ArticlePDF Available

Artificial intelligence in supply chain management: Theory and applications

Authors:

Abstract and Figures

Artificial intelligence (AI) was introduced to develop and create “thinking machines” that are capable of mimicking, learning, and replacing human intelligence. Since the late 1970s, AI has shown great promise in improving human decision-making processes and the subsequent productivity in various business endeavors due to its ability to recognise business patterns, learn business phenomena, seek information, and analyse data intelligently. Despite its widespread acceptance as a decision-aid tool, AI has seen limited application in supply chain management (SCM). To fully exploit the potential benefits of AI for SCM, this paper explores various sub-fields of AI that are most suitable for solving practical problems relevant to SCM. In so doing, this paper reviews the past record of success in AI applications to SCM and identifies the most fruitful areas of SCM in which to apply AI.
Content may be subject to copyright.
PLEASE SCROLL DOWN FOR ARTICLE
This article was downloaded by:
[Min, Hokey]
On:
12 February 2010
Access details:
Access Details: [subscription number 919227105]
Publisher
Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK
International Journal of Logistics Research and Applications
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713435362
Artificial intelligence in supply chain management: theory and
applications
Hokey Min
a
a
Management, College of Business Administration, Bowling Green State University, Bowling Green,
OH, USA
First published on: 24 March 2009
To cite this Article Min, Hokey(2010) 'Artificial intelligence in supply chain management: theory and applications',
International Journal of Logistics Research and Applications, 13: 1, 13 — 39, First published on: 24 March 2009 (iFirst)
To link to this Article: DOI: 10.1080/13675560902736537
URL: http://dx.doi.org/10.1080/13675560902736537
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
International Journal of Logistics: Research and Applications
Vol. 13, No. 1, February 2010, 13–39
Artificial intelligence in supply chain management: theory
and applications
Hokey Min*
Management, College of Business Administration, Bowling Green State University, Bowling Green,
OH 43403, USA
(Received 12 June 2008; final version received 29 October 2008)
Artificial intelligence (AI) was introduced to develop and create “thinking machines” that are capable of
mimicking,learning, andreplacinghuman intelligence.Since thelate 1970s,AIhasshowngreat promise in
improvinghumandecision-makingprocessesandthesubsequentproductivityinvariousbusinessendeavors
due to its ability to recognise business patterns, learn business phenomena, seek information, and analyse
data intelligently. Despite its widespread acceptance as a decision-aid tool,AI has seen limited application
in supply chain management (SCM). To fully exploit the potential benefits of AI for SCM, this paper
explores various sub-fields of AI that are most suitable for solving practical problems relevant to SCM. In
so doing, this paper reviews the past record of success in AI applications to SCM and identifies the most
fruitful areas of SCM in which to apply AI.
Keywords: artificial intelligence; supply chain management; knowledge management; literature review
1. Introduction
In an era of greater demand uncertainty, higher supply risk, and increasing competitive intensity,
supplychain (SC) excellenceoften hinges on the organisation’sability to integrateand orchestrate
the entire spectrum of end-to-end processes of acquiring materials or components, converting
them into finished goods, and delivering them to customers. Since such ability can be enhanced
by increased visibility across the end-to-end SC processes, many leading-edge organisations have
attempted to enrich their information sources and share real-time information with SC partners.
Thus, SC management (SCM) is becoming more information intensive and its focus has been
directed toward the substitution of assets (e.g., inventory, warehouses, transportation equipment)
with information. Recognising the increasing significance of information to SC success, SC
professionals have explored various ways to better manage information and leverage it to make
better business decisions. One of those ways may include artificial intelligence (AI) that has been
in existence for decades, but has not been fully utilised in the area of SCM.
In general, AI is referred to as the use of computers for reasoning, recognising patterns, learn-
ing or understanding certain behaviors from experience, acquiring and retaining knowledge, and
*Email: hmin@bgsu.edu
ISSN 1367-5567 print/ISSN 1469-848X online
© 2010 Taylor & Francis
DOI: 10.1080/13675560902736537
http://www.informaworld.com
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
14 H. Min
developing various forms of inference to solve problems in decision-making situations where
optimal or exact solutions are either too expensive or difficult to produce (Nilsson 1980, Russell
and Norvig 1995, Luger 2002). Put simply, the main objectives of AI are to understand the phe-
nomenonof human intelligence andto design computer systems that can mimichuman behavioral
patterns and create knowledge relevant to problem-solving. Thus, AI should have the ability to
learn and comprehend new concepts, learn from experience (“on-their-own”), perform reasoning,
draw conclusions, impute meaning, and interpret symbols in context. Due to such ability, AI has
been successfully applied in areas such as game playing, semantic modeling, human performance
modeling, robotics, machine learning, data mining, neural networks, genetic algorithms (GAs),
and expert systems (Russell and Norvig 1995, Luger 2002).
One area of AI’s potential application that has not yet been fully explored is the emerging
management philosophy of SCM, which requires the comprehension of complex, interrelated
decision-making processes and the creation of intelligent knowledge bases crucial for joint
problem-solving. For example, Eastman Kodak once structured the thinking processes of expe-
rienced order pickers and then developed a rule-based expert system to select the optimal
order-picking path in a warehouse (Allen and Helferich 1990). Also, in an effort to synchronise
a series of interrelated but different stages of joint demand planning and forecasting processes in
the SC, Min and Yu (2008) proposed an agent-based forecasting system that has the capability
to predict end customer demand through information exchange among multiple SC partners and
learn from the past forecasting experience. As illustrated by these examples, some sub-fields of
AI such as expert systems and agent-based systems can be useful for dealing with various aspects
(e.g., warehousing, joint demand planning, inventory control) of the SC. With this illustration in
mind, the main objectives of this paper are to:
(1) Identify the sub-fields ofAI that are most suitable for SCM applications and then characterise
those sub-fields in terms of their usefulness for improving SC efficiency.
(2) Synthesise the existing literature dealing with the applications of AI to SCM with respect to
their practical implications and technical merits.
(3) Develop a hierarchical taxonomy for the existing AI literature and categorise it according to
its SCM application areas, problem scope, and methodology.
(4) SummariseAI research trends and identify the potential SCM application areas that have not
been explored.
(5) Discuss the future outlook for extensions of existing AI literature and untapped AI research
topics relevant to SCM.
2. The taxonomy of the AI literature
To gaina birds-eyeviewofpastAI studies,we develop ataxonomy usingthree broadclassification
schemes: (1) problem scope as a criterion for measuring the breadth and depth of the SCM
problems that theAI study attempted to handle; (2) the methodology as a criterion for evaluating
the theoretical advances in AI studies and the suitability of particular AI sub-fields for SCM
applications; and (3) the implementation status as a criterion for assessing the practicality of
AI technology. These broad classifications will be further subdivided into smaller categories as
discussed below in greater detail.
2.1. Problem scope
The problem scope is categorised with respect to the three-level decision-making hierarchy:
(1) strategic decisions that deal with long-term, executive-level issues such as strategic alliances,
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 15
facility location, and capital investment; (2) tactical decisions that deal with intermediate term,
mid-manager-level issues such as joint demand planning, supplier selection, and inventory plan-
ning;and(3)operationaldecisionsthatdealwithshort-term,routineissuessuchasvehiclerouting,
order picking, and cycle counting.
2.2. Methodology
AI is known for its ability to think like humans, act like humans, think rationally, and act ratio-
nally (Russell and Norvig 1995).Thus, with respect to these distinctive features,AI can be further
classified into a number of sub-fields: (1) artificial neural networks (ANN) and rough set the-
ory (“thinking humanly”); (2) machine learning, expert systems, and GAs (“acting humanly”);
(3) fuzzy logic (“thinking rationally”); and (4) agent-based systems (“acting rationally”). These
sub-fields are discussed below.
2.2.1. Artificial neural networks
The theory of an ANN was predicated on the way that the living organ’s brain cells, namely
neurons, function. Using the interconnected network of computer memories,ANN can learn from
experience, distinguish features, recognise patterns, cluster objects, and process ambiguous or
abstract information. To elaborate, anANN is composed of a number of nodes which correspond
to biological neurons. Those nodes are connected to each other by links. Each link has a numeric
weight assigned toit. The links and their weights are the primary means of the long-term memory
storage. The network processes information in such way that the output of one neuron is an input
to another neuron linked to it. The weights are responsible for the strengthening or weakening
of the information passed via the link. The links are placed and the values of weights are set in
a process called learning. ANN can be taught to respond to various data patterns according to
our wishes or to learn hidden interrelationships among the data. Once the network is initialised,
ANN can be modified to improve its performance using an inductive learning algorithm and be
trained in either supervised or unsupervised environments (McCulloch and Pitts 1943, Russell
and Norvig 1995).
ANN has been proved to be useful for semantic modeling due to its ability to learn to pronounce
English vocabularies. In the logistics field, ANN can be useful for maneuvering autonomous
vehicles using its image processing technique. Indeed, Pomerleau (1993) utilisedANN to steer a
land vehicle along a single lane on a highway by mimicking the performance of a human driver.
Although the application ofANN to auto-piloting land vehicles is still limited to a certain type of
road condition and traffic environment, it showed promise in autonomous vehicle navigation.
ANN has also been applied to a traditional lot-sizing problem (Gaafar and Choueiki 2000).
In a broader context, ANN was successfully utilised to develop hierarchical SC planning that
determined the time/capacity needed for setups, estimated optimal lot-size between successive
SC processes, and linked inventory and scheduling decisions at the lower level to demand and
production planning decisions at the higher level (Rohde 2004). As such, ANN is designed to
reflect the interconnectivity and interdependence of SC planning processes better than traditional
operational research (OR) techniques that were primarily intended for solving less-integrated
sub-problems (e.g., inventory or production or transportation planning) of SC planning.
2.2.2. Rough set theory
Rough set theory was introduced by Pawlak (1982) as a way of synthesising the approx-
imation of concepts from the acquired data using a data table comprising of one or more
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
16 H. Min
classificationattributes.Theseattributesinclude equivalence classes, indiscernibility relations,set
approximation, and rough membership. These classification attributesare necessary to implement
mechanisms similar to those that humans use for object classification and recognition.Also, based
on the description of common features of indiscernible objects, they can be useful for developing
decision rules (Pawlak 1984, 1989, 1997).As such, rough set theory can be employed to classify
decision criteria and then develop decision rules relevant to SCM. For example, Li et al. (2007)
used rough set theory to select the most desirable supplier among a pool of qualified suppliers
with respect to multiple but conflicting supplier selection criteria. Zheng and Lai (2008) also
utilised rough set theory to develop multiple criteria decision rules for measuring dynamic SC
performance.
2.2.3. Machine learning
Machine learning, coined by Samuel (1995), was designed to provide computers with the ability
to learn without being explicitly programmed. In other words, machine learning investigates
ways in which the computer can acquire knowledge directly from data and thus learn to solve
problems(Ratner 2000).Depending onthe methodof learningtasks, machinelearning canbe sub-
classified into a number of categories: (1) concept learning that is designed to correctly recognise
orconstructconceptsrelevanttofuturedecision-making processesfollowingan inductive learning
process; (2) decision tree learning that aims to classify all the objects by testing their values for
certain properties and then constructing a decision tree; (3) perceptron learning that aims to
acquire useful knowledge, reduce the error, and solve decision problems using a single layer
of the network called a “perceptron”; (4) Bayesian learning that trains the computer to learn
representations of probabilisticfunctions; and (5) reinforcement learning that trains the computer
to perform at high levels by giving constant feedback in a form of rewards (see e.g., Luger 2002
for various machine learning processes).
Regardlessof differencesinlearning tasks, machine learningtechniques often attempt tomimic
nature based upon the knowledge and experience that the human race has amassed over the eons
of its existence. Some of the machine learning techniques were motivated by the neurological
studiesofthe humanbrainfunction, somebythe processesdictatinghuman evolution, somebythe
mathematical theory ofhuman knowledge acquisition and reasoning and some by the sociological
theory behind human collaborative behavior. In particular, machine learning can be a useful tool
for understanding the motivation behind collaborative behavior among SC partners for sharing
critical information and improving ways of strengthening the partnership among SC partners
through the organisational learning process. For instance, Carbonneau et al. (2008) recently used
machine learning to forecast the distorted demand information at the end of a SC, namely the
bullwhip effect, if that demand information was not shared among the SC partners due to lack of
collaboration.
