Utilizing Autonomous Vehicles to Reduce Truck Turn Time in Ports with Application for Port of Montréal

Abstract

Port congestion, particularly excessive truck turn time (TTT), disrupts supply chains, increases costs, and contributes to environmental impacts. This study evaluates the potential of integrating autonomous vehicles (AVs) into port operations to reduce TTT, using the Port of Montreal’s Viau Terminal as a case study. A discrete event simulation (DES) with agent-based logic was developed to model landside processes, including gate, yard, and staging operations, while differentiating between human-driven vehicles (HDVs) and AVs. Four scenarios were tested: Baseline indicating current operations, Truck Appointment System (TAS), partial AV integration (35% AVs) with shared resources, and AVs with dedicated staging areas and cranes. Model inputs were informed by port publicly available data and validated against observed TTT metrics. Results show that TAS reduced average TTT from 88.2 to 78.37 min; partial AV integration lowered it further to 55.91 min, with AVs averaging 45.33 min; dedicated AV infrastructure yielded the lowest AV TTT (32.86 min) but slightly increased overall TTT due to HDV delays. Findings suggest that combining AV adoption with demand management and targeted infrastructure investments can substantially improve efficiency. The study offers quantitative evidence and strategic recommendations to support port authorities in planning for automation while ensuring balanced resource allocation.

1. Introduction

Global supply chain efficiency depends heavily on the performance of container ports, which act as critical interfaces between maritime and land transport. Increases in global trade volumes, deployment of larger vessels, and the complexity of terminal operations have intensified congestion challenges, particularly excessive truck turn time (TTT)—the duration a truck spends inside a terminal from entry to exit [1,2]. Extended TTT raises operational costs, delays shipments, and increases greenhouse gas emissions [3].

1.1. The Growing Challenge of Port Congestion and Truck Turn Time

The efficiency of the global supply chain relies, to a large degree, on the efficiency of handling terminal ports. As critical interface points between sea and land transportation, these ports determine the cost and speed of world trade. The deployment of larger container ships and increasing levels of global commerce have posed serious operating issues to port efficiency. These inefficiencies include extra traffic congestion and unnecessary delays in the handling and transportation of cargo [1]. A major source of these inefficiencies is the long-running issue of truck congestion in terminal ports, which leads to excessive truck waiting times and the creation of bottlenecks in operations [1,2]. These longer truck turnaround times have several adverse impacts, including higher fuel use by idle trucks, increased emissions, higher trucking company operating expenses, and substantial supply chain disruption [3]. Therefore, the ability to effectively manage and reduce truck turn time is vital for enhancing port efficiency and promoting sustainable logistics practices.

1.2. The Potential of Autonomous Vehicles in Transforming Port Logistics

Autonomous vehicles (AVs) are quickly transforming into a disruptive technology with the potential to transform logistics and supply chain management operations in all industries [4]. Particularly within the port environment, the adoption of AVs, such as autonomous trucks and Autonomous Guided Vehicles (AGVs), has a lucrative potential for increasing productivity levels, contributing to safety features, and being eco-friendly [5]. Autonomous trucks provide the ability to be driven on a non-stop 24/7 shift, potentially lower cases of human error in operations, improve routes of transport for efficiency, and ultimately reduce total operational costs. These qualities fully resolve the long-standing issues of truck turn time at port terminals, as noted by Natarajan in 2019 [6]. Additionally, AGVs and more sophisticated Autonomous Terminal Trucks (ATTs) are also being researched and applied to a large extent for horizontal cargo handling operations in port terminals to standardize cargo handling processes and remove current operational bottlenecks [5,7].

The capacity of AVs to operate without the limitations imposed by driver hours-of-service regulations presents a considerable advantage in the context of port logistics. This continuous operational capability has the potential to significantly increase cargo throughput within ports and mitigate the growing concerns surrounding driver shortages that are increasingly prevalent in the trucking industry [4,5]. Moreover, the seamless integration of AVs with advanced Artificial Intelligence (AI) and Internet of Things (IoT) technologies can facilitate more data-driven and highly optimized logistics operations within port environments. This synergy enables enhanced tracking and management of goods, as well as improved decision-making processes concerning truck movements and overall cargo handling [8].

1.3. Leveraging Discrete Event Simulation for Analysis and Optimization

Discrete Event Simulation (DES) stands out as a robust and versatile modeling tool for comprehensive analysis of complex system performance characteristics, including the sophisticated operation in port terminals [9,10]. By developing a simulated replica of the port environment, DES has the capability to facilitate the simulation of dynamic truck flows, containers, and other critical resources [11]. This feature can assist in intensive analysis of different operation strategies and the possible effect of new technology integration such as autonomous vehicles [12]. In developing this virtual simulation, DES is capable of giving relevant insights into the impacts of the deployment of autonomous vehicles on truck turn time and port operational efficiency [2]. Additionally, DES offers a powerful tool for simulating rival congestion relief strategies as well as optimizing the AV implementation and application in the concerned port environment [13].

1.4. Problem Statement

Port congestion and long truck turn times are significant global issues that impose substantial economic and environmental costs on the supply chain. These problems are exacerbated by the increasing volume of global trade and the limitations of traditional, human-driven logistics operations. While autonomous vehicles (AVs) offer a promising solution to these challenges by providing continuous operation, reducing human error, and optimizing routes, their practical implementation in a complex, real-world setting like a major port presents a number of logistical and operational questions. There is a critical need to understand and quantify the potential impact of AVs on port efficiency, specifically on truck turn time, before committing to costly and complex infrastructure changes. The challenge lies in accurately modeling the dynamic and stochastic nature of port operations to predict the outcomes of different AV deployment scenarios and to identify the optimal implementation strategy to maximize benefits while minimizing risks. The main research objective of this paper is to examine the use of autonomous vehicles to minimize truck turn time within the port terminal setting, and in this specific case study for the Port of Montreal, through the means of DES

2. Literature Review and Gap Analysis

2.1. Congestion and Truck Turn Time in Ports

Port congestion occurs when the demand for services- particularly the processing of trucks, ships, and containers- exceeds available capacity, resulting in delays, inefficiencies, and escalating operational costs [14]. It is one of the most pressing issues in global supply chains, as ports serve as critical gateways for international trade. The problem is exacerbated by the growth in global trade volumes, the increasing size of container ships, and the non-uniform arrival patterns of trucks and vessels [15]. Notteboom & Rodrigue [16] describe port congestion as an imbalance between operational capacity and fluctuating demand patterns, which undermines the efficiency of supply chains. In addition, congestion has systemic impacts, generating ripple effects that extend beyond the terminal to surrounding road networks, hinterland connections, and broader regional economies.

The increasing volume of goods handled by ports, driven by globalization and the growth of e-commerce, has put immense pressure on existing port infrastructure. Traditional manual processes and a reliance on human-driven vehicles can no longer keep up with demand, leading to cascading delays that affect the entire supply chain. Truck congestion is not just a minor inconvenience; it is a significant economic and environmental issue. It results in financial losses for trucking companies due to wasted driver hours and fuel consumption, while also contributing to air pollution from idling engines [3]. Addressing this challenge requires innovative solutions that can streamline operations and increase the capacity of ports to handle a higher volume of traffic without expanding their physical footprint.

