1. Introduction
The transition from traditional on-site construction methods to off-site construction signifies a pivotal evolution in the construction industry [
1]. Off-site construction (OSC) encompasses several levels of prefabrication, each offering distinct advantages in terms of efficiency, delay, cost, and environmental impact [
2]. The simplest form, construction based on prefabricated components (PCs), involves the off-site manufacturing of basic structural elements like beams, columns, and slabs, using materials such as reinforced concrete, steel, and wood. This method, while lessening some on-site activities, still requires significant on-site construction efforts [
3]. More advanced is panelized construction, where major structural components such as walls, floors, and roofs are produced in factories as prefabricated panels [
4]. These are then transported to construction sites for assembly, improving the construction process by reducing the construction timeline and on-site labor requirements. Further along the spectrum lies modular construction (MC), where entire sections of a building, such as hotel rooms or classrooms, are fully constructed off-site in a factory setting [
5]. These modules are then transported to the site and assembled into the final structure. This method allows for building components to be stacked, joined side by side, or layered, depending on the architectural design. At the top of OSC technology is modular integrated construction (MiC) [
3]. MiC not only employs the benefits of modular construction but also integrates mechanical, electrical, and plumbing components along with interior finishes during the factory phase. This integration significantly expedites the overall construction process and minimizes the need for on-site installation, offering the highest level of prefabrication. Given its comprehensive approach, MiC presents a unique set of logistical challenges, particularly in the transport of these fully integrated modules.
Figure 1 illustrates the distinct stages of OSC. This progression from the simplest form of prefabrication, using basic structural components, through more complex modular constructions highlights the incremental advancements in prefabrication techniques that enhance efficiency, reduce delays, and mitigate environmental impacts across the construction industry.
Recent studies in MiC and construction supply chain management (CSC) have highlighted a series of significant challenges necessitating in-depth research [
3,
6,
7,
8,
9]. These works underline the critical need for resilient CSC frameworks capable of mitigating impacts from disruptions, a common feature in today’s dynamic construction environments [
10]. Enhancing CSC resilience relies crucially on robust information sharing and collaborative efforts among all stakeholders, including suppliers, manufacturers, transporters, contractors, and project managers [
6].
Advanced technologies, such as Building Information Modeling (BIM) and Internet of Things (IoT), are essential throughout the entire building lifecycle, including the design, construction, and exploitation phases [
11]. BIM provides detailed digital models that enhance decision-making, coordination, and efficiency at each stage [
12]. IoT complements BIM by enabling real-time data collection and communication among devices, which allows for the dynamic monitoring and control of construction processes [
13]. Additionally, IoT allows for the continuous monitoring and management of building systems in the exploitation phase (e.g., energy use, environmental conditions, and equipment status), ensuring optimal performance and proactive maintenance [
14]. Together, BIM and IoT support the creation of digital twins—virtual replicas of physical assets—that offer comprehensive insights and predictive analytics to optimize both construction and post-construction phases [
11,
13]. In the context of MiC, these technologies are particularly impactful. The detailed digital representations of BIM facilitate precise prefabrication and assembly, improving coordination and reducing errors. IoT enhances these capabilities by providing real-time data on the status and condition of components and modules during transportation and installation [
15,
16,
17,
18]. Moreover, Artificial Intelligence (AI) is recognized for its potential to significantly improve MiC operations. AI can automate complex data management tasks and improve traceability throughout the MiC supply chain, providing stakeholders with real-time visibility of logistics operations. This transparency is essential for making informed decisions quickly, improving operational efficiency and reducing delays. AI can also facilitate predictive analytics, enabling potential delays or problems to be anticipated before they occur [
8].
Recent works in the literature emphasize that MiC is not only about improving efficiency and productivity but also about fostering sustainability and environmental responsibility [
3]. MiC supports the principles of the circular economy by promoting resource efficiency, reducing construction waste, and enabling the reuse and recycling of building components [
19]. This is achieved through design for disassembly (DfD) and the use of recycled materials, which significantly lower the environmental impact of construction projects [
20,
21]. By integrating circular economy strategies such as reducing, reusing, and recycling materials, MiC helps create a more sustainable and resource-efficient built environment [
22]. Moreover, from a MiC supply chain perspective, optimizing components and modules logistics and adopting closed-loop supply chains can significantly minimize CO
2 emissions, as efficient transport logistics inherently reduce fuel consumption and emissions [
23]. These practices align with the broader objectives of the 2030 Agenda for Sustainable Development adopted by the United Nations [
24]. Despite the advancements in MiC methods, the literature reveals a notable lack of quantitative analyses concerning the cost-effectiveness and sustainability benefits of MiC [
9,
25]. In addition, sustainability assessments are often qualitative rather than quantitative, limiting the ability of decision-makers to make comprehensive assessments based on solid empirical data [
9].
