The Role of Digital Transformation in Manufacturing: Discrete Event Simulation to Reshape Industrial Landscapes

Abstract

In the era of Industry 4.0, the integration of intelligent systems with human elements presents both opportunities and challenges. This study explores this interplay through the application of an industrial engineering technique to a real process issue, demonstrating originality in problem selection and solution tools, as well as the relevance of the results. An operational framework is proposed to drive digital transformation in manufacturing by balancing automated systems efficiency with the complexity of human activities, which include decision-making flexibility, adaptability, tacit knowledge and collaborative interaction. It examines Industry 4.0 domains to find solutions that use smart technology while enhancing human experience. A key element is the use of discrete-event simulation to create a digital replica of the existing process. This enabled a detailed analysis and the development of innovative, validated approaches through what-if scenarios. The implemented solutions led to a significant annual increase in productivity, the result of an overall improvement in process efficiency, which was also achieved through the identification and resolution of key process bottlenecks, confirming the method’s effectiveness. The research offers a scalable model for various sectors, emphasizing the need to integrate human aspects into intelligent systems. It highlights how technological progress should enrich, not overshadow, human contribution, contributing to a deeper understanding of digital transformation in intelligent manufacturing and service systems, where technology and humanity evolve together.

1. Introduction

Until recently, digitization was seen as a strategic lever to innovate, stand out, and attract market attention [1]. Today, it has become a necessity for all companies [2], as evidenced by the value of the global digital transformation market, estimated at USD 731.13 billion in 2022, with a projected annual growth of 26.7 percent until 2030 [3]. It is no longer a choice, but a compulsory shift [4], accelerated by the COVID-19 pandemic that has driven every industry to adopt digital solutions [5].

To be competitive, companies must embrace digital transformation [6]. In manufacturing, this means integrating information systems, production environments, and operational technologies [7], aiming for more agile, flexible, and intelligent infrastructures capable of adapting to the changing market [8,9]. In this context, digital transformation is one of the main outcomes of Industry 4.0 [10] and requires a profound overhaul of the entire business organization [11].

Integrating digital technologies into all processes is not easy, and many companies are not yet ready [12]. Reliable tools are needed to analyze system performance, manage complex changes, and support decision-making [13]. Indeed, moving to automation involves comparing different operational configurations before final adoption [14], evaluating options in terms of cost, flexibility, and throughput [15], as well as the optimal allocation of buffering resources in often constrained contexts [16].

In a resource-constrained scenario, it is critical to invest purposefully and use fast and accurate analytical tools [17,18]. Among the enabling technologies, discrete event simulation (DES) offers key support for digital transformation [19]. In addition to digitizing processes, DES makes it possible to analyze, optimize, and test alternative scenarios in a virtual environment without operational impacts [20,21]. This makes it a strategic tool in initial planning, helping companies understand the real impact of new technologies on their processes [22].

In this scenario, the present research proposes a case study in which DES is employed to guide and support the digital transformation process of an enterprise operating in the manufacturing sector. Specifically, WITNESS simulation software v 14.0 is used to digitally reconstruct the manufacturing process of an Italian company that produces aluminum cabinets for electric charging stations for automobiles. In a first phase, the simulation allowed for a detailed study and analysis of the process, identifying strengths and weaknesses [23,24]. The results obtained were crucial for the next stage, in which several “what-if” analyses were conducted to evaluate the performance of various technological solutions to be implemented in the process, with the aim of optimizing it and initiating the development of smart manufacturing.

The goal of this study is to propose a scientifically rigorous approach to support companies in their digital transition. DES is the tool used to guide business managers through such a radical and delicate change. An innovative approach is proposed using a framework and specific guidelines.

In detail, it is intended to clarify the following:

• Q1: What is the evolutionary scenario and current status of DES becoming a Key Enabling Technology (KET) for Industry 4.0.?
• Q2: What are the distinguishing features of DES that offer significant benefits to manufacturing companies in terms of operational efficiency, cost reduction, and product quality improvement?
• Q3: How can DES effectively drive digital transformation in processes, facilitating the adoption of emerging technologies and optimizing the entire production chain?

The proposed framework provides a clear and methodological approach for addressing digital transformation, offering companies a defined path. The practical guidelines, based on experience and empirical evidence, show how to apply the theoretical concepts of digital transformation in the real context of production processes. This reduces the operational and financial risks associated with direct experimentation on business processes, allowing companies to more confidently evaluate the effectiveness of proposed solutions. The strength and success of the framework developed in this research are demonstrated by the practical implementation of the proposed methodology in a real production process. The conformity of the results with the estimates obtained from the simulation confirms the novelty and scientific relevance. The article presents original contributions focused on the development of new computer-based methodologies for solving industrial engineering problems and applications of these methodologies to problems of interest in the broad field of industrial engineering and related communities. It also expands the frontiers of fundamental theories and concepts that form the basis of industrial engineering techniques.

The rest of the work is structured as follows: Section 2 introduces background and research gaps, while Section 3 presents materials and methods, detailing the tools used and the applied methodology for research. Section 4 defines the reference scenario. The results obtained are discussed in Section 5, where possible process optimization is evaluated through what-if analysis. Finally, Section 6 summarizes the future developments and conclusions.

