1 Introduction

The progress in simulation modelling has been consistent with the growth of information technology and the increasing demand for quick and reliable solutions to complex problems. This progress was also harnessed by the port industry in response to the mounting pressure for improved shipping operations in general and effective integration of shipping into the logistics of freight in particular. It is interesting to note that Steer and Page (1961) and Beattie et al. (1971) were two pioneering studies based on the Montecode package specifically for Monte Carlo statistics (produced by the British Iron and Steel Research Association). Subsequently, Lawrence (1973) and Borovits and Ein-Dor (1975) developed port simulations in FORTRAN addressing the emerging challenges of containerisation. By the end of the 1970s, David and Collier (1979) simulated the operation of container handling equipment (fork-lifts and cranes) in conjunction with a ship’s turnaround time in order to study the optimisation of cargo handling rates. Therefore, through the advances in computing including the first introduction of microcomputers during the 1970s, port simulation models experienced a step improvement and by the end of the 1980s they became sector and operation specific which inevitably facilitated the rapid expansion of simulation modelling in port development. During the last decade, the accelerated use of simulation models (SM) in port development and particularly in container terminals (CT) presents a clear demonstration of its appeal towards optimised port development, whether this involves the establishment of new or the expansion of existing facilities and services. This impetus not only justifies the treatment of port SM as a distinct research area, but also dictates the need to produce a clearer picture of the relevant knowledge base in order to promote its effective utilisation by all stakeholders and especially amongst researchers.

A total of 226 papers are analyzed here. Among these, 209 papers present a simulation model of a port or container terminal operation; seven papers are literature reviews that are not included in Table 1 in Appendix; and 10 papers are literature reviews that are included in Appendix Table 1. Thus, a total of 219 papers are listed in Table 1 indicating the number of journal papers in our database.

The bottom of Table 1 in Appendix shows a list of ten literature review papers (Vis and de Koster 2003; Steenken et al. 2004; Günther and Kim 2006; Stahlbock and Voss 2008; Angeloudis and Bell 2011; Luo et al. 2011; Rashidi and Tsang 2013; Carlo et al. 2014a, b; Carlo et al. 2015) published between 2003 and 2015 which cover a wide range of operations research (OR) techniques applied to CTs. They generally offer a concise description for each of the reviewed items, whilst the work by Angeloudis and Bell (2011) extends into the presentation of the empirical aspect of 30 individual SM applications. In addition to these, there are seven other literature reviews that are not listed in Table 1 since they consider related topics. Bierwirth and Meisel (2010) presented a first survey of berth allocation and quay crane scheduling problems in CTs up to 2009 and in that sense they provided an update of their paper in Bierwirth and Meisel (2015). The problem of loading and unloading container stacks as is faced in yard operations and stowage planning was investigated in Lehnfeld and Knust (2014). In building the multidisciplinary picture of port research, the literature reviews by Pallis et al. (2010, 2011) on port economics, policy and management over 12 years (1997–2008) are most informative, whilst those by Woo et al. (2011, 2012) present a decadal analysis of trends and themes since the 1980s.

Amongst the previous review papers, the most extensive was the work conducted by Angeloudis and Bell (2011) which included 30 journal papers and that by Rashidi and Tsang (2013) with an even shorter reference of papers. Both papers considered limited number of simulation studies that investigate the impact of management decisions on the performance of CTs.

Besides the use of the extensive database of 219 papers, the major contribution of the current review stems from the development and presentation of several tables and figures which enable the reader to obtain a concise and thorough knowledge of this research. This entails: (1) the journals in which these papers have been published; (2) the type of software that was used to develop the SMs; (3) the main features of the SMs; (4) the specific port (if any) that inspired the development of the SMs, and (5) the general topic (application area) investigated.

The current paper is the first comprehensive review article in this area, since the classification of published SM research according to areas of application and tools employed provides a comprehensive presentation of the structure of the attained knowledge base, so far. In this manner, the evaluation of the current research literature ultimately facilitates the identification of gaps in the simulation modelling of ports amidst the trend of port operations with increasing complexity, range and interdependence.

