HYSYS Topping unit distillation crude 2

[paulinaote]

Hello everyone:
I simulate topping unit where in current operating conditions. After reconstituting the oil improve the simulate data. But I do not matches the pour point of diesel.
The pour point (API 1974) is -8 ° C and should be 0 ° C.
Urgent need that this simulation is as real as possible because I have design a simple vacuum flasher to recover the diesel that is lost as reduced crude.
Attached is my simulation.
Hopefully someone can help me, my mind is somewhat blocked.
The product specifications are as follows:

Naphta:
API:65,2

Kerosene:
API: 46,8
Flash Point °C: 42
Freeze Point °C: -50,2

Diesel:
API: 38,2
Flash Point °C: 79.

 

[EmmanuelTop]

I don't know how much different these crude oils are. It could be a big deal, it may not be a big deal. What I know for sure is that using only ASTM D86 for crude oil characterization will not be very much useful. So I guess you did the best possible thing by re-blending products and introducing them as feed, instead of relying on D86 results for the total crude charge.

A useful tip: never specify properties and yields of a product at the same time. In simple words, if you want to produce e.g. 30 m3/h of Diesel, don't specify 95% D86 or Pour Point of Diesel if you have specified the flow/yield already. Property is a direct function of draw-off rate so if you are forcing software to match both criteria, it will likely fail to converge.

Dejan IVANOVIC
Process Engineer, MSChE

 

[paulinaote]

Thanks for the advice, I think it's exactly the mistake I make.
Then keep the characterization as is.

 

[paulinaote]

Not for iterating...I've tried in many many many ways, but there is something I'm not seeing.
H E L P

 

[EmmanuelTop]

I think you should focus on one crude source only, for a start. There seems to be a problem with phase envelope of the crude charge stream - and this could be due to million reasons. For simulation purpose try to use the crude oil for which you have the most comprehensive information (the most detailed assay) and start from there. Start with a blank flowsheet and this will eliminate all the excessive pseudo-components.

I think your column cannot converge because of inability to perform T-P flashes. You can see that from the phase envelope - it is a single straight line (like a pure component). This needs to be fixed.

Dejan IVANOVIC
Process Engineer, MSChE

 

[paulinaote]

OK.. will try again tested mixtures. I shall guide from phase evelope utility as you mention.
Has you a reference picture as it should be that curve?
Regards

 

[paulinaote]

you mean the green line, right? This should not be a single straight line

 

[EmmanuelTop]

Use e.g. only "Crudo X" in the model since you said you have the crude assay with you and "Crudo X" represents 35% of the feed blend. Make it 100%.

There needs to be a clear 2-phase region in the phase envelope. See examples for different types of oil/fluids:

http://www.fekete.com/SAN/WebHelp/F...al/General_Concepts/Reservoir_Fluid_Types.htm

and

https://www.metu.edu.tr/~kok/pete110/PETE110_CHAPTER6.pdf

Dejan IVANOVIC
Process Engineer, MSChE

 

[paulinaote]

I did the simulation again with a feed:
CRUDO X: 50% date: TBP - API - Viscosity
CRUDO Y: 50%: TBP - API - Viscosity
a total 2650 m3/d (april average).

The result is far from the actual (Cold properties product - T Condenser - T tope - T bottom - D86 HYSYS v.s Lab)
Which is the reason this BIG difference? A master phase envelope? Do not generate the bubble point line, i don´t know why.

can you advise me?
Regards

 

[paulinaote]

Master phase envelope

 

[paulinaote]

comparison of curves D86

 

[EmmanuelTop]

Attached is an updated model (native file in 7.3 version). I spent solid 2 hours playing with it, so you owe me a beer whenever I visit your country.

- I've changed the fluid package model to SRK. It yields more accurate results sometimes. Now the crude oil phase envelope looks more realistic.
- I managed to obtain desired product properties (speaking of D86). Targeting IBP or FBP can lead to larger errors. That is why 5%, 10%, 90%, and 95% D86 are more commonly used.
- By matching properties, there aren't many degrees of freedom to play with the flows/yields. Naphtha cut point is determined by the tower top temperature, so if you want to match the cut point (lets use 90% D86), then you have to leave the simulator to calculate the flow. There is more room to play with Diesel and Kero, but ultimately the combined yield of these two products will be more or less constant if you have fixed the product properties.
- Diesel stream has a very low flash point. Is there any stripping steam at all? A good practice is to leave steam control valves on manual, not on automatic - especially if meters are not calibrated regularly.