2. Background and Research Gaps

Discrete event simulation is a Key Enabling Technology (KET) of Industry 4.0 [25]. DES, which originated in the previous century, saw its first applications during World War II in the military field [26]. In the 1950s and 1960s, with the development of early programming languages and simulation systems such as SIMSCRIPT and GPSS, it became possible to define discrete events and simulate them on early computers [27,28]. Ten years later, in the 1970s, the first commercial simulation software, which greatly simplified the modeling and simulation of organizational systems using DES [29]. Over time, the progression of DES modeling techniques has continued [30]. Researchers and practitioners have developed more sophisticated modeling constructs, such as process-oriented simulation, object-oriented simulation, and the integration of DES with other methodologies, until today, where it is accessible and effective due to the prevalence of powerful computers and software tools [31]. In the era of Industry 4.0, DES is increasingly integrated with digital technologies, real-time data, and the Internet of Things [23]. This enables more dynamic and data-driven simulations in smart factories and supply chains [32]. Versatility and the ability to model complex systems have made DES a valuable tool in various areas, which are summarized in Table 1.

Table 1. Main fields of application of DES.

Field of Application	Main Benefits	Main Challenges	Main References
1. Manufacturing and Production	DES is extensively used in manufacturing to optimize production processes, manage resources, reduce bottlenecks, and improve overall efficiency. It helps in the design and layout of production facilities, scheduling, and quality control.	Complex system dynamics, high variability in processes, need for detailed data	[23,33,34]
2. Logistics and Supply Chain Management	DES is essential for modeling and analyzing supply chain operations, including inventory management, order processing, transportation, and distribution. It aids in optimizing logistics networks and improving delivery schedules.	Dynamic and uncertain environment, complex interaction of factors, need for real-time data	[35,36,37]
3. Healthcare	DES is employed to model patient flow, hospital operations, and emergency department scenarios. It helps in optimizing healthcare system designs, resource allocation, and patient throughput	Sensitive data handling, complex patient pathways, staff scheduling complexities	[38,39,40]
4. Transportation and Traffic Management	DES is used to simulate traffic flow, public transportation systems, and logistics networks. It aids in designing efficient traffic management systems, evaluating congestion scenarios, and optimizing transportation schedules.	Complex urban layouts, unpredictable traffic patterns, coordination with multiple agencies	[41,42,43]
5. Aerospace and Defense	The aerospace and defense industries use DES to model complex systems, such as air traffic control, military operations, and logistics. It helps in optimizing resource allocation and decision-making in these critical sectors.	Highly complex systems, security concerns, regulatory compliance	[44,45,46]
6. Retail and Customer Service	DES is employed in retail and customer service to model customer queues, store layouts, and service operations. It assists in improving customer experiences and optimizing staff allocation.	Demand variability, inventory management complexities, customer behavior prediction	[47,48,49]
7. Energy and Utilities	DES is used to model power generation and distribution systems, allowing for analysis of energy demand, reliability, and potential issues.	Complex infrastructure, regulatory challenges, fluctuating demand	[50,51,52]
8. Environmental and Ecological Modeling	DES can be used to model ecological systems and environmental processes, such as water management, ecosystem dynamics, and pollution control.	Complex interactions, long-term data requirements, unpredictability of natural systems	[53,54,55]
9. Financial Services	DES is applied in financial services for modeling trading systems, portfolio management, and risk assessment. It helps with understanding the impact of different trading strategies and market conditions.	High market volatility, reliance on predictive models, regulatory constraints	[56,57,58]
10. Simulation Gaming and Entertainment	DES is employed in the development of video games, serious games, and simulations for entertainment and training purposes. It provides a realistic and dynamic environment for users.	High demand for realism, computational intensity, rapidly evolving technology	[47,59,60]
11. Education and Training	Educational institutions and organizations use DES for training purposes, such as simulating business processes, medical procedures, and military training exercises.	Diverse educational needs, resource constraints, evolving educational methods	[61,62,63]
12. Urban Planning and Smart Cities	DES helps urban planners simulate the impact of various city development scenarios, traffic management, and infrastructure changes to optimize city designs and services.	Complex social dynamics, integrating diverse systems, long-term planning challenges	[64,65,66,67,68,69,70,71,72,73,74,75,76]
13. Telecommunications	DES assists in modeling network traffic, call centers, and communication systems to optimize resource allocation and improve service quality.	Rapid technological changes, high data traffic, complex network management	[67,68,69]
14. Pharmaceuticals and Drug Development	DES is employed for modeling drug manufacturing processes, clinical trial designs, and supply chain management.	Regulatory compliance, high accuracy requirement, complex biological systems	[70,71,72]
15. Space Exploration	NASA and other space agencies use DES for mission planning, spacecraft operations, and exploration simulations.	Extreme environmental conditions, limited data, high risk and cost	[73,74,75]

Table 1. Main fields of application of DES.

