Increasing Operational Efficiency Through The Integration of An Oil Refinery and An Ethylene Production Plant

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

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

Increasing operational efficiency through the


integration of an oil refinery and an ethylene
production plant

Elham Ketabchi ∗ , Evgenia Mechleri, Harvey Arellano-Garcia


Department of Chemical and Process Engineering, University of Surrey, Guildford, GU27XH, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: In this work, the optimal integration between an oil refinery and an ethylene production
Received 11 February 2019 plant has been investigated. Both plants are connected using intermediate materials aim-
Received in revised form 31 July 2019 ing to remove, at least partially, the reliance on external sourcing. This integration has been
Accepted 23 September 2019 proven to be beneficial in terms of quality and profit increase for both production systems.
Available online 1 October 2019 Thus, three mathematical models have been formulated and implemented for each plant
individually as well as for the integrated system as MINLP models aiming to optimise all
Keywords: three systems. Moreover, a case study using practical data is presented to verify the fea-
Integration sibility of the integration within an industrial environment. Promising results have been
Optimisation obtained demonstrating significant profit increase and enhanced operability in both plants.
Modelling © 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Oil refinery
Petrochemical plant
Ethylene production plant

1. Introduction industry while also investigating solutions to maintain the industry’s


operation.
The importance of petroleum refining and refineries are shown in our Moreover, fluctuations in oil refinery and petrochemical plants’
everyday lives in various forms such as transportation, usage in house- profitability has led these industries to look for new ways to maintain
hold and commercial products and even in pharmaceuticals. These their profit while staying competitive.
industries are mainly dependant on crude oil as it is their main feed, Taking all of this into account, to maintain the operability of refiner-
therefore, any change in crude oil price will have a direct impact on ies, improve profit margins and product quality while also considering
these industries. An increase in feed price encourages the industry environmental regulations, one valid approach represents the integra-
to develop new ways to increase the profit margin. Following this, all tion with compatible systems, which may result in a beneficial outcome
resources are required to be used in a highly efficient way which can for the considered production systems. So for instance, Shell Global
be achievable through adequate integration. Other than the impact of (2017) recently proposed a hydrocracker-ethylene cracker integration to
crude oil price, other considerations that have substantial effects on reduce ethylene cracker feedstocks while using the hydrogen produced
the plants are the strict environmental regulations which have raised from the ethylene plant in the hydrocracking unit in the refinery. This
the cost of producing clean fuels (Fahim et al., 2010). was carried out at industrial scale in Singapore, Germany, Netherlands,
Most of the technological change in refineries over the last 20 and USA.
years has been because of environmental concerns that lead to the The study of modelling oil refineries is well established. However,
change and alteration of the existing processes. This “cause and effect” they are intrinsically non-linear and complex to simulate, therefore,
response to environmental mandates have had a large impact on most of the proposed models aim at reducing this challenging problem.
the refineries’ economic performance, causing increased costs (Leffler, In an oil refinery the conventional production planning is usually
2008). Nevertheless, these mandates have opened avenues to seek conducted focusing on the optimisation of individual or group of units
alternative pathways to step towards a more environmentally friendly within the production but not considering the utility system (Burkhard
et al., 2017). For example, a detailed formulation of a refinery model was
carried out by Kancijan et al. (2015) in which the modelling of the main

Corresponding author. units was focused on while also modelling 6 stages of the distillation
E-mail address: e.ketabchi@surrey.ac.uk (E. Ketabchi). unit and merging them into the general pattern of the refinery. This
https://doi.org/10.1016/j.cherd.2019.09.028
0263-8762/© 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
86 Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94

Nomenclature
Emc Fixed maintenance cost regarding either pro-
Sets and Subsets duction unit or utility equipment u when the
C Set of commodities operation is on
u Units Ec Activation energy of coking reaction
m Operation mode Mpr Molar concentration of propylene
f(u) Cracking furnaces-subset of u SEC Material switching cost coefficient regarding
RR(c) Subset of C of raw materials of the oil refinery cracking furnace f
RE(c) Subset of C- raw material of the ethylene plant FS Separation factor
CF(c) Subset of C of fuel oil or fuel gas ␪ Coke density
CE(c) Subset of C of ethylene plant product ␬ Ideal gas constant
CU(c) Subset of C of intermediates DP(c,t) Market demand for final product c
CFGE(c) Subset of C of fuel gas flowing out of the ethy- pri(c) Price of material c
lene plant SM(c,t) Supply of material c
CP(c) Subset of C of production product of the oil Dfc(c) Constant value regarding aggregated model of
refinery fuel consumption of cracking furnace f when
FGHER(c) Subset of C of fuel oil and gasoline and hydro- cracking material
gen flowing out of the ethylene plant ␥(u,c) Pre-linear coefficient of unit load on product
FGRE(c) Subset of C of fuel gas flowing out of the refinery yield
FORE(c) Subset of C of fuel oil flowing out of the refinery Cfp(u,c) Pre-linear coefficient of separation unit top
CHRE(c) Subset of C of chemical products flowing into pressure on product yield
the ethylene plant pi(u,t) Operation cost of the production unit
CRE(c) Subset of C of raw materials flowing into the ␣(u,m,c) Yield ratio of the material c of unit u in opera-
ethylene plant tion mode m
IPR(c) Subset of C of inventorial commodities (produc- ␣p(u,c) Fixed yield of the product c of cracking furnace
tion product) in the refinery f
IER(c) Subset of C of intermediate materials flowing SIL(c,t) Safety inventory level of commodity c
into the refinery IC(c,t) Inventory cost of commodity c
IRE(c) Subset of C of intermediate materials flowing INC(c,t) Inventory capacity of commodities c
out of the refinery PU(c,p) Property of intermediates-upper limit
HER(c) Subset of C of hydrogen flowing into the refin- PL(c,p) Property of intermediates-lower limit
ery PROI(c, p) Property p of intermediate product c
OMU(m) Subset of operation mode on unit
BL(u) Subset of U of blending headers Variables
UP(u) Subset of U of processing units Ct (u, t) Coke thickness of furnace
FB(u) Set of cracking furnaces and boilers FOT (u, c, t) Outlet temperature of furnace in period t
r(c) Raw material-subset of c in cracking furnace FB (u, c, t) Fuel c consumed by furnace or boiler of period
SO(c) Set of output products c of separation unit u t
US(u) Set of separation units in ethylene plant FC (u, c, t) Amount of commodities c consumed in
P Property period t in unit u
t Time horizon FC1 (u, m, c, t) Amount of commodities c consumed in
SI(c) Set of input material c of separation unit u period t in unit u on operation m
CI(c) Set of feed material of operation m on unit u FF (u, m, t) Flow rate of unit u of period t with operation
CO(c) Set of products of operation m on unit u mode m
FI(c) Set of input material c of furnace f FP (u, m, c, t) Amount of commodities c produced in
FO(c) Set of output products c of furnace f period t of unit u on operation m in the refinery
FP1 (u, c, t) Amount of commodities c produced in period
Parameters t of unit u in the ethylene plant
CDF Pre-exponential factor regarding impact of cok- FPP(c, t) Amount of commodities c produced of period t
ing deposition on product yields FU (u, t) Flow rate of unit u of period t
LCF Linearized coke factor MC(c, t) Raw material consumed in period t
FCCF Pre-linear coefficient fuel consumption of MER(c, t) Amount of material c from the ethylene plant
cracking furnace to the refinery
CRF Pre-linear factor for coking reaction IN(c, t) Material inventory of c of period t
FOTF Pre-linearized factor regarding impact of outlet MRE(c, t) Amount of material c from the refinery to the
temperature of furnace on product yields ethylene plant
FCCFT Pre-linear coefficient regarding fuel consump- PC(c, t) Amount of commodity c purchased of period t
tion of cracking furnace f related to coil outlet PEN(c, t) Penalty difference between real inventory and
temperature expected inventory of commodity c
Cfd Pre-linear coefficient regarding fuel consump- Prof Overall profit
tion of cracking furnace f related to dilution Rc (c, u, t) Coking reaction rate regarding furnace f pro-
steam cessing material in period t
Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94 87

