Preliminary investigation of the sea-rail intermodal system's efficiency using a simulation approach: case of the Port of Trois-Rivieres

Efficient cargo transportation from the port to its destination is critical due to the high cost associated with transportation activities, such as shipping and port activities. Therefore, it is imperative to carefully consider the transportation system to ensure it is cost-effective and efficient. Thus, an effective and efficient transportation system is required (Putra et al. 2018).

Due to the high frequency and flexibility of the road transportation mode (Bouchery et al. 2020), its market share is steadily growing worldwide (Hanssen et al. 2012). In the port of Trois Rivieres, about 70% of cargo is transported by trucks, and 30% is transported by train; the port plans to increase the train share by improving the current rail network within the port and delimiting the bottleneck in the port. In the context of a climate emergency, moving from trucks to trains is urgent since rail is considered a more sustainable transportation mode (Ng and Talley 2020). Various countries are planning to mitigate congestion and environmental effects, especially from urban traffic, including port-related trucking transportation. For example, Rotterdam Port has planned to reduce its traffic by 20% during city peak periods; the Los Angeles and New York ports also have their own strategy (Fan et al. 2020). Port of Trois Rivieres has an initiative with the collaborative partnership announced in 2022 between the ports of Montreal, Quebec City, and Trois-Rivières, aimed in particular at greening their facilities, operations, and supply chains. It also supports the three ports' efforts to join Canada's government's carbon–neutral challenge. Nowadays, road congestion is a serious problem in urban port cities worldwide, causing major delays in shipments and disturbing the entire supply chain (Selmoune et al. 2020). Recently, many researchers have been studying sea-rail intermodal transportation because of its lower cost and carbon footprint compared to other transportation modes. Many container ports are trying to enhance connectivity between the terminal and hinterland by investing in a rail terminal located at seaports. Efficiently accelerating container flow between container ships and trains is one of the significant problems that must be addressed (Yan et al. 2020). The entrance to and exit from the train area clogs because trains arrive, depart, or are processed on the same tracks. A common solution to this problem would be to increase the facility size. However, in many cases, expansion is not feasible due to a lack of available land near the port (Hdrinc 2021). This challenge represents just one of the many issues the Port of Trois-Rivières faced. Fang (2016) pointed out two main reasons for the lack of smooth sea-rail connections: lack of connecting conditions between railway container yards and ports and incomplete sea-rail intermodal transport information system construction.

Examining the research perspective on sea–rail intermodal transportation, Liu and Wang (2023) conducted a study aimed at assessing the service capacity of port-centric intermodal transhipment hubs. The evaluation was categorised into three dimensions: radiation scale capacity, transportation connection capacity, and resource integration capacity. Utilising the fuzzy matter element method, the service capability of these hubs was evaluated, and the results were quantified through the Euclidean closeness degree. The findings indicate that Tianjin Port possesses the highest service capacity, followed by Ningbo Zhoushan Port. The dedicated rail line mileage is a critical area that requires attention in Ningbo Zhoushan Port and Qingdao Port. Tianjin Port should enhance its container sea–rail transportation volume, while Guangzhou Port and Xiamen Port should focus on improving sea–rail container handling capacity.

The aforementioned studies distinctly underscore the significance of enhancing sea-rail intermodal operations within the terminal to facilitate cargo flow and mitigate the adverse effects of cargo transportation between the port and its hinterland.

Despite their vital role in the global supply chain, ports have a significant negative environmental impact, primarily stemming from terminal operations and activities (Hossain et al. 2019). Consequently, mitigating this impact and enhancing terminal performance sustainability is imperative (Vacca 2010). Reducing the carbon footprint associated with transport logistics at seaports is a complex challenge. Meng (2018) analysed the overall benefits of sea-rail intermodal transportation and the status of combined rail and water transport in containers in China.

The increase in container traffic and environment rules in the recent decade have forced the involved parties to pay more attention to the negative influence on their operational activities. Recently, considerable attention has been paid to the sea-rail intermodal container transportation sector because of its cost and environmental benefits (Ashrafi et al. 2019; Di Vaio and Varriale 2018; Acciaro 2015; Roh et al. 2016; Sislian et al. 2016; Kang and Kim 2017; Langenus and Dooms 2018; Oh et al. 2018).