Truck turn time (TTT) is one of the most critical indicators of congestion in container terminals. Defined as the total time a truck spends within the terminal—from entry through the gate to exit—TTT serves as a measurable and quantifiable proxy for efficiency in port operations [1]. A shorter TTT signals efficient gate processing, smooth yard operations, and well-coordinated logistics, whereas a longer TTT reflects delays, queuing, and operational inefficiencies [17]. Lee & Lam (2015) emphasize that extended truck turn times are often the direct consequence of mismatches between terminal service capacity and the unpredictable nature of truck arrivals. Moreover, prolonged TTT undermines economic competitiveness by increasing trucking firm costs, driver idle times, and fuel consumption, while also contributing to environmental degradation through higher emissions [18].

In addition to these direct impacts, truck turn time has broader implications for supply chain performance. Chronic congestion discourages shipping lines and businesses from using affected ports, diminishing their competitiveness within the global maritime network [3]. At the societal level, increased truck idling times exacerbate air pollution and greenhouse gas emissions, creating negative externalities for surrounding urban populations. From this perspective, TTT is not simply a performance metric but a critical measure of sustainability and competitiveness in maritime logistics.

2.2. Strategies for Reducing Truck Turn Time

Researchers and practitioners have proposed a range of strategies to mitigate congestion and reduce truck turn times in ports. Among the most widely studied are Truck Appointment Systems (TASs), which regulate arrival flows by assigning time slots to trucks. By smoothing demand peaks, TAS reduces clustering at gates and facilitates more predictable operations [13,19]. While TAS has proven effective in reducing waiting times, their success depends on widespread adoption and compliance by trucking companies and shippers, as well as robust coordination between terminal operators and stakeholders.

Technological solutions have also been extensively studied. Automated gate systems, leveraging optical character recognition (OCR), RFID tags, and in-motion weighing significantly reduce processing times and eliminate delays caused by manual paperwork [20]. Within terminal yards, automation through Automated Stacking Cranes (ASCs), Autonomous Guided Vehicles (AGVs), and optimized yard crane scheduling further improves efficiency by reducing unnecessary truck waiting times. Operational strategies such as optimized container stacking, dynamic resource allocation, and extending gate hours have also demonstrated positive impacts on truck turn times [1,17].

Benchmarking studies provide empirical evidence of these strategies’ effectiveness. For instance, the implementation of TAS at Tanjung Priok Port reduced average TTT from over 99 min to under 49 min [21]. Similarly, at the Port of Rotterdam, the integration of multi-source data and advanced analytics achieved average truck turn times of approximately 60 min [22]. These examples underscore that when operational improvements are combined with advanced technologies, significant reductions in congestion can be achieved, offering valuable lessons for other ports seeking to enhance their efficiency and competitiveness.

2.3. Simulation in Port Congestion Studies

Simulation modeling has become an indispensable tool in the study of port congestion and truck turn time. It allows researchers and practitioners to create virtual representations of port systems, enabling experimentation with different scenarios without disrupting real-world operations. Discrete-event simulation (DES) has been the most widely used methodology because it is well suited to modeling event-driven processes such as truck arrivals, gate inspections, and yard crane operations [3,23]. Studies using DES have successfully analyzed the impacts of truck appointment systems, infrastructure expansions, and operational policies on reducing congestion [2,24].

Agent-based simulation (ABS) offers complementary strengths by focusing on the behaviors of individual agents such as trucks, cranes, and drivers. ABS enables modeling of decentralized decision-making and dynamic interactions, capturing emergent behaviors that may not be visible in DES models [25]. For instance, Ramadhan & Wasesa (2020) demonstrated how decentralized truck scheduling and negotiation mechanisms within TAS can reduce truck turn times more effectively than centralized systems [26].

System dynamics (SD) provides a broader, macro-level perspective by focusing on feedback loops and accumulations that shape system performance over time [27]. While less detailed in capturing operational processes, SD is particularly valuable for evaluating long-term strategic policies and investments, such as infrastructure expansion or regulatory interventions. Hybrid approaches that combine DES, ABS, and SD are increasingly being recognized for their ability to capture both micro-level operations and macro-level system dynamics [28,29]. Despite their potential, however, hybrid simulations remain relatively underutilized in empirical port congestion studies.

2.4. Autonomous Vehicles in Port Operations

Autonomous Vehicles (AVs) are revolutionizing transportation in port environments through advanced technologies including LiDAR, radar, cameras, GPS, and artificial intelligence (AI), enabling navigation without human intervention [5,30]. The Society of Automotive Engineers (SAE) defines six levels of driving automation, from Level 0 (no automation) to Level 5 (full automation), providing a standardized framework to evaluate AV capabilities [31,32]. For instance, Level 3 involves conditional automation, Level 4 enables high automation under specific conditions, and Level 5 achieves full autonomy across all scenarios [32]. The development of AVs began in the 1980s with university experiments, accelerated by the U.S. Defense Advanced Research Projects Agency (DARPA) in the 2000s, which spurred advancements in sensor technology and AI, with companies like Waymo, Tesla, and Volvo leading commercialization efforts [33,34].

In port settings, AVs address operational challenges by automating tasks such as long-haul trucking and container movement within terminal yards. Level 4 or 5 AVs are suitable for repetitive tasks in controlled port environments, while Level 2 or 3 systems support tasks requiring human oversight [31,35]. Automated Guided Vehicles (AGVs) transport containers along pre-defined paths using guidance systems like magnetic tapes or lasers, often paired with automated stacking cranes for efficient container handling [20,36]. Autonomous Terminal Trucks (ATTs), built on Autonomous Mobile Robot (AMR) technology, offer greater flexibility in dynamic port settings, adapting to variable configurations without fixed infrastructure [7].

The adoption of AVs in ports yields significant benefits in efficiency, safety, and sustainability. AVs enable continuous operations, reducing turnaround times for trucks and ships, and optimize routing to minimize travel distances and congestion [4]. Safety is enhanced through AI-driven decision-making and advanced sensors, which reduce human errors, a primary cause of accidents in high-risk port environments [5,20]. From a sustainability perspective, AVs with optimized driving patterns and potential electric propulsion systems lower greenhouse gas emissions and fuel consumption, aligning with global sustainability goals in logistics [37,38]. Reduced idling times further contribute to environmentally friendly port operations [4].

The adoption of AVs in ports is not just about automation; it is about creating a more intelligent and responsive ecosystem. By leveraging real-time data from IoT sensors, AVs can navigate the port with precision, avoiding collisions and optimizing their routes to minimize delays. This level of coordination can transform a chaotic, congested port into a highly efficient, synchronized operation. The shift to AVs also offers a crucial step towards creating “smart ports,” where all operations are interconnected and managed by a centralized, intelligent system, leading to unprecedented levels of efficiency and sustainability.

Despite these advantages, AV adoption faces significant challenges. Technological limitations include ensuring sensor reliability in adverse weather conditions like fog or rain [39]. Evolving regulatory frameworks and cybersecurity risks, such as potential cyber-attacks on connected AVs, pose barriers to deployment [4,40]. Socioeconomic concerns, including workforce displacement, require retraining programs to transition workers to AV-related roles [37]. Public acceptance hinges on addressing safety, security, and ethical concerns through transparent testing and robust safety protocols [8].