In addition to the economic and environmental considerations, the literature indicates an absence of advanced, suitable models for optimizing transport planning and supply chain configurations in the context of MiC [
3,
7]. In particular, there are gaps in strategies for efficiently managing the logistics of large, integrated modules requiring special handling and routing procedures. Stochastic programming appears to be a recommended method for filling these gaps [
7]. This approach makes it possible to model the uncertainties and variabilities inherent in construction logistics, such as delivery delays, variable module sizes and fluctuating costs, thereby improving supply chain resilience. Moreover, the recent literature review in MiC identifies research opportunities across all phases of construction [
3]. This includes the use of decision-making systems in design, and the adoption of Industry 4.0 technologies in module production for increased operational effectiveness. In module logistics, advanced models optimize the location of module fabrication centers and vehicle routes, enhancing performance while considering environmental and economical impacts. On-site, sophisticated planning models accelerate assembly and optimize project durations, demonstrating a shift towards more technologically integrated construction practices.
To address these challenges, this paper explores the potential of multi-agent simulation using AnyLogic to enhance the MiC supply chain. Multi-agent simulation offers a sophisticated approach to modeling the interactions of various agents involved in the process, such as suppliers of prefabricated components, fabrication centers for integrated modules, construction sites, and transportation vehicles [
26,
27]. This simulation paradigm provides a comprehensive tool for optimizing transport configurations and supply chain operations [
23]. By simulating the complex behaviors and interactions within the logistics network, this approach enables the detailed analysis and optimization of MiC transport logistics. The study introduces a multi-agent-based model specifically designed to optimize transportation logistics for MiC, focusing on reducing costs and improving operational efficiency. Results from the model application demonstrate enhanced logistics performance and underscore the value of multi-agent simulation in addressing the complexities of MiC transport logistics.
Therefore, the article is structured as follows:
Section 2 provides a comprehensive review of the existing literature related to MiC and simulation-based models, and examines the specific transportation challenges faced by MiC.
Section 3 details the developed agent-based model and the interactions between each agent, utilizing AnyLogic to efficiently address these challenges.
Section 4 analyzes the results obtained from implementing the multi-agent simulation model, focusing on Key Performance Indicators (KPIs) related to MiC transport. Finally,
Section 5 concludes the paper by summarizing the key findings and discussing future research avenues and potential advancements in the MiC supply chain.
4. Results and Discussion
In this section, we present the results of the proposed multi-agent simulation model and provide an analysis of the findings.
4.1. Study Scenarios
To evaluate the efficiency and effectiveness of logistics in MiC, we consider two construction configurations: prefabricated component construction without MiC factories, and fully MiC process.
We chose parameters that reflect the typical mid-sized project, logistics capacities, and operational constraints encountered in the construction industry. The number of suppliers, vehicles, production capacities, and cost elements were chosen based on common practices and the literature to ensure the model’s relevance and accuracy. For instance, the numbers of suppliers and vehicles were determined to capture the logistical complexity to maintain a steady flow of components to the construction site, while the cost parameters were based on standard economic considerations to reflect the financial implications of logistics decisions (e.g.,
https://www.truckingdive.com/ (accessed on 30 May 2024)). Additionally, incorporating stochastic elements such as vehicle speed and production rates helps simulate the variability and uncertainty inherent in real-world scenarios.
4.1.1. Scenario 1: Prefabricated Component Construction without MiC Factories
In this scenario, prefabricated components are supplied directly to the construction site by a number of suppliers. This represents traditional prefabricated construction rather than MiC. In this scenario, we consider the following parameters:
Number of suppliers: For a mid-sized project, we assume five suppliers to provide different components such as beams, columns, slabs, and exterior panels. These components are crucial for the structural integrity of the construction.
Number of vehicles: Each supplier uses vehicles to deliver components to the construction site, based on the typical logistics capacity required to maintain a steady flow of components. We will conduct a sensitivity analysis by varying the number of vehicles per supplier from one to five to find the optimal number that minimizes the total cost.
Demand generation: The demand for components is deterministic and based on project management derived from BIM, with an average of 60 orders per day. This demand is influenced by the construction schedule and project planning, ensuring that the supply chain remains responsive to the dynamic needs of the construction project.
