- Email: [email protected]
Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains
Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Simulation Modelling Practice and Theory journal homepage: www.elsevier.com/locate/simpat
Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains Aida Huerta-Barrientos ⇑, Mayra Elizondo-Cortés, Idalia Flores de la Mota Department of Operations Research, National Autonomous Mexico University, Ciudad Universitaria, 04510 Mexico, DF, Mexico
a r t i c l e
i n f o
Article history: Available online xxxx Keywords: Simulation Optimization Network analysis Scientific collaboration Supply chain
a b s t r a c t In the 1970s, a co-authorship network in the field of Simulation Optimization of supply chains was established, supported by local associations. Then, the development of this network was favored by the foundation of new co-authorships and the consolidation of already existing. The purpose of this study is to analyze the structure, collaboration patterns and the time-evolution of the co-authorship network of Simulation Optimization of supply chains. Data are based upon 202 peer-reviewed contributions published from 1970 to August 2012 in relevant journals indexed in the ISI/Web of Science database and International Conferences. The analysis is conducted using exploratory social network analysis technique. Results indicate that the development of knowledge in Simulation Optimization of supply chains has been carried out mainly by 353 authors from 35 countries. Also, there have been proposed over forty Simulation Optimization methods by different authors however the most usual is response surface methodology, followed by gradient based search method and genetic algorithms. In addition, applications of Simulation Optimization methods and techniques are found mainly in areas as health care, management, transport, airline, telecommunications, aerospace, and financial. Although research in Simulation Optimization of supply chains has received much attention by the simulation community, its application in key industries continues to be still small, limiting its support in decision-making. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction Simulation Optimization is a structured approach that is useful to determine optimal settings for input parameters associated with a simulation model. In this case, the optimality is measured by a (steady-state or transient) function of output variables [1]. As suggested by Fu [2], the general optimization problem consists of finding a setting of controllable parameters that minimizes a given objective function, i.e.
min JðhÞ h2H
ð1Þ
where h e H represents the vector of input variables, J(h) is the objective function, and H is the constraint set, which may be either explicitly or implicitly defined. The assumption in the Simulation Optimization setting is that J(h) is not available directly, but must be estimated via simulation [2]. ⇑ Corresponding author. Address: Faculty of Engineering, Av. Universidad 3000, Ciudad Universitaria, C.P. 04510, Mexico. Tel.: +52 5558419565. E-mail address: [email protected] (A. Huerta-Barrientos). http://dx.doi.org/10.1016/j.simpat.2014.02.007 1569-190X/Ó 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
2
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
The key difficulty in Simulation Optimization involves a trade-off between allocating computational resources for searching h e H versus conducting additional simulation replications for better estimating the performance of current promising solutions [3]. Since the 1970s several Simulation Optimization techniques have been proposed for finding optimal settings for input parameters of a simulation model. Some of these methods have been very sophisticated while others have been naïve just generating several input values at random and running the simulation model at each of these input values. None of Simulation Optimization techniques is sure to work for all problems, but each one has characteristics which make it useful for certain types of problems [4]. Extensive reviews and surveys about Simulation Optimization methods and techniques have been carried out by different authors and can be found in [5–14,3,15,16]. However, at present little attention in the literature has been paid to the co-authorship network of Simulation Optimization and its time-evolution and very little has been written about main applications of Simulation Optimization techniques in industry. Peer-reviewed contributions are considered an important source of information that let us to identify the development of knowledge in a specific field, and pinpoint how research is organized and structured [17]. In this direction, specific research questions addressed by this study are: how do authors of peer-reviewed articles in the field of Simulation Optimization of supply chains collaborate? How is structured the co-authorship network of Simulation Optimization of supply chain? Which Simulation Optimization methods have been proposed by researchers? And last but not least, which industries have been supported by Simulation Optimization methods? In line with this, the aim of this study is to analyze the structure, collaboration patterns and the time-evolution of the co-authorship network of Simulation Optimization of supply chains. In order to achieve this aim 202 peer-reviewed articles were selected using the purposive sampling method, which is defined in Section 2. Based on the information about authors, we developed the co-authorship network and analyze its structural properties and time-evolution using exploratory social network analysis technique. This technique has the advantage of determine at global level structural features of a network, detecting its patterns. Then Simulation Optimization methods were classified based on their application areas. This classification is of relevance because provides guidance for industrials, academics and practitioners in optimization method selection. After that, industries that have been supported by Simulation Optimization methods in making decisions processes were listed. Although theoretical Simulation Optimization methods are numerous, key industrial areas of application are still small. We highlight this gap and recommend how to increase the application of Simulation Optimization methods in real-world problems solutions. The remainder of this paper is organized as follows. In Section 2, the data collection method is presented. The exploratory social network analysis technique is described in Section 3. The co-authorship network of Simulation Optimization of supply chains is characterized in Section 4. The structure and time-evolution of the co-authorship network of Simulation Optimization of supply chains is analyzed in Section 5. Main conclusions and future research are drawn in Section 6.
2. The data collection We collected 202 peer-reviewed contributions in Simulation Optimization of supply chains from relevant journals indexed in the ISI/Web of Science database and International Conferences. These contributions were selected based on purposive sampling method. This method is also referred to as qualitative sampling that involves certain units or cases based on a specific purpose rather than randomly [18]. Three broad categories of purposive sampling techniques are well known: sampling to achieve a representativeness or comparability, sampling special or unique cases, and sequential sampling [18]. In this study, peer-reviewed contributions were sampled based on the first category of the purposive sampling techniques to achieve a representativeness of the application of Simulation Optimization in supply chain field. The criterion to filter a contribution was the inclusion of the phrase ‘‘simulation optimization’’ in its title but with its application in supply chain field. The period of publication taken in account was from 1970s to August 2012. As the first contribution was published in the 1970s, additional queries regarding five periods were placed in data as follows: Period I, 1970–1979; Period II, 1980–1989; Period III, 1990–1999; Period IV, 2000–2009; and Period V, 2010–2012. As it is noted from Table 1, for each period peer-reviewed contributions were quantified as well as International Journals and Conferences, authors and their countries. The tendency in the number of peer-reviewed contributions published over five periods already mentioned, suggests the attracting increasing interest from the simulation community in Simulation Optimization field since the last decades. The distribution of peer-reviewed contributions based on the number of co-authors is summarized in Table 2. Two-authored contributions represent the biggest proportion with 41% of the total. In contrast to this, the single-authored contributions represent only 15%.
