial boiling point (IBP). The required relieving area is increases as it is boiled until it reach peak, then it will required relieving area will drop. The peak varies with mixtures, pressure, temperature, etc. This solely with my limited experience and observation.

Why i stated may (or may not) as i have experienced some combination of liquid mixtures at P and T, the peak is at the initial boiling point (IBP).

Problem is, I do not have a crude assay so I have no idea what the first cuts are. All I have are pseudocomponents generated by my simulator using a black oil model. The lowest cut from the simulator is NBP513. Not exactly a light cut, that...

Eventhough you have crude assay, you may get the relieving area but there are still some degree of uncertainties if the the maximum relieving area is at it initial boiling point.

On the other hand, API 2000, section 4.3.3.2, Tank without weak Roof-to-Shell attachment, you may calculated the relieving rate using equations in 4.3.2.2.1. However, if you need lesser degree of accuraccy, you may choose to use method & table as stated in 4.3.3.2.2.

I would suggest you may consider to use 4.3.3.2.2 until you have a good definitin of the composition.

 

[JoeTank]

The basic API-2000 rules assume a very light product (hexane, I think?). Can you survive using that very conservatve assumption and not worry about a specific assay of the oil?

Joe Tank

 

[KernOily]

Hi guys. Thanks for the great replies.

JoeWong - I appreciate your point. The vapor composition will of course be controlled by equilibrium. I can't really calculate it because I don't have an assay. If I had an assay I could do the calcs. Blackoil pseudocomponent models are really only good to estimate bulk properties.

Joe Tank - Great idea. That is in fact what I did. I actually used n-heptane because that is the component that has Pvap close to my tank operating pressure and temperature (210 F and 0.1 psig) so that will be (mostly) the first component to flash upon the application of heat from an external fire. This method gives me a relief load about 1/6 that of the method in 4.3.3.2.2, which suggests a HUGE relief load. Thanks! Pete