85 replies to this topic

[PingPong]

Why should you want to add water as one of the light components before the crude is cut into fractions? That is nonsens.

And if that has a big impact on the fraction yield in Hysys, then Hysys has a big problem.

 

The crude from the assay has to be cut into a mixture of lights and narrow boiling pseudofractions.

 

You can let Hysys do the cutting, which so far has been a failure,

or you can do the cutting yourself, as I already advised two weeks ago.

Your assay has very few points, so maybe that's why Hysys does a lousy job.

 

That mixture is the dry crude feed to the CDU.

You can add water to that dry crude feed before it enters the various unit operations.

[NAP]

In post#20, I had asked you the meaning of "cutting the crude yourself" and to check if my understanding was correct. I will ask again.

 

When you say to cut crude in pseudofractions ourself, I understand the following steps are to be done.

 

1. Plot the TBP vs Vol % curve  as provided in the assay which is like 5-6 points obtained from actual experiment

2. Devide this curve in narrow boiling pseudofractions (say each of 15 C interval). Record each pseudofraction's vol% and API gravity.

3. For each pseudofraction, calculate the molecular weight, Tc, Pc and accentricity factor using correlations such as TWU method

4. Using the estimated MW, Tc, Pc, AF etc, create a hypothetical component or pseudocomponent.

5. Create a crude oil stream by selecting these pseudocomponents and entering their vol% (recorded in step 2) as composition

 

The initial part of assay is the light ends whose data is available (vol% yield of light ends). We have to extrapolate the available TBP vs Vol% curve using mathematical functions from the last available experimental point to the FBP (not given in assay, which we are using as 850C).

 

Is my approach correct?

[PingPong]

For step [2] I would plot (curvefit) the UOP-K (calculated from each given assay point) versus TBP, and calculate gravity of each pseudofraction using its UOP-K and NBP.

 

Iw ould expect that steps [3] and [4] need not really be done by hand in Hysys, but merely specifying your preferred method is sufficient. It would be so in PRO/II.

[NAP]

For each point of the assay data, API gravity is also given along with vol% distilled. So I can calculate UOPK for each assay data point by the following formula.

 

UOPK = TBP^(1/3) / SG

 

So once I have the UOPK for each assay datapoint, I plot these against the corresponding TBP values.

 

Then I  create a smooth TBP vs UOPK curve by curve fit and lay over my pseudofractions to get each pseudofractions UOPK.

 

Using each pseudofraction's UOPK, I calculate SG for each pseudofraction from the above formula.

 

 

Alternatively,

 

When the assay data gives API gravity for each point, why not just curve fit for TBP vs API to get each pseudofraction's API value?

 

 

I am trying to understand the significance of evaluating SG from UOPK instead of getting it directly from a TBP vs API curve fit.

[PingPong]

In the assay API gravity varies from 93.7 to 3.6

 

whereas UOP-K's from that assay vary only from 13.55 to 12.07 which is a much smoother curve and with very little difference between adjacent fractions.

 

But you can do it any way you prefer, I only tell you how I do it.

 

Crude cutting is not an exact science. When doing it by hand you know what you are doing, you see the degrees of freedom (inaccuracy due to rather arbitrary way of drawing a curve between points far apart) when drawing, or fitting, a curve through only a few points.

Hysys uses a method that we don't know, and it is only one method. By hand you can tweak as much as you like, which is an advantage, or disadvantage, depending on your view (and experience).

[NAP]

Fair enough....I don't mind extra work as long as I know what I am doing it for.

 

I will come back with the new results from the manually cut crude...

[NAP]

I have done the manual pseudocuts as follows

 

Temp Range     No of Cuts

70 - 430            30

430 - 650          20 

650 - 870          10

 

What I have realised is that HYSYS does a good fit of predicted TBP curve to the actual assay TBP curve.

 

However, the corresponding predicted SG curve is a bad fit to the actual assay SG curve.

