See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232390117

Optimization of crude distillation system using aspen plus: Effect of binary

feed selection on grass-root design

Article  in  Chemical Engineering Research and Design · February 2010

DOI: 10.1016/j.cherd.2009.08.004

CITATIONS READS

50 812

4 authors, including:

Vijaya Kumar Bulasara

C. V. Raman College of Engineering
35 PUBLICATIONS   343 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Vijaya Kumar Bulasara on 02 April 2019.

The user has requested enhancement of the downloaded file.

 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Optimization of crude distillation system using aspen plus:

Effect of binary feed selection on grass-root design

Raja Kumar More, Vijaya Kumar Bulasara, Ramgopal Uppaluri ∗ , Vikas R. Banjara
Department of Chemical Engineering, Indian Institute of Technology Guwahati, Guwahati 781039, India

a b s t r a c t

With an objective to supplement guidelines available as general rules of thumb for the grass-root design of crude
distillation unit (CDU) using binary crude mixtures, this work presents the optimization of crude distillation unit
using commercial Aspen Plus software. The crude distillation unit constituted a pre-flash tower (PF), an atmospheric
distillation unit (ADU) and a vacuum distillation unit (VDU). Optimization model constituted a rigorous simulation
model supplemented with suitable objective functions with and without product flow rate constraints. Three dif-
ferent feed stocks namely Bombay crude, Araby crude and Nigeria crude were considered in this work along with
various binary combinations of these crudes. The objective function considered was profit function (subjected to
maximization) for cases without product flow rate constraints and raw-materials and energy cost (subjected to mini-
mization) for cases with product flow rate constraints. Parametric study pertaining to feed selection and composition
has been conducted in this work to further benefit refinery planning and scheduling. Simulation study inferred that
the product flow rate constraints sensitively affect atmospheric distillation column diameter and crude feed flow
rate calculations. Based on all simulation studies, a generalized inference confirms that it is difficult to judge upon
the quality of the solutions obtained as far as their global optimality is concerned.
Crown Copyright © 2009 Published by Elsevier B.V. on behalf of The Institution of Chemical Engineers. All rights
reserved.

Keywords: Crude distillation; Pre-flash; Vacuum distillation; Aspen Plus; Product flow rate; Feed selection; Grass-root
design; Optimization

1. Introduction constraints, a feature that is more prominent in the energy

economy past globalization era.
Distillation of crude oil is regarded to be the most funda- To date, the analysis of crude distillation system has been
mental process in the petroleum refining and petrochemical presented by various academic contributions. Research in the
industries. A crude distillation unit (CDU) consists of an analysis of crude distillation system is dovetailed towards
optional pre-flash tower (PF) followed by atmospheric distil- three major areas namely heat exchanger networks associated
lation unit (ADU) and vacuum distillation unit (VDU). Typical to crude distillation unit (Klemeš and Ptácník, 1985; Sunden,
products from crude distillation system include light ends, 1988; Nilsson and Sundén, 1994; Plesu et al., 2003; Gadalla et
light, heavy and medium naphtha, kerosene, diesel, atmo- al., 2003) refinery planning and scheduling (Cao and Xin, Cao et
spheric gas oil (AGO), light vacuum gas oil (LVGO), heavy al., 2009; Rivero et al., 2004; Aires et al., 2004; Dave and Zhang,
vacuum gas oil (HVGO) and vacuum residue. Analyzing the 2003; Göthe-Lundgren et al., 2002; Pinto et al., 2000) and crude
performance of crude distillation system is beneficial to simul- distillation modeling, simulation and optimization (Inamdar
taneously achieve higher process efficiency and lower process et al., 2004; Liau et al., 2004; Dave et al., 2003; Kumar et al.,
cost. The industrial operation of a crude distillation system is 2001; Seo et al., 2008).
subjected to perform under various operational factors such as The optimization of crude distillation unit has been favor-
type and quality of feed stock. Many a times, refineries involve ably addressed by Inamdar et al. (2004), Liau et al. (2004), Seo
blending of crudes due to operational and feed availability et al. (2008). While Inamdar et al. (2004) reported to the devel-

∗
Corresponding author. Tel.: +91 361 2582260; fax: +91 361 2582260.
E-mail addresses: b.vijaya@iitg.ernet.in (V.K. Bulasara), ramgopalu@iitg.ernet.in (R. Uppaluri).
Received 16 January 2009; Received in revised form 28 July 2009; Accepted 9 August 2009
0263-8762/$ – see front matter Crown Copyright © 2009 Published by Elsevier B.V. on behalf of The Institution of Chemical Engineers. All rights reserved.
doi:10.1016/j.cherd.2009.08.004
122 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134

