3 replies to this topic

[Dmitry]

Hello, somebody would you mind to help me with next issue.

I work on a gas plant and have been working with Hysys since last time. I am trying to simulate LPG recovery unit in Hysys. I got propane refrigeration unit to recover HC from export gas due reducing temperature of inlet gas from -15 C to- 38 C and LT separator to separate glycol+water, HC from gas. I simulated it in Hysys and specified inlet feed gas composition at operating temperature and pressure.

Inlet gas to the chiller:
T = - 15C
P= 400 psig
Q= 34MMSCFD
COMPONENTS MOLE MASS FLOW
FRACTION (kg/h)


Methane 0.856 23300.09
Ethane 0.082 4183.63
Propane 0.0321 2398.429
i-Butane 0.0035 347.24
n-Butane 0.0054 532.83
i-Pentane 0.0013 163.2
n-Pentane 0.0009 105.95
H2O 0 1.513
Helium 0 0
Nitrogen 0.0135 643.10
CO2 0.0053 394.65
EGlycol 0 0.0074


After LTS I got a gas stream with next composition (by Hysys):

T = - 38 C
P= 390 psig

COMPONENTS MOLE MASS FLOW
FRACTION (kg/h)

Methane 0.8711 23136.46
Ethane 0.0793 3948.678
Propane 0.0256 1872.44
i-Butane 0.0020 189.9
n-Butane 0.0024 232.56
i-Pentane 0.0003 34.58
n-Pentane 0.0001 16.27
H2O 0 0.066
Helium 0 0
Nitrogen 0.0138 641.96
CO2 0.0053 385.02
EGlycol 0 0.0009


As you can see I still got a lot of C3-C5 in gas stream from LTS. And as result Hysys calculated 2% more C1-C2 product, 50 % less C3-C4 product abd 35% less C5+ product after fractionation train if compare with my gas plant report.
I think it depends with excess of C3-C5 components in the export gas (gas stream from LTS). Recovery of HC depends from operations T and P, all conditions are the same ason plant. What it is wrong?

Regards

Edited by Dmitry, 11 May 2012 - 04:50 AM.

[Zauberberg]

Your results look good. LTS units are seldom used for C3+ recovery as they cannot recover much of heavier components from a gas stream. For such deep recovery of C3+ components another technology is used - a turboexpander plant. LTS units are normally employed for Hydrocarbon dew pointing, it's their main purpose.

For turboexpander plant, look at Ortloff's website - they are the licensors of many Propane+ and Ethane+ recovery plants worldwide.

http://www.ortloff.com/

[Dmitry]

Yes, we use LTS to recovery C3+ from natural gas and then deethanizer and debutanisez columns to separater LPG from C5+. Now I don't look information about another methods of recovery LPG, I just want to simulate existing LPG precovery plant in Hysys. Thanks for the link, it's always useful to find new inf.

[Fletch]

Okay, so interpretting what you have provided, HYSYS is underpredicting your recover that you are seeing in the operational plant (in most cases).
Matching the simulation to the real world is often tricky.

On the simulation side, equation of states do their best to predict the fluid behaviour. A different EOS can produce quite different results, not to the extreme you are seeing, but say 2-5% differences.
However, real world data can often be misleading, as it is taken at one point in time.
Does your composition vary to the LTS?
What is the purity of the recovered C2 and C3/C4 ? are the purities inline with how your columns are modelled, you could be simulating a very pure product, but in fact you are slipping more light or heavy ends into them, increasing the product flow compared to the simulation.

To better understand how well or poorly Hysys is predicting you might need to compare smaller windows. i.e. can you get gas samples around the LTS or get the flow rate out of the LTS and compare
You might find that HYSYS is predicting very well here, and the larger descrepancy occurs in your fractionation unit.
But going by your above results, it seems that the descrepancy starts with the separation at the LTS. So maybe the EOS is not prediciting the phase envelope correctly and therefore the you are actually acheiving a deeper cut into the vapour fraction.

If you can generate some more localised data around the LTS, and Hysys continues to underpredict, you could look at adjusting the binary parameters of the fluid package for your case.
This would allow you to tailor simulation to reality... but this is not a simple task (and not one that i have done)