chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

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

Integrated design of diesel hydrotreating processes

Muhammad Imran Ahmad a,∗ , Nan Zhang b , Megan Jobson b

aDepartment of Chemical Engineering, NWFP University of Engineering and Technology, 25000 Peshawar, Pakistan
bCentre for Process Integration, School of Chemical Engineering and Analytical Science, The University of Manchester,
PO Box 88, Sackville Street, Manchester, M60 1QD, UK

a b s t r a c t

This work presents an integrated approach for the design of diesel hydrotreating processes employing a simu-
lated annealing optimisation algorithm. The modelling of reactor, separation and heat recovery system for diesel
hydrotreating processes is discussed, and a novel optimisation framework is developed for the design of complex
refinery processes. A comparison with conventional approach to process design, i.e. sequential evolution of design,
is given to illustrate the ability of proposed approach to obtain overall hydrotreating process designs with minimum
total annualised costs. The proposed integrated approach takes into account the trade-offs between capital and oper-
ating costs, as well as interactions between the hydrotreater, distillation column, and the associated heat exchanger
network.
© 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Integrated design; Hydrotreating; Process optimisation

1. Introduction
A hydrotreating process consists of three subsystems, i.e. a
reactor, a separation system and an associated heat recovery
system. A typical diesel hydrotreating process flow sheet is
shown in Fig. 1.
The feed to the diesel hydrotreating process is mixed
with hydrogen-rich gas and heated to the reactor operating
temperature. This mixture is fed to the diesel hydrotreater, a
fixed-bed catalytic reactor. The diesel hydrotreating reactions
occurring in the presence of metal-oxide catalyst produce
hydrogen sulphide, ammonia, and saturated hydrocarbons
as the main products (Gary and Handwerk, 2001). The reactor
effluent, consisting of the reaction products and the un-
reacted feedstock along with hydrogen, is cooled before being
sent to the separation system. The separation system of the
diesel hydrotreating process consists of high and low pressure
flash separation units and a distillation column. There are two
high pressure flash separation units namely the hot high pres-
sure separator (HHPS) and the cold high pressure separator
(CHPS). The first high pressure flash separator recovers hydro-
gen and makes a rough split between light and heavy reaction
products. The liquid from this high pressure flash unit is sent
to the low pressure separator and then to the distillation col-
umn for the separation of the final diesel product, taken out as
the bottom product of the distillation column (Meyers, 1997).
The vapour from the first high pressure flash separator is
cooled and fed to the cold high pressure separator after water
injection, in order to absorb hydrogen sulphide, and ammonia.
The liquid light hydrocarbon is separated from the vapour and
liquid water in the cold high pressure separator and sent to
the low pressure separator. The hydrogen-rich gas from this
flash unit is sent to the amine unit for further purification, i.e.
removal of hydrogen sulphide, that was not absorbed in water,
through contact with lean amine solution. The hydrogen-rich
gas after purification is recycled back to the diesel hydrotreater
after necessary compression. A fraction of this gas is purged,
and fresh hydrogen is mixed with the recycle gas, to avoid
build-up of light hydrocarbons and to maintain the desired
hydrogen partial pressure in the reactor.
Design of refinery processes such as diesel hydrotreating
process is conventionally carried out sequentially due to its
hierarchical nature, i.e. the design calculations start with the
reactor moving out to the separation system. The reactor and
separation system, along with recycle, establish the material
and energy balance of the process thereby defining the heating
and cooling duties for the heat recovery system (Smith, 2005).
The overall process design is evolved in this step-by-step

Corresponding author. Tel.: +92 91 9218180; fax: +92 91 9218180.

∗

E-mail address: Imran.Ahmad@nwfpuet.edu.pk (M.I. Ahmad).

Received 11 March 2010; Received in revised form 23 November 2010; Accepted 29 November 2010
0263-8762/$ – see front matter © 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.cherd.2010.11.021
1026 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036

approach. However, the design and operating conditions of

Nomenclature individual components of a process may influence each other
and contribute to the overall performance of the process
a capital cost coefficient (Smith, 2005; Sadighi et al., 2009; Budzianowski, 2010a). For
AF annualisation factor example, recycle of carbon monoxide and water to oxyform-
AHX total area of heat exchangers, m2 ing reactor, after separation of hydrogen and carbon dioxide,
C area cost coefficient for heat exchangers helps overcome the thermodynamic limitations of water
Cm concentration of matrix entries in Eq. (1), mol/L gas shift reaction and improves hydrogen production with
Cni concentration of nitrogen species in Eq. (6) carbon dioxide capture in auto thermal catalytic oxyforming
Cf fixed charge for heat exchanger unit of methane (Budzianowski, 2010b). Such interactions are not
Cost capital or operating cost accounted for fully if a sequential design strategy is followed.
Cp mean molar heat capacity of a stream A novel approach is developed in this paper for integrated
CP stream heat capacity flow rate, kW/K design of diesel hydrotreating processes. In the proposed
Cut cost per energy unit for utility framework, individual sub-systems of the hydrotreating pro-
Ea,g activation energy of hydrodesulphurisation cess namely the reactor, separation and heat recovery systems
reactions in Eq. (3) are handled simultaneously in order to capture the overall
Km adsorption constant of molecule m in Eq. (1), picture and achieve optimal hydrotreating process design.
L/mol
Kni adsorption constant of nitrogen species in Eq.
2. Integrated approach for process design
(5)
kni rate constant of hydrodenitrogenation reaction
This section introduces a novel approach for integrated
in Eq. (5)
design of hydrotreating processes using the molecular type
kop combined rate parameter in Eq. (1), L/kgcat s
and homologous series (MTHS) matrix representation of
Kop equilibrium ratio in Eq. (1)
petroleum fractions. Previous research studies using MTHS
NHX number of heat exchangers
matrix representation (Peng, 1999; Zhang, 1999; Sun, 2004)
PH2 hydrogen partial pressure
have been focused on individual subsystems of refinery pro-
PH2S partial pressure of hydrogen sulphide
cesses such as modelling of reactor or separation systems.
R reflux ratio of distillation column. Also used to
However, the interaction between hierarchical levels of pro-
represent the radius of diesel hydrotreater in
cess design, namely the reactor, separation and heat recovery
Eq. (9)
system has not been explored in previous studies employing
rg rate of reaction of sulphur compound
the molecular type and homologous series matrix represen-
rni rate of hydrodenitrogenation reaction in Eq. (5)
tation. The optimisation framework developed for integrated
rop rate of aromatic hydrogenation reaction in Eq.
design of hydrotreating processes is presented in this paper.
(1), mol/kgcat s
A detailed account of the modelling of individual sub-systems
SRGO straight run gas oil
of diesel hydrotreating processes may be found elsewhere
T temperature, ◦ C
(Ahmad, 2009).
T10 true boiling temperature at which 10% liquid is
vaporised, ◦ C
T50 true boiling temperature at which 50% liquid is 2.1. Reactor design for diesel hydrotreating processes
vaporised, ◦ C
T90 true boiling temperature at which 90% liquid is In this section the model for diesel hydrotreater, employed for
vaporised, ◦ C the integrated design and optimisation of diesel hydrotreat-
TAC total annualised cost ing process is presented. The three main reaction groups
x mole fraction of a component in diesel hydrotreating process are aromatics hydrogenation
Z length of reactor, m (AH), hydrodesulphurisation (HDS), and hydrodenitrogenation
Hr heat of reaction in Eq. (8) (HDN) (Girgis and Gates, 1991).
H˛,g heat of adsorption for hydrodesulphurisation
reactions in Eq. (4) 2.1.1. Aromatic hydrogenation
 catalyst site for hydrogenolysis pathway of Aromatic hydrogenation reactions are reversible and exother-
hydrodesulphurisation reactions mic with equilibrium conversions of hydrocarbons often less
 catalyst site for hydrogenation pathway of than 100% under practical processing conditions (Girgis and
hydrodesulphurisation reactions Gates, 1991). Analysis of petroleum fractions shows that the
aromatic compounds in petroleum mixtures mainly fall into
Subscripts four groups, namely monoaromatics, diaromatics, triaromat-
 catalyst site for hydrogenation pathway of ics and polycyclic aromatics (Stanislaus and Cooper, 1994).
hydrodesulphurisation reactions The polycyclic aromatics with four or more condensed ben-
g sulphur compounds zene rings are usually present in high-boiling fractions (boiling
m molecules in molecular type and homologous range >350 ◦ C), while the first three types are present in mid-
series matrix, also represented by s in Eqs. (7) dle distillates. The reaction networks for aromatic compounds
and (8) of diesel hydrotreating feedstocks are obtained from the lit-
ni nitrogen compounds erature (Girgis and Gates, 1991; Korre et al., 1994, 1995) and
presented in Fig. 2.
The rate of aromatic hydrogenation reactions may be
expressed in a simplified form of a Langmuir–Hinshelwood–
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036 1027