Field of Application	Main Benefits	Main Challenges	Main References
1. Manufacturing and Production	DES is extensively used in manufacturing to optimize production processes, manage resources, reduce bottlenecks, and improve overall efficiency. It helps in the design and layout of production facilities, scheduling, and quality control.	Complex system dynamics, high variability in processes, need for detailed data	[23,33,34]
2. Logistics and Supply Chain Management	DES is essential for modeling and analyzing supply chain operations, including inventory management, order processing, transportation, and distribution. It aids in optimizing logistics networks and improving delivery schedules.	Dynamic and uncertain environment, complex interaction of factors, need for real-time data	[35,36,37]
3. Healthcare	DES is employed to model patient flow, hospital operations, and emergency department scenarios. It helps in optimizing healthcare system designs, resource allocation, and patient throughput	Sensitive data handling, complex patient pathways, staff scheduling complexities	[38,39,40]
4. Transportation and Traffic Management	DES is used to simulate traffic flow, public transportation systems, and logistics networks. It aids in designing efficient traffic management systems, evaluating congestion scenarios, and optimizing transportation schedules.	Complex urban layouts, unpredictable traffic patterns, coordination with multiple agencies	[41,42,43]
5. Aerospace and Defense	The aerospace and defense industries use DES to model complex systems, such as air traffic control, military operations, and logistics. It helps in optimizing resource allocation and decision-making in these critical sectors.	Highly complex systems, security concerns, regulatory compliance	[44,45,46]
6. Retail and Customer Service	DES is employed in retail and customer service to model customer queues, store layouts, and service operations. It assists in improving customer experiences and optimizing staff allocation.	Demand variability, inventory management complexities, customer behavior prediction	[47,48,49]
7. Energy and Utilities	DES is used to model power generation and distribution systems, allowing for analysis of energy demand, reliability, and potential issues.	Complex infrastructure, regulatory challenges, fluctuating demand	[50,51,52]
8. Environmental and Ecological Modeling	DES can be used to model ecological systems and environmental processes, such as water management, ecosystem dynamics, and pollution control.	Complex interactions, long-term data requirements, unpredictability of natural systems	[53,54,55]
9. Financial Services	DES is applied in financial services for modeling trading systems, portfolio management, and risk assessment. It helps with understanding the impact of different trading strategies and market conditions.	High market volatility, reliance on predictive models, regulatory constraints	[56,57,58]
10. Simulation Gaming and Entertainment	DES is employed in the development of video games, serious games, and simulations for entertainment and training purposes. It provides a realistic and dynamic environment for users.	High demand for realism, computational intensity, rapidly evolving technology	[47,59,60]
11. Education and Training	Educational institutions and organizations use DES for training purposes, such as simulating business processes, medical procedures, and military training exercises.	Diverse educational needs, resource constraints, evolving educational methods	[61,62,63]
12. Urban Planning and Smart Cities	DES helps urban planners simulate the impact of various city development scenarios, traffic management, and infrastructure changes to optimize city designs and services.	Complex social dynamics, integrating diverse systems, long-term planning challenges	[64,65,66,67,68,69,70,71,72,73,74,75,76]
13. Telecommunications	DES assists in modeling network traffic, call centers, and communication systems to optimize resource allocation and improve service quality.	Rapid technological changes, high data traffic, complex network management	[67,68,69]
14. Pharmaceuticals and Drug Development	DES is employed for modeling drug manufacturing processes, clinical trial designs, and supply chain management.	Regulatory compliance, high accuracy requirement, complex biological systems	[70,71,72]
15. Space Exploration	NASA and other space agencies use DES for mission planning, spacecraft operations, and exploration simulations.	Extreme environmental conditions, limited data, high risk and cost	[73,74,75]

In this research, simulation is applied in the manufacturing field. However, the use of KET in production is widespread, as it allows the characteristics of a product or process to be studied and tested digitally at an early stage, preventing design errors and reducing time to market [76]. The results obtained from simulations help increase the company’s knowledge. At the strategic level, it is an important tool for constantly seeking innovation and competitive advantage, reducing the costs and risks associated with these activities [77]. Research activity on the topic has grown over the years, as evidenced by the large number of articles in the scientific literature. A quick search conducted in the Scopus scientific library, shown in Figure 1, carried out with two keywords, “Smart Manufacturing” and “DES”, returned more than one hundred articles published mainly from 2019 to the present. This has been an 85% increase from 2001, when the first publication was recorded.

Despite the wide-ranging debate on digital transformation and Industry 4.0, there is still a lack of standards and operational guidelines that can support companies in defining clear strategies for automation [78]. This deficiency is a real challenge, hindering the effective integration of technologies into production processes [79]. In the absence of a shared framework, it is difficult to identify the best practices and ensure a sustainable transition to advanced production models [80].

The problem is particularly critical for small and medium-sized enterprises (SMEs), which often operate with limited resources. Without accurate guidance, they risk making erroneous investments, such as purchasing unsuitable technologies or poorly integrating them into existing processes [9,81], with negative consequences for competitiveness [82].