2.2.4. Expert systems
Expert systems represent computer programs capable of emulating human cognitive skills such as
problem-solving, visual perception and language understanding, and are capable of performing
reasoningabout a problem domaincomplex enoughfor a considerable amount ofhuman expertise
(Jackson 1999). Expert systems are comprised of four components: (1) knowledge base, (2)
inference engine, (3) justifier/scheduler, and (4) user interface. To elaborate, the knowledge base
is the repository of the rules, facts, and knowledge acquired from the human expert.The inference
engine is a cluster of problem-solving programs (the “brain” of the expert system) that coordinate
the searching, reasoning, and inference based on the rules of the knowledge base. The justifier
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 17
explains how and why an expert arrives at a solution, while the scheduler is set up to coordinate
and control the sequencing rules. The user interface facilitates communication and interaction
between the system and its user through a series of user queries (Awad 1996).
Since an expert system operates at a level and in terms and concepts with which the user can
feel affinity, the expert system is much more easily understood by the practitioner and thus more
applicable to practical SC problems (Basden 1984). In particular, the expert system is known
to increase productivity in managing logistics (Eom and Karathanos 1996). For instance, Allen
(1986) successfully solved multiple echelon inventory control problems in the US Air Force
Logistics Center using more than 400 rules and heuristics created by a number of experts and
verifiedthe effectivenessand efficiencyofthe proposed expertsystem in terms ofimprovedinven-
tory accuracy. Sullivan and Fordyce (1990) also developed a real time, transaction-based expert
system that aimed to schedule, monitor, and control the logistics flowof the IBM’s semiconductor
facilitynear Burlington,Vermont.They reported thatthe use of the expert system increased IBM’s
product output by 35% and saved approximately $10 million of capital expenditure.
The application potential of the expert system is limitless in the SC field, as evidenced by
its successful application to air traffic control, spatial mapping, airline yield management, and
vehicle repair and maintenance scheduling (Findler 1987, Jefferies andYeap 2008). In addition,
an expert system can be more useful than a single more sophisticated forecasting method (or a
mixture of such methods) for demand forecasting at every stage of the SC, in terms of forecasting
accuracy,computationalspeed,userunderstanding,andcosteffectiveness(DeLurgio1998).Other
intriguingapplicationsoftheexpertsystemtotheSCdisciplineinclude:concurrentproductdesign
and planning (Zha et al. 1999); gas pipeline operations (Uraikul et al. 2000); supplier evaluation
(Kwong et al. 2002); evaluation and selection of third-party logistics (3PLs) providers (Yan
et al. 2003); formulation of a logistics strategy (Chow et al. 2005); and production planning and
control (Lawrynowicz 2007).
2.2.5. Genetic algorithms
As a branch of evolution programs, a GA imitates the principles of natural evolution and derives
a set of rules from natural selection processes that create organisms that most fit the surrounding
environment. GAs have often been used to solve combinatorial optimisation problems for which
it is possible to construct a function that can estimate a fitness of a given representative (solution )
to agivenenvironment (problem).The GA encodes possible solutions to the problem in numerical
strings called chromosomes. By iterative application of genetic operators (crossover, mutation,
and selection) to a whole population of such chromosomes, the GA produces solutions that are
not necessarily optimal, but quite satisfactory in terms of the fitness to the optimisation problem.
In general, a GA is referred to as a stochastic AI technique that utilises a solution search
process that mimics natural evolutionaryphenomena: genetic inheritance and Darwinian struggle
for survival (Holland 1975, Michalewicz 1999). The GA typically comprises five components
(Michalewicz 1999, Gen and Cheng 2000):
(1) A genetic representation of potential solutions to the problem.
(2) A way to create a population (an initial set of potential solutions).
(3) An evaluation function measuring the fitness of solutions to see whether they will survive.
(4) Genetic operators that alter the genetic composition of offspring. These operators include
reproduction, crossover, and mutation. Reproduction is a process in which individuals (solu-
tions) are copied through the selection of individualsthat are the most fit. Crossovercombines
the features of two parent chromosomes (potential solutions) to form two similar offspring
by exchanging corresponding attributes of the parents. Mutation randomly alters one or more
features of a selected solution to introduce extra variability.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
18 H. Min
(5) Parameter values that determine population size (how many individuals should be in the
population); crossover rate (the probability that the individual will crossover); and mutation
rate (the probability that a certain gene will mutate).
Although the GA aims to produce global optimal solutions efficiently, its population size cannot
be infinite. Thus, its final solution may be biased due to the finite sampling of potential solu-
tions.Also, the performance of a GA may depend on the specific rates of crossover and mutation.
As a result, a GA may suffer from premature convergence. Furthermore, the size of population
does not usually improve the performance of a GA with respect to the speed of finding solu-
tions (see, e.g., Holland 1975 and Goldberg 1989 for distinctive features of a GA). Nevertheless,
GAs have been applied successfully to a variety of challenging SC network design problems.
These problems include: vehicle routing and scheduling (Malmborg 1996, Potvin et al. 1996,
Chen et al. 1998, Park 2001); minimum spanning tree (Zhou and Gen 1998, 1999); delivery and
pickup (Jung and Haghani 2000); bus network optimisation (Bielli et al. 2002); and location–
allocation problems (Hosage and Goodchild 1986, Jaramillo et al. 2002, Zhou et al. 2002,
2003, Min et al. 2005). In addition, a GA was employed to solve well-known logistics and
purchasing problems involving facility layout (Tam and Chan 1998, Balamurugan et al. 2006);
pallet loading (Fontanili et al. 2000); inventory control (Disney et al. 2000, Haq and Kannan
2006); container loading (Gehring and Bortfeldt 1997, Bortfeldt and Gehring 2001); material
handling (Wu and Appleton 2002), delivery reliability assurance (Antony et al. 2006); freight
consolidation (Min et al. 2006a, b); supplier selection (Rao 2007); and express courier services
(Ko et al. 2007).
2.2.6. Fuzzy logic
Fuzzy logic can be a powerful tool for building knowledge bases for particular domains and
acquiring knowledge from the experts. In a broad sense, fuzzy logic uses expert opinions as an
input to specify “good” and “bad” areas of each variable and then determines the likelihood
of “goodness” and “badness” levels after comparing the input variable with the expert opinion
(Tanaka 1997). Fuzzy logic is an extension of Boolean logic that was designed to conceptualise
partial truth – somewhere between definitely true and definitely false. Typically, fuzzy logic con-
sists of five basic components: (1) linguistic variables, (2) linguistic values, (3) fuzzy sets, (4)
membership functions, and (5) fuzzy IF-THEN rules.Thus, fuzzy logic is in contrast to crisp logic
that is predicated on clear distinction between objects (or values). In other words, fuzzy logic can
handle ambiguity, imprecision, and uncertainty of objects. For instance, fuzzy logic may help us
answer questions of how cold the temperature is, how heavy a person is, or how expensive the
product price is, without setting a clear-cut boundary.
In fuzzy logic, since the membership of an object in a fuzzy set takes a value between 0 and 1,
the transition from membership to non-membership in the set is gradual. This gradual transition
allowsforthemathematical expressionofobjects with varyingconditionsand states. Forexample,
a given temperature (object) can be expressed as cold to the degree of 0.1 and warm to the degree
of 0.8; a car can belong to the set of a cheap car to the degree of 0.2 and an expensive car to
the degree of 0.6. Thus, fuzzy logic can be useful for developing a set of rules for SC decision
environments where subjective performance criteria have to be employed. For instance, in the
airport location decision, a transportation planner may not know exactly how convenient the
location of the airport is to the nearby shippers and carriers. Fuzzy logic has also been applied to
solve the well-known traveling salesman problem (TSP) in a SC setting (Michalewicz and Fogel
2000). Other applications of the fuzzy logic to SCM include: supplier performance evaluation
(Lau et al. 2002); inventory cost control (Wang and Shu 2005); measurement of the bullwhip
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 19
effect (Balan et al. 2007); agro-industry SC planning (Yandra et al. 2007); supplier selection
(Carrera and Mayorga 2008); and order fulfillment (Amer et al. 2008).
2.2.7. Agent-based systems
Anagent-based systemis oneof thedistributedproblem-solvingtechniques thatdividesadecision
problem into sub-problems and solves those sub-problems using independent entities called
agents.Eachagentcanusedifferentmethodology,knowledgeandrecoursestoprocessgiventasks.
According to Reis (1999), an agent refers to an autonomous entity that can take certain actions
to accomplish a set of goals and can compete and cooperate with other agents while pursuing its
individual goals.An agent is characterised by its ability to exploit significant amounts of domain
knowledge,overcomeerroneous input, use symbols and abstraction, learn from the decision envi-
ronment, operate in real time and communicate with others in natural language (Newell 1989).
Exploiting such characteristics, an agent-based system has often been employed to handle various
SC issues including shop floor control (Van Dyke Parunak 1998, Shen et al. 2000, Usher 2003,
Wang and Shen 2003); logistics planning (Satapathy et al. 1998); air traffic control (Iordanova
2003); aggregate demand planning and forecasting (Yu et al. 2002, Liang and Huang 2006); joint
production planning (Lima et al. 2006); newproduct development(Liang and Huang 2002); order
monitoring(Chen andWei2007); business-to-business negotiation(Itoand Saleh2000, Lenar and
Sobecki 2007); bidding evaluation (Kang and Han 2002); outsourcing relationship management
(Logan 2000); customer relationship management (CRM) (Baxter et al. 2003); SC relationship
management (Ghiassi and Spera 2003); SC performance assessment (Gjerdrum et al. 2001);
SC coordination (Swaminathan 1998, Fox et al. 2000, Nissen 2001, Sadeh et al. 2001, Ono
et al. 2003, Chan and Chan 2004, Lou et al. 2004, Xue et al. 2005); SC collaboration under uncer-
tainty (Kwon et al. 2007); information exchange among SC partners (Garcia-Flores et al. 2000,
Turowski 2002); information tracking across the SC (Zimmermann et al. 2001); material han-
dling (Ito and Mousavi Jahan Abadi 2002); retail merchandise purchasing (Park and Park 2003);
e-logistics (Santos et al. 2003); strategic e-procurement (Cheung et al. 2004); e-supply chains
(Singh et al. 2005); traffic incident management (Tarver and Fae 2007); and the procurement of
maintenance, repair, and operating (MRO) supplies (Nissen and Sengupta 2006).
Abyproductofagent-basedmethodologiesthatmaybeusefulfor solvingcomplexSCproblems
is ant colony optimisation. This algorithm mimics the social behavior of ants, who often find the
shortest path to their food sources and nests using their innate capability to follow pheromone
trails (hormone deposited by ants reflecting their collective memory) released by other ants.
To elaborate, an ant colony optimisation algorithm is a meta-heuristic inspired by knowledge-
sharing behaviors of ants in solving combinatorial problems pertaining to different realms. The
ant colony optimisation algorithm is known to stabilise the solution with a reasonable amount of
computationaltime withoutdetriment tothe solutionaccuracy,by exploitingthepositivefeedback
providedby armies of ants workingas multiple agents (Dorigo et al. 1996). Due to its past success
in handling NP-hard problems, the ant colony optimisation algorithm has been employed to solve
TSP, vehiclerouting problems, sequential orderingproblems, and process planselection problems
within the SC framework (Gambardella and Dorigo 2000, Tiwari et al. 2006).
2.3. Implementation status
Since SC managers may be interested in determining the applicability of the proposed AI tech-
nique, we included the third dimension of the taxonomy indicating whether the proposed AI
technique has been applied to the real-world decision environment using actual data, and whether
the AI technique was successfully implemented in the SC setting.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
20 H. Min
3. The synthesis of AI applications in SCM
Despite the long history of AI, the potential of AI as a means of solving complex problems and
searching for information in the SCM area has not been fully exploited in the past. However,
some pioneering efforts have been made to initiateAI applications in the SCM area. In particular,
certain sub-disciplines of AI such as expert systems and GAs have been increasingly utilised
to address SCM issues involving inventory management, purchasing, location planning, freight
consolidation, and routing/scheduling problems. In this section we outline those SCM areas
that have been explored for AI applications, identify specific sub-disciplines of AI that have
been proved to be useful for improving SC decisions, and assess their contributions to the SC
decision-making process.