Fig. 1 – A typical diesel hydrotreating process flow sheet (Meyers, 1997).

Hougen–Watson (LHHW) rate expression to account for reac- et al., 1978; Gates et al., 1979; Nag et al., 1979; Vrinat, 1983;
tivity, equilibrium and competitive adsorption of all aromatic Aubert et al., 1986; Van Parijs et al., 1986; Moreau et al., 1987;
hydrocarbons (Korre et al., 1994): Sauer et al., 1989).
A model for hydrodesulphurisation reactions proposed by
kop (Co − Cp /Kop ) Hidalgo-Vivas (1999) is incorporated in the diesel hydrotreater
rop =  (1)
1+ K C
m m m model. The model for hydrodesulphurisation reactions pro-
posed by Hidalgo-Vivas (1999) is derived from the LHHW
where rop (mol/kgcat /s) is the rate of reaction of aromatic formalism, assuming the surface reaction is the rate deter-
molecule o to naphthenic molecule p; Co , Cp , Cm (mol/L) are mining step. The hydrodesulphurisation model assumes
the concentrations of molecules o, p and m, respectively; kop non-competitive adsorption of sulphur containing com-
(L/kgcat /s) is the combined rate parameter; Kop is the equi- pounds and hydrogen, and molecular adsorption of hydrogen:
librium ratio and Km (L/mol) is the adsorption constant of
molecule m, m represents molecule in the MTHS matrix. Reac-
Kg, Cg KH2 , CH2
tion rate parameters for aromatic hydrogenation reactions of rg = kg,  2

the aromatic species existing in the middle distillate fractions 1+ K C + KH2 S CH2 S
g g, g
(1 + KH2 CH2 )
are also available in the literature (Aubert et al., 1988; Moreau
et al., 1990; Korre et al., 1995). Kg, Cg KH2 , CH2
+ kg,    (2)
1+ K C + KAro CAro
g g, g
2.1.2. Hydrodesulphurisation
The sulphur compounds in middle distillates may be clas-
sified into five groups in molecular type and homologous  −E 
series matrix representation (Ahmad et al., 2010): SI to a,g
kg = kog exp (3)
SV. SI homologous series includes mercaptans and similar RT
compounds while the other four groups account for the thio-
phenic sulphur compounds namely alkyl benzothiophenes,
 −H 
a,g
Kg = Kgo exp (4)
dibenzothiophenes and alkyl dibenzothiophenes without sub- RT
stituents at the 4- and 6-positions, alkyl dibenzothiophenes
with only one of substituents at either the 4- or 6-position, and where rg is the rate of reaction of sulphur compound g, Cg , CH2 ,
alkyl dibenzothiophenes with two of the alkyl substituents at CH2 S , CAro , are the concentrations; kg is the rate constant; Ea,g
the 4- and 6-positions (Ma et al., 1994). is the activation energy; Kg , KH2 , KAro , KH2 S are the adsorption
The reaction networks and rate expressions for constants and Ha,g is the heat of adsorption.
hydrodesulphurisation reactions have been investigated
extensively, based on model compounds such as thiophene,
benzothiophene, dibenzothiophene, and isomers of benzon- 2.1.3. Hydrodenitrogenation
aphthothiophene, and are available in the literature (Houalla Nitrogen compounds in middle distillate fractions are present
as heterocyclic aromatic compounds and nonheterocyclic
organonitrogen compounds (Girgis and Gates, 1991). Nonhete-
rocyclic organonitrogen compounds such as anilines, aliphatic
(C8,1A) (C8,1N) amines and nitriles are usually in very minute quantities and
are denitrogenated very rapidly. Thus they are of very lit-
tle importance in hydroprocessing of heavy feeds (Girgis and
(C10,AA) (C10,AN) (C10,NN)
Gates, 1991). The nitrogen heterocyclic compounds are classi-
fied as basic nitrogen compounds such as pyridine, quinoline
and non-basic nitrogen compounds, e.g. indole, and carbazole
(C14,AAA) (C14,AAN) (C14,NAN) (C14,NNN)
(Klein et al., 2006).
Fig. 2 – Aromatic hydrogenation reactions in diesel Mathur et al. (1982) determined the hydrodenitrogenation
hydrotreating process (Sun, 2004). rates of several basic nitrogen heterocyclic compounds. The
1028 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036
hydrodenitrogenation reaction rate can be expressed as:
k Cni ni
rni = ni PH2 ˇ
1+ K C
m ni ni
where rni is the rate of hydrodenitrogenation reaction, kni is
the rate constant, Cni is the concentration of nitrogen species
ni
ni, Kni is the adsorption constant, PH2 ˇ is the hydrogen partial
ni
pressure with ˇ as a coefficient (Sun, 2004).
2.1.4. Cracking reactions
The formation of light gases due to cracking of paraffinic
molecules is taken into account in this work. The production
of light gases is correlated with the concentration of H2 S and
is given as (Hidalgo-Vivas, 1999):
 
