You are on page 1of 6

Rigorous Distillation

There are three modules (IO, SURE and CHEMDIST, in a way they are just subroutines) that can perform
rigorous distillation calculations. The modules solve the general heat and mass balances of equilibrium
staged operations and thus apply to gas absorption, stripping operations and liquid-liquid extractions as
well as distillations. (Of course, we are only applying the concept of theoretical stages to absorptions or
liquid-liquid extractions that use a continuous column.) The differences among the modules are mainly in
the iterative algorithms and how they handle the thermodynamic properties. The SURE method uses a
Newton-Raphson convergence scheme with matrix partitioning, while the CHEMDIST method uses the
modified Naphtali-Sandholm algorithm which involves stage grouping as a variation of the basic Newton-
Raphson procedure. Both of these methods and how to handle liquid-liquid extraction simulations are
explained in detailed in Seader and Henley. The IO (inside-out) algorithm uses the stripping factors as the
iteration variables. The convergence is fast because the column is always in mass balance. Simplified
thermodynamic models are also used to solve the column heat balances in earlier trials (inner loops).
Simple guidelines on which one to use:
Always try the IO algorithm first; this is the default method in PRO/II. It is fast and supports most
conventional fractionators, absorbers, and strippers (Section 72). Except for a pure water decant
from the condenser, IO does not support two liquid phases.
When IO fails or when you have non-ideal systems, systems that form two liquid phases or
azeotropes, or systems that have free water, use CHEMDIST. You will also have to use a
thermodynamic method that uses liquid activity coefficients; CHEMDIST does not work with
generalized K-value predictors such as GS or SRK (Section 74).
When neither IO nor CHEMIDST converges or applies, use SURE. It has better convergence and is
more versatile, but slow (Section 73). SURE is the only algorithm that can handle total
pumparounds and water decant on any tray.
Use the LLEX option when you deal with liquid-liquid extractions (Section 79).
A few words on the conventions:
Trays are always numbered from top to bottom.
If there is a condenser, it is tray number 1. (Essentially it is the first theoretical stage with an
associated cooler. See Henley & Seader, Chapter 15.)
If there is a reboiler, it is the bottom tray.
There is no upper limit on the number of trays, but there must be at least two stages.
There is no limit on the number of feed or product streams.
The pressure of each tray must be specified and is invariant except in CASE STUDIES.
The configuration of the column must be completely defined, including the total number of
theoretical trays, and the locations of all the feeds, product draws, pumparounds and heater/coolers.
By default, the entire feed is added directly to the liquid portion of the feed tray. Using the
SEPARATE key word will flash the feed adiabatically and any vapor if exist will be fed to the tray
directly above; this option simulates a typical feed nozzle projecting into the vapor space.

Before we go through the key words, here is a sample COLUMN input from Section 71:
COLUMN UID=DI S3, NAME=MOONSHI NE
PARAMETER TRAY = 21 $use I O, 21 t heor et i cal st ages
$
FEED F1, 17, SEPARATE $f l ashi ng t he f eed at st age 17
PROD OVHD = D3, 153, BTMS=B3 $2 pr oduct s; t op est i mat ed at 153
$
COND TYPE=BUBB, PRES=65 $t ot al condenser wi t h bubbl e pt . l i q.
PRESSURE 2, 70/ 21, 75 $def i nes pr essur e i n r est of col umn
DUTY 1, 1/ 2, 21 $def i nes l ocat i on of heat er / cool er
VARY DUTY=1, 2 $l et PRO/ I I var y t he heat dut i es
Column http://chemelab.ucsd.edu/CAPE/keyword/column.html
1 of 6 25-04-2014 12:27
TEFF( MURPHREE) 2, 0. 9/ 20, 0. 9 $90%Mur phr ee t r ay ef f i ci ency
$
PRINT PROP=BRI EF, RECOVERY $pr i nt opt i ons
ESTIMATE MODEL=CONV $l et PRO/ I I make i ni t i al guesses
TSIZE SECTION( 1) =2, 20, SIEVE $si ze t he col umn
$
SPEC STREAM=D3, COMP=5, 6, RATE, VALUE=0. 3
SPEC STREAM=B3, RVP( API N) , VALUE=12
$t wo col umn specs; t he f i r st speci f i es t he f l ow r at es of component s 5 pl us
6 at t he
$t op; t he second speci f i es t he Rei d vapor pr essur e at t he bot t om
Now, we'll try to go through the key words with special focus on the IO method. These notes will get you
started, but you really have to read the Input Manual to see all the other fancy options and items specific
to CHEMDIST and SURE. Section 71 of the Manual covers features general to all methods. Sections 72 to
74 contain information specific to each method.