In response to these difficulties, the literature proposes useful decision-making tools to guide technology choices. The study by [83], for example, shows how discrete event simulation (DES) helps reduce decision uncertainty by identifying KPIs through simulated scenarios. Similarly, ref. [84] employs DES to optimize buffer allocation in logistics flows, while [85] uses it to validate the improvement of an assembly line for PCBA. Also making a relevant contribution is [86], which integrates DES and Throughput Accounting to support medium-term decisions in manufacturing system development. Applied to a roof window manufacturing company, the study demonstrates how simulation provides strategic information that is difficult to access with traditional tools.

Despite the growing number of contributions, the literature review still shows relevant methodological and conceptual gaps. This research aims to fill in these gaps, as summarized in

Table 2. Progresses beyond the state of the art.

Literature Gap	Goal of This Article	Impact on Industry 4.0
Digital Transformation Framework	Develop a comprehensive framework guiding digital transformation in manufacturing, tailored to Industry 4.0.	Provides a roadmap for digital transition in line with emerging industrial standards.
Human-Centric Design Integration	Prioritize the integration of human factors in system design, enhancing the worker experience and ergonomic aspects.	Improves worker satisfaction and productivity, reducing risks of automation-related alienation.
Discrete-Event Simulation Application	Utilize DES to create a digital replica of the existing process, enabling detailed analysis and scenario testing.	Facilitates informed decision-making through accurate simulation and analysis.
Innovative Manufacturing Approaches	Design and validate innovative manufacturing approaches through what-if scenarios, leading to smart manufacturing solutions.	Enhances efficiency and adaptability in manufacturing processes.
Scalability Across Sectors	Demonstrate a scalable model that can be applied in various sectors, not limited to manufacturing.	Offers versatile solutions adaptable to different industries, increasing the reach of Industry 4.0 principles.
Balancing Technology and Humanity	Emphasize the importance of harmonizing technological advancements with human elements, ensuring a synergistic growth.	Ensures that technological progress supports and enhances human roles and experiences.

, contributing to advancing knowledge and improving practices in the field of industrial digitization.

However, despite the large amount of research available in the literature on this topic, a thorough analysis of published articles reveals the existence of several significant gaps. These methodological and conceptual shortcomings represent critical areas that this research aims to address. Table 2 summarizes the main gaps identified and what this work aims to do to advance knowledge and improve practices in the field of digital transformation and Industry 4.0.

In summary, the literature highlights the value of discrete event simulation (DES) as a Key Enabling Technology for digital transformation in manufacturing. However, significant gaps remain, particularly in terms of integrated methodologies, practical guidelines, and applications tailored to SMEs. These limitations underline the need for more concrete and replicable approaches. This study addresses these gaps by proposing an operational DES-based model, validated through a real case study, to support both strategic and operational decision-making in the digitalization process.

3. Materials and Methods

Discrete-event simulation is a computer modeling technique that offers an intuitive and versatile approach to reproducing the dynamic behaviors of complex systems [87]. This methodology is implemented in Witness, an interactive visual simulation software used in this research to reproduce the production process [88]. Model construction in the Witness Horizon simulation environment is characterized by an iterative process featuring four macro phases. The research methodology that enabled the development of the digital model of the real production process is divided into several consecutive stages, as shown in Figure 2.

The methodology followed in this study represents a structured and replicable approach for the analysis, modeling, and optimization of a production process through digital simulation. Below, each step is described, with emphasis on the technical aspects required to ensure the reproducibility of the entire process.

• Step 1: Process characterization and input analysis

The first phase consists of a detailed assessment of the existing production process. During a period of direct observation at the company, a systematic collection of data and information regarding the daily operation of the system was carried out. These data include the following:

• Operation cycle times: Precise measurements of the time taken to complete each stage of the production process.
• Material and product flows: Monitoring the movement of materials through the various workstations.
• Scrap levels and rework: Collecting data on quantities of defective products and rework rates.
• Resource utilization: Analysis of the use of machines, operators, and buffers along the process.

This information was used to construct a detailed flowchart of the production process, shown in Figure 3. This diagram represents the basis for subsequent digital modeling. Thorough analysis of the inputs ensures that the model accurately reflects the behavior of the real system [89]. Accuracy is ensured by verifying randomness (independence of events), homogeneity (uniformity of the data population), and goodness of fit (consistency with the theoretical distribution), through specific statistical tests. Below is a description of each test:

• Randomness: Randomness in the input data is essential to ensure that simulations are not affected by unanticipated patterns or correlations. In a discrete-event simulation, events should occur independently, reflecting the stochastic nature of the real system. To analyze randomness, statistical tests such as Pearson’s test for independence are used, which tests whether two variables are independent of each other [90]. If the variables show strong dependence, it could indicate the presence of a pattern that must be eliminated to obtain a correct simulation. Another method is the runs test, which examines the sequence of the data to detect the presence of patterns. Finally, the autocorrelation test helps identify any temporal dependencies between successive data points, ensuring that the inputs are not influenced by previous events [91].
• Homogeneity: Homogeneity of input data is crucial to ensure that all data come from the same population or that their variances are equivalent. This is especially important when modeling complex systems with multiple components or operational steps. To check for homogeneity, statistical tests such as Bartlett’s test, which assess whether multiple samples come from populations with the same variance, are used. However, because this test can be sensitive to deviations from normality, Levene’s Test offers a more robust alternative [92]. Homogeneity is further analyzed using nonparametric tests such as the Kruskal–Wallis Test, which compares the averages of multiple groups without assuming a normal distribution. In addition, visual tools such as box plots and Q-Q plots make it possible to observe the distributions of the data and visually verify the equality of variances between groups [93].
• Goodness of fit: The goodness of fit of the input data refers to how well the observed data follow an expected theoretical distribution. This is critical to ensure that the models used in the simulation accurately represent the behavior of the real system. The chi-square test is commonly used to compare the observed distribution with an expected distribution, detecting significant discrepancies. The Kolmogorov–Smirnov test, on the other hand, compares a sample distribution with a reference distribution, checking their congruence over all points in the distribution [94]. A more sensitive test, the Anderson–Darling, places more emphasis on the tails of the distribution, where discrepancies can have a significant impact on simulation results. Using these goodness of fit tests helps ensure that the input data are well represented by the theoretical models used in the simulation [95].

• Step 2: Model development and validation

Once the process analysis was completed, a digital model was constructed using Witness Horizon simulation software v.14. Model construction is an iterative process that consists of the following technical steps:

•

Insertion of physical elements: The first step is to insert physical elements representing the system’s resources into the model. These include the following:

▪

Machines: Identification and modeling of the machines used in the process, with parameters such as capacity, utilization rate, and setup times.

▪

Buffers: Definition of buffers between operations, specifying maximum capacity and material flow management policies.

▪

Operators: Modeling the workforce, with the inclusion of parameters such as availability, work times, and movement between locations.

•

Defining logical elements: After modeling the physical elements, move on to configuring the logical elements, which are needed to simulate resource behavior:

▪

Processing and waiting times: Specifying cycle times for each operation and waiting or transit times between operations.

▪

Resource capacity and availability: Configuration of resource operating limits and their time availability.

▪

Maintenance and failure parameters: Inclusion of variables to simulate scheduled maintenance events and sudden failures.

•

Implementation of connections among resources: Defining the paths between different resources, creating a network that allows simulating multiple flows and different operational sequences within the same model. This step is crucial to reproduce the complexity of real production processes.

•

Setting input/output rules: Finally, the rules governing the flow of materials and products within the system are established. These rules determine how product units move from one stage to another, what conditions must be met for transfer, and how to handle any bottlenecks.

After construction, the digital model undergoes a rigorous validation phase to ensure its accuracy. The process is inherently iterative, as the insights gained from the analysis phase feed back into the modeling phase, leading to model refinements and subsequent re-execution [96]. This cycle continues until the model accurately reflects the system or the desired insights are achieved according to the following iteration loop. This iterative, event-driven approach allows DES to model complex systems with a high level of detail and accuracy, adapting to changes and providing deep insights into system dynamics over time [97].

It involves the following:

•

Comparison with real data: The behavior of the model is compared with real data collected during the analysis phase. Matching cycle times, production rates, and resource utilization are checked.

•

Statistical analysis: Statistical tests are performed to ensure that the model results are statistically aligned with the real system, including analysis of variance and significance tests.

If the simulated model is not statistically equivalent to the real-world system, the modeler needs to identify the gaps and perform the necessary adjustments until the model’s reliability is achieved.

• Step 3: Performance analysis

Once validated, the model is used to run simulations of different operational scenarios and draw insights into the system’s performance and behavior [98]. Simulations are conducted over extended time intervals to capture operational variabilities and identify the following:

•

Bottlenecks: Identification of areas where production flow stops or slows down, causing inefficiencies.

•

Strengths: Recognition of areas of the process that work optimally, to maintain and exploit these advantages.

Simulation results, automatically generated by the software, are analyzed in detail to provide useful information for process optimization.

• Step 4: What-if analysis

With the results of simulations, “what-if” analyses are conducted to test various optimization strategies. This step includes the following:

•

Testing alternative solutions: Simulation of specific changes to the process, such as changing buffer capacities, adding resources, or changing operating shifts.

•

Evaluating the impact of changes: Each change is evaluated in terms of its impact on productivity, operating costs, and product quality before implementation in the wild.

The best optimization solution, selected through what-if analyses, is implemented in the actual production process. This final stage involves the following:

•

Implementation planning: Development of a detailed plan to introduce the necessary changes in the production system.

•

Post-implementation monitoring: Once implemented, the changes are monitored for effectiveness and to make any adjustments.

4. Reference Scenario

4.1. Company Background and Industrial Challenges

The use of simulation comes from the need to develop intelligent production in accordance with the Industry 4.0 paradigm [99]. In this research, a simulated model of the production process is developed using Witness Horizon software. The main objective is to evaluate the efficiency of the production activity and identify its strengths and weaknesses to improve the accuracy and effectiveness of business decisions, leading to the emergence and development of smart manufacturing. The case study concerns a company, located in southern Italy, that specializes in the production of aluminum cabinets intended for electric vehicle charging stations. Founded about five years ago, it has seen exponential growth in recent years due to the increased adoption of electric vehicles and government policies favoring sustainable mobility. The rapid expansion in demand has led the company to increase production from 100 cabinets per month to 100 cabinets per week, highlighting several weaknesses in the production system.