1
ln = 4.603 × 10−6 · PH2 S
1 − Xcracking
where Xcracking is the yield of light gas, and PH2 S is the par-
tial pressure of hydrogen sulphide. The composition of light
gas is assumed to be constant due to mild operating condi-
tions generally employed in diesel hydrotreating process. A
detailed account of the cracking model can be found elsewhere
(Hidalgo-Vivas, 1999; Sun, 2004).
2.1.5. Reactor modelling of diesel hydrotreater
A one-dimensional pseudo homogeneous reactor model of
Sun (2004) is employed in this work based on the following
assumptions (Sun, 2004):
1. The reactor operates adiabatically under isobaric and
steady-state conditions.
2. There is no radial profile of concentration within the reac-
tor.
3. The reactor behaves like a plug-flow reactor.
4. Evaporation of diesel oil occurs instantaneously only at
the entrance of the reactor, liquid volume in the reactor
remains constant.
5. The catalyst particles are completely wetted.
The rate of mass transfer between gas, liquid and solid
phases has been ignored as the hydrogenation reaction is
kinetically controlled (Cooper and Donnis, 1996). Also, ade-
quate gas and liquid distribution is assumed for complete
wetting of the catalyst particles as is commonly accepted for
commercial hydroprocessing units (Shah and Paraskos, 1975).
A detailed account of modelling techniques can be found else-
where (Froment and Bischoff, 1990). Both mass and energy
balance are considered:
dCs rs B
= (7)
dz US

dT (−H ) r
=  rss
dz Cps Cs
where rs is the rate of reaction of molecule s, B is the catalyst
bulk density, i.e. catalyst weight per unit volume of reactor,
US is the superficial velocity, CPs is the heat capacity, Cs is the
concentration, z is the reactor length, T is the temperature of
the bed, Hr is the heat of reaction. The differential equations
of pseudo-homogeneous model are integrated numerically
using fourth order Runge–Kutta method (NAG, D02PCF) for the
design of a new reactor.
Table 1 – Economic data of diesel hydrotreater (Sun,
2004).
(5) Atmospheric gas oil 131 $US/m3
Light cycle oil 121 $US/m3
Diesel 141.03 $US/m3
H2 price 0.002 $US/scf
Catalyst cost 9679 $US/day
Cost of catalyst replacement 1,500,000 $US
2.1.6. Estimation of annualised catalyst cost
The catalysts in diesel hydrotreating processes are primarily
de-activated due to carbon containing deposits. Therefore, the
de-activation model estimates the catalyst activity based on
the amount of coke, deposited on the catalyst, calculated as a
function of time and process conditions. The de-activation of
the catalyst in diesel hydrotreater is calculated using the coke
(6) formation model and reactivity-coke relationship developed
by De Jong (1994a,b).
The run length calculation for the diesel hydrotreater oper-
ation, i.e. the estimation of catalyst life, is carried out using
dynamic programming approach for optimisation of catalyst
life of fixed bed reactors (Kuchcinski, 2002). The strategy is
based on the breakdown of catalyst life into fixed interval
operating periods, e.g. two weeks. The diesel hydrotreater
model calculates the reactor temperature needed to satisfy
the product specifications, such as the sulphur content, for
each operating period, starting with a new catalyst. Based on
the reactor temperature and the aromatic content of the feed-
stock, the de-activation model predicts the coke formation
and the de-activation coefficient. This de-activation coeffi-
cient serves to represent the loss in catalyst activity resulting
in decrease in rate of diesel hydrotreating reactions. The cat-
alyst activity for the consecutive period is updated based on
the de-activation coefficient. The reactor operating tempera-
ture for this current operating period is calculated again using
the diesel hydrotreater model discussed. These calculations
are repeated sequentially until the termination criterion for
shutting down the process is satisfied (Sun, 2004). The termi-
nation criterion may be based on an upper limit of the reactor
temperature and/or a lower limit on the catalyst activity.
The cost data for the reactor and catalyst are taken from
Sun (2004) and presented in Table 1. The annualised catalyst
cost is estimated as a function of the size of reactor and the
catalyst life:
1
 
Cat.Cost($US/year) = a × R2 × Z × (9)
Cat.life(year)
where R is the radius of the reactor, Z is the length of reac-
tor, and a = 62018.452 $US/m3 is the catalyst cost coefficient. A
linear correlation is developed for estimation of catalyst life
from the simulation results of the diesel hydrotreater model
for run-length calculation using the method discussed above:
(8)
Cat.life(year) = 9.843 − 0.01172 × T(K) (10)
where T is the operating temperature of the diesel
hydrotreater at the start-of-run, i.e. when the catalyst is 100%
active. The catalyst life depends on the start-of-run temper-
ature of the reactor, which influences the rate of catalyst
de-activation, hydrogen partial pressure in the reactor, and the
aromatic content of the feedstock. The above correlation has
been developed based on the feedstock for diesel hydrotreater,
i.e. straight run gas oil (SRGO), employed in the case study in
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036 1029

correction factor for material, Fp is the correction factor for

Table 2 – Key factors of diesel hydrotreater (Sun, 2004).
pressure.
Catalyst volume (m3 ) 374
The cost of column trays is estimated using the following
Catalyst weight (kg) 330,000
correlation:
Reactor inlet Press. (bar) 34
Purge rate (Nm3 /h) 6400  M&S.Index 
1.55
H2 composition (volume%) 81.6 ColumnTray .Cost(US$) = 4.7(Dcolumn ) HTrays Fc
280
C1 3.5
C2 3.1 (13)
C3 4.1
C4 4.0
C5 2.8
>C5 0.74 Fc = Fs + Ft + Fm (14)
H2 S 0.07