3.1. Inventory control and planning
Inventory represents idle resources that are required to maintain high levels of customer service
but which incur substantial costs. In fact, the annual cost of holding a single unit of inventory
might range from 15% to 35% of its product value (Timme and Williams-Timme 2003). Thus,
the firm’s success in a competitive market often hinges on its ability to control and plan inventory
at minimum cost, while making inventory constantly available for customers when needed. Such
an ability can be enhanced by the presence of accurate, real-time information about expected
customer demands, the size and type of inventory at hand and the amount of order cycle time to
fulfill the customer order. However, since this kind of information is often difficult to estimate,
predict and obtain, traditional decision rules based on mathematical models such as economic
order quantity cannot reflect the very essence of inventory management. That is to say, a tool
such as an expert system, which can replace the sound judgment and intellect of experienced
inventory managers and deal with the unexpected, is better suited to handling inventory control
andplanningdecisions.Recognisingthispotential,Allen(1986)developedanexpertsystemcalled
the Inventory Management Assistant (IMA) that was designed to aid the US Air Force Logistics
Command in replenishing various types of spare aircraft parts and reducing safety stocks. The
IMA was reported to improve the effectiveness of inventory management by 8–18% by reducing
the inventory errors.
As illustrated above, AI techniques such as expert systems offer a promising new approach to
inventory control and planning problems of great magnitude and complexity due to their powerful
knowledge representation language that is capable of capturing inventory patterns throughout the
entire SC at all levels of detail. The capturing of such dynamic complexity in the inventory
data base enables human experts such as inventory managers to estimate the desirable level of
inventory at eachstocking point without causing a bullwhip effect. For example, an expert system
may be incorporated into the material requirement planning system so that it can store data bases
regarding historic master production schedules, bills of materials, and order patterns and then
develop systematic lot-sizing rules to estimate the optimal level of future orders and the optimal
timing of inventory replenishments. Another intriguing application ofAI techniques to inventory
control and planning includes the recent study of Teodorovic et al. (2002) who developed fuzzy
logic rules to make online, intelligent, airline seat inventory control decisions as to whether to
accept or reject any passenger request for seating arrangements.
3.2. Transportation network design
So far one of the most popular applications of AI techniques to a particular SC area has been
to a class of the transportation network design problems that are intrinsically combinatorial and
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 21
for which global optimal solutions are thus difficult to find. This class of problems include: the
TSP, the vehicle routing and scheduling problem, the minimum spanning tree problem, the freight
consolidation problem, and the intermodal connection problem. Other related problems include:
road network design, gas distribution pipeline network design, parking space utilisation, traffic
assignment, and ramp metering in freeway networks. In particular, due to the combinatorial
nature of these problems, GA turns out to be one of the most popular forms of AI techniques
employed to handle these various aspects of transportation network design problems (Chambers
2001). Another AI technique that has emerged as an increasingly popular meta-heuristic is the
ant colony optimisation algorithm. This algorithm has been applied successfully to handle well-
known network design problems such as the TSP, the vehicle routing problem, and the minimum
spanningtree problem (Dorigo and Gambardella 1997, Bullnheimer et al. 1999, Shyu et al. 2003).
UnliketraditionalORtechniquesorheuristics,bothGAsandantcolonyoptimisationalgorithms
belong to a class of meta-heuristics that are viewed as a general algorithmic framework that can
be applied to a wide set of different combinatorial optimisation problems with relatively few
modifications to make them adapted to a specific transportation network design problem (see
e.g., Glover and Kochenberger 2003 for details of meta-heuristics). Thus, they are more flexible
than the traditional OR techniques and heuristics in accommodating variations in transportation
problem structure. However, it is worth noting that other meta-heuristics such as tabu search,
simulated annealing, scatter search, and iterative local search can be as effective as GAs and ant
colony optimisation for solving a TSP and its variants.
3.3. Purchasing and supply management
A make-or-buy decision is primarily concerned with weighing the options of producing goods or
services internally or purchasing those from the external sources of supply to better utilise the
firm’s given resources (e.g., capacity and personnel) and focus on its core competency. Although
the make-or-buy decision sounds simple and straightforward, it should factor into various “what-
if” scenarios as illustrated below (see e.g., Baily et al. 2005 for issues involving the make-or-buy
decision):
What volume of goods does the company expect to produce?
How much capital investment is needed to produce goods or render services?
How much risk is involved in developing new products or innovating technology to stay
competitive in the market?
Hastheproductthatthecompanyisconsideringmakingreacheditspeakdemandorthematurity
stage of its life cycle?
What business is the company in?
What is the key strength of the company?
Do the company employees have the expertise and skill to produce goods that the customers
desire?
Due to the complexity and dynamics of the above scenarios, the make-or-buy decision calls
for systematic decision-aid tools. Such tools include an expert system. For example, Humphreys
etal. (2002) developed an expertsystem thatcould assist the purchasing manager in evaluatingthe
performanceofprospectivesuppliers,enhancinginformationexchangeamongthepurchasingper-
sonnel and reducing the time to make the make-or-buy decision. To handle a broader spectrum of
purchasing decisions, Kim et al. (2002b) proposed an agent-based purchasing system to automate
the on-line ordering process involved in the acquisition of shoe materials from the global supply
base. Similarly, Cheung et al. (2004) developed a hybrid agent- and knowledge-based system to
evaluate on-line bids and the performance of the bid-winning suppliers in fulfilling orders. More
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
22 H. Min
recently,Nissen and Sengupta(2006) proposed intelligentsoftware agentsthat could automatethe
processesof searchingfor prospectivesuppliers through onlinecatalogs, evaluating supplierswith
respect to multiple attributes, screening qualified suppliers and completing the purchase order.
Preventing specification ambiguity, they discovered that the proposed agent-based purchasing
system can substitute the role of the human decision-maker. As illustrated above, agent-based
systems can aid the purchasing manager in a series of strategic and tactical purchasing decisions,
while traditional OR techniques such as analytic hierarchy process and multiple attribute theory
can handle only one aspect of purchasing decisions (e.g., supplier selection).
3.4. Demand planning and forecasting
Informationabout futuredemand isa basis forthe firm’scapacity planning, workforcescheduling,
inventorycontrol, newproduct development,andpromotional campaigns.However,its usefulness
often depends on its accuracy that, in turn, rests with the firm’s ability to reduce the uncertainty
and variability inherent in future demand. Given the volatile nature of future demand coupled
with the varying degree of uncertainty and variability associated with such demand, it has been a
dauntingtaskto developaccurate forecastingtechniquesand/or selecta forecastingtechnique that
is most suitable for particular business environments. For example, some forecasting techniques
are intended for a short-term projection whereas others work better for a long-term projection.
Regardless, a common denominator among most traditional forecasting techniques such as expo-
nential smoothing, moving average, time series, and Box–Jenkins methods is their underlying
premise that future demand will follow the pattern of past demand. Under such a premise, these
traditional forecasting techniques have relied heavily on the accuracy and validity of historical
data. Although historical data is still invaluable in predicting the future demand of existing prod-
ucts and services, it is not available for the prediction of the future demand of new products and
innovative services that were not extant in the past. To overcome such a drawback of traditional
forecasting techniques, AI techniques have recently been introduced as viable alternatives for
demand forecasting and planning.
Forinstance,Yu et al. (2002) proposed a dynamic pattern matching procedure within the agent-
based system framework that combines human expertise and data mining techniques to predict
the demand for new products. Their experiments indicated that the dynamic pattern matching
procedure outperformed exponential smoothing techniques with respect to forecasting accuracy.
In contrast to exponential smoothing, which merely relies on historical data, the dynamic pattern
matching procedure utilised multiple agents to capture past (base-line agent), current (causal
agent),and future(pattern agent) customerbehaviorsthat helpedimproveits forecastingaccuracy.
Similarly, Jeong et al. (2002) improved forecasting accuracy without relying heavily on historical
data by introducing a genetic algorithm-based causal forecasting technique that outperformed
traditionalregressionanalysis.Asillustratedabove,AItechniquessuchasagent-basedsystemsand
GAs can be useful for predicting future demand for new products or innovativeproducts/services
that have not yet been introduced in the market and thus have no historical demand data.
3.5. Order-picking problems
Put simply, order picking involves selecting the items that have been placed on order. Due to its
labor-intensive operations, order picking typically accounts for the largestportion of warehousing
operating expenditure (Frazelle 2002). Thus, it affects warehousing productivity significantly.
Considering its significant role in warehousing operations,warehousing managers have attempted
to devise ways to improve order-picking efficiency. Such ways include the computerisation and
subsequent automation of sequencing and filling the orders. As part of the automation process,
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 23
Kim et al. (2002a) developed an intelligent agent-based system that optimally assigned workers
to a specified zone from which orders were picked. It was also designed to adjust conveyor speed
dynamically to minimise queuing time for order picking intervals and maximise order picking
throughput. Although the order-picking problem has often been tackled by simulation models
and mathematical models in the past, the use ofAI techniques such as an intelligent agent-based
system may better handle the added complexity caused by the increasing adoption of value-added
services and e-fulfillments due to their inherent learning capability.
3.6. Customer relationship management
To retain customers, the firm should make its customers trust its manufacturing and service
capabilitiesand makecustomersbelieveitcandeliverexactlywhat theywant.Suchtrust cannotbe
instilled without constantly communicating and building a long-termrelationship with customers.
Thus, CRM is an important prerequisite to demand creation that drives SC activities. In general,
CRM is referred to as the business practice that is intended to improve service delivery, build
social bonds with customers and secure customer loyalty by nurturing a long-term, mutually
beneficial relationship with valued customers selected from a pool of more than a few customers
(Min 2006).
Since CRM has a profound impact on the firm’s profitability, it would be necessary for the firm
toassess the costs of sustaining CRM andweigh its benefits against costs. Baxter etal. (2003) pro-
posed anagent-based model that simulated interaction between members of customer populations
and business environments in which they were contained.Their agent-based model considered the
communication of customer experiences between members of a social network and then incor-
porated the powerful influence of word-of-mouth reputation on the purchase of products and
services. By doing so, it aided the firm in assessing the extent of its return on investment in CRM
and enhancing its customer acquisition efforts.
3.7. e-synchronised SCM
To facilitate the coordination and integration of SC activities, SC partners often share information
regardingdemandforecasting,jointproductionanddistributionplanningthroughelectronicmedia
such as Internet websites and electronic data interchange. Abundance of such information in the
cyber space provides a fertile ground for applying machine learning techniques such as web
mining and text mining. Web mining generally refers to the search, classification and analysis
of all web-related data, including web content, hyperlink structure, and web access statistics
(Fayyad et al. 1996). In particular, web mining can be used to extract new patterns or previously
unknown patterns of data regarding customer profiles, supplier profiles, sales trends, sourcing
trends,revenuetrends,anddemandfluctuationsstoredinvariouswebsites.Discoveringknowledge
through web mining can help multinational firms such asAmazon.com and e-Bay identify future
customer bases, develop pricing strategies, evaluate trading partners, and increase revenue. For
example,Symeonidis et al. (2008)utilised data mining techniques to evaluate the performances of
intelligenttradingagentsand thenmaximiserevenuepotential ine-synchronisedSCenvironments
including electronic bidding.
4. Concluding remarks and future research directions
Since SCM requires the comprehension of complex, interrelated decision-making processes and
the creation of intelligent knowledge bases essential for joint problem-solving, SCM has evolved
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
24 H. Min
Table 1. Categorised list of the selectedANN in SCM literature.
