Petroleum Refining Operations: Key Issues, Advances, and Opportunities
Petroleum Refining Operations: Key Issues, Advances, and Opportunities
Petroleum Refining Operations: Key Issues, Advances, and Opportunities
Enterprise-wide optimization for the petroleum refining industry involves optimization of the supply chain
involving manufacturing and distribution with emphasis on integration of the different decision making levels.
The key manufacturing operations include crude oil loading and unloading, mixing of crude oil, production
unit operations of conversion and separation, operations of blending, and distribution of products. Other
components of the petroleum supply chain network include oil explorations, crude oil procurement, and sales
and distribution of products. The main issues present in the petroleum industry across various decision levels
(strategic, tactical, and operational) and within oil refinery operations are discussed. This paper presents an
extensive literature review of methodologies for addressing scheduling, planning, and supply chain management
of oil refinery operations. An attempt is also made to identify the future challenges in efficiently solving
these problems.
logistics and finished product distribution management remain with wellhead operation should not be underestimated,13 it is
important parts of the refinery supply chain. not the focus of this review.
2.1. Petroleum Industry Supply-Chain Components. As 2.1.1. Crude Oil Procurement. A typical crude oil procure-
shown in Figure 1, a typical petroleum supply chain involves ment process in a refinery involves three phases: crude selection
the exploration phase, crude procurement, storage logistics, and purchase, crude transportation and storage, and crude
transportation to the oil refineries, refinery operations, distribu- processing, and these phases are explained in detail by Julka
tion, and transportation of the final products. The problem of et al.5 The crude selection is an extremely important step in
investment and operations in offshore oil field development is refinery supply chain since it has direct impact on finished
characterized by a long planning horizon (typically 10 years) product quality, quantity, and it accounts for about 90% of the
where a large number of choices in platforms, wells, field refinery input cost. Crude oil selection and purchase is done
locations, and pipeline infrastructure is present.12 The explora- periodically at fixed or variable intervals and is purchased for
tion phase involves significant investment costs; thus economical production over procurement cycle (of few days to a month) in
drive exists to maximize the return on investment. Rights for advance. After the crudes are selected and purchased, the focus
extracted crude oil are then purchased for the production of shifts to optimizing delivery costs from the suppliers to the
finished products to satisfy market demand, while respecting refinery jetties. The crude oil is carried to the port by tankers,
product quality specifications. The nature of the petroleum where it is unloaded into tanks when the jetty becomes available
supply chain is such that its economics are extremely complex for docking. This crude oil unloading problem is limited by
and heavily linked. For example, the quality and quantity of tank capacity, pumping rate, setup costs, and demurrage. Care
the finished products are impacted by different crude-oil mixes should be taken when transporting high-fusion-point crude oil
and by oil refinery configuration, capacity, and yield constraints. to avoid large residency time in pipelines.14 The crude oil
As mentioned before, oil refinery production operations are the loading and unloading problem is further described next.
most complex components of an oil refinery supply chain, and, 2.1.2. Crude Oil Operations. Crude oil scheduling problem
to effectively tackle scheduling and planning of single site or is one of the most challenging problems in an oil refinery plant.
multisite optimization, it is necessary to decompose the refinery An optimal crude oil scheduling is predetermined based on crude
production operations into smaller manageable subproblems. arrival data, production targets, and operational constraints. Once
Generally, oil refinery operations are decomposed into three crude oil is loaded into the storage tanks, crude oil mixtures
subproblems based on the structure of the refinery configuration are prepared to charge crude distillation units (CDU) for
as shown in Figure 2. These problems are (1) crude-oil production. Generally, there are two different types of mixture
unloading, mixing, and inventory control, (2) scheduling of preparation: (a) in charging tanks and (b) using pipelines in a
production units, and (3) finish products blending and distribu- place called manifold just before the CDU. The complexity in
tion.4 The crude oil procurement problem and oil refinery the scheduling problem arises from a bilinear term (flow and
operations subproblems are described in detail next. Although inventory) present in mass balance constraints where the
the upstream operations including the safety issues associated intensive properties of the crude oil mixtures charging the CDU
Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1163
must be regulated to meet the downstream product quality after a brief discussion of different decision levels in an oil-
specifications. Saharidis and Ierapetritou15 solved the crude oil refinery supply chain.
scheduling problem using discrete time presentation where they 2.2. Integration of Different Decision-Making Levels in
address both types of crude mixture preparation configurations. the Supply Chain. Management of the petroleum supply chain
The economic and operability benefits of the crude oil blend is a complex task due to the large size of the physical supply
scheduling problem in a refinery is discussed in detail by Kelly network dispersed over vast geography, complex refinery
and Mann.16,17 production operations, and inherent uncertainty involved. There
2.1.3. Production Unit Operations. Once the crude prepared are three main decision levels in a traditional supply chain: (1)
mixture is prepared, it is charged to CDUs for distillation. The strategic, (2) tactical, and (3) operational. Strategic level
distillation cuts from CDU are then sent to other production decisions are made for a few years, tactical level decisions range
units for fractionation and reaction to produce blend components from several months to a few days, and control level decisions
for finished products. The most common refinery process range from a few minutes to seconds. Different decision levels
includes catalytic, hydro, and thermal cracking units to convert in a traditional supply chain and their complexity are described
heavy hydrocarbons into light hydrocarbons. They also include in detail by Simchi-Levi, Kaminsky et al.22 The main issues in
other process units like continuous catalytic reforming, hy- decision making at different levels are related to temporal
drotreating, and hydrodesulfurization units. This problem is integration, where they involve coordination across different
characterized by continuous processes, intermediate component time scales.
storage, and a recycle stream. The production units scheduling The goal of supply-chain design is to determine optimal
problem is addressed by Jia and Ierapetritou4 and Shah et al.18 infrastructure (location, number, and capacity of wellheads and
where they applied spatial decomposition to solve the scheduling refineries, and flow of material through logistics network). The
problem. Gary, Handwerk et al.19 give a comprehensive strategic level decisions are made for several years in a highly
reference for refinery production unit operations. uncertain environment. In many instances, the supply-chain
2.1.4. Blending and Distribution Operations. Product network design can have extensive impact on the profitability
blending is a crucial step in refinery operations, for example, and risk uncertainty of the global supply chain. In the oil refinery
gasoline can yield 60-70% of a refinery’s profit. The finished industry, the supply-chain network is composed of shipping via
product blending process is a mixing of various intermediate vessel, oil tankers, and pipelines that may run across multiple
products, along with some additives, to produce blends with countries. This network is used to transport crude from wellhead
certain qualities in order to minimize cost and maximize to refinery for processing, to transport intermediates between
performance while satisfying product quality and demand multisite refining facilities, and to transport finished products
requirements. The finished blend products are stored in storage from product storage tanks to distribution centers and finally to
tanks, and then the products are lifted during a specific time the customers. Any disruptions arising in the global supply chain
period to satisfy demand orders. While blending is a critical can have tremendous adverse effects in achieving operational
step in refinery operations, it is also the most complex of the efficiency, maintaining quality, profitability, and customer
subproblems. The blend scheduling problem gives rise to a satisfaction. The adverse events may happen due to uncertainty
nonlinear mixed integer model due to bilinear terms. The in supply of crude, demand, transportation, market volatility,
problem is further complicated by including some important and political climate. To effectively model a supply-chain design
logistics details. These logistics details are sequence dependent problem, the dynamics of the supply chain ought to be
switchovers, multipurpose product tanks and blender units, considered and data aggregation techniques for the extensive
minimum run-length requirements, fill-draw-delay, constant data set should be employed.