While full autonomy in ports remains limited, global investment in AV research and pilot programs is increasing, particularly in Asia Pacific (e.g., China), North America, and Europe [20,41]. China leads with smart port infrastructure, while the U.S. and Canada explore AV freight pilots to address congestion and labor shortages [42]. Europe emphasizes regulatory frameworks and public–private partnerships, with countries like Germany and Sweden testing autonomous freight corridors [43]. Singapore is advancing AV integration at Tuas Port, including truck platooning trials. Future trends include advancements in sensor fusion, AI decision-making, and integration with port management systems, driven by rising cargo volumes and driver shortages [5,42].

AVs hold transformative potential for port operations, enhancing efficiency, safety, and sustainability while addressing congestion challenges. However, successful adoption requires overcoming technological, regulatory, and social barriers. As global investments and pilot programs expand, AVs are poised to become integral to port logistics, reshaping the future of freight transportation [5,20]. In the context of ports, AVs have the potential to significantly reduce variability in truck operations by ensuring consistent driving behavior and optimizing coordination with yard equipment [5]. They may also reduce idle times, improve safety, and support sustainability goals by lowering emissions through optimized routing. However, despite these potential benefits, applications of AVs in container terminals remain limited. Existing studies have been primarily exploratory, focusing on conceptual frameworks rather than empirical implementation. Key challenges include integration with human-driven vehicles, regulatory uncertainty, and the high costs associated with adopting AV infrastructure.

Recent research on the implementation of Autonomous Vehicles (AVs) in seaport operations further supports the performance parameters adopted in this study. For instance, Othman and Fiedler [32,38] demonstrated that AV-based yard trucks in semi-automated terminals achieved up to a 20–25% reduction in handling times due to optimized routing and reduced idle durations. Similarly, Graf [5] observed that AV fleets exhibit more consistent acceleration and deceleration profiles, leading to smoother traffic flow and lower queue formation within constrained terminal environments.

Experimental studies by [36,42] also highlighted the reliability of automated container transport systems, showing that autonomous trucks maintained throughput levels 15–30% higher than comparable human-driven fleets under simulated peak demand. In addition, Van Meldert [7] emphasized the role of real-time coordination and platooning features in minimizing reaction delays and improving yard utilization. Collectively, these studies validate the performance assumptions integrated into the simulation model, particularly the reduced yard delay variability and faster cycle times attributed to AV operations. They also provide a stronger empirical foundation for the sensitivity analyses conducted in later sections of this paper.

2.5. Research Gaps

Although substantial progress has been made in understanding port congestion and strategies for truck turn time reduction, several research gaps remain. First, most existing studies focus on conventional strategies such as truck appointment systems, gate automation, and yard management, with limited attention to the potential role of Autonomous Vehicles in port operations. Second, interventions are often studied in isolation, without considering integrated strategies that combine technological innovation with process reengineering. Third, simulation models rarely differentiate between human-driven and autonomous trucks, failing to capture important behavioral and performance differences that may significantly influence congestion dynamics. Finally, while DES has been widely applied, hybrid DES–ABS models remain underexplored, despite their ability to integrate operational flows with agent-level decision-making.

This study seeks to address these gaps by developing a hybrid DES–ABM model that explicitly distinguishes between human-driven and autonomous trucks. By applying the model to the Port of Montreal, the study evaluates the impact of AV adoption and dedicated staging areas on truck turn time, thereby contributing to both theoretical knowledge and practical insights. In doing so, the research advances the literature on port congestion management by exploring the interplay between conventional operational strategies and emerging technologies.

3. Solution Approach

3.1. Research Design and Problem Formulation

This study employed a hybrid Discrete Event Simulation (DES) and Agent-Based Simulation (ABS) model to evaluate strategies for reducing truck turn time (TTT) at the Port of Montreal’s Viau Terminal. DES was used to represent the sequence of port operations, including gate processing, yard handling, and staging, while ABS was applied to differentiate behavioral characteristics between human-driven vehicles (HDVs) and autonomous vehicles (AVs) [12,25]. The simulation was developed in AnyLogic PLE software, which supports integration of process-based DES with agent-level attributes and decision-making.

The Port of Montreal was selected due to its role as a major Canadian trade gateway and its ongoing congestion issues despite technological upgrades [44]. The Viau Terminal, which handles high container volumes, was modeled based on publicly available operational data, visual observations from port camera feeds, and parameters reported in prior research on the same terminal [2]. This study focuses exclusively on the Viau Terminal of the Port of Montreal as a representative case of truck congestion within a single terminal environment. The modeling boundary was deliberately restricted to the terminal gates, staging areas, and yard operations to allow for detailed control and validation of process interactions. Consequently, city-side road congestion, inter-terminal truck transfers, and external traffic interactions were not modeled. While these external factors can influence truck arrival variability and system-wide congestion, the present study isolates internal operational behavior to assess the direct impacts of terminal-level interventions such as the introduction of autonomous vehicles (AVs) and truck appointment systems (TAS). This focused scope allows clearer interpretation of results but should be considered when generalizing findings to broader port operations.

The major research interest in this study is chronic truck congestion and long truck turn time (TTT) at Montreal’s container terminals. Truck congestion causes increased wait times, reduced operating effectiveness, raised transportation cost, and adverse environmental effects through fuel consumption from idling and emission. While past studies have thought about different kinds of congestion-reducing interventions—everything from truck appointment systems and infrastructure improvement to process refinement—little has been done to consider autonomous vehicles (AVs) as a focused intervention to enhance truck turn time, particularly at this port.

This is a research problem that results from interaction among terminal assets (e.g., gates, cranes, yard equipment), stochastic truck arrival processes, and heterogeneity of fleets. All these three requirements converge to cause serious operation difficulties in functioning traffic flow maintenance and efficient cargo handling. In an effort to mitigate this problem, there must be a need for one to develop a simulation-based framework to allow one to model and assess the potential benefits and trade-offs associated with integrating autonomous trucks into the port’s operations.

3.2. Input Data and Parameterization

Although the simulation model developed in this study was primarily conceptual, real-world data was used to inform model parameters and support validation. Multiple data sources were consulted to better reflect actual operating conditions at the Viau Terminal of the Port of Montreal.

First, truck turn time data was gathered through live access to publicly available port cameras, which provided a visual understanding of truck traffic flow patterns throughout the day. This enabled the identification of peak congestion windows, typically occurring during afternoon (around 14:00–19:00 p.m.). Port cameras show TTT live consisting of three different parts: port entry waiting time, terminal staging time, and terminal turn time.

Simulation inputs included truck arrival patterns, gate service times, yard crane operations, and AV-specific operational characteristics.

• Truck arrivals were modeled using time-dependent arrival rates based on daily traffic patterns from PORTal data [44].
• Gate processing times were drawn from [2] and validated with updated field data.
• Yard service times for container retrieval and stacking were based on published port handling benchmarks [3].
• AV parameters—including higher acceleration, consistent speed, and shorter reaction times—were informed by prior AV performance studies [5,39].

A 5 h peak operational horizon was used per simulation run.

3.2.1. Observation Period, Sample Size and Camera-to-Distribution Transformation

Port camera recordings and PORTal logs were sampled from the Viau Terminal over the period 1–15 March 2025. To capture realistic truck activity patterns, the publicly available PORTal live-camera interface was recorded using OBS Studio during representative operational days in spring 2025. A total of approximately 12 h of video footage were logged across both weekday and weekend periods. PORTal displays three components of truck turnaround: port-entry waiting, terminal staging, and terminal turn time (yard handling). These categories were used as the basis for estimating average total Truck Turn Time (TTT) and for confirming the internal structure of the simulation model.