Figure 4 illustrates different activities in the production process. Photo “A” shows the spot welding applied to the doors, while photo “B” depicts the insertion phase of the inserts to which they are subjected. Photo “C” shows the laser cutting machine used to process the large aluminum sheets. Finally, photo “D” shows the press used to bend the sheets.

The company applied a zero-stock policy, making it crucial to balance production according to market demand. As a result, it is necessary to speed up production processes and reorganize production capacity to respond effectively to demand. Process management has become more complex; proper planning is needed to avoid the slowing down of production, putting stress on existing equipment, and increasing the risk of breakdowns and downtime. Maintaining high quality standards with such a high production pace is a challenge; increased production has led to an increase in defects, causing delays and additional costs. Personnel management has also become crucial, with the need to increase the workforce and ensure adequate training to maintain production efficiency and quality. Finally, logistics and distribution need to be reviewed to ensure that cabinets are delivered to customers in a timely manner, meeting growing demand.

4.2. Strategic Use of DES in Production Systems

The use of DES can offer several advantages to address these challenges [100]. DES allows the entire production process to be modeled, identifying bottlenecks and areas of inefficiency, allowing different configurations and solutions to be experimented with without disrupting real operations [101]. It also helps to plan production capacity by predicting future equipment and human resource needs [102]. Realistic simulation models can also be used to train personnel, reducing the time needed to achieve full operational efficiency [103]. The market for electric vehicle charging stations is expected to grow further in the coming years, fueled by stricter environmental policies, government incentives, growing popularity of electric vehicles among consumers, and technological innovation in charging stations [104]. The company is exploring opportunities to expand into new international markets where demand for charging infrastructure is increasing. Therefore, the company has an opportunity to consolidate its market position and address production challenges through DES implementation. Optimizing and reorganizing production capacity will enable it to respond effectively to growing demand while improving quality and reducing costs, ensuring a sustainable competitive advantage in the electric vehicle charging station market.

4.3. Production Process and Input Data Analysis

The case study in this research is about the production process involved in making the TX model, shown in Figure 5, for the 50 kW DC fast charging station, which can charge all electric and plug-in hybrid vehicles. Typical charging times range from 15 to 30 min for DC charging.

The case study of this research focuses on the production process of the “TX model” cabinet, shown in Figure 5. This model represents the most significant part of the company’s production due to its compatibility with all types of electric vehicles and plug-in hybrids, and is suitable for both indoor and outdoor installations; it is the most requested in the market. Typical charging times range from 15 to 30 min for DC charging. Maximum output power is 50 kW DC and 22 or 43 kW AC. Output voltage for charging ranges from 150 to 500 V in direct current and 400 V ± 10% in alternating current.

The production process leading up to the making of the cabinet has been schematized through a flowchart shown in Figure 4.

The cabinet emerges from the assembly of macro-elements: base, roof, profiles, doors, and covers. Considering these elements, the production process is divided into two parallel activity flows that converge into a final flow (Figure 6). The first flow is dedicated to the production of the furniture ‘skeleton’ composed of the base, roof, and profiles (blue outline in Figure 6). The activities of the second flow return to the three doors (right, left, and front) and the cover (light blue outline in Figure 6). An assembly activity combines doors and a skeleton to give life to the cabinet (gray outline in Figure 6).

4.4. Data Analysis for Process Modeling Accuracy

Cycle times related to each process activity were collected through eight iterations conducted in the field. The results of these iterations were subjected to rigorous statistical tests for compliance with three basic criteria: randomness (independence among observed events), homogeneity (uniformity of the data population), and goodness of fit (alignment with the theoretical distribution model). The results of these analyses are shown in

Table 3. Results of tests of randomness.

Test	Flow	K	p-Value	Conclusion
Runs Test (Randomness)	Structure	−0.573	0.567	Random
	Doors	−1.909	0.056	Random
	Cover	−0.382	0.703	Random
	Cabinet	−0.206	0.837	Random

,

Table 4. Results of tests of homogeneity.

Test	Flow	F-Value	p-Value	Conclusion
Levene’s Test (Variance)	All Flow	2.782	0.058	Homogeneous variances
Kruskal–Wallis Test (Distribution)	All Flow	7.564	0.056	Homogeneous distributions

and

Table 5. Results of tests of goodness of fit.

Test	Flow	Min Value	Max Value	p-Value	Conclusion
Kolmogorov–Smirnov Test (Uniformity)	Structure	9	12	0.124	Uniform
	Doors	5	10	0.212	Uniform
	Cover	8	11	0.748	Uniform
	Cabinet	10	13	0.424	Uniform

.

Statistical analysis of the cycle times of process activities provided insight into the nature of the data and their distribution, giving useful indications for accurate modeling of the processes analyzed.