Nomenclature
Qf,p pre-flash furnace duty (kJ/a)
A annualization factor for process equipments Qf,v vacuum column furnace duty (kJ/a)
(a−1 ) Qp1,a ADU pump-1 around duty (kJ/a)
Ac heat transfer area of the condenser (m2 ) Qp1,v VDU pump-1 around duty (kJ/a)
Af heat transfer area of the furnace (m2 ) Qp2,a ADU pump-2 around duty (kJ/a)
Ca,e annualized cost of energy ($/a) Qp2,v VDU pump-2 around duty (kJ/a)
Ca,p annualized cost of products ($/a) r rate of annual interest
Ca,r annualized cost of all raw-materials ($/a) R annualized raw-materials and energy cost ($/a)
Cc cost of crude ($/m3 ) Tc differential log mean temperature difference
CD product value of diesel & atmospheric gas oil for sizing condenser (K)
($/m3 ) Tf differential log mean temperature difference
Ce energy cost ($/kJ) for sizing furnace (K)
Cf fixed cost of all process equipments ($) Uc overall heat transfer coefficient for the con-
Cf,a fixed cost of atmospheric distillation column ($) denser (kW m−2 K−1 )
Cf,ac fixed cost of condenser in atmospheric distilla- Uf overall heat transfer coefficient for the furnace
tion column ($) (kW m−2 K−1 )
Cf,af fixed cost of atmospheric distillation column
furnace ($)
Cf,c fixed cost of the condenser ($)
opment of rigorous mathematical model coupled with genetic
Cf,d fixed cost of the distillation column ($)
algorithm for the optimization of crude distillation unit, Liau
Cf,f fixed cost of the furnace ($)
et al. (2004) developed artificial neural networks that can be
Cf,p fixed cost of pre-flash column ($)
used for simulation and optimization studies. It is further
Cf,pc fixed cost of pre-flash condenser ($)
interesting to note that Inamdar et al. (2004) reported better
Cf,pf fixed cost of pre-flash process furnace ($)
optimal conditions using multi-objective genetic algorithms
Cf,v fixed cost of vacuum column ($)
than those reported using local optimization algorithms. Sim-
Cf,vf fixed cost of vacuum unit furnace ($)
ilarly, Seo et al. (2008) addressed the design optimization of
CK product value of kerosene ($/m3 )
crude distillation column using mixed integer non-linear pro-
CN product value of naphtha and heavy naphtha
gramming method and have inferred better objective function
($/m3 )
values and hence a reduction in energy costs for an existing
CR product value of residue ($/m3 )
CDU system in a typical refinery.
Cs cost of steam ($/kg)
A critical observation of the literatures cited above infers
CV product value of light vacuum gas oil, heavy
that the inherent relationships between design variables (such
vacuum gas oil ($/m3 )
as column diameters, steam flow rates, crude feed flow rates)
D diameter of the distillation column (m)
and cost/economics have not been studied for binary feed
Fc crude flow rate (m3 /a)
systems. Traditional grass-root design rule of thumb for CDU
Fs,a steam flow rate into the atmospheric crude dis-
systems involves the simpler translation of crude assay data
tillation unit (kg/a)
into suitable product distributions and thereby the evalua-
Fs,p steam flow rate into the pre-flash tower (kg/a)
tion of the associated refinery profit margins and trade-offs.
Fs,s1 steam flow rate into stripper-1 (kg/a)
This however does not translate into trends associated in
Fs,s2 steam flow rate into stripper-2 (kg/a)
the performance characteristics. In addition, it is opined that
Fs,s3 steam flow rate into stripper-3 (kg/a)
the performance characteristics of binary crude feed CDU
Fs,v steam flow rate into the vacuum distillation
systems can be extrapolated from the performance character-
unit (kg/a)
istics of single crude feed CDU system using volume fractions
G annualized gross profit ($/a)
of the feeds as the basis. However, the impact of binary crude
L length of the distillation column (m)
composition upon several issues such as optimality of col-
n years of operation (a)
umn diameters, steam flow rates, cost and economics cannot
N number of stages in the distillation column
be extrapolated due to associated non-linearities in mod-
PA AGO flow rate (m3 /a)
eling expressions. Therefore, a thorough investigation upon
PD diesel flow rate (m3 /a)
the effect of binary feed selection on grass-root design be
PH HVGO flow rate (m3 /a)
addressed and is the objective of this article. Such studies are
PHN heavy naphtha flow rate (m3 /a)
also regarded to be beneficial for refinery engineers engaged
PK kerosene flow rate (m3 /a)
in refinery operations, planning and scheduling in the midst
PL LVGO flow rate (m3 /a)
of tighter crude supply–demand constraints.
PN naphtha flow rate (m3 /a)
This work addresses optimization studies for crude distilla-
PR residue flow rate (m3 /a)
tion system using Aspen Plus, a commercial process modeling
Qc condenser duty (kW)
software. The methodology adopted in this work involves
Qc,a ADU condenser heat duty (kJ/a)
design optimization of a chosen CDU system configuration
Qc,p pre-flash condenser heat duty (kJ/a)
with specified values for number of trays (and hence column
Qf furnace duty (kW)
height) in various distillation columns using SQP optimiza-
Qf,a atmospheric distillation column furnace duty
tion solver built in the software. In due course of optimization,
(kJ/a)
column diameters, crude feed flow rate and steam flow rates
are regarded as the key optimization variables that influence
 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134 123

Fig. 1 – True boiling point (TBP) curves for (a) Bombay, (b) Araby and (c) Nigeria crudes.
124 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134

Fig. 2 – Process Block Diagram for CDU simulation and optimization.

the performance characteristics and optimality of the binary unit and vacuum distillation unit. The feed streams to these
crude fed CDU systems. Several finer objectives of this article columns have been subjected to undergo heating via process
are addressed as follows: furnaces. Based on the available literature data (Kumar et al.,
2001; Stojic et al., 2004; William, 2006), Table 1 presents the
(a) Impact of crude selection on refinery profits and CDU opti- summary of various operating parameters for the CDU system
mality including feed and steam flow rates to the CDU represented in Aspen Plus. A brief description of the process
system. configuration is presented below.
(b) Effect of binary crude compositions on refinery profit mar-
gins as well as CDU design variables such as steam and 2.2.1. Pre-flash tower
crude feed flow rates and diameters of various columns. Crude stream sent to pre-flash furnace undergoes partial
(c) Trade-offs associated to raw-materials and energy cost vaporization. The furnace operating pressure and tempera-
and profit margins of the CDU system for various choices ture are taken as 345 kPa and 232 ◦ C. The pre-flash tower
of feeds. constitutes 10 theoretical stages (chosen for the study) with
(d) Impact of product flow rate constraints on the design opti- an average tower pressure drop of 21 kPa and steam stripping
mality and performance characteristics of the CDU. is considered using steam at 204 ◦ C and 414 kPa. The pre-flash
column produces light ends along with naphtha product that
The next section outlines the representation of the CDU is removed as the top product using a partial condenser that
system problem in Aspen Plus. operates at 77 ◦ C and 274 kPa with a pressure drop of 14 kPa.
The product specification desired corresponds to naphtha cut
2. Problem summary with ASTM 95% temperature of 191 ◦ C.

2.2.2. Atmospheric distillation unit

2.1. Crude assay
The bottom product of the pre-flash tower enters the process
furnace and undergoes a further vaporization of about 3% by
Three crude assays namely BOMBAYHG (Bombay crude),
volume. The pressure of the furnace is maintained at 167 kPa.
ARABY (Mideast Crude) and BONNYLT (Nigeria crude) are con-
The atmospheric distillation column constitutes 25 theoretical
sidered in this work as different choices of single and binary
crude feed choice to the CDU system. Fig. 1a–c presents the
true boiling point (TBP) curves of these crudes, respectively. Table 1 – Parameters for CDU optimization problem.
As shown in these figures, naphtha range corresponds to
Parameter Value
34.2 vol% in Bombay crude, 20 vol% in the Mideast crude and
32 vol% in Nigeria crude, respectively. A critical analysis of Pre-flash unit
these crudes indicated that Bombay and Nigeria crudes have No. of stages 10
similar TBP profiles where as Araby crude possesses more high Pressure 273.7 kPa (top stage)
boiling components. Based on these TBP curves, the mixed 308.2 kPa (bottom stage)
crude TBP curves for various volume fractions of different ADU
single crudes can be as well generated. No. of stages 25

Pressures 108.2 kPa (top stage)

