12 replies to this topic

[Wendy]

Hello Everyone,

I am working for an oil and gas consultant company as a process engineer, not experienced tho. I found this website very useful and am happy to post my first question here. Hope I can receive valueable comments from the forum. Much appreciated

One part of the PSV fire sizing requires to calculate the effect of fire on the wetted surface of a vessel, and hence needs to estimate the Latent heat of Vaporization of the liquid phase. I am always unsure what to use for the latent heat of vaporization for Hydrocarbons.

When I do PSV sizing on scrubbers, the HC liquid contains mixtures of components that may have a wide boiling range. I often refer to the Chat D3 in API 520, although its for pure single component paraffin HC liquids. Do you think this would be a good estimate for the latent heat of HC mixtures? Should I use "a rigorous series of equilibrium calculations" as discussed in API520 D6.1, which I am still not too sure...

Many thanks

[pleckner]

Wendy,

It all depends on how accurate you want to be. As you boil a mult-component mixture the lights boil off first. Therefore, the very first boil-off will give you the most conservative latent heat so I would typically flash the saturated liquid to 1% vapor and use this heat value for the latent heat. But depending on the composition of the mixture, one can actually devleop a larger size relief valve sometime after this initial boil off.

"Guidelines for Pressure Relief and Effluent Handling Systems", Center for Chemical Process Safety (CCPS) of the American Institute of Chemical Engineers, 1998 recommends one of four approaches depending on the degree of detail and accuracy of results:

1. Determine the single-component venting requiremnts for each component alone

2. Develop the vapor-liquid equilibrium relationship and caluate the intitial equilibrium phase composition (pretty much what I describe above)

3. Obtain the vaporization rate over the course of the venting incident from a multicomponent batch vaporization computer routine using the given heat input rate

4. Simulate the incident using a multicomponent pressure relief computer routine such as SAFIRE or SuperChems for DIERS

[Wendy]

Thanks a lot, Phil

I have played around it on the simulation. I put the saturated liquid stream through a cooler and add a small amount of heat to flash off the liquid - use this value as the latent heat.

And thanks for the reference.

[latexman]

Wendy,

Having been sizing PSV's for about 28 years, I would like to recommend method # 1 in pleckner's reference. It is conservative and the simplest and quickest of all 4 methods. You do not have to worry about vapor-liquid equilibrium, flash calculations, or run a simulator, and that is a time saver!

If you go through the sizing of a PSV by hand for a single pure component, you will see that you can develop a parameter made up of the component’s physical properties that is proportional to the nozzle area of the PSV. If you calculate this "physical property parameter" for every component in a mixture of concern (Excel is your friend here!), the worst case component will be obvious upon inspection. Then, using good engineering judgment, you can decide if you want to use that component as your sizing basis.

Most of the time, I have found you will probably want to use this worst case component as your sizing basis, but there will be times when you do not want to use this worst case component. For example, that component may be in the mixture in very small quantities and it may impact the sizing of the PSV significantly. If it's concentration in the mixture is < 1% (my rule of thumb), you probably do not want to use it as your sizing basis, but you need to think it through. If it's concentration in the mixture is > 10% (again, my rule of thumb), you probably will want to use it as your sizing basis, but, again, you need to be happy with that decision.

There will still be cases where you need to get more sophisticated to size a PSV, especially if the PSV is going to be large and expensive or if there is a significant risk that justifies it, but, in my experience, method # 1 can be used the majority of the time.

[Kryz]

Hi All,

I think I have similar problem to Wendy. Need to do fire sizing for filter - fire vapour generation case.

The liquid inside vessel is gasoline (simply unleaded fuel for car) or naphtha in more technical terms. This is mixture of hydrocarbons as all we know. How obtain data needed for sizing. Apart from heat of vaporization, also molecular weight is needed and Cp, and even Z (compressibility). I am using Farris Size master.

I found in literature some number for gasoline heat of vaporization like 140 BTU/lb or other source 184 cal per gram. Both numbers are close if one converts the units. My concern is however as Phil described that there are light components, which flash first. How is serious to worry about this.

Currently I determined from data I have that I need 4 sq millimetres. How sensitive it can be if I try to use conservative component as pleckner/latexman advice. I can apply now simply PSV with C orifice for example model 27CA13-M20 by Farris. This one has 44 sq mm - a lot of spare capacity. I wonder if it is worth to spent more time and do Hysys simulation as described. And for that I need light end components and TBP curve, to produce assay, making pseudo components and determine properties. It is tedious work. I believe in this case not worth to do it. The filters are small dimension and to be beyond 44 sq millimetres is unlikely to be happen.