Table 1 Dissemination of peer-reviewed contributions of Simulation Optimization of supply chains. Period Years
I 1970–1979
II 1980–1989
III 1990–1999
IV 2000–2009
V 2010–2012
Authors Peer-reviewed contributions Int. Journals/Conference Countries
10 5 1 1
20 15 2 2
32 20 1 2
232 118 23 29
100 44 17 18
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
3
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx Table 2 The distribution of peer-reviewed contributions of Simulation Optimization of supply chains by number of authors. Description
Peer-reviewed contributions
% Peer-reviewed contributions
Two-authored Three-authored Four-authored Single-authored
82 52 37 31
41 26 18 15
Table 3 Journals and international conferences accounting for at least one peer-reviewed contribution in Simulation Optimization of supply chains from 1970 to 2012. Journal/conference
Peer-reviewed contributions
Winter Simulation Conference Simulation Modelling Practice and Theory Int. J. Production Economics Computers and Chemical Engineering Computers and Operations Research European Journal of Operational Research Computers and Industrial Engineering IIE Transactions ACM Transactions on Modeling and Computer Simulation Applied Soft Computing Automatica Engineering Applications of Artificial Intelligence Engineering Optimization European Simulation Symposium Expert Systems with Applications Handbooks in Operations Research and Management Science Industrial and Engineering Chemistry Research INFORMS Journal on Computing INFORMS Simulation Society Research Workshop International conference on Intelligent Systems Modelling and Simulation International conference on Service Systems and Service Management International Journal of Computer Integrated Manufacturing International Journal of Industrial Engineering: Theory Applications and Practice International Journal of Production Economics International Journal of Production Research Irish Journal of Management Journal of Mining Science Mathematics and Computers in simulation Naval Research Logistics OR Spectrum Simulation Practice and Theory Technological and Economic Development of Economy The Arabian Journal for Science and Engineering
154 5 4 3 3 3 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Table 3 indicates 33 journals and international conferences that include at least one peer-reviewed contribution sampled for this study. The top contributor is International Conference Winter Simulation Conference followed by International Journal Simulation Modelling Practice and Theory. Involved international journals and conferences, suggest a multidisciplinary interest in the application of Simulation Optimization methods in supply chain field, as they belong to different industrial areas such as Production, Operations Research, Chemical Engineering, Industrial Engineering, Computer Engineering, Economy, Management, and Logistics. 3. The exploratory social network analysis technique Social networks can be analyzed using different techniques. Important results have been obtained using the exploratory social network analysis technique. Through this technique, it is possible to analyze the time-evolution of a network. Also it is possible to detect and interpret patterns of social ties. In general, the exploratory social network analysis technique consists of four parts [19]: 1. 2. 3. 4.
The The The The
definition of a network; network manipulation; determination of structural features; visual inspection.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
4
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
Fig. 1. The pattern of the co-authorship network of Simulation Optimization of supply chains from 1970 to 2012.
Fig. 2. The pattern of the co-authorship network of Simulation Optimization of supply chains independently of the year from 1970 to 2012.
The definition of a set of vertices and a set of lines where each line connects two vertices is included in the definition of a network. A vertex is considered the smallest unit in a network, while a line is a tie between two vertices. In social network analysis, a vertex represents an actor, and a line represents any social relation. The network manipulation is a very powerful tool that lets us to modify a network according with our requirements. Considering an entire network, by social network analysis technique, its structural features can be determined at global level. And considering a sub network or a single vertex extracted from the entire network, its structural features can be determined at individual level. Additionally, the visual inspection of a network by means of the social network analysis technique facilitates the intuitive understanding of its features, helping us to trace and represent patterns of ties [19]. It is important to stress that visual inspection of a network can be carried out using the Kamada–Kawai free algorithm, well supported by Pajek software. This algorithm is useful for drawing undirected and weighted graphs. The basic idea of this algorithm consists on regarding the desirable ‘‘geometric’’ (Euclidian) distance between two vertices in the drawing as the ‘‘graph theoretic’’ distance between them in the corresponding graph and introducing a virtual dynamic system in which every two vertices are connected by a ‘‘spring’’ of such desirable length. So, the optimal layout of vertices indicates us the state in which the total spring energy of the system is minimal [20]. 4. The co-authorship network of Simulation Optimization of supply chains The co-authorship network of Simulation Optimization of supply chains is represented by authors of peer-reviewed contributions and their interrelations. An author is represented by a vertex, while the interrelations between authors are represented by edges in the co-authorship network. Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
5
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
Fig. 3. The giant component of the co-authorship network of Simulation Optimization of supply chains.
Table 4 Authors who integrate the giant component of the co-authorship network of Simulation Optimization of supply chains.
a
Author
Index
Luo Y. Cho H. Jacobson S. Hyden P. Keng N. Shi L. Carson J. Chen Ch. Healy K. Eren Ultanir A. Bektas T. Ilenda V. Hall J. Better M. Yücesan E. Yoo T. Swisher J. Prudius A. Laguna M. Kleinman N. Kelly J. Jarugumili S. Hutchison D. Hill S. Fu M. Dengiz B. Bowden R. Boesel J. April J. Andradóttir S.
352 348 334 333 267 248 247 246 245 233 232 231 212 202 182 179 160 138 102 91 86 82 79 76 59 48 27 26 15 12
a
The index used to identify authors in the co-authorship network of Simulation Optimization of supply chains.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
6
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
4.1. Trends in the co-authorship network of Simulation Optimization of supply chains The primary constraint on the pattern of the co-authorship network is the number of authors on a contribution. In this direction, the first co-authorship network was defined to detect the total pattern of co-authorships between authors in the timeline from 1970 to 2012, described in Section 2. In this case, each author in 1 year is represented by one vertex. It is important to note that for each contribution the co-authorships were established between first and second author, between first and third author, and between first and fourth author, respectively. Using the Kamada–Kawai free algorithm, the visualization of the co-authorship network of Simulation Optimization of supply chains is showed in Fig. 1. In other case, considering that one vertex represents one and just one author, independently of time on which its contribution was published, the co-authorship network was developed, see Fig. 2. That means that an author is represented by one vertex, independently if the same author has others co-authorships in the timeline considered. To really interpret patterns of co-authorship of Simulation Optimization of supply chains, it was necessary to determine three of the main structural features of the co-authorship network: the giant component, the centralization and structural holes. On one hand, the giant component represents a subnet formed by the largest share of vertices (authors) interconnected within a network. The importance of the giant component lies on the possibility to reach a large number of other authors of the same collaboration network starting by one author of the giant component and moving along its connections. The
Table 5 The top ten centers of the co-authorship network of Simulation Optimization of supply chains correlated with their weighted values and index.
a
Author
Weighted values
Index
Fu M. Biles W. Eskandari H. Yücesan E. Ding H. Hay L. Azadivar F. Chen Y. Othman S. Dengiz B.
40 34 14 13 13 12 8 8 8 7
59 24 56 182 51 72 19 37 129 48
a
The index used to identify authors in the co-authorship network.
Fig. 4. Centers of the co-authorship network of Simulation Optimization of supply chains.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
7
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
appearance of the giant component with 30 authors is verified by black vertices of the co-authorship network presented in Fig. 3. The giant component embraced 8.78% of vertices in the network. It means that it was possible to reach a large number of other authors starting by one of vertices that conforms the giant component and moving along its connections. As the vertices in the co-authorship network were enumerated by its index, we listed in Table 4 the authors included in the giant component based on their index. On another hand, the centrality is one of the main concepts used in network analysis and refers to positions of individual vertices within a network. As is pointed out in [21], the centrality, concept proposed by Freeman, supports the measurement of centralization and can be calculated in terms of the degree to which a vertex falls on the shortest path between others vertices and therefore has a potential for controlling communication in a network. The importance of centrality lies on the idea that information may easily reach central vertices (authors) in a communication network. It is important to note that authors can benefit from serving as intermediaries between others authors who are not directly connected in a network. Through such intermediation, some authors potentially can broke the information flow and synthesize ideas arising in different parts of a network. These principles form the underpinning for structural holes theory. This theory indicate us ways in which some vertices (authors) fill ‘‘holes’’ between groups or sub networks that are not otherwise interacting in a network [22]. Nowadays, two classic centralization measures are applied to characterize a network: closeness and between’s. First, closeness centralization is defined as the variation in the closeness centrality of vertices divided by the maximum variation in closeness centrality. Second, between’s centralization is defined as the variation in the between’s centrality of vertices divided by the maximum variation in between’s centrality. Using Pajek software, it was not possible to obtain measurements of closeness centralization for the co-authorship network of Simulation Optimization of supply chains because this network is not strongly connected. This fact means that there are not paths between all authors in the co-authorship network. In
Fig. 5. Structural holes of the co-authorship network of Simulation Optimization of supply chains.