2.2. Process configuration 170.3 kPa (bottom stage)

VDU
Fig. 2 presents the process block diagram for the crude distilla-
No. of stages 6
tion system represented in Aspen Plus commercial software. A
typical representation of the crude distillation system consti- Pressures 8.0 kPa (top stage)
9.3 kPa (bottom stage)
tutes a pre-flash unit followed by the atmospheric distillation
 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134 125

stages (chosen for the study) and consists of a total condenser,

Table 2 – Unit prices of various commodities (WPS, 2009).
three coupled side strippers and two pump-around circuits.
The feed to the atmospheric distillation column enters on Price Units
stage 22 (stage numbering is from top to bottom). Steam strip- Cc 628.97 $/m3
ping that aids enhancement of volatility via the reduction of Cs 0.0055 $/kg
partial pressure is facilitated by using steam at 204 ◦ C and CN 1003.84 $/m3
414 kPa. The average pressure drop of the atmospheric dis- CK 861.06 $/m3
CV 718.28 $/m3
tillation tower is taken as 28 kPa with the first stage at 108 kPa,
CR 449.08 $/m3
and the bottom stage pressure of 170 kPa. Two pump-around Ce 4.74 × 10−6 $/kJ
circuits are facilitated to provide internal reflux at various sec-
tions of the tower. The location of the first pump around is
those with the specification of maximum product flow rate
from stage 8 to stage 6 with a base case heat duty of −11.7 MW.
inequality constraints.
The second pump around is located from stage 14 to stage
13 with a base case heat duty of −4.4 MW. The condenser is
3.1. Parameters
operated at 108 kPa and a pressure drop of 34 kPa.
The first side stripper consisting of 4 equilibrium stages is
All design specifications mentioned in Table 1 along with those
fed with liquid drawn from stage 6 of the main column. The
presented in Table 2 (WPS, 2009) are regarded as design param-
lighter product produced in the first side stripper is fed to stage
eters of the mathematical formulation. These include number
5 of the main column. The second side stripper consisting of 3
of theoretical stages, column pressures and column height as
equilibrium stages produces diesel product with feed fed from
well as cost parameters.
the 13th stage of the main column and the vapor product fed
to stage 12 of the main column. The third stripper consisting
3.2. Variables
of 2 equilibrium stages produces atmospheric gas oil as the
bottom product and is fed with liquid drawn from stage 18
All cases corresponding to the negation of product flow rate
of the main column. The vapor from the third stripper is fed
inequality constraints involve maximum number of variables
to stage 17 of the main column. The residue product is fur-
that need to be optimized simultaneously. For these cases, the
ther subjected to distillation in the vacuum distillation unit.
variables that are subjected to optimization include crude feed
Various products produced from the atmospheric distillation
flow rate, steam flow rates associated to atmospheric column,
column supplemented with the three side strippers include
side strippers associated to atmospheric column, pre-flash
naphtha, kerosene, diesel, atmospheric gas oil and the bottom
and vacuum distillation column, product flow rates of light
product.
naphtha, heavy naphtha, kerosene, AGO, LVGO, HVGO and
residue products, column reflux ratios, condenser duties, col-
2.2.3. Vacuum distillation unit
umn diameters, furnace duties, pump around duties, furnace
The bottom product of the atmospheric distillation column
and condenser areas.
is fed to the vacuum distillation which is facilitated with jet
Those cases that correspond to the specification of product
ejectors that enable the generation of vacuum in the unit. The
flow rate inequality constraints in the mathematical model
vacuum unit separates the atmospheric column bottom prod-
involve the optimization of variables mentioned above except
uct into off-gas, light vacuum gas oil, heavy vacuum gas oil
the product flow rates.
and residual oil. The vacuum column constitutes 6 theoreti-
cal stages (chosen for the study) with pressure maintained at
3.3. Inequality constraints
8 kPa (stage 1) and 9.3 kPa (stage 6). The process furnace asso-
ciated to the vacuum column operates at a pressure of 14 kPa
The upper and lower bounds for variables such as steam flow
and provides an overflash of 0.6 vol%. The vacuum tower is
rates associated to pre-flash, ADU, strippers S1-3 and VDU
enabled with two pump-around circuits, with the first located
and crude feed flow rate have been specified between a valid
from stages 2 to 1 and the second located from stage 4 to 3.
lower and upper bound. All other variables have been treated
Light vacuum gas oil is taken out from stage 2 as a total draw.
as non-negative variables. Further, for all cases with prod-
The heavy vacuum gas oil is withdrawn from stage 4 and bitu-
uct flow rate constraints, upper bounds of naphtha, heavy
men is taken as bottom product of the vacuum distillation
naphtha, kerosene and diesel products have been specified
unit.
that impose additional inequality constraints. Table 3 presents
all inequality constraints corresponding to all cases of opti-
3. Mathematical formulation
mization considered in this work. Table 4 summarizes upper
bounds of product flow rates that have been additionally
This section illustrates various components of the optimiza-
tion model formulated in Aspen Plus environment to evaluate
the impact of the feed choice on the grass-root design. A Table 3 – Inequality constraints for the CDU optimization
typical optimization model consists of an objective function problem.
supplemented with equality and inequality constraints. In the Variable Lower bound Upper bound
present case study, the equality constraints relate to the mass
Crude feed flow rate (m3 /d) 11924.25 31798.00
balances facilitated by the rigorous process simulation mod- ADU steam (kg/h) 3175.13 9071.80
els built in Aspen Plus. The inequality constraints are imposed Stripper-1 steam (kg/h) 907.18 2721.54
in the form of upper bounds for product flow rates for differ- Stripper-2 steam (kg/h) 226.80 1360.77
ent cases. Broad grass-root design studies were carried out for Stripper-3 steam (kg/h) 226.80 680.39
two different classes namely those without the specification VDU steam (kg/h) 6803.85 13607.70
PF steam (kg/h) 1360.77 3628.72
of maximum product flow rates as inequality constraints and
126 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134

Table 4 – Upper bounds for the flow-rates of different products for various single and binary crude choices.
Units Bombay Araby Nigeria Araby-Bombay Araby-Nigeria Bombay-Nigeria
3
Naphtha m /d 4769.70 3179.80 4769.70 4769.70 4769.70 5246.67
H. naphtha m3 /d 1271.92 1907.88 1271.92 1271.92 1271.92 1271.92
Kerosene m3 /d 3974.75 2305.36 3974.75 3020.81 3974.75 3179.80
Diesel m3 /d 3179.80 2225.86 3179.80 2384.85 3179.80 3179.80
AGO m3 /d 1271.92 1271.92 1271.92 1271.92 1271.92 1271.92
LVGO m3 /d 3179.80 1589.90 3179.80 3179.80 3179.80 3179.80
HVGO m3 /d 2384.85 2384.85 2384.85 2702.83 2702.83 2702.83

The evaluation of objective functions for both classes of

Table 5 – Design parameters/variables and expressions
for the evaluation of fixed costs of various process optimization problems was based on the following expres-
equipments (William, 2006). sions:

Parameter/variable Value/expression
Ca,r = Cc Fc + Cs (Fs,p + Fs,a + Fs,s1 + Fs,s2 + Fs,s3 + Fs,v ) (1)
Condensers
Uc (kW m−2 K−1 ) 0.852
Tc (K) 13.9 Ca,e = Ce (Qc,p + Qc,a + Qp1,a + Qp2,a + Qp1,v + Qp2,v + Qf,p
Ac (m2 ) Qc /(Uc Tc )
Cf,c ($) 7296(Ac )0.65 + Qf,a + Qf,v ) (2)
Furnaces
Uf (kW m−2 K−1 ) 0.568
Tf (K) 34.8
Af (m2 ) Qf /(Uf Tf ) Ca,p = CN (PN + PHN ) + CK PK + CD (PD + PA ) + CV (PL + PH ) + CR PR
Cf,f ($) 7296(Af )0.65
(3)
Distillation columns
L (m) 0.73N
Cf,d ($) 17640D1.066 L0.802