blend rate, one-flow out of blender, maximum heel quantity, As in the other chemical industries, production planning in
and downgrading of product to lower quality. The variables the petroleum refining industry relies on the following sources
associated with logistics details are combinatorial variables and of information: process topology, plant capacity, operating
have an exactly one-to-one correspondence with quantity sizing constraints, market demand, and costs involved in the processes.
variables. This combinatorial characteristic of the logistics Planning level decisions are made based on the strategic level
problem makes the optimization problem NP hard. Kelly20 decisions. The objective of planning the production is to
emphasized the importance of logic details in the blending and efficiently take advantage of the company infrastructure so that
distribution problem and proposed a decomposition technique its products may be competitively offered to customers. Specif-
to deal with complexity present in the model due to logistics ically, the refining industry is under immense pressure to
and bilinear terms. They decompose the problem into two produce cleaner products but faces low economic margins
subproblems: quality and logistic. A simultaneous production because of stricter environmental regulations and depressed
and blending and distribution scheduling problem for large scale market demand. In this situation, refinery planning becomes very
refinery with logistics details is addressed by Shah and Iera- important as it can exploit all potential opportunities to push
petritou,21 in which they assume linear blend property relations the economic margin to the maximum limit. A planning problem
to avoid any nonlinearity. They propose valid inequalities to determines how to best allocate the production, distribution, and
reduce the computational effort of a large scale mixed integer storage resources in the chain to respond to market demand
programming model. orders and forecasts in an economically efficient manner, and a
Overall, the problem of refinery operations is very compli- scheduling problem determines the detailed schedule for shorter
cated and if simultaneous optimization of all the components period (typically 10 days) by taking into consideration the
of a refinery supply chain is attempted, the resulting scheduling operational constraints of the system. Scheduling models use
and planning models will be highly intractable. Thus, in practice small bucket time-periods which means that only one activity,
each of the subproblems is addressed individually, and tech- task, job, or operation can be performed at any given time on
niques are proposed to integrate the solutions of the subprob- a piece of equipment. Scheduling models are in general order-
lems. An overview of the existing work in the area of enterprise- driven, which means that scheduling problems are driven by
wide optimization for refinery operations is given in next section real product demand and typically accompanied by a customer
1164 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011
purchase order. Owing to the complexity of refinery operations, flow rate, the recoverable oil or gas volume, and the water
commercial tools for production scheduling are few and these breakthrough time of the reservoir, and those uncertainties are
do not allow a rigorous representation of plant particularities. assumed to be gradually revealed as a function of design and
Uncertainty also persists at the planning and scheduling level operation decisions.
due to machine breakdown, rush orders, and price fluctuation. 3.1.2. Multisite Supply-Chain Planning. In a typical pe-
A real time optimization model deals with process level troleum supply chain, a set of crude oil suppliers and refineries
optimization to determine the product yields, temperatures, and are interconnected by intermediate and final product streams
pressures at each process units. These real times level decisions and a set of distribution centers. Thus multisite supply and
are made for very short times (seconds), and the process distribution planning is an essential part of an oil-refinery
operational complexities give rise to nonlinear dynamic process supply-chain design and has also received lots of attention.
models. Considering the details of operations at different sites and
An integrated optimization of strategic, tactical, operational, integrating the interconnectivity between sites and the appropri-
and inventory planning and scheduling level decisions under ate logistics are the main issues on addressing this problem.
uncertainty gives an intractable computationally intensive large- Julka et al.5,27 proposed an agent-based framework for supply-
scale nonlinear model. A further complexity in implementing a chain decision support systems (DSSs) and demonstrated its
globally optimal enterprise solution is the conflicting objectives application through a prototype system for crude procurement.
employed by different components of the supply chain. The best The system serves as a central DSS through which all processes
way to address decisions at different levels in an integrated of a refinery can be studied and enables integrated decisions
manner is through decomposition techniques, heuristics, and with respect to the overall refinery business. Rejowski and
simulation-based optimization methods. Pinto28 proposed a general framework for modeling petroleum
supply chains. Nodes of the chain are considered as grouped
3. Literature Review elementary entities that are interconnected by intermediate
streams. The supply-chain topology is then built by connecting
3.1. Petroleum Supply-Chain Design and Planning. The
the nodes representing refineries, terminals, and pipeline
petroleum industry can be characterized as a typical supply chain.
networks. Decision variables include streamflow rates, proper-
All levels of decisions (strategic, tactical, and operational) arise in
ties, operational variables, and inventory and facilities assign-
such a supply chain. In the literature, optimization models deal
ment. Persson and Gothe-Lundgren29 suggested an optimization
with planning and scheduling of several subsystems of the
model which integrates the shipment planning and the process
petroleum supply chain such as oil field infrastructure, crude oil
scheduling at the refineries. This problem concerns the simul-
supply, refinery operations, and product transportation. In what
taneous planning of how to route a fleet of ships and the
follows, developments on the petroleum supply-chain design and
planning are reviewed on the basis of the following two classifica- planning of which products to transport in these ships. Nishi
tions: oil field infrastructure investments and operations and et al.30 proposed a framework for distributed optimization of
petroleum supply-chain planning including crude oil worldwide supply-chain planning using an augmented Lagrangian decom-
transportation and multisite distribution planning. position and coordination approach. The proposed method is
3.1.1. Oil Field Infrastructure Design and Operations. Oil applied to supply-chain planning problems for a petroleum
field development represents a complex and expensive under- complex and a midterm planning problem for multiple compa-
taking in the oil industry. The problem is characterized by long nies, and a near-optimal solution was derived. MirHassani31
planning horizons and a large number of choices of platforms, showed how the capacitated network may be used to analyze
wells, and fields, and their interconnecting pipeline infrastruc- possible long-term transportation of oil derivatives by pipeline,
ture. The main challenges in resolving this problem is the truck, railway, and ship and reduce the distribution cost. The
complexity of the models usually characterized by uncertainty capacitated network can deal with the scheduling of a multi-
and the large dimensionality of the problem due to large time product, multidepot system receiving a number of petroleum
horizon and spatial domain. products from different refineries and distributes them among
Van den Heever and Grossmann23 presented a multiperiod several depots and market areas while the demand is an
mixed-integer nonlinear programming model for offshore oil uncertain parameter. Pitty et al.32 presented a dynamic model
field infrastructure planning, where nonlinear reservoir behavior of an integrated refinery supply chain. The model explicitly
is incorporated directly into the formulation. Goel and Gross- considers the various supply-chain activities such as crude oil
mann24 considered the optimal investment and an operational supply and transportation along with intrarefinery supply-chain
planning of gas field developments under uncertainty in gas activities such as procurement planning, scheduling, and opera-
reserves, where the shape of the scenario tree associated with tions management. Stochastic variations in transportation, yields,
the problem depends on the investment decisions. They prices, and operational problems are also considered in the
presented a stochastic programming model that incorporates the proposed model.
