Determination of Optimal Cut Point Temperatures at Crude Distillation Unit using the Taguchi Method
ABSTRACT This paper proposes a technique for optimizing crude cut points. Taguchi method is applied in the selection of the significant variables and their respective optimal levels for a fractionation process. The variables considered were the cut point temperatures of products, namely, naphtha, kerosene, light and heavy diesels, atmospheric gas oil and the residue. The variables were assigned lower and upper bounds with a difference of ± 13.9°C (± 25°F) from the standard straightrun cut points. A steadystate model of a Crude Distillation Unit (CDU)is used as a virtual plant to carry out the fractionation of 100 kilo barrels per day of crude oil. S traightrun cases comprising of three Malaysian crude oils, namely, Bunga Kekwa, Bintulu and Tembungo, were analyzed as single, binary and ternary crude feeds. In each case, an optimal configuration of variables was determined by minimizing the energy required for the production of one kilo barrel/day of diesel.
 Citations (17)
 Cited In (0)
 [Show abstract] [Hide abstract]
ABSTRACT: With tighter emission requirements on fuels and the changing world supply of crude oil, more detailed methods for characterizing refinery feed and product streams are needed to make the complicated decisions required to control and optimize refinery operations. Changes in analytical measurement technology provide additional data for fingerprinting the molecular structure of these hydrocarbon mixtures. Integrating this molecular characterization of the materials with advances in compositional modeling methods, model reduction techniques, optimization, and equationsolving methods, along with continued implementation of model predictive control, improves operational effectiveness at all levels of the refinery from the individual process level to the refinerywide planning optimization. The examples presented highlight progress toward capturing the highest value of every molecule at every point in the refining complexIEEE control systems 01/2007; · 2.37 Impact Factor  [Show abstract] [Hide abstract]
ABSTRACT: For those refineries which have to deal with different types of crude oil, blending is an attractive solution to obtain a quality feedstock. In this paper, a novel scheduling strategy is proposed for a practical crude oil blending process. The objective is to keep the property of feedstock, mainly described by the true boiling point (TBP) data, consistent and suitable. Firstly, the mathematical model is established. Then, a heuristically initialized hybrid iterative (HIHI) algorithm based on a twolevel optimization structure, in which tabu search (TS) and differential evolution (DE) are used for upperlevel and lowerlevel optimization, respectively, is proposed to get the model solution. Finally, the effectiveness and efficiency of the scheduling strategy is validated via real data from a certain refinery.Chinese Journal of Chemical Engineering. 01/2010;  SourceAvailable from: Jose Romagnoli[Show abstract] [Hide abstract]
ABSTRACT: This paper presents an integrated optimization approach for crude operations scheduling and production for refineries. The production process is composed of a prefractionator, crude, and vacuum distillation columns. It is modeled as an NLP. The scheduling problem is composed of unloading operations and simultaneous blending and charging of CDUs. It is modeled as a MILP. The nonlinear simulation model for the production process is used to derive individual crude costs for the two crudes considered (Dubai and Masila). This is performed using multiple linear regressions of the individual crude oil flow rates around the crude oil percentage range allowed by the production facility. These individual crude costs are then used to derive a linear cost function that is optimized in the MILP scheduling model, along with logistics costs. Results show that this integrated approach can lead to a 0.53 M$ decrease in production and logistics costs in a 15 day time horizon.01/2010;
Page 1
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 36
12106063939IJETIJENS © December 2012 IJENS I J E N S
Determination of Optimal Cut Point
Temperatures at Crude Distillation Unit using the
Taguchi Method
Syed Faizan Ali , Nooryusmiza Yusoff
Abstract— This paper proposes a technique for optimizing
crude cut points. Taguchi method is applied in the selection of the
significant variables and their respective optimal levels for a
fractionation process. The variables considered were the cut
point temperatures of products, namely, naphtha, kerosene, light
and heavy diesels, atmospheric gas oil and the residue. The
variables were assigned lower and upper bounds with a
difference of ± 13.9°C (± 25°F) from the standard straightrun
cut points. A steadystate model of a Crude Distillation Unit
(CDU)is used as a virtual plant to carry out the fractionation of
100 kilo barrels per day of crude oil. Straightrun cases
comprising of three Malaysian crude oils, namely, Bunga Kekwa,
Bintulu and Tembungo, were analyzed as single, binary and
ternary crude feeds. In each case, an optimal configuration of
variables was determined by minimizing the energy required for
the production of one kilo barrel/day of diesel.
Index Term Crude Distillation Unit, Cutpoint Optimization,
Diesel Production, Taguchi Method
I.
INTRODUCTION
Petroleum refineries have advanced periodically with the
passage of time. Refinery operations for the creation of
products such as Naphtha, Diesel, Kerosene and Gasoline
have grown complex affecting the refinery profit. Limitations
such as safety and environmental regulations, for maintaining
plants to run at cleaner processes, are some constraints to
achieve such profits [1].
Crude oil trapped in different reservoirs of the world of a
specific field hold unique characteristics from one another on
a physical and chemical basis [2]. The first classification of
crude can be done in „light‟ and „heavy‟ crudes having
respective importance with regard to profit extraction.