Cf = Cf,p + Cf,a + Cf,v + Cf,pc + Cf,ac + Cf,pf + Cf,af + Cf,vf (4)

imposed for all cases where product flow rate constraints were Here the fixed costs of columns (pre-flash, atmospheric
considered. distillation column and vacuum column), condensers (pre-
flash and atmospheric distillation column condensers)
3.4. Objective function and furnaces (associated to pre-flash, atmospheric distil-
lation column and vacuum column) were evaluated using
For all cases without product flow rate constraints, the objec- parameters/variables and design expressions (William, 2006)
tive function (annualized profit) to be maximized is evaluated summarized in Table 5.
as a function of annualized values of the products cost, feed For all cases without product flow rate constraints, gross
cost, capital and energy costs. For those cases where prod- profit function was evaluated using the following expression
uct flow rate constraints were further imposed, the objective and is taken as the objective function for maximization:
function (raw-materials and energy cost) to be minimized was
evaluated as a function of feed and energy costs. G = Ca,p − Ca,r − Ca,e − Cf A (5)

Table 6 – Optimal product flow rates and costs for the case study with single crude and without product flow rate
constraints.
Variable Units Araby Bombay Nigeria
3
Crude m /d 22067.00 23848.50 23848.50
ADU steam kg/h 8493.52 5511.80 5511.80
ADU S-1 steam kg/h 2541.96 1527.74 1527.74
ADU S-2 steam kg/h 1246.06 226.80 226.80
ADU S-3 steam kg/h 635.07 426.42 426.42
PF steam kg/h 3628.72 3628.72 3628.72
VDU steam kg/h 9071.80 9071.80 9071.80
Naphtha m3 /d 3179.80 5340.51 5340.70
H. naphtha m3 /d 1911.20 1359.84 1359.70
Kerosene m3 /d 2337.22 4579.83 4579.12
Diesel m3 /d 2226.15 3563.89 3562.22
AGO m3 /d 1351.45 1351.42 1351.42
LVGO m3 /d 1625.78 3994.31 3994.88
HVGO m3 /d 2702.83 2702.83 2702.83
Residue m3 /d 6442.26 631.08 632.81
Total feed cost/annum M$/a 5066.80 5475.70 5475.70
Income/annum M$/a 5303.80 6384.90 6384.70
Total annual cost M$/a 10.60 14.50 15.60
Total annual profit (objective function) M$/a 226.40 894.70 893.40
 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134 127

Table 7 – Optimal column variables for the case study with single crude and without product flow rate constraints.
Nigeria Bombay Araby

Pre-flash ADU VDU Preflash ADU VDU Preflash ADU VDU

Reflux ratio 0.4 6.8 1.3 0.4 3.3 0.9 0.4 6.8 1.2
Condenser heat duty (MW) 35.4 47.2 0.0 34.0 44.3 0.0 20.2 20.4 0.0
Column diameter (m) 4.7 29.9 15.1 4.0 17.7 13.0 4.7 29.9 15.0
Furnace heat duty (MW) 93.1 71.9 32.9 90.1 72.8 32.0 43.3 53.5 22.6
Pump-1 around duty (MW) – 11.7 18.9 – 11.7 19.8 – 11.7 3.6
Pump-2 around duty (MW) – 4.4 23.4 – 4.4 23.4 – 4.4 19.0
Condenser area (m2 ) 2986.3 3981.0 0.0 2862.7 3736.0 0.0 1699.5 1716.8 0.0
Furnace area (m2 ) 4690.2 3624.8 1657.9 4542.4 3671.5 1613.2 2184.0 2696.4 1140.4
Heat exchangers cost (M$) 3.10 3.10 0.90 3.03 3.05 0.89 2.00 2.16 0.71
Column vessel cap cost (M$) 0.45 6.78 1.04 0.38 3.89 0.89 0.45 6.78 1.04
Energy cost (M$) 10.69 10.49 6.27
Capital cost (M$) 15.38 12.13 13.14

Here “A” refers to annualization factor for process equipment

which is given by:
n
r (1 + r)
A = n (6)
(1 + r) − 1

For ten years of operation and at an annual interest of

8%, the annualization factor would be 0.15 a−1 . However,
in this work the annualization factor has been chosen as
0.33 a−1 . This is due to two reasons. Firstly, higher annualiza-
tion factor enables consideration of all auxiliary processes and
equipments associated to the CDU system and hence allows
estimation of cost values that are close to those existing in
the reality. Secondly, since feed flow rate is chosen as an opti-
mization variable, there is a tendency to enhance feed flow
rates and hence column diameters to maximize profit. There-
fore, larger column diameters are difficult to realize in a single
Fig. 3 – Variation of CDU total annual cost and profit for
column and the column diameter needs to be treated as an
various single crudes without product flow rate constraints.
equivalent diameter. Always, it is well known that the cost
of two columns is higher than that of a single column with
same equivalent diameter and therefore the chosen higher
value of annualization factor is anticipated to estimate the
costs of more number of columns using the simple concept of
equivalent diameter and higher value of annualization factor.
For all cases where product flow rate constraints were
imposed, the overall raw-materials and energy cost was eval-
uated using the following expression is taken as the objective
function for minimization:

R = Ca,r + Ca,e (7)

It is important to note here that the raw-materials and

energy cost forces the feed flow rate to take an optimal (lowest
possible) value such that the product flow rate constraints are
satisfied and minimal overall energy is required.

3.5. Optimization methodology

The optimization of the CDU system has been conducted using

an in-built sequential quadratic programming (SQP) solver in
the Aspen Plus environment (version 12) (ATAP, 2009). SQP
method is a typical non-linear programming method that
requires a feasible set of initial values for all process variables
(Singiresu, 1998). Thereby, the SQP proceeds to yield optimal
process variables of the CDU optimization problem. Since var-
ious data sets are involved, different feasible sets of initial Fig. 4 – Contributions of various key cost functions to
values are also required. All initial feasible data sets have been optimized objective function for the case with Bombay
provided based on the expertise in process simulation. crude as feed stream.
128 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134

Table 8 – Optimal product flow rates and costs for the case study with binary crudes and without product flow rate
constraints.
Variable Units Bombay-Nigeria Araby-Nigeria Araby-Bombay

Optimal composition – 0.9 (Bombay) 0.9 (Nigeria) 0.2 (Araby)