decision-dependence of the scenario tree and proposed a Al-Qahtani and Elkamel33 addressed the design and analysis
decomposition-based approximation algorithm for the solution of multisite integration and coordination strategies within a
of this model. Carvalho and Pinto25 proposed an optimization network of petroleum refineries using different crude combina-
model for the planning of an infrastructure in offshore oil fields. tion alternatives. Recently, they extended the work to address
The proposed model determines the existence of a given set of the design of optimal integration and coordination within a
platforms and their potential connection with wells, as well as multisite refinery and petrochemical system.34 The refinery and
the timing of extraction and production rates. Tarhan et al.26 petrochemical systems were modeled as a mixed-integer prob-
proposed a multistage stochastic programming model that lem with the objective of minimizing the annualized cost over
captures the economic objectives and nonlinear reservoir a given time horizon among the refineries and maximizing the
behavior and simultaneously optimizes the investment and added value of the petrochemical network. Carneiro et al.35
operating decisions over the entire planning horizon. The analyzed the strategic planning of an oil supply chain. To
uncertainties considered are in the initial maximum oil or gas optimize this chain, a two-stage stochastic model with fixed
Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1165
recourse and incorporation of risk management was developed. and also local consumer markets which receive large amounts
The model took a scenario-based approach and addressed three of refinery products. Herran et al.44 also proposed a discrete
sources of uncertainty. mathematical approach to solve short-term operational planning
Although significant progress has been made in the supply- of multipipeline systems for refined products. Jetlund an
chain design and planning for petroleum refining industry, the Karimi45 developed an optimization model to obtain the
production plan of an industrial supply chain is in general maximum-profit scheduling of a fleet of multiparcel tankers
created first and a compatible schedule is then identified engaged in shipping bulk liquid chemicals.
accordingly. Because the detailed scheduling constraints are Comillier et al.46,47 studied the petrol station replenishment
often ignored in the planning model, there is no guarantee that problem with time windows (PSRPTW) which aims to optimize
an operable schedule can be obtained with this hierarchical the delivery of several petroleum products to a set of petrol
approach. To address this issue, it is necessary to propose an stations using a limited heterogeneous fleet of tank-trucks. They
efficient formulation to coordinate various planning and sched- described two heuristics based on arc preselection and on route
uling decisions for optimizing the supply-chain performance.36 preselection. Extensive computational tests on randomly gener-
Solving this kind of model can yield the proper procurement ated instances confirm the efficiency of the proposed heuristics.
scheme for crude oils, the schedules for producing various Abraham and Rao48 made a study on the existing practices of
petrochemical products, and the corresponding logistics. The production planning, scheduling and prevailing constraints in
appropriate sources (suppliers) of raw materials, the economic the six plants of a lube oil section in a petroleum refinery. On
order quantities, the best purchasing intervals, and also the the basis of the data collected from these plants, some generative
transportation schedules can be identified accordingly. and evaluative models were developed. The generative models
Finally, it is worth mentioning that most of the oil industry developed were flow network optimization model and binary
companies still operate their planning, central engineering, integer linear programming model. The evaluative model
upstream operations, refining, and supply and transportation developed was simulation.
groups as complete separate entities. Therefore, systematic 3.2. Refinery Planning. Planning models, or more generally
methods for efficiently managing the petroleum supply chain time incremented microeconomic models, are in general mul-
as one entity must be exploited. In the next sections the work tiperiod and traditionally use what are known as big-bucket time-
that has appeared in the literature to cover these separate period models compared to small-bucket time-period models
problems are reviewed. for scheduling operations. Planning models are forecast-driven,
3.1.3. Pipeline and Transportation Scheduling. The pipe- which means that the quantity and quality specifications of the
line network plays a key role in the petroleum business. These demanded products are only forecasted or estimates for the time-
operational systems provide connections between ports and/or periods into the future.
oil fields and refineries (upstream), as well as between these 3.2.1. Production Planning Model. The availability of LP-
and consumer markets (downstream). Transportation is among based commercial software for refinery production planning,
the basic challenges in a refinery supply chain with the such as RPMS49 and PIMS,50 has allowed the development of
dimensionality of the problem and the specificities of each general production plans of the whole refinery, which can be
individual implementation to be the main issues. interpreted as general trends. However, inaccuracy caused by
The system discussed by Rejowski and Pinto37 is composed nonrigorous linear models may reduce the overall profitability
of a petroleum refinery, a multiproduct pipeline connected to or sacrifice product quality. Major advances in this area will be
several depots, and the corresponding consumer markets that on model refinement, notably through the use of nonlinear
receive large amounts of gasoline, diesel, LPG, and aviation programming.
fuel. An MILP optimization model that is based on a convex- As pointed out by Kelly,51 there are four major driving-forces
hull formulation is proposed for the scheduling system. In their pressing us to formulate planning models with nonlinearities:
later work, Rejowski and Pinto38 divided the pipeline into complex government regulations, increasingly expensive raw
segments that connect two consecutive depots and packs that materials of poorer quality, new and more sophisticated produc-
contain one product and that compose the segments. Recently, tion processes, and higher energy, chemical, and utility costs.The
Rejowski and Pinto39 proposed a MINLP formulation based on refinery planning model uses the nlp process models and blend
a continuous time representation for the scheduling of multi- relations. Pinto and Moro52 developed a nonlinear planning
product pipeline systems that must supply multiple consumer model for refinery production. The model allows the imple-
markets. The proposed continuous time representation is mentation of nonlinear process models as well as blending
compared with the previously developed discrete time repre- relations. Joly et al.53 presented a nonlinear planning model that
sentation of Rejowski and Pinto28 in terms of solution quality represents a general refinery topology and allows implementation
and computational performance. of nonlinear process models as well as blending relations. The
Cafaro and Cerda40 studied the scheduling of a multiproduct optimization model is able to define new operating points, thus
pipeline system receiving a number of liquid products from a increasing the production of the more valuable products and
single refinery source to distribute them among several depots. simultaneously satisfying all specification constraints. Neiro and
The problem of scheduling a transmission pipeline carrying Pinto54 proposed a nonlinear petroleum production planning
several petroleum products from a single oil refinery to a unique model, which incorporates multiple planning periods and the
distribution center over a monthly horizon is studied in their selection of different crude oil types. Uncertainties related to
later work.41 Recently, they further introduced a continuous petroleum and product prices as well as demand are also
formulation for the scheduling of multiple-source pipelines included as a set of discrete probabilities. Li et al.55 presented
operating on the fungible or segregated mode.42 a refinery planning model that utilizes simplified empirical
MirHassani and Ghobanalizadeh43 presented an integer nonlinear process models with considerations for crude char-
programming approach to oil derivative transportation schedul- acteristics, product yields, and qualities, etc.