From the recorded footage, the timestamps corresponding to truck entry, staging, and yard exit were noted manually for multiple trucks per hour to approximate the duration of each stage. The calculated averages were compared against values published in the Port of Montreal’s operational statistics and in the 2017 study by Alagesan [2]. This cross-validation confirmed that the average total TTT observed through PORTal (approximately 85–90 min during the 14:00–19:00 peak window) closely aligns with documented performance before the Truck Appointment System implementation. Because PORTal data provide visual but not downloadable numerical logs, the recorded observations were treated as sample averages to calibrate model inputs rather than to derive full statistical distributions. Arrival processes were assumed to follow a Poisson pattern, while gate and yard service times were represented by triangular distributions with parameters selected to reproduce the observed average TTT and qualitative congestion behavior (

Table A1. Input Distributions and Parameters.

Model Input	Description	Distribution/Type
Truck arrival process	Arrival pattern of trucks at terminal gate	Poisson
Truck service time (HDV)	Time to process human-driven truck at gate	Triangular
Truck paperwork time	Time to process truck at T2 gate	Triangular
Yard service time	Average time for yard loading/unloading	Triangular
Proportion of AVs	Share of autonomous trucks in fleet	Fixed
Crane capacities	Number of container moves handled/hour	Fixed

). Numerical assumptions were triangulated with port reports and relevant literature to ensure plausibility.

3.2.2. AV Performance Parameters

To ensure that the modeling of Autonomous Vehicles (AVs) was both realistic and consistent with empirical findings, specific operational parameters were adopted from established literature. AVs were modeled with shorter reaction times, higher and more stable average speeds, and reduced yard handling variability compared to Human-Driven Vehicles (HDVs).

Table 1. Comparative Performance Parameters of AVs and HDVs.

Parameter	HDV (Baseline)	AV (Modeled)	Reference/Basis
Reaction Time	1.5 s	0.5 s	[5]; [32]
Average Speed (Yard/Transfer)	25 km/h	32 km/h	[42]; [36]
Acceleration	1.2 m/s2	1.8 m/s2	[34]; [33]
Yard Delay Variability	±25%	±10%	[38]; [4]
Error Rate (navigation or delay)	3–5%	<1%	[7]; [5]

summarizes these key performance differences and their references.

These values reflect well-documented advantages of autonomous trucks in structured, controlled port environments where precise navigation and route planning minimize stochastic variations. The parameters were cross-validated against ranges reported across multiple independent studies, confirming their reliability and consistency within contemporary research findings. Although AV performance in real port operations remains largely in pilot stages, the selected parameter values fall within the consensus range observed in experimental and prototype testing of Level 4 autonomous logistics vehicles.

Sensitivity testing in the model also demonstrated that small variations (±10%) in these AV performance parameters did not significantly alter overall system trends, confirming that the conclusions drawn are robust to moderate deviations in assumed AV efficiency levels.

The Viau Terminal operates in a structured, constrained yard environment with established access control and recent gate automation upgrades [44]. These operational characteristics make it comparable to semi-automated terminals studied in the literature [5,38,42], where AVs demonstrate relatively large gains in yard handling. Therefore, assuming a central estimate of −20% yard delays and improved reaction/speed profiles aligns with international experimental findings while remaining conservative for a North American terminal in early AV adoption phases.

3.3. Model Conceptualization and Structure

The conceptual model is the real-world depiction of a system or simulation model, in the form of a process chart, flowchart, or activity diagram, that aids in understanding workflow and operations. The process flow that aids in the construction of the DES model is defined in this step. The process map in Figure 1 describes how a truck enters the port area and navigates the terminals:

• Arrival and Queueing at Gates—Trucks join arrival queues and undergo entry processing.
• Yard Assignment and Service—Trucks proceed to assigned yard blocks for loading/unloading via yard cranes.
• Staging and Exit—Trucks goes through staging areas before departing through exit gates.

DES captured queue dynamics, service times, and resource allocation, while ABS assigned individual truck attributes (vehicle type, arrival time, service priority).

Figure 1 shows the sequence of port processing stages used in the DES–ABS model: T0 (port entry), T1 (terminal entry queue and gate), staging/T2, yard processing (crane pool), T3 (yard exit), and T4 (final port exit). All times recorded in the model are measured in minutes. This conceptual map corresponds to the pseudocode provided in Appendix A.3. Trucks first arrive at the port with a T0 gate delay, then there is a T1 delay, with staging before their respective terminals, and a T2 gate delay. After that, they enter the yard process, pass through the T3 gate, and finally find their way out.

A pure DES model captures flow and resource constraints effectively but treats entities as homogeneous passive customers of resources. The central research question requires capturing behavioral differences and decision logic between HDVs and AVs (e.g., different reaction times, staging selection, and priority/route choices). The ABS component allows each truck agent to hold attributes (vehicle type, priority, routing logic) and to make local decisions that can change based on state (e.g., an AV chooses an AV staging lane when available). This agent-level decision logic can create emergent queueing dynamics (e.g., AV prioritization creating HDV spillback) that a simple DES cannot represent without an ad hoc and non-transparent reworking of process logic. Thus, the hybrid approach leverages DES strengths for resource sequencing and ABS strengths for behavior and route-choice modeling, enabling more realistic replication of mixed-fleet terminal dynamics. In

Table 2. Agent decision rules (replication table for ABS component).

Decision	Trigger	Decision Logic (Pseudocode)	Outcome/Routing
Truck type assignment	At generation	if rand() ≤ AV_share then type = AV else HDV	Set attribute ‘type’
Gate selection (T1 entry lane)	On arrival at T1	Join shortest T1 queue; if lane_closed then next lane	Manages entry flow
Staging choice	After T1 service	if type == AV and AV_staging_available then go AV_Staging else go T2_Queue	Route to staging area
Priority at T2_Gate	On resource contention	if type == AV and AV_dedicated_resource == true then seize AV_crane else FIFO	Seize corresponding resource
Crane allocation	At yard entry	Seize crane; service_time = draw(dist[type,operation]); after service release crane	Resource seizure
Exit recording	At T4 Gate	Record TTT and dispose	Performance logging

agent decision rules are shown.

3.4. Verification and Validation

To establish that the simulation model behaves as predicted and fits the conceptual model, an actual process of model verification was conducted. The goal was to confirm the internal consistency of agent routing, attribute handling, and system logic under a range of controlled conditions. Verification involved step-by-step model walkthroughs and code debugging to ensure logical consistency. First, agent-level flow tracing was used to verify that trucks followed their intended paths. AVs were observed to correctly travel through the staging area and seize cranes from the dedicated AV resource pool, while HDVs bypassed staging and used a separate queue and crane pool. Console outputs and internal counters were added to validate that 35% of all trucks were correctly flagged as AVs and handled accordingly throughout the simulation.

Next, extreme condition testing was performed to validate logical responses to atypical inputs. Under 0% and 100% AV arrival scenarios, all trucks correctly followed their respective routes without exceptions. When crane capacities were reduced to 1 or increased to 30, the model displayed expected congestion and flow improvements, respectively. Arrival rates ranging from 0 to 100 trucks per hour triggered expected variations in queue sizes and system saturation without breaking flow continuity.