Classification of crude oil is based upon the difference in
specific gravities and the proportion with which they form.
Product demand is met by proportioned blending of crude.
Seasonal scheduling of the crudes is carried out in order to
produce optimized cuts [3]. The type of product requiring
greater operating cost resulting in lesser profit is subtracted
from the crude to meet the demand supply.
Syed Faizan Ali and Nooryusmiza Yusoff are with the Chemical Engineering
Department, Universiti Teknologi PETRONAS, 31750 Tronoh, Perak,
Malaysia (email: nooryus@petronas.com.my)
Refineries fractionate these barrels of crude by their boiling
points in order to obtain high value products such as gasoline,
diesel, jet fuel, heating oil, fuel oil, lubricants, asphalt, coke,
wax, and chemical feed stocks [4].
Many studies have been published related to crude
distillation unit (CDU) study with reference to refinery
planning and scheduling [5][7], estimation of product
properties [8][9] and process control, modeling, simulation
and optimization [10][12]. Optimization of a crude
distillation unit using a binary feed was carried out on the
basis of the gross profit instead of the costs inferred by energy
and raw materials [10]. An atmospheric distillation unit
subjected to transient behavior due to changes in the operating
conditions can be improved by a suitable control strategy to
obtain better operations [11]. An expert system was designed
for a CDU to predict the product flow and temperature values
by minimizing the model output error by genetic algorithm
framework and maximizing the oil production subjected to
control parameters [12].
Optimization has been previously carried out by devising a
process control strategy, rigorous modeling and improved
design specifications. The performance of the CDU using
straight run cut points for several types of crude oil such as
Tapis (44.8), Bintulu (36.0) and Terengganu (47.4) where the
optimization objective was a profit function (total product
value – feed cost – utility cost) using AspenTech DMC+ (an
APC technology) [13]. The literature cited above deduces that
the optimization based on the crude cut points have not been
given much importance. Previously, straight run temperatures
have been applied on every type of crude in order to optimize
a crude distillation system. The objective of this article is a
detailed design constituting the crude cut points to minimize
the amount of energy utilized to produce a kilobarrel per day
of a product (for e.g. diesel) using optimized cut point
temperatures.
This study can be regarded valuable for the operations
personnel concerned with the planning and scheduling of the
crude feed involving the blending of different crudes to reduce
the supplydemand gap of the refinery products. It deals with a
crude distillation unit modeled in Aspen HYSYS environment.
The methodology devised for the crude optimization is based
on a design of experiments technique known as Taguchi
method. The optimization variable involved is the overall
Page 2
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 37
12106063939IJETIJENS © December 2012 IJENS I J E N S
energy required in order to produce a kilo barrel of a product,
which in this study is diesel.
Section II deals with the background study of the crude
assays considered followed with the process description for
simulating the process. Section III discusses the methodology
of the paper formulated, initiating from the crude specification
procedure in HYSYS, an overview to the Taguchi design of
study and the devised case studies. In Section IV, the
optimized results are outlined with emphasis on determining
the cutpoint recipe and shifting of TBP curves to produce
such beneficial results. Finally, the last section concludes the
study of this work and recommended future studies to this
work.
II.
BACKGROUND
A. Crude Assay
Refining engineers analyze the True Boiling Points (TBP)
curves of the cuts present to determine the behavior of the
crude distilled and various saleable products [14]. The crude
assays considered are taken from the different fields of
Malaysia namely Bintulu, Tembungo and Bunga Kekwa.
These crude assays have been obtained from the Assay
manual published in December 2003 by KBC Advanced
Technologies Inc. The samples date back to October 1994
(Bintulu), November 1997 (Bunga Kekwa) and October 1986
(Tembungo). With 0.5% sulfur content as a limit, sweet crude
has been observed ranging from 0 to 0.08% by weight. The
specific gravities and the sulfur content are shown in Tab. I:
TABLE I
CRUDE DATA
Specific
Gravity (15°C)
0.8445
0.8415
0.8400
Crude Assay
API
Values
36.0
36.6
36.9
Sulfur content
(wt. %)
0.020
0.064
0.049
Bintulu
Tembungo
Bunga Kekwa
Fig. 1. True boiling point curves
From the HYSYS oil manager, the straight run analyses yield
25.6% diesel from Bintulu crude, 37.6% from Tembungo
crude and 29.1% from the Bunga Kekwa crude; thus
indicating that the Tembungo crude produces more diesel as
compared to other two crudes. Fig. 1 shows that Bintulu and
Tembungo have comparable TBP profiles whereas Bunga
Kekwa crude comprises heavier components.
B. Process description
Fig. 2 shows the process flow diagram for a CDU simulated
in Aspen HYSYS. The diagram consists of a preflash unit
followed by a CDU. The feed to the CDU is preheated in a
furnace.
Crude oil at a rate of 100 kilo barrel per day (kBPD) is fed
to the preflash tower at a temperature of 232.2˚C and a
pressure 517.1 kPa. The preflash tower is responsible for
separating the crude vapors and liquid entering into the crude
column as a bottom feed. This is carried out to reduce the duty
of the furnace to devise an economical process [15].