Section 4. The case study in Section 4 is based on the example
case from Sun (2004). Therefore, all key factors in relation to
diesel hydrotreater catalyst such as properties of the catalyst,
void fraction, and hydrogen partial pressure in the reactor are
same as in the work of Sun (2004) and are presented in Table 2.
It is worthwhile to note that the above correlation, i.e. Eqs. (10),
is limited to the specific feed composition investigated in the
case study in Section 4. The reason for this simplification is
to avoid the iterative procedure for the calculation of catalyst
life.
2.2. Separation system design for diesel hydrotreating
process
The separation system of a diesel hydrotreating process may
consist of high and low pressure flash units and a distillation
column. In this work the flash separation is modelled using
the method for calculation of vapour–liquid equilibrium from
Sandler (1989). The size and cost of high and low pressure flash
units have not been taken into account in this framework.
The reason for this simplification is that the operating condi-
tions of flash separation units have been assumed to remain
constant while comparing the proposed integrated design
approach with the conventional sequential design approach.
The amount of diesel feedstock to be treated in the design
calculations also remains constant. Therefore, this simplifi-
cation may be justified for comparing the integrated design
approach developed in this work with the sequential approach
for design of overall diesel hydrotreating process.
The distillation column is modelled using the
Fenske–Underwood–Gilliland (FUG) short-cut method based
on the models developed by Fenske (1932), Gilliland (1940),
and Underwood (1948). The capital cost of the distillation
column is calculated based on the cost of the column shell
and the cost of column trays. The cost correlations have been
taken from Douglas (1988).
The cost of column shell is calculated using the following
equation:
ColumnShell .Cost(US$)
 M&S.Index 
1.066 0.802
= 101.9(DColumn ) (HColumn )
280
Fc = Fm · Fp
where M&S.Index represents the Marshall and Swift cost index,
DColumn is the diameter of the column in ft, HColumn is the height
of the column in ft, Fc is the cost correction factor, Fm is the
where HTrays is the height from top to the bottom tray in ft,
Fs is the correction factor for tray spacing, Ft is the correction
factor for tray type.
The height of the distillation column is calculated based on
the number of trays and the tray spacing:
Hcolumn = (Nreal − 1) · htray + htop + hbottom (15)
where Nreal is the number of real trays.
The tray spacing, top and bottom spacing used are:
• htray = 0.46 m
• htop = hbottom = 2.0 m.
The number of real stages is calculated based on the theo-
retical number of trays and tray efficiency:
Ntheoretical
Nreal = (16)
efficiencytray
The tray efficiency is calculated using the following correlation
(Douglas, 1988):
efficiencytray = 101.6975−0.252 log(·˛) (17)
where  is the viscosity of feed in cp, and ˛ is the average
relative volatility between the top and bottom section.
The diameter of the distillation column is calculated based
on the vapour flow and the vapour velocity in the column:

4V
Dcolumn = (18)
 · V · (1 − KXS)
where V is the vapour flow, V is the vapour velocity, and KXS
is the area of down comer as a fraction of the cross sectional
area of the column (usually 12%).
The vapour velocity is calculated from the maximum allow-
able vapour velocity calculated using the following correlation
(Sounders and Brown, 1934):

L − V
V,Max = KV · (19)
V
(2.18 + Fc ) (11) where KV is the empirical constant.
KV = 0.243 for the tray spacing chosen. The vapour veloc-
ity is assumed to be 70% of the maximum vapour velocity
(Triantafyllou, 1991). The diameter of the column is calculated
(12) for the top and bottom section and the maximum value is used
in cost estimation.
The short-cut model employed in this work requires
key components and their recoveries to be specified for a
desired separation. However, in refinery processes, such as
1030 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036
diesel hydrotreating process, the products of distillation pro-
cess are specified in terms of the boiling range and bulk
physical properties (Watkins, 1979). In order to model the
distillation column of a diesel hydrotreating process, using
short-cut models, the product specifications in terms of boil-
ing curve temperatures need to be translated in terms of key
components and their recoveries. Chen (2008) developed a
methodology for specifying boiling curve temperatures, and
flow rates for products of a distillation column. This method-
ology employs the Levenberg–Marquardt algorithm (Gill and
Murray, 1978). The methodology developed by Chen (2008)
treats key components and recovery of key components as
optimisation variables, i.e. degrees of freedom, and not as
specifications for a desired separation. A detailed account of
the methodology for specifying boiling curve temperatures of
products of distillation can be found elsewhere (Chen, 2008).
This methodology is employed along with the short-cut model
for distillation so that the diesel product, i.e. bottom product of
distillation column may be specified in terms of boiling range
and flow rate.
2.3. Heat recovery system design for diesel
hydrotreating process
Heat exchanger networks provide a means of recovering the
heat available from hot process streams and transferring to
cold process streams, thereby reducing the heating and cool-
ing duties of external utilities such as steam and cooling water.
The heating and cooling duties of the heat recovery system
depend on the design and operating conditions of reactor and
separation system.
Capital and operating costs may be estimated for a given
heat recovery system using the material and energy balances
of process streams. These estimates provide an opportunity
to screen design options conveniently without the actual
design of heat exchanger networks (Smith, 2005). Pinch anal-
ysis (Linnhoff and Hindmarsh, 1983) is a widely used tool for
estimation of minimum hot and cold utility demand, mini-
mum number of units and minimum investment cost for a
given set of process streams. In this paper, capital and energy
cost targeting for a heat recovery system is briefly introduced
for the benefit of understanding the integrated approach for
design of diesel hydrotreating process.
The stream data for hot and cold process streams of a
heat recovery system depends on the reactor and separation
system design. This stream data, i.e. stream temperatures,
flow rates and enthalpies, are used to estimate the minimum
hot and cold utility requirement using the pinch analysis. In
order to estimate the capital cost of the heat recovery sys-
tem, the number of heat exchangers and the heat transfer area
required for the heat recovery system needs to be estimated.
The minimum number of heat exchangers required for the
heat recovery system is estimated from the number of process
streams and the pinch point (Hohman, 1971; Linnhoff et al.,
1979). The heat transfer area required for a heat recovery sys-
tem is estimated assuming vertical heat transfer (Townsend
and Linnhoff, 1984; Linnhoff and Ahmad, 1990).
The integrated approach for the design of diesel hydrotreat-
ing processes developed in this work is the first of its kind
using the molecular type and homologous series matrix repre-
sentation. Therefore, a methodology is needed to translate the
detailed molecular level information employed in the diesel
hydrotreater model to bulk macroscopic thermal properties,
such as heat capacities of process streams, required in the
design and the cost estimation of the heat recovery system.
The heat capacities of diesel hydrotreater streams are calcu-
lated using the heat capacities and mole fractions of individual
components:
n