ing. The system reported is composed of an oil refinery, one Refinery production planning is usually associated with the
multibranch multiproduct pipeline connected to several depots crude oil unloading and product distribution problems. These
1166 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011
problems have traditionally been solved separately because an 3.2.2. Uncertainty in Refinery Planning. Uncertainty arises
overall optimization of the problems together could render the in realistic decision making processes and has huge impact on
solution of the planning problem intractable. However, with the refinery planning activities. Three major uncertainties that
the development of computation algorithm and hardware, the should be considered in refinery production planning include
integrated solution received lots of attention in the recent past. (1) market demand for products; (2) prices of crude oil and the
Gao et al.56 addressed a production planning optimization saleable products; and (3) product (or production) yields of crude
problem of overall refinery using a mixed integer linear oil from chemical reactions in the primary crude distillation unit.
programming model, which considers the main factors for Dempster et al.63 applied stochastic programming modeling
optimizing the production plan of overall refinery related to the and solution techniques to planning problems for a consortium
use of run-modes of processing units. The aim of this planning of oil companies. A multiperiod supply, transformation, and
is to decide which run-mode to use in each processing unit in distribution scheduling problem is formulated for strategic or
each period of a given horizon, to satisfy the demand, while tactical level planning of the consortium’s activities. This
the total cost of production and inventory is minimized. deterministic model is used as a basis for implementing a
Guyonnet et al.57 explored benefits of the integration of stochastic programming formulation with uncertainty in the
production planning with these two models. Alabi and Castro58 product demands and spot supply costs. Li et al.64 presented an
presented a mathematical model of the refinery operations approach to address refinery planning under uncertainty. They
characterized by complete horizontal integration of subsystems proposed “loss function” to calculate the expectation of plant
from crude oil purchase through product distribution. To avoid revenues. The decision maker’s service objectives: confidence
complexity of the overall planning problem, after the identifica- level and fill rate are applied to handle possible unmet customer
tion of the nonzero structure of the constraints matrix, structure- demands. Neiro and Pinto65 presented a stochastic multiperiod
exploiting techniques such as the Dantzig-Wolfe technique and model for representing a petroleum refinery. Uncertainty is taken
block coordinate-descent decomposition were applied. into account in parameters such as demands, product sale prices,
Generally, a refinery complex consists of a process system and crude oil prices. Lagrangean decomposition was applied to
and a utility system. The process system not only produces exploit the block-diagonal structure of the problem and to reduce
liquefied petroleum gas, gasoline, diesel, and so on, but also solution time by decomposing the model on a temporal basis.
some byproducts, such as fuel gas and residual fuel oil, which Khor et al.66 proposed a two-stage stochastic programming
supply the utility system as fuel. The utility system converts model with fixed recourse via scenario analysis with incorpora-
fuel gas and fuel oil to high pressure or medium pressure steam tion of risk management for an optimal midterm refinery
and electricity to meet the energy demand of the process system. planning that addresses three factors of uncertainty: prices of
The integrated process system and utility system optimization crude oil and saleable products, product demands, and product
has also received attention recently. Zhang and Hua59 presented yields. Li et al.67 developed a stochastic programming model
an approach to address the integration of the process system for refinery production planning under demand uncertainty with
and utility system for better energy utilization. A plant-wide uniform distribution assumption, and then a hybrid programming
multiperiod planning mathematical model is proposed with the model incorporating the linear programming model with the
process unit energy-consumed model embedded in the plant- stochastic programming model by a weight factor is proposed.
wide model to gain the overall optimization and for better energy Subsequently, piecewise linear approximation functions are
efficiency. Zhang et al.60 presented a method for overall refinery derived and applied to solve the hybrid programming model
optimization through integration of the hydrogen network and under a uniform distribution assumption.
the utility system with the material processing system. To make In the literature, several studies also addressed the issue of
the problem of overall optimization solvable, they adopted a uncertainty and studied the financial risk aspects in refinery
decomposition approach, in which material processing is operations planning. Pongsakdi et al.68 studied the problem of
optimized first using linear programming (LP) techniques to determining what crude to purchase and to decide on the
maximize the overall profit, and supporting systems, including production level of different products given forecasts of
the hydrogen network and the utility system, are optimized next demands. The profit is maximized taking into account revenues,
to reduce operating costs for the fixed process conditions crude oil costs, inventory costs, and cost of unsatisfied demand.
determined from the LP optimization. Lakkhanawat and Bagajewicz69 addressed the similar problem.
The capacity of the world petroleum refining industry has Park et al.70 developed an integrated model based on two-stage
increased rapidly during the past decades. On the one hand, it stochastic programming for operational planning and financial
plays a very important role in international economics and in risk management of a refinery. Downside risk, which rationally
our daily life, and on the other hand environmental regulations quantifies financial risk, is selected as the objective function to
and risks of climate change are pressuring the refinery industry be minimized. Subsequently, the contract sizes and the opera-
to minimize its greenhouse gas emissions. Elkamel et al.61 tional plan are optimized on the basis of the developed model
proposed a mixed-integer nonlinear programming (MINLP) and the price scenarios.
model for the production planning of refinery processes to In summary, with the development in mathematical program-
achieve maximum operational profit while reducing CO2 emis- ming and computation facility, the refinery planning operations
sions to a given target through the use of different CO2 tend to consider more accurate and complex nonlinear process
mitigation options. The objective of the MINLP model is to models, implementing overall optimization among different
determine suitable CO2 mitigation options for a given reduction subsystems in the whole plant, and also tend to systematically
target while meeting the demand of each final product and its address the various uncertainties, with financial risk and finally
quality specifications and simultaneously maximizing profit. environmental aspect considerations.