Table 6 Intermediary authors who are not directly connected in the co-authorship network of Simulation Optimization of supply chains.
a
Author
Indexa
April J. Chang K. Fu M. Glover F. Hong L. Noack D. Syberfeldt A. Wan H. Yücesan E. Ng A. Rose O. Chen Ch. Nelson B. Tiwari M.K.
15 36 59 61 77 127 161 174 182 195 201 246 262 284
The index used to identify authors in the co-authorship network of Simulation Optimization of supply chains.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
8
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
contrast to this, the between’s centralization value was calculated equal to 0.00569. This means that few crucial authors in the co-authorship network of Simulation Optimization of supply chains promote the information transmission through the entire co-authorship network established, i.e. Jacobson S., Fu M., and Andradóttir S. It is important to note that these crucial authors were also elements of the giant component of the same network. Additionally, centers of the co-authorship network in Simulation Optimization of supply chains were identified as vertices with the most weighted values based on the degree of
Fig. 6. The network of author’s countries.
Optimization Techniques Response surface methodology 11% Others 25%
Gradient based search methods 8% Genetic algorithm 8%
Mathematical programming 3% Simulated annealing 6%
Stochastic approximation 6% Tabu search 5% Scatter search 5%
Hooke-Jeeves pattern search method 3% Sample path method 4% Evolution strategy 4% Random search algorithms 4% Heuristics Ranking & selection 4% 4%
Fig. 7. Optimization techniques most usual in peer-reviewed contributions of Simulation Optimization of supply chains.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
9
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
vertices. Normally these weighted values are calculated using a weighted algorithm. Ten centers (authors) with the highest degree of vertices in the co-authorship network of Simulation Optimization of supply chains are listed in Table 5 and showed graphically in Fig. 4, represented by biggest black vertices. Co-authorships that potentially can break the flow of information and synthesize ideas of the co-authorship network of Simulation Optimization of supply chains are relationships represented by lines in grey color in Fig. 5. It is important to note that authors, who represent vertices of these co-authorships (see Table 6), served as intermediaries between others authors who are not directly connected in the co-authorship network.
4.2. The network of author’s countries The network of author’s countries also was built. A vertex of this network represented an author’s country and edges represented the collaboration between authors of different countries. On one hand, we observed the central role of authors from USA in this network and their co-authorships with authors from Germany, China, Chile, Belgium, Turkey, Taiwan, France, Singapore, Mexico, Thailand, Korea, Netherlands, Iran and India, mainly. On another hand, authors from United Arab Emirates have collaborated in the field of Simulation Optimization of supply chains only with authors from Saudi Arabia. The same situation is observed for authors from Ireland who have collaborated just with authors from Austria as is illustrated in Fig. 6.
4.3. Optimization techniques most usual in Simulation Optimization of supply chains Other interesting part of this study is the analysis of optimization techniques used by authors of the co-authorship network of Simulation Optimization of supply chains. We counted over forty different optimization techniques. The most usual in peer-reviewed contributions was the response surface methodology followed by gradient based search methods, genetic algorithms, simulated annealing, stochastic approximation, tabu search, scatter search, and others, as is presented in Fig. 7.
Industries Airline 9%
Pharmaceutical 2%
Aerospace 7%
Sport 2%
Energy 2%
Financial 7%
Telecommunications 11%
Newspaper 2%
SD 2%
Other 22% Mining export 2%
Transport 15%
Textil 2%
Military 2% Management 15%
Health care 15%
Fishery 2%
Chemical 2%
Fig. 8. Industries supported by Simulation Optimization methods.
Table 7 Supply chain and manufacturing activities supported by Simulation Optimization methods. Supply chain
Manufacturing
Design Planning Physical distribution planning Management Scheduling Resources allocation Storage and retrieval policies Flow Shop Logistics Inventory
Production Planning Production line Process plant design Flexible manufacturing Automated manufacturing system Lean manufacturing Robot manufacturing cell Buffer allocation Scheduling Assembly line
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
10
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
4.4. Industries supported by the application of simulation optimization methods A fundamental aspect of Simulation Optimization of supply chains is its application to support the solution of realworld problems. In this direction, industries that have been supported by Simulation Optimization methods and techniques from 1970 to 2012 are showed in Fig. 8. Industries in which these methods have been more applied are: health care, management, transport, telecommunications, airline, aerospace, and financial. Few applications in areas of energy, newspaper, mining export, military, fishery, chemical, textile, sustainable development, sport and pharmaceutical were found. Although Simulation Optimization methods are numerous, as we observed in Section 4.3, industrial areas of application are still small. 4.5. Supply chain areas supported by the application of Simulation Optimization methods and techniques The application of Simulation Optimization methods in supply chain and manufacturing activities has been focalized just in certain activities, see Table 7. Although the technological potential of these techniques, they have not been applied extensively to support the decision-making.
Fig. 9. The co-authorship network of Simulation Optimization of supply chains in Period I (1970–1979).
Fig. 10. The co-authorship network of Simulation Optimization of supply chains in Period II (1980–1989).
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
11
Fig. 11. The co-authorship network of Simulation Optimization of supply chains in Period III (1990–1999).
Fig. 12. The co-authorship network of Simulation Optimization of supply chains in Period IV (2000–2009).
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
12
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
5. The structure and time-evolution of the co-authorship network of Simulation Optimization of supply chains Since the 1970s, the structure of the co-authorship network of Simulation Optimization of supply chains started to be developed based on several local associations, forming little communication groups between its members. In Period I, from 1970 to 1979, five co-authorships were established, see Fig. 9. Optimization techniques mainly used by authors in this period were coordinate search, gradient based search, heuristics, Hooke–Jeeves pattern search, random search, and response surface methodology, in health care and transport industries. In Period II, from 1980–1989, co-authorships were incremented to eight (see Fig. 10) and optimization techniques mainly used were Nelder method, perturbation analysis, response surface methodology, simulated annealing, and stochastic approximation supporting decision problems in the aerospace industry. In Period III, from 1990 to 1999, co-authorships of Simulation Optimization of supply chains were incremented to 17 (see Fig. 11). Optimization techniques used in this period were mathematical programming, Nested partitions method, neural networks, perturbation analysis, random search, ranking and selection, response surface methodology, sample path, scatter search, simulated annealing, stochastic approximation and tabu search, to support the decision-making in airline and transport industries. In Period IV, from 2000 to 2009, co-authorships of Simulation Optimization of supply chains were incremented to 195, see Fig. 12. In this period, authors used optimization techniques such as adaptive partitioning search, approximate dynamic programming, brute force method, coordinate search, evolution strategy, genetic algorithms, golden region search, gradient based search, heuristics, Hooke–Jeeves pattern search, kriging methodology, mathematical programming, multiple comparison, Nested partitions, neural networks, particle swarm optimization, perturbation analysis, random search, ranking and selection, response surface methodology, retrospective approximation, sample average approximation, sample path, scatter search, sequential selection, simulated annealing, stochastic approximation and tabu search. Industries of interest in applying these techniques were aerospace, airline, telecommunications, management, financial, heath care, military, sustainable development, chemical, newspaper, energy, textile, and fishery.
Fig. 13. The co-authorship network of Simulation Optimization of supply chains in Period V (2010–2012).