Sea-rail intermodal transportation is a crucial component of international trade, gaining global recognition because of numerous benefits, including the ability to handle large volumes, low cost, and low energy consumption. Liu (2020) proposed a route optimisation model for container sea-rail intermodal transport by minimising the total cost as the objective function. In the Zhao et al. study (2020), a multi-objective optimisation model was established to reduce the total cost of the transportation process, which meant maximising resource utilisation and ensuring it was environmentally friendly. Most studies conducted in the field of intermodal transportation have primarily emphasised the reduction of operating costs as a primary objective (Ghane-Ezabadi and Vergara 2016; Hanssen et al. 2012; Ishfaq and Sox 2011; G. Chen et al. 2013).

The better environmental performance of sea-rail intermodal versus road transport is often presented as evidence by policymakers to encourage the modal shift to sea-rail intermodal. Because low-carbon emissions and mass capacity are major benefits, sea-rail combined container transportation has been believed to be a promising way to mitigate air pollution and traffic congestion (Yan and Xu 2021). In that regard, the study of Zhang et al. (2021a, b) assesses the environmental benefits of the modal shift of port-connecting freight transportation by increasing the use of sea-rail intermodal in Shenzhen. Abu Aisha et al. (2020) compared GHG emissions produced by different modes of transportation to increase train dependence on transport containers in sea-rail intermodal.

Although sea-rail intermodal is a tremendous opportunity to improve seaport competitiveness, various factors determine their efficiency. Factors affecting the port container sea-rail transportation system, as described in (Chen and Zhang 2021), can be divided into internal and external factors. The interior sea-rail intermodal transport factors compositions in the container port are infrastructure, production equipment, resource scheduling, and transmission service subsystem. In contrast, the external compositions are geographical location, economic level, transport policy changes, and weather characteristics. The impact of most of these factors is notably evident in our case study, which is the port of Trois Rivieres. The system's infrastructure is considerably aged, and the rail network is characterised by numerous intersections, impeding the smooth flow of wagons within the port, prompting a closer examination of its existing infrastructure and potential improvements needed to enhance efficiency.

Additionally, the port's geographical location, surrounded by the city of Trois Rivieres, constrains train movements through the urban area during daylight hours to prevent street blockages. Understanding and addressing these constraints is crucial for enhancing sea-rail intermodal operations within the port. The study by Zhang et al. (2021a, b) demonstrated that logistics facility location and layout can significantly affect seaport connectivity with other transportation modes. The study further has shown that location rationality and logistics facility layout reduce costs by delivering an economy of scale and maximising transportation efficiency and service quality by planning efficient multimodal networks.

Similarly, Naiyu Wang and Wei (2020) summarised factors affecting rail-sea intermodal transportation handling equipment configuration and put forward the principles that should be considered when configuring the equipment. In another research, the model of Zhao et al. (2018) considered many factors, such as transshipping capacity, network capacity, and the importance of containers, to minimise total container hours in the coordination area, reflecting the efficiency of inbound container distribution organisation.

Feng et al. (2014) defined dry bulk's key sea-rail intermodal transport factors. Dynamic system and modelling software VENSIM were used to establish a dynamic system model of sea-rail intermodal transport of dry bulk. The case of sea-rail intermodal transport of dry bulk at Meizhou Bay Port in Fujian was used to simulate and check the model; the result showed that the model was feasible and effective.

Han et al. (2020) analysed vital factors influencing the level of multimodal transportation development from a sea-rail intermodal transport perspective. He concluded that the volume of intermodal container transportation and railway mileage were the core factors affecting the level of multimodal transportation development. The evaluation model results can objectively reflect the level of multimodal transportation development and problems in each city in China.

The main operations of the chain's rail-sea link are train transfers between the railway station and the maritime terminals, train loading and unloading, and the storage management of goods in dedicated yards. Trains approaching the area may sometimes have to split into cars depending on specific ports; these cars are then transferred to destination terminals (and vice versa for the import cycle).

Grishin et al. (2022) addressed the problem of optimising cargo transfer from ships to trains at the sea-rail terminal in Russia, aiming to reduce the overall delivery time to the destination and minimise the cost associated with train formation. Two models were formulated for this purpose: a binary model and an integer model, both of which were compared. These models were executed using the Gurobi optimiser.