The furnace feed is the bottom product of the preflash
tower operating at a pressure drop of 68.95 kPa with the crude
being heated to 343.6˚C. As shown in Fig. 2 the column
consists of 29 stages with a partial condenser, three side
strippers and three pumparounds. The heated crude is sent to
in the tray 28. Side strippers comprising 3 stages have been
utilized for diesel and atmospheric gas oil (AGO).
Fractionation is increased by reducing the partial pressures
with the aid of steam and a reboiler for Kerosene. The
pressure drop of the CDU is 62.05 kPa with a top and bottom
stage pressure of 144.8 kPa and 255.5 kPa, respectively.
Internal reflux has been ensured by the installation of three
pumparounds as in Tab II (A).
TABLE II (A)
PUMPAROUND SPECIFICATIONS
Location between trays
1 and 2
16 and 17
21 and 22
Pumparound
PA1
PA2
PA3
Duty (kW)
16,124
10,258
10,258
Flow (kBPD)
50
30
30
TABLE II (B)
SIDESTRIPPER SPECIFICATIONS
Location between trays
8 and 9
16 and 17
21 and 22
Side Stripper
SS1
SS2
SS3
Stripped by
Reboiler
Steam
Steam
Flow / Duty
2199 kW
1361 kg/hr
1134 kg/hr
0
100
200
300
400
500
600
700
800
900
0 20 406080100
Temperature (˚C)
Percentage of Distilled Products (Volume %)
Bungka Kekwa
Tembungo
Bintulu
Page 3
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 38
12106063939IJETIJENS © December 2012 IJENS I J E N S
Fig. 2. Process Flow Diagram of CDU
The partial condenser is operated at a pressure of 135.83
kPa with waste water as the side product. The distillate
(naphtha) rate is maintained at 20 to 25 kBPD. The reflux
ratio is fixed at 1.7 corresponding to a reflux flow rate of 34
to 42 kBPD. The bottom steam entering at tray 28 is
exchanging heat twice, i.e. absorbing heat from the liquid
flowing down the trays and then exchanging heat with the
upward flowing vapors, entered at a rate 3402 kg/hr at
190.6˚C and 1034 kPa. The over flash is specified at the tray
27 with 3.5% of the feed.
Straight run analyses have been carried out to determine
the volume percent of the distilled products namely Offgas,
Naphtha, Kerosene, Diesel, AGO and Residue. These
products are further subjected to physical and chemical
separation processes for better quality in, for example; the
vacuum distillation unit (VDU), fluid catalytic cracking unit
(FCCU) and hydrocracker. The summary of the process
configuration is shown in the Tab. III.
TABLE III
CDU SPECIFICATIONS
Parameter Value
No. of ideal trays
Temperature
Pressure
No. of pumparounds
No. of side strippers
Feed rate
Feed Location
Feed Temperature
Feed Pressure
Reflux ratio
Condenser type
29
169.2˚C (top stage)
355.6˚C (bottom stage)
144.8 kPa (top stage)
255.5 kPa (bottom stage)
3
3
100 kBPD
Tray 28
343.6˚C
448.2 kPa
1.7
Partial Condenser
III. METHODOLOGY
A. HYSYS Crude Specification
In HYSYS, components and the thermodynamic fluid
package (PengRobinson) are defined to create the
simulation basis. In HYSYS Oil manager characterizes the
crude assay to generate the hypothetical components with
their respective physical and critical properties. The
correlations and calculations performed in the oil manager
are all in accordance with the API technical data book.
The crude assay is defined with the TBP assay type to
generate an internal TBP curve at atmospheric conditions
Atmospheric
Distillation Unit
Furnace
Residue
AGO
Diesel
Kerosene
Naphtha
Crude
Middle . Pumparound
Top Pumparound
Bottom Pumparound
Condenser
Cooling Water
SS  1
SS  2
SS  3
Reboiler
Reflux
Separator
Crude Vapor
Tower feed
Mixer
Bottom Steam
Steam
Steam
1
28
29
Page 4
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 39
12106063939IJETIJENS © December 2012 IJENS I J E N S
for the characterization method. The assay definition is
based on the API gravity, distillation data and the input
composition of the components from Methane Pentane (C1
C5). However, sulfur contents are not specified in this work.
The extreme ends of the distillation TBP curve are
extrapolated.
Aspen HYSYS calculates the blend of a single or multiple
crudes with their respective flow rates defined. The numbers
of cuts are defined using the values for boiling point ranges
as shown in the Tab. IV. It generates the hypothetical
components on the „light ends free‟ basis to obtain more
accurate results. Half of the crude is fractionated into the
cuts ranging from gas oil to a low boiling point gas. Initial
boiling point (IBP) has been determined as a weighted
average boiling point from the components present in the
first volume percent if the distilled crude. The boiling points
of these components are utilized as 1% of the IBP and 98%
of the final boiling point (FBP). The FBP is determined
similarly using the components found from 98100% of the
distilled crude. Later the assay is distributed by specifying
the cut input information before it can be installed into the
simulation environment.
The calculation of the energy per diesel product flow has
been carried out by considering the pure energy streams
entering and exiting into the system and not by the energy
withheld by the material streams, thus only accounting for
the energy consumption due to utilities only.