Zhang et al.62 proposed a model-centered approach with an Furthermore, in the past work, uncertainty is in general
eight-step procedure for the early planning and design of an considered as a set of discrete scenarios, each representing a
eco-industrial park around an oil refinery, considering technical possible shifting of market expectations. Every environment is
and environmental factors. weighted through an expected probability of occurrence. Previous
Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1167
work revealed that the computational effort of uncertain multiperiod The product blending process involves mixing various stocks,
refinery production planning models grows exponentially with the which are the intermediate products from the refinery, along
number of time periods and scenarios. Therefore, in order to reduce with some additives, such as antioxidants and corrosion inhibi-
the computational effort over uncertain long-planning horizons, tors, to produce blends with certain properties. A variety of
special techniques must be employed. support systems have been also developed to address the refinery
3.3. Refinery Scheduling. The literature generally addressed blending operations. Jia and Ierapetritou82 studied the problem
the refinery scheduling in the form of smaller subproblems: of gasoline blending and distribution. The problem involves the
crude oil operations from unloading to the charging into a crude optimal operation of gasoline blending, the transfer to product
distillation unit; blending of intermediate products from the CDU stock tanks, and the delivering schedule to satisfy all of the
into finished products; lifting or delivery of the finished products, orders. An efficient mixed-integer linear programming formula-
etc. In what follows, we review literature in the above scheduling tion is developed based on continuous representation of the time
applications. domain and on the assumption of a fixed recipe. Joly and Pinto83
3.3.1. Crude Oil Scheduling. The scheduling of crude oil developed mixed-integer programming (MIP) models for a real-
operations is crucial to petroleum refining, which includes world fuel oil and asphalt production scheduling problem.
determining the times and sequences of crude oil unloading, Mendez et al.84 presents a MILP-based method that addresses
blending, and CDU feeding. The infrastructure that serves to the simultaneous optimization of the off-line blending and the
facilitate crude oil loading and unloading operations usually short-term scheduling problem in oil-refinery applications.
comprises (a) one or more jetties, (b) crude storage tanks at Depending on the problem characteristics as well as the required
port and refinery, (c) intermediate tanks between port and flexibility in the solution, the model can be based on either a
refinery, and (d) pipeline infrastructure connecting port tanks discrete or a continuous time domain representation. To preserve
to refinery. In the past decade, many approaches have been the model’s linearity, an iterative procedure is proposed to
proposed for solving this problem. effectively deal with nonlinear gasoline properties and variable
Shah71 presented one of the earliest mathematical formula- recipes for different product grades. Lee et al.85 addressed a
tions for the crude oil scheduling problem. Their model is based naphtha feeding problem for the Naphtha Cracking Center
on a discrete time presentation. Not in their earlier studies,72-74 (NCC). The naphtha feeding problem involves two key opera-
but in a recent work of Pinto et al.75 were the scheduling tions: delivering naphtha from refineries to NCC and blending
problems in oil refineries that are formulated as mixed integer naphtha in storage tanks before feeding it to NCC. This paper
optimization models and rely on both continuous and discrete considers both issues simultaneously by transforming them into
time representations discussed. Chryssolouris et al.76 addressed a single mixed linear integer programming problem of minimiz-
primarily the scheduling of a refinery importing various types ing the cost function of naphtha prices, shipping expenses, and
of crude oil. Karuppiah et al.77 presented an outer-approximation unloading costs, etc. In a recent publication Li et al.86 developed
a slot-based MILP formulation for an integrated treatment of
algorithm to obtain the global optimum of a nonconvex mixed-
recipe, specifications, blending, and storage, considering real-
integer nonlinear programming (MINLP) model for the crude
life features such as multipurpose product tanks, parallel
oil scheduling problem. The model relies on a continuous time
nonidentical blenders, minimum run lengths, changeovers,
representation making use of transfer events. The proposed
piecewise constant profiles for blend component qualities, and
algorithm focuses on effectively solving a MILP relaxation of
feed rates, etc.
the nonconvex MINLP to obtain a rigorous lower bound (LB)
In most of the existing refinery scheduling studies, the
of the global optimum. The solution of this relaxation is used
problem data are assumed to be deterministic. However, under
as a heuristic to obtain a feasible solution to the MINLP which
the current situations of unsteady supply of crude oil, variations
serves as an upper bound (UB). The lower and upper bounds
of feedstock qualities and yield levels, pressure of intense time,
are made to converge to within a specified tolerance in the
low inventory flexibility, etc., it is very important for refinery
proposed outer-approximation algorithm. Pan et al.78 considered
companies to assess the extra cost of these changes during
the coastal and marine-access refineries with simplified work-
making their short-term crude oil scheduling operations. Cao
flow and proposed a mixed integer nonlinear programming
et al.87 proposed chance constrained mixed-integer nonlinear
formulation for crude oil scheduling. Finally, well scheduling
stochastic and fuzzy programming models for refinery short-
in petroleum fields is a very important activity related to crude
term crude oil scheduling problems under demand uncertainties
oil production, where decisions include the operational status
of distillation units. The scheduling problem studied has the
of wells (open or closed), the allocation of wells to manifolds characteristics of discrete event and continuous event coexist-
or separators, and the allocation of flow lines to separators. ence, multistage, multiproduct, nonlinear, uncertainty, and large
However, the work in that area is not reviewed in this paper, scale.
but it can be referred to in Kosmidis et al.79
In summary, a refinery scheduling problem is more complex
3.3.2. Refinery Production and Blend Scheduling. The than a planning problem since in general more operational
production unit scheduling problem is addressed by Jia and constraints should be considered and many combinatorial
Ierapetritou.4 Shah et al.80 presented a decomposition strategy decisions need to be optimized; thus it has been hard to
for solving a large scale refinery scheduling problem. They implement an overall scheduling for the whole refinery opera-
proposed a spatial decomposition scheme that generates smaller tions. An efficient decomposition-based solution strategy is still
subsystems that can be solved to global optimality instead of necessary to implement overall refinery scheduling optimization.
formulating a huge one for the centralized problem. Gothe- Finally, efficient reactive scheduling, rescheduling methodolo-
Lundgren et al.81 describe a production planning and scheduling gies are also imperative for refinery scheduling operations.8,88
problem in an oil refinery company. The aim of the scheduling
is to decide which mode of operation to use in each processing 4. Future Challenges
unit at each point in time in order to satisfy the demand while
minimizing the production cost and taking storage capacities This article has shown that a very large amount of work was
into account. undertaken to address the problems in refinery supply-chain
1168 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011
management, especially in the areas of the scheduling and (2) Varma, V. A.; Reklaitis, G. V.; Blau, G. E.; Pekny, J. F. Enterprise-
planning of refinery operations, crude oil procurement, and wide modeling and optimizationsAn overview of emerging research
challenges and opportunities. Comput. Chem. Eng. 2007, 31, 692–711.
petroleum supply-chain design and planning. However, a large
(3) Papageorgiou, L. G. Supply chain optimization for the process
number of issues still exist which provides challenges for industries: Advances and opportunities. Comput. Chem. Eng. 2009, 33,
ongoing research. 1931–1938.