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
13
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx Table 8 Optimization methods used in Simulation Optimization of supply chains from 1970 to 2012. Optimization method
Period I
Period II
Period III
Period IV
Period V
Gradient based search methods Response surface methodology Hooke–Jeeves pattern search method Heuristics Random search algorithms Simulated annealing Stochastic approximation Genetic algorithm Mathematical programming Nested partitions method Neural networks Perturbation analysis Ranking and selection Sample path method Scatter search Tabu search Frequency domain method Nelder method Evolution strategy Simultaneous Perturbation Stochastic Approximation Kriging methodology Particle swarm optimization Sequential selection Coordinate search Geoffrion and Graves method Discrete stochastic optimization Adaptive Partitioning search Approximate dynamic programming Brute force method Cross-entropy method Estimation of distribution algorithms Golden region search Greedy heuristics Indifference-zone ranking Mixed integer programming Model reference adaptive search Multiple comparison Non-monotonic search Retrospective approximation Sample average approximation Branch -and bound method, Linear programming Optimal computing budget allocation
X X X X X
X X X
X X X X X X X X X X X X X X X X X
X X X X X X X X X X X X X X X X
X X
X X
X X X X X
X X
X
X X
X X X X X X X X X X X X X
X X X
X X X X X X X X X X X X X X X X X X X
Over time, the structure of the co-authorship network of Simulation Optimization of supply chains has been favored mainly by the foundation of new co-authorships and the consolidation of already existing. It is important to stress that in Period V, from 2010 to 2011, a total of 71 co-authorships were established, see Fig. 13. Optimization techniques used by authors in this period were genetic algorithms, gradient based search, heuristics, kriging methodology, mathematical programming, Nelder method, Nested partitions, neural networks, optimal computing budget allocation, particle swarm optimization, random search, ranking and selection, response surface methodology, sample path, scatter search, sequential selection, simulated annealing, stochastic approximation, and tabu search to support the decision-making in aerospace, telecommunications, health care, port, transport, mining export, and pharmaceutical industries. Table 8 presents an overview of optimization techniques used in all five periods (Period I, Period I, Period III, Period IV and Period V). The optimization method most usual is response surface methodology. It is important to outline that response surface methodology has been applied mainly in transport and health industries, in areas of inventory and manufacturing. Also this methodology has been combined with other optimization techniques such as genetic algorithms and Taguchi.
6. Conclusions and future research This study demonstrates that since the 1970s the knowledge dissemination of Simulation Optimization of supply chain has involved a large number of authors from different countries, peer-reviewed contributions, international journals and conferences. We analyzed 202 peer-reviewed contributions in Simulation Optimization of supply chains from relevant journals indexed in the ISI/Web of Science database and International Conferences. From the analysis, we observed that about 85% of peer-reviewed contributions were developed by more than one author. That means that the development of Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
14
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
knowledge in this field is mainly based on co-authorships. The co-authorship network of Simulation Optimization of supply chains was represented by authors of peer-reviewed contributions and their interrelations. The findings in the analysis of the co-authorship network indicate that it was possible to reach a large number of other authors in the network through 30 authors moving along the edges of the subnet that form these authors. Also, few authors were crucial to the transmission of information through the co-authorship network and some of these authors also had the function of centers in the network as Fu M. and Yücesan E. Additionally, we observed that the optimization technique most usual in peer-reviewed contributions analyzed was response surface methodology followed by gradient based search methods, genetic algorithms, simulated annealing, stochastic approximation, tabu search and scatter search. Response surface methodology has not been used alone instead it has been combined with other optimization techniques, incrementing in this way the potential of optimization. Although Simulation Optimization methods are numerous, they have not been applied extensively to support the decision-making in many areas of supply chain and manufacturing. As the structure of the collaboration network between authors has been favored mainly by the foundation of new co-authorships and the consolidation of existing, it can be a good strategy to increment the number of applications of simulation optimization in more supply chain areas based on new coauthorships, collaborating. Future research is needed into co-authorship network analysis based on impact factors of journals indexed in the ISI/Web of Science database. References [1] J.R. Swisher, E. Yücesan, Discrete-event simulation optimization using ranking, selection, and multiple comparison procedures: a survey, ACM Trans. Model. Comput. Simul. 13 (2003) 134–154. [2] M.C. Fu, Feature article: optimization for simulation: theory vs. practice, INFORMS J. Comput. 14 (2002) 192–215. [3] M.C. Fu, Ch. Chen, Some topics for simulation optimization, in: Winter Simulation Conference, 2008, pp. 27–38. [4] W. Farrell, Literature review and bibliography of simulation optimization, in: Winter Simulation Conference, 1977, pp. 117–124. [5] M.S. Meketon, Optimization in simulation: a survey of recent results, in: Winter Simulation Conference, 1987, pp. 58–67. [6] R.O. Bowden, J.D. Hall, Simulation optimization research and development, in: Winter Simulation Conference, 1998, pp. 1693–1698. [7] Y. Carson, A. Maria, Simulation Optimization: methods and applications, in: Winter Simulation Conference, 1997, pp. 118–126. [8] S. Andradóttir, A review of Simulation Optimization Techniques, in: Winter Simulation Conference, 1998, pp. 151–158. [9] F. Azadivar, Simulation optimization methodologies, in: Winter Simulation Conference, 1999, pp. 93–100. [10] F. Glover, J.P. Kelly, M. Laguna, New advances for wedding optimization and simulation, in: Winter Simulation Conference, 1999, pp. 255–260. [11] M.C. Fu, Simulation optimization, in: Winter Simulation Conference, 2001, pp. 53–61. [12] J.R. Swisher, P.D. Hyden, A survey of simulation optimization techniques and procedures, in: Winter Simulation Conference, 2000, pp. 119–128. [13] S. Ólafsson, J. Kim, Simulation optimization, in: Winter Simulation Conference, 2002, pp. 79–84. [14] M.C. Fu, F.W. Glover, Simulation optimization: a review, new developments, and applications, in: Winter Simulation Conferences, 2005, pp. 83–95. [15] M.C. Fu, Ch. Chen, Some topics for simulation optimization, in: Winter Simulation Conference, 2008, pp. 27–38. [16] E. Tekin, I. Sabuncuoglu, Simulation optimization: a comprehensive review on theory and applications, IIE Trans. 36 (2004) 1067–1081. [17] C. Gomes de Souza, R. Garcia, Knowledge diffusion and collaboration networks of life cycle assessment, Int. J. Life Cycle Asses. 16 (2011) 561–568. [18] C. Teddlie, F. Yu, Mixed methods sampling: a typology with examples, J. Mixed Meth. Res. (2007) 77–100. [19] W. de Nooy, A. Mrvar, V. Batagelj, Exploratory Social Network Analysis with Pajek, Cambridge University Press, New York, 2005. [20] T. Kamada, S. Kawai, An algorithm for drawing general undirected graphs, Inform. Process. Lett. 31 (1989) 7–15. [21] E. Yan, Y. Ding, Applying centrality measures to impact analysis: a coauthorship network analysis, J. Am. Soc. Inform. Sci. Technol. 60 (2009) 2107– 2118. [22] J. Kleinberg, S. Suri, E. Tardos, T. Wexler, Strategic network formation with structural holes, ACM SIGecom Exchanges 7 (2008) 1–4.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
Contents lists available at ScienceDirect
Simulation Modelling Practice and Theory journal homepage: www.elsevier.com/locate/simpat
Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains Aida Huerta-Barrientos ⇑, Mayra Elizondo-Cortés, Idalia Flores de la Mota Department of Operations Research, National Autonomous Mexico University, Ciudad Universitaria, 04510 Mexico, DF, Mexico
a r t i c l e
i n f o
Article history: Available online xxxx Keywords: Simulation Optimization Network analysis Scientific collaboration Supply chain
a b s t r a c t In the 1970s, a co-authorship network in the field of Simulation Optimization of supply chains was established, supported by local associations. Then, the development of this network was favored by the foundation of new co-authorships and the consolidation of already existing. The purpose of this study is to analyze the structure, collaboration patterns and the time-evolution of the co-authorship network of Simulation Optimization of supply chains. Data are based upon 202 peer-reviewed contributions published from 1970 to August 2012 in relevant journals indexed in the ISI/Web of Science database and International Conferences. The analysis is conducted using exploratory social network analysis technique. Results indicate that the development of knowledge in Simulation Optimization of supply chains has been carried out mainly by 353 authors from 35 countries. Also, there have been proposed over forty Simulation Optimization methods by different authors however the most usual is response surface methodology, followed by gradient based search method and genetic algorithms. In addition, applications of Simulation Optimization methods and techniques are found mainly in areas as health care, management, transport, airline, telecommunications, aerospace, and financial. Although research in Simulation Optimization of supply chains has received much attention by the simulation community, its application in key industries continues to be still small, limiting its support in decision-making. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction Simulation Optimization is a structured approach that is useful to determine optimal settings for input parameters associated with a simulation model. In this case, the optimality is measured by a (steady-state or transient) function of output variables [1]. As suggested by Fu [2], the general optimization problem consists of finding a setting of controllable parameters that minimizes a given objective function, i.e.