feedstock management in terms of vessel arrivals and storage tanks


alongside the consideration of operation conditions for the plant side
RD (u, c, t) Dilution steam consumption of furnace f scheduling (Tjoa et al., 1997). Other relevant modelling aspects and
when consuming material details of an ethylene plant have been studied such as the optimal
SC(c,t) Amount of commodity c sold or delivered in reaction conditions for the steam cracking of ethane to obtain a higher
ethylene yield (van Goethem et al., 2013). Improved operating condi-
period t
tions were also investigated in a non-linear real-world and validated
B (u, c, t) Binary variable denotes whether the unit u
operational planning model aiming to maximise product revenues
operates with consuming material
while considering utility and feedstock cost (Gubitoso and Pinto, 2007).
r in period t Zhao et al. (2016) incorporated a short-term multi-period MINLP
Bm (u, m, t) Binary variable that denotes whether pro- planning model for the ethylene production that incorporated the use
cessing unit u is on with operation m of t of energy in both the thermal cracking and the downstream process to
Z (u, t) Binary variable denoting whether the material explore the potential for an increase in the production margin and the
consumed of cracking furnace is changed in reduction of energy losses.
period t Considering that a fair amount of research has been carried out
␣ (u,c,t) Yield ratio of the product c of cracking furnace on the individual plants in terms of modelling, it is of importance to
f consuming raw material investigate work carried out on the combination of these plants to also
identify the existing gap in this area.
Tp (u, t) Top pressure of separation unit
Al-Qahtani and Elkamel have published a paper regarding the
EDF(c, t) Fuel gas consumed in utility generation equip-
integration of the aforementioned plants with the consideration of a
ment
relevant aspect of refinery planning using different crude combination
alternatives. They have also considered production capacity expansion
while focussing on the simultaneous analysis of process network inte-
was then simulated using nonlinear constraints. Based on simulation gration alternatives in a multisite refining system and a petrochemical
studies, Kancijan et al.’s objective was to maximise the profit while tak- system. Through a mixed-integer linear program (MILP), their overall
ing the market requirements into account. Another example regarding objective was minimizing total annualized cost. In their formulation
studies on nonlinear modelling of the oil refinery is Guerra et al.’s work they have considered individual component flows rather than bilinear
which consists of two papers, one concerning the formulation of the mixing equations in order to avoid complexities. However, the finer
process models and the other presenting case studies. In Guerra et al.’s details of the petrochemical plant is not presented in the work and the
formulation section, the crude distillation unit (CDU) and the fluid cat- main focus is on the oil refinery (Al-Qahtani and Elkamel, 2008).
alytic cracking unit (FCC) in the oil refinery were modelled empirically In addition, Zhao et al. developed an integrated optimisation
and nonlinearly with the success of being validated through rigorous approach connecting an oil refinery with an ethylene plant while con-
process simulators (Guerra and Le Roux, 2011a, b). Further develop- sidering the potential of increasing the overall profit, formulating a
ments of the CDU model were achieved by Alattas et al. aiming to MINLP model to optimise the production planning of process units in
reflect the true nature of the model through adding nonlinearity to the oil refinery and the ethylene plant concurrently. They proposed
existing linear refinery planning models. This has resulted in higher though to simplify the complex model using the Lagrangian algorithm
profit margins when compared to common models, such as swing-cut in order to obtain a MILP problem (Zhao et al., 2017).
models, but with an increase in order of magnitude (Alattas et al., 2011). This work aims at implementing the proposed integration by using
Menezes et al. have also tackled this issue by improving the swing-cut relevant recent data as well as conducting model-based studies and
model for improved accuracy of the model. They have achieved this via optimization analyses in order to obtain an improved understanding
taking into consideration that the model requires corresponding light of those parameters that have a bigger impact on the integration of
and heavy swing-cuts avoiding overestimation of the profit (Menezes a conventional oil refinery with a petrochemical plant, in this case,
et al., 2013). an ethylene plant. This integration requires upgrading the refinery’s
Guerra and Le Roux however, were able to formulate a nonlinear by-products and main products by reducing operating costs while
refinery planning with empirical models for CDU and FCC by repre- upgrading the feed availability and quality of petrochemical plants.
senting accurate models to overcome limitations of both linear and This combination is beneficial for both plants in terms of lowering costs
nonlinear empirical models for CDUs and FCC units (Guerra and Le and improving the plant’s efficiency as well as supplying feed for the
Roux, 2011b). petrochemical plant. However, based on various research presented
Despite favoring accurate models for improved results, their com- and available, even though the accuracy of the models is of impor-
plexities must be considered in which various studies have aimed to tance, their complexities could cause problems that would hinder the
tackle this such as Shah’s research. They have proposed approaches for progression of implementing this idea. Therefore, our aim is to prove
linearization and model decomposition due to the high computational this concept while demonstrating a model that not only has the key
expense to reach an optimal solution for real-life refinery applications requirements of representing the plants to a good level of accuracy but
when they developed a short-term scheduling model for large-scale would also ensure the success of the model through a simpler model
refineries (Shah, 2015). This leads to the fact that although an accurate compared to previous work. It is of importance that through this con-
model is required for the representation of an oil refinery, the balance nection, intermediate material of both plants is utilised for improving
between accuracy and non-complexity is an important, yet challeng- and producing products as well as feeding the utility system demand.
ing aspect of this area of research that is also highly dependent on the This should be done in a way so that the suitable interaction between
application and end goal of the work. the plants followed by efficient material utilization are applied. More-
Petrochemical plants have also been studied in terms of produc- over, since both plants are prone to market volatility, yield maximising
tion planning and optimisation. One seminal example is the work of and profit increase alongside plant optimisation are key differentiators
Diaz and Bandoni in which they have presented the operational opti- for the competitiveness of refineries.
mization of a real ethylene production plant in operation using the
outer-approximation method for a mixed integer nonlinear program-
2. Problem statement
ming (MINLP) problem. They were able to reach convergence in few
iterations while also opening avenues to introduce this method to be
considered as a standard option in large scale conditions (Díaz and The key contribution in this work is the synergy between
Bandoni, 1996). complex plants in a way to achieve a beneficial outcome, in
Moreover, different approaches have been implemented regard- this case, an oil refinery and an ethylene production plant.
ing the ethylene production plant via MINLP modelling, including Through the proposed integration, economic benefit would be
88 Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94