This is certainly more closer to reality but hard to say whether it is decently away from reality, or not. I would have to see the plant and look into operating parameters myself. Also, remember that for producing a quality model not only the feed should be characterized properly, but also the instruments and meters around the column need to be zeroed and calibrated. What they are showing currently does not necessarily need to be true.

Dejan IVANOVIC
Process Engineer, MSChE

 

[paulinaote]

Many many thank you very much for your time. Consider a friend in Patagonia.

I worked all day in the simulation,followind your advice. Finally arrive at more realistic results (envelope phase and operating parameters).

These are the actual data:
PRODUCTION m3/d
NAFTA: 570
KERO: 450
DIESEL: 520
CRUDO REDUCIDO: 1110
FEED TO PLANT: 2650

PROPERTIES:
NAFTA KEROSENE DIESEL
63.36 °API 47 °API 38.08 °API
ASTM D86 °C Flash point 41°C Flash Point 79 °C
0% 38 Freeze Point -51 °C Pour Point -6 °C
10% 69 ASTM D86 °C ASTM D86 °C
50% 101 0% 147 0% 179.3
90% 129 10% 165 10% 225
100% 161 50% 192 50% 274
90% 230 90% 316.3
100% 256 100% 340

In EXCEL attached plant data.
I have 3 simulation with different convergence and tray distribution.
My problem is that I have result in very light products (sim 17-sim 18) and very heavy products (sim 19, especially naphtha)

I need to prioritize a realistic result for diesel and naphtha, because i need simulate a splitter of nafta and a vacuum column to recover diesel from reduced crude.

I'm stuck in the simulation of topping, please help.
By the way do you have information to simulate HYSYS a vacuum tower? Because I have no idea how to do it(....for now). I never simulate a vacuum tower.

Hopefully you can help, I owe you a beer.

 

[paulinaote]

EXCEL

 

[paulinaote]

PROPERTIES:
NAFTA KEROSENE DIESEL
63.36 °API 47 °API 38.08 °API
ASTM D86 °C Flash point 41°C Flash Point 79 °C
0% 38 Freeze Point -51 °C Pour Point -6 °C
10% 69 ASTM D86 °C ASTM D86 °C
50% 101 0% 147 0% 179.3
90% 129 10% 165 10% 225
100% 161 50% 192 50% 274
90% 230 90% 316.3
100% 256 100% 340

Sorry for this

 

[paulinaote]

PROPERTIES

 

[EmmanuelTop]

I would gladly help further but I do not have the time to fine tune the simulation. Also note that I don't have access to plant operating data, and for a model to be considered successful some back and forth information flow is required. One cannot develop a reliable model unless it is calibrated and verified against multiple sets of input and plant data.

If you have obtained decently accurate simulation model (in terms of product yields and properties), that's a pretty good start. I would not bother with cold properties if you managed to match the yields and distillation cuts. Hysys isn't so good in predicting cold properties, as mentioned by others in this thread.

As for the vacuum distillation simulation, remember to use Brown K10 or Esso K thermodynamic package. I think there should be a simulation template/sample on Vacuum distillation in Hysys files - you can start from there. I am attaching one paper of oil characterization, and below is the link for Aspen property packages.

Good luck.

http://sites.poli.usp.br/d/pqi2408/bestpracticesoptimumsimulationshysyspropertypackages.pdf

Dejan IVANOVIC
Process Engineer, MSChE

 

[paulinaote]

Thanks, can you send me esta paper via e-mail, there a problem with the proxy.
paulinabarrientos.s@gmail.com

 

[paulinaote]

Hello!!....again.
I´m tunning my simulation and I originated a question. Is there a problem if the number of theoretical trays exceeds the number of actual tray??

Regards

 

[poli60]

In turn, the answer is in another question: did you ever hear of a tray efficiency greater than 1?
good luck

 

[paulinaote]

(Y)