Article/year Implementation Key Key Findings
of publication Problem scope Methodology status Contribution(s) Limitation(s) (if any) Comments
Gaafar and Choueiki
(2000)
Lot sizing in
Materials
Requirement
Planning
ANN Not implemented in
practice
Solved the problem
of determining the
optimal lot size to
order in discrete
time periods of a
single item over
multiple time periods
to satisfy a certain
demand pattern while
minimising the sum
of ordering and
inventory carrying
costs. Combined
ANN with simulation.
The study was limited
to deterministic
time-varying demand
patterns.
The proposed ANN was
robust to changing
demand patterns over
time periods.
The proposed ANN
outperformed the
existing lot-sizing
heuristics such as
Silver–Meal and
Periodic Order
Quantity approaches
in terms of finding
the optimal order
quantities.
Rohde (2004) Hierarchical SC
planning
ANN Not implemented in
practice
One of the first to use
an ANN to develop
three-level SC plans
involving demand
planning, purchasing,
transportation, pro-
duction scheduling,
and lot sizing.
Considered a single-
stage flow in
deterministic
environments.
The proposed ANN
outperformed an
economic order
quantity model
in calculating the
effective lot-size.
The generalisation
capability of the ANN
can be improved by
performing network
training demand
series with increased
average demand and
decreased seasonality.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 25
Table 2. Categorised list of the selected expert systems in SCM literature.
Article/year Implementation Key Key Findings
of publication Problem scope Methodology status Contribution(s) Limitation(s) (if any) Comments
Allen (1986) Inventory
management
Expert system Implemented in
Sacramento
and Ogden Air
Logistics Center
One of the first studies
which utilised a
rule-based expert
system to assist the
decisions involving
practical inventory
management.
The proposed nominal
group technique can
create difficulty in
forming consensus
among the experts.
The study was limited
to relatively small
inventory tasks.
The proposed expert
system resulted in a
greater performance
improvement on a
complex case than
on a simple case.
User acceptance of the
expert system was
high.
The proposed expert
system led to 15%
productivity gains.
Yan et al. (2003) Evaluation and
selection of 3PLs
providers
Expert system Not implemented in
practice
Presented a case-based
reasoning approach
to facilitate the
decision regarding
3PL evaluation and
selection.
The proposed system
mainly relied on the
expertise of internal
managers within
the firm rather than
external consultants.
The benefits of the
proposed system
were neither
assessed nor
documented.
The proposed case
reasoning is
based on human
decision-making
behavior.
Cheung et al. (2004) Strategic e-
procurement
Agent-based
system Expert
system
Implemented
by a small
home appliance
manufacturing
company in Hong
Kong
Combined an agent-
oriented approach
with a knowledge-
based system to
issue requests for
quotations, evaluate
on-line bids, and
automate on-line
purchase orders.
The proposed system
was validated only
by trials and runs.
The proposed system
helped to reduce
the time and steps
of searching and
evaluating suppliers.
The proposed system
reduced paperwork
and human errors.
The proposed system
leveraged the
knowledge bases
of experienced
purchasing
professionals
to automate
e-procurement
processes.
Chow et al. 2005) Logistics strategy
formulation under
various customer
demands
Expert system Applied to a Hong
Kong-based
freight forwarder
Proposed an expert
system that
could collect,
transform, and
store organisational
knowledge to
formulate logistics
strategy.
The proposed system
may not be
generalised to
other logistics
environments.
Case-based reasoning
relied heavily on
human judgment
that was subject to
error.
The proposed system
improved both
the inbound and
outbound logistics
efficiency by 15%.
Reduced operating
costs and customer
claims. Reduced
logistics planning
time by 70%.
The proposed system
incorporated case-
based reasoning
into data mining
techniques to
facilitate the strategy
development
process.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
26 H. Min
Table 3. Categorised list of the selected GAs in SCM literature.
Article/year Implementation Key Key Findings
of publication Problem scope Methodology status Contribution(s) Limitation(s) (if any) Comments
Jeong et al. (2002) Demand fore-
casting in the
SC
GA Applied to glass
manufacturing lines
and residential
construction
One of the first studies
which used GA
to determine the
coefficients of the
linear forecasting
model.
The forecasting
accuracy was
measured only by
the mean absolute
deviation.
The forecasting model
is predicated on
the premise of a
causal relationship
between customer
demand and price
index.
The proposed GA
(i.e., guided GA)
outperformed
regression analysis
with respect to
its forecasting
accuracy.
The length of the
GA training
period can affect
the forecasting
accuracy of the
guided GA.
Disney et al. (2000) Inventory control GA Never implemented in
practice
Combined theclassical
inventory control
theory with GA to
optimally make a
tradeoff between
inventory levels and
factory orders.
The study was limited
to a simplified linear
SC.
If inventory costs are
significant, work-in-
process information
significantly affects
the performance of
the ordering system.
The hybrid GA and
simulation model
can substantially
improve an
inventory control
system in the SC.
Park (2001) Vehicle scheduling GA Not implemented in
practice
Developed GA to
solve multiple
objective vehicle
scheduling problem
with service due
times and deadlines.
The solution was
confined to a single
depot problem.
The proposed GA
outperformed a
well-known savings
method for solving
some test problems.
The GA combined
with a greedy
interchange
heuristics improved
its computational
efficiency.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 27
Zhou et al. (2002) Allocation of
customers to
warehouses
GA Applied to the
distribution problem
facingan actual yard
fence component
supplier
Formulated a naïve
balanced star
spanning forest and
developed GA to
solve the warehouse
allocation problem.
The proposed solution
was confined to
a single objective
problem with equal
warehouse capacity.
The proposed GA
allocated customers
to multiple
warehouses with
more than 98%
balanced level.
The balanced
warehouse
allocation can
reduce stock-outs
and late deliveries.
Bielli et al. (2002) Urban bus
transportation
network
GA Applied to a small city
bus network in Italy
Proposed both GA and
neural network to
optimally design
the urban bus
transportation
network. Multiple
criteria were
considered.
Sensitivity analysis
was not conducted.
The best solution
found the GA was a
bus network formed
by a set of lines
corresponding to
the genes having the
switch set on.
The proposed
algorithm can
reduce the number
of buses needed.
Liang and Huang
(2006)
Demand
forecasting
Agent-based
system Rough
set theory GA
Applied to the
hypothetical Beer
Game
Applied agent
technology to
simulate inventory
levels throughout
the SC and
determine optimal
order quantity for
every echelon of the
SC.
Used the prior expert
knowledge to make
a demand better
forecast.
Experiments with the
system were based
on a small number
of MBA student
subjects.
Assumed that SC
partners were
always willing to
share information
among themselves.
The proposed agent-
based system
mitigated the
bullwhip effect.
The proposed agent-
based system is less
costly to make a
demand forecast
than conventional
forecasting
techniques such
as exponential
smoothing.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
28 H. Min
Table 4. Categorised list of the selected fuzzy logic in SCM literature.
Article/year Implementation Key Key Findings
of publication Problem scope Methodology status Contribution(s) Limitation(s) (if any) Comments
Yandra et al. (2007) Bio-diesel SC
planning
Fuzzy logic and
GAs
Applied to the
hypothetical
agricultural SC
problem
Solved the problem
of determining the
optimal lot size to
order in discrete
time periods of a
single item over
multiple time
periods to satisfy
a certain demand
pattern while
minimising the sum
of ordering and
inventory carrying
costs.
Combined GAs with
fuzzy logic.
The study was limited
to a flow of a single
product made from
coconut or palm oil.
The proposed GA and
fuzzy logic turned
out to be robust and
reliable and could
produce promising
results.
First, a GA was
employed to create
a Pareto front that
is comprised of
non-dominated
solutions. Then a
fuzzy logicwas used
to select the most
preferred solution.
Amer et al. (2008) Order fulfillment
Supplier
performance
monitoring
Fuzzy set theory Not implemented
in practice
Presented the design
for six sigma to
monitor and control
the SC process of
order fulfillment.
Developed a transfer
function that
mathematically
represented a
relationship
between input
and output
order fulfillment
processes.
Only considered
three performance
metrics: delivery
time, order quantity,
and order quality
to evaluate the
order fulfillment
process. Thus,
other intangible
service performance
criteria such as
delivery reliability
were not taken into
consideration.
The maximum score
for the order
occurs when both
the quantity and
delivery time are at
target.
Fuzzy logic was
used to develop a
transfer function
and provide a
means of putting
quantitative value
on qualitative (or
vague) performance
measurement.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 29
Table 5. Categorised list of the selected agent-based systems in SCM literature.
Article/year Implementation Key Key Findings
of publication Problem scope Methodology status Contribution(s) Limitation(s) (if any) Comments
Swaminathan et al.
(1998)
SC reengi-
neering and
coordination
Agent-based
system
Applied to IBM for
inventory control
Developed a library
of software for
modeling SC
processes.
Combined simulation
with a multi-agent
paradigmto address
SC coordination
issues under
uncertainty.
The study focussed on
generic outbound
logistics activities.
The system is more
appropriate for large
corporations.
More adaptive agents
are needed.
The proposed system
was useful for
evaluating both
short- and long-term
SC reengineering
efforts.
The multi-agent-based
system reduced the
development time
for SC models.
Logan (2000) Transportation
outsourcing
Agency theory Never implemented
in practice
Engaged agency
theory to help
design the types
of transportation
outsourcing
contracts and
relationships.
Agency theory was
proposed as a viable
alternative to the
resource-based view
of the firm and
transaction cost
economics.
The proposed agency
concept was not
fully computerised.
Prescriptions
for success in
transportation
outsourcing (such
as communica-
tion, training,
performance
measurement, and
motivation for con-
tinual improvement)
are similar to those
in other industries.
Agency theory can be
used to determine
which outsourcing
service providers
and user are best
suited to manage
the conflicting
relationship.
(Continued)
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
30 H. Min
Table 5. Continued
Article/year Implementation Key Key Findings
of publication Problem scope Methodology status Contribution(s) Limitation(s) (if any) Comments
Fox et al. (2000) Inter-
organisational
SC integration
Agent-based
system
Implemented
in Perfect
Minicomputer
Corporation in
Toronto, Canada
One of the first
studies to use
multiple agents
to integrate and
coordinate various
SC activities. The
proposed system is
capable of handling
unexpected SC
disruptions.
Since the simulation
experiments
were conducted
in simplified
and controlled
environments,
the proposed
system needs to be
expanded to validate
the usefulness of the
agent-based system.
The simple notification
of the delivery
plans to the
downstream SC
partner substantially
reduces the
inventory level.
A good basis for future
case studies.
Nissen (2001) Inter-
organisational
SC inte-
gration and
supplier–buyer
relationships
Agent-based
system
Not implemented in
practice
Developed Petri
Net-based (so-
called Grafcet)
intelligent agents
that facilitated
SC mapping and
supplier–buyer
collaboration.
Only the preliminary
results were
presented.
The performance
metrics of the
intelligent agents
were not clearly
defined.
The agent-based SC
integration led to
SC cost savings
and cycle time
reduction.
Intelligent agents are
superior to web-
based technology
for SC integration.
Ito and Mousavi
Jahan Abadi
(2002)
Material handling
Inventory
planning and
control
Agent-based
system
Not implemented in
practice
Developed agent-
based warehouse
systems comprised
of three sub-
systems: (1)
communication, (2)
material handling,
and (3) inventory
planning andcontrol
systems to enhance
warehousing
efficiency.
The proposed system
that containsmanual
components was
tested only based on
limited simulation
experiments.
The efficient use of the
automated guided
vehicles improved
the delivery cycles.
The agent-based
system automates
warehousing
operations.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 31
Kim et al. (2002a) Order picking Agent-based
system
Applied to unknown
actual ware-
housing
operations
Considered order-
picking problems
with multiple order
pickers in multiple
storage areas.
Considered both
hierarchical and
heterarchical
frameworks.
Assumed a fixed
conveyor speed.
The proposed system
reduced picking
errors.
Theproposed systemis
robust when facing
with unforeseen
events such as
machine failures.
Kim et al. (2002b) Global sourcing Agent-based
system
Implemented by a
shoe manufacturer
One of the first to
develop a multi-
agent-based system
to automate the
process of ordering
and procuring goods
via online.
The actual benefits
of the proposed
system were neither
measured nor
documented.