With regard to the structure of the paper, the next section delivers a concise presentation of port operations. The third section presents the selection and classification of the reviewed literature, followed by a detailed analysis and discussion on the review results, whilst the final section presents the conclusions. With regard to the terminology, the terms are used in full on their first reference in the text followed by their acronym or abbreviation in brackets, whilst their explanation for the remainder of the text may also be obtained through the list of terms shown in Appendix Table 2.

2 Background of port operations

The significance of the application of SM in ports is relevant to their pivotal role in supporting international seaborne trade. With particular reference to the fastest growing shipping sector of containerised freight, the main function of a port system is to improve intermodality through securing the seamless transfer of containers within the port, as well as between the sea and land modes of transport. A port represents a complex system consisting of different types of terminals with highly dynamic interactions between the various handling, transportation, and storage units. The port process can be divided into the following principal links with the specification of port operations, summarized in Fig. 1: anchorage-ship-berth link with the loading/unloading stage of ships, an internal transport link for moving cargo from apron area to storage area and vice versa, cargo storage and receiving/delivery operations from/to external vehicles (EVs) (shore-side link). SM is applied to these links either separately or complementary to each other. The most explored area of simulated port operations refers to the anchorage-ship-berth link. The handling processes at a CT includes: ship operations, receiving/delivery operations from/to EVs and container handling and storage operations on the container yard (CY). Loaded and unloaded containers are temporarily stored awaiting a new journey. Inbound containers arrive by ship and quay cranes (QC) transfer containers to a yard truck or automated guided vehicle (AGV) [or buffer space for straddle carriers (SC), shuttle carriers and automated lifting vehicles (ALV)]. The transfer equipment then delivers the inbound container to a yard crane (YC) which picks it off the yard truck or AGV (or from CY buffer space) and stacks it into the respective storage location, which moves back to the quay crane to receive the next unloaded container. The storage location is given by row, bay and tier within the block and is assigned in real time upon arrival of the container in the terminal. Inbound as well as outbound containers are stocked in a CY which is divided into a number of blocks. Each block consists of the container stacks which are reserved for reefer containers or to store hazardous goods among others. For the loading operation, the process is reversed. This is an indirect transfer system (ITS) in which container handling equipment delivers a container between the apron area and the CY. Rubber-tired or rail-mounted gantry cranes store the containers at yard stacks in CY. In direct transfer system (DTS), no YCs are needed because SC is used to pick up (put down) containers from (into) the CY, deliver it to (from) the apron, and transfer it to (from) a quay crane. European CTs are based upon the DTS in which transport of containers from ship to stack and vice versa and stacking (retrieval) of containers into (from) the slots assigned in the CY are performed by SCs. It is pure SC systems or SC terminals. DTS requires the larger area due to dedicated lanes required for SCs to access slot position, while the ITS minimizes yard area requirements (Stahlbock and Voss 2008).

Fig. 1
figure 1

Port system with main subsystems

Full size image

According to the different types of handling equipment two main categories with containers being stacked on top of each other by either SCs or YCs are common in Europe and Asia. A third type which is used quite often in North America, is an on-chassis system where every containers are stacked on a chassis and a tractor pulls the chassis between the apron and the CY.

CTs are greatly differing by the type of transfer and handling equipment used. Improving CTs operations and attaining handling processes excellence to gain competitive advantage has attracted a lot of attention in the last few years. Many CT operations and many of the technological advances in port like Decision Support Systems, AGVs, ALVs, Automated Stacking Cranes, Global Positioning Systems (GPS), differential GPS, Radio Frequency Identification, Real-time Location System, Information Technologies, QC double cycling, indented container berth, dual-hoist tandem QC, dual-hoist triple tandem QC and automated CT admit modelling via SMs. In this paper, the literature available on the application of SMs in port is reviewed.

3 Review methodology

The literature database is shown in Table 1 of the Appendix and contains 219 journal papers. Among these, 209 papers present a SM of a port or CT operation and 10 are review papers. Their temporal distribution covers 32 papers from 1961 to 1999, which are substantially lower than the 187 published papers in the period of 2000–2015. Enriching and extending the work of the ten previous reviews (of Table 1), the current paper also constitutes a follow-up review of individual SM applications in ports and CTs from 1961 to 2015. The total number of papers in the previous ten reviews that contain the application SM in ports was equal to 65 and all of them were also included in the current review.