Cp = xi · Cpi (20)
i=1
where Cp is the heat capacity of the stream, xi and Cpi rep-
resent the mole fraction and heat capacity of component i,
respectively. The individual component heat capacities are
estimated using group contribution methods, e.g. Joback and
Reid (1987). Pinch analysis is based on the assumption of
constant thermal properties. Therefore, mean molar heat
capacities are calculated for diesel hydrotreater streams:
T2
CpdT
Cp =
T1
(21)
T2 − T1
where Cp is the mean molar heat capacity of a stream.
2.4. Optimisation framework for integrated process
design
The optimisation of the design of diesel hydrotreating pro-
cesses is a complex problem due to the simultaneous handling
of different non-linear models such as the diesel hydrotreater
model, the short-cut distillation model, and the flash model
for high pressure and low pressure flash separation. A
Stochastic optimisation method is applied for the optimi-
sation of the design of diesel hydrotreating process due
to the non-linear nature and complexity of the problem.
Stochastic optimisation techniques have higher probability of
escaping local optima and finding solutions in the vicinity
of the global optimum for complex non-linear and non-
convex problems compared to deterministic methods (Arora,
2004).
Simulated annealing is implemented in this framework for
optimisation of the integrated design of diesel hydrotreating
process because of its successful applications in a wide range
of problems relating to process synthesis and optimisation
(e.g. Dolan et al., 1989; Floquet et al., 1994; Athier et al., 1996,
1997; Linke, 2001). The optimisation algorithm makes use
of the values of the objective function at randomly chosen
points in the search space. It is worth mentioning here that
one of the drawbacks of simulated annealing is the higher
computation time required to search for solutions in the
vicinity of global optimum. The higher computation time of
simulated annealing optimisation may be attributed to two
factors. Firstly, the random nature of search requires higher
number of evaluations of objective function. Secondly, black-
box treatment in simulated annealing algorithm may also
result in excessive computation time, i.e. the time required
for solving complex models after each perturbation by the
optimisation algorithm (Chen, 2008). This drawback can be
a limiting factor in application of simulated annealing in
treating larger problems. The simulated annealing algorithm
is discussed in detail in Section 3.
3. Simulated annealing algorithm
Simulated annealing was first employed for the optimisa-
tion of combinatorial problems by Kirkpatrick et al. (1983).
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036
Fig. 3 – Simulated annealing algorithm (Rodriguez, 2005).
The simulated annealing algorithm of Kirkpatrick et al. (1983)
employed in this work is shown in Fig. 3.
The optimisation process starts with an initial guess repre-
sented as a feasible initial solution in Fig. 3. This initial design
is modified by a random change, termed as random move.
The objective function, i.e. total annualised cost of a design, is
evaluated against that of the previous trial solution. Any ran-
dom change, made to a current design, may be accepted or
rejected based on the acceptance criterion. This procedure is
carried out a number of times (N) as specified by the designer
through a parameter known as Markov chain length (Nmax ).
The annealing temperature, a control parameter for the opti-
misation algorithm, is reduced after a number of candidate
designs have been obtained. This cycle is continued until the
termination criterion is satisfied.
The constraints for the optimisation of integrated design
of diesel hydrotreating process are the modelling equations
for the reactor, the de-activation model, equations for estima-
tion of catalyst life, short-cut distillation model, equations for
flash calculations, and the equations for modelling of the heat
recovery system.
3.1. Simulated annealing parameters
Simulated annealing parameters govern the optimisation pro-
cess and effect the performance of an optimisation algorithm.
The annealing temperature, acceptance criterion, Markov
1031
chain length and termination criterion constitute the simu-
lated annealing parameters. Metropolis acceptance criterion
(Metropolis et al., 1953) is employed in this work. A detailed
discussion on simulated annealing parameters can be found
elsewhere (Rodriguez, 2005; Chen, 2008; Ahmad, 2009).
3.2. Simulated annealing moves
The modifications or changes made by the simulated anneal-
ing algorithm to a current design are known as random
moves. The random moves made by the optimisation algo-
rithm depend on the nature of the optimisation problem and
the variables involved. Table 3 presents a list of simulated
annealing moves for modifications to the integrated design
of diesel hydrotreating process along with the range for each
variable searched by the optimisation algorithm. The variables
presented in Table 3 serve as the degrees of freedom in the
optimisation of integrated design of diesel hydrotreating pro-
cess.
The integrated design problem consists of two sub-
systems: the diesel hydrotreater and the separation system.
The heat recovery system depends on the design of the diesel
hydrotreater and the separation system, since only targeting
of heat recovery system is employed and not the actual heat
exchanger network design. The move probabilities are clas-
sified into two groups based on these sub-systems with the
probability p for a diesel hydrotreater move and 1 − p for dis-
1032 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036

Table 3 – Simulated annealing moves for integrated design of diesel hydrotreating process.
Integrated design moves for diesel hydrotreating process

Diesel hydrotreater move (p) Distillation column move (1 − p)

Reactor operating temperature move (320–380 C) ◦

Column feed temperature move (100–370 ◦ C)
Hydrogen partial pressure move Column operating pressure move (1–10 bar)
R/Rmin of column move (1.05–2)
Key components move
Key components recovery move (60–99.7%)

Table 4 – Properties of feedstock of diesel hydrotreater Table 6 – Operating conditions of the flash separation
for case study on design of diesel hydrotreating process. units of diesel hydrotreating process.
Feed SRGO Operating conditions of flash separation units

Density (kg/m3 ) 830 High Low

Distillation (ASTM D86) pressure pressure
Initial boiling point 155 flash unit flash unit
0.1 230
0.3 260 Temperature (◦ C) 50.0 50.0
0.5 275 Pressure (bar) 32.0 8.0
0.7 300
0.9 325
Final boiling point 350 heat exchanger network, CostColumn is the capital cost of dis-
Total sulphur (wt%) 0.3 tillation column, Cat.Cost is the catalyst cost calculated using
Eq. (9), Cut is the operating costs of utilities based on the load,
and QUt is the total load, i.e. the heating and cooling provided
tillation column move. The probabilities for random moves by steam and cooling water in distillation column and heat
of individual optimisation variables provide flexibility to the exchanger network.
optimisation procedure and are specified on the basis of the
influence of a design variable on the overall performance of 4. Case study
the process (Chen, 2008).

A case study is presented in this section to illustrate the appli-

3.3. Objective function for integrated design of diesel
cation and benefits of the proposed integrated approach for
hydrotreating process
the design of diesel hydrotreating processes. The objective of
the case study is to design the diesel hydrotreating process for
The objective function for the optimisation of integrated
a given feedstock satisfying the commercial product specifica-
design of diesel hydrotreating process is the total annualised
tions for diesel fuel. The case study is taken from Sun (2004) for
cost. The total annualised cost of the process consists of the
a diesel hydrotreating process using straight run gas oil (SRGO)
sum of all annualised capital costs and operating costs. The
as the diesel hydrotreater feed. The properties of the feedstock
capital costs include the catalyst cost of the reactor, invest-
are shown in Table 4. The sulphur content of the diesel product
ment costs of the distillation column and heat exchangers
is specified to be at most 10 ppm, i.e. ultra-low sulphur diesel.
while the operating costs account for the energy costs. The
The specifications for the bottom product of the distillation
energy costs consist of the cost of hot utility and cold util-
column, i.e. the diesel stream are shown in Table 5.
ities, i.e. steam and cooling water respectively, required by
The cost data for the heat recovery system are taken from
the distillation column and heat exchanger network of the
Verheyen and Zhang (2006). The heat recovery system target-
diesel hydrotreating process. The objective function may be
expressed as:
Table 7 – Sequential design of diesel hydrotreating
TAC = AF · Cf · NHX + AF.C.AHX + AF · CostColumn + Cat.Cost process.
Diesel hydrotreater Sequential Integrated
+ Cut · QUt (22) approach approach

where TAC is the total annualised cost of the overall process, Reactor operating 322.52 324.74
temperature (◦ C)
AF is the annualisation factor, Cf is the fixed cost of heat
Length of reactor (m) 11.522 11.522
exchanger units, NHX is the number of heat exchanger units, C Distillation column
is the area cost coefficient, AHX is the total area required by the Column pressure (bar) 5.926 1.296
Reflux ratio 1.795 1.064
Light key recovery (%) 86.95 74.44
Table 5 – Bottom product specifications for the
Heavy key recovery (%) 61.09 64.86
distillation column of diesel hydrotreating process.
Total number of stages 44 16
Flow rate specification (kmol/s) 0.127 Column feed temperature 232.43 152.56
True boiling point specification (◦ C) (◦ C)
T10 201.10 Feed pre-heat duty (kW) 2417 1322
T50 251.70 Total condenser duty (kW) 6090 5343
T90 310.70 Reboiler duty (kW) 7311 6294
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036 1033