COLUMN Just like all other modules, define the 4 alphanumeric
UID and the optional 12 alphanumeric NAME.
PARAMETER Defines the tray number, the calculation method and
the maximum number of iterations. Examples:
PARA TRAY=32, IO=20, DAMP=0. 7
PARA TRAY=32, CHEMDIST
PARA TRAY=32, SURE
You have to get the tray number from a shortcut
calculation first. Count the condenser and the
reboiler in the TRAY number.
For IO, the default number of iterations is 15.
The example above reset it to be 20. When
convergence appears to be difficult, say for
heavy-ends refinery fractionators and non-ideal
mixtures, lower the relaxation factor DAMP to
0.7 or 0.8 (default is 1). You can also set
ERRINC slightly larger than unity to allow
inner loop errors to increase during iterations.
For CHEMDIST, the default number of
iterations is 20. Calculation will stop if there is
no improvement after 5 successive iterations
(STOP=5); it can be reset with STOP=value.
Use of DAMP is same as IO, but the default
value for ERRINC is 100.
For SURE, the default number of iterations is
10. Calculation will stop if there is no
improvement after 5 successive iterations
(STOP=5). DAMP and ERRINC do not apply
here.
FEED Identify the feed streams and the feed tray numbers.
Example:
FEED F1, 3/ F2, 6/ ...etc.
Adding the SEPARATE key word will flash each of
the feed stream adiabatically into its vapor and liquid
Column http://chemelab.ucsd.edu/CAPE/keyword/column.html
2 of 6 25-04-2014 12:27
phases.
PRODUCT Identify the product streams and the flow rates. You
must use the key words OVHD, BTMS, LDRAW and
VDRAW
Example with top and bottom products plus two liquid
sidedraws at stages 6 and 9 and a vapor sidedraw at
stage 4:
PROD OVHD=OV, 121, BTMS=BT, &
LDRAW=LD1, 6, 100/ LD2, 9, 120, &
VDRAW=VD1, 4, 60
Like SHORTCUT, you must provide the estimated
flow rates to all but one product streams. Output
usually will try to stick to the molar flow rates given
unless they are specified as variables.
CHEMDIST uses different key words for the
sidedraws. Check the Manual if you have to use them.
CONDENSER Just like the SHORTCUT module
COND TYPE=PART, MI XED, BUBB, TFI X, or DTBB
If the PRESSURE statement does not specify the
pressure of tray 1, make sure you specify the
condenser pressure here.
DUTY Identifies the number and the location of condenser,
reboiler and side heaters or coolers. You must provide
the heat duty if it is not going to be varied by the
VARY statement. The default dimension is millions of
energy unit per time unit. Example of a 30-plate
column with a condenser, a side cooler (ve duty)
and a reboiler:
DUTY 1, 1/ 2, 12, 6/ 3, 32
VARY Define the variables. For simple distillation columns,
we vary the heat duties:
VARY DUTY=1, 32
For say a gas absorber, we vary the liquid feed rate:
VARY FEED=2 $say we' ve def i ned f eed number
2 as l i qui d
PRESSURE
PSPEC
The column pressure profile must be defined. If only
the pressures at the top and bottom are given, a linear
profile is assumed for the plates in between. The
follow forms are equivalent:
COND TYPE=. . . , PRES=65
PRESSURE 2, 70/ 21, 75
or
Column http://chemelab.ucsd.edu/CAPE/keyword/column.html
3 of 6 25-04-2014 12:27
COND TYPE=. . .
PRESSURE 1, 65/ 2, 70/ 21, 75
or
COND TYPE=. . . , PRES=65 $PTOP i s f or t r ay
no. 2
PSPEC PTOP=70, DPCOL=5 $pr essur e dr op
acr oss col .
or
COND TYPE=. . . , PRES=65 $pr essur e dr op per
t r ay
PSPEC PTOP=70, DPTRAY=0. 26 $i s
5/ ( 212) =0. 263
ESTIMATE All column calculations are inherently iterative.
ESTIMATE is an initial estimate generator for the
temperature and vapor flow profiles. A Fenske
calculation will be used under the option