min JðhÞ h2H
ð1Þ
where h e H represents the vector of input variables, J(h) is the objective function, and H is the constraint set, which may be either explicitly or implicitly defined. The assumption in the Simulation Optimization setting is that J(h) is not available directly, but must be estimated via simulation [2]. ⇑ Corresponding author. Address: Faculty of Engineering, Av. Universidad 3000, Ciudad Universitaria, C.P. 04510, Mexico. Tel.: +52 5558419565. E-mail address: [email protected] (A. Huerta-Barrientos). http://dx.doi.org/10.1016/j.simpat.2014.02.007 1569-190X/Ó 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
2
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
The key difficulty in Simulation Optimization involves a trade-off between allocating computational resources for searching h e H versus conducting additional simulation replications for better estimating the performance of current promising solutions [3]. Since the 1970s several Simulation Optimization techniques have been proposed for finding optimal settings for input parameters of a simulation model. Some of these methods have been very sophisticated while others have been naïve just generating several input values at random and running the simulation model at each of these input values. None of Simulation Optimization techniques is sure to work for all problems, but each one has characteristics which make it useful for certain types of problems [4]. Extensive reviews and surveys about Simulation Optimization methods and techniques have been carried out by different authors and can be found in [5–14,3,15,16]. However, at present little attention in the literature has been paid to the co-authorship network of Simulation Optimization and its time-evolution and very little has been written about main applications of Simulation Optimization techniques in industry. Peer-reviewed contributions are considered an important source of information that let us to identify the development of knowledge in a specific field, and pinpoint how research is organized and structured [17]. In this direction, specific research questions addressed by this study are: how do authors of peer-reviewed articles in the field of Simulation Optimization of supply chains collaborate? How is structured the co-authorship network of Simulation Optimization of supply chain? Which Simulation Optimization methods have been proposed by researchers? And last but not least, which industries have been supported by Simulation Optimization methods? In line with this, the aim of this study is to analyze the structure, collaboration patterns and the time-evolution of the co-authorship network of Simulation Optimization of supply chains. In order to achieve this aim 202 peer-reviewed articles were selected using the purposive sampling method, which is defined in Section 2. Based on the information about authors, we developed the co-authorship network and analyze its structural properties and time-evolution using exploratory social network analysis technique. This technique has the advantage of determine at global level structural features of a network, detecting its patterns. Then Simulation Optimization methods were classified based on their application areas. This classification is of relevance because provides guidance for industrials, academics and practitioners in optimization method selection. After that, industries that have been supported by Simulation Optimization methods in making decisions processes were listed. Although theoretical Simulation Optimization methods are numerous, key industrial areas of application are still small. We highlight this gap and recommend how to increase the application of Simulation Optimization methods in real-world problems solutions. The remainder of this paper is organized as follows. In Section 2, the data collection method is presented. The exploratory social network analysis technique is described in Section 3. The co-authorship network of Simulation Optimization of supply chains is characterized in Section 4. The structure and time-evolution of the co-authorship network of Simulation Optimization of supply chains is analyzed in Section 5. Main conclusions and future research are drawn in Section 6.
2. The data collection We collected 202 peer-reviewed contributions in Simulation Optimization of supply chains from relevant journals indexed in the ISI/Web of Science database and International Conferences. These contributions were selected based on purposive sampling method. This method is also referred to as qualitative sampling that involves certain units or cases based on a specific purpose rather than randomly [18]. Three broad categories of purposive sampling techniques are well known: sampling to achieve a representativeness or comparability, sampling special or unique cases, and sequential sampling [18]. In this study, peer-reviewed contributions were sampled based on the first category of the purposive sampling techniques to achieve a representativeness of the application of Simulation Optimization in supply chain field. The criterion to filter a contribution was the inclusion of the phrase ‘‘simulation optimization’’ in its title but with its application in supply chain field. The period of publication taken in account was from 1970s to August 2012. As the first contribution was published in the 1970s, additional queries regarding five periods were placed in data as follows: Period I, 1970–1979; Period II, 1980–1989; Period III, 1990–1999; Period IV, 2000–2009; and Period V, 2010–2012. As it is noted from Table 1, for each period peer-reviewed contributions were quantified as well as International Journals and Conferences, authors and their countries. The tendency in the number of peer-reviewed contributions published over five periods already mentioned, suggests the attracting increasing interest from the simulation community in Simulation Optimization field since the last decades. The distribution of peer-reviewed contributions based on the number of co-authors is summarized in Table 2. Two-authored contributions represent the biggest proportion with 41% of the total. In contrast to this, the single-authored contributions represent only 15%.