The proposed agent-
based purchasing
system facili-
tates information
exchange between
buyer and sup-
pliers in global
environments.
The agent-based
system was
incorporatedintothe
cooperative (joint)
problem-solving
mechanism.
Iordanova (2003) Air traffic control Agent-based
system
Not implemented in
practice
An agent-based
architecture was
developed to ensure
efficient flights and
utilise air-space
time in the wake
of the increased air
traffic.
The proposed system
was not fully tested
in a real-world
setting.
The proposed system
was expected to
reduce cancelled
and delayed flights.
The agent-based
system can be
embedded within
the knowledge
management system
dealing with air
traffic control and
planning.
Santos et al. (2003) Intra-
organisational
logistics
management
Agent-based
system
Never implemented
in practice
Developed a multi-
agent-based system
that was useful
for coordinating
logistics planning
and scheduling as
well as allocating
logistics resources
efficiently.
Combined
Lagrangian
relaxation heuristics
with theagent-based
system.
The proposed system
is intended to
solve relatively
small intra-logistics
problems. The
duality gap
produced by the
proposed solution
method was often
large.
As the problem
complexity
increased, the
proposed heuristics
did not perform
well.
The proposed system
can be extended to
solve air mission
planning.
(Continued)
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
32 H. Min
Table 5. Continued
Article/year Implementation Key Key Findings
of publication Problem scope Methodology status Contribution(s) Limitation(s) (if any) Comments
Shyu et al. (2003) Minimum
spanning tree
Ant colony
optimisation
(agent-based
system)
Theory development
without actual
applications
Developed an ant
colony optimisation
technique to solve
the generalised
minimum spanning
tree problem that
is known to be NP
hard.
The computational
experiments with
the proposed
algorithm were
confined to the
pre-tests of the
hypothetical data.
Although the
proposed ant
colony optimisation
technique was not
necessarily more
accurate than GA, it
solved the problem
faster than GA.
The proposed ant
colony optimisation
technique is
capable of solving
large-sized
problems.
Chan and Chan
(2004)
SC integration and
coordination
Agent-based
system
The proposed
system was tested
with simulated
hypothetical
examples.
Proposed the dis-
tributed modeling
philosophy within
the multi-agent-
based system that
was designed to
coordinate the
manufacturing SC.
The simulated
examples were
confined to simple
and deterministic
cases.
The system per-
formance was
measured mainly
using the order fill
rate.
With the presence of
uncertainty (or high
system variance),
the proposed agent-
based system can
be degraded. The
decentralised multi-
agent-based system
(MAS) generally
outperformed the
centralised MAS.
To make the proposed
MAS work, trust
among the SC
partners must be
built.
Cheung et al.
(2004)
Strategic
e-procurement
Agent-based
system Expert
system
Implemented
by a small
home appliance
manufacturing
company in Hong
Kong
Combined an agent-
oriented approach
with a knowledge-
based system to
issue requests for
quotations, evaluate
on-line bids and
automate on-line
purchase orders.
The proposed system
was validated only
by trials and runs.
The proposed system
helped to reduce
the time and steps
of searching and
evaluating suppliers.
The proposed system
reduced paperwork
and human errors.
The proposed system
leveraged the
knowledge bases
of experienced
purchasing
professionals
to automate
e-procurement
processes.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 33
Nissen and
Sengupta (2006)
Procurement of
MRO supplies
Agent-based
system
Applied to the
hypothetical
Intelligent Mall
Compared the
performance of
human to those of
software agents
across varying
levels of ambiguity
associated with
the procurement
process.
Experiments with the
system were based
on a limited number
of student subjects.
An integration of the
SC ontology and
a specification of
product knowledge
improved the
sophistication of
agents.
The proposed
intelligent software
agents can improve
the performance of
the procurement
task.
Liang and Huang
(2006)
Demand
forecasting
Agent-based
system Rough
set theory GA
Applied to the
hypothetical Beer
Game
Applied agent
technology to
simulate inventory
level throughout the
SC and determine
optimal order
quantity for every
echelon of the SC.
Used prior expert
knowledge to make
a demand better
forecast.
Experiments with the
system were based
on a small number
of MBA student
subjects. Assumed
that SC partners
were always willing
to share information
among themselves.
The proposed agent-
based system
mitigated the
bullwhip effect.
The proposed agent-
based system is
less costly for
demand forecasting
than conventional
forecasting
techniques such
as exponential
smoothing.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
34 H. Min
Figure 1. Link between popular AI tools and their SCM applications areas.
into knowledge management. In other words, it is increasingly important for SC partners to learn
fromtheincreasedknowledgebasesandautomatetheSCdecision-makingprocesses.Thus,AIhas
been put forward as a useful decision-aid tool that helps the firm connect its customers, suppliers,
and SC partners by facilitating information exchange among various business entities across the
SC, while replacing assets (e.g., inventory, facilities, transportation equipment) with information.
DespitethepresenceofAIforthelasthalf-centuryanditsrecentemergenceintheSCMarea,AIhas
not been fully exploited to solve SC problems whose solutions are either too expensive or difficult
to produce due to their inherent complexity and ill-structured nature. Indeed, we have discovered
a pattern that most AI applications in the SCM area remain limited to relatively well structured
(welldefined),tacticalandoperationalSCproblemsasrecapitulatedinTables1–5.However,some
recent AI studies have shown the great potential of AI tools (especially agent-based systems) for
addressingavarietyofsoftbutstrategicissuesinvolvingCRM,outsourcingrelationships,strategic
alliances among SC partners, SC coordination, collaborative demand planning, and business-to-
business negotiations that have often been overlooked by more traditional analytical models (Min
and Zhou 2002). Another finding is that an agent-based system has emerged as one of the most
popular AI tools for tackling various aspects of SC problems (Figure 1). One of the reasons for
the paucity of AI applications in the SCM area may be the relative youth and broad spectrum of
the SCM discipline. Other challenges for AI applications to SCM include:
AI does not have free will and thus relies heavily on the computer software, which may lead to
wrong decisions, if it is programmed incorrectly;
AI solutionsmay notbe easyto implement becausetheyare soesoteric and difficult forordinary
decision-makers to comprehend;
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 35
AlthoughAI usually works best for specific, narrowly focussed SC problems,AI may not work
wellforhandlingriskanduncertaintyinvolvedincross-functionalandcross-borderSCdecision
environments due to its knowledge acquisition bottlenecks.
Despite these challenges, as SCM continues to draw more attention from both practitioners
and academicians alike and begins to mature as an academic discipline,AI will have a promising
future in the SCM area. Based on projectedAI research trends, we suggest the following selected
line of AI research topic areas that can advance the SCM decision-making processes.
Multiple agent-based systems that can manage complexity better can be applied to a new set
of strategic SC problems such as SC integration and SC risk (disaster) management.
Intelligent agents can be utilised for real-time pricing and reverse auctioning involving SC
partners.
Game theory can be incorporated into agent-based systems to understand SC dynamics and
form strategic SC partnerships.
Profiles of desirable SC partners, including suppliers and 3PLs providers, can be developed
using knowledge discovery techniques.
Rule-based expert systems can be developed to assist in logistics outsourcing or contract
manufacturing decisions.
Expert systems that improve airline revenue management can be developed.
Hybrid meta-heuristics can be developed to integrate the AI traits of GAs with those of ant
colony optimisation, to solve combinatorial transportation network design problems.
The fuzzy logic approach can be integrated with the GA or ANN approaches to control total
logistics costs.
Arule-basedexpertsystemcanbeincorporatedintotheSCknowledgemanagementframework.
Rough set theory and/or machine learning can be further explored to address the evaluation
and selection issues of foreign suppliers or 3PLs.
AI can be integrated with existing legacy systems of various SC partners without disrupting
information flows across the SC.
References
Allen, M.K., 1986. The development of an artificial intelligence system for inventory management using multiple experts.
Unpublished PhD dissertation. Columbus, Ohio: The Ohio State University.
Allen, M.K. and Helferich, O.K., 1990. Putting expert systems to work in logistics. Oak Brook, IL: Council of Logistics
Management.
Amer, Y., et al., 2008. Optimizing order fulfillment using design for six sigma and fuzzy logic. International Journal of
Management Science and Engineering Management, 3 (2), 83–99.
Antony, J., et al., 2006. Design of synchronized supply chain: a genetic algorithm based six sigma constrained approach.
International Journal of Systems and Management, 2 (2), 120–141.
Awad, E.M., 1996. Building expert systems: principles, procedures, and applications. St. Paul, MN: West Publishing
Company.
Baily, P., et al., 2005. Purchasing principles and management. 9th ed. Essex, England: Pearson Education Limited.
Balamurugan, K., Selladurai, V., and Ilamathi, B., 2006. Solving unequal area facility layout problems using genetic
algorithm. International Journal of Logistics Systems and Management, 2 (3), 281–301.
Balan, S., Vrat, P., and Kumar, P., 2007. Reducing the Bullwhip effect in a supply chain with fuzzy logic approach.
International Journal of Integrated Supply Management, 3 (3), 261–282.
Basden, A., 1984. Developments in expert systems. London, England: Academic Press.
Baxter, N., Collings, D., and Adjali, I., 2003.Agent-based modeling – intelligent customer relationship management. BT
Technology Journal, 21 (2), 126–132.
Bielli, M., Caramia, M., and Carotenuto, P., 2002. Genetic algorithms in bus network optimization. Transportation
Research Part C, 10, 19–34.
Bortfeldt, A. and Gehring, H., 2001. A hybrid genetic algorithm for the container loading problem. European Journal of
Operational Research, 131, 143–161.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
36 H. Min
Bullnheimer, B., Hartl, R.F., and Strauss, C., 1999. Apply the ant system to the vehicle routing problem. In: S. Voss, ed.
Meta-heuristics: advances and trends in local search paradigms for optimization. Norwell, MA: Kluwer Academic
Publishers.
Carbonneau, R., Laframboise, K., and Vahidov, R., 2008. Application of machine learning techniques for supply chain
demand forecasting. European Journal of Operational Research, 184 (3), 1140–1154.
Carrera,D.A. andMayorga,R.V.,2008. Supplychainmanagement: amodular fuzzyinference systemapproachin supplier
selection for new product development. Journal of Intelligent Manufacturing, 19 (1), 1–12.
Chambers, L., 2001. The practical handbook of genetic algorithms: applications. 2nd ed. Boca Raton, FL: Chapman &
Hall/CRC.
Chan, F.T.S. and Chan, H.K., 2004. A new model for manufacturing supply chain networks: a multiagent approach.
Proceedings of Institution of Mechanical Engineers 218, 443–454.
Chen,Y.M. and Wei, C., 2007. Multi-agent-oriented approach to supply chain planning and scheduling in make-to-order
manufacturing. International Journal of Electronic Business, 5 (4), 427–454.
Chen, X., Wan, W., and Xu, X., 1998. Modeling rolling batch planning as a vehicle routing problem with time windows.
Computers and Operations Research, 25 (12), 1127–1136.
Cheung, C.F., et al., 2004. An agent-oriented and knowledge-based system for strategic e-procurement. Expert Systems,
21 (1), 11–21.
Chow, H.K.H., et al., 2005. Design a case-based logistics strategy system – an integrated approach. Expert Systems,22
(4), 173–192.
DeLurgio, S.L., 1998. Forecasting principles and applications. Boston, MA: Irwin/McGraw-Hill.
Disney,S.M.,Naim,M.M.,andTowill,D.R.,2000.Geneticalgorithmoptimizationofaclassofinventorycontrolproblems.
International Journal of Production Economics, 68, 259–278.
Dorigo, M. and Gambardella, L., 1997. Ant colony system: a cooperative learning approach to the traveling salesman
problem. IEEE Transactions on Evolutionary Computation, 1, 53–66.
Dorigo, M., Maniezzo,V., and Colorni,A., 1996. The ant systems: optimization by a colony of cooperative agents. IEEE
Transactions in Man, Machine and Cybernetics Part B, 26 (1), 29–41.
Eom, S.B.and Karathanos, D., 1996.The role of expertsystems in improvingthe management of processes intotal quality
management organizations. SAM Advanced Management Journal, 61 (4), 8–15.