MODEL=CONV. If the column is difficult to
converge, use MODEL=CHEM; a multiflash
technique will bring the profiles closer to the final
solution before the column algorithm takes over.
(Chapter 15 of Henley & Seader uses linear profiles;
MODEL=SIMPLE)
The default reflux ratio is 3. So if you know of a better
value, you can specific it here. Example:
ESTIMATE MODEL=CHEM, RRATIO=1. 1
You can provide other estimates or your own profiles,
but they are generally not necessary.
SPECIFICATION
The number of SPECIFICATIONS must match the
number of variables in the VARY statement. Just like
the shortcut, purity or recovery specifications under
most circumstances (See Manual for all the other
options if you need to). You can also specify the reflux
ratio or the heat duty. Example:
SPEC RRATIO, VALUE=2. 0
SPEC DUTY( 2) , RATIO, DUTY, FLASH=F1,
VALUE=1
VARY DUTY=2
$Thi s speci f i es t he heat dut y 2 ( say
r eboi l er ) as mi nus t he val ue of t he
$dut y of FLASH dr umF1. DUTY 2 must be a
var i abl e i n t hi s case.
Other options:
TEFF Tray efficiency. The example below provides the
simplest scenario which is a constant Murphree
efficiency for all the plates in the column.
Column http://chemelab.ucsd.edu/CAPE/keyword/column.html
4 of 6 25-04-2014 12:27
TEFF( MURPHREE) 2, 0. 9/ 20, 0. 9 $say 21 i s
r eboi l er
PRINT Print options to get extra information or to help figure
why column does not converge. Options:
$pr i nt s i ni t i al est i mat es and br i ef
i t er at i on r epor t
ITER=ESTI
$pr i nt s r epor t of most usef ul col umn
pr of i l es
PROP=BRI EF
$pr i nt s t he per cent r ecover y of each
component i n
$each pr oduct st r eam
RECOVERY
$pr i nt s t r ay by t r ay composi t i on r epor t by
M or WT
COMP=M
$pr i nt s t r ay by t r ay K- val ues of al l
component s
KVALUE
TSIZE PRO/II can do tray sizing and rating for trayed
columns, and HETP and pressure drop calculations for
packed columns. We'll only provide the simplest
example for tray sizing, TSIZE, i.e., the topic covered
by Chapter 13 in Henley & Seader.
Example: We have a column with TRAY=21. From
trays 2 to 9, we want valve trays spaced 20 inches
apart. From trays 10 to 20, we want sieve trays,
default spacing (which is 24") and a system factor of
0.8 to account for foaming (Default is SF=1).
TSIZE SECTION( 1) =2, 9, VALVE,
SPACE( TRAY, I N) =20
TSIZE SECTION( 1) =10, 20, SIEVE, SF=0. 8
Choices of trays are VALVE, SIEVE and CAP.
You can define the number of flow paths with
PASSES=1, 2 or 4. The default is 1.
The default spacing is 24 inches.
The system factor to account for foaming is
tabulated in Section 77. Or refer to Henley &
Seader or Perry's.
By default, sizing is done after convergence.
Using the DPCALC=value key word will force
the calculations to be performed during
flowsheet solution. This enables the pressure
drop to be reflected in the column solution and
gives CONTROLLER access to the flooding
factor.
The value is used to adjust for tray efficiency.
For example, a column is modeled as having 8
theoretical trays but in fact has 10 actual trays.
Set DPCALC=0.8.
Column http://chemelab.ucsd.edu/CAPE/keyword/column.html
5 of 6 25-04-2014 12:27
You can set the flooding factors with FF=value.
The default is in Section 77: 70% for tray
diameters 0-2 ft, 75% for trays 2-4 ft, 78% for
trays 4-10 ft and 80% for trays 10-50 ft.
Minimum column diameter DMIN is 15 inches.
DEFINE Other than to replace SPEC, column parameters like
DUTY, PTOP, or PCOND can be defined. (See Section
71 for details.)

Column http://chemelab.ucsd.edu/CAPE/keyword/column.html
6 of 6 25-04-2014 12:27

You might also like