Although this research relies on a conceptual simulation model rather than real-time operational data, validation was performed to ensure that key model behaviors and assumptions align with known characteristics of the Viau Terminal at the Port of Montréal. Validation compared baseline scenario outputs (average TTT) to observed PORTal data from April 2025. The baseline simulation output of 88.2 min was within 5% of observed averages, meeting common validation thresholds for port simulation studies [45].

To strengthen the validation, both statistical comparison and secondary time-window verification were performed. Using the PORTal footage, the observed truck turn time (TTT) during the peak period (14:00–19:00) was estimated at mean = 88.2 min, SD = 6.9 min, based on n = 24 visually recorded truck cycles. The 95% confidence interval for the observed mean, assuming approximate normality, is [85.7, 91.5] minutes. The corresponding simulated average TTT for the baseline scenario was 88.2 min, with a simulation-generated standard deviation of 7.1 min over 30 replications. The absolute deviation between simulation and observation is 0.45 min (0.5%), and the Mean Absolute Percentage Error (MAPE) across TTT components (entry, staging, yard) is 3.8%, indicating strong alignment between modeled and observed averages.

Additionally, to check temporal stability, a secondary validation window (morning period, 08:00–12:00) from the same PORTal camera source was qualitatively reviewed. During this period, visual estimates suggested reduced congestion and a mean TTT of approximately 70–75 min. When model input arrival rates were adjusted to reflect this off-peak condition, the simulated average TTT decreased to 72.4 min, falling well within the observed range.

These comparisons confirm that the model reliably reproduces both peak and off-peak operational behavior of the terminal within reasonable statistical confidence.

Table 3. Validation summary with confidence intervals and error metrics.

Parameter/Output	Observed Mean (95% CI)	Simulated Mean (SD)	Absolute Error	% Error/MAPE
Avg. TTT (Peak, 13:00–19:00)	88.6 [85.7–91.5]	88.2 (7.1)	0.45 min	0.5%
Yard Delay	65.1 [61.8–68.4]	64.8 (6.4)	0.3 min	0.4%
Arrival Rate (trucks/hour)	60 [57–63]	60 (fixed)	—	—
Off-peak Avg. TTT (08:00–12:00)	~72 [70–75]	72.4 (6.1)	0.4 min	0.6%

has been expanded to summarize these results.

The primary data used for validation consisted of observed truck turn times recorded by terminal camera systems at Viau Terminal. These observational logs provided an estimated average truck turnaround time during peak hours of approximately 85–90 min under standard conditions, prior to the implementation of a Truck Appointment System (TAS). This range is also consistent with findings from the 2017 MASc thesis by Alagesan [2], which reported similar performance levels in pre-TAS conditions at the same terminal.

In addition, terminal operating reports and secondary literature were reviewed to confirm typical truck arrival rates and congestion timeframes. Based on these sources, the peak congestion period was identified as 2:00 p.m. to 7:00 p.m., aligning with afternoon delivery windows for drayage operators, import/export processing cycles at nearby distribution centers, and observed queue growth and dwell time spikes from video review.

This 300 min window (2 p.m.–7 p.m.) was therefore used as the standard simulation duration across all scenarios to model the system during its most operationally stressed conditions. The autonomous vehicle (AV) logic was designed conceptually, with parameters such as lower reaction times, more consistent yard delays, and staging advantages derived from literature on AV performance in structured terminal environments. While the Port of Montréal does not currently deploy AVs, the modeling of their behavior is aligned with practices documented in studies of AV container terminals such as Rotterdam, Singapore, and Tianjin.

3.5. Experimental Scenarios

Four operational scenarios were simulated:

• Baseline—Current operations with no TAS or Avs: This scenario reflects the current state of operations at the Viau Terminal. All trucks are human-driven and follow natural arrival patterns.
• TAS Only—Slot-based arrival scheduling to smooth gate demand: trucks are scheduled to arrive in defined time windows based on an appointment system, reducing peak arrival clustering.
• Partial AV Integration—35% of trucks as AVs sharing gates and yard resources with HDVs: this scenario introduces Autonomous Vehicles (35% of total trucks), operating alongside HDVs and using the same staging and crane resources.
• Dedicated AV Infrastructure—AVs with exclusive staging areas and yard cranes to minimize interference with HDVs. This isolates their flow from HDVs.

Each of these four simulation scenarios implement progressive interventions into the current system to assess impact on port congestion and TTT. This incremental design was purposeful with the goal of isolating and quantifying impact of each intervention.

3.6. Output Analysis

The baseline simulation results for the Viau Terminal truck operations over a five-hour period are summarized in

Table 4. Output results of the simulation modeled.

Output Metric	Description	Results
Average Truck Turn Time (minutes)	Total time from port entry to final exit for each truck	88.2
Truck Throughput	Total number of trucks processed by the terminal during the simulation period	144
Queue Size at Terminal Entry Gate (T1_Queue)	Average number of trucks waiting before the terminal entry	3.01
Queue Size at Yard Entry Gate (T2_Queue)	Average number of trucks waiting before the yard entry	12.01
Yard Delay Time (minutes)	Average delay experienced during yard processing (load/unload operations)	64.81
Resource Utilization	Utilization rates of yard cranes	96%

. The model produced an average truck turn time (TTT) of 88.2 min, with a total of 144 trucks processed during the simulation. Queueing behavior was moderate at the terminal entry (3.01 trucks) but significantly higher at the yard entry (12.01 trucks), reflecting a bottleneck in yard operations. Yard delay time was found to be the dominant component of TTT, averaging 64.81 min, while yard crane utilization reached a high level of 96%, indicating near-capacity operation. These results align with historical operational reports and previous studies (e.g., [2]), supporting the plausibility of the model’s behavior.

Several simplifications influenced these outputs. First, the model assumed uniform truck characteristics (size, speed, and acceleration), thereby reducing complexity but underestimating real-world variability caused by a heterogeneous fleet. Second, truck arrivals were modeled as an averaged pattern over the study period rather than replicating daily peaks and troughs, which may affect the accuracy of congestion patterns during high-demand intervals. Third, resource availability was simplified; operational adjustments such as adding temporary gate lanes or reprioritizing truck flows were not included, potentially leading to understated system flexibility.

Despite these limitations, the outputs provide valuable insights into the terminal’s current operational performance and highlight key areas of congestion, particularly in the yard. The findings establish a credible baseline against which alternative scenarios can be evaluated. Subsequently, scenario analysis was conducted to compare these baseline results with configurations that incorporate autonomous vehicles (AVs) and adjusted yard processing capacities, thereby assessing their potential to reduce truck turn time and improve overall system throughput.

4. Case Study and Results

The Viau Terminal was specifically chosen for this study due to its specialization in container handling and its relevance to truck-based freight flows. It is operated by Termont Montréal since 2016 and has seen increased investment in recent years, including gate automation and expanded yard space. Located in the Mercier–Hochelaga–Maisonneuve borough, it handles approximately 600,000 TEUs and serves as a key gateway for global container flows, primarily via Mediterranean Shipping Company (MSC).

Despite these improvements, truck congestion remains a significant operational challenge, particularly during peak hours when long queues at the gates and staging areas cause delays, reduce efficiency, and contribute to environmental impacts.

The simulation experiments evaluated the impact of different operational strategies on truck turn time (TTT), resource utilization, and queuing dynamics at the Viau Terminal. Results are presented for the baseline system, Truck Appointment System (TAS), partial Autonomous Vehicle (AV) integration, dedicated AV infrastructure, and sensitivity analyses, with detailed tables and figures highlighting key performance outcomes. The scenarios were chosen to reflect both current operational realities and proposed near-future interventions that are relevant to the Port of Montréal’s Viau Terminal.