Table 1 Dissemination of peer-reviewed contributions of Simulation Optimization of supply chains. Period Years
I 1970–1979
II 1980–1989
III 1990–1999
IV 2000–2009
V 2010–2012
Authors Peer-reviewed contributions Int. Journals/Conference Countries
10 5 1 1
20 15 2 2
32 20 1 2
232 118 23 29
100 44 17 18
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
3
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx Table 2 The distribution of peer-reviewed contributions of Simulation Optimization of supply chains by number of authors. Description
Peer-reviewed contributions
% Peer-reviewed contributions
Two-authored Three-authored Four-authored Single-authored
82 52 37 31
41 26 18 15
Table 3 Journals and international conferences accounting for at least one peer-reviewed contribution in Simulation Optimization of supply chains from 1970 to 2012. Journal/conference
Peer-reviewed contributions
Winter Simulation Conference Simulation Modelling Practice and Theory Int. J. Production Economics Computers and Chemical Engineering Computers and Operations Research European Journal of Operational Research Computers and Industrial Engineering IIE Transactions ACM Transactions on Modeling and Computer Simulation Applied Soft Computing Automatica Engineering Applications of Artificial Intelligence Engineering Optimization European Simulation Symposium Expert Systems with Applications Handbooks in Operations Research and Management Science Industrial and Engineering Chemistry Research INFORMS Journal on Computing INFORMS Simulation Society Research Workshop International conference on Intelligent Systems Modelling and Simulation International conference on Service Systems and Service Management International Journal of Computer Integrated Manufacturing International Journal of Industrial Engineering: Theory Applications and Practice International Journal of Production Economics International Journal of Production Research Irish Journal of Management Journal of Mining Science Mathematics and Computers in simulation Naval Research Logistics OR Spectrum Simulation Practice and Theory Technological and Economic Development of Economy The Arabian Journal for Science and Engineering
154 5 4 3 3 3 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Table 3 indicates 33 journals and international conferences that include at least one peer-reviewed contribution sampled for this study. The top contributor is International Conference Winter Simulation Conference followed by International Journal Simulation Modelling Practice and Theory. Involved international journals and conferences, suggest a multidisciplinary interest in the application of Simulation Optimization methods in supply chain field, as they belong to different industrial areas such as Production, Operations Research, Chemical Engineering, Industrial Engineering, Computer Engineering, Economy, Management, and Logistics. 3. The exploratory social network analysis technique Social networks can be analyzed using different techniques. Important results have been obtained using the exploratory social network analysis technique. Through this technique, it is possible to analyze the time-evolution of a network. Also it is possible to detect and interpret patterns of social ties. In general, the exploratory social network analysis technique consists of four parts [19]: 1. 2. 3. 4.
The The The The
definition of a network; network manipulation; determination of structural features; visual inspection.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
4
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
Fig. 1. The pattern of the co-authorship network of Simulation Optimization of supply chains from 1970 to 2012.
Fig. 2. The pattern of the co-authorship network of Simulation Optimization of supply chains independently of the year from 1970 to 2012.
The definition of a set of vertices and a set of lines where each line connects two vertices is included in the definition of a network. A vertex is considered the smallest unit in a network, while a line is a tie between two vertices. In social network analysis, a vertex represents an actor, and a line represents any social relation. The network manipulation is a very powerful tool that lets us to modify a network according with our requirements. Considering an entire network, by social network analysis technique, its structural features can be determined at global level. And considering a sub network or a single vertex extracted from the entire network, its structural features can be determined at individual level. Additionally, the visual inspection of a network by means of the social network analysis technique facilitates the intuitive understanding of its features, helping us to trace and represent patterns of ties [19]. It is important to stress that visual inspection of a network can be carried out using the Kamada–Kawai free algorithm, well supported by Pajek software. This algorithm is useful for drawing undirected and weighted graphs. The basic idea of this algorithm consists on regarding the desirable ‘‘geometric’’ (Euclidian) distance between two vertices in the drawing as the ‘‘graph theoretic’’ distance between them in the corresponding graph and introducing a virtual dynamic system in which every two vertices are connected by a ‘‘spring’’ of such desirable length. So, the optimal layout of vertices indicates us the state in which the total spring energy of the system is minimal [20]. 4. The co-authorship network of Simulation Optimization of supply chains The co-authorship network of Simulation Optimization of supply chains is represented by authors of peer-reviewed contributions and their interrelations. An author is represented by a vertex, while the interrelations between authors are represented by edges in the co-authorship network. Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
5
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
Fig. 3. The giant component of the co-authorship network of Simulation Optimization of supply chains.
Table 4 Authors who integrate the giant component of the co-authorship network of Simulation Optimization of supply chains.
a
Author
Index
Luo Y. Cho H. Jacobson S. Hyden P. Keng N. Shi L. Carson J. Chen Ch. Healy K. Eren Ultanir A. Bektas T. Ilenda V. Hall J. Better M. Yücesan E. Yoo T. Swisher J. Prudius A. Laguna M. Kleinman N. Kelly J. Jarugumili S. Hutchison D. Hill S. Fu M. Dengiz B. Bowden R. Boesel J. April J. Andradóttir S.
352 348 334 333 267 248 247 246 245 233 232 231 212 202 182 179 160 138 102 91 86 82 79 76 59 48 27 26 15 12
a
The index used to identify authors in the co-authorship network of Simulation Optimization of supply chains.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
6
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
4.1. Trends in the co-authorship network of Simulation Optimization of supply chains The primary constraint on the pattern of the co-authorship network is the number of authors on a contribution. In this direction, the first co-authorship network was defined to detect the total pattern of co-authorships between authors in the timeline from 1970 to 2012, described in Section 2. In this case, each author in 1 year is represented by one vertex. It is important to note that for each contribution the co-authorships were established between first and second author, between first and third author, and between first and fourth author, respectively. Using the Kamada–Kawai free algorithm, the visualization of the co-authorship network of Simulation Optimization of supply chains is showed in Fig. 1. In other case, considering that one vertex represents one and just one author, independently of time on which its contribution was published, the co-authorship network was developed, see Fig. 2. That means that an author is represented by one vertex, independently if the same author has others co-authorships in the timeline considered. To really interpret patterns of co-authorship of Simulation Optimization of supply chains, it was necessary to determine three of the main structural features of the co-authorship network: the giant component, the centralization and structural holes. On one hand, the giant component represents a subnet formed by the largest share of vertices (authors) interconnected within a network. The importance of the giant component lies on the possibility to reach a large number of other authors of the same collaboration network starting by one author of the giant component and moving along its connections. The
Table 5 The top ten centers of the co-authorship network of Simulation Optimization of supply chains correlated with their weighted values and index.
a
Author
Weighted values
Index
Fu M. Biles W. Eskandari H. Yücesan E. Ding H. Hay L. Azadivar F. Chen Y. Othman S. Dengiz B.
40 34 14 13 13 12 8 8 8 7
59 24 56 182 51 72 19 37 129 48
a
The index used to identify authors in the co-authorship network.
Fig. 4. Centers of the co-authorship network of Simulation Optimization of supply chains.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
7
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
appearance of the giant component with 30 authors is verified by black vertices of the co-authorship network presented in Fig. 3. The giant component embraced 8.78% of vertices in the network. It means that it was possible to reach a large number of other authors starting by one of vertices that conforms the giant component and moving along its connections. As the vertices in the co-authorship network were enumerated by its index, we listed in Table 4 the authors included in the giant component based on their index. On another hand, the centrality is one of the main concepts used in network analysis and refers to positions of individual vertices within a network. As is pointed out in [21], the centrality, concept proposed by Freeman, supports the measurement of centralization and can be calculated in terms of the degree to which a vertex falls on the shortest path between others vertices and therefore has a potential for controlling communication in a network. The importance of centrality lies on the idea that information may easily reach central vertices (authors) in a communication network. It is important to note that authors can benefit from serving as intermediaries between others authors who are not directly connected in a network. Through such intermediation, some authors potentially can broke the information flow and synthesize ideas arising in different parts of a network. These principles form the underpinning for structural holes theory. This theory indicate us ways in which some vertices (authors) fill ‘‘holes’’ between groups or sub networks that are not otherwise interacting in a network [22]. Nowadays, two classic centralization measures are applied to characterize a network: closeness and between’s. First, closeness centralization is defined as the variation in the closeness centrality of vertices divided by the maximum variation in closeness centrality. Second, between’s centralization is defined as the variation in the between’s centrality of vertices divided by the maximum variation in between’s centrality. Using Pajek software, it was not possible to obtain measurements of closeness centralization for the co-authorship network of Simulation Optimization of supply chains because this network is not strongly connected. This fact means that there are not paths between all authors in the co-authorship network. In
Fig. 5. Structural holes of the co-authorship network of Simulation Optimization of supply chains.