TABLE IV
TBP RANGES AND CUTS[6]
TBP range
< 38°C (<100°F)
38 – 427°C (100800°F)
427 – 649°C (8001200°F)
649 – 871 °C (12001600°F)
ΔT No. of cuts
Pentanes and Lighter
28
8
4
14 °C
28 °C
56 °C
B. Taguchi Method
Taguchi design is a statistical technique to optimize the
process design problems in different engineering disciplines.
It implies the formation of a layout of the experiments by
the combination of factors to obtain the number of tests to
be performed. Besides Taguchi method, the factorial design
methods are also utilized to examine the combination of
factors involved to arrive at the best product. It has certain
shortfalls with respect to the number of parameters involved,
which are as follows:
Time consuming and costly
Different outputs of the different configuration
of the same experiment
Influence of factors cannot be determined
Taguchi method is considered as an offline optimization
method for the designing of system from the parameters
involved. System design can be considered as the formation
of the experiments on the basis the engineering principles
employed and process limitations. Parametric design
determines the optimum state due to the design factors at a
specific point.
In the beginning, Taguchi design has been only used in
the manufacturing sector to devise the most economical way
of building equipments with the best performance. Later it
evolved and proved quite helpful in the process industry to
characterize the performance of an operation and has been
proven highly effective in order to study the effects of the
multiple factors involved and the influence of the factors on
the performance of an experiment.
Reliable results can be obtained via standard tables
known as orthogonal arrays which are helpful in designing
the experiments. The factors represent the design parameters
that influence the performance whereas levels are the values
by which an experiment is conducted. With two levels of
factors the behavior is fundamentally assumed to be linear
where the nonlinear response can be observed with the
latter levels [16].
Taguchi method is used for the selection of optimization
variables to minimize the energy consumption in crude
distillation unit [17]. A systematic procedure was designed
for the selection of optimization variables in a refrigerated
gas plant (RGP), later validated on HYSYS, and showed
noteworthy concurrence amongst all cases [18]. It was
proposed that the concentration of a surfactant, zeolite bed
height and flow rate of the waste water are the significant
factors in detecting color removal from textile dye bath
effluents in a zeolite fixed bed reactor [19].
The consistency of the performance is analyzed by
determining the effects of the factors involved and obtaining
the optimum configuration for the best performance by
analysis of means (ANOM) or analysis of variance
(ANOVA). The model factors are found by calculating the
Signal to Noise (S/N) ratio or weight factors to come up
with a robust design [20]. The measured result of these
experiments constitutes a quality characteristic which in this
case is the minimum energy requirement for the unit
production of diesel.
In the current research, Taguchi method has been used to
devise a recipe for increased diesel production in a crude oil
as in Fig. 3. The objective (Eqn. 1) is to select the optimal
cut point temperatures for maximizing diesel yield at the
lowest energy utilization. In this case, 3 levels namely
nominal as well as lower and upper bounds have been
considered.
V
E
FlowProduct Diesel
Energy
(1)
Page 5
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 40
12106063939IJETIJENS © December 2012 IJENS I J E N S
Fig. 3. Proposed methodology cut point optimization
The experiments to be conducted using the proposed
methodology are designed using the Taguchi method
explained below;
1) Design of Experiments
Since the model of CDU is based on first principles, the
significance of the factors involved can be systematically
established through Taguchi method. The next step is the
designing of experiments and simulating them on HYSYS.
The selection of standard orthogonal array is based on the
number of factors and the levels involved. Three levels have
been considered because Aspen HYSYS calculates the
blend using the following Tab. III for boiling point ranges
with their designated number of cuts. Since the cut points
fall in the range of the 426.7°C, therefore the difference of
the upper and lower bounds from the reference cut points
has been taken as 13.9°C (25°F).
For 8 factors and 3 levels, an L18 array is selected as
shown in the Tab. V. The rows and columns represent the
test runs and factors involved respectively.
TABLE V
ARRAY SELECTOR[21]
Number of Factors
4 5
2 3 6 7 8 9 10
Number of Levels
2
L4 L4 L8 L8 L8 L8 L12 L12 L12
3
L9 L9 L9 L18 L18 L18 L18 L27 L27
4
L‟16
L‟16
L‟16
L‟16
L‟32
L‟32
L‟32
L‟32
L‟32
5
L25 L25 L25 L25 L25 L50 L50 L50 L50
Note: „apostrophe accounts for the modified arrays from the standard ones.
Since three crudes are involved, the number of case
studies is seven (231=7). It is noteworthy that only 18
experiments needs to be conducted as in Tab. VI, for each of
the 7 runs in order to determine the responses of the 8
factors involved. Therefore, a total of 126 experiments need
to be carried out. This is more beneficial and easy as
compared to 7 × (37) + 7 × (21) = 15,323 experiments
designed by full factorial design approach.