4.1. Baseline Scenario

The baseline simulation, representing current Viau Terminal operations without a Truck Appointment System (TAS) or Autonomous Vehicles (AVs), yielded an average truck turn time (TTT) of 88.20 ± 26.2 min (

Table 5. Truck Turn Time results for the baseline scenario.

Scenario	Average TTT (minutes)	T1 Queue Size	T2 Queue (Staging) Size	Yarding Time (minutes)	Throughput (Trucks)	Yard Resource Utilization
Scenario Baseline	88.2	3.01	12.01	64.81	144	96%

). Minimum and maximum TTT values were 42.5 min and 165.8 min, respectively, indicating high variability.

The average Truck Turn Time (TTT) of the baseline scenario was 88.2 ± 26.2 min. This aligns closely with real-world observational data from terminal camera systems and prior studies, which reported average TTTs of approximately 85–90 min under similar conditions. This validation confirms the model’s ability to accurately represent the existing operational challenges.

Analysis of congestion indicators showed substantial queue lengths and waiting times, particularly at the T2 Queue (staging area before yard operations), with an average size of 12.01 trucks, and within the yard processing area, indicated by a high Yarding Time of 64.81 min. These two points emerged as the primary bottlenecks, consistent with observations from port camera data. T1 Queue, with an average size of 3.01, is another congestion indicator which happens in the entry to the terminal. The unmanaged, continuous arrival of trucks led to unpredictable surges in demand, overwhelming the fixed capacities of these critical resources.

The throughput for the baseline scenario was 144 trucks, observed to be lower than optimal, directly constrained by the extended TTT and the inability of the system to efficiently process trucks during peak demand. Yard Resource Utilization was 96%, which, while seemingly high, was often indicative of congestion rather than efficient flow, as cranes were frequently occupied, but trucks were still experiencing long wait times due to upstream bottlenecks. The high TTT and significant queuing underscore the urgent need for intervention to improve efficiency and reduce operational costs at the Port of Montreal.

4.2. TAS Scenario

Implementation of a TAS reduced the average TTT to 78.37 ± 21.1 min, an 11.15% decrease relative to baseline. Standard deviation also dropped from 26.15 to 21.08 min, suggesting improved predictability (

Table 6. Truck Turn Time results under TAS implementation.

Scenario	Average TTT	T1 Queue Size	T2 Queue (Staging) Size	Yarding Time	Throughput (Trucks)	Yard Resource Utilization
Scenario 2-TAS	78.37	0.38	11.83	52.44	171	96%

). Minimum and maximum TTT values narrowed to 41.2 min and 142.7 min, respectively.

TAS implementation reduced Average Truck Turn Time (TTT) sharply to 78.37 ± 21.1 min. This is improved compared to the baseline of 88.2 min, which shows that the appointment scheduling was sufficient to manage demand. The TAS was able to handle the uneven arrival patterns, and therefore there was more even truck flow into the terminal.

Congestion metrics such as T1 Queue Size were significantly enhanced to 0.38 trucks from 3.01 in the base line. This shows that TAS was good at controlling truck arrival at the first gate. T2 Queue (Staging) Size remained high at 11.83 trucks, showing that although TAS enhanced entry flow, the internal staging area prior to yard operation still had significant queuing. Noticeably, the Yarding-Time was reduced to 52.44 min, reflecting more effective utilization of yard resources as a result of smoother arrivals.

Scenario 2 throughput was 171 trucks, showing that by regulating arrivals, the terminal would be able to handle more trucks over the same 300 min. Yard Resource Utilization was still 96%, but the system itself was more efficient since even arrival patterns left it with some space to schedule loading/unloading operations, minimizing idle time caused by random truck queues. This scenario confirms that even a basic TAS can yield substantial operational benefits by improving flow predictability and reducing congestion at critical junctures.

4.3. Partial AV Integration (35% AV Share)

The introduction of AVs into 35% of truck operations produced the most significant improvement in TTT. The average system TTT decreased to 55.91 ± 18.5 min, an approximately 37% reduction relative to baseline. AVs recorded an average TTT of 45.33 min, while HDVs averaged 61.62 ± 20.1 min (

Table 7. Truck Turn Time results under 35% AV integration.

Scenario	Average TTT			T1 Queue Size	T2 Queue (Staging) Size	Yarding Time		Throughput (Trucks)	Yard Resource Utilization
Scenario 3-AV Integration	Total	AV	HDV	0.35	4.81	HDV	AV	223	97%
	55.91	45.33	61.62			39.64	29.54

).

The average Truck Turn Time experienced a further significant reduction, dropping to 55.91 ± 18.5 min. This improvement over Scenario 2 (78.37 min) highlights the inherent efficiencies brought by AVs. Specifically, the AV turn time was 45.33 ± 15.2 min, significantly lower than the HDV turn time of 61.62 ± 20.1 min. This clear differentiation in performance underscores the advantages of AVs, attributed to their reduced reaction times, more consistent yard delay durations (modeled with a 20% reduction), and faster acceleration and speed. The T1 Queue Size remained low at 0.35 trucks, consistent with the continued effectiveness of the TAS. The most notable improvement in this scenario was T2 Queue size. It was reduced to 4.81 trucks on average indicating that AVs with their inherent efficiencies, can decongest bottlenecks in terminals. Although there was a 59% improvement in staging area bottleneck, this suggests that while AVs are faster once they seize a resource, their queuing behavior in a shared environment can still be a bottleneck. The Yarding Time HDV was 39.64 min, while the Yarding Time for AVs was 29.54. These results were expected as AVs are faster in transiting in yard area, faster once they seize and release a resource, and have a lower error rate in comparison to HDVs.

Throughput continued to increase to 223 trucks, reflecting the combined positive impact of TAS and the introduction of AVs. Yard Resource Utilization increased slightly to 97%, indicating a higher demand on the shared crane pool. This scenario demonstrates that even without dedicated infrastructure, the integration of AVs can significantly enhance terminal efficiency, particularly for the autonomous fleet itself, though shared resource contention can still lead to unexpected queuing for AVs.

4.4. Dedicated AV Infrastructure

When AVs were allocated dedicated staging areas and cranes, their efficiency improved further. The average AV TTT decreased to 32.86 ± 12.1 min, a 27.5% improvement compared to shared-resource AV operations (45.33 ± 15.2 min). However, HDV TTT increased to 70.20 ± 22.4 min, raising the overall average system TTT to 57.13 ± 19.3 min (

Table 8. Truck Turn Time results under dedicated AV infrastructure.

Scenario	Average TTT			T1 Queue Size	T2 Queue (Staging) Size		Yarding Time		Throughput (Trucks)	Yard Resource Utilization
Scenario 4-AV + Staging Area	Total	AV	HDV	0.48	HDV	AV	HDV	AV	224	95%
	57.13	32.86	70.20		3.01	2.88	54.68	19.88

).

The simulation results for Scenario 4, as detailed in Table 5, demonstrated a complex impact on terminal efficiency. The Average Truck Turn Time (TTT) for the entire system was 57.13 ± 19.3 min. While this is an improvement over the baseline (88.2 min) and Scenario 2 (78.37 min), it is unexpectedly higher than Scenario 3 (55.91 min). This counter-intuitive result warrants further discussion in Section 5.