Crude m3 /d 23848.50 23848.50 23848.50
ADU steam kg/h 3628.72 6181.75 2396.18
ADU S-1 steam kg/h 9071.80 2075.95 8526.90
ADU S-2 steam kg/h 2721.54 344.32 2601.57
ADU S-3 steam kg/h 226.80 471.10 1212.13
PF steam kg/h 680.39 3628.72 647.00
VDU steam kg/h 9071.80 9071.80 9071.80
Naphtha m3 /d 5366.06 5139.37 5240.49
H. naphtha m3 /d 1345.04 1360.33 1200.47
Kerosene m3 /d 4500.42 4493.28 4013.34
Diesel m3 /d 3685.61 3443.15 2684.05
AGO m3 /d 1351.42 1351.42 1351.42
LVGO m3 /d 3740.02 3751.80 4730.51
HVGO m3 /d 2702.83 2702.83 2702.83
Residue m3 /d 750.05 1279.44 1160.18
Total annual feed cost M$/a 5476.2 5476.1 5476.2
Income (annual) M$/a 6373.6 6319.1 6088.5
Total annual cost M$/a 14.2 (13.1 lowest) 14.2 (13.7 lowest) 16.3 (14.5 lowest)
Binary crude composition for minimum total cost – 0.4 (Bombay) 0.1 (Nigeria) 0.7–0.9 (Araby)
Total annual profit (objective function) M$/a 883.2 828.8 596.0

3.6. Cases considered flow rate constraints was found to be at least 50% lower than
those conducted with product flow rate constraints.
The following cases have been considered in this work:

4. Results and discussion

(a) Optimization without product flow rate specification con-
straints: For this case, the optimization of the CDU system
4.1. Optimization without product flow rate
has been considered for
constraints
a. Single crudes namely Bombay, Nigeria and Araby
crudes.
4.1.1. Performance of single crude fed CDU system
b. Binary crudes namely Bombay-Nigeria, Nigeria-Araby
Tables 6 and 7 summarize various optimized process variables
and Araby-Bombay with different volume % composi-
for the CDU system with different single crudes namely Bom-
tions (10–90%).
bay, Nigeria and Araby crudes. Based on the obtained results,
(b) Optimization with product flow rate specification con-
the following inferences have been made.
straints: For this case, the optimization of the CDU system
has been considered for
(a) Bombay crude provided highest optimal gross profit of
a. Single crudes namely Bombay, Nigeria and Araby crudes. 894.7 million United States Dollars (M$) as opposed to 893.4
b. Binary crudes namely Bombay-Nigeria, Nigeria-Araby and M$ for Nigeria crude and 227.4 M$ for Araby crude (Fig. 3).
Araby-Bombay with different volume % compositions (b) The contributions of various key cost functions to opti-
(10–90%). mized profit (Fig. 4) infers that only 2% was contributed
by equipment related costs and 98% was contributed by
The CPU time for all simulations was evaluated to be less process economy. The distribution between annualized
than 180 s on 2.8 GHz Intel Pentium D processor with 1 GB RAM. energy and capital costs relate to 32% and 68% of the total
As expected, the CPU time for all simulations without product annualized costs.

Table 9 – Optimal column variables for the case study with binary crudes and without product flow rate constraints.
Bombay-Nigeria Araby-Nigeria Araby-Bombay

Preflash ADU VDU Preflash ADU VDU Preflash ADU VDU

Reflux ratio 0.4 6.8 1.3 0.4 0.5 1.3 0.4 6.7 1.2
Condenser heat duty (MW) 34.3 48.5 0.0 32.7 45.7 0.0 33.2 44.1 0.0
Column diameter (m) 4.8 12.3 14.7 19.2 12.3 15.0 4.8 11.5 16.4
Furnace heat duty (MW) 90.5 70.7 31.7 89.7 74.1 32.2 91.8 71.6 41.4
Pump-1 around duty (MW) 0.0 11.7 18.5 0.0 11.7 18.7 0.0 11.7 24.5
Pump-2 around duty (MW) 0.0 4.4 23.4 0.0 4.4 23.4 0.0 4.4 24.9
Condenser area (m2 ) 2893.1 4089.2 0.0 2753.3 3851.3 0.0 2795.1 12685.3 0.0
Furnace area (m2 ) 4562.1 3563.2 1595.2 4524.0 3733.7 1621.7 4626.0 12322.5 7122.1
Heat exchangers cap cost (M$) 3.04 3.11 0.88 2.99 3.09 0.89 3.03 6.72 2.33
Column vessel cap cost (M$) 0.46 2.64 1.01 0.41 2.64 1.04 0.46 2.44 1.14
Energy Cost (M$) 10.52 10.49 10.96
Capital Cost (M$) 11.15 11.06 16.12
 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134 129

Table 10 – Optimal product flow rates and costs for the case study with single crude and with product flow rate
constraints.
Variable Units Araby Bombay Nigeria

Naphtha m3 /d 3179.8 5246.7 5340.5

H. naphtha m3 /d 1913.4 1276.9 1359.8
Kerosene m3 /d 2341.5 3973.4 4579.8
Diesel m3 /d 2225.7 1755.5 3563.9
AGO m3 /d 1351.4 1351.4 1351.4
LVGO m3 /d 1624.4 1351.4 3994.3
HVGO m3 /d 2702.8 4392.2 2702.8
Residue m3 /d 6437.5 2833.9 631.1
Crude m3 /d 22067.0 22181.3 23848.5
ADU steam kg/h 8463.1 5438.0 5511.8
ADU S-1 steam kg/h 2552.0 1496.4 1527.7
ADU S-2 steam kg/h 1247.8 453.5 226.8
ADU S-3 steam kg/h 638.0 362.8 426.4
PF steam kg/h 3628.7 1360.8 3628.7
VDU steam kg/h 9071.8 9071.8 9071.8
Feed cost (annual) M$/a 5067.2 5092.3 5476
Total income (annual) M$/a 5304.9 5860.0 6384.9
Total annual cost (objective function) M$/a 12.9 14.0 16.0
Annual profit M$/a 224.7 753.7 892.9

crude stream undergoes parallel processing in several

distillation columns. The obtained high values of distil-
lation column diameters from Aspen simulation appear
to be justified based on the literature values presented
by Gadalla et al. (2003), who indicated column diameters
varying between 5 and 8 m with a crude feed flow rate of
15,899 m3 /d for an atmospheric distillation column con-
stituting forty-one stages and three pump-around units.
Also, lower values of column diameters would have been
realized if the distillation column heights (or number of
trays) are taken more which facilitates additional degrees
of freedom in the complex columns. In either case, it is
important to note that the cost of the units would be appar-
ently similar due to the associated trade-offs between
column height and diameter. Therefore, considering these
Fig. 5 – Comparison of diameters of various units for
trade-offs, the cost implication upon obtained diameters
different single crudes.
is anticipated to provide a good conceptual understand-
ing upon the relationships between operational variables
(c) Optimal pre-flash unit diameters varied around 4–4.7 m, (such as flow rates) and economics (cost and profit).
and that of the atmospheric distillation column varied (d) The optimal feed flow rate to achieve maximum gross
between 17.7 and 29 m while for the vacuum distilla- profit was found to lie between 22,067 and 23,849 m3 /d
tion column they were about 13–15.1 m (Fig. 5). These (Table 6).
diameters as such are infeasible to achieve industri- (e) The optimal reflux ratios for different columns varied as
ally in a single column and are therefore referred to as follows: Pre-flash 0.4, atmospheric distillation unit 3.3–6.8
equivalent diameters for various columns assuming the and vacuum unit 0.9–1.3.