Fig. 1 – Schematic representation of the integration between an oil refinery and an ethylene production plant (Ketabchi
et al., 2018).

DCU that produces additional VGO. RKERO is then consumed


Table 1 – List of abbreviations used in Fig. 1 with their
corresponding description. in the HDS that produces kerosene (KERO) as a final product
that is then stored in a product tank.
Abbreviation used in Figure 1 Description
Furthermore, the produced VGO alongside AGO then enters
DM Demethaniser the FCC which generates the main products of Cracked Gas Oil
DE Deethaniser (CGO), Cracked Gasoline (CG) and Fuel Oil (FO). Parallel to these
E-FR Ethylene Fractionator products, by-products of propylene, Fuel Gas (FG) and ethane
DPRO Depropaniser
are also produced from the FCC in which the ethane finds use
PRO-FR Propylene Fractionator
DB Debutaniser
in the cracking furnaces of the ethylene production plant. On
DPEN Depentaniser the other hand, the propylene produced will be stored in a tank
as a final product in the aforementioned plant For example, FO
and FG are used in the oil refinery to fuel the utility system
achieved for both plants in terms of providing an alternate though their initially high sulphur content, which requires
source of feedstock as well as intermediate material required these to be desulphurised in the DS prior to use.
for both involved production plants. For this purpose, each Furthermore, the CGO from the FCC enters the HT produc-
plant is to be modelled and optimised individually as well ing diesel and FO as well as FG which is a by-product of this
as the final integrated plant. Thus, three mathematical mod- process. The diesel from the HT, gathered with portions of LGO
els have been derived and implemented considering their and AGO are all blended in the diesel blender (DB). The DB then
integration method as a MINLP problem. A schematic of the produces two types of diesel. The former blend produces −10#
integration can be seen in Fig. 1. A Table is also provided for diesel, while when blending diesel with portions of CGO will
the abbreviations used in the aforementioned figure (Table 1). result in 0# diesel product.
The pathway of integrating the two plants will be explored Another unit previously mentioned, the CRU, produces FG,
while defining the units used in each plant as well as the cor- Naphtha and Reformer Gasoline (RG). This RG is then blended
responding streams. Firstly, the oil refinery process and its with Methyl tert-butyl ether (MTBE) as well as LSRN in the
units will be discussed. This plant consists of the main feed gasoline blender (GB) that gives the output of 90# gasoline.
of crude oil being converted into the final products. These Concurrently, CG, HSRN and MTBE are blended in the GB
units are as follows: CDU, FCC, Delayed Coking Unit (DCU), producing 93# gasoline. The gasoline and diesel blending pro-
Catalytic Reforming Unit (CRU), Hydro-Treating unit (HT), Gas cesses have specifications in terms of octane number and pour
Desulfurization unit (DS) and blenders. point. For example, the octane number of gasoline 90# and 93#
The primary process being the distillation process occurs in should be higher than 90 and 93, respectively. While diesel
the CDU, in which the products are lightest to heaviest: Light −10# and 0# should have a pour point (the temperature below
Straight-Run Naphtha (LSRN), Heavy Straight-Run Naphtha which the flow characteristics of the substance are lost) lower
(HSRN), Raw Kerosene (RKERO), Light Gas Oil (LGO), Atmo- than −10 ◦ C and 0 ◦ C, respectively. The most desired products
spheric Gas Oil (AGO), Vacuum Gas Oil (VGO) and the heaviest of the refinery consist of two kinds of gasoline (90# gasoline
product, residue (RESID). The produced RESID then enters the and 93# gasoline), two kinds of DIE (−10# diesel and 0# diesel),
Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94 89

KERO and FO. In more detail, −10# diesel means that −10 ◦ C operation mode m. Eq. (8) stipulates that only one operation
is the lowest temperature that diesel remains pourable and in of unit, u (processing units in the refinery) is allowed in each
this process the requirement is to have a pour point less than time period, t. Another equation that is critical in this model is
−10 ◦ C for −10# diesel. the sum of the flow rate under all operation modes equals the
Streams produced in the oil refinery can either be sold as flow rate of each processing unit, as shown in Eq. (9). Addition-
individual products or used beneficially in the ethylene plant. ally, the flow rate of the processing unit in the refinery is equal
One of the key objectives of this integration is to maintain a to the sum of commodities consumed, represented in Eq. (10)
balance of material usage in both plants. This is especially the and the fraction of flow rate is also equal to the commodities
case with ethylene and ethane that are wasted through heat- produced in the processing units of the refinery, Eq. (11).
ing and boiling within the refinery that could otherwise be
used in the ethylene production plant. Additionally, the propy- DP (c, t) = SC (c, t) ∀c ∈ CP, t (1)
lene produced by the FCC unit can be deployed in the ethylene
 