Yan et al. (2020a, b) conducted a study on the sea-rail transhipment operation problem regarding seaport rail terminals, which included two key sub-problems involved in sea-rail intermodal container transportation: namely, train schedule templates and inbound container transhipment plans. In another study, Yan et al. (2020a) investigated transhipment operations between vessels and trains in seaport rail terminals. Results showed that handling capacity significantly affected transfer plan performance, and an increase in the storage cost of import containers led to a more effective transhipment plan. Pingping et al. (2013) analysed the challenges of Ningbo sea-rail combined transport in China. These challenges included connecting the port with the hinterland and expanding the expected hinterland of rail-sea intermodal transportation in Ningbo. Canadian ports respond to record cargo volume growth through major expansions and far-reaching transformation of their facilities (Zatylny 2020). One example of investigating rail-seaport intermodal issues in a Canadian port is studied by Gillen et al. (2018), demonstrating that enhancing train services and increasing train frequency is the best way to improve port performance.

Although the principle of dry port is not new, its implementation and proficient operations have been only recently exploited and implemented. In this respect, Borruso et al. (2023) conducted an analytical study to highlight the importance of railway connections between the seaport and its hinterland, particularly in terms of the reconstruction of the rail links among the Port of Trieste and its major inland destinations. However, it is essential to emphasise that the research has focused almost exclusively on the geographical aspects of evaluation.

Various authors have investigated numerous strategies to improve sea-rail intermodal efficiency. For example, Abu Aisha et al. (2020) suggested changing the container terminal layout and connecting it to a dry port with a rail track to enhance sea-rail intermodal efficiency. This suggestion was approved according to results by Tadić et al. (2021). The study concluded that the best intermodal transportation system development scenario referred to establishing dry port terminals for Danube River ports with improved network connectivity between terminals via rail transportation mode. Another strategy to improve sea-rail intermodal, suggested by Li and Ye (2009), was to develop a management information system for railway-sea intermodal transportation. The synthetic logistic information platform, including information on railway-sea intermodal transportation, should be established to execute information sharing among the railway, port, customs clearance, three inspections, intermodal transport companies, ship-owners, and consigners. Jarašūnienė and Čižiūnienė (2021) covered the need for applying information systems in the field of maritime and rail transport. Insights from these studies could be beneficial in developing strategies for the Port of Trois-Rivieres that promote efficient cargo transfer, minimise overall delivery times, and reduce associated costs.

Many container ports worldwide are trying to enhance connectivity between railway transportation and the shipping area by investing in rail terminals located at seaports. Due to the multi-management of combined railway and port operations, a series of problems may cause low operation efficiency. While the studies in the literature review provide a broader understanding of sea-rail intermodal systems, applying these insights to the specific context of the Port of Trois-Rivieres would require a detailed analysis of the port's current situation, operational challenges, and goals. It serves as a valuable resource for identifying relevant factors and potential strategies that could contribute to improving sea-rail intermodal operations in Trois-Rivières.

Although many studies have been carried out to assess and improve port operation efficiency, most research has emphasised the container terminal. However, non-containerised cargo has specific characteristics and needs to be transported in general cargo ships or bulk carriers. Despite the prevalence of general cargo ports handling raw materials, there is a notable absence of research investigating these specific types of ports. The Port of Trois-Rivières, situated in Quebec, is one of the general cargo ports, and it is a case study of this research. This study aims to fill the research gap and contribute valuable insights that can benefit ports throughout Quebec. Particularly, these insights can support the alignment of Quebec's ports with the provincial government's plan to achieve a substantial 37.5% reduction in greenhouse gas emissions below 1990 levels by 2030 (Québec 2022).

Most of the previous research focused on addressing the impact of particular factors on the performance of the system within the container terminals, and mathematical models were widely used to solve these problems. However, the current study focuses on general cargo ports, and complexity is characteristic of these types of ports. Since handled cargo in general cargo ports is different in shape, condition, and size, the operations in general cargo ports are more critical and complex than those in container ports. Therefore, the simulation approach will be the proper tool for the current study to solve problems in the complicated system of the port of Trois Rivieres. Leveraging a simulation model, we aim to discern factors contributing to bottlenecks and propose viable enhancements for the intricate system.

Based on the literature review, no comprehensive research has been conducted to analyse how wagons operate within ports and increase rail share in any ports. Therefore, this study aims to examine operations of moving wagons between rail tracks and handling points in an actual general cargo port to identify improvement points and analyse their impact on the port operation using discrete event simulation in the ARENA simulation software, version 16.0. In this study, one of the most active general cargo ports along Canada's St. Lawrence River has been selected to be analysed in railway cargo functioning. Although the Port of Trois-Rivieres has a railway system infrastructure and the benefits of using this transportation method are clear, its railway performance seems to have potential for improvement.

