Proceedings of the 6th International Conference on Process Systems Engineering (PSE ASIA)

25 - 27 June 2013, Kuala Lumpur.

Simultaneous optimization for heat-integrated
crude oil distillation system

Yiqing LUO, He WANG and Xigang YUAN


State Key Laboratory of Chemical Engineering, School of Chemical Engineering and
Technology, Tianjin University, Tianjin 300072, CHINA
Abstract

Crude oil distillation is a process of major importance in the refining industry. The
operating variables of the distillation process have a critical effect on the process profit
and energy recovery. However, the objective of heat recovery and product profit of the
system does not always coordinate with each other. In this paper, a systematic approach
which has considered the product profit and energy recovery simultaneously during the
optimization of the crude oil distillation systems is presented, while meeting required
product qualities by searching for an optimal operating condition by manipulating the
operating variables. The shortcut model (SCFrac) in ASPEN Plus was used to describe
the crude oil distillation process, and the pinch analysis was adopted to identify the
target of energy recovery without knowing exact heat exchanger network structure. The
optimization problem was solved by applying a particle swarm optimization (PSO)
algorithm to achieve the objective of maximum net profit. Finally, a case study was
used to illustrate the application of the approach for crude oil distillation systems and
the results showed good agreement with rigorous simulation.

Keywords: Crude oil distillation; Shortcut model; Optimization; PSO; Net profit
1. Introduction
Crude oil distillation is the most fundamental and energy-intensive process in the
petroleum refining and petrochemical industry. The operating variables of the
distillation process as well as the heat exchanger network have a critical effect on the
process profit and energy recovery. However, the objective of heat recovery and product
profit of the crude oil distillation system do not always coordinate with each other, more
heat recovery of the system does not mean more profit for the system. For example,
increasing heat duties of the pump-around near to the bottom of the atmospheric column
in the system will enable more heat to be recovered, but the naphtha product yield of the
column will decrease, as a result, the product profit from higher price product will
decrease.
So far, there have been lots of advances in crude oil distillation process, such as
increasing heat integration opportunities and obtaining an optimal heat exchanger
network structure (Friedler, 2010; Promvitak, et al, 2009), decreasing CO
2
emission
(Gadalla, et al, 2006), considering the availability of heavy crude oils by the crude oil
blending to increase the economic profit of refinery (Mittal, et al, 2011; More, et al,
2010), optimizing crude oil distillation operation by simulating and analyzing the
operating variables (Zhang, et al, 2013; Gadalla, et al, 2003; Inamdar, et al, 2004), and
so on. However, few researchers considered the product profit and energy recovery
860 Y. Q. Luo, et al
simultaneously during the optimization of the crude oil distillation systems. So in this
paper, based on the SCFrac shortcut model of ASPEN Plus and particle swarm
optimization (PSO) algorithm (Luo, et al, 2007), a systematic approach was presented
that aimed to obtain optimum operating conditions of the existing distillation process
with considering energy recovery and product profit simultaneously in the system. All
operating parameters to be optimized include product yields, stripping steam flow rate
and pump-around duties. Finally, a case study was used to illustrate the application of
the approach for crude oil distillation systems, with respect to energy demand and net
profit improvement.
2. Process Description
A typical process flow sheet for crude oil distillation system includes an atmospheric
distillation unit and a vacuum distillation unit. In this paper, all of the equipment
conditions and the range of the operating variable value are based on an existing crude
oil distillation unit in China.
The crude oil from the storage tank is preheated up to around 300 after passing
through the preheat trains and the desalter unit, and then routed to the atmospheric
furnace in which the crude oil is heated to around 370 at 145kPa and enters
atmospheric column flash zone at stage 29. The atmospheric column contains 31
theoretical stages with pressure maintained at 131kPa (top) and 147kPa (bottom). The
temperature of the top condenser is 47, and a steam stream (STC) feeds to the bottom
of atmospheric column. There are 3 pump-arounds, and each of them is located from
stage 3 to 1, from stage 12 to 10, and from stage 21 to 19 respectively. The top distillate
product is naphtha. The side products from 3 steam strippers (ST1, ST2, ST3) are
kerosene, diesel and atmospheric gas oil (AGO), which draw from stage 10, 13 and 24
respectively.