Table 8 – Stream data of diesel hydrotreating process from sequential design approach.
Stream Inlet temperature Outlet Enthalpy
(◦ C) temperature (◦ C) difference (kW)

Reactor feed 25.0 322.52 33240.9

Reactor product 322.52 50.0 −29724.6
Column feed 50.0 232.43 2416.93
Condenser stream 238.95 230.48 −6089.88
Distillate product 230.48 50.0 −764.782
Reboiler stream 328.50 340.57 7310.74
Bottom product 340.57 25.0 −3115.08

Table 9 – Comparison of cost breakdown of integrated Table 11 – Design results obtained using integrated
design with sequential design for diesel hydrotreating approach with an additional constraint on distillation
process. column operating pressure.
Sequential Integrated Diesel hydrotreater
design design
Reactor operating temperature (◦ C) 324.74
Catalyst cost (£/year) 2,733,000 2,758,000 Length of reactor (m) 11.522
Column capital cost 366,000 182,000 Distillation column
(£/year) Column pressure (bar) 5.0
Heat exchanger area 1,209,000 1,277,000 Reflux ratio 1.428
cost (£/year) Light key recovery (%) 60.53
Energy cost (£/year) 978,000 911,000 Heavy key recovery (%) 63.18
Total annualised cost 5,287,000 5,128,000 Total number of stages 14
(£/year) Column feed temperature (◦ C) 167.26
Feed pre-heat duty (kW) 1515
Total condenser duty (kW) 5254
Reboiler duty (kW) 7164

ing is carried out for a minimum approach temperature of

20 ◦ C. The operating conditions of the high pressure and low
pressure flash separation units are fixed for both the design
approaches and correspond to the range of operating condi-
tions in commercial hydroprocessing units (Meyers, 1997). The
operating conditions of the flash separation units are shown
in Table 6.
The sequential approach to process design is taken as
the base case in this case study for evaluation and compar-
ison of the novel integrated approach for design of diesel
hydrotreating process developed in this work. For the sequen-
tial evolution of design of diesel hydrotreating process, first
a quadratic interpolation method is used to minimise the
diesel hydrotreater cost (NAG, E04ABF). The separation system
is designed for the reactor products using COLOM software
version 2.1 minimising the capital cost of distillation column.
The simulated annealing algorithm is used for optimisation
of capital cost of distillation column. Finally, the capital and
energy costs are estimated using the methods discussed in
Section 2.3. The major features of the overall process design
obtained from the sequential approach for the system under
consideration are shown in Table 7. The process stream data
of this design are shown in Table 8.
It can be seen from Table 7 that in sequential approach for
design the start-of-run temperature, i.e. the reactor operating
temperature at the beginning of catalyst life, is as low as pos-
sible while satisfying the product specifications, e.g. sulphur
content. In this way a maximum catalyst life and conse-
quently, a minimum catalyst cost is ensured. The capital cost
of the distillation column is then minimised while satisfying
the flow rate and boiling range specification of diesel product.
However, this sequential approach of process design neglects
the interactions of individual subsystems of the overall pro-
cess. For example, the reactor operating temperature may
affect the product composition, boiling range and enthalpies
of the product streams. These factors in turn may influence
the capital and operating costs of the distillation column and
the heat recovery system.
The cost breakdown of the design of diesel hydrotreat-
ing process obtained using the sequential design approach
is shown in Table 9. The currency exchange rate, for con-
version of US dollars to British pounds, used in this work is
1 GBP = 0.666 $US.
The major features of the design, obtained using the pro-
posed integrated approach, are also shown in Table 7. The

Table 10 – Stream data of diesel hydrotreating process from integrated approach for design of diesel hydrotreating
process.
Stream Inlet temperature Outlet Enthalpy
(◦ C) temperature (◦ C) difference (kW)

Reactor feed 25.0 324.74 33528.9

Reactor product 324.74 50.0 −29999.7
Column feed 50.0 152.56 1321.66
Condenser stream 167.26 156.44 −5342.71
Distillate product 156.44 50.0 −409.193
Reboiler stream 238.41 251.48 6293.74
Bottom product 251.48 25.0 −2124.32
1034 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036

Table 12 – Stream data of diesel hydrotreating process from integrated approach with an additional constraint on
distillation column operating pressure.
Stream Inlet temperature Outlet Enthalpy
(◦ C) temperature (◦ C) difference (kW)

Reactor feed 25.0 324.74 33528.9

Reactor product 324.74 50.0 −29999.7
Column feed 50.0 167.26 1515.49
Condenser stream 228.66 219.87 −5254.02
Distillate product 219.87 50.0 −682.627
Reboiler stream 313.59 326.73 7164.14
Bottom product 326.73 25.0 −2992.05

Table 13 – Comparison of cost breakdown of integrated design, with an additional constraint on distillation column
operating pressure, with sequential design for diesel hydrotreating process.
Sequential Integrated design
design (pressure
constraint)

Catalyst cost (£/year) 2,733,000 2,758,000

Column capital cost (£/year) 366,000 198,000
Heat exchanger area cost (£/year) 1,209,000 1,205,000
Energy cost (£/year) 978,000 980,000
Total annualised cost (£/year) 5,287,000 5,141,000