 

Right now, I am having to adjust some of the UOPK factors to adjust the densities of the pseudo-cuts to make the predicted SG curve failry match the assay SG curve.

[gegio1960]

very long discussion, difficult to follow. Anyway, I'd suggest the following procedure (based on my direct experience):

- if you have the TBP curve by vol and the SG curve you can heasily obtain, with excel, the TBP curve by weight

- input in hysys both the wt TBP points and the SG points (note: this is the minimum requirement to avoid big troubles)

- perform the "controlled" breakdown according to your assumptions in post #57 (note: you're considering one cut every 12°C in the light section and every 11°C in the middle section... it can be improved!)

- compare the generated curves (wt, vol and SG) vs the input ones. This time they should fairly agree (this couldn't be true in case you input only the vol TBP)

- if you want you can readjust the controlled breakdown ranges and repeat the procedure

Moreover:

- replace the initial 1-3 %wt of your TBP by a "fictitious" light end

- water content at the tower inlet shall be the one specified at desalter outlet

- add some nitrogen (representing uncondensables) at the top stage (see the effect on temp profiles)

- don't be too much optimistic with the pressure profile

- use pumparounds and side strippers, otherwise your product characterizations could result very different from the reality (this is not good ofr subsequent utilization in a Linear Programming model)

- show us some (significant) graphs/tables of your simulation results....

Good luck!

[NAP]

@ PingPong,

 

So far I have done the following.

 

1. Using assay TBP vs Vol% data, let HYSYS fit a TBP curve which matches closely to the assay curve

2. Devided this HYSYS TBP curve in narrow boiling pseudofractions

3. Calculated the UOPK values for 6 datapoints of the assay and plotted TBP vs UOPK curve

4. From the TBP vs UOPK curve, tabulated t the UOPK values for the pseudofractions I created earlier

5. From the UOPK values, calculated the density for each pseudofraction

6. Plotted the density vs vol% for pseudofractions and overlaid this with actual assay density vs vol% (curve attached)

 

Question: Is it OK to manipulate the initial values of calculated UOPK to bring these two curves closer?

 

 Crude TBP Distribution.xlsx   167.27KB   114 downloads

[PingPong]

That's an enormous amount of data that you dump on us....

 

There should not be such a mismatch between initial values and calculated UOP-K. The curvefit results therefor seem not quite right.

For example: I see that you use 12.58 voor TBP=70 and 13.464 for TBP = 80 whereas the assay gives 13.55 for TBP=70.

 

Obviously you can, and should, tweak all the calculated results to exactly fit the assay data.

 

Note however that in my opinion the API gravity data in the assay can not be correct, they are too high, resulting in cuts that all have higher UOP-K's (13.55 - 12.07) than that of the crude itself (11.86).

In other words: if you would mix all the calculated pseudofractions, you would obtain a crude with a few percent lower density than that in the assay (876.5 kg/m3). That is a problem that you need to tweak away, by adjusting the UOP-K of all fractions a few percent lower, until the mixed crude has exactly the same density as the crude in the assay. If the assay would not only contain the API gravity curve, but also densities of several TBP products (such as naphtha, kero, gasoil, AR, VGO, VR), then you would need to tweak the UOP-K's of the pseudofractions in each of those TBP products so that you get exactly the right density, not only for the whole crude mixture, but also for each TBP product mix.

Edited by PingPong, 03 October 2013 - 04:29 AM.

[NAP]

I did the crude cutting myself in HYSYS and brought the charecterized crude to as close as 11.7 UOPK vs 11.84 of assay. I have converged the simulation on draw rates which has resulted in slightly different cut points from 165-220-360 set but so be it as I dont have to publish cut points in the simulation results.

 

I have passed on the HYSYS model to a HYSYS expert to investiage why I couldnt charecterize crude to match the one in assay in HYSYS.

 

Now, I have shifted to PRO-II with an objective of getting more accurate results than HYSYS. Doing the same exercise of crude-cutting has resulted in a crude of 11.81 UOPK vs 11.84 of assay and 875.6 kg/m3 of density vs 876.5 of assay which I believe is fairly close.