Table 11 – Optimal column variables for the case study with single crude and with product flow rate constraints.
Araby Bombay Nigeria

Preflash ADU VDU Preflash ADU VDU Preflash ADU VDU

Reflux ratio 0.4 7.0 1.1 0.4 0.6 0.9 0.4 6.5 1.4
Condenser heat duty (MW) 20.2 25.6 0.0 33.9 13.2 0.0 34.0 44.3 0.0
Column Dia (m) 4.3 10.1 13.1 4.7 32.5 14.0 4.7 29.2 15.1
Furnace heat duty (MW) 80.2 86.2 23.4 83.7 60.7 2.0 90.1 72.8 32.0
Pump-1 around duty (MW) 0.0 35.3 9.4 0.0 35.3 9.4 0.0 35.3 9.4
Pump-2 around duty (MW) 0.0 10.1 19.0 0.0 10.1 19.0 0.0 10.1 19.0
Condenser area (m2 ) 1699.5 2156.5 0.0 2861.0 3813.2 0.0 2862.7 3736.0 0.0
Furnace area (m2 ) 4041.9 4346.5 1179.0 4220.9 10440.2 346.3 4542.4 3671.5 1613.2
Heat exchangers cost (M$) 2.53 2.76 0.72 2.95 4.54 0.33 3.03 3.05 0.89
Column vessel cap cost (M$) 0.41 2.14 0.90 0.45 7.42 0.96 0.45 6.62 1.04
Energy cost (M$) 9.76 8.44 10.95
Capital cost (M$) 9.46 16.65 15.07
130 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134

Fig. 7 – Variation of total annual profit with binary crude

volume fractions for all cases without product flow rate
constraints (BN: Bombay crude in a mixture of
Bombay-Nigeria crudes; NA: Nigeria crude in a mixture of
Nigeria-Araby crudes; BA: Bombay crude in a mixture of
Bombay-Araby crudes).

(b) As illustrated in Fig. 6a, the grass-root designs of the

CDU for Nigeria-Bombay crudes adopted different equiv-
alent diameter profiles with respect to volume fractions
for various binary crude systems. Low pre-flash diameters
(4 m), medium range equivalent diameters for the atmo-
spheric column (about 11 m) and highest vacuum unit
equivalent diameters (16 m) are indicated by the simu-
lation study for Nigeria-Bombay crude system. However,
lower atmospheric column equivalent diameters followed
by vacuum unit diameters and then the pre-flash unit
equivalent diameters is indicated for the Araby-Nigeria
crude system (Fig. 6b). On the other hand, lower pre-
flash unit equivalent diameters followed by atmospheric
column diameters and then vacuum unit diameters is
observed for Araby-Bombay crude system (Fig. 6c). These
observations are analyzed to be consistent with the
binary crude-assay data generated for various binary sys-
tems. In addition, equivalent diameter sensitivity was
observed to be more predominant for Araby-Bombay and
Araby-Nigeria crude systems, both that feature lighter
Fig. 6 – Variation of diameter of different columns with
and heavier crude combinations. The Araby-Nigeria crude
volume fraction of (a) Nigeria crude in mixture of Nigeria
involved maximum equivalent diameters for the pre-flash
and Bombay crudes, (b) Nigeria crude in a mixture of
Nigeria and Araby and (c) Araby crude in a mixture of
Araby-Bombay crude for optimization studies conducted
without product flow rate constraints.

4.1.2. Performance of binary crude fed CDU system

Tables 8–9 and Figs. 6–8 summarize results obtained from the
simulation studies involving various compositions of binary
crude combinations from Nigeria, Bombay and Araby crudes.
Based on these results, the following inferences have been
drawn:

(a) As summarized in Table 8, the binary crude composition

of 0.9 volume fraction Bombay crude in Bombay-Nigeria
crude provided maximum gross annual profit of 883.2 M$
amongst all binary compositions studied, followed by 0.9 Fig. 8 – Variation of optimal crude feed flow rate with
volume fraction of Nigeria crude in Nigeria-Araby crude volume fraction of various crudes in binary crudes
(828.8 M$) and 0.2 volume fraction of Araby crude in Araby- corresponding to optimization cases without product flow
Bombay crude (596 M$). Incidentally, these optimal binary rate constraints (BN: Bombay crude in a mixture of
compositions do not refer to the lowest raw-materials and Bombay-Nigeria crudes; NA: Nigeria crude in a mixture of
energy cost of the CDU system as also indicated in the Nigeria-Araby crudes; BA: Bombay crude in a mixture of
table. Bombay-Araby crudes).
 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134 131

(e) The optimal feed flow rate varied between 21,425 and
23,849 m3 /d (Table 10).

4.2.2. Performance of binary crude fed CDU systems

Tables 12 and 13 along with Figs. 10 and 11 summarize results
obtained from the simulation studies of binary crude fed CDU
systems in Aspen Plus. Based on these results, the following
inferences have been drawn:

(a) As summarized in Table 12, the binary crude composition

of 0.2 volume fraction of Nigeria crude in Bombay-Nigeria
crude requires minimum raw-materials and energy cost
of 13.4 M$ followed by 0.2 Araby crude in Araby-Bombay
crude (13.5 M$) and 0.7 Nigeria crude in Araby-Nigeria
Fig. 9 – Variation of total annual cost and profit for single crude (15.3 M$). Incidentally, the maximum annual profit
crudes with product flow rate constraints. obtained for these cases corresponds to 865.9 M$ for
0.9 Nigeria crude in Bombay-Nigeria crude system fol-
unit being the heaviest crude combination (Fig. 6b), where
as higher vacuum column diameters profiles are observed
for Nigeria-Bombay and Araby-Bombay crude systems.
(c) The total annual gross profit was observed to vary linearly
with enhancement in volume fraction of the lighter crude
as presented in Fig. 7. The linear trend was attributed due
to lack of any product flow rate constraints that enabled
higher profit margins at higher volume fractions of the
lighter crudes in the chosen binary systems.
(d) Except for a few cases, almost all compositions indicated
optimal crude feed flow rate of 23,849 m3 /d (Fig. 8). This
indicates that a high crude feed flow rate enables maxi-
mum profits generated due to higher rates of producing
the petroleum products.
(e) The optimal reflux ratios for various optimal binary com-
positions varied as follows: pre-flash 0.4; atmospheric
distillation column 0.5–6.8 and vacuum column 1.2–1.3.

In summary, the single crudes have provided highest

annual profits compared to the binary crudes. This is asso-
ciated due to the heavier components contributed by heavier
crudes such as Araby crude.