Other possible properties in the oval can also be specified
Page 6
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 41
12106063939IJETIJENS © December 2012 IJENS I J E N S
TABLE VI
L18 ARRAY
T2 T3
1 1
1 2
1 3
2 1
2 2
2 3
3 1
3 2
3 3
1 1
1 2
1 3
2 1
2 2
3 3
3 1
3 2
3 3
Experiments
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
T1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
T4
1
2
3
1
2
3
2
3
1
3
1
2
2
3
1
3
1
2
T5
1
2
3
2
3
1
1
2
3
3
1
2
3
1
2
2
3
1
T6
1
2
3
2
3
1
3
1
2
2
3
1
1
2
3
3
1
2
T7
1
2
3
3
1
2
2
3
1
2
3
1
3
1
2
1
2
3
T8
1
2
3
3
1
2
3
1
2
1
2
3
2
3
1
2
3
1
2) Analysis of Results
The final step is the analysis of results with the help of the
statistical means such as ANOM and ANOVA, averages of
factor k at level l in case m,
kl
x
factorial values.
Where;
K= 7 and L=3 are correspondingly numbers of factors and
levels.
Average of factor k in each case m, xk is calculated as:
klxx
kl
klkl
,, 1;
These two averages are used to calculate
denominator is called degrees of freedom of factor k over all
levels L in case m,
k
DOF)(
.
2
m
is taken as sum of respective
m
kl
m
kl
xx
k=1,…,K; l=1, …, L; m=1, …, M (2)
K
L
mm
(3)
m
k
V
. The
m
m
k
l
m
kl
m
kl
m
k
DOF
xx
V
)(
)(
1
(4)
Percentage contribution
V
m
k
)(
1
The validation of the optimized recipe obtained is based
upon the ANOM and the ANOVA rankings. If the rankings
are in compliance with each other, an optimized run is
performed subjected to the optimal configuration of the cut
point temperatures to calculate the objective function for a
respective case.
m
k
C is determined as follows:
Kk
V
C
k
m
k
m
k
,, 1;
100
(5)
C. Case Studies
The cases considered for the optimization of the CDU are:
Single crude: Bintulu (Case I), Tembungo (Case II)
and Bunga Kekwa (Case III)
Binary crudes: Bintulu–Tembungo (Case IV),
Tembungo–Bunga Kekwa (Case V) and Bunga
Kekwa–Bintulu (Case VI)
Ternary crudes: BintuluTembungo–Bunga Kekwa
(Case VII)
The amount of different flow rates calculated for crude
blending is based on the ratio of the diesel production of the
crudes considered in order to produce maximum amount of
diesel. From the straight run analysis of the different crudes
that Tembungo was found to be the best crude for diesel
production followed by Bunga Kekwa and Bintulu.
IV. RESULTS AND DISCUSSIONS
The simulations have been performed using specifications
such as product flow rate constraints and duties of
condenser, pumparounds and reboiler. All experiments are
conducted using steadystate model developed under
HYSYS environment. Consistency has been maintained in
the units prior to the calculations. The experiments are
conducted using the default cut point temperatures in the
HYSYS environment as shown in the Tab. VII. The
temperature difference between the cuts in Tab. IV is
considered as a reference for setting the lower and the upper
bound for the experiments.
TABLE VII
CUT POINT TEMPERATURES AND VALUES AT EACH LEVEL
S.R Cut
points (˚C)
Offgas A 10
Light S.R B 70
Naphtha C 180
Kerosene D 240
Light Diesel E 290
Heavy Diesel F 340
Atmospheric
Gas Oil
Residue H 1200
Products Factor
Level 1
(˚C)
10
56
166
226
276
326
Level 2
(˚C)
23.9
70
180
240
290
340
Level 3
(˚C)

84
194
254
304
354
G 370 356 370 384
1144.5 1175 1200
Note: S.R: Straight run method
These cutpoint temperatures are used to predict the
amount of extract of the petroleum fraction to be made, and
define the pseudo component as midpoint normal boiling
point (NBP).
A. Taguchi Results
For the base case, the objective function values (Eq. 1)
are 1714.8, 2646.2, 1762.3, 2146.8, 2321.3, 2003.3 and
2191.2 in kW/kBPD. To calculate the optimal recipe, the
factors are ranked using the analysis of mean (ANOM)
results. This way the significance of factors can be
determined. Eqs. (2) and (3) are used to calculate the
averages of the factors k which are ranked from 1 to 9.The
ranking is based on the difference of the lowest from the
global mean. This difference is noted as Dk, where highest
value of Dk is assigned the highest ranking.
Page 7
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 42
12106063939IJETIJENS © December 2012 IJENS I J E N S
Case I Case II
Case III Case IV
Case V Case VI
Case VII
Fig. 4. Response plots for Cases: (I) Tembungo , (II) Bintulu , (III) Bunga Kekwa, (IV) Tembungo – Bintulu, (V) Tembungo – Bunga Kekwa, (VI) Bunga
Kekwa – Bintulu, (VII) Tembungo – Bintulu – Bunga Kekwa
Ranking from ANOM is verified from the (analysis of
variance) ANOVA results. The sum of the squares of the
factor involved, represent the variation of the experimental
result from the data average where the larger value
represents a significant factor and vice versa. The degree of
freedom (DOF)k is calculated to be one less than the number
of levels. It is used to obtain the population variance of all
the possible experiments. Finally, the relative important
factor k is indicated with regard to the percentage
contribution.