In order to collect all relevant publications of port or container terminal SMs, an extensive literature search was conducted in the online-resources of the publishers Inderscience, Springer, Elsevier, ASCE, SAGE, Hindawi, Taylor & Francis, Palgrave and scientific database Google Scholar. The keywords used for our search were port, port operation and container terminal from which we only selected the ones addressing the following items: simulation, simulation model, simulation–optimisation, simulation-based optimisation, simulation-based heuristic method and agent-based system or model. Altogether, this search produced the database of 219 papers which have been published in 71 international scientific journals, i.e. on average 3.08 papers in each journal. In total, the authorship of these papers was shared by 614 researchers. The average number of authors per paper was 2.8, whilst 32 (14.6 %) of the papers were written by a single author, whilst the maximum number of authors in a single paper was six. For each paper, the affiliation of the first author was located as follows: Europe 115 papers, Asia 49 papers, North America 42 papers, Australia seven papers, Africa three papers and South America three papers. In total, the first authorship of the papers under the current review was distributed in 37 different countries, namely 35 papers from the United States of America, 22 from Netherlands, 20 from Italy, 18 from China, 12 from South Korea, 10 from Germany, 10 from Turkey, 7 from Australia, 6 from Greece, 6 from Singapore, 5 from Spain, whilst the first author of the 68 remaining papers was located in 26 other countries.

In order to improve the reliability of the literature review process towards fulfilling its objectives, the full text of each paper was screened and included in this review after meeting the following criteria: (1) exclusive coverage of SM applications in ports; (2) does not contain SM with reference to navigation, maritime environment and conditions, geographical space considerations and port uses other than shipping e.g. fishing and (3) is not a “grey” literature item (i.e. textbook, conference paper, monograph, doctoral dissertation and book chapter). Furthermore, it should be noted that although the papers of the WinterSim Proceedings constitute an important outlet of simulation papers which covers applications in various ports (http://informs-sim.org/), the current review was focused on papers mainly (by almost 90 %) published in SCI or SSCI Journals. The aforementioned criteria are considered appropriate as they offer the ability to surpass the exemplary approach in reliably reviewing the evolution of major advances, significant gaps, current debates and emerging challenges on this topic.

Port SM literature has been published in journals covering various research fields ranging from generic informatics (e.g. computer science) to transportation and sector specific (i.e. maritime or shipping) and port SM literature finds application in various areas ranging from port project planning to management (incl. performance evaluation of terminal services and operations). Therefore, with regard to the SM application area, each paper was classified according to its general description (conveyed through the title of paper, abstract, keywords), its detailed content (based on its literature review, methodology and results) and its research contribution (outlined in conclusions). Furthermore, taking into account that SM involves computer programming written in a general purpose language or in a special simulation-oriented language, the reviewed literature was analysed on the basis of the tools employed.

By adopting a review approach which focuses on the identification of research fields, application areas, simulation tools and other critical features, a systematic assessment of the research literature on port SM was ensured, whilst its citation analysis captured the impact of the various tools in pushing the research agenda forward and the level of coherence achieved amongst the various application areas. Finally, the review results were summarised according to:

  • the journals in which these papers have been published,

  • the area of application to port operations,

  • the port simulation model statements by application area in respect to importance, contribution, benefits and output results of SMs,

  • the specific port (if any) that inspired the development of the SMs,

  • the most widely used simulation tools and SM features, and

  • the main features of the SMs.

4 Review results and discussion

4.1 Analysis by research field

The distribution of the reviewed literature according to the journals in which it was published is presented in Fig. 2. Furthermore, a research field classification was produced through grouping journals on the basis of their theme (i.e. computer Science, OR, maritime transport, transportation and other research fields).

Fig. 2
figure 2

Distribution of port SM research literature according to journal field

Full size image

As shown, all collected port SM papers appeared in journals of different research fields; 31 journals included 179 papers, each containing two papers at least, whilst the remaining 40 papers were published in one journal. According to further analysis, it was found that 18.7 % of port SM papers were associated with the computer science journals, 20.1 % were published in OR journals, 16.0 % of the papers in maritime journals and 12.3 % of the papers were included in journals specialising in the field of transportation research. The remaining 32.9 % of port SM papers were attached to other research field journals.