Fayyad, U., Piatetsky-Shapiro, G., and Smyth, P., 1996. From data mining to knowledge discovery: an overview. In:
Advances in knowledge discovery and data mining. Cambridge, MA: MIT Press, 1–38.
Findler, N.V., 1987. Air traffic control: a challenge for AI. AI Expert, 2 (1), 59–66.
Fontanili, F., Vincent, A., and Ponsonnet, R., 2000. Flow simulation and genetic algorithm as optimization tools.
International Journal of Production Economics, 64, 91–100.
Fox, M.S., Barbiceanu, M., and Teigen, R., 2000. Agent-oriented supply chain management. International Journal of
Flexible Manufacturing Systems, 12, 165–188.
Frazelle, E.H., 2002. World-class warehousing and material handling. NewYork, NY: McGraw-Hill.
Gaafar, L.K. and Choueiki, M.H., 2000. A neural network for solving the lot-sizing problem. Omega, 28, 175–184.
Gambardella,L.M. andDorigo,M., 2000.Antcolonysystem hybridizedwitha newlocal searchfor thesequentialordering
problem. INFORMS Journal on Computing, 12 (3), 237–255.
Garcia-Flores, R., Wang, X.Z., and Goltz, G.E., 2000. Agent-based information flow for process industries’supply chain
modeling. Computers and Chemical Engineering, 24, 1135–1141.
Gehring, H. and Bortfeldt, A., 1997. A genetic algorithm for solving the container loading problem. International
Transactions in Operational Research,4(5/6), 401–418.
Gen, M. and Cheng, R., 2000. Genetic algorithm and engineering optimization
. NewYork: John Wiley and Sons.
Ghiassi,M. andSpera,C., 2003.Definingthe Internet-basedsupplychain systemformass customizedmarkets.Computers
and Industrial Engineering, 45, 17–41.
Gjerdrum, J., Shah, N., and Papageorgiou, L.G., 2001. A combined optimization and agent-based approach to supply
chain modeling and performance assessment. Production Planning and Control, 12 (1), 81–88.
Glover, F. and Kochenberger, G.A., 2003. Handbook of metaheuristics. Berlin, Germany: Springer.
Goldberg, D.E., 1989. Genetic algorithms in search, optimization and machine learning. Boston, MA: Addison-Wesley.
Haq, A.N. and Kannan, G., 2006. Two-echelon distribution-inventory supply chain model for the bread industry using
genetic algorithm. International Journal of Logistics Systems and Management, 2 (2), 177–193.
Holland, J., 1975. Adaptation in natural and artificial systems. AnnArbor, MI: University of Michigan Press.
Hosage, C.M. and Goodchild, M.F., 1986. Discrete space location-allocation solutions from genetic algorithms. Annals
of Operations Research, 6 (2), 35–46.
Humphreys, P., Huang, G., and McIvor, R., 2002. An expert system for evaluating the make or buy decision. Computers
and Industrial Engineering,42(2/4), 567–585.
Iordanova, B.N., 2003. Air traffic knowledge management policy. European Journal of Operational Research, 146,
83–100.
Ito, T. and Mousavi Jahan Abadi, S.M., 2002. Agent-based material handling and inventory planning in a warehouse.
Journal of Intelligent Manufacturing, 13, 201–210.
Ito, T. and Saleh, M.R., 2000.A blackboard-based negotiationfor collaborative supply chain systems.Journal of Material
Processing Technology, 107, 398–403.
Jackson, P., 1999. Introduction to expert systems. 3rd ed. Essex, England: Addison-Wesley.
Jaramillo, J.H., Bhadury, J., and Batta, R., 2002. On the use of genetic algorithms to solve location problems. Computers
and Operations Research, 29, 761–779.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 37
Jefferies, M.E. andYeap, W.K., 2008. Robots and cognitive approaches to spatial mapping. Berlin, Germany: Springer.
Jeong, B., Jung, H., and Park, N., 2002. A computerized forecasting system using genetic algorithm in supply chain
management. The Journal of Systems and Software, 60 (3), 223–237.
Jung, S. and Haghani, A., 2000. A genetic algorithm for pick-up and delivery problem with time windows. Journal of
Transportation Research Board, 1733, 1–7.
Kang, N. and Han, S., 2002. Agent-based e-marketplace system for more fair and efficient transaction. Decision Support
Systems, 34, 157–165.
Kim, B.I., et al., 2002a. Intelligent agent modeling of an industrial warehousing problem. IIE Transactions, 34 (7),
601–612.
Kim, T.W., Ko, C.S., and Kim, B.N., 2002b. An agent-based framework for global purchasing and manufacturing in a
shoe industry. Computers and Industrial Engineering,42(2/4), 495–506.
Ko, C.S., Min, H., and Ko, H.J., 2007. The determination of cutoff time for express courier services: a genetic algorithm
approach. International Transactions in Operational Research, 14 (2), 159–177.
Kwon, O., Im, G.P., and Lee, K.C., 2007. MACE-SCM: a multiagent and case-based reasoning collaboration mechanism
forsupplychainmanagementundersupplyanddemanduncertainties.ExpertSystemswithApplications,33,690–705.
Kwong,C.K.,Ip,W.H.,and Chan,J.W.K., 2002.Combining scoringmethodand fuzzyexpertsystemsapproach tosupplier
assessment: a case study. Integrated Manufacturing Systems, 13 (7), 512–519.
Lau, H.C.W., Pang,W.K., and Wong, C.W.Y., 2002. Methodology for monitoring supply chain performance: a fuzzy logic
approach. Logistics Information Systems, 15 (4), 271–280.
Lawrynowicz, A., 2007. Production planning and control with outsourcing using artificial intelligence. International
Journal of Services and Operations Management, 3 (2), 193–209.
Lenar, M. and Sobecki, J., 2007. Using recommendation to improve negotiations in agent-based systems. Journal of
Universal Computer Science, 13 (2), 267–286.
Li, G.D., Yamaguchi, D., and Nagai, M., 2007. Application of grey-based rough decision-making approach to supplier
selection. Journal of Modelling in Management, 2 (2), 131–142.
Liang,W. and Huang, C., 2002. The agent-based collaboration information system of product development. International
Journal of Information Management, 22, 211–224.
Liang, W. and Huang, C., 2006.Agent-based demand forecast in multi-echelon supply chain. Decision Support Systems,
42, 390–407.
Lima, R.M., Sousa, R.M., and Martins, P.J., 2006. Distributed production planning and control agent-based system.
International Journal of Production Research,44(18/19), 3693–3709.
Logan, M.S., 2000. Using agent theory to design successful outsourcing relationships. The International Journal of
Logistics Management, 11 (2), 21–32.
Lou, P., et al., 2004. Study on multi-agent agile supply chain management. International Journal of Advanced
Manufacturing Technology, 23, 197–203.
Luger, G.F., 2002. Artificial intelligence: structures and strategies for complex problem solving. 4th ed. Essex, England:
Pearson Education Limited.
Malmborg, C., 1996. A genetic algorithm for service level based vehicle scheduling. European Journal of Operational
Research, 93, 121–134.
McCulloch,W.S. and Pitts,W., 1943.A logicalcalculus of the idea imminent in nervousactivity. Bulletin of Mathematical
Biophysics, 5, 115–137.
Michalewicz, Z., 1999. Genetic algorithms + data structures = evolution programs. Berlin: Springer.
Michalewicz, Z. and Fogel, D.B., 2000. How to solve it: modern heuristics. Heidelberg, Germany: Springer-Verlag.
Min, H., 2006. Developing the profiles of supermarket customers through data mining. The Service Industries Journal,
26 (7), 1–17.
Min, H. and Yu, W., 2008. Collaborative planning, forecasting and replenishment: demand planning in supply chain
management. International Journal of Information Technology and Management, 7 (1), 4–20.
Min, H. and Zhou, G., 2002. Supply chain modeling: past, present and future. Computers and Industrial Engineering,
43 (1), 231–249.
Min, H., et al., 2005. A genetic algorithm approach to the balanced allocation of customers to multiple warehouses with
varying capacities. International Journal of Logistics: Research and Applications, 8 (3), 181–192.
Min, H., Ko, H.J., and Ko, C.S., 2006a. A genetic algorithm approach to developing the multi-echelon reverse logistics
network for product returns. Omega, 34 (1), 56–69.
Min, H., Ko, C.S., and Ko, H.J., 2006b. The spatial and temporal consolidation of returned products in a closed-loop
supply chain network. Computers and Industrial Engineering, 51 (2), 309–320.
Newell,A., 1989. Putting it all together. In: D. Klahr and K. Kotovsky, eds. Complex information processing: the impact
of Herbert A. Simon. Hillsdale, NJ: Lawrence ErlbaumAssociates.
Nilsson, N.J., 1980. Principles of artificial intelligence. Los Altos, CA: Morgan Kaufmann Publishers, Inc.
Nissen, M.E. 2001. Agent-based supply chain integration. Journal of Information Technology and Management, 2 (3),
289–312.
Nissen, M.E. and Sengupta, K., 2006. Incorporating software agents into supply chains: experimental investigation with
a procurement task. MIS Quarterly, 30 (1), 145–166.
Ono, C., et al., 2003. Trust-based facilitator: handling word-of-mouth trust for agent-based e-commerce. Electronic
Commerce Research, 3, 201–220.
Park, B.-Y., 2001. A hybrid genetic algorithm for the vehicle scheduling problem with due times and time deadlines.
International Journal of Production Economics, 73 (2), 175–188.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
38 H. Min
Park,J. and Park,S., 2003.Agent-based merchandisemanagement in business-to-business electronic commerce.Decision
Support Systems, 35, 311–333.
Pawlak, Z., 1982. Rough sets. International Journal of Computer and Information Science, 11, 341–356.
Pawlak, Z., 1984. Rough classification. International Journal of Man-Machine Studies, 20, 469–483.
Pawlak,Z., 1989.Knowledge, reasoning and classification a roughset perspective. Bulletin of the EuropeanAssociation
for Theoretical Computer Science, 38, 199–210.
Pawlak, Z., 1997. Rough set approach to knowledge-based decision support. European Journal of Operational Research,
293 (3), 1–10.
Pomerleau, D.A., 1993. Neural network perception for mobile robot guidance. Dordrecht, The Netherlands: Kluwer.
Potvin, J., Dube, D., and Robillard, C., 1996. A hybrid approach to vehicle routing using neural networks and genetic
algorithms. Applied Intelligence, 6, 241–252.
Rao, R.V., 2007. Vendor selection in a supply chain using analytic hierarchy process and genetic algorithm methods.
International Journal of Services and Operations Management, 3 (3), 355–369.
Ratner, B., 2000. A comparison of two popular machine learning methods. DM STAT-1 Online Newsletter about
Quantitative Methods in Direct Marketing,4.
Reis, B.Y., 1999. A multi-agent system for on-line modeling, parsing and prediction of discrete time series data.
In: M. Mohammadian, ed. Computational intelligence for modeling, control and automation. Amsterdam,
The Netherlands: IOS Press.
Rohde, J., 2004. Hierarchical supply chain planning using artificial neural networks to anticipate base-level outcomes.
OR Spectrum, 26, 471–492.
Russell, S. and Norvig, P., 1995. Artificial intelligence: a modern approach. Upper Saddle River, NJ: Prentice-Hall.
Sadeh, N.M., et al., 2001. MASCOT: an agent-based architecture for dynamic supply chain creation and coordination in
the Internet economy. Production Planning and Control, 12 (3), 212–223.
Samuel, A.L., 1995. Some studies in machine learning using the game of checkers. In: G. F. Luger, ed. Computation and
intelligence: collected readings. Menlo Park, CA:AAAI Press/The MIT Press.
Santos, Jr., E., Zhang, F., and Li, P.B., 2003. Intra-organizational logistics management through multi-agent systems.
Electronic Commerce Research, 3, 337–364.
Satapathy, G., Kumara, S.R.T., and Moore, L.M., 1998. Distributed intelligent architecture for logistics (DIAL). Expert
Systems with Applications, 14, 409–424.