The optimal configuration of the cut point temperatures was
based upon the „the lower the better‟ principle of quality
through averaged energy/product response plots. Thus,
another set of experiments subjected to the optimal
configuration has been performed known as the optimized
run. Optimal recipe shown in Fig. 4 infer that steep slopes
1500
1600
1700
1800
1900
2000
2100
2200
ABCDEFGH
Energy/ Product Flow (kW / kBPD)
2000
2200
2400
2600
2800
3000
3200
ABCDEFGH
1700
1800
1900
2000
2100
2200
2300
ABCDEFGH
Energy/ Product Flow (kW / kBPD)
1800
1925
2050
2175
2300
2425
2550
2675
2800
ABCDEFGH
1800
1900
2000
2100
2200
2300
2400
2500
2600
ABCDEFGH
Energy/ Product Flow (kW / kBPD)
1700
1800
1900
2000
2100
2200
2300
2400
ABCDEFGH
1700
1800
1900
2000
2100
2200
2300
2400
ABCDEFGH
Energy / Product Flow (kW/kBPD)
Average Factors at Respective Levels
Page 8
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 43
12106063939IJETIJENS © December 2012 IJENS I J E N S
can reveal the significance of the factors involved. The
steeper the slope the more significant a variable is. The
averages of the energy/product flow (kW/kBPD) are based
upon the ANOM results. The optimal configuration of the
cut point temperatures are shown in Tab. VIII.
TABLE VIII
RECIPE FOR OPTIMUM DIESEL YIELD
Cases Recipe
Factors
Base Case
I
II
III
IV
V
VI
VII
A
1
1
1
2
2
1
1
2
B
2
3
3
3
3
3
3
3
C
2
1
1
1
1
3
1
1
D
2
1
1
1
1
1
1
1
E
2
3
3
1
3
1
3
3
F
2
3
3
3
3
3
3
3
G
2
3
1
1
3
2
3
1
H
3
2
2
1
3
2
3
3
Tab. VIII shows great deal of consistency from a specific
level involved of the factors B, D, and F. These factors
correspond to cut point temperatures of the light straight run,
kerosene and heavy diesel, respectively. The contributing
factors calculated from the ANOM results revealed that only
the cutpoint temperature of kerosene and heavy diesel are
responsible for the optimization of diesel in a crude
distillation unit. This can also be validated from the
steepness of the slopes shown by the two factors. The
shifting of the levels caused the diesel range to expand
much wider as compared to their base cases. The final
boiling point of the kerosene cut was reduced and for heavy
diesel it increased. This resulted in the higher production of
diesel resulting and the vapors climbing up to the condenser
discharged much greater energy from the system.
The minimum value of energy/product flow (kW/kBPD)
for the optimum configuration for the all the cases was
calculated after conducting simulation of similar process
configuration but different cutpoints as shown in Tab. VII.
Thorough inspection of the above configuration reveal that
the optimum recipe of the Case I and II are alike except
factor G which switches from level 3 to 1. This trend is
observed due to the fact that the TBP curves for Bintulu and
Tembungo are quite alike. And later in their TBP curves
they deviate from one another, thus causing the switching of
the level G. Similarly Case VI also comprises of identical
level of factors with Case I except for factor H which
changes from level 2 to 3. The resemblance of these two
cases deduces an interesting point that Case I has the best
optimum value in the category of single crude feed.
Similarly, Case VI shows the finest results amongst binary
crude feed in terms of energy utilization for a kilobarrel of
production. On the contrary Case II reflects higher objective
function because the straight run analysis of Bintulu crude
produced the lowest amount of diesel production in
comparison to the other two crudes.
B. Cases Studies
The global means for the cases of single crude feed (I, II
and III) were 1861.2, 2563.8 and 2003.2, respectively, in the
unit of kW/kBPD. The diesel production increased up by
6.9, 7.4 and 8.9 kBPD respectively. For optimal runs in
Cases I and II the energy usage was reduced by 32.4% and
22.9%, respectively. Case III reported an increase by 25.5%
from its straight run analysis as shown in the Fig. 5.
The global mean for the binary crude feed Cases (IV, V
and VI) were found to be 2230.3, 2243.3 and 2036.9
respectively in kW/kBPD. As for the ternary crude feed
(Case VII), the objective function was calculated as 2256
kW/kBPD from the 18 experiments performed. The diesel
yield was also increased by 7.5kBPD because the usage of
such crude feed to a comparatively greater federate as
compared to the others. This ratio was calculated on the
basis of the S.R analysis of the single crude feed. Tembungo
crude was kept 27.3% and 27.5% higher than Bintulu and
Bunga Kekwa respectively. Bunga Kekwa feed was kept
47.5% higher than Bintulu to produce greater diesel.
According to the HYSYS oil manager, the first two cases
have shown commendable increase in the light diesel
fraction. The remaining of the cases notified a greater
increase in the heavy diesel fraction as compared to the light
diesel fraction off the entire diesel production.