4.2 Analysis by application area

The application of SM in ports was found to be very concentrated in the CT sector represented by the highest number of papers (166 papers), followed by applications to ports in general (24 papers), port traffic (15 papers) and bulk cargo terminals (14 papers).

As shown in Fig. 3, all papers were classified into four port SM application areas: bulk terminal operations (6.4 %), ports in general (11 %), port traffic (6.8 %) and CT operations (75.8 %). It can be seen that SM has been used to solve a problem (whether real-life or hypothetical) in over 72.2 % of the papers, whereas the remaining 27.8 % have explored methodological issues. Further analysis of the papers reviewed per port application area is presented in Table 1 of the Appendix.

Fig. 3
figure 3

Distribution of port SM research literature according to application area

Full size image

The predominance of SM applications in the CT sector has been manifested during the last decade through its rapid expansion in specific CT operations. In this context, through the current review it was possible to identify 11 subareas of SM application, i.e. intermodal operations (18 papers), automated operations (21 papers), performance evaluation (33 papers), comparison of analytical models (AM) and SM (10 papers), transfer and storage equipment (23 papers), storage policies at CY (6 papers), logistics planning (16 papers), integration of simulation and optimisation models (OM) (21 papers), operational policies (OP) (4 papers), education and training (E&T) (4 papers), and literature review (10 papers).

Figure 4 presents a schematic classification of the application of SM within the CT sector and provides an indication of the duration of research engagement in each area through reference to the first and last relevant paper.

Fig. 4
figure 4

Classification of literature on SM at CT

Full size image

Table 1 of the Appendix presents the main results of the research literature review, showing the application area, the specific focus (port system) and the employed tools of the SMs, as well as other basic features. These are highly consistent with the results of previous review papers, indicating that many SMs are increasingly incorporating knowledge of port operations, modelling and analysis into port system to provide users the capability of intelligent assistance. Very often, computations are performed using various types of AM to generate solution while SM components into port systems provide necessary data, monitoring the calculation process, and evaluating AM results to assist users on a real time basis.

The distribution of the reviewed literature produced during the pre- and post-2000 period according to SM application areas and subareas is shown in Fig. 5. The majority of papers published between 1961 and 2000 addressed bulk operations and ports in general by 50 and 41.6 %, respectively. In contrast, this earlier period was poor with regard to the consideration of stacking and operational policies, education and training, whilst no literature review was delivered. During 2001–2015, most papers were related to performance evaluation (84.8 % per subarea), automated operations (95.2 % per subarea) and transfer and storage equipment (95.6 % per subarea) while the minimum number of published papers were in relation to operational policies, as well as education and training (four papers per subarea).

Fig. 5
figure 5

Distribution of publications according to their application

Full size image

Figure 5 reveals an increase in SM application with regard to real-life port issues, compared to the findings reported in the previous reviews. This is attributed to the improved dissemination of the accumulated knowledge and experience gained in port SM, whilst the demand for prompt and reliable solutions to several port problems is increasing rapidly driven by the stronger competition within the sector. Unlike the past, most ports recently build databases through recording and analysing massive data, from transaction processing systems to support port resource assignment and planning, which provide essential support to SM utilisation.

Furthermore, Fig. 6 shows the annual distribution of papers by SM application area and subarea, according to which with an average of seven applications per year a total of 201 port SMs were published during the period from 1995 to 2015. This is a remarkable increase compared to the preceding period of 1961–1994 which was associated with less than one application per year. During 2011–2015, 17.2 publications were recorded annually in comparison to 11.4, 7.6 and 3.8 for 2006–2010, 2001–2005 and 1996–2000, respectively. Therefore, it is evident that with the passage of time the application areas of port SM have diversified mostly into CT subareas (23 papers in 2015) with the predominant subareas being integration of SM and OM, transfer and storage equipment and performance evaluation. Analyses show that the number of journal papers published per year on simulation modeling in seaports and seaport container terminals continues to increase on an annual basis with a peak of 25 papers in 2015.