Regarding independence, Runs Test confirmed that the “Structure”, “Doors”, “Cover”, and “Cabinet” streams are random, highlighting the absence of sequential dependencies. Values of p above the significance level support the hypothesis of independence of the data, ensuring that there are no repetitive patterns or hidden structures.

Homogeneity among the streams was tested using the Levene’s Test and the Kruskal–Wallis Test, the results of which show that the variances and distributions of the data are homogeneous. The absence of significant differences confirms that the streams share similar statistical characteristics, allowing them to be considered as belonging to a single coherent system.

Finally, the Kolmogorov–Smirnov Test showed that all streams follow a uniform distribution. High p-values for “Structure”, “Doors”, “Cover”, and “Cabinet” indicate that the data are well distributed over a uniform range, confirming a good statistical representation of the observed behavior.

These results provide a solid basis for further analysis and intervention, ensuring that the conclusions drawn are robust and reliable.

4.5. Digital Model: Development and Validation

The production process was reproduced in detail digitally using DES Witness software v.14. This reflects the actual production activity: two macro streams of activities that proceed in parallel and then come together in a final stream through an assembly activity. Considering the elements that compose the cabinet (roof, base, profiles, doors, and cover), the company produces the base and roof, while the other elements are purchased in their unfinished state and processed in-house. The time spent at the company enabled the acquisition of information necessary for the construction of the model. In detail, the times of each activity were timed, the interconnections between activities were defined, the distribution of workers, the type of machines used, the accumulation systems present, and the systems and times of material handling were noted. The type of production was studied, an analysis of discarded or defective products was made, and the failures and setups of automatic machines in the process were quantified. The model is shown in Figure 6. However, the model was constructed based on several assumptions:

-

Each machine represents a process activity that, based on the number of inputs and outputs of each, is distinguished into a single machine, an assembly machine, and a production machine.

-

Process inputs are modeled through parts. Some start the process; others are called precise process steps.

-

The time of each machine is equal to a uniform distribution between the maximum and minimum values measured by the timing activity.

-

Buffers, which represent accumulation systems, have a maximum capacity of 50 units.

-

In-process material handling times were not considered.

-

The screen-printing activity was not simulated in the process because of its short duration (less than one minute).

4.6. Analysis of Simulation Results

To assess the veracity and consistency of the simulation results, a statistical analysis was conducted through a detailed comparison between the process data collected in the same operational phase and the results obtained from the digital model simulation. This approach was chosen to verify the model’s ability to accurately replicate the behavior of the real system, allowing accurate validation of its performance. Process data were collected during specific operational phases of the real system, ensuring that operating conditions were well documented and reproducible. In parallel, simulation results were generated under the same operating conditions, ensuring a direct and meaningful basis for comparison. After a warm-up phase of the digital model of about 25 days, the system simulated an entire year’s production of the actual process, recording outputs monthly.

Table 6. Outputs of the digital model compared with those of the real production process.

Year 2022	Output of Real Production	Output of Digital Model
January	78	110
February	184	130
March	213	151
April	100	152
May	153	155
June	150	153
July	145	153
August	140	153
September	155	152
October	138	153
November	140	153
December	120	153

compares the outputs obtained from the simulated model with those from the real process.

To test whether the digital model estimate is indeed in line with the results observed in reality, the paired t-test was performed. A statistical tool used to compare two closely related or paired data sets to determine whether there is a significant difference between their averages [105]. The goal is to determine whether discrepancies between the two data sets are due to chance or represent a systematic difference [106].

In this case, the actual production data of a company is compared with the output of the digital model over the 12-month period of 2022 to test the accuracy of the model in predicting actual outcomes. Paired t test for samples produced the following results:

• Observed t value: −0.454
• p-value: 0.65
• Mean difference (đ): −4.33
• Standard deviation of differences: 33.03
• Degrees of freedom: 11

The comparison of the observed t-value (−0.454) with the theoretical t-value (±2.201) enhances the reliability of the simulated model, demonstrating the absence of significant differences from the real output data with respect to the digital model outputs. This is also confirmed by the p-value of 0.658, which is much higher than the commonly used significance level (α = 0.05), thus being unable to reject the null hypothesis. The direct comparison approach between the collected process data and the simulation results proved effective in validating the digital model. This allowed the analysis of the simulation results to proceed with an awareness of their validity, enabling well-founded conclusions to be drawn.

4.7. Results of Simulation: Performance Analysis

The company’s operational strategy is focused on reducing inventory to the bare minimum. As a result, production is aligned with market demand of around 500 cabinets per month. The model reflects this productive approach. To ensure an accurate representation of actual conditions, experts conducted a test by defining 2 days as the initial break-in time. This period allows the system to stabilize and optimize processes, ensuring that simulation results are reliable and consistent with the company’s actual production activity.

To have a clear and reliable view of the production activity, this was simulated for three different time intervals representing a day, a week, a month, and six months of production activity. The outputs of the digital model, representing the number of cabinets produced in these specific ranges, are shown in

Table 7. Output of simulation for different interval times.

Simulation Time	Output (Number of Cabinets)
One Working Day	30
One Working week	120
One Working month	550
Six Working months	3450

.