Fig. 5. Trends of energy requirement for a barrel of product for various
case studies
Since diesel is fractionated through ncetane, it is
recognized with its cetane index, which is a measure of its
knocking tendency. The cetane index has been kept within a
range of 49 to 55 with the percentage change to its
corresponding straight run is Tab. VII.
The pour points for the diesel stream were estimated
using the ASTM D97 method. The pour points for the
straight run experiments are reported in Tab. XI along with
the temperature deviation of the optimized runs with their
respective straight runs.
0
500
1000
1500
2000
2500
3000
IIIIIIIVVVIVII
Energy/Product (kW / kBPD)
Straight runOptimized run
Page 9
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 44
12106063939IJETIJENS © December 2012 IJENS I J E N S
Some experiments indicated reduced AGO flow rates
depicting that a major portion of the AGO cut has been
shifted to the diesel yield as shown in the Fig. 6 below.
TABLE XI
RESULTS FOR OPTIMIZED RUNS
Cases I II III IV V VI VII
Optimized run Energy/Barrel of
Product (kW/kBPD)
1352.6 1972.8 1358.2 1572.6 1842.3 1187.4 1729.2
Percent Optimized (%)
S.R Cetane Index C.I (C.I %)
S.R ASTM D97 Pour point ˚C (P.P)
Note: C.I% = Percentage difference of the Cetane index; P.P = Difference of Pour point values within straight and optimized run
21.1 25.4 22.9 26.7 20.6 40.7 21.1
52.07 (0.71)
6.37 (1.1)
48.59 (1.2)
20.05 (0.8)
50.08(0.8)
25.2 (1.0)
52.12 (0.7)
13.16 (0.5)
52.95 (0.3)
14.06 (0.9)
53.33 (0.5)
11.58 (0.8)
53.52 (0.4)
6.10 (1.7)
Fig. 6. Curve Shifting of Bintulu Crude: Straight run ( ) v/s Optimized run ()
For Case II, the increase in the diesel production is
observed due to shifting of all the product TBP curves. As
shown in Fig. 6, diesel range has expanded both ways by
extracting 4.9 kBPD from Kerosene and 1.9 kBPD from
AGO. Shift in the IBP of diesel from 140.9˚C to 136.3˚C
and FBP 360.6˚C to 387.4˚C resulted in a greater lighter
fraction of diesel.
The increase of the diesel TBP curve was up to 6.8 kBPD as
compared to the straight run product TBP curve. The shift
from the dashed line to the solid line can be observed in
each of the product against the overall TBP curve to the
optimized product configuration.
CONCLUSION
The fractionation of light crudes and their blends have
been optimized by manipulating 8 cutpoint temperatures. In
general, the utilization of Taguchi method has increased the
diesel production whilst decreasing energy consumption.
The optimized recipes were significantly different from their
respective base cases. The optimal recipes of all cases
showed that 20 to 41% benefits can be achieved as
compared to straight run temperatures. Two factors namely
D and F, i.e., the cut points of Kerosene and Heavy Diesel,
respectively, are significant for the optimization of the
objective function.
NOMENCLATURE
Abbreviations
API
AGO
ANOM
ANOVA
CDU
American Petroleum Institute
Atmospheric Gas Oil
Analysis of Means
Analysis of Variance
Crude Distillation Unit
Page 10
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 45
12106063939IJETIJENS © December 2012 IJENS I J E N S
BPD
FBP
FCCU
IBP
TBP
NBP
VDU
Variables
Barrels per day
Final Boiling Point
Fluid Catalytic Cracker Unit
Initial Boiling Point
True Boiling Point
Normal Boiling Point
Vacuum Distillation Unit
Percentage contribution of factor k in case
m (%)
Difference between the lowest and the
global mean
Degree of Freedom at level k in each case m
Number of factors
Total number of factors
kilo Watt
Number of levels
Total number of levels
Number of experiments in an array
Total number of experiments
Variance of factor k at level l
Average of objective function value due to
factor k at level l in each case m
Average of objective function value due to
factor k over all levels L in each case m
Average of objective function value in each
case m
Dk
(DOF)k
k
K
kW
l
L
m
M
m
̅
xm
ACKNOWLEDGMENT
The authors appreciate the financial supports from the
Universiti Teknologi PETRONAS and MOSTI (Science
Fund 030202SF0113).
REFERENCES
[1] M. A. Fahim, T. A. AlSahhaf and A. S. Elkilani, Fundamentals of
Petroleum Processing, 1st ed., Elsevier, Oxford, 2010.
[2] R.Boulet, “Composition of Crude Oil and Petroleum Products”, in
Crude Oil Petroleum Products Process Flow sheets, Ch. 1,
TECHNIP, France, 2001.
[3] J. H. Gary and G. E. Hendwerk, Petroleum Refining: Technology an
Economics, 5th ed., CRC Press, Florida, 2007.
[4] R. E. Young, “Petroleum refining process control and realtime
optimization”, Control Systems, IEEE, 26(6), pp. 7383, Dec. 2006.
[5] B. A. I. Liang, Y. Jiang, D. Huang and X. Liu, “A Novel Scheduling
Strategy for Crude Oil Blending”, Chinese Journal of Chemical
Engineering, 18(5), pp. 77786, Oct. 2010.