process stream data of this design are shown in Table 10. The
computation time for the design, using simulated annealing
optimisation is 1 h, 2 min, and 13 CPU seconds (Pentium® 4
CPU 3.00 GHz processor and 512 MB RAM). The cost break-
down of the design obtained using the proposed integrated
approach is shown in Table 9 in comparison with the design
obtained using the sequential approach.
The design obtained using the integrated approach pro-
vides a saving of 0.16 MM £/year corresponding to 3% reduction
compared to the design obtained using sequential approach.
In the integrated design the distillation column satisfies the
bottom product specification while operating at a lower pres-
sure and reflux ratio. Therefore, the column capital cost and
energy costs are lower compared to the design obtained using
sequential approach to process design.
Another scenario may be analysed for the given system as
it may be argued that the compression costs would be higher
for downstream processing of the distillation column products
for the design obtained using the novel integrated approach.
In order to account for this scenario, an additional constraint
of minimum operating pressure of the distillation column of
5 bar is enforced on the integrated design framework.
The major features of the design, obtained using the
integrated approach with an additional constraint on the
operating pressure of the distillation column, are shown in
Table 11. The process stream data of the design for this new
scenario are shown in Table 12. The computation time for
this second design, obtained using the proposed integrated
approach is 2 h, 31 min, and 50 CPU seconds (Pentium® 4 CPU
3.00 GHz processor and 512 MB RAM).
The cost breakdown of the design obtained using the pro-
posed integrated approach for this new scenario is shown in
Table 13 in comparison with the design obtained using the
sequential approach. The design obtained using the proposed
integrated approach with an additional constraint on the oper-
ating pressure of distillation column provides a saving of 0.15
MM £/year corresponding to 2.8% reduction compared to the
design obtained using sequential approach. It can be seen
from Table 11 that the start-of-run temperature of the diesel
hydrotreater is the same as in the design, obtained using the
integrated approach, without the pressure constraint for dis-
tillation column. However, the energy cost is increased due to
higher reflux ratio resulting in slightly lower savings compared
to the original integrated design.
The savings suggested in the above case study may require
critical analysis as the cost calculations, and models employed
in this work remain subject to a certain degree of inaccu-
racy. The annealing schedule and random nature of search
in simulated annealing also contributes to obtaining a range
of solutions. However, the focus of this paper is to present
a novel approach to account for the interactions of sub-
systems in the design of overall diesel hydrotreating process.
These issues may be addressed in future extensions to
this integrated approach for design of diesel hydrotreating
processes.
5. Conclusions
The approach developed for integrated design of diesel
hydrotreating processes helps obtain designs with lower total
annualised costs as illustrated by the case study. The novel
integrated approach developed in this work for design of diesel
hydrotreating processes takes into account the interactions of
individual subsystems, i.e. the diesel hydrotreater, the sepa-
ration system, and the heat recovery system. The proposed
framework may serve the purpose of analysing the effects of
modifications to the operating conditions of individual sub-
systems, e.g. diesel hydrotreater, and distillation column, on
the overall costs of diesel hydrotreating process at conceptual
design stage.
References
Ahmad, M.I., 2009. Integrated and multi-period design of diesel
hydrotreating process. PhD Thesis, The University of
Manchester, Manchester, UK.
Ahmad, M.I., Zhang, N., Jobson, M., 2010. Molecular
components-based representation of petroleum fractions.
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036
Chemical Engineering Research and Design,
doi:10.1016/j.cherd. 2010.07.016.
Arora, J.S., 2004. Introduction to Optimum Design, 2nd Edition.
Academic Press.
Athier, G., Floquet, P., Pibouleau, L., Domenech, S., 1996.
Optimization of heat exchanger networks by coupled
simulated annealing and NLP procedures. Computers &
Chemical Engineering 20, S13–S18.
Athier, G., Floquet, P., Pibouleau, L., Domenech, S., 1997. Process
optimization by simulated annealing and NLP procedures.
Application to heat exchanger network synthesis. Computers
& Chemical Engineering 21, S475–S480.
Aubert, C., Durand, R., Geneste, P., Moreau, C., 1986.
Hydroprocessing of dibenzothiophene, phenothiazine,
phenoxanthiin, thianthrene, and thioxanthene on a sulfided
NiO–MoO3 /Gamma-Al2 O3 catalyst. Journal of Catalysis 97,
169–176.
Aubert, C., Durand, R., Geneste, P., Moreau, C., 1988. Factors
affecting the hydrogenation of substituted benzenes and
phenols over a sulphided NiO–MoO3 /Gamma-Al2 O3 catalyst.
Journal of Catalysis 112, 20.
Budzianowski, W.M., 2010a. Mass-recirculating systems in CO2
capture technologies: a review. Recent Patents on Engineering
4, 15–43.
Budzianowski, W.M., 2010b. An oxy-fuel mass-recirculating
process for H2 production with CO2 capture by autothermal
catalytic oxyforming of methane. International Journal of
Hydrogen Energy 35, 7454–7469.
Chen, L., 2008. Heat-integrated crude oil distillation system
design. PhD Thesis, The University of Manchester,
Manchester, UK.
Cooper, B.H., Donnis, B.B.L., 1996. Aromatic saturation of
distillates: an overview. Applied Catalysis 137, 203.
COLOM, Version 2.1. Centre for Process Integration, The
University of Manchester, Manchester, UK.
De Jong, K.P., 1994a. Effects of vapour-liquid equilibria on coke
deposition in trickle-bed reactors during heavy oil processing.
1. Experimental results. Industrial & Engineering Chemistry
Research 33, 821–824.
De Jong, K.P., 1994b. Effects of vapour–liquid equilibria on coke
deposition in tricke-bed reactors during heavy oil processing.
2. Modelling. Industrial & Engineering Chemistry Research 33,
3141–3145.
Dolan, W.B., Cummings, P.T., Levan, M.D., 1989. Process
optimisation via simulated annealing: application to network
design. AIChE Journal 35, 725–736.
Douglas, J.M., 1988. Conceptual Design of Chemical Processes.
McGraw-Hill.
Fenske, M.R., 1932. Fractionation of straight-run Pennsylvania
gasoline. Industrial and Engineering Chemistry 24, 428.
Floquet, P., Pibouleau, L., Domenech, S., 1994. Separation
sequence synthesis: how to use simulated annealing
procedure. Computers & Chemical Engineering 18,
1141–1148.
Froment, G., Bischoff, K., 1990. Chemical Reactor Analysis and