4.1. Improved Models for Refinery Operations. The crude (4) Jia, Z.; Ierapetritou, M. Efficient short-term scheduling of refinery
oil scheduling and planning problem has been traditionally operations based on a continuous time formulation. Comput. Chem. Eng.
2004, 28, 1001–1019.
addressed using linear relations for CDU yield prediction as a
(5) Julka, N.; Karimi, I.; Srinivasan, R. Agent-based supply chain
function of the crude feed. Improvements in the swing cut model managements2: A refinery application. Comput. Chem. Eng. 2002, 26,
for CDU yield in recent years provide an opportunity to develop 1771–1781.
nonlinear planning models that capture the nonlinearity of the (6) Neiro, S. M.; Pinto, J. M. A general modeling framework for the
process. Implementing nonlinear process model equations for operational planning of petroleum supply chains. Comput. Chem. Eng. 2002,
planning problems using the latest NLP algorithms provides true 28, 871–896.
(7) Arief, A.; Rajagopalan, S.; Karimi, I. A. Heuristic rescheduling of
optimal and accurate solutions to the planning problem. crude oil operations to manage abnormal supply chain events. AIChE J.
Efficient models and solution algorithms for the blending and 2007, 53, 397–422.
distribution problem are needed to simultaneously address the (8) Adhitya, A.; Srinivasan, R.; Karimi, I. A. A model-based reschedul-
optimization of blend recipe and blend logistics. The bilinear ing framework for managing abnormal supply chain events. Comput. Chem.
Eng. 2007, 31, 496–518.
terms in the quality problem are a result of nonlinear property
(9) Kim, Y.; Yun, C.; Park, S. B.; Park, S.; Fan, L. T. An integrated
relations, while the bilinear term in the logistics problem is a model of supply network and production planning for multiple fuel products
result of a constant blending rate where blend duration and of multisite refineries. Comput. Chem. Eng. 2008, 32, 2529–2535.
amount can vary. Some research has been done to address the (10) Pitty, S. S.; Li, W.; Adhitya, A.; Srinivasan, R.; Karimi, I. A.
simultaneous optimization of quality, quantity, and logistics.20,86 Decision support for integrated refinery supply chains: Part 1. Dynamic
4.2. Integrated Models for Supply-Chain Management. simulation. Comput. Chem. Eng. 2008, 32, 2767–2786.
(11) Yang, J.; Gu, H.; Rong, G. Supply chain optimization for refinery
There is a great economic potential in enterprise-wide optimiza- with considerations of operation mode changeover and yield fluctuations.
tion; however, a lack of comprehensive optimization models Ind. Eng. Chem. Res. 2009, 49, 276–287.
and computational tools are one of the major issues that must (12) Iyer, R. R.; Grossmann, I. E.; Vasantharajan, S.; Cullick, A. S.
addressed. Optimal planning and scheduling of offshore oil field infrastructure
investment and operations. Ind. Eng. Chem. Res. 1998, 37, 1380–1397.
(1) Simultaneous refinery operation scheduling (production
(13) Cunha, J. C.; Demirdal, B.; Gui, P. Use of quantitative risk analysis
unit, blend, and distribution scheduling) models are very for uncertainty quantification on drilling operationssReview and lessons
sparse,89,21 and they do not incorporate accurate process models learned. SPE Latin American and Caribbean Petroleum Engineering
(e.g., assumptions of linear blending rules). Moreover the Conference, Rio de Janeiro, Brazil, 2005.
consideration of detailed dynamics and feedback loops will (14) Wu, N.; Chu, F.; Chu, C.; Zhou, M. Tank cycling and scheduling
provide a better representation of the realistic problem. analysis of high fusion point oil transportation for crude oil operations in
refinery. Comput. Chem. Eng. 2010, 34, 529–543.
(2) Integrated scheduling planning models with nonlinear (15) Saharidis, G. K. D.; Ierapetritou, M. G. Scheduling of loading and
scheduling process operations complexity and coordinated unloading of crude oil in a refinery with optimal mixture preparation. Ind.
control/real time and scheduling decisions making models are Eng. Chem. Res. 2009, 48, 2624–2633.
still not adequately explored. Efficient decomposition algorithms (16) Kelly, J. D.; Mann, J. L. Crude-oil blend scheduling optimization:
An application with multi-million dollar benefits: Part II. Hydrocarbon
and computational models that integrate the decision making
Process. 2003b, 72–79.
at different time scales are required. (17) Kelly, J. D.; Mann, J. L. Crude-oil blend scheduling optimization:
(3) Coordinated decision making between multisite production An application with multi-million dollar benefits: Part I. Hydrocarbon
facilities and the design of the supply chain is vital for improving Process. 2003a, 47–53.
profit margin while maintaining customer satisfaction. The (18) Shah, N.; Saharidis, G. K. D.; Jia, Z.; Ierapetritou, M. G.
Centralized-decentralized optimization for refinery scheduling. Comput.
challenge in this area is to develop optimization models and Chem. Eng. 2009, 33, 2091–2105.
computational tools that capture inherent complexities and (19) Gary, J. H.; Handwerk, G. E.; Kaiser, M. J. Petroleum Refining:
uncertainties present across modern supply chains. Technology and Economics, 5th ed.; CRC Press: Boca Raton, FL, 2007.
New comprehensive models and algorithms can significantly (20) Kelly, J. D. Logistics: The missing link in blend scheduling
optimization. Hydrocarbon Process. 2006, 45–51.
improve the economic performance of refinery operations by
(21) Shah, N. K.; Ierapetritou, M. G. Short-term scheduling of a large-
reducing costs and increasing profits while providing solutions scale oil-refinery operations: Incorporating logistics details. AIChE J. 2010,
for real world applications. Moving away from simulation-based DOI 10.1002/aic.12359.
approaches to optimization models will likely occur in the next (22) Simchi-Levi, D.; Kaminsky, P.; Simchi-Levi, E. Managing the
years. However there are still many issues, including issues of Supply Chain: Concepts, Strategies, And Case Studies; McGraw-Hill: New
York, 2004.
interfaces and communication barriers among different refinery
(23) van den Heever, S. A.; Grossmann, I. E. An iterative aggregation/
units. disaggregation approach for the solution of a mixed-integer nonlinear oil
field infrastructure planning model. Ind. Eng. Chem. Res. 2000, 39, 1955–
1971.
Acknowledgment (24) Goel, V.; Grossmann, I. E. A stochastic programming approach to
planning of offshore gas field developments under uncertainty in reserves.