Initial components in storage: Each supplier starts with an initial inventory between 5 and 10 components, providing a buffer to meet the immediate demand.
Production capacity/rate: Each supplier has a stochastic production capacity, with an average of 10–15 components per day depending on the complexity and type of component, with a production rate that ensures steady output to meet the ongoing demand.
Cost Parameters:
- –
Fixed cost per vehicle per day (=EUR 150):This cost per vehicle per day is a set amount that must be paid to maintain the vehicle in the fleet, regardless of whether it is actively used for transportation on that day. It is incurred regardless of the level of activity and the distance traveled. For examples, this includes costs such as vehicle leasing or rental fees, insurance, permits, and salaries for drivers and staff.
- –
Variable cost per kilometer (=EUR 2): This cost reflects the additional expenses incurred for every kilometer the vehicle travels. This includes fuel consumption, wear and tear, and other distance-dependent costs.
- –
Project delay penalty per day (=EUR 500): Refers to a financial penalty that is imposed for each day a project extends beyond its scheduled completion date. For instance, if a construction project is supposed to be completed in 120 days but takes 130 days to finish, a penalty of EUR 500 will be charged for each of the 10 extra days, resulting in a total penalty of EUR 5000. This penalty is intended to incentivize timely project completion and to compensate for the additional costs and potential losses incurred due to the delay.
Vehicle speed: The speed of the vehicles is stochastic, with an average speed of 50–70 km/h.
Geo-locations:
- –
Construction site is located in Paris, France;
- –
Supplier 1 is 85 km from the construction site;
- –
Supplier 2 is 130 km from the construction site;
- –
Supplier 3 is 145 km from the construction site;
- –
Supplier 4 is 75 km from the construction site;
- –
Supplier 5 is 135 km from the construction site.
The normal project completion time for this scenario is set at 120 days. Any delays beyond this period will incur a project delay penalty as mentioned in the cost parameters.
4.1.2. Scenario 2: Fully MiC Process
In this scenario, suppliers provide prefabricated components to MiC factories, which then supply integrated construction modules to the construction site. This configuration leverages the benefits of MiC to enhance efficiency and reduce the on-site construction time. The parameters for this scenario are as follows:
Number of Suppliers: Similar to Scenario 1, we assume five suppliers providing different components required for the modules. This allows for a diverse range of components, ensuring that all necessary parts are available for the integrated modules.
Number of MiC Factories: We assume one MiC factory responsible for assembling the integrated modules from the supplied components. Centralized assembly at a single factory can improve efficiency and reduce complexity in logistics.
Number of Vehicles:
- –
Transporter 1 () for components: Similar to Scenario 1, each supplier uses vehicles for transportation of prefabricated components to the factories. We will conduct a sensitivity analysis by varying the number of vehicles per supplier from one to five to find the optimal number that minimizes the total cost.
- –
Transporter 2 () for modules: The MiC factory uses one vehicle for module delivery to the construction site. This is based on the need to maintain a steady flow of modules while minimizing transportation costs.
Demand generation: The demand for components and modules is deterministic and based on project management derived from BIM, with an average of 60 component requests per day by the MiC factory and three integrated module requests per day by the construction site.
Initial components in storage: Each supplier starts with an initial inventory between 5 and 10 components, providing a buffer to meet the immediate demand for construction components.
Production Capacity/Rate:
- –
Each supplier has a stochastic production capacity, with an average of 10–15 components per day depending on the complexity and type of component.
- –
Each factory has a production capacity of three modules per day. This rate ensures a continuous supply of modules to the construction site while maintaining high-quality standards.
Cost Parameters:
- –
Fixed cost per vehicle per day: EUR 150 for , EUR 200 for . The higher cost for reflects the more complex logistics and handling required for transporting integrated modules.
- –
Variable cost per kilometer: EUR 2 for , EUR 3 for . Transporting integrated modules incurs higher variable costs due to their size and weight.
- –
Project delay penalty per day: EUR 500. This penalty is consistent across both scenarios, incentivizing timely project completion and compensating for potential financial losses due to delays.
Vehicle Speed: The speed of the vehicles is stochastic, with an average speed of 50–70 km/h for and 40–60 km/h for . The lower speed for reflects the additional care and slower speeds required for transporting larger, more delicate modules.