plant to produce a wide range of chemicals. IN (c, t) = IN (c, t − 1) + FP (u, m, c, t)
The considered ethylene production plant is presented as u ∈ UPm ∈ OMU(m)
follows: a series of parallel cracking furnaces having the feed  
of Naphtha, Ethane, AGO, and Hydrocracking Vent Gas Oil − FC1 (u, m, c, t) ∀c ∈ IPR(C), t (2)
(HVGO), which are used directly from the oil refinery. Follow- u ∈ UPm ∈ OMU(m)
ing this unit in which cracking takes place, quenching and
 
compressing occurs followed by separation units. These units IN (c, t) = IN (c, t − 1) + FP (u, m, c, t)
are namely, a demethaniser, deethaniser, depropaniser, debu-
u ∈ BL(U)m ∈ OMU(m)
tanizer, and depentaniser alongside a propylene and ethylene
fractionator to further separate ethane and propane from −SC(c, t)∀c ∈ CP, t (3)
ethylene and propylene. Other products from these units are
hydrogen, butadiene, benzene, C4 and C5, which have also 
partly use in the oil refinery. The produced hydrogen is more FC1 (u, m, c, t)
valuable to the refinery plant than to the ethylene plant. c ∈ CI(c)

Moreover, a portion of hydrogen can be transferred to the oil 


= FP (u, m, c, t) ∀u ∈ UBL, m ∈ OMU(m), t (4)
refinery, as well as to FO and CG from the furnace output.
c ∈ CO(c)
The proposed integration is beneficial in many ways, imple-
menting changes in requirements for each plant that should  
lead to mutual improvements while also increasing produc- PROI (c, p) *FC1(u, m, c, t) ≥ FP (u, m, c, t) *PL (c, p)
tion levels using available process technologies. Each plant c ∈ CI(c) c ∈ CO(c)
model is formulated separately including their integration to
∀u ∈ BL (U), m ∈ OMU(m) , p ∈ P, t (5)
minimise externally purchased feed material, with the overall
model based on the previous work of Zhao et al. (2017). Here,
the time dependency of some variables and parameters were
 
PROI (c, p) *FC1(u, m, c, t) ≤ FP (u, m, c, t) *PU (c, p)
removed in addition to assigning values to some parameters,
c ∈ CI(c) c ∈ CO(c)
instead of a range, to streamline the problem such as the sup-
ply demand relationship. Regarding time dependency, we have ∀u ∈ BL (U) , m ∈ OMU (m) , p ∈ P, t (6)
chosen the price of commodities sold and purchased not to be
time dependent in our study. This is beneficial in a way that
would lead to less complications when simulating the model, Bm ( u, m, t) *FU (u, t) = FF(u, m, t)∀u ∈ UP, m ∈ OMU(m), t (7)
less infeasibilities and compilation time. 
Bm (u, m, t) ≤ 1∀u ∈ UP, t (8)
2.1. Oil refinery model m ∈ OMU(m)

A model is proposed to represent the process of refining crude FF (u, m, t) = FU (u, t) ∀u ∈ UP, t (9)
oil to the manufacturing of the products in accordance with m ∈ OMU(m)

the process description. Eq. (1) defines the relation between 


commodities sold and the market demand and have been FF (u, m, t) = FC1 (u, m, c, t) ∀u ∈ UP, m ∈ OMU (M), t
chosen to be equal in this case. Material inventory balance c ∈ CI(c)

for the final products is presented in Eq. (2), being equal to (10)
the sum of commodities produced in the processing units of
the refinery considering the commodities consumed. Eq. (3) FP (u, m, c, t) = ˛ (u, m, c) *FF (u, m, t) ∀u ∈ UP, m ∈
shows the inventory balance for final products in the blender,
which is the sum of the amount of commodities produced in OMU (M), c ∈ CO(c), t (11)
the blending headers subtracting the number of production
materials sold. Eq. (4) evidences the commodities consumed 2.2. Ethylene production plant model
for blending is equal to commodities produced in the blending
headers, while Eq.s (5) and (6) represent inequality constraints The ethylene plant has been considered to consist of crack-
for blending processes. Eq. (7) denotes that the flowrate of unit, ing processes taking place in parallel furnaces and multiple
u on operation mode, m is equal to the unit capacity (flow rate separation units to obtain the desired final products. Here,
in unit u-processing units in the oil refinery) with Bm being Eq. (12) represents the inventory balance for raw material in
the binary variable indicating whether the unit is active with the ethylene plant specifically that it is equal to the sum of
90 Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94

commodities produced in the plant considering the recycle


flow rate of ethane produced in the separation train, plus 
the supply of raw material while subtracting the commodi- IN (c, t) = IN (c, t − 1) + FP1 (u, c, t) − SC(c, t)
ties consumed in the furnace. Inventory balance and demand u ∈ SO(u,c)
constraints for the final products are shown in Eq. (13). The 
inventory is equal to the sum of the amount of commodi- − FB (u, c, t) ∀c ∈ CE, t (13)
ties produced in the ethylene plant (recycle flowrate of ethane u ∈ FB(U)

produced in the separation train) minus the amount of sold


commodities and minus the fuel consumption of the boiler. PEN (c, t) ≥ IN (c, t) − SIL(c, t)∀c, t (14)
Moreover, the economic penalty, (PEN(c, t)), is introduced in
Eqs. (14) and (15) as a necessary factor to be considered in the −PEN (c, t) ≤ IN (c, t) − SIL(c, t)∀c, t (15)
plant inventory, when the inventory of the material or prod-
uct exceeds the security level, MIS. Exceeding the security level IN (c, t) ≤ INC(c, t)∀c, t (16)
will lead to a loss in the overall profit as the economic penalty

is included in the objective function. It should also be noted PC (c, t) + FPP (c, t) ≥ FB (u, c, t) ∀c ∈ CF, t (17)
that the inventory capacity must exceed the material inven-
u ∈ FB(u)
tory as shown in Eq. (16). Another aspect that needs to be
considered is the material balance. Material balance for Fuel 
FPP (c, t) = FP1(u, methane, t)
gas/fuel oil is presented in Eq. (17), indicating the fact that the
u
sum of Fuel gas/fuel oil purchased and produced is greater or