 

My simulation model is ready to the point of heated crude stream from the heater outlet.

 

I tried the shortcut distillation to find out min. trays and reflux but to no luck of convergence. The intent here is to find out the min. trays required as you had commented that 45 trays for only 2 side draws is an overkill.

 

I gave LK as ethane and HK as NBP_840 component. For specs, I specified overhead Naphtha bbl/d rate and bottoms AR bbl/d rate.

 

Is shortcut distillation in Pro-II applicable for crude tower to find out min.trays and reflux or does it work well only for binary systems?

[gegio1960]

It seems you're completely forgetting the PAs... They impact both the "reflux" and the "number of trays"... but maybe these concepts are too much old for today design.

good luck

[NAP]

It seems you're completely forgetting the PAs... They impact both the "reflux" and the "number of trays"... but maybe these concepts are too much old for today design.

good luck

 

No I am not forgetting reflux or pump around.

 

Adding PA will reduce the reflux in the column and the trays between PA draw and return will not contribute to the the seperation there by increasing the number of trays required for the seperation.

 

But I need to know the min. parameters like reflux ratio, trays required for seperation so that I can build on it further to include P/A and side strippers. By this way, I know I havent installed too many trays in the column without being any benefit to the separation.

 

I am taking simple steps one by one so that my simulation doesnt get into an endless loop of non-convergence where I cant figure out as to what is going wrong. Hence the start with shortcut distillation

[PingPong]

Forget about trying to use shortcut distillation for a crude column with multiple products and multiple PA's.

 

Number of trays required depends on final PA duties, so you probably cannot finalise trays right now.

 

To determine trays required, simply use a trial-and-error approach.

 

For example: topsection of column, separation between Naphtha and Kero. Simply reduce the number of trays in that section until you approach the required 95-5 gap, while still meeting all other product specifications.

 

Then move down to section between Kero and GO drawoff and do the same.

 

Et cetera.

Edited by PingPong, 09 October 2013 - 02:24 AM.

[NAP]

I will do that once I get convergence atleast once, lol.

 

I dont understand the logic of "variables" under column performance specification section.

 

I mean I get the idea of specifications, the parameters that I want to target in my simulation. But why do I have to give variables corresponding to the specifications?

 

I am trying to specify draw rates of GO (42000 bbl/d), Naphtha (42000 bbl/d) and Kero (17000 bbl/d) as my performance specs for the column. So Pro-II is prompting me to specify variables like heat duty or draw rate of the same streams I am targeting spec for!

 

This particular bit is quite different between Pro-II and HYSYS

[PingPong]

PRO/II is for experienced people who know what they are doing, and understand degrees of freedom and how all variables are related.

HYSYS is for inexperienced people who have no clue what they are doing, but think that HYSYS knows everything that they don't know.

 

You should specify the bottoms products of the sidestrippers by varying the draw rates from the main column (to the sidestrippers).

Specify Naphtha Minus quantity by varying condensor duty (assuming you have a partial condensor with Naphtha Minus as dewpoint vapor product).

[processengbd]

I found Scfrac model to do shortcut model for CDU in Aspen Plus which is similar to Pro II Shortcut method.

Attached Files

•  Scfrac (2).pdf   23.81KB   86 downloads
•  SCFrac.pdf   57.52KB   90 downloads
•  CDU Simulation using PROII.ppt   2.94MB   114 downloads

[curious_cat]

HYSYS is for inexperienced people who have no clue what they are doing, but think that HYSYS knows everything that they don't know.

 

Love that. Nominated for the hall of fame. 

[gegio1960]

I use and like both (Hysys and Pro2). I think Pro2 is for people that have a lot of time available and Hysys is not for inexperienced people. Anyway, for both, it's valid the dear old statement "garbage in, garbage out". More than Hysys, I liked Hysim.