Table 6 Intermediary authors who are not directly connected in the co-authorship network of Simulation Optimization of supply chains.
a
Author
Indexa
April J. Chang K. Fu M. Glover F. Hong L. Noack D. Syberfeldt A. Wan H. Yücesan E. Ng A. Rose O. Chen Ch. Nelson B. Tiwari M.K.
15 36 59 61 77 127 161 174 182 195 201 246 262 284
The index used to identify authors in the co-authorship network of Simulation Optimization of supply chains.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
8
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
contrast to this, the between’s centralization value was calculated equal to 0.00569. This means that few crucial authors in the co-authorship network of Simulation Optimization of supply chains promote the information transmission through the entire co-authorship network established, i.e. Jacobson S., Fu M., and Andradóttir S. It is important to note that these crucial authors were also elements of the giant component of the same network. Additionally, centers of the co-authorship network in Simulation Optimization of supply chains were identified as vertices with the most weighted values based on the degree of
Fig. 6. The network of author’s countries.
Optimization Techniques Response surface methodology 11% Others 25%
Gradient based search methods 8% Genetic algorithm 8%
Mathematical programming 3% Simulated annealing 6%
Stochastic approximation 6% Tabu search 5% Scatter search 5%
Hooke-Jeeves pattern search method 3% Sample path method 4% Evolution strategy 4% Random search algorithms 4% Heuristics Ranking & selection 4% 4%
Fig. 7. Optimization techniques most usual in peer-reviewed contributions of Simulation Optimization of supply chains.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
9
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
vertices. Normally these weighted values are calculated using a weighted algorithm. Ten centers (authors) with the highest degree of vertices in the co-authorship network of Simulation Optimization of supply chains are listed in Table 5 and showed graphically in Fig. 4, represented by biggest black vertices. Co-authorships that potentially can break the flow of information and synthesize ideas of the co-authorship network of Simulation Optimization of supply chains are relationships represented by lines in grey color in Fig. 5. It is important to note that authors, who represent vertices of these co-authorships (see Table 6), served as intermediaries between others authors who are not directly connected in the co-authorship network.
4.2. The network of author’s countries The network of author’s countries also was built. A vertex of this network represented an author’s country and edges represented the collaboration between authors of different countries. On one hand, we observed the central role of authors from USA in this network and their co-authorships with authors from Germany, China, Chile, Belgium, Turkey, Taiwan, France, Singapore, Mexico, Thailand, Korea, Netherlands, Iran and India, mainly. On another hand, authors from United Arab Emirates have collaborated in the field of Simulation Optimization of supply chains only with authors from Saudi Arabia. The same situation is observed for authors from Ireland who have collaborated just with authors from Austria as is illustrated in Fig. 6.
4.3. Optimization techniques most usual in Simulation Optimization of supply chains Other interesting part of this study is the analysis of optimization techniques used by authors of the co-authorship network of Simulation Optimization of supply chains. We counted over forty different optimization techniques. The most usual in peer-reviewed contributions was the response surface methodology followed by gradient based search methods, genetic algorithms, simulated annealing, stochastic approximation, tabu search, scatter search, and others, as is presented in Fig. 7.
Industries Airline 9%
Pharmaceutical 2%
Aerospace 7%
Sport 2%
Energy 2%
Financial 7%
Telecommunications 11%
Newspaper 2%
SD 2%
Other 22% Mining export 2%
Transport 15%
Textil 2%
Military 2% Management 15%
Health care 15%
Fishery 2%
Chemical 2%
Fig. 8. Industries supported by Simulation Optimization methods.
Table 7 Supply chain and manufacturing activities supported by Simulation Optimization methods. Supply chain
Manufacturing
Design Planning Physical distribution planning Management Scheduling Resources allocation Storage and retrieval policies Flow Shop Logistics Inventory
Production Planning Production line Process plant design Flexible manufacturing Automated manufacturing system Lean manufacturing Robot manufacturing cell Buffer allocation Scheduling Assembly line
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
10
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
4.4. Industries supported by the application of simulation optimization methods A fundamental aspect of Simulation Optimization of supply chains is its application to support the solution of realworld problems. In this direction, industries that have been supported by Simulation Optimization methods and techniques from 1970 to 2012 are showed in Fig. 8. Industries in which these methods have been more applied are: health care, management, transport, telecommunications, airline, aerospace, and financial. Few applications in areas of energy, newspaper, mining export, military, fishery, chemical, textile, sustainable development, sport and pharmaceutical were found. Although Simulation Optimization methods are numerous, as we observed in Section 4.3, industrial areas of application are still small. 4.5. Supply chain areas supported by the application of Simulation Optimization methods and techniques The application of Simulation Optimization methods in supply chain and manufacturing activities has been focalized just in certain activities, see Table 7. Although the technological potential of these techniques, they have not been applied extensively to support the decision-making.
Fig. 9. The co-authorship network of Simulation Optimization of supply chains in Period I (1970–1979).
Fig. 10. The co-authorship network of Simulation Optimization of supply chains in Period II (1980–1989).
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
11
Fig. 11. The co-authorship network of Simulation Optimization of supply chains in Period III (1990–1999).
Fig. 12. The co-authorship network of Simulation Optimization of supply chains in Period IV (2000–2009).
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
12
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
5. The structure and time-evolution of the co-authorship network of Simulation Optimization of supply chains Since the 1970s, the structure of the co-authorship network of Simulation Optimization of supply chains started to be developed based on several local associations, forming little communication groups between its members. In Period I, from 1970 to 1979, five co-authorships were established, see Fig. 9. Optimization techniques mainly used by authors in this period were coordinate search, gradient based search, heuristics, Hooke–Jeeves pattern search, random search, and response surface methodology, in health care and transport industries. In Period II, from 1980–1989, co-authorships were incremented to eight (see Fig. 10) and optimization techniques mainly used were Nelder method, perturbation analysis, response surface methodology, simulated annealing, and stochastic approximation supporting decision problems in the aerospace industry. In Period III, from 1990 to 1999, co-authorships of Simulation Optimization of supply chains were incremented to 17 (see Fig. 11). Optimization techniques used in this period were mathematical programming, Nested partitions method, neural networks, perturbation analysis, random search, ranking and selection, response surface methodology, sample path, scatter search, simulated annealing, stochastic approximation and tabu search, to support the decision-making in airline and transport industries. In Period IV, from 2000 to 2009, co-authorships of Simulation Optimization of supply chains were incremented to 195, see Fig. 12. In this period, authors used optimization techniques such as adaptive partitioning search, approximate dynamic programming, brute force method, coordinate search, evolution strategy, genetic algorithms, golden region search, gradient based search, heuristics, Hooke–Jeeves pattern search, kriging methodology, mathematical programming, multiple comparison, Nested partitions, neural networks, particle swarm optimization, perturbation analysis, random search, ranking and selection, response surface methodology, retrospective approximation, sample average approximation, sample path, scatter search, sequential selection, simulated annealing, stochastic approximation and tabu search. Industries of interest in applying these techniques were aerospace, airline, telecommunications, management, financial, heath care, military, sustainable development, chemical, newspaper, energy, textile, and fishery.
Fig. 13. The co-authorship network of Simulation Optimization of supply chains in Period V (2010–2012).