Figure 1. Simplified process flow sheet

The reduce crude oil from the bottom of the atmospheric column is heated up to about
394 in the vacuum furnace and enters the vacuum column flash zone. The vacuum
column has 14 theoretical stages with pressure maintained at 2.7kPa (top) and 3.9kPa
(bottom). The vacuum column is enabled with 3 pump-arounds, with PAV1 located
from stage 2 to 1, PAV2 located from stage 7 to 6 and PAV3 located from stage 9 to 8.
And the side products, vacuum diesel (VD), light vacuum gas oil (LVGO) and heavy
vacuum gas oil (HVGO) withdraw from stage 2, 7 and 9 respectively. The residue is the
bottom product. The simplified process flow sheet is shown in Figure 1.
Simultaneous optimization for heat-integrated crude oil distillation system 861
3. Crude oil distillation model
3.1. Shortcut model
In the design and retrofit problem of crude oil distillation system, shortcut models
(Gadalla et al, 2003) are more suitable for optimization and can be solved in a quicker
way, especially when multiple variables are considered simultaneously. In this paper,
the SCFrac shortcut model in ASPEN Plus is used to model and simulate the crude oil
distillation. The SCFrac divides an n-product column into n-1 sections. By specifying
the column pressure, estimated product flow, number of theoretical stage, steam flow
and the value of product property (such as D86 95% point), the SCFrac can give
relatively accurate results of the crude oil distillation system including the product
status and heat duty (heat duty of each pump-around) in each section.
3.2. Heat recovery calculation
For the calculation of heat recovery from the heat exchanger networks, the pinch
analysis (Linnhoff, 1983) is adopted to identify the target of energy recovery without
knowing exact heat exchanger network structure. In crude oil distillation, the crude oil
feed is the only one cold stream, and the products and pump-arounds are all hot stream.
By using pinch analysis, relatively accurate results of theoretical hot (required furnace
heat) and cold (required cooling duty) utility consumption can be obtained in a quicker
way. In this work, the minimum temperature approach is 20.
4. Optimization framework
4.1. Variables
For each product of crude oil distillation, the range of D86 95% point is specified as in
Table 1, within which the quality of the product can be regarded as satisfaction
guaranteed. The variables selected to optimization include stripping steam flow rates
(ST1, ST2, ST3, STC), and D86 95% points of naphtha (D
N
), kerosene (D
K
), diesel (D
D
),
AGO (D
A
) and vacuum diesel (D
VD
), as shown in Table 1.
Table 1. Bounds of optimization variable value
Variable Range () Variable Range (kg/h)
D
N
170~185 ST1 0~1000
D
K
250~260 ST2 0~1000
D
D
310~320 ST3 0~1000
D
A
355~365 STC 4000~9000
D
VD
360~365
4.2. Objective function
In industrial process, the profit of whole system is concerned about most. So in this
work, the objective function of the optimization is maximizing the annualized net profit
that the product profit reduces crude oil cost, steam cost, and energy cost of furnace fuel
oil and cooling water respectively. Details of product/utility prices (Table 2) and the
objective function are given below. Work hour of refinery per year is 8400h.
Objective function:
8400 ] [
ST STj C C i i
P F Erengy P F P F MaxZ
(1)
3600 ) (
CW CW F F
Q P P Q Energy (2)
Where, F
i
is the flow rate of a particular product (m
3
/h); F
C
is the flow rate of crude oil
feed (m
3
/h); P
i
is the price of a particular product; Q
F
and Q
CW
are the required furnace
heat and required cooling duty calculated from heat recovery system (MW); F
STj
is the
flow rate of a particular steam (kg/h); Energy is the system energy cost. And all these
value are the function of stripping steam flow rates and D86 95% points of each product.
862 Y. Q. Luo, et al
Table 2. Prices of products and utilities (More, et al, 2010)
Item Price Unit Item Price Unit
P
N
(Naphtha) 1003.8 $/m
3
P
F
(Fuel oil) 0.011 $/MJ
P
K
(Kerosene) 861.1 $/m
3
P
CW
(Cooling
water)
4.7410
-3
$/MJ
P
D
(Diesel) 810.2 $/m
3