[6] G. Robertson, A. Palazoglu and J. A. Romagnoli, “Refining
Scheduling of Crude Oil Unloading, Storing, and Processing
Considering Production Level Cost”, Computer Aided Chemical
Engineering, vol. 28, pp. 115964, 2010.
[7] G. K. D. Saharidis,M. Minoux and Y. Dallery, “Scheduling of
loading and unloading of crude oil in a refinery using eventbased
discrete time formulation”, Chemical Engineering, 33(8), pp. 1413
26, Aug. 2009.
[8] P. Behrenbruch and T.
characterization of crude oils based on distillation properties”,
Dedigama,
“Classification and
Journal of Petroleum Science and Engineering, 57(12), pp. 16680,
May 2007.
[9] T. Chatterjee and D. N. Saraf, “Online estimation of product
properties for crude distillation units”, Journal of Process Control, 1st
Ed., vol. 14, pp. 6177, Feb. 2004.
[10] R. K. More, V. K. Bulasara, R. Uppaluri and V. R.
Banjara,“Optimization of crude distillation system using aspen plus:
Effect of binary feed selection on grassroot design”, Chemical
Engineering Research and Design, 88(2), pp. 12134, Feb. 2010.
[11] D. D. Gonçalves and F. G. Martins,“Dynamic Simulation and
Control: Application to Atmospheric Distillation Unit of Crude Oil
Refinery”, Computer Aided Chemical Engineering, vol. 28, 2010.
[12] S. Motlaghi, F. Jalali and M. Nili Ahmadabadi, “An expert system
design for a crude oil distillation column with the neural networks
model and the process optimization using genetic algorithm
framework”, Expert Systems with Applications, 35(4), pp. 154045,
Nov. 2008.
[13] M. Tran, D. K. Varvarezos and M. Nasir, “The importance of first
principles, modelbased steadystate gain calculations in model
predictive control – a refinery case study”, Control Engineering
Practice, vol. 13, pp. 136982, 2005.
[14] Marvin and S. Rakow, Petroleum Oil Refining. In Fuels and
lubricants handbook: technology, properties, performance, and
testing, Ch.1, ASTM International, U.S.A, 2003.
[15] T.Benali, D. Tondeur and J.N. Jaubert, “An improved crude oil
atmospheric distillation process for energy integration: Part II: New
approach for energy saving by use of residual heat”, Applied Thermal
Engineering, vol. 40, pp. 13244, 2012.
[16] R. K. Roy, Design of Experiments Using the Taguchi Approach: 16
Steps to Product and Process Improvement, John Wiley and Sons,
New York, 2001.
[17] M. Zainal Abidin, and N. Yusoff, “Selection of Optimization
Variables to Minimize Energy Consumption in Crude Distillation
Unit using Taguchi Method”, ICPEAM 2012.
[18] N. Yusoff, M. Ramasamy and S. Yusup,“Taguchi‟s parametric design
approach for the selection of optimization variables in a refrigerated
gas plant”, Chemical Engineering Research and Design, vol. 89, pp.
66575, 2011.
[19] A. B. Engin, O. Özdemir, M. Turan and Z. T. Abdullah, “Color
removal from textile dye bath effluents in a zeolite fixed bed reactor:
Determination of optimum process conditions using Taguchi
method”, Journal of Hazardous Materials, 159(2), pp. 348–53, Nov.
2008.
[20] R. K. Roy, A Primer on the Taguchi Method, Society of
Manufacturing Engineers, Dearborn, Michigan, 1990.
[21] S. Fraley, M. Oom, B. Terrien and J. Zalewski, viewed at 3 July
2012, UMICH, https://controls.engin.umich.edu, 2007.
S. Faizan Ali is an MSc. Research Scholar in the
department of Chemical Engineering, Universiti
Teknologi PETRONAS (UTP), Malaysia since
November 2011. He received his Bachelors degree
in Chemical Engineering from NED University of
Engineering and Technology, Karachi, Pakistan in
2010.Currently, he has been working as a Teaching
Assistant to
Nooryusmiza
Dynamics and Control and Process Optimization courses in the department
of Chemical Engineering, UTP. Previously, he worked as a Process
Engineer for Zishan Engineers Pvt. Ltd., Karachi, Pakistan in the year
2011.
his
Yusoff, for
research supervisor,
Chemical Process
Dr.
Page 11
International Journal of Engineering & Technology IJETIJENS Vol:12 No:06 46
12106063939IJETIJENS © December 2012 IJENS I J E N S
Nooryusmiza Yusoff graduated from Northwestern
University, USA with BSc Degree in Chemical
Engineering (1997) and subsequently became a
member of the American Chemical Engineering
Honors Society “Omega Chi Epsilon”. He received
MSc Degree (2001) from the University of Calgary,
Canada with a thesis on applying geostatistical
analyses in predicting ozone temporal trends. He
obtained PhD (2010) from the Universiti Teknologi PETRONAS (UTP),
Malaysia after completing a research work on the integrated framework of
scheduling and realtime optimization in a large industrial plant. His areas
of research interest centers on process modeling and simulation as well as
process systems engineering.