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
13
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx Table 8 Optimization methods used in Simulation Optimization of supply chains from 1970 to 2012. Optimization method
Period I
Period II
Period III
Period IV
Period V
Gradient based search methods Response surface methodology Hooke–Jeeves pattern search method Heuristics Random search algorithms Simulated annealing Stochastic approximation Genetic algorithm Mathematical programming Nested partitions method Neural networks Perturbation analysis Ranking and selection Sample path method Scatter search Tabu search Frequency domain method Nelder method Evolution strategy Simultaneous Perturbation Stochastic Approximation Kriging methodology Particle swarm optimization Sequential selection Coordinate search Geoffrion and Graves method Discrete stochastic optimization Adaptive Partitioning search Approximate dynamic programming Brute force method Cross-entropy method Estimation of distribution algorithms Golden region search Greedy heuristics Indifference-zone ranking Mixed integer programming Model reference adaptive search Multiple comparison Non-monotonic search Retrospective approximation Sample average approximation Branch -and bound method, Linear programming Optimal computing budget allocation
X X X X X
X X X
X X X X X X X X X X X X X X X X X
X X X X X X X X X X X X X X X X
X X
X X
X X X X X
X X
X
X X
X X X X X X X X X X X X X
X X X
X X X X X X X X X X X X X X X X X X X
Over time, the structure of the co-authorship network of Simulation Optimization of supply chains has been favored mainly by the foundation of new co-authorships and the consolidation of already existing. It is important to stress that in Period V, from 2010 to 2011, a total of 71 co-authorships were established, see Fig. 13. Optimization techniques used by authors in this period were genetic algorithms, gradient based search, heuristics, kriging methodology, mathematical programming, Nelder method, Nested partitions, neural networks, optimal computing budget allocation, particle swarm optimization, random search, ranking and selection, response surface methodology, sample path, scatter search, sequential selection, simulated annealing, stochastic approximation, and tabu search to support the decision-making in aerospace, telecommunications, health care, port, transport, mining export, and pharmaceutical industries. Table 8 presents an overview of optimization techniques used in all five periods (Period I, Period I, Period III, Period IV and Period V). The optimization method most usual is response surface methodology. It is important to outline that response surface methodology has been applied mainly in transport and health industries, in areas of inventory and manufacturing. Also this methodology has been combined with other optimization techniques such as genetic algorithms and Taguchi.
6. Conclusions and future research This study demonstrates that since the 1970s the knowledge dissemination of Simulation Optimization of supply chain has involved a large number of authors from different countries, peer-reviewed contributions, international journals and conferences. We analyzed 202 peer-reviewed contributions in Simulation Optimization of supply chains from relevant journals indexed in the ISI/Web of Science database and International Conferences. From the analysis, we observed that about 85% of peer-reviewed contributions were developed by more than one author. That means that the development of Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007
14
A. Huerta-Barrientos et al. / Simulation Modelling Practice and Theory xxx (2014) xxx–xxx
knowledge in this field is mainly based on co-authorships. The co-authorship network of Simulation Optimization of supply chains was represented by authors of peer-reviewed contributions and their interrelations. The findings in the analysis of the co-authorship network indicate that it was possible to reach a large number of other authors in the network through 30 authors moving along the edges of the subnet that form these authors. Also, few authors were crucial to the transmission of information through the co-authorship network and some of these authors also had the function of centers in the network as Fu M. and Yücesan E. Additionally, we observed that the optimization technique most usual in peer-reviewed contributions analyzed was response surface methodology followed by gradient based search methods, genetic algorithms, simulated annealing, stochastic approximation, tabu search and scatter search. Response surface methodology has not been used alone instead it has been combined with other optimization techniques, incrementing in this way the potential of optimization. Although Simulation Optimization methods are numerous, they have not been applied extensively to support the decision-making in many areas of supply chain and manufacturing. As the structure of the collaboration network between authors has been favored mainly by the foundation of new co-authorships and the consolidation of existing, it can be a good strategy to increment the number of applications of simulation optimization in more supply chain areas based on new coauthorships, collaborating. Future research is needed into co-authorship network analysis based on impact factors of journals indexed in the ISI/Web of Science database. References [1] J.R. Swisher, E. Yücesan, Discrete-event simulation optimization using ranking, selection, and multiple comparison procedures: a survey, ACM Trans. Model. Comput. Simul. 13 (2003) 134–154. [2] M.C. Fu, Feature article: optimization for simulation: theory vs. practice, INFORMS J. Comput. 14 (2002) 192–215. [3] M.C. Fu, Ch. Chen, Some topics for simulation optimization, in: Winter Simulation Conference, 2008, pp. 27–38. [4] W. Farrell, Literature review and bibliography of simulation optimization, in: Winter Simulation Conference, 1977, pp. 117–124. [5] M.S. Meketon, Optimization in simulation: a survey of recent results, in: Winter Simulation Conference, 1987, pp. 58–67. [6] R.O. Bowden, J.D. Hall, Simulation optimization research and development, in: Winter Simulation Conference, 1998, pp. 1693–1698. [7] Y. Carson, A. Maria, Simulation Optimization: methods and applications, in: Winter Simulation Conference, 1997, pp. 118–126. [8] S. Andradóttir, A review of Simulation Optimization Techniques, in: Winter Simulation Conference, 1998, pp. 151–158. [9] F. Azadivar, Simulation optimization methodologies, in: Winter Simulation Conference, 1999, pp. 93–100. [10] F. Glover, J.P. Kelly, M. Laguna, New advances for wedding optimization and simulation, in: Winter Simulation Conference, 1999, pp. 255–260. [11] M.C. Fu, Simulation optimization, in: Winter Simulation Conference, 2001, pp. 53–61. [12] J.R. Swisher, P.D. Hyden, A survey of simulation optimization techniques and procedures, in: Winter Simulation Conference, 2000, pp. 119–128. [13] S. Ólafsson, J. Kim, Simulation optimization, in: Winter Simulation Conference, 2002, pp. 79–84. [14] M.C. Fu, F.W. Glover, Simulation optimization: a review, new developments, and applications, in: Winter Simulation Conferences, 2005, pp. 83–95. [15] M.C. Fu, Ch. Chen, Some topics for simulation optimization, in: Winter Simulation Conference, 2008, pp. 27–38. [16] E. Tekin, I. Sabuncuoglu, Simulation optimization: a comprehensive review on theory and applications, IIE Trans. 36 (2004) 1067–1081. [17] C. Gomes de Souza, R. Garcia, Knowledge diffusion and collaboration networks of life cycle assessment, Int. J. Life Cycle Asses. 16 (2011) 561–568. [18] C. Teddlie, F. Yu, Mixed methods sampling: a typology with examples, J. Mixed Meth. Res. (2007) 77–100. [19] W. de Nooy, A. Mrvar, V. Batagelj, Exploratory Social Network Analysis with Pajek, Cambridge University Press, New York, 2005. [20] T. Kamada, S. Kawai, An algorithm for drawing general undirected graphs, Inform. Process. Lett. 31 (1989) 7–15. [21] E. Yan, Y. Ding, Applying centrality measures to impact analysis: a coauthorship network analysis, J. Am. Soc. Inform. Sci. Technol. 60 (2009) 2107– 2118. [22] J. Kleinberg, S. Suri, E. Tardos, T. Wexler, Strategic network formation with structural holes, ACM SIGecom Exchanges 7 (2008) 1–4.
Please cite this article in press as: A. Huerta-Barrientos et al., Analysis of scientific collaboration patterns in the co-authorship network of Simulation–Optimization of supply chains, Simulat. Modell. Pract. Theory (2014), http://dx.doi.org/10.1016/j.simpat.2014.02.007