Shen, W., Maturana, F., and Norrie, D.H., 2000. Enhancing the performance of an agent-based manufacturing system
through learning and forecasting. Journal of Intelligent Manufacturing, 11, 365–380.
Shyu, S.J., et al., 2003.Ant-tree: an ant colony optimization approach to the generalizedminimum spanningtree problem.
Journal of Experimental and Theoretical Artificial Intelligence, 15, 103–112.
Singh, R., Salam, A.F., and Iyer, L., 2005. Agents in e-supply chains. Communication of the ACM, 48 (6), 109–115.
Sullivan, G. and Fordyce, K., 1990. IBM Burlington’s logistics management system. Interfaces, 20 (1), 43–64.
Swaminathan, J.M., 1998. Modeling supply chain dynamics: a multiagent approach. Decision Sciences, 29 (3), 607–632.
Symeonidis,A.L.,Nikolaidou,V.,andMitkas, P.A.,2008. Sketchingamethodologyfor efficientsupply chainmanagement
agents enhanced through data mining. International Journal of Intelligent Information and Database Systems, 2 (1),
49–68.
Tam, K.Y. and Chan, S.K., 1998. Solving facility layout problems with geometric constraints using parallel genetic
algorithms: experimentation and findings. International Journal of Production Research, 36 (12), 3253–3272.
Tanaka, K., 1997. An introduction to fuzzy logic for practical applications. NewYork, NY: Springer-Verlag.
Tarver, M. and Fae, M.I., 2007. Applications of multiagents in transport and logistics. International Journal of Logistics
Systems and Management
, 3 (4), 405–418.
Teodorovic, D., et al., 2002. Intelligent airline seat inventory control system. Transportation Planning and Technology,
25 (3), 155–173.
Timme, S.G. and Williams-Timme, C., 2003. The real costs of holding inventory. Supply Chain Management Review,7,
30–37.
Tiwari, M.K., et al., 2006. Ant colony optimization to select the best process plan in an automated manufacturing envi-
ronment. Proceedings of Institute of Mechanical Engineers Part B: Journal of Engineering Manufacture, 220,
1457–1472.
Turowski, K., 2002. Agent-based e-commerce in case of mass customization. International Journal of Production
Economics, 75, 69–81.
Uraikul, V., Chan, C.W., and Tontiwachwuthikul, P., 2000. Development of an expert system for optimizing natural gas
pipeline operations. Expert Systems with Applications, 18, 271–282.
Usher, J.M., 2003. Negotiation-based routing in job shops via collaborative agents. Journal of Intelligent Manufacturing,
14, 485–499.
Van Dyke Parunak, H., 1998. Practical and industrial applications of agent-based systems. White Paper, Ann Arbor, MI:
Industrial Technology Institute.
Wang, J. and Shu, Y.F., 2005. Fuzzy decision modeling for supply chain management. Fuzzy Sets and Systems, 150 (1),
107–127.
Wang, L. and Shen, W., 2003. DPP: an agent-based approach for distributed process planning. Journal of Intelligent
Manufacturing, 14, 429–439.
Wu, Y. and Appleton, E., 2002. The optimization of block layout and aisle structure by a genetic algorithm. Computers
and Industrial Engineering, 41, 371–387.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
International Journal of Logistics: Research and Applications 39
Xue,X.,etal.,2005.Anagent-basedframeworkforsupplychaincoordinationinconstruction.AutomationinConstruction,
14, 413–430.
Yan, J., Chaudhry, P.E., and Chaudhry, S.S., 2003.A model of adecision supportsystem based on reasoningfor third-party
logistics evaluation. Expert Systems, 20 (4), 196–207.
Yandra, Y., et al., 2007. An integration of multi-objective genetic algorithm and fuzzy logic for optimization of agro-
industry supply chain design. Proceedings of the 51st annual meeting of the International Society for the System
Sciences, 5–10 August. Tokyo, Japan, 1–15.
Yu, W., Graham, J.H., and Min, H., 2002. Dynamic pattern matching for demand forecasting using temporal data mining.
Proceedings of the 2nd International Conference on Electronic Business, December. Taipei, Taiwan, 400–402.
Zha, X.F., Lim, S.Y.E., and Fok, S.C., 1999. Development of expert system for concurrent product design and planning
for assembly. International Journal of Advanced Manufacturing Technology, 15, 153–162.
Zheng, P. and Lai, K.K., 2008.A rough set approach on supply chain dynamic performance measurement. Lecture Notes
in Computer Science, 4953, 312–322.
Zhou, G. and Gen, M., 1998. An effective genetic algorithm approach to the quadratic minimum spanning tree problem.
Computers and Operations Research, 25 (3), 229–237.
Zhou, G. and Gen, M., 1999. Genetic algorithm approach on multi-criteria minimum spanning tree problem. European
Journal of Operational Research, 114, 141–152.
Zhou, G., Min, H., and Gen, M., 2002. The balanced allocation of customers to multiple distribution centers in the supply
chain network: a genetic algorithm approach. Computers and Industrial Engineering,43(1/2), 251–261.
Zhou,G.,Min,H.,andGen,M.,2003.Ageneticalgorithmapproachtothebi-criteriaallocationofcustomerstowarehouses.
International Journal of Production Economics, 86, 35–45.
Zimmermann, R., Hofmann, O., and Bodendorf, F., 2001.Agent-based tracking and tracing in inter-organizational supply
chains. Proceedings of the 7th Americas Conference on Information Systems, 409–411.
Downloaded By: [Min, Hokey] At: 16:26 12 February 2010
... The essential idea of AI is learning [30] and constant improvement over time. It refers to computer systems capable of mimicking, learning, and replacing human intelligence [31], but the transition from BA/I to AI is not self-starting and requires conceptually reasoned, empirically proven grounds. Despite its long history and widespread acceptance as a decision-aid tool, AI has seen limited application in SCRM. ...
... Despite its long history and widespread acceptance as a decision-aid tool, AI has seen limited application in SCRM. The potential of AI to solve complex problems and search for SCRM information has not been fully exploited [31]. ...
... AI analysts (e.g., Akerkar [14][15][16]; Davenprot [17]; Min [31]) point out that AI solutions often do not need to be developed (i.e., coded) from scratch. The set of SCRM-relevant indicators defined here will also require extensive use of other AI techniques. ...
Article
Full-text available
As supply chains (SCs) become more complex globally, businesses are looking for efficient business analytics (BA), business intelligence (BI), and artificial intelligence (AI) tools for managing supply-chain risk. The tools and methodologies proposed by the supply-chain risk management (SCRM) literature are mostly based on experts’ judgments, their knowledge, and past data. The expert evaluation-based approach could be partly or fully replaced by AI solutions, increasing objectivity, impartiality, and impersonality, reducing sources of human mistakes, biases, and inefficiencies in SCRM. However, the transition from BA to AI in SCRM is not a self-contained process; though attractive as a vision, it is not straightforward as a management or implementation process. The purpose of this research is to explore and define the conceptual grounds for transitioning from BA to AI in SCRM. The conceptual SCRM structure, its AI suitability, and implementation terms are defined theoretically based on a literature review. A single, in-depth business case study is employed to explore the theoretically defined terms of AI-based SCRM implementation. The proposed conceptual AI-suitable SCRM structure is defined by five principal building blocks: risk events, risk-event indicators, data-processing rules and algorithms, analytical techniques, and risk event probability forecasts. The study concludes that the business environment meets AI-based SCRM-implementation terms of data existence and access. Since data on risk events and negative outcomes are limited for machine learning, experts’ experience and knowledge might be utilised to build initial rules and data-processing algorithms for AI.
... However, this return strategy is not simply the return of production factors, but the upgrading of traditional manufacturing [25]. Intelligent technology has a wide range of permeability characteristics, which can improve the original dominant technology and production methods of traditional industries, and significantly improve their productivity and resource allocation efficiency [26][27][28]. Therefore, intelligent technology enhances the rationalization of the industrial structure by improving the added value of traditional industries, narrowing the technological gap between industries, and enhancing the flow of resource factors. ...
Article
Full-text available
In recent years, the application of artificial intelligence (AI) has had a significant impact on economic development. This study examined the relationship between the level of AI development and economic growth in 28 Chinese provinces from 2005 to 2018, and we focused on the mediating role of the industrial structure. We found that the unreasonable state of the structure is an important reason behind the slowdown of China’s economic growth. The development of AI not only has a direct effect on economic growth, but can also improve economic slowdown by inhibiting industrial structure upgrading. Taking into account regional heterogeneity, we also conducted sub-regional regressions, and the results show that this mediating effect is particularly significant in the eastern, central, and western areas of China; the regression results also show that the development of AI technologies did not boost the economy before the 2008 financial crisis, but during the economic recovery period, the R&D and application of AI helped China’s economy to rebound. Thus, AI has gradually become an important power engine for high-quality and sustainable growth in China’s economy.
... An organisation's resilience encompasses a proactive stage and a reactive stage (Ivanov and Dolgui 2019;Modgil, Singh, and Hannibal 2022). In a proactive stage, AI has been put forward as a useful tool that helps the firm connect its customers, suppliers, and supply chain partners by facilitating information exchange among various business entities (Cadden et al. 2021;Min 2010). In a reactive stage, resilience deals with the company's ability to restore its performance following a significant change in environmental conditions (Aven 2017;Belhadi et al., 2021b). ...
Article
Artificial intelligence (AI) is considered a mechanism that can improve supply chain resilience. Organisations around the world are investing in implementing AI systems to improve their supply chain and become more resilient to pandemics and disruption. At the same time, practitioners are not fully aware of the factors that impact the implementation of these systems. Alongside this, the extant literature lacks a comprehensive study that evaluates the enablers impacting the implementation of AI in production systems. This research fills this gap by identifying, defining, and evaluating the critical enablers influencing the adoption and implementation of AI in production systems. We extracted twelve enablers, created a conceptual model, and categorised the enablers based on the Technology , Organization, and Environment (TOE) framework. After categorisation, we used the analytical hierarchy process to assess the importance of the enablers presented in the model using data collected from eight experts. The results revealed that technology, as a category, is more crucial than organisation or environment. The findings also indicated that project management is the most critical of all twelve enablers. We discuss the implications of the analysis for practitioners and researchers. We also offer twelve propositions that researchers can empirically assess in future studies.
... First, AI facilitates the development of more proactive global supply chains (Min, 2010;Baryannis et al., 2019). Data is already vital to the success of supply chains from the very beginning to the very end. ...
Article
Full-text available
Technology has historically played a role in shaping international trade, but the current explosion in Artificial Intelligence has the potential to radically alter global commerce in the years ahead. In this research, we hypothesized that the AI capability of a nation has a major impact on international trade. This study discusses different ways in which technological advancements in the AI domain are improving global trade. We tested the hypothesis using the WDI, Government AI Readiness Index panel dataset of 150 countries for the years 2018-2021. Fixed effect, and Random effect panel models were applied. The results show that the AI capability of a nation has a major positive influence on trade. The findings also show that GDP and exchange rate have significant positive impacts, and inflation and trade restrictions have negative and significant impacts on trade. The findings of this study recommend strengthening the nation's AI capacity to increase its trade volume. AI will stimulate better economic development and open up new avenues for international trade to the extent that it fosters productivity growth. However, governments will need time to adapt and employ new AI technology, since doing so requires significant financial investments, access to skilled people, and a shift in how international companies are operated.
Article
Bu makalede bibliyometrik analiz kullanılarak lojistik ve tedarik zincirinde yapay zekâ çalışmaları ile ilgili bilimsel araştırmaların kapsamlı bir görünümünü sunmak amaçlanmıştır. Veriler 10 Mayıs 2022 tarihinde WoS veri tabanında yayınlanan 98 çalışmadan elde edilmiştir. Elde edilen veriler Excel ve VOSviewer yazılımı kullanılarak değerlendirilmiştir. Sonuçta lojistik ve tedarik zincirinde yapay zekâ ile ilgili çalışmaların 2018 yılından itibaren hızlı bir şekilde arttığı, 55 adet çalışma ile en çok araştırma makalesinin yayınlandığı görülmüştür. En üretken yazarlar Gunasekaran A. ve Klumpp, M.’dir. Çin 19 çalışma ile ortak yazarlıkta en fazla işbirliği yapan ülkedir. Türkiye’nin ortak yazarlıkta en fazla işbirliği yaptığı ülkeler Çin ve Hindistan’dır. Klumpp, M. en fazla alıntılanan yazar, ABD en fazla alıntılanan ülkedir. Türkiye en fazla alıntılanan beşinci ülkedir.