4.2. Optimization with product flow rate constraints

4.2.1. Performance of single crude fed CDU system

Tables 10 and 11 along with Fig. 9 present the summary of
various optimized variables obtained from the optimization
of CDU system with product flow rate constraints. From these
results, the following inferences are highlighted.

(a) As illustrated in Fig. 9, Araby crude indicated the lowest

raw-materials and energy cost of 12.9 M$ followed by Bom-
bay crude (14 M$) and Nigeria crude (16 M$). However,
highest annual profit has been obtained for the case of
Nigeria crude (892.9 M$).
(b) Percentage contribution of various costs was similar to
those presented for single crude CDU fed systems without
product flow rate constraints.
(c) The optimal reflux ratios of various columns varied as fol-
lows: pre-flash 0.4; atmospheric distillation column 0.6–7; Fig. 10 – Variation of diameter of different columns with
vacuum column 0.9–1.4. volume fraction of (a) Nigeria crude in mixture of Nigeria
(d) Optimal pre-flash unit equivalent diameters varied around and Bombay crudes, (b) Nigeria crude in a mixture of
4.3–4.7 m, and that of the atmospheric distillation column Nigeria and Araby and (c) Araby crude in a mixture of
varied between 10.1 and 32.5 m and those of the vacuum Araby-Bombay crude for the case with product flow rate
column varied in the range of 13.1–15.1 m. constraints.
132 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134

Table 12 – Optimal product flow rates and costs for the case study with binary crudes and with product flow rate
constraints.
Variable Units Bombay-Nigeria Araby-Nigeria Araby-Bombay

Optimal composition – 0.2 (Nigeria) 0.7 (Nigeria) 0.2 (Araby)

Crude m3 /d 21732.3 23677.8 20800.9
ADU steam kg/h 7027.1 8479.5 5439.6
ADU S-1 steam kg/h 1496.8 2539.8 1496.6
ADU S-2 steam kg/h 453.5 1247.3 453.5
ADU S-3 steam kg/h 362.9 635.0 362.9
PF steam kg/h 3628.7 3628.7 3628.7
VDU steam kg/h 9071.8 9071.8 9071.8
Naphtha m3 /d 5326.2 4769.7 4769.7
H. naphtha m3 /d 1271.9 1271.9 1271.9
Kerosene m3 /d 3139.8 4330.6 3037.4
Diesel m3 /d 3179.8 3179.8 2384.9
AGO m3 /d 1351.4 1351.4 1351.4
LVGO m3 /d 4047.0 3405.0 3765.9
HVGO m3 /d 2751.6 2702.8 2702.8
Residue m3 /d 205.9 2358.7 882.9
Total annual feed cost M$/a 4990.2 5437 4776.3
Income (annual) M$/a 5741 6151.1 5353.4
Total annual cost (objective function) M$/a 13.4 15.3 13.5
Total annual profit M$/a 737.4 698.7 563.5
Maximum annual profit M$/a 865.9 797.8 563.5
Composition corresponding to maximum profit – 0.9 (Nigeria) 0.9 (Nigeria) 0.2 (Araby)
Total annual cost corresponding to above composition M$/a 16.1 16.0 13.5

Table 13 – Optimal column variables for the case study with binary crudes and with product flow rate constraints.
Bombay-Nigeria Araby-Nigeria Araby-Bombay

Preflash ADU VDU Preflash ADU VDU Preflash ADU VDU

Optimal composition 0.2 (Nigeria) 0.7 (Nigeria) 0.2 (Araby)

Reflux ratio 0.4 0.8 1.1 0.4 3.5 1.2 0.4 1.0 1.1
Condenser heat duty (MW) 34.8 15.2 0.0 30.1 29.5 0.0 31.2 15.7 0.0
Column diameter (m) 4.9 13.1 15.4 4.2 23.4 14.8 4.6 10.9 15.8
Furnace heat duty (MW) 84.5 64.7 33.8 88.4 75.4 34.2 80.0 66.3 35.6
Pump-1 around duty (MW) 0.0 24.4 21.2 0.0 27.8 17.3 0.0 36.3 20.1
Pump-2 around duty (MW) 0.0 15.2 22.0 0.0 19.0 23.4 0.0 14.4 24.9
Condenser area (m2 ) 2931.2 1282.3 0.0 2533.5 2487.5 0.0 2627.3 1321.8 0.0
Furnace area (m2 ) 4257.1 3258.6 1706.1 4453.9 3801.4 1724.7 4034.9 3341.6 1794.3
Heat exchangers cap. cost (M$) 2.98 2.17 0.92 2.91 2.72 0.93 2.83 2.20 0.95
Column vessel cap. cost (M$) 0.47 2.81 1.06 0.41 5.23 1.02 0.44 2.31 1.10
Energy cost (M$) 9.95 10.88 10.23
Capital cost (M$) 10.40 13.22 9.83

lowed by 797.8 M$ for 0.9 Nigeria crude in Araby-Nigeria

crude and 563.5 M$ for 0.2 Araby in Araby-Bombay
crude.
(b) The optimal column equivalent diameters for the opti-
mal compositions that yielded minimum raw-materials
and energy cost indicate pre-flash equivalent diameters
of 4.2–4.9 m; atmospheric distillation column diameters
of 10.9–23.4 m; vacuum column equivalent diameters of
14.8–15.8 m.
(c) The variation of column equivalent diameters with the
volume fraction of various crudes in binary systems indi-
cates that atmospheric distillation column equivalent
diameter is the most sensitive amongst all diameters.
These atmospheric distillation column diameters are
observed to be quite different from those obtained for
Fig. 11 – Variation of total annual profit with binary crude similar cases without any product flow rate specification
volume fractions for all cases with product flow rate constraints. In other words, product flow rate speci-
constraints (NB: Nigeria crude in a mixture of fication constraints enable tighter optimization model
Nigeria-Bombay crudes; NA: Nigeria crude in a mixture of formulation that tightens crude feed flow rate and
Nigeria-Araby crudes; BA: Bombay crude in a mixture of hence the diameters of the atmospheric distillation
Bombay-Araby crudes). column.
 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134 133

that for all cases with product flow rate constraints, the total
utility consumption has been equal to or lower than those
obtained without product flow rate constraints. These obser-
vations further confirm that it is difficult to judge upon the role
of product flow rate constraints due to the difficulty in assess-
ing the quality of the solutions to be either local or global.