equal to the sum of Fuel gas/fuel oil consumed by the cracking + FP1(u, propane, t)∀c ∈ CF, u ∈ US, t (18)
furnaces and boiler. Furthermore, the fuel oil produced in the u
furnace is assumed to be equal to the amount of methane
and propane produced, which is demonstrated in Eq. (18).
Rc (r (c) , f (u) , t) = CRF ∗ *FOT (f (u) , r, t) *Ec*Mpr∀f (u),
During the cracking process in the furnace, coke is formed.
The consideration of this phenomenon is included in this r(c), t (19)
model in Eq. (19). This equation calculates the reaction rate
of coke formation through the use of Arrhenius’ Law as well
as coke thickness demonstrated in Es. (19) and (20) developed 
t 
Ct(f (u) , t) = LCF*Rc (r(c), f (u) , t) **B(u, r(c), t)
by Berreni and Wang (2011). Eq. (21) shows that the product
1 r
yield model is equal to the sum of fixed yield, fractions of fur-
nace outlet temperature and coke thickness. Another material ∀f (u), r(c), t (20)
balance is presented in Eqs. (22) and (23) concerning the rela-
tionship between the commodities consumed/produced in
the furnace and the flow rate in the furnace. In more detail, ˛ (f (u), c, t) = ˛p (f (u), c) + CDF (r(c), f (u), c) *Ct (f (u) , t)
they portrait the commodities produced in the ethylene plant, +FOTF*FOT(f (u), r(c), t)∀f (u), r(c), c ∈ FO(c) (21)
which is equal to the commodities consumed multiplied by
their fixed yield in Eq. (22) while Eq. (23) depicts that the sum
of commodities consumed (raw material in this case) should
FP1(f (u), c, t) = ˛p (r(c), f (u) , c) *FC(f (u), r(c), t)∀f (u), r(c),
be equal to the flow rate of the corresponding unit. Eq. (24)
regards a constraint for yield fractions that should have a sum c ∈ FO(c), t (22)
of less than one, while Eq. (25) shows that the commodities
consumed specifically in the cracking furnace should be equal 
to the flowrate. Eq. (26) defines that for each furnace, only one FC (f (u), r(c), t) = FU(f (u), t)∀f (u), r(c) ∈ FI(c), t (23)
material can process in any given period with B (u, c, t) being r

the binary variable that denotes whether the unit u operates 


with consuming material r in period t for the ethylene produc- ˛p (r, f (u), c) ≤ 1∀f (u), r(c), c ∈ FO(c) (24)
tion plant. Eq. (27) details the changeover condition variable, z c
(u, t), for cracking furnaces and is introduced in the objective
function as a penalty in case of unstable operation. Eq. (28) B (u, r(c), t) *FU (u, r(c)) = FC(u, r(c), t)∀u ∈ f (u), r(c), t (25)
shows the production of each product, FP1 (u,c,t), from the 
corresponding separation unit as a linear function of flowrate. B (f, r(c), t) ≤ 1∀f (u), t (26)
Operating condition, Tp (u,t), corresponds to the top pressure r(c)
of the separation column and unit separation factor, Fs. Lastly,
Eq. (29) shows that fuel consumed by furnace or boiler is equal Z (u, t + 1) ≥ B (u, r(c), t + 1) − B (u, r(c), t) ∀u ∈ f (u), r(c), t (27)
to the sum of the linear function of flowrate, outlet tempera-
ture of furnace and dilution steam. FP1 (u, c, t) =  (u, c) *FU (u, t) + Cfp (u, c) *Tp2 (u, t)

+Fs ∀u ∈ US, c ∈ SO (u, t), t (28)


IN (c, t) = IN (c, t − 1) + FC (f (u), c, t) + SM (c, t) FB (f, c, t) = FCCF*FC (f (u) , r (c) , t) + FCCFT*FOT (f (u) , r (c) , t)
f (u) ∈ FI(u,c)
 +Cfd*RD (f (u), r(c), t) + Dfc(c)∀ f (u), c ∈ CF, t (29)
− FC (f (u), c, t) ∀c ∈ RE(C), f (u), t (12)
f (u) ∈ FI(u,c)
Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94 91

2.3. Interconnection section model — inventory


 
balance for ethylene production plant − FC1 (u, m, c, t) + MER(c, t)∀c ∈ IER(C), t (35)
u ∈ UPm ∈ MU
This section covers the material balance of transferred mate-
rial from the refinery to the ethylene plant and vice versa.  
Eq. (30) shows the inventory balance of raw material to IN (c, t) = IN (c, t − 1) + FP (u, m, c, t)
u ∈ UPm ∈ OMU(m)
the ethylene plant from the refinery, while Eq. (31) demon-
strates the inventory balance of chemicals flowing into the +MER (c, t) − SC(c, t)∀c ∈ FORE, t (36)
ethylene plant from the oil refinery as depicted in Fig. 1. The
supply of fuel gas is presented in Eq. (32). This supply consists
 
of the amount purchased from the market, transported from IN (c, t) = IN (c, t − 1) + FP (u, m, c, t)
the refinery and produced by the ethylene complex. Eq. (33)
u ∈ BL(U)m ∈ OMU(m)
denotes the relationship of flow streams for fuel oil, cracked
gasoline and hydrogen. −MRE (c, t) − EDF (c, t) ∀c ∈ FGRE, t (37)


IN (c, t) = IN (c, t − 1) + MRE (c, t) + FP1 (u, c, t) + SM (c, t)
IN (c, t) = IN (c, t − 1) + MER (c, t)
u ∈ SO(c)
 
 − FC1 (u, m, c, t) + PC (c, t) ∀c ∈ HER(C), t (38)
− FC (f, c, t) ∀c ∈ RE, f, t (30)
u ∈ UPm ∈ OMU(m)
f ∈ FI(c)

 MRE (lsrn, t) = MRE (naphtha, t) ∀t (39)


IN (c, t) = IN (c, t − 1) + MRE (c, t) + FP1 (u, c, t)
u ∈ SO(c)  
FP (u, m, ethane, t) = MRE (ethane, t) ∀t (40)
−SC (c, t) ∀c ∈ CHRE, t (31) u ∈ UPm ∈ OMU(m)

 
 FP (u, m, propylene, t) = MRE (propylene, t) ∀t (41)
PC (c, t) + MRE (c, t) + FPP(c, t) ≥ FB(u, c, t)∀c ∈ CFGE, t u ∈ UPm ∈ OMU(m)
u ∈ FB(u)
(32)
2.5. Objective function definition