Conference Paper
Capacitance-Resistance model (CRM) has been a useful tool for fast production forecasts for decades. The unique combination of simplicity and physics-based nature in this data-driven approach allowed it to stay as an object of scientific interest and get its own place among other types of models capable of giving predictions on flow rates, such as full-scale 3D reservoir models. However, the model simplicity, assumptions, and limitations does not allow wide application of a conventional CRM in complex field cases. A vast majority of studies on CRM are about overcoming its limitations by introducing new coefficients, modifying the analytical form of the equation, etc. Integrating CRM with rapidly developing artificial intelligence (AI) methods seems to be a logical continuation of model evolution. Recently introduced Physics-Informed Neural Networks (PINN) can preserve CRM's governing equations and coefficients that gives some insights about wells and formations standing out from other popular machine learning and deep learning methods. Moreover, PINN type models give certain flexibility in the choice of architectures – it means that the model architecture can be changed in a way that may assist in solving different problems. Thus, we introduce end-to-end learning of neural networks (NN) while implying some physical constraints. It is intended to overcome one of the major limitations of CRM, which is obtaining predictions for oil and water production rates from total liquid. This way, the additional training of rough approximation fractional flow models that are either not suitable for the case or may require the knowledge of reservoir properties is not needed. In this work, the well-known concept of Capacitance-Resistance models appears in a new form, which allows performing history matching rather rapidly, achieving robustness and forecasting liquid, oil and water production rates simultaneously. To test this new approach, several datasets (both synthetic and real) were used. The results obtained by PINN are compared to those obtained by a conventional CRM. By conventional we mean the analytical solution, which was modified by our research group to take into consideration common real field cases such as shut-in wells, workover operations, etc. by introducing dynamic characteristic coefficients [1].
Conference Paper
Accurately forecasting demand is one of the most undervalued and complex strategies that can significantly impact organizations bottom line. This industrial field study was co-conducted with Sumitomo Corporation's Tubular Division which primarily deals with high-grade Oil Country Tubular Goods (OCTG) globally. The presented solution demonstrates how with the right data set (drilling sequence data, stock data and consumption data), artificial intelligence can be used to build out a model that can quantify and predict future demand accurately thereby reducing cost, working capital and emissions. Multiple multi-layered machine learning models were built to compare and analyze a wide variety of data inputs for bill of materials, operational/project schedules; This includes (a) ‘product movement data’ which describes the changes in demand and supply of a product, (b) ‘product specification data’ which describes the characteristics of a product, and (c) ‘activity specification data’ which describes the characteristics of an activity. The models follow the base temporal map design with different weighting on model inputs. With a temporal map, a sequence of monthly data values (called lags) is used to predict the next monthly value in the sequence. The lags are rolled so that there are six months of data for the model to predict on. All models also use boosted decision-tree-based ensemble machine learning algorithm. It is critical to understand how product movement metrics (actual and safety stock levels, historical forecasts, and consumption patterns), product specification data (lead time, product grade, well function, well category, work center), and external factors (oil price, rig counts, national budget, production targets) can be utilized together to better understand future product demand. Using historical data acquired from drilling operations and supply chain over an eight-year period, multiple machine learning models were trained to predict one year of demand across the most consumed products. Across five years of predictions (2016 to 2019), the models were able to predict with 78% average accuracy for the top 10 products by volume which represents 75% of inventory volume. Across the same time-period, they were able to predict with 73% average accuracy on all 17 products which account for 80% percent of inventory volume. Further iterative updates with additional data led to improvement in results and the model where the model predicted with an improved accuracy of 83% on the top 17 products and an accuracy of 86% on the top 10 products. Moreover, the data can also be used to generate dashboards featuring metrics on material uncertainty / velocity and expected differences between the internally predicted forecasts and actual sales. The results further indicate that, on average, and within a simulated environment (where shipping delays were not considered for instance,) the AI model can maintain a lower inventory than the originally planned stock levels at lowest cost and footprint. This would not only lead to less resource consumption, but also reduce the embodied carbon and emissions within the overall process. This novel study presents the success of a validated tailored AI model for inventory forecast with field data and commercial implementation. Such a tool can be integrated into other value adding digital tools, such as integrated schedule optimization, logistics optimization and management systems to make overall operations more efficient and sustainable with lower costs, inventory, wastage, and reduced emissions.
Conference Paper
Artificial Intelligence (AI) has significant potential to optimize practices, processes, and energy consumption along with maximizing yield, quality, and uptime. This has substantial impact on putting organizations on the path to net-zero, as such optimizations can reduce greenhouse gas emissions by 20% with minimal capital investments. This comprehensive study presents proven industrial case studies that delivered economically strong strategies coupled with sustainability practice and providing strategic insights to identify, manage and/or attenuate the associated impacts. Environomics presented in this study is a novel framework which deals with unifying economic strategies with sustainability practices (through artificial intelligence) for optimal business performance in terms of finances but also environmental impact. This is achieved through a track, trace, and optimize approach for resources (particularly emissions, energy, water, waste, materials,, and safety) This was achieved through a combination of AI methods such as unsupervised machine learning, multi-variate optimization, and the implementation of similarity measures. A few of the inputs included well data (including production data, drilling data, completion data etc.), logistics/supply chain data (scheduling data, production inventory, mobilization data etc.), safety data (near-miss, observations, hazards, disciplines and insights etc.) with associated costs and emission data. Multiple industrial case studies are presented where sustainability metrics are identified through validated AI models to optimize productivity while reducing emissions and inventory. For instance, well profiling can be used to identify historical parameters that have maximized production potential while optimizing for aspects such as cost or emissions. Furthermore, we can identify the optimal completion parameters for a new well which satisfies carbon targets, use well profiles to build an optimized drilling schedule that meets budget or production criteria while still achieving production targets and optimizing drilling rig routes. Thus, the approach can quickly (within run time) solve interrelated environomic challenges in the reservoir studies space and the field development space. Further case studies indicate that the supply chain can have immense optimization impact on scope 3 aspects with results indicating 30-50% asset utilization improvement with respect to fleets (Vessel, Truck, Rigs). With respect to materials, a 10-20% reduction of material inventory levels all improved through AI. As the workforce are also part of the environment it has been observed that identifying unsafe behaviors within a large operation, also leads to enhanced sustainability behaviors. The models indicate potential of overall emission reduction ranging from 12-20%. This led to the comprehensive framework presented in this study to support sustainable practices that are also economically feasible and deployable. The real-time sustainability metrics generated has immense values in terms of decision-making processes and scenario generation in a fraction of the time that is required using traditional approaches. In addition to assessing the scope of impact, a novel multidisciplinary study and framework is presented to analyze environomic strategies to propose a market-oriented approach through the application of artificial intelligence. Furthermore, industrial, and academic case studies have been evaluated to identify, predict, and optimize the crucial parameters within such workflows that are effective in reducing resources utilized and associated emissions.
Article
This paper investigates the transformation of information sharing and multi-tier supply chain management (MSCM) from the perspective of small and medium-sized enterprises (SMEs) in the context of Industry 4.0. SMEs are typically sharing less information digitally, are less active in conducting MSCM, but must be integrated in digital supplier networks to achieve supply chain transparency or traceability. For this purpose, this paper empirically analyzes SMEs from the manufacturing industry in Austria and Germany are investigated regarding information sharing and MSCM. The responses are analyzed to test three hypotheses, using 81 replies from manufacturing SMEs in Austria and Germany. The three hypotheses investigate the upstream and downstream MSCM practices of SMEs as well as which kind of information is shared with and by SMEs. The results help to understand and reveal potential barriers for information sharing besides technical possibilities to SMEs in order to be actively involved in MSCM.
Article
Purpose Although scholars argue that artificial intelligence (AI) represents a tool to potentially smoothen the bullwhip effect in the supply chain, only little research has examined this phenomenon. In this article, the authors conceptualize a framework that allows for a more structured management approach to examine the bullwhip effect using AI. In addition, the authors conduct a systematic literature review of this current status of how management can use AI to reduce the bullwhip effect and locate opportunities for future research. Design/methodology/approach Guided by the systematic literature review approach from Durach et al. (2017), the authors review and analyze key attributes and characteristics of both AI and the bullwhip effect from a management perspective. Findings The authors' findings reveal that literature examining how management can use AI to smoothen the bullwhip effect is a rather under-researched area that provides an abundance of research avenues. Based on identified AI capabilities, the authors propose three key management pillars that form the basis of the authors' Bullwhip-Smoothing-Framework (BSF): (1) digital skills, (2) leadership and (3) collaboration. The authors also critically assess current research efforts and offer suggestions for future research. Originality/value By providing a structured management approach to examine the link between AI and the bullwhip phenomena, this study offers scholars and managers a foundation for the advancement of theorizing how to smoothen the bullwhip effect along the supply chain.
Article
Full-text available
Traditional time series methods are designed to analyze historical data and develop models to explain the observed behaviors and then predict future value(s) through the extrapolation from the models. The underlying premise is that the future values should follow the path of the historical data analyzed by the time series methods, and as such, these methods necessitate a significant amount of historical data to validate the model. However, this assumption may not make sense for applications, such as demand forecasting, where the characteristics of the time series may alter frequently because of the changes of consumers' behavior and/or cooperate strategies such as promotions. As the product life cycle gets shorter as it tends to be in today's e-business, it becomes increasingly difficult to make a forecast using traditional time series methods. In response to this challenge, this paper proposes a novel pattern matching procedure to decide whether one or combination of several patterns actually represents the development of the time series and then to use the patterns in forecasting. Several pattern transformation algorithms are also proposed to facilitate a flexible match. Rematching through dynamic reevaluation of the new data may be needed until the true development of the time series is discovered. Initial evaluation indicates superior performance in predicting the demand of a new product.
Chapter
We’ve discussed a few traditional problem-solving strategies. Some of them guarantee finding the global solution, others don’t, but they all share a common pattern. Either they guarantee discovering the global solution, but are too expensive (i.e., too time consuming) for solving typical real-world problems, or else they have a tendency of “getting stuck” in local optima. Since there is almost no chance to speed up algorithms that guarantee finding the global solution, i.e., there is almost no chance of finding polynomial-time algorithms for most real problems (as they tend to be NP-hard), the other remaining option aims at designing algorithms that are capable of escaping local optima.
Article
A mathematical model for a multivehicle pickup and delivery problem with time windows is presented, and a genetic algorithm (GA) for solving it is proposed. The mathematical model is formulated as a mixed-integer linear programming problem. The objective of the proposed model is to minimize the total cost, which consists of the fixed cost for the vehicles, the routing cost, and the customer inconvenience cost, which is modeled as a penalty cost for violation of the time windows of each customer. Like other combinatorial problems, solving this pickup and delivery problem is time-consuming, and sometimes it is impossible to find an exact solution. The problem is solved exactly for up to six demands (12 nodes), and GA is used for larger problems with more than six demands. The proposed GA can solve a pickup and delivery problem in an extremely short time compared with the exact solution procedure. It also produces excellent results for small problems. A GA used to solve a 30-demands problem (60 nodes) with 10 vehicles is illustrated.
Article
This paper explores the application of the multi-agent system paradigm to modelling, parsing and prediction of multi-dimensional discrete time-series data. A reactive, grammar-like hierarchical model is inferred from the data. The model is instantiated as a collection of pattern agents, each representing a different pattern. When analysing data, each agent uses its own view to generate predictions about what is to come next. The outputs of the various agents are then integrated to form the prediction of the system as a whole. Additionally, a distributed algorithm similar to bottom-up chart parsing is developed in which the agents compete and co-operate with each other on-line to arrive at an optinial parsing. Results with musical data indicate that this approach is well suited for the explicit handling of multiple hypotheses, especially when modelling on-line data analysis processes.