15 replies to this topic

[shahad]

Hi 

I am doing a flash calculation around a reflux drum (inlet is vapor, and the outlet is 2 phase) at 38oC and 2 bar

Components in the inlet include: 

 

Hydrogen             Hydrogen Sulfide     Methane              Ethane               Propane              I-Butane             N-Butane             I-Pentane            N-Pentane            Water                MEA   Nitrogen             Naphtha

 

I need to find the Saturated vapor pressures to find the Ki values, but I am having difficulties with the pressures as all the Antoine coefficients are out of the temperature range, except isopentane and npentane. So how can I get them?

Also does anyone have any recommendations for Naphtha? I know its molecular weight, accentic factor, critical temperature and pressure, so should I just match it to the closest hydrocarbon in terms of boiling point or MW?

[MrShorty]

38 C is above the critical temperature of 4 of those compounds (H2, CH4, C2H6, N2), so there is no way to talk about the saturated vapor pressures of those 4 compounds. As limited as the Antoine equation is (usually limited to pressures below 1 to 2 atm), I'm not surprised that the vapor pressures for many of the other compounds are considered outside the range of their respective Antoine equations. For these compounds, you would need a better vapor pressure equation -- extended Antoine, Riedel, Wagner, or other (https://chejunkie.co...and-estimation/ ).

 

What kind of thermodynamic/phase equilibrium model are you using? Are you assuming Raoult's Law and/or Henry's Law? Incorporating activity coefficients? Maybe using an equation of state for both phases? You ask about finding K values, but what models/assumptions are you using for determining K values?

[shahad]

I have a simulation using SRK, and I am doing confirmatory calculations, so they dont need to be very accurate. I am assuming ideal gas behavior, so I am wondering where and how I can find the K values for these components?

So maybe solubility? But I cannot find any literature 

[MrShorty]

Ideal gas behavior can describe the vapor phase, but cannot describe the liquid phase. I assume you want to use some ideal solution (Raoult's Law and/or Henry's law) for the liquid phase?

 

For Raoult's Law, a suitable vapor pressure equation is all that is needed to calculate partial pressures and K values (when total pressure is known). I'm not sure what sources you have available for vapor pressure data. My first choice is usually DIPPR's database, but you can also get vapor pressure information from other sources. NIST's website might be as good as any for the well known natural gas hydrocarbons https://webbook.nist...hemistry/fluid/Henry's constants are harder to come by -- especially in free internet sources. My go to source for Henry's constant information is DeChema's database or the Dortmund Data Bank, but I don't know if you have access to those.

 

An internet search reminded me that we have previously talked about using GPSA engineering data book which is a graphical approach to getting K values and/or Henry's constants: https://www.cheresou...e-in-petroleum/The discussion was specific to H2S, but the data book should have charts for most natural gas type compounds.

 

I'm not sure what to specifically recommend because I don't know what resources you have access to. If you can find resources you like for vapor pressures and Henry's law constants, you should be able to make quick VLE estimates from that information to check your SRK calculations against.

[SilverShaded]

There are spreadsheets available online which use Peng-Robinson, this should give you a similar result to SRK for the components you mention and would do as a check.

[SilverShaded]

You could also use Wilsons equation to estimate k.

Edited by SilverShaded, 21 April 2021 - 02:41 AM.

[MrShorty]

@SilverShaded: How are you expecting to use Wilsons equations for estimating K? I usually associate Wilson's equation with an activity coefficient model which is going to bump up against the same problem mentioned in the OP -- what to use for pure component vapor pressures and/or Henry's constants along the way to calculating K.

[SilverShaded]

Different Wilsons equation than the one your thinking off probably; (probably more correctly described as Wilsons correlation)

 

K = Pcrit / P * Exp(5.373 * (1.0 + Omega) * (1.0 - Tcrit / T))

Edited by SilverShaded, 23 April 2021 - 11:17 AM.

[MrShorty]

That is one I had not heard of before. Whenever I see "Wilson's equation" without any other qualification, I assume one means his expression for excess Gibbs energy which leads into the activity coeffcient equations. This looks like a useful approximation in that it only needs critical properties and acentric factors.

[breizh]

Hi,

Consider the doc I prepared (flash using EOS) , you will find an estimation of Kis known as Wilson correlation, valid for P< 35 bars

 

Good luck

Breizh 

[SilverShaded]

That is one I had not heard of before. Whenever I see "Wilson's equation" without any other qualification, I assume one means his expression for excess Gibbs energy which leads into the activity coeffcient equations. This looks like a useful approximation in that it only needs critical properties and acentric factors.

Its often used in commeircal simulators to get a first estimate of liquid/vapour composition before the algrothm switches to something more rigorous.

[SilverShaded]

Hi,

Consider the doc I prepared (flash using EOS) , you will find an estimation of Kis known as Wilson correlation, valid for P< 35 bars

 

Good luck

Breizh 

Did you consider solving the cubic directly?

[breizh]

Hi SilverShaded,

Probably yes .

Breizh 

[SilverShaded]

Hi SilverShaded,

Probably yes .

Breizh 

I guess it solves quite quick the way you did it anyway.  I had to do something similar to you for Lee-Kesler Enthalpy/Entropy (5 order etc).

[breizh]

Hi,

Sorry I did not pick up your query in the previous post . Instead of Newton Raphson , you can use Halley Method ( 1st & 2nd derivative) to get a faster convergence . 

BTW , domain is quite small  [0,1].

If you want to know more about the method consider Wikipedia or your favorite search engine . 

You should be able to find an example in this forum related to the calculation of the compressibility factor Z .

Good luck

Breizh 

Edited by breizh, 26 April 2021 - 08:38 PM.

[SilverShaded]

Hi,

Sorry I did not pick up your query in the previous post . Instead of Newton Raphson , you can use Halley Method ( 1st & 2nd derivative) to get a faster convergence . 

BTW , domain is quite small  [0,1].

If you want to know more about the method consider Wikipedia or your favorite search engine . 

You should be able to find an example in this forum related to the calculation of the compressibility factor Z .

Good luck

Breizh 

 

Thanks, however I allready have my own simulator.  For cubic compressbility its more efficient to solve directly rather than iteratively.  I found some commercial (and a well known open source) simulators struggle to solve Lee-Kesler correctly close to the critical point as they often solve to the wrong root, it is however very tricky to solve correctly.