P
A
(AGO and VD) 767.5 $/m
3
P
ST
(Steam) 0.0055 $/kg
P
V
(LVGO and HVGO) 718.2 $/m
3
P
C
629.0 $/m
3

P
R
(residue) 449.0 $/m
3

4.3. Optimization methodology

Figure 2. The methodology framework

The optimization problem of crude oil distillation system can be formulated as a NLP
model, which has been solved by using a particle swarm optimization (PSO) algorithm
in this paper. The PSO is a population based stochastic optimization technique. In PSO,
the potential solutions, named as particles, fly through the problem space following
the current optimum particles. Each particle is regarded as a point in a D-dimensional
space that updates according to its own experience and of other particles. The algorithm
is simple in concept and can get better results in a faster and cheaper way. The basic
equations of PSO are given below.
1 1
i i i
k k k
x x v (3)
1 1 1 2 2
( ) ( )
i i i i g i
k k k k k k k
c r c r

v v p x p x (4)
In each update step k, for each particle i, the position x
i
is updated as equation (3), and
the step velocity is calculated by equation (4), based on the particles best position p
k
i

and groups best position p
k
g
. c
1
and c
2
are two positive parameters, in this work,
c
1
=c
2
=2; = 0.729, is the inert weight; r
1
and r
2
represent uniform random numbers
between 0 and 1.
In this work, PSO algorithm has connected with the SCFrac model by using the Visual
Basic and ASPEN Plus ActiveX automation sever. The framework of the optimization
process is shown in Figure 2.
5. Case study and results
A case study has carried out based on an existing crude oil distillation system and the
conditions of the unit have been mentioned above. The feed oil of the unit is DAGANG
crude oil, and the throughput is 5208000m
3
/a (4.5Mt/a). The optimization approach
Build SCFrac simulation flow sheet in ASPEN
Set k=1, randomly initialize particle positions of variable x
i

Write the values of variable x
i
into ASPEN and run simulation
Get results from ASPEN and calculate the objective function
Update particle and group best position p
k
i
and p
k
g


Update velocity V
k
i
for all particles

Update particles positions
Stopping criterion satisfied?
k=k+1
Output results and stop
YES
NO
Simultaneous optimization for heat-integrated crude oil distillation system 863
presented in this paper was applied to achieve the maximum annualized profit by
optimizing the variables in Table 1. And the results of the optimization are shown below.
Table 3. Results of optimization variables
Variable Value () Variable Value (kg/h)
D
N
184.0 ST1 503.6
D
K
254.6 ST2 594.5
D
D
310.2 ST3 501.2
D
A
362.1 STC 5860.1
D
VD
363.6
Table 4. Results of optimal operating parameters
Parameter Value Parameter Value
F
N
(m
3
/h) 58.5 F
VD
(m
3
/h) 14.8
F
K
(m
3
/h) 42.5 F
LVGO
(m
3
/h) 120.5
F
D
(m
3
/h) 48.4 F
HVGO
(m
3
/h) 84.2
F
A
(m
3
/h) 39.7 F
R
(m
3
/h) 184.0
Duty(PAC1) (kW) 4212 Duty(PAV1) (kW) 3919
Duty(PAC2) (kW) 5499 Duty(PAV2) (kW) 7676
Duty(PAC3) (kW) 9826 Duty(PAV3) (kW) 17094
Condenser duty (kW) 10231 Energy cost (M$/a) 14.28
Energy cost (M$/a) 14.28 Annualized profit (M$/a) 284.45
Table 5. Comparison of product flow rate with rigorous simulation
Parameter Optimal
result
Rigorous
simulation result
Parameter Optimal
result
Rigorous
simulation result
F
N
(m
3
/h) 58.5 56.6 F
VD
(m
3
/h) 14.8 12.8
F
K
(m
3
/h) 42.5 41.7 F
LVGO
(m
3
/h) 120.5 125.6
F
D
(m
3
/h) 48.4 48.5 F
HVGO
(m
3
/h) 84.2 82.1
F
A
(m
3
/h) 39.7 41.4 F
R
(m
3
/h) 184.0 180.9