Three sources of information were employed during this research to understand all cargo operations and wagon movements in the port. The first source was field visits to the Port of Trois-Rivieres in July, August 2022, and February 2023. During each visit, all operations were investigated and scheduled, including loading and unloading cargo, transferring cargo to storage facilities, loading trucks and wagons, and sending wagons to the train yard in the port to leave the port.

During the port visits and sharing files with the port authority, we obtained data files such as maps and port layout, historical data of ship arrival and associated data, as well as types and amounts of cargo. Other data was related to trains, such as arrival and departure times and dates, as well as amounts and types of cargo. Some of the data was taken from the port's statistical reports. Additionally, many meetings were scheduled with the port authority to discuss the format and required data and clarify some port operations. Some meetings took place at the port, and others were held via Zoom.

Moreover, security cameras installed in the port were used to observe the time to transport wagons between the train yard and the loading point. The recorded video was meticulously transferred to the laboratory for in-depth analysis and time study. Given the exploratory nature of our study, as stated in the title, we had limited observations from video records to determine the time it takes to transport wagons between the train yard and the loading point. Therefore, we decided to model this transportation time using the normal distribution. This choice was influenced by findings from our literature review, which showed that transportation and transhipment times often follow this distribution pattern (Xu et al. 2021). The accuracy and consistency of data were verified using the base scenario.

The Port of Trois-Rivieres is the case study used in this work. This port is strategically located halfway between Montreal and Quebec City and has been an active Canadian port since 1882. The port plays a crucial role in local, national, and international economic development, particularly in major industrial sectors such as aluminium, forestry, and agri-food. The Port of Trois-Rivieres is a general cargo port that receives more than 200 ships from over 100 different ports in more than 40 countries worldwide. The port has ten berths with a length of 1458 m. The port also receives 11,000 wagons and 55,000 trucks annually. In all, more than 3.5 million metric tons are handled by the port. The Port of Trois-Rivieres is a medium-sized urban port specialising in the storage and multimodal transport of a wide variety of bulk products and general cargo. We visited the port many times to build a model of the Port of Trois Rivieres; the port layout is shown in Fig. 1.

Once the ship arrives at the port and is berthed, ship unloading operations to the storage facility or ship loading from the storage facility begin. The ship leaves the port after ending these operations. Trucks and trains transport cargo between the port and the hinterland. Trucks arrive at the port based on the availability of cargo that needs to be picked up by trucks. The train entry has been restricted to only 7 a.m. and 11 p.m. to minimise potential disruptions to public roads within the city.

The train arrives at the port vertically with the berths and runs parallel to the berth, as shown in Fig. 1. The wagons stop in the train yard located inside the port. From the train yard, wagons are sent to storage facilities to load or unload the cargo in groups of wagons.

Since a general cargo port is a complex system consisting of many subsystems and various overlapping operations that undoubtedly affect outputs, a model that simulates such a system could provide a significant analytical benefit. A complex, large-scale, discrete event-based simulation model was developed to implement and validate the developed framework. The model was created after a comprehensive understanding of the actual system and all operations related to handling and transporting the cargo in the Port of Trois-Rivières. While the simulation code is not included in the supplementary materials, it is available upon request for researchers. We modelled wagons and cargo flow movement, starting from the ship's arrival at berth to unload or load its cargo and the truck and train arrival to load or unload the cargo and leave the port, as shown in Fig. 2. Since this study is a preliminary investigation, the model of this study does not consider seasonality in cargo flow; it will be our focus of attention in the next port project.

Entities that move through the simulation model include ships, cargo, trucks, and wagons. Resources include six berths, two loaders to load/unload the ship, one loader to load/unload trucks, one loader to load/unload wagons, and a storage facility. The processes are created to represent port operations for loading and unloading cargo and moving wagons in the terminal. In this context, four key system parameters were selected based on their significant impacts on the system performance: (1) ship arrivals, (2) train arrivals, (3) truck arrivals, and (4) the amount of cargo, in addition to (1) time to move wagons from train yard to storage facility and (2) time to load and unload wagons. Including these parameters reflects their critical roles in evaluating port congestion, resource utilisation, and overall cargo flow efficiency. The initial values of ship, train, and truck arrivals are zero, but during the simulation run, these values are subject to change based on real data, which is correlated to the day and time of the year. We assumed a consistent load of 15,000 tons per ship for cargo volume, derived from the average cargo capacity observed in ships visiting the port. The initial values for the time required to move wagons from the train yard and for loading and unloading wagons are set at zero. Still, they are susceptible to real-data adjustments throughout the simulation run. Table 1 presents justifications to clarify the rationale behind the parameter selection and to emphasise their relevance to real-world port operations.