Design, 2nd Edition. John Wiley & Sons, Singapore.
Gary, J., Handwerk, G., 2001. Petroleum Refining Technology and
Economics, 4th Edition. Marcel Dekker Inc., New York.
Gates, B.C., Katzer, J.R., Schuit, G.C.A., 1979. Chemistry of
Catalytic Processes. McGraw-Hill, New York.
Gill, P.E., Murray, W., 1978. Algorithms for the solution of the
non-linear least-squares problem. SIAM Journal on Numerical
Analysis 15, 977–992.
Gilliland, E.R., 1940. Multi-component rectification: estimation of
the number of theoretical plates as a function of the reflux
ratio. Industrial and Engineering Chemistry 32, 1220.
Girgis, M.J., Gates, B.C., 1991. Reactivities, reaction networks, and
kinetics in high pressure catalytic hydroprocessing. Industrial
& Engineering Chemistry Research 30, 2021.
Hidalgo-Vivas, A., 1999. Reactive distillation for distillates
hydrotreatment. PhD Thesis, UMIST, Manchester, UK.
Hohman, E.C., 1971. Optimum networks for heat exchange. PhD
Thesis, University of Southern California.
1035
Houalla, M., Broderick, D.H., Sapre, A.V., Nag, N.K., de Beer, V.H.J.,
Gates, B.C., Kwart, H., 1978. Hydrodesulphurisation of
methyl-substituted dibenzothiophenes catalysed by
sulphided CoMo/Gamma-Al2 O3 , the reaction network. AIChE
Journal 24, 1015.
Joback, K.G., Reid, R.C., 1987. Estimation of pure-component
properties from group-contributions. Chemical Engineering
Communications 57, 233–243.
Kirkpatrick, S., Gellat, C.D., Vecchi, M.P., 1983. Optimisation
by simulated annealing. Science 220 (4598),
671–680.
Klein, M.T., Hou, G., Bertolacini, R.J., Broadbelt, L.J., Kumar, A.,
2006. Molecular Modelling in Heavy Hydrocarbon
Conversions. Taylor & Francis.
Korre, S.C., Klein, M.T., Quann, R.J., 1995. Polynuclear aromatic
hydrocarbons hydrogenation. 1. Experimental reaction
pathways and kinetics. Industrial & Engineering Chemistry
Research 34, 101–117.
Korre, S.C., Neurock, M., Klein, M.T., Quann, R.J., 1994.
Hydrogenation of polynuclear aromatic hydrocarbons. 2.
Quantitative structure/reactivity correlations. Chemical
Engineering Science 49, 4191.
Kuchcinski, G., 2002, September. Dynamic programming
determines optimum catalyst life strategy. Oil & Gas Journal.
Linke, P., 2001. Reaction and separation process integration. PhD
Thesis, UMIST, Manchester, UK.
Linnhoff, B., Ahmad, S., 1990. Cost optimum heat exchanger
networks. 1. Minimum energy and capital using simple
models for capital cost. Computers and Chemical Engineering
14, 729.
Linnhoff, B., Hindmarsh, E., 1983. The pinch design method for
heat-exchanger networks. Chemical Engineering Science 38,
745–763.
Linnhoff, B., Mason, D.R., Wardle, I., 1979. Understanding heat
exchanger networks. Computers and Chemical Engineering 3,
295.
Ma, X., Sakanishi, K., Mochida, I., 1994. Hydrodesulphurisation
reactivites of various sulphur compounds in diesel fuel.
Industrial & Engineering Chemistry Process Design and
Development 33, 218.
Mathur, K.N., Schrenk, M.D., Kwart, H., and Katzer, J.R., 1982.
Development of unique catalysts for hydrodenitrogenation of
coal-derived liquids. Final report for the period September 15,
1978 to September 1981. Prepared for office of fossil energy:
Department of Energy, Washington, DC.
Metropolis, N., Rosenbluth, A.W., Rosenbluth, M.N., Teller, A.H.,
Teller, E., 1953. Equation of state calculations by fast
computing machines. Journal of Chemical Physics 21,
1087–1091.
Meyers, R.A., 1997. Handbook of Petroleum Refining Processes.
McGraw-Hill.
Moreau, C., Durand, R., Geneste, P., Olive, J., Bachelier, J., Cornet,
D., Duchet, J.C., Lavalley, J.C., Bonnelle, J.P., Grimplot, J.,
Kasztelan, S., Payen, E., Breysse, M., Leclerq, C., De Mourgues,
L., Portefaix, J.L., Vrinat, M., Engelhard, P., Gueguen, C.,
Toulhoat, H., 1987. Chemical evidence for the existence of two
types of catalytic sites for hydroprocessing of substituted
benzenes over NiW–(Mo)/Gamma-Al2O3 hydrotreating
catalyst. Abstracts of Papers of the American Chemical
Society 193, 14, PETR.
Moreau, C., Joffre, J., Saenz, C., Geneste, P., 1990. Hydroprocessing
of substituted benzenes over a sulphided
CoO–MoO3 /Gamma-Al2 O3 catalyst. Journal of Catalysis 122,
451.
Nag, N.K., Sapre, A.V., Broderick, D.H., Gates, B.C., 1979.
Hydrodesulfurization of polycyclic aromatics catalyzed by
sulfided CoO–MoO3 /Gamma-Al2 O3 : the relative reactivities.
Journal of Catalysis 57, 509–512.
Peng, B., 1999. Molecular modeling of refinery processes. PhD
Thesis, UMIST, Manchester, UK.
Rodriguez, C.A., 2005. Fouling mitigation strategies for heat
exchanger networks. PhD Thesis, The University of
Manchester, UK.
1036 chemical engineering research and design 8 9 ( 2 0 1 1 ) 1025–1036
Sadighi, S., Mohaddecy, S.R.S., Ghabouli, O., Bahmani, M., 2009.
Revamp of naphtha hydrotreating process in an Iranian
refinery. Petroleum & Coal 51 (1), 45–50.
Sandler, I.S., 1989. Chemical and Engineering Thermodynamics.
John Wiley & Sons.
Sauer, N.N., Markel, E.J., Scharder, G.L., Angelici, R.J., 1989. Studies
of the mechanism of thiophene hydrodesulphurisation:
conversion of 2,3 and 2,5-dihydrothiophene and model
organometallic compounds. Journal of Catalysis 117,
295.
Shah, Y.T., Paraskos, J.A., 1975. Criteria for axial dispersion effects
in adiabatic trickle bed hydroprocessing reactors. Chemical
Engineering Science 30, 1168.
Smith, R., 2005. Chemical Process Design and Integration. John
Wiley & Sons.
Sounders, M., Brown, G., 1934. Design of fractionating columns I.
Entrainment and capacity. Industrial & Engineering
Chemistry 26, 98.
Stanislaus, A., Cooper, B.H., 1994. Aromatic hydrogenation
catalysis: a review. Catalysis Reviews: Science & Engineering
36, 75–132.
Sun, J., 2004. Molecular modelling and integration analysis of
hydroprocessors. PhD Thesis, University of Manchester, UK.
Townsend, D.W., Linnhoff, B., 1984. Surface area targets for heat
exchanger networks. In: IChemE Annual Research Meeting ,
Bath, UK.
Triantafyllou, C., 1991. The design optimisation and integration
of dividing wall distillation columns. PhD Thesis, UMIST,
Manchester, UK.
Underwood, A.J.V., 1948. Fractional distillation of
multi-component mixtures. Chemical Engineering Progress
44, 603.
Van Parijs, I.A., Hosten, L.H., Froment, G.F., 1986. Kinetics of
hyrodesulphurisation on a CoMo/Al2 O3 catalyst. 2. Kinetics of
the hydrogenolysis of dibenzothiophene. Industrial &
Engineering Chemistry Product Research & Development 25,
437.
Verheyen, W., Zhang, N., 2006. Design of flexible heat exchanger
network for multi-period operation. Chemical Engineering
Science 61, 7730–7753.
Vrinat, M.L., 1983. The kinetics of the hydrodesulphurisation
process: a review. Applied Catalysis 136, 137.
Watkins, R.N., 1979. Petroleum Refinery Distillation. Gulf
Publishing Company, TX, USA.
Zhang, Y., 1999. Molecular modelling of petroleum processes.
PhD Thesis, UMIST, Manchester, UK.