But for future reference if meet the similar problems I wonder how your experience telling you about sensitivity. How the components may impact. For example molecular weight – I selected something like MW = 130 (it was naphtha stream from on example in process simulator) but who know what number should be used. It is mixture; I have no any particular data like TBP curve. Do you believe that variation in thermo properties has still less significant impact on orifice fire size than vessel dimensions, liquid level in vessel and height above the ground, and abviously evironmental factor, which those are more serious driving factors for PSV size.

Perhaps big difference in heat of vaporization could impact seriously.

But I check my other calcs for liquid ethane accumulator fire sizing for example. Heat of vaporization quoted above for naphtha I compared to pure ethane and it is very similar – the number is 358.5 Joules per kg, or I know you prefer imperial units so it would be 154 BTU/lb. Very similar to 144/140 BTU/lb.

[demank]

Hello,,
for latent heat determination, of course we need the assay data of hydrocarbon mixture. With HYSYS software, you can simulate it by for example 10 degree Fanrenheit simulation.

[JoeWong]

Hello,,
for latent heat determination, o course we need the assay data of hydrocarbon mixture. With HYSYS software, you can simulate it by for example 100 degree Fanrenheit simulation.

Hi Demank,
Can you please elaborate a bit more in detail the method you proposed ?

[demank]

The idea is from GPSA chapter 5

The following steps to be performed to determine the relief rates at interval of 10 oF temperature increase using HYSYS:
a. Pressure of equipment (vessel) is first increased to the relieving pressure.
b. Simulation is carried out based on isochoric condition.
c. Temperature is then increased at increment of 10 oF to provide gradual change in the vapor-ized liquid rate.
d. Dividing the amount of heat input into the equipment (vessel) for every increase of 10 oF (from HYSYS) by liquid vaporized rate gives latent heat and relief rate changes within the system at every 10 oF of temperature increment.
e. Latent heat and relief rate calculation is stop when the calculated vapor flow stat to decrease with the increasing of temperature.
f. The largest relief rate among those at each 10 oF of temperature rise and the corresponding latent heat will be selected for PSV sizing/ design.

[JoeWong]

Thanks Demank.

Probably this post "Determine Latent Heat for Multi-Component and Relieving Area Using Rigorous Method in HYSYS" may have slightly different approach.

What determine 10 degF... my experience is dT subject to system pressure and temperature. Less dT for initial step but higher dT for later step...

Largest relief rate DOES NOT mean largest relief area. Properties change along the steps will also affects the relief area. Please take note.

[demank]

"What determine 10 degF... my experience is dT subject to system pressure and temperature. Less dT for initial step but higher dT for later step..."
I absolutely agree with above. (Although 10 degree is enough regarding the purpose for this simulation)

"Largest relief rate DOES NOT mean largest relief area. Properties change along the steps will also affects the relief area."
Those HYSYS step is utilized by me to get the value of relieving temperature and the latent heat.
This latent heat is used to calculate relief rate, W = Q / latent-heat.
Then largest relief rate in HYSYS simulation is chosen to get the properties of relief condition.

[JoeWong]

"Largest relief rate DOES NOT mean largest relief area. Properties change along the steps will also affects the relief area."
Those HYSYS step is utilized by me to get the value of relieving temperature and the latent heat.
This latent heat is used to calculate relief rate, W = Q / latent-heat.
Then largest relief rate in HYSYS simulation is chosen to get the properties of relief condition.

Suggest you to calculate relief area for each steps. Then you will understand what i mean.

[Kunal_DNV]

Hi,

Can anyone please email me "Guidelines for Pressure Relief and Effluent Handling Systems", Center for Chemical Process Safety (CCPS) of the American Institute of Chemical Engineers, 1998 on my person id kunal14284@gmail.com.

Regards
Kunal

[amitkatyal.asim]

For finding out the latent heat of a hydrocarbon mixture, you can visit www.eq-comp.com/liq-prop.htm . Though the software LIQ-PROP is not freely downloadable but you can send your query by filling the downloadable excel sheet and sending it to info@eq-comp.com . You will get a reply within 24 hours.
For your conveinience I am attaching the query excel sheet downloaded from this site. Kindly fill it and send it to info@eq-comp.com .  INPUT SHEET - LIQ-PROP.xls   21KB   240 downloads