The cargo operation flow chart in the port is illustrated in Fig. 2. With the help of this chart, the model was used to test various scenarios to accelerate cargo flow between the port and the hinterland.

A discrete simulation model was developed using the SIMAN simulation language and implemented through the ARENA software application. We used this software to structure the conceptual and simulation model for the port layout, including ship arrivals, cargo operations, truck movement inside the port, and train operations. The objective of the model is to reflect the system's functioning and assess its performance to discover bottlenecks that negatively affect port capacity and assess various scenarios to improve port efficiency.

System performance consists of calculating the number of trucks, wagons, ships, and the amount of cargo transported by many modes of transportation to discover bottlenecks in the system. These can be defined based on the average number of wagons in the system, the average waiting time to enter the port, and the amount of cargo remaining in the storage facility that needs to be transported out of the port or loaded on the ship for export.

In order to analyse and evaluate system performance, we conducted a variety of simulation tests aimed at demonstrating bottlenecks in the system. The model has three components: ship movement and activity, truck movement, and wagon movement and activity in the terminal. Each component has many operations, and each operation performs a specific task or event in the system (for example, ship arrival and departure, truck arrival and departure, truck loading/unloading, train arrival and departure, wagon grouping and separation, transport, loading, and unloading cargo). The study will focus on analysing the movement of the wagons in the terminal.

The simulation model considered the Port of Trois-Rivieres business hours, i.e., 8 a.m. to 5 p.m., Monday to Friday. In other words, loading and unloading operations and truck movement to and from the port take place from Monday to Friday from 8 a.m. to 5 p.m., whereas the trains arrive and leave the port at 7 a.m. and 11 p.m. However, the simulation model was developed to model generic port operations as well as ship, truck, and train movement in the port.

Because of uncertainties such as random truck, train, and ship delays, separate simulation replications were needed to determine the necessary time for the system to reach its steady state. If we consider only a single replication, the results may be influenced by specific or unusual events within the system, such as the arrival of trains, trucks, or ships, as well as loading/unloading and time of sending wagons to storage facility. However, these events are stochastic in nature, meaning they are subject to randomness and variability. To ensure the accuracy and generalizability of the results over multiple years, it is essential to conduct multiple replications of the simulation. By running numerous replications, we can better capture the variability inherent in the system and obtain a more robust understanding of its behavior. Consequently, the duration of simulation runs was set to 365 days. On average, each run takes 2.65 min on a computer with a 2.00 GHz CPU. In 2019, the port handled 217 ships, 159 imported ships, and 58 exported ships. The port can receive 150 trucks and handle 40 wagons daily, but the number and arrival of trucks and trains are based on ship arrival. The processing time to load and unload modes of transportation was calculated based on the number of each mode the port can handle per day, considering weekends and business hours.

After implementing the system in ARENA software, many steps are conducted to assess and validate the simulation model's accuracy, including model operation monitoring, animation displays, and debug features in the simulation software. Model operation monitoring involves observing and analysing a simulation model's performance while running to ensure that it behaves as expected, represents the real-world system, and identifies any issues. Animation displays visually represent the simulation model, allowing you to observe the system's dynamic behaviour over time. As the simulation ran, we observed the movement of entities through the system and the utilisation of resources in real-time. In addition, we observed whether resources were appropriately assigned, whether queues were behaving as expected, and whether resources were idle or busy as per the model logic to ensure that entities moved through the simulation model according to the defined logic and that there were no unexpected delays, blockages, or other issues. Also, the ARENA software's debug features were used to monitor the value of specific variables, and the breakpoints feature was used to inspect the model and variables at a given breakpoint. Therefore, This helped identify the cause of unexpected behaviour. System bottlenecks appeared after the simulation run ended. The block diagram representation of the simulation model is illustrated in Fig. 2.