 The integrated model’s objective function aims to maximize


MER (c, t) = FP1(u, c, t) ∀c ∈ FGHER(C), t (33) the profit.
u ∈ SO(c) This function consists of the revenue of refinery and
ethylene plant sales, the raw material cost for both plants,
2.4. Interconnection section model — inventory inventory costs, equipment and maintenance costs. This
balance for oil refinery plant objective function is detailed in Eq. (42) where pri (c) repre-
sents the price of material and SC (c, t) the amount of material
This section covers the material balance of transferred mate- sold. The second and third terms are the revenue of the ethy-
rial from the refinery to the ethylene plant and vice versa. lene plant products and intermediates, respectively. The next
The first equation in this section represents the inven- three terms correspond to the costs for raw material, inventory
tory balance of products, including the portion that will be and processing into account with PC (c, t) being the amount
transferred to the ethylene production plant. Eq. (35) is the of material purchased, IN (c, t) the material inventory, INC (c,
inventory balance of cracked gasoline flowing back to the refin- t) inventory cost, FR (u, t) the flow rate and OPC (u) the price
ery as intermediate products from the ethylene plant whereas of unit operation. The remaining terms show the cost for the
Eq. (36) shows the inventory balance of fuel oil. The amount of supply for the ethylene plant, penalty and material switching
fuel oil in the inventory balance is comprised of the amount costs as well as the cost of operating the separation columns.
produced in the refinery plus the amount produced in the
ethylene plant, subtracting the amount sold. Similar to Eq. (36),    
Eq. (37) represents the inventory balance for fuel gas. Eq. (38) Prof = pri (c) *SC (c, t) + pri (c) *SC (c, t) +
is the inventory balance for hydrogen that is mostly produced t c ∈ CP(C) t c ∈ CE(C)
in the ethylene production plant and utilised in the Hydro-    
Treating Units of the refinery. Eqs. (39–41) represent the mixing pri (c) *SC (c, t) − pri (c) *PC (c, t) −
relationship for flow streams. t c ∈ CU(C) c ∈ RR(C) t
   
  IN (c, t) *IC (c) − FU (u, t) *pi (u, t) −
IN (c, t) = IN (c, t − 1) + FP (u, m, c, t) t c ∈ CU(C) t u ∈ UP(U)
u ∈ UPm ∈ OMU(m)    
  pri (c) *SM (c, t) − PEN (c) *IC (c) −
− FC1 (u, m, c, t) − MER (c, t) ∀c ∈ IRE(C), t (34)
c ∈ RE(C) t t c ∈ RE(C)
u ∈ UPm ∈ MU   
FU (u, t) *pi (u, t) − Z (u, t) *SEC(f (u)) (42)
  t u ∈ US(U) t f
IN (c, t) = IN (c, t − 1) + FP (u, m, c, t)
u ∈ UPm ∈ OMU(m)
92 Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94

2.6. Model implementation


Table 2 – Amount of material produced in the oil refinery
sent to the ethylene production plant- comparison
This model has been implemented in GAMS version 24.8.5 for before and after integration.
optimisation. Each model has been implemented separately
Material Amount-before Amount-after
for comparison and to also optimise the plants individually. All integration (t) integration (t)
three models were mathematically formulated while the inte-
Ethane 15 10
grated version was formulated as an MINLP problem, solved
Naphtha 20 30
with the BARON solver (Tawarmalani and Sahinidis, 2004) with
AGO 345 345
a compilation time of 33.12 s for the oil refinery model, 3.4 min HVGO 385 380
for the ethylene plant and 6.7 min for the integration. Propane 5 2
The compilation time is significantly lower when compared FG 20 29
to other studies, mainly due to the simplification of the model
leading to the attainment of results in short order. Below is
Table 3 – The capacity of units chosen by the solver to
presented a graph demonstrating various papers and their
optimise the plant before and after integration.
compilation time while modelling an oil refinery and an ethy-
lene production plant. Unit Capacity-before Capacity-after
integration (t) integration (t)
As can be seen in Fig. 2 a series of papers have been
compared in terms of compilation times which includes both CDU 200000 200000
pre-processing and optimization time for the modelling and FCC 80000 85,824
optimization of an oil refinery and an ethylene production HDS 70000 85,824
CRU 82000 82,460
plant separately, while the time for the model in this study has
GB 200000 237,551
also been marked amongst the previous studies. It is evident
DB 200000 237,551
that the compilation time for this paper is considered to be in
a good range, being among the studies that have been able to
conduct the simulation at impressively low times. However, it propylene, butadiene, benzene, C4 and C5 . It must be noted
must be noted that due to the complexities of each model the that due to one of the aims being the avoidance of model com-
time can differ which is not necessarily due to any shortcom- plexity while also maintaining accuracy as much as possible,
ings of the prior research. Nonetheless, it is an advantage of there were various assumptions considered in this work. Apart
this research to be able to conduct the simulation in a rela- from the few mentioned in the previous paragraph, no utility
tively short time when compared to other studies making this system such as cooling unit models, energy consumption and
simulation easier to conduct. generation of furnaces and compressors or boiler and turbine
models were considered. Various parameters were given fixed
values or were calculated to produce a fixed number rather
3. Case study than having a range and lastly, only mass integration is being
investigated in this research.
The proposed model has been constructed in a way to cover To describe the decision variables, the feed required for the
the full period of both plants however, in this case study, a spe- ethylene plant is sourced from the refinery products having
cial case of integration has been considered while regarding the values obtained from the optimisation of the oil refinery
the model as a multi-period network. presented below while comparing their values prior and after
This study will explore the integration of a conventional the integration. The products obtained in the ethylene produc-
UK oil refinery and an ethylene production plant on a single tion plant are also demonstrated through comparison before
period aiming to maximise profits of both plants as well as and after the integration. To investigate the impact of integra-
the integration. The proposed mathematical model requires a tion on the oil refinery, the changes in the unit capacity of the
number of parameters such as price, capacity, properties, sup- modelled oil refinery units are also presented in Table 3.
ply, demand, product specifications and process yield, which As seen in Table 2, there is a slight increase in the pro-
are not calculated but defined using various databases for UK duction of Naphtha and FG after the integration taking place.
refineries and petrochemical plants (BOC, 2017; Deloitte, 2012; This is due to the introduction of streams from the ethylene
Oil and Gas Journal, 2014). This model contains a total of 42 production plant to the oil refinery. For instance, introduc-
equations with 23 variables while considering the demand of ing Hydrogen from the ethylene production plant entering the
material to be equal to the amount of material sold as well CRU would result in the increase of Naphtha and FG produc-
as using only one type of crude oil for the oil refinery sec- tion that can be seen in the results of this integration. This
tion, assuming the maximum capacity of 200,000 bbl/day in increase is beneficial for the system as it can be used in the
the CDU with the decision variables of FP1 (u, c, t), FPP (c, t) and cracking furnaces producing a cycle that produces more inter-
FU (u, t) being the products obtained from the Ethylene plant, mediates for the oil refinery while providing more feed for the
oil refinery and the capacity of the units involved in this opti- ethylene plant all over again. However, some of the materials
mization problem, respectively. Furthermore, the physical and produced in the oil refinery have decreased such as Ethane,
chemical properties of the materials and their intermediates HVGO and propane. They are used in the ethylene production
from the blending processes were considered. plant, as feed for the cracking furnaces and product tanks in
The main units implemented in GAMS for the oil refinery the aforementioned plant. Due to the change in stream levels
were the CDU, FCC, CRU, DS and blenders having the prod- after the integration, the capacity and flows in the oil refinery
ucts and intermediates of gasoline, diesel, kerosene, fuel oil, units, which have been modelled, have also changed respec-
fuel gas, ethane and propylene. Meanwhile, the main units tively as noted in Table 3.
modelled for the ethylene production plant are the series of The results demonstrated in Table 4 are indicative of the
furnaces and separation units. These units produce ethylene, changes occurring after integration regarding the ethylene
Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94 93