Geo-locations:
- –
Construction site is located in Paris, France;
- –
MiC factory is 20 km from the construction site;
- –
Supplier 1 is 70 km from the MiC factory;
- –
Supplier 2 is 125 km from the MiC factory;
- –
Supplier 3 is 155 km from the MiC factory;
- –
Supplier 4 is 75 km from the MiC factory;
- –
Supplier 5 is 125 km from the MiC factory.
The normal project completion time for this scenario is set at 120 days. Any delays beyond this period will incur a project delay penalty.
Table 3 provides a summary of the parameters for both scenarios.
4.2. Results
In this section, we introduce the Key Performance Indicators (KPIs) used to evaluate the logistics performance in each scenario. The KPIs considered are total variable cost, total fixed cost, total penalty, and total cost. We will also present the results of the simulations for both scenarios.
Total variable cost (CV): This cost is calculated using the following formula:
Total fixed cost (CF): This cost is calculated as follows:
Total penalty (P): This is the financial penalty imposed for each delay day to complete the project. It is calculated as follows:
Total cost (CT): This is the sum of the total variable cost, total fixed cost, and total penalty:
4.2.1. Results of Scenario 1
In this scenario, prefabricated components are supplied directly to the construction site by five suppliers.
Figure 7 presents the GIS map from AnyLogic’s simulation showing the locations of each supplier and construction site.
The results for Scenario 1 are presented in
Table 4.
In Scenario 1, we analyze the costs associated with the delivery of prefabricated components.
Table 4 shows how the total cost varies with the number of vehicles used by each supplier.
The variable cost remains constant at EUR 1,938,540 across all configurations. This indicates that the total distance traveled by all vehicles does not change with the number of vehicles, suggesting a stable demand and consistent routing efficiency.
The fixed cost increases with the number of vehicles. For instance, with one vehicle per supplier, the total fixed cost is EUR 237,000, which rises to EUR 322,500 when five vehicles are used per supplier.
Penalties are incurred due to project delays. When only one vehicle per supplier is used, the delay costs EUR 98,000, reflecting significant delays. As the number of vehicles increases, the penalty decreases, eventually reaching zero when three or more vehicles are used. This demonstrates that having more vehicles reduces the likelihood of project delays, ensuring the timely delivery of components.
The optimal number of vehicles appears to be three, as this configuration has the lowest total cost of EUR 2,177,040, with no penalty costs and balanced fixed costs. Adding more vehicles increases the fixed costs without further reducing penalties, leading to higher total costs.
4.2.2. Results of Scenario 2
In this scenario, suppliers provide prefabricated components to MiC factories, which then supply integrated construction modules to the construction site.
Figure 8 presents the GIS map from AnyLogic’s simulation showing the locations of each supplier, MiC factory, and construction site.
The results for Scenario 2 are presented in
Table 5.
In Scenario 2, we analyze the costs associated with the delivery of prefabricated components to the MiC factory and the subsequent delivery of integrated construction modules from the factory to the construction site.
Table 5 shows how the total cost varies with the number of vehicles used by each supplier.
The total variable cost includes the variable costs for both and . For example, with one vehicle per supplier, the variable cost for is EUR 1,870,500, and for , it is EUR 43,320. This cost remains stable for and across all configurations, indicating consistent routing efficiency.
The fixed cost comprises the fixed costs for both and . As the number of vehicles increases, the total fixed cost for rises from EUR 220,500 to EUR 322,500, and the reverse with respect to fixed costs for , which varies between EUR 58,800 and EUR 17,200, because in all configurations, we use one vehicle, and so the fixed costs depend only on days of use of this vehicle.
Related to penalties, with 1 vehicle per supplier, the delay costs EUR 87,000, reflecting significant delays. As the number of vehicles increases, the penalty decreases, eventually reaching zero when three or more vehicles by each supplier are used.
The optimal number of vehicles appears to be three, as this configuration has the lowest total cost of EUR 2,153,920.
4.3. Discussion
In this section, we compare the results obtained for the two scenarios. The analysis of the results as shown in
Figure 9 from both scenarios highlights several key points.
In both scenarios, the total cost initially decreases as the number of vehicles per supplier increases, reaching a minimum at three vehicles per supplier. Beyond this point, further increasing the number of vehicles leads to higher total costs. This trend suggests that there is an optimal number of vehicles for each supplier that balances the fixed and variable costs while minimizing penalties. Using three vehicles per supplier is the optimal configuration. This setup achieves the lowest total cost by eliminating penalties. This result underscores the importance of optimizing the number of logistical resources to ensure cost efficiency and timely project completion. This finding aligns with the work of [
57], which highlights the significance of optimal vehicle allocation in prefabricated construction logistics. Their study demonstrates that an automated approach to vehicle allocation can substantially reduce the transportation costs associated with inefficient vehicle use.