The authors gratefully acknowledge financial support from Comput. Chem. Eng. 2004, 28, 1409–1429.
National Science Foundation under Grants CBET 0625515 and (25) Carvalho, M. C. A.; Pinto, J. M. An MILP model and solution
0966861. technique for the planning of infrastructure in offshore oil fields. J. Pet.
Sci. Eng. 2006, 51, 97–110.
(26) Tarhan, B.; Grossmann, I. E.; Goel, V. Stochastic programming
Literature Cited approach for the planning of offshore oil or gas field infrastructure under
decision-dependent uncertainty. Ind. Eng. Chem. Res. 2009, 48, 3078–3097.
(1) Grossmann, I. Enterprise-wide optimization: A new frontier in (27) Julka, N.; Srinivasan, R.; Karimi, I. Agent-based supply chain
process systems engineering. AIChE J. 2005, 51, 1846–1857. managements1: Framework. Comput. Chem. Eng. 2002, 26, 1755–1769.
Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011 1169
(28) Rejowski, R.; Pinto, J. M. Efficient MILP formulations and valid (57) Guyonnet, P.; Grant, F. H.; Lagajewicz, M. J. Integrated model
cuts for multiproduct pipeline scheduling. Comput. Chem. Eng. 2004, 28, for refinery planning, oil procuring, and product distribution. Ind. Eng. Chem.
1511–1528. Res. 2009, 48, 463–482.
(29) Persson, J. A.; Gothe-Lundgren, M. Shipment planning at oil (58) Alabi, A.; Castro, J. Dantzig-Wolfe and block coordinate-descent
refineries using column generation and valid inequalities. Eur. J. Operat. decomposition in large-scale integrated refinery-planning. Comput. Operat.
Res. 2005, 163, 631–652. Res. 2009, 36, 2472–2483.
(30) Nishi, T.; Shinozaki, R.; Konishi, M. An augmented Lagrangian (59) Zhang, B. J.; Hua, B. Effective MILP model for oil refinery-wide
approach for distributed supply chain planning for multiple companies. IEEE production planning and better energy utilization. J. Clean. Prod. 2007,
Trans. Autom. Sci. Eng. 2008, 5, 259–274. 15, 439–448.
(31) MirHassani, S. A. An operational planning model for petroleum (60) Zhang, J.; Zhu, X. X.; Towler, G. P. A simultaneous optimization
products logistics under uncertainty. Appl. Math. Comput. 2008, 196, 744– strategy for overall integration in refinery planning. Ind. Eng. Chem. Res.
751. 2001, 40, 2640–2653.
(32) Pitty, S. S.; Li, W. K.; Adhitya, A.; Srinivasan, R.; Karimi, I. A. (61) Elkamel, A.; Ba-Shammakh, M.; Douglas, P.; Croiset, E. An
Decision support for integrated refinery supply chains. Part 1. Dynamic optimization approach for integrating planning and CO2 emission reduction
simulation. Comput. Chem. Eng. 2008, 32, 2767–2786. in the petroleum refining industry. Ind. Eng. Chem. Res. 2008, 47, 760–
(33) Al-Qahtani, K.; Elkamel, A. Multisite facility network integration 776.
design and coordination: An application to the refining industry. Comput. (62) Zhang, X. P.; Stromman, A. H.; Solli, C.; Hertwich, E. G. Model-
Chem. Eng. 2008, 32, 2189–2202. centered approach to early planning and design of an eco-industrial park
(34) Al-Qahtani, K.; Elkamel, A. Multisite refinery and petrochemical around an oil refinery. EnViron. Sci. Technol. 2008, 42, 4958–4963.
network design: optimal integration and coordination. Ind. Eng. Chem. Res. (63) Dempster, M. A. H.; Pedron, N. H.; Medova, E. A.; Scott, J. E.;
2009, 48, 814–826. Sembos, A. Planning logistics operations in the oil industry. J. Operat. Res.
(35) Carneiro, M. C.; Ribas, G. P.; Hamacher, S. Risk management in Soc. 2000, 51, 1271–1288.
the oil supply chain: A CVaR approach. Ind. Eng. Chem. Res. 2010, 49, (64) Li, W. K.; Hui, C. W.; Li, P.; Li, A. X. Refinery planning under
3286–3294. uncertainty. Ind. Eng. Chem. Res. 2004, 43, 6742–6755.
(36) Kuo, T. H.; Chang, C. T. Application of a mathematic programming (65) Neiro, S. M. S.; Pinto, J. M. Lagrangean decomposition applied to
model for integrated planning and scheduling of petroleum supply networks. multiperiod planning of petroleum refineries under uncertainty. Latin Am.
Ind. Eng. Chem. Res. 2008, 47, 1935–1954. Appl. Res. 2006, 36, 213–220.
(37) Rejowski, R.; Pinto, J. M. An MILP formulation for the scheduling (66) Khor, C. S.; Elkamel, A.; Douglas, P. L. Stochastic refinery planning
of multiproduct pipeline systems. Braz. J. Chem. Eng. 2002, 19, 467–474. with risk management. Pet. Sci. Technol. 2008, 26, 1726–1740.
(38) Rejowski, R.; Pinto, J. M. Scheduling of a multiproduct pipeline (67) Li, C. F.; He, X. R.; Chen, B. Z.; Xu, Q.; Liu, C. W. A hybrid
system. Comput. Chem. Eng. 2003, 27, 1229–1246. programming model for optimal production planning under demand
(39) Rejowski, R.; Pinto, J. M. A novel continuous time representation uncertainty in refinery. Chin. J. Chem. Eng. 2008, 16, 241–246.
for the scheduling of pipeline systems with pumping yield rate constraints. (68) Pongsakdi, A.; Rangsunvigit, P.; Siemanond, K.; Bagajewicz, M. J.
Comput. Chem. Eng. 2008, 32, 1042–1066. Financial risk management in the planning of refinery operations. Int. J.
Prod. Econ. 2006, 103, 64–86.
(40) Cafaro, D. C.; Cerda, J. Optimal scheduling of multiproduct pipeline
systems using a non-discrete MILP formulation. Comput. Chem. Eng. 2004, (69) Lakkhanawat, H.; Bagajewicz, M. J. Financial risk management
28, 2053–2068. with product pricing in the planning of refinery operations. Ind. Eng. Chem.
Res. 2008, 47, 6622–6639.
(41) Cafaro, D. C.; Cerda, J. Efficient Tool for the Scheduling of
(70) Park, J.; Park, S.; Yun, C.; Kim, Y. Integrated model for financial
Multiproduct Pipelines and Terminal Operations. Ind. Eng. Chem. Res. 2008,
risk management in refinery planning. Ind. Eng. Chem. Res. 2010, 49, 374–
47, 9941–9956.
380.