Fig. 2 – Comparison of compilation time regarding the modelling of an oil refinery and an ethylene production plant with
previous studies (Shah et al., 2015) (Guyonnet et al., 2009) (Zhao et al., 2014) (Menezes et al., 2013) (Alattas et al., 2011) (Zhao
et al., 2016) (Díaz and Bandoni, 1996) (Gubitoso and Pinto, 2007) (Zhao et al., 2011) (Tjoa et al., 1997).

Table 4 – Amount of intermediate material produced in Table 6 – Operating costs used in the model for units in
the ethylene production plant- comparison before and the ethylene plant, oil refinery and their integration.
after integration.
Process unit Operating cost ($/bbl feed)
Material Amount-before Amount-after
CDU 0.15
integration (t) integration (t)
CRU 0.6
CG 300 370 FCC 0.65
FO 200 150 HDS 0.65
Hydrogen 50 50 GB 0.2
DB 0.2
Separation unit 0.3
Boiler 0.1
Table 5 – Profit of the oil refinery, ethylene production
plant, and the proposed integration.
Plant Profit (million UK pounds)
Table 7 – Inventory costs used in model for ethylene
plant, oil refinery and their integration.
Oil refinery 1.28
Material Inventory cost ($/day)
Ethylene production 0.475
Integrated oil refinery 13.46 Gasoline 1427.868
and ethylene Diesel 255.95
production plant FO 869.0475
Kerosene 91.66
Propylene 260.035
Crude oil 714257.8
FG 869.0475
production plant intermediates. There is an increase in the Ethylene 190.255
production of CG, which is due to the increase of FG, Naph- C4 238.095
tha, AGO, HVGO and Ethane feed to the cracking furnaces. Butadiene 119.0475
This leads to the beneficial use of CG in the GB units in the C5 5714.28
oil refinery. The levels of FO produced from this plant have Benzene 238.095
Hydrogen 62500
decreased, which can be explained through the fact that it
enters the oil refinery and is blended in the product tank
that could be utilised for further processes. This highlights 4. Conclusion
the close and successful connection between the two plants
that has proven that the proposed model has in fact comple- In this work, a mixed-integer nonlinear programming model
mented both plants as explained regarding Tables 2–4. The has been proposed for the integration of an oil refinery and
maximization of the profit of each plant including the inte- an ethylene production plant. Using data from a typical oil
gration, the outcome is presented in Table 5. As seen, there refinery and ethylene plant in the UK, the models were imple-
is a significant increase in the profit after integration, which mented in GAMS and using different relaxation techniques to
demonstrates the relevance of the proposed plant integration. tackle complexities and errors. Through our proposed model,
This high increase is due to the interaction in both plants and we were able to achieve less complexity while considering
the use of intermediate streams for the benefit of each plant, accuracy to have the possibility of achieving the intended
and thus, aiming to replace the externally purchased feed. outcome of higher profit, operational efficiency and less
Nevertheless, it must be noted that not taking utility costs into dependency on fossil fuel which is maintained through the
consideration would be part of the substantial profit increase. introduced concept of industrial symbiosis where both plants
A breakdown of the process costs for each plant, material benefit through the connection having a reduced need to pur-
inventory cost, inventory capacity and material price utilised chase feed for one of the plants. This simplified approach
in the models are presented in Tables 6–8. enables a smoother pathway for application, removing many
94 Chemical Engineering Research and Design 1 5 2 ( 2 0 1 9 ) 85–94