A closer examination reveals that the impact on AVs was particularly pronounced and positive: AV TTT dropped significantly to 32.86 ± 12.1 min, representing the lowest AV TTT across all scenarios. This is a substantial improvement from Scenario 3’s AV TTT of 45.33 ± 15.2 min. This benefit is directly attributable to the dedicated staging area for AVs and the reserved crane allocation. The T2 Queue (Staging) size for AVs was drastically reduced to 2.88 trucks, an approximately 50% improvement, and Yarding Time AV was 19.88 min, an approximately 50% improvement, both indicating virtually unobstructed flow for autonomous vehicles within the terminal.

In this scenario, 35% of trucks were designated as autonomous vehicles (AVs) and assigned to a separate staging area and a dedicated subset of cranes (4 out of the total 15 cranes). The remaining 11 cranes and staging area were reserved exclusively for human-driven vehicles (HDVs). This configuration was designed to evaluate whether physical and operational segregation of AVs would enhance their performance and reduce overall congestion. The HDV turnaround time increased to 70.20 ± 22.4 min in Scenario 4, which is 12% higher than the HDV turnaround time of 61.62 ± 20.1 min in Scenario 3. This suggests that while dedicating resources to AVs greatly benefits the autonomous fleet, it may come at the expense of HDV efficiency if the remaining shared resources (11 cranes for HDVs) become more constrained. The T2 Queue (Staging) size for HDVs was 3.01 trucks, and Yarding Time HDV was 54.68 min. This outcome is consistent with principles observed in queueing systems and resource allocation studies. By dedicating cranes and staging infrastructure to Avs The effective processing capacity available to HDVs was reduced (from 15 cranes shared to only 11 cranes), even though the total truck arrival rate remained constant. Also, the variability inherent in human-driven truck arrivals and processing times was no longer smoothed by the more predictable AV flows sharing the same resources and thus, HDVs had to wait for longer queues and spend more waiting and yarding times. Throughput was still high at 224 trucks, a modest rise from Scenario 3, which showed the ability of the terminal to handle large numbers of trucks. In general, the system throughput was enhanced, but non-uniformly, and at the cost of AV operation at the expense of HDV performance. This finding suggests that, in practice, dedicated infrastructure for AVs can produce operational gains for the AV fleet but at the cost of degrading service for human-driven vehicles unless compensatory capacity adjustments are implemented.

In this scenario, yard crane resources were divided between human-driven and autonomous truck operations to reflect their respective fleet shares. A total of 15 yard cranes were modeled, with 4 cranes dedicated to AVs and 11 to HDVs. This split corresponds closely to the 35% AV fleet share used in the scenario, ensuring resource allocation remained proportional to demand while still allowing AVs to demonstrate operational efficiency gains. This allocation also reflects practical port management strategies in early automation phases, where a limited number of cranes are dedicated to autonomous operations for testing and control. The 4–11 configuration was therefore chosen to represent a moderate, scalable deployment of AV-handled operations consistent with pilot implementations at semi-automated terminals [5,38].

To verify that results are not overly dependent on this assumption, two alternative allocations were simulated:

• 3–12 split (reduced AV capacity);
• 5–10 split (increased AV capacity).

The resulting average truck turn times (TTT) for the 35% AV scenario were 57.13 min, 56.63 min, and 56.06 min for the 3–12, 4–11, and 5–10 cases, respectively. The largest variation across configurations was 1.1 min (≈1.2%), indicating that system-wide performance is only marginally sensitive to reasonable changes in crane allocation. Therefore, the 4–11 split was retained as a realistic and balanced configuration representing partial automation conditions at the Port of Montreal.

This scenario clearly illustrates that a synergistic approach, combining demand management (TAS), advanced vehicle technology (AVs), and targeted infrastructure investment, yields the most significant improvements in port operational efficiency and truck turn time for the AV fleet, but careful consideration of resource partitioning is needed to avoid negative impacts on the HDV fleet.

4.5. Sensitivity Analysis

Three sensitivity experiments tested the robustness of results.

• AV Fleet Share (25–75%): Increasing AV share produced diminishing returns. TTT decreased sharply between 35% and 50% AV penetration (Figure 2a).
• Crane Capacity: Adding one crane per yard block reduced HDV and AV delays, especially under shared-resource conditions (Figure 2b).
• Truck Arrival Rates: Increasing arrivals by 15% raised TTT across all scenarios. TAS and AV scenarios were more resilient, with TAS maintaining an average TTT below 80 min and AV integration below 60 min, compared to 88 min in the baseline (Figure 2c).

The AV Fleet Share range of 25–75% was chosen to reflect practical adoption scenarios consistent with global projections for autonomous logistics technologies. Studies such as [5,41] indicate that full automation (100% AV deployment) is not expected in the short to medium term, while partial adoption between 25% and 75% represents realistic transitional stages as ports progressively integrate AV fleets alongside human-driven vehicles. This range thus allows assessment of system performance under both early and advanced phases of AV integration, without assuming complete fleet automation.

4.6. Summary of Findings

The simulation experiments revealed distinct performance patterns across the four operational scenarios. The baseline scenario confirmed that yard operations are the main bottleneck, with high average truck turn times (88.2 ± 26.2 min) and heavy crane utilization (96%), consistent with observed port congestion. The implementation of a Truck Appointment System (TAS) improved flow regularity, reducing average TTT by approximately 11% and increasing throughput by 18%, demonstrating the value of demand management in mitigating peak congestion.

The partial AV integration scenario (35% AVs) achieved the most significant overall improvement, lowering system-wide TTT by about 37% relative to the baseline. The differentiation between AV and HDV performance highlighted the operational advantages of automation, particularly reduced yard delays and smoother flow through the terminal.

When dedicated AV infrastructure was introduced, AVs achieved their lowest individual TTT (32.86 ± 12.1 min), but HDVs experienced increased delays due to resource partitioning, resulting in a slight rise in the system’s overall average TTT. This outcome underscores the importance of balancing resource allocation to avoid efficient trade-offs between vehicle types.

The sensitivity analyses confirmed that AV adoption beyond 50% yields diminishing returns unless accompanied by proportional resource expansion, such as increased crane capacity. Higher truck arrival rates amplified congestion across all scenarios, but both TAS and AV configurations exhibited resilience, maintaining considerably lower TTTs than the baseline.

Overall, the results demonstrate that the combination of demand management and partial automation offers the most effective and balanced strategy for congestion reduction at the Port of Montreal’s Viau Terminal. These findings provide a foundation for the subsequent discussion and recommendations on port automation strategies.

5. Discussion

This study demonstrated that port congestion at the Viau Terminal, expressed through elevated truck turn time (TTT), can be significantly mitigated through a combination of digital demand management and emerging automation technologies. The baseline results confirmed that yard operations remain a persistent bottleneck, consistent with earlier findings that highlight yard crane utilization and queue accumulation as critical drivers of TTT variability [1,3].

The implementation of a Truck Appointment System (TAS) reduced both mean TTT and its variability, aligning with prior studies that emphasize the effectiveness of appointment systems in smoothing truck arrivals and lowering peak-hour congestion [13,19]. However, the observed improvements were modest compared to the integration of Autonomous Vehicles (AVs), suggesting that digital demand management alone may not be sufficient in high-volume terminals without complementary infrastructure and process optimization.