[NAP]

I am using overall tray efficiency of 0.5 across the column. Is that reasonably conservative?

 

Another suggested approach is to use zone-wise varying efficiency such as

 

AR stripping section: 0.3

GO - Flash zone: 0.4

Kero - GO: 0.5

Column top - Kero: 0.6

 

At the basic design stage, is it relavent to use zone-wise efficiency approach?

[PingPong]

Wise engineers base their simulation on theoretical stages. In that way they do not have to worry about tray efficiency and tray type while they still have other, more important, things on their mind. When actually designing the column later on, number of actual trays in a section is: number of theoretical trays divided by efficiency.

 

The fact that one can specify tray efficiency in a simulator does not mean that one has to use it. That applies to a lot of useless features in modern simulators that especially young inexperienced engineers feel they should use as much as possible....

 

Moreover efficiency depends also on type of trays (or packing) used, and column will no doubt also have chimney trays, so tray numbers in a simulation with efficiencies specified will not match actual trays on datasheet anyway.

[NAP]

My Pro-II simulation is based on theoretical stages only.

 

I was trying to get an idea of how the column is going to look like in terms of total height after I factor in the total trays, P/A trays, draw trays etc etc hence was wondering about the actual number of trays

[PingPong]

At the basic design stage, is it relavent to use zone-wise efficiency approach?

I am not sure what you mean by basic design stage. In my opinion that includes AFD (approved for design) heat & mass balances and equipment datasheets, so you should use the tray efficiencies as quoted in the design manual of your company. Once you give a column length, diameter, whatever, to other departments and/or the client it becomes very difficult to justify any change to the bigger, so best do it right the first time.

 

Column length not only depends on number of trays but also on chosen tray spacing(s), the heigth of the flash zone, and liquid hold-up in the bottom. And finally there is the skirt height, which depends on the NPSHR of the bottoms pump(s).

[NAP]

By basic design stage, I simply meant the initial simulation stage that I am at right now. I didnt mean the basic design stage in terms of a milestone context typically used by licensors during basic engineering package stage or EPC companies during FEED or early engineering in EPC

[NAP]

I think I have a decent solution now...please let me know if something looks unusual.

 

Charecterized Crude Properties

Std. Density - 875.6 kg/m3 (Assay - 876.5 kg/m3)

UOPK - 11.8 (Assay 11.84)

 

Flash Zone

Heater COT - 358 C

Overflash - 10.5 wt% (DWO definition)

FZ Pressure = 1.17 barg

FZ Temperature = 347 C

 

Tower configuration (16 ideal trays for fractionation + 1 ideal tray each for 4 pumparounds )

Tray 1 - Partial condenser (Temp @ 51 C)

Tray 2 - 6: Naphtha - Kero separation (includes 1 tray for top PA and 1 tray for kero PA)

Tray 7 - 12: Kero - AGO separation (includes 1 tray for AGO PA)

Tray 13-17: AGO - AR separation (includes 1 tray for HAGO PA..no HAGO product draw)

Tray 18 - Feed inlet, Flash Zone

Tray 19 - 20: AR stripping

 

Stripping Steam

Kero SS - 4000 kg/hr (Column draw flash point of 25 C raised to 42.5 C of side stripper Kero product)

Kero SS - 2 ideal stages

AGO SS - 8000 kg/hr (Column draw flash point of 50 C raised to 85.6 C of side stripper AGO product)

AGO SS - 2 ideal stages

AR stripping - 10000 kg/hr (To meet AGO D86 spec and AR quantity in bottom draw)

 

Product

LPG - 9170 kg/h

Naphtha - 42000 bpsd ( D86 90%T @ 150 C, spec limit is 155 C max)

Kero - 17000 bpsd (D86 90%T @ 217 C, spec limit is 230 C max)

AGO- 43000 bpsd (D86 90%T @ 356 C, spec limit is 360 C max)

AR- 95800 bpsd (Should have been 96000 bpsd..but this is the best I could get)