planning optimization. Ind. Eng. Chem. Res. 50 (11),


Table 8 – Material costs used in the model for the
6860–6870, http://dx.doi.org/10.1021/ie200151e.
ethylene plant, oil refinery and their integration.
Menezes, B.C., Kelly, J.D., Grossmann, I.E., 2013. Improved
Material Price ($/bbl) swing-cut modeling for planning and scheduling of
oil-refinery distillation units. Ind. Eng. Chem. Res. 52 (51),
Gasoline 55.18
18324–18333, http://dx.doi.org/10.1021/ie4025775.
Diesel 51.85
Shah, N.K., 2015. Decomposition Approaches for Enterprise-Wide
FO 12.88
Optimization in Process Industry.
Kerosene 83.33
Díaz, M.S., Bandoni, J.A., 1996. A mixed integer optimization
Propylene 5602.9
strategy for a large scale chemical plant in operation. Comput.
Ethylene 150.8
Chem. Eng. 20 (5), 531–545,
C4 22.2
http://dx.doi.org/10.1016/0098-1354(95)00209-X.
Butadiene 244.4
Tjoa, I.B., Ota, Y., Matsuo, H., Natori, Y., 1997. Ethylene plant
C5 12333.3
scheduling system based on a MINLP formulation. Comput.
Benzene 99.28
Chem. Eng. 21 (Supplement 1), S1073–S1077,
Ethane 0.465
http://dx.doi.org/10.1016/S0098-1354(97)87645-4.
Naphtha 4058.8
van Goethem, M.W.M., Barendregt, S., Grievink, J., Verheijen, P.J.T.,
AGO 10
Dente, M., Ranzi, E., 2013. A kinetic modelling study of ethane
Hydrogen 37.037
cracking for optimal ethylene yield. Chem. Eng. Res. Des. 91
Crude oil 47
(6), 1106–1110, http://dx.doi.org/10.1016/j.cherd.2013.01.006.
FG 13
Gubitoso, F., Pinto, J.M., 2007. A planning model for the optimal
production of a real-world ethylene plant. Chem. Eng. Process.
obstacles that were present in previous works ensuring oper- Process Intensif. 46 (11), 1141–1150,
http://dx.doi.org/10.1016/j.cep.2007.02.022.
ational efficiency and a step towards an environmentally
Zhao, H., Ierapetritou, M.G., Rong, G., 2016. Production planning
friendly industry. optimization of an ethylene plant considering process
Each plant was optimised individually followed by their operation and energy utilization. Comput. Chem. Eng. 87,
integration which has presented valuable results in terms of 1–12, http://dx.doi.org/10.1016/j.compchemeng.2016.01.002.
producing higher value products while maximising the profit Al-Qahtani, K., Elkamel, A., 2008. Multisite facility network
for the plants individually. Moreover, after the integration of integration design and coordination: an application to the
refining industry. Comput. Chem. Eng. 32 (10), 2189–2202,
the plants, there is a significant profit increase showing the
http://dx.doi.org/10.1016/j.compchemeng.2007.10.017.
integration to be of considerable advantage while benefiting
Zhao, H., Ierapetritou, M.G., Shah, N.K., Rong, G., 2017. Integrated
from product availability, synergy from joint infrastructure model of refining and petrochemical plant for enterprise-wide
while dampening the influence of fluctuations of feed and optimization. Comput. Chem. Eng. 97, 194–207,
product costs. The high increase in profit is mainly due to http://dx.doi.org/10.1016/j.compchemeng.2016.11.020.
the use of intermediates from the oil refinery as opposed to Berreni, M., Wang, M., 2011. Modelling and dynamic optimization
purchasing them from an outside source. Not only does the of thermal cracking of propane for ethylene manufacturing.
Comput. Chem. Eng. 35 (12), 2876–2885,
plant benefit from reducing the external feed purchases, it also
http://dx.doi.org/10.1016/j.compchemeng.2011.05.010.
benefits from upgrading the quality of the final products. Tawarmalani, M., Sahinidis, N., 2004. Global optimization of
mixed-integer nonlinear programs: a theoretical and
References computational study. MathProg 99, 563–591.
BOC, 2017. Industrial Gases UK. Propylene Surcharge Monthly
Update,. Published. (Accessed 10 September 2017)
Fahim, M.F., Alsahhaf, T.A., Elkilan, A.E., 2010. Fundamentals of
www.boconline.co.uk.
Petroleum Refining, 1st, ed. Elsevier, Oxford.
Deloitte, London 2012. Study of the UK Petroleum Retail Market.
Leffler, W.L., 2008. Petroleum Refining in Nontechnical Language,
http://www.ukpia.com/Libraries/Download/UK Petroleum Retail
4th ed. PenWell, Tulsa, Oklahoma.
Market Study Final Report v3 STC.sflb.ashx.
Shell Global, 2017. Hydrocracker-Ethylene Cracker Integration.
Oil and Gas Journal, 2014. Worldwide report. Oil Gas J., 45,
Shell Global Solutions, . Published. (Accessed 15 August 2017)
December.
http://www.shell.com/business-customers/global-solutions/
Ketabchi, E., Mechleri, E., Gu, S., Arellano-Garcia, H., 2018.
industry-focus/hydrocracker-ethylene-cracker-integration.
Modelling and optimisation approach of an integrated oil
html.
refinery and a petrochemical plant. In: Eden, M.R.,
Burkhard, J., Lewandowski, S., Eramo, M., 2017. Understanding
Ierapetritou, M., Towler, G.P. (Eds.), 13th International
How Fluctuating Crude Oil Prices Impact Petrochemical
Symposium on Process Systems Engineering (PSE2018).
Investment Strategies. IHS Markit,. Published. (Accessed 10
Elsevier B.V., California, pp. 1081–1086.
October 2018)
Shah, N.K., Sahay, N., Ierapetritou, M.G., 2015. Efficient
https://ihsmarkit.com/research-analysis/understanding-how-
decomposition approach for large-scale refinery scheduling.
fluctuating-crude-oil-prices-impact-petrochemical-
Ind. Eng. Chem. Res. 54 (41), 9964–9991,
investment-strategies.html.
http://dx.doi.org/10.1021/ie504835b.
Kancijan, M., Ivanjko, M., Ilak, P., Krajcar, S., 2015. An oil refinery
Guyonnet, P., Grant, F.H., Bagajewicz, M.J., 2009. Integrated Model
production optimization. 2015 5th International Youth
for Refinery Planning, Oil Procuring, and Product Distribution
Conference on Energy, 1–11,
Integrated Model for Refinery Planning, Oil Procuring, and
http://dx.doi.org/10.1109/IYCE.2015.7180772.
Product Distribution., pp. 463–482.
Guerra, O.J., Le Roux, G.A.C., 2011a. Improvements in petroleum
Zhao, H., Rong, G., Feng, Y., 2014. Multiperiod planning model for
refinery planning: 1. Formulation of process models. Ind. Eng.
integrated optimization of a refinery production and utility
Chem. Res. 50 (23), 13403–13418,
system. Ind. Eng. Chem. Res. 53 (41), 16107–16122,
http://dx.doi.org/10.1021/ie200303m.
http://dx.doi.org/10.1021/ie502717e.
Guerra, O.J., Le Roux, G.A.C., 2011b. Improvements in petroleum
Zhao, C., Liu, C., Xu, Q., 2011. Dynamic scheduling for ethylene
refinery planning: 2. Case studies. Ind. Eng. Chem. Res. 50 (23),
cracking furnace system. Ind. Eng. Chem. Res. 50,
13419–13426, http://dx.doi.org/10.1021/ie200304v.
12026–12040.
Alattas, A.M., Grossmann, I.E., Palou-Rivera, I., 2011. Integration
of nonlinear crude distillation unit models in refinery

You might also like