Partial AV integration provided the most substantial efficiency gains, reducing TTT by over 37% relative to the baseline. This finding supports prior simulation-based studies that indicate AVs can outperform human-driven vehicles through consistent speeds, reduced maneuvering times, and greater resilience to congestion [5,39]. The results also highlight positive spillover effects for HDVs, suggesting that AVs can alleviate system-wide congestion even without full adoption.

Interestingly, the dedicated AV infrastructure scenario revealed a trade-off: while AVs achieved their lowest TTT, HDVs experienced delays due to reallocation of shared resources. Similar concerns have been noted in the literature, where automation-focused designs can unintentionally disadvantage legacy systems if not carefully balanced [7,40]. This finding underscores the need for equitable resource planning to ensure that incremental automation benefits do not disproportionately burden traditional operators.

The sensitivity analysis reinforced the robustness of these findings. Increasing AV penetration showed diminishing returns beyond 40%, a trend consistent with prior adoption modeling studies [20]. Similarly, crane capacity expansions improved system performance primarily in AV scenarios, while increased arrival rates amplified congestion across all scenarios, validating the hypothesis that demand management and automation are most effective when combined.

From a managerial perspective, the results suggest that a hybrid approach—integrating TAS with partial AV deployment—offers a balanced strategy for ports transitioning toward automation. This combination provides meaningful reductions in TTT without imposing the trade-offs observed in dedicated AV infrastructure. For policymakers and port authorities, these findings support investment in both digital scheduling systems and gradual AV adoption, rather than isolated reliance on one measure. Stakeholders should plan for socio-economic impacts—especially workforce displacement and develop retraining programs for workers to foster public acceptance. To leverage AV efficiencies specially in shared infrastructure, port authorities need to consider flow paths to avoid new bottlenecks and rigorously model and phase investments to avoid negative impacts on HDV efficiency and consider dynamic resource allocation. Also, optimizing all critical resources and processes to ensure system-wide benefits, not just localized gains seems crucial.

While the simulation framework provides valuable insights, several limitations should be acknowledged. First, the model was conceptual and relies on simulated AV performance parameters rather than real-world deployment data, as large-scale AV adoption in ports remains limited. Second, the analysis focused on a single terminal and did not capture interactions across multiple terminals or with external urban traffic conditions, which can significantly influence TTT. It is important to note that the simulation model represents a single-terminal abstraction of port operations, specifically focusing on the Viau Terminal. Multi-terminal interactions, shared resource dependencies, and city-side traffic effects were intentionally excluded to maintain model tractability and focus on internal flow efficiency. As such, the results should be interpreted as indicative of within-terminal performance improvements, rather than system-wide congestion reduction across the entire port complex.

Third, behavioral responses from stakeholders, such as truck operators’ compliance with TAS, were not modeled explicitly. Future research should address these gaps by incorporating empirical AV performance data, extending the scope to multi-terminal networks, and integrating behavioral modeling to capture a wider range of operational dynamics.

6. Conclusions and Future Works

6.1. Conclusions

This study developed a hybrid Discrete Event Simulation–Agent-Based model to analyze truck congestion at the Port of Montreal’s Viau Terminal and evaluate the potential of both digital and automation-based solutions. Four operational scenarios were examined: current baseline operations, a Truck Appointment System (TAS), partial integration of Autonomous Vehicles (AVs), and AVs supported by dedicated infrastructure. Sensitivity analyses were also conducted to test robustness under varying conditions of AV penetration, crane capacity, and truck arrival rates.

The results demonstrated that TAS provides moderate improvements in truck turn time (TTT) and system stability, while partial AV integration yields the most significant overall benefits, reducing TTT by more than one-third. Dedicated AV infrastructure further enhances AV performance but creates trade-offs for human-driven vehicles. Sensitivity experiments confirmed the resilience of TAS and AV strategies to fluctuations in demand, while also highlighting diminishing returns beyond certain adoption thresholds.

The findings suggest that a balanced approach—gradual AV integration combined with demand management systems—offers the most effective strategy for reducing congestion without disadvantaging existing stakeholders. The study contributes to the growing body of port automation research by quantifying the operational impacts of AVs in a North American port context.

While this study primarily focused on operational efficiency rather than financial performance, the findings have direct implications for investment planning. A qualitative cost–benefit reflection was therefore conducted based on typical cost estimates and the model’s observed performance gains. Published industry reports estimate that implementing autonomous truck operations in container terminals requires capital investments ranging from USD 250,000–400,000 per AV unit, plus infrastructure upgrades (sensors, communication systems, and staging modifications) totaling approximately USD 5–8 million for a medium-sized terminal [5,32]. Operating costs per hour for AVs are typically 20–30% lower than for human-driven trucks due to reduced labor and fuel inefficiencies [38].

In the modeled scenarios, partial AV integration (35% of fleet) reduced average Truck Turn Time (TTT) by roughly 37% compared to the baseline, leading to increased throughput and reduced idling. Assuming a truck fleet of 200 vehicles, this efficiency translates to potential time savings of 25–30 truck-hours per day during peak operations. Based on an estimated cost of USD 80–100 per truck-hour (including fuel, labor, and demurrage), the operational savings could reach USD 0.7–1 million annually under partial automation.

When compared to indicative infrastructure costs, these savings suggest a medium-term payback horizon of 6–8 years, excluding additional environmental and reliability benefits. Although a complete financial model was beyond the scope of this study, these estimates support the conclusion that AV adoption in the Port of Montreal context is economically promising when implemented incrementally alongside demand-management strategies such as TAS.

6.2. Future Work

Though this study offers practical observations on the ability of autonomous vehicles and operational strategies for reducing truck turn time at the Port of Montreal, several avenues for further research can enhance the model’s realism, scope, and applicability.

6.2.1. Using Real Data to Validate Findings

The existing simulation model, although informed by publicly available data and prior studies, was still conceptual. One of the most important steps in the process would be to work closely with the Port of Montreal or Termont Montréal to obtain high-resolution, real-time operating data. This would include precise distributions for gate processing times, yard service times, actual truck arrival patterns, and detailed resource utilization logs. All these data would allow for closer calibration and validation of the model, going beyond qualitative fit and nearing quantitative accuracy. This additional verification would greatly increase the validity and applicability of the findings, so the port authorities would be more directly able to apply the recommendations.

6.2.2. Including Multi-Terminal Behavior

The current model focuses exclusively on the Viau Terminal. The Port of Montreal possesses multiple container terminals, and truck traffic tends to cross among these terminals. Subsequent research could expand the model to accommodate multi-terminal behavior, representation of truck travel between terminals, inter-terminal transfer impacts, and potential spillover congestion. This would provide a more complete representation of port-wide logistics network and allow coordinated strategy to be tested on all terminals, such as a port-wide TAS or shared AV fleet.

6.2.3. Expanding Model to Account for Last-Mile Delivery

The scope of this study was intentionally confined to the internal landside operations of the Viau Terminal. The value addition would be to integrate the “last-mile” delivery feature, mirroring truck movement from port of departure right through to their ultimate destination in the urban or regional hinterland. External road network conditions, city traffic, and last-mile logistics special features would be added. This expansion would create a larger perspective on the overall drayage process to permit consideration of how port-side efficiencies are being translated into supply chain advantages more broadly and how AVs would affect the port-to-consignee trip overall. This can also enable consideration of the environmental advantages of less urban congestion and greater efficiency.