(42) Cafaro, D. C.; Cerda, J. Optimal Scheduling of Refined Products
(71) Shah, N. Mathematical programming techniques for crude oil
Pipelines with Multiple Sources. Ind. Eng. Chem. Res. 2009, 48, 6675–
scheduling. Comput. Chem. Eng. 1996, 20, S1227–1232.
6689.
(72) Lee, H. M.; Pinto, J. M.; Grossmann, I. E.; Park, S. Mixed-integer
(43) MirHassani, S. A.; Ghorbanalizadeh, M. The multiproduct pipeline
linear programming model for refinery short-term scheduling of crude oil
scheduling system. Comput. Math. Appl. 2008, 56, 891–897.
unloading with inventory management. Ind. Eng. Chem. Res. 1996, 35,
(44) Herran, A.; de la Cruz, J. M.; de Andres, B. A mathematical model 1630–1641.
for planning transportation of multiple petroleum products in a multipipeline (73) Jia, Z. Y.; Ierapetritou, M.; Kelly, J. D. Refinery short-term
system. Comput. Chem. Eng. 2010, 34, 401–413. scheduling using continuous time formulation: Crude-oil operations. Ind.
(45) Jetlund, A. S.; Karimi, I. A. Improving the logistics of multicom- Eng. Chem. Res. 2003, 42, 3085–3097.
partment chemical tankers. Comput. Chem. Eng. 2004, 28, 1267–1283. (74) Jia, Z. Y.; Ierapetritou, M. Efficient short-term scheduling of refinery
(46) Cornillier, F.; Boctor, F. F.; Laporte, G.; Renaud, J. A heuristic operations based on a continuous time formulation. Comput. Chem. Eng.
for the multiperiodpetrol station replenishment problem. Eur. J. Operat. 2004, 28, 1001–1019.
Res. 2008, 191, 295–305. (75) Pinto, J. M.; Joly, M.; Moro, L. F. L. Planning and scheduling
(47) Cornillier, F.; Laporte, G.; Boctor, F. F.; Renaud, J. The petrol models for refinery operations. Comput. Chem. Eng. 2000, 24, 2259–2276.
station replenishment problem with time windows. Comput. Operat. Res. (76) Chryssolouris, G.; Papakostas, N.; Mourtzis, D. Refinery short-
2009, 36, 919–935. term scheduling with tank farm, inventory, and distillation management:
(48) Abraham, J. N.; Rao, K. S. Integration of generative and evaluative An integrated simulation-based approach. Eur. J. Operat. Res. 2005, 166,
models for production scheduling of lube oil plants in a petroleum refinery. 812–827.
Appl. Math. Model. 2009, 33, 1213–1227. (77) Karuppiah, R.; Furman, K. C.; Grossmann, I. E. Global optimization
(49) RPMS (Refinery and Petrochemical Modeling Systems): A System for scheduling refinery crude oil operations. Comput. Chem. Eng. 2008,
Description; Bonner and Moore: Houston, TX, 1979. 32, 2745–2766.
(50) PIMS (Process Industry Modeling System) User’s Manual; Betchel (78) Pan, M.; Li, X. X.; Qian, Y. New approach for scheduling crude
Corp.: Houston, TX, 1993. oil operations. Chem. Eng. Sci. 2009, 64, 965–983.
(51) Kelly, J. D. Formulating production planning models. Chem. Eng. (79) Kosmidis, V. D.; Perkins, J. D.; Pistikopoulos, E. N. A mixed
Progress. 2004, 100, 43–50. integer optimization formulation for the well scheduling problem on
(52) Pinto, J. M.; Moro, L. F. L. A planning model for petroleum petroleum fields. Comput. Chem. Eng. 2005, 29, 1523–1541.
refineries. Braz. J. Chem. Eng. 2000, 17, 575–585. (80) Shah, N.; Saharidis, G. K. D.; Jia, Z. Y.; Ierapetritou, M. G.
(53) Joly, M.; Moro, L. F. L.; Pinto, J. M. Planning and scheduling for Centralized-decentralized optimization for refinery scheduling. Comput.
petroleum refineries using mathematical programming. Brazi. J. Chem. Eng. Chem. Eng. 2009, 33, 2091–2105.
2002, 19, 207–228. (81) Gothe-Lundgren, M.; Lundgren, J. T.; Persson, J. A. An optimiza-
(54) Neiro, S. M. S.; Pinto, J. M. Multiperiod optimization for production tion model for refinery production scheduling. Int. J. Product. Econ. 2002,
planning of petroleum refineries. Chem. Eng. Commun. 2005, 192, 62–88. 78, 255–270.
(55) Li, W. K.; Hui, C. W.; Li, A. X. Integrating CDU, FCC, and product (82) Jia, Z. Y.; Ierapetritou, M. Mixed-integer linear programming model
blending models into refinery planning. Comput. Chem. Eng. 2005, 29, for gasoline blending and distribution scheduling. Ind. Eng. Chem. Res.
2010–2028. 2003, 42, 825–835.
(56) Gao, Z.; Tang, L.; Jin, H.; Xu, N. N. An optimization model for (83) Joly, M.; Pinto, J. M. Mixed-integer programming techniques for
the production planning of overall refinery. Chin. J. Chem. Eng. 2008, 16, the scheduling of fuel oil and asphalt production. Chem. Eng. Res. Des.
67–70. 2003, 81, 427–447.
1170 Ind. Eng. Chem. Res., Vol. 50, No. 3, 2011
(84) Mendez, C. A.; Grossmann, I. E.; Harjunkoski, I.; Kabore, P. A (88) Li, Z. K.; Ierapetritou, M. Process scheduling under uncertainty:
simultaneous optimization approach for off-line blending and scheduling Review and challenges. Comput. Chem. Eng. 2008, 32, 715–727.
of oil-refinery operations. Comput. Chem. Eng. 2006, 30, 614–634. (89) Luo, C.; Rong, G. Hierarchical Approach for Short-Term Scheduling
(85) Lee, T.; Ryu, J. H.; Lee, H. K.; Lee, I. B. A study on scheduling in Refineries. Ind. Eng. Chem. Res. 2007, 46, 3656–3668.
of naphtha transportation and storage systems for naphtha cracking center.
Chem. Eng. Res. Des. 2010, 88, 189–196.
(86) Li, J.; Karimi, I. A.; Srinivasan, R. Recipe determination and ReceiVed for reView May 1, 2010
scheduling of gasoline blending operations. AIChE J. 2010, 56, 441–465. ReVised manuscript receiVed December 1, 2010
(87) Cao, C. W.; Gu, X. S.; Xin, Z. Chance constrained programming Accepted December 1, 2010
models for refinery short-term crude oil scheduling problem. Appl. Math.
Model. 2009, 33, 1696–1707. IE1010004