(a) (b)
Figure 3. Results of the optimization. (a) Maximum annualized profit in each searching
step k; (b) Optimized Energy cost in each searching step k

The results of optimization variables and optimal operating parameters are shown in
Table 3 and Table 4. Figure 3 (a) and Figure 3 (b) show the results of optimization
problem, which are the optimized annualized profit result and the value of energy cost
in each searching step k. It is observed that the objective of maximum profit and
minimum energy cost of the crude oil distillation system does not coordinate with each
other. The maximum annualized profit is 284.45M$/a and the corresponding energy
cost is 14.28M$/a. But when the energy cost is minimum (13.91M$/a), the annualized
profit is only 240.97M$/a. So less energy cost represents more heat recovery, but it
doesnt mean more net profit. And Figure 3 also shows that the PSO algorithm got the
maximum annualized profit during less than 50 times optimization (k=46). It proved
that the PSO algorithm presented in this work can solve this problem in a fast way. By
864 Y. Q. Luo, et al
specifying the product D86 95% points and pump-around heat duty of the optimization
results as the design specification, a rigorous simulation in ASPEN Plus was carried out.
Compared with the results of product flow rate in Table 5, it showed good agreement
between the optimization approach and rigorous simulation. It proved that the
optimization approach could provide relatively accurate and feasible results.
6. Conclusions
In this paper, a systematic approach which has considered the product profit and energy
recovery simultaneously during the optimization of the crude oil distillation systems
was presented, while meeting required product qualities through searching for an
optimal operating condition by manipulating the operating variables. The shortcut
model (SCFrac) in ASPEN Plus was used to describe the crude oil distillation process,
and the pinch analysis was adopted to identify the target of energy recovery without
knowing exact heat exchanger network structure. The result of optimization shows that
the operating variables of the crude distillation process such as product yields, pump-
around duties, stripping steam flow rates, etc. have a critical effect on the process profit
and energy recovery. And the objective of maximum profit and minimum energy cost
does not coordinate with each other, they should be optimized simultaneously. The
methodology developed in this paper can be solved with PSO algorithm in a fast way
and provide a quantitative support for the optimization of the crude oil distillation
systems. Compared with the results of rigorous simulation, it can be proved that the
optimization approach can give relatively accurate results.
References
B. Linnhoff, E. Hindmarsh, 1983, The pinch design method of heat exchanger networks, Chem
Eng Sci, 38, 745-763
F. Friedler, 2010, Process integration, modelling and optimisation for energy saving and pollution
reduction, Appl Therm Eng, 30, 2270-2280.
M. Gadalla, M. Jobson, R. Smith, 2003, Shortcut models for retrofit design of distillation columns,
Trans IChemE, 81, Part A, 971-986
M. Gadalla, . Oluji, M. Jobson, R. Smith, 2006, Estimation and reduction of CO2 emissions
from crude oil distillation units, Energy, 31, 2398-2408
N. Zhang, R. Smith, I. Bulatov, J. Kleme, 2013, Sustaining high energy efficiency in existing
processes with advanced process integration technology, Applied Energy, 101, 26-32
P. Promvitak, K. Siemanond, S. Bunluluesriruang, V. Raghareutai, 2009, Retrofit design of heat
exchanger networks of crude oil distillation unit, Chemical Engineering Transactions, 18, 99-
104
R. K. More, V. K. Bulasara, R. Uppaluri, V. R.Banjara, 2010, Optimization of crude distillation
system using aspen plus: Effect of binary feed selection on grass-root design, Chem. Eng. Res.
Des., 88, 121-134
S. V. Inamdar, K. S. Gupta, D. N. Saraf, 2004, Multi-objective optimization of an industrial crude
distillation unit suing the elitist non-dominated sorting genetic algorithm, Chemical
Engineering Research and Design, 82(A5), 611623
V. Mittal, J. Zhang, X. T. Yang, Q. Xu, 2011, E3 Analysis for crude and Vacuum Distillation
System, Chem. Eng. Technol, 34, No. 11, 1854-1863
Y. Q. Luo, X. G. Yuan, Y. J. Liu, 2007, An improvec PSO algorithm for solving non-convex
NLP/MINLP problems with equality constraints, Comput. Chem. Eng., 31, 153-162