Fig. 6
figure 6

Distribution of SM papers by year of publication

Full size image

4.3 Analysis by simulation tool

The list of simulation languages and tools that have been used for modelling port processes and operations are presented in Table 1 of the Appendix. In a synoptic form, Fig. 7 shows a plethora of employed simulation tools per port application area.

Fig. 7
figure 7

Distribution of employed simulation tools per port application area

Full size image

There are 26 simulation tools used at least twice whilst other 29 are used only once. Approximately 21.4 % of port SMs are developed in ARENA, which reveals the user’s animation attempts to realistically represent object movement in terms of speed, direction, orientation and interaction with other objects. About 9.6 % of port SMs are built in C and C++, 5.9 % in FORTRAN, 6.4 % in Java, 5.9 % in MATLAB, 3.75 % in AweSim, 3.2 % in (Flexsim, Witness) separately, 2.1 % (PORTSIM, Agent Based) for each of them, 1.6 % in (MODSIM, SLAM, GPSS/H, VISCOT, Monte Carlo, Visual Basic, eM-Plant) for each of them and 1.1 % (Pascal, Taylor II, Siman, PortModel, iThink & Micro World (MW), MUST, SimPort, AnyLogic and NetLogo), respectively. The remaining 15.5 % of SMs are developed in other simulation languages as already shown in Fig. 7.

To distinguish the presence of simulation tools in all application areas, a classification of main elements of SMs was developed which is defined by six binary variables A, B, C, D, E and F (see Appendix Table 1—Column Main features of SMs). Each binary variable served as an indicator of whether an SM possesses a particular feature. For example, A = 1 indicates that the SM has a flow diagram. Otherwise, the SM does not possess a flow diagram. These binary variables are explained below:

Variable

Value

Interpretation

A

0

There is no flow diagram of the SM

1

There is flow diagram of the SM

B

0

There is no screen or code of the models and submodels

1

There is screen or code of the models and submodels

C

0

There is no port and terminal layout

1

There is port and terminal layout

D

0

There is no simulation graphics and animations

1

There is simulation graphics and animations

E

0

There is no integration among terminal subsystems

1

There is integration among terminal subsystems

F

0

There is no experimental strategy and sensitivity analysis

1

There is experimental strategy and sensitivity analysis

The tuple (A, B, C, D, E and F) was used to describe features of simulations processes for each paper and complete the application of SMs. Moreover, defined binary variables were sufficient to describe a physical layout of ports and their operational features.

5 Conclusions

Port SM has made meaningful progress over the past five decades and has strengthened its impetus in port and terminal research, despite the increasing prominence of other modelling techniques based on mixed integer programming, various heuristics and so on. Based upon the statistics of the 219 reviewed items, it was found that during 2000–2015 the number of papers on port SM literature is almost sixfold in comparison to the pre-2000 period, showing an 85.4 and 14.6 % distribution, respectively. By virtue of its ability to produce prompt and reliable solutions to a multitude of port related problems, SM has been a very important and popular decision support tool for port developers, whilst its effectiveness has been mostly demonstrated in the area of port operations.

The results of this port SM literature review are summarised as follows:

  • Five research fields have been identified, with the predominance of OR with 44 papers published in six journals.

  • Four areas and 11 subareas of application have been identified, with the predominance of port performance evaluation and container terminals, respectively.

  • The observed application diversification of port SM demonstrates its wide-ranging capabilities.

Discrete-event simulation remains one of the most popular techniques in port operations modelling (more than 20 % SMs were built in ARENA), despite the introduction of new techniques such as agent based modelling, network based modelling, simulation-based education, web-based simulation and so on.

The current literature review provides evidence that future port SM will continue to follow the trend of developing a general simulation platform, broadening simulation into new domains of applications and integrating with other simulation approaches. This trend would undoubtedly fulfil the need for future port SM to be concentrated in developing tools and techniques which would be able to meet the emerging challenges in the management of port development and operation. Amongst these challenges the most important is that of improving the influence of ports and particularly of container terminals in the optimisation of the freight transport chain.