5. Conclusions

This work presented the impact of the feed choice on the

optimality of grass-root design of crude distillation system
comprising of a pre-flash, atmospheric distillation column
and a vacuum column. The case study chosen constitutes a
basic representation of all columns with modest number of
theoretical stages and pump-arounds for the columns. Var-
ious case studies indicate that always gross profit shall be
Fig. 12 – Variation of optimal crude feed flow rate with
regarded as the objective function rather than raw-materials
volume fraction of various crudes in binary crudes
and energy cost, as the later could ignore design variable
corresponding to optimization cases with product flow rate
envelopes that provide maximum profit, an objective that is
constraints (AB: Araby crude in a mixture of Araby-Bombay
very much desired by industrial petroleum refineries.
crudes; NB: Nigeria crude in a mixture of Nigeria-Bombay
Case study specific inferences are presented as follows.
crudes; NA: Nigeria crude in a mixture of Nigeria-Araby
crudes).
(a) For all cases other than the case of single crudes with prod-
uct flow rate constraints, Bombay crude provided highest
profits followed by Nigeria crude and Araby crude.
(b) For all binary crude systems, Bombay-Nigeria crude char-
acterized with similar crude assay yielded good sets of
optimal results.
(c) The optimal crude feed flow rate to the distillation sys-
tem was evaluated to be around 23,849 m3 /d for most of
the cases without product flow rate constraints (including
binary crudes) and varied between 20,801 and 32,802 m3 /d
for all cases with product flow rate constraints. A stronger
impact of product flow rate constraints on the crude feed
stream flow rate has been observed.
(d) Overall process economy dominated the cost of refining
while the capital and operating costs do not contribute
more than 2% of the overall profit margins of the refin-
Fig. 13 – Total optimal steam consumption for various ery system. These observations are in strong agreement
single and binary crudes (optimal composition). with the general trends associated with the refinery
operation.
(d) The optimal reflux-ratios varied as follows: pre-flash 0.4;
atmospheric distillation column 0.8–3.5 and vacuum col- As far as Aspen Plus is concerned, the case studies further
umn 1.1–1.2. inferred that Aspen Plus is not suitable to conduct process
(e) Unlike cases reported earlier in this work, the crude stream optimization that involves both continuous and binary vari-
feed flow rate was observed to be significantly influenced ables (such as pump around location, feed stream location and
with the volume fraction of various crudes in the binary side stream location, etc). Even for the case of binary crude
crude mixtures and indicated various non-linear trends systems, it has been observed that initialization has been
(Fig. 12). These trends have been found to be in accordance extremely difficult for all problems, and much difficult prepo-
with the crude assays. sition has been posed by cases that have additional product
flow rate constraints.
4.3. Total utility consumption Based on the case studies considered, it has been inferred
that the case of product flow rate constraints accomplishes
Fig. 13 illustrates the total optimal utility consumption for higher non-linear complexity of the grass-root design prob-
both single and binary-crudes (optimal composition) with lems. While equivalent diameters obtained in this work
and without product flow rate constraints. As shown in the appear to be naive for an industrial design, the major purpose
figure, the total optimal steam consumption of the CDU sys- of evaluating the impact of crude quality and composition on
tem without product flow rate constraints corresponded to process performance has been achieved in this work. These
20,393–25,617 kg/h for both single and binary crudes (with inferences along with design data indicated in this work is
optimal composition). For all cases where product flow rate anticipated to serve as a supplement to understand the per-
constraints have been additionally imposed, the total optimal formance characteristics of the grass-root design of crude
steam consumption varied between 18,183 and 25,601 kg/h for distillation units involving processing with binary crude sys-
all crudes. Based on these observations it is hereby inferred tems.
134 chemical engineering research and design 8 8 ( 2 0 1 0 ) 121–134

References Liau, L.C., Yang, T.C. and Tsai, M., 2004, Expert system of a crude
oil distillation unit for process optimization using neural
networks. Expert Syst Appl, 26(2): 247–255.
Aires, M., Lucena, A., Rocha, R., Santiago, C. and Simonetti, L.,
Nilsson, K. and Sundén, B., 1994, Optimizing a refinery using the
2004, Optimizing the petroleum supply chain at Petrobras.
pinch technology and the mind method. Heat Recov Syst CHP,
Comput Aided Chem Eng, 18: 871–876.
14(2): 211–220.
Aspen Technology Aspen Plus user guide volume 1 and 2, 2009,
Pinto, J.M., Joly, M. and Moro, L.F.L., 2000, Planning and scheduling
Available at: http://internet.processmanual.com/
models for refinery operations. Comput Chem Eng, 24:
ipmcontent/AES 12.htm.
2259–2276.
Cao, C., Gu, X. and Xin, Z., 2009, Chance constrained
Plesu, V., Bumbac, G., Iancu, P., Ivanescu, I. and Popescu, D.C.,
programming models for refinery short-term crude oil
2003, Thermal coupling between crude distillation and
scheduling problem. Appl Math Modell, 33: 1696–1707.
delayed coking units. Appl Therm Eng, 23(14): 1857–
Dave, D.J., Dabhiya, M.Z., Satyadev, S.V.K., Ganguly, S. and Saraf,
1869.
D.N., 2003, Online tuning of a steady state crude distillation
Rivero, R., Rendón, C. and Gallegos, S., 2004, Exergy and
unit model for real time applications. J Process Control, 13(3):
exergoeconomic analysis of a crude oil combined distillation
267–282.
unit. Energy, 29: 1909–1927.
Dave, D. and Zhang, N., 2003, Hybrid methods using genetic
Seo, J.W., Oh, M. and Lee, T.H., 2008, Design optimization of a
algorithm approach for crude distillation unit scheduling.
crude oil distillation process. Chem Eng Technol, 23(2):
Comput Aided Chem Eng, 15(1): 445–450.
157–164.
Gadalla, M., Jobson, M. and Smith, R., 2003, Optimization of
Singiresu, S.R., (1998). Engineering Optimization Theory and Practice.
existing heat-integrated refinery distillation systems. Chem
(New Age International (P) Limited, India).
Eng Res Des, 81(1): 147–152.
Stojic, M.M., Nedeljkov, S.L., Krstic, D.M. and Mauhar, S., 2004,
Göthe-Lundgren, M., Lundgren, J.T. and Persson, J.A., 2002, An
Simulation of atmospheric crude unit ‘Badger’ using Aspen
optimization model for refinery production scheduling. Int J
Plus. Petroleum Coal, 46(2): 57–62.
Production Econom, 78(3): 255–270.
Sunden, B., 1988, Analysis of the heat recovery in two crude
Inamdar, S.V., Santosh, K., Gupta, D.N. and Saraf., 2004,
distillation units. Heat Recov Syst CHP, 8(5): 483–488.
Multi-objective optimization of an industrial crude distillation
William, L.L., (2006). Distillation Design and Control using ASPEN
unit using the elitist non-dominated sorting genetic
Simulation. (John Wiley & Sons, New York).
algorithm. Chem Eng Res Des, 82(5): 611–623.
Weekly Petroleum Status Report from Energy Information
Klemeš, J. and Ptácník, R., 1985, Computer-aided synthesis of
Administration, 2009, Available at: http://www.eia.
heat exchange network. J Heat Recov Syst, 5(5): 425–435.
doe.gov/oil gas/petroleum/data publications/weekly
Kumar, V., Sharma, A., Chowdhury, I.R., Ganguly, S. and Saraf,
petroleum status report/wpsr.html.
D.N., 2001, A crude distillation unit model suitable for online
applications. Fuel Process Technol, 73(1): 1–21.

View publication stats