Moreover, we can observe that Scenario 2 with the fully MiC process generally results in lower total costs compared to Scenario 1 across various vehicle configurations. This indicates that the MiC approach offers better cost savings due to centralized assembly and optimized transportation logistics. This finding aligns with the results of comparative studies which highlight that modular construction can lead to a 10–25% decrease in construction costs. One of the primary factors contributing to these cost savings is reduced material transportation, as components are produced off-site and then assembled on-site, leading to more efficient logistics and lower overall expenses [
58].
On the other side, the presence of penalties significantly affects the total cost when the number of vehicles is insufficient to meet demand promptly. In both scenarios, configurations with fewer than three vehicles per supplier incur penalties, which substantially increase the total cost. This highlights the critical impact of ensuring adequate logistical capacity to avoid project delays. This observation is consistent with findings in the literature, which identify poor supply chain capacity, including delays in the delivery of modular components, as critical risk factors that can derail the success of MiC projects [
59].
A notable observation is that the difference in total costs between the two scenarios becomes constant beyond the optimal number of vehicles. This phenomenon can be attributed to several factors. Once the number of vehicles per supplier reaches the optimal point of three vehicles, penalties for delays are eliminated in both scenarios. This means that any additional vehicles do not contribute to reducing penalties further, as there are no penalties left to reduce. Furthermore, the variable costs, which depend on the distance traveled and the number of trips, stabilize once the logistics operations reach optimal efficiency. Both scenarios achieve a similar level of efficiency in terms of transportation distance and frequency, leading to a constant difference in variable costs. Additionally, beyond the optimal point, the additional fixed costs of deploying more vehicles dominate the total cost structure. These fixed costs are similar in both scenarios, leading to a constant difference in total costs. The fixed costs per vehicle per day become the primary factor of increased costs, and since they are applied uniformly across both scenarios, the difference remains constant.
The findings from this study provide valuable insights for construction project managers and logistics planners in the MiC industry. Understanding the optimal number of vehicles per supplier helps in making informed decisions about resource allocation, ultimately leading to cost savings and timely project completion. The comparison between Scenario 1 and Scenario 2 demonstrates the potential cost-efficiency benefits of adopting MiC over prefabricated component construction [
58]. The centralized assembly in MiC allows for more efficient transportation logistics, reducing the total cost [
60]. While the MiC approach shows significant advantages, Scenario 1 may still be preferable for projects with simpler logistics requirements and fewer components.
4.4. Limitations and Future Research Directions
While this study offers valuable insights, several limitations must be acknowledged. The model assumes a fixed demand for both modules and components, which does not fully capture the variability and uncertainty present in real-world scenarios. Additionally, it does not consider real-time traffic data or weather conditions, both of which significantly impact transportation logistics. Ignoring these elements can lead to less accurate simulations and suboptimal logistical planning. Also, in this model, we have not taken into account multi-modal transport (e.g., train, air, river and sea, and vehicles) which could optimize logistics. Furthermore, the current model does not evaluate the environmental impact of different scenarios, thus ignoring important sustainability factors such as carbon emissions and energy consumption. Lastly, the simulation is based on a specific geographic area, which may not be representative of other regions with different logistical challenges, regulatory environments, and infrastructure conditions.
Building on the insights gained from this study, several avenues for future research and practical advancements in the MiC supply chain can be identified. Future studies should incorporate real-time traffic data, weather conditions, and dynamic demand variations to enhance the accuracy and responsiveness of logistics models. Leveraging real-time information can help anticipate and mitigate disruptions, ensuring smoother supply chain operations. Additionally, incorporating environmental impact assessments into the logistics models can provide a more holistic evaluation of the benefits of MiC. Moreover, utilizing advanced optimization techniques, such as machine learning algorithms and genetic algorithms, can further refine logistics planning. These methods can identify the most efficient routes, schedules, and resource allocations under various constraints and uncertainties. The integration of IoT and blockchain technologies can enhance transparency, traceability, and coordination among stakeholders. IoT devices can provide the real-time tracking and monitoring of logistics operations, while blockchain can ensure secure and immutable data sharing across the supply chain. Finally, developing digital twin models of the MiC supply chain can enable real-time simulation and predictive analytics. Digital twins can provide a virtual replica of the physical supply chain, allowing stakeholders to test different scenarios, optimize processes, and make informed decisions.