The extraordinary capabilities of simulation are evident at this stage. The purpose of the simulation model is to define failures and bottlenecks in the process to optimize it through the implementation of technologically advanced tools. Witness software is used to identify important processes and verify actual results based on production data. For process analysis and evaluation, detailed process performance reports, including management of external variability, are generated without any impact on actual production and in a very short time. Among the elements of analysis in the software, “statistics” are taken into consideration for this study. These are performance indicators that provide detailed information on different aspects of the simulated system [107]. In detail, in the simulation model, each component element has its own statistics that provide differentiated information, varying according to the model element analyzed [108]. Machine statistics provide, as a percentage of the total simulation duration, detailed data on several operational aspects: the actual working time of the machine, the time it was stopped due to failure, setup, or waiting for the operator. But also, whether it has been stopped due to causes related to the other elements or waiting to work. These data not only provide insight into the operational efficiency of machines, but also highlight interactions and dependencies with other elements of the system, such as parts, buffers, and workers. This brings out the specific reasons why interruptions may occur, helping to identify and resolve bottlenecks in the production process (see Figure 7).

The graph is composed of bars of different colors. Each color shows, in percentage form, the operating status of the machine. Yellow indicates the idle state of the machine and therefore waiting to work, red represents the fault state, blue shows a blocked state due to a lack of availability of material from the upstream machine, pink indicates waiting due to the unavailability of the operator who is needed for the process in certain phases, light blue indicates the setup phase to which the machine is subject, and finally, green represents the processing phase. However, given the large number of machines present in the process, for better analysis and a more in-depth study, the three macro-flows that make up the process were also studied in detail separately: structure flow statistics, statistics of doors/cover flow, statistics cabinet flow. The flow statistics are shown below in Figure 8.

Figure 9a shows the machine statistics of the process flow that characterizes the processing of doors/covers while Figure 9b represents the machine status of the shortest assembly flow of the complete cabinet.

The results show that idle state and blocking of machines occur in significant percentages. Specifically, considering the average versus the number of machines in each stream, 34% for the structure stream and 15% for the door/cover stream are blocked. These two operational flows represent the heart of the production process as they aggregate the set of fundamental activities required for cabinet assembly and construction. They are crucial not only for the optimal sequencing of operations, but also to ensure the efficiency and quality of the final product. Going deeper with the statistics broken down by flow, it is observed that the blocking state is dominant in a specific section of the process. In detail, in the machines preceding welding and adjusting, initial activities of the cabinet flow. The idle state, on the other hand, occurs in most of the machines of the door/cover flow.

For the other elements of the model, statistics are in tabular form.

Table 8. Statistics of parts.

Name	No. Entered	No. Assembled	W.I.P.	Avg W.I.P.	Avg Time	Sigma Rating
Accessories	50	0.000	50.000	49.971	457.224.350	0.000
Alluminium_sheet	339	262.000	77.000	76.866	103.732.245	6.000
Base	138	134.000	4.000	4.642	15.388.366	6.000
Console	184	134.000	50.000	50.007	124.335.138	6.000
Profile	552	536.000	16.000	15.965	13.231.583	6.000
Roof	134	134.000	0.000	0.317	1.081.590	6.000
Sheet_door	955	402.000	419.000	417.341	199.925.980	6.000

shows the statistics of the parties.

Table 9. Statistics of buffers.

Name	Total In	Total Out	Avg Size	Avg Time	Min Time	Max Time
Assemblay_Queue	34	134	0.000	0.000	0.000	0.000
Base_Xtig	138	138	0.312	1.035.764	0.000	2.261.599
Cabinet_Final	134	134	0.000	0.143	0.000	5.885
Leveling	134	134	0.000	0.053	0.000	2.611
Painting_D	549	449	99.359	82.797.515	0.000	454.048.464
Painting_In	134	134	0.000	0.022	0.000	1.168
Painting_Out	134	134	0.000	1.000	1.000	1.000
Queue_Insertion	134	134	0.000	0.000	0.000	0.000
Queue_Tf	134	134	0.000	0.000	0.000	0.000
Queue_Wt	134	134	0.000	0.000	0.000	0.000
Racker	651	601	49.478	34.770.525	0.000	452.893.952
Racker01	600	550	49.542	37.774.961	0.000	453.277.756
Racker_Cover	120	90	29.931	114.108.955	0.000	454.371.994
Racker_Cover01	165	135	30.000	83.179.062	0.000	453.853.964
Racker_Front	111	90	20.976	86.452.493	0.000	454.294.951
Racker_Front01	165	135	30.000	83.180.000	0.000	453.853.964
Racker_Left	105	91	14.024	61.103.115	0.000	453.889.478
Racker_Left01	165	135	30.000	83.180.000	0.000	453.853.964
Racker_Right	112	91	21.020	85.861.002	0.000	454.098.632
Racker_Right01	165	135	30.000	83.180.000	0.000	453.853.964
Repair_Cabinet	17	17	0.007	180.000	180.00	180.000
Repair_D	18	18	0.007	180.602	180.00	190.828
Repair_S	11	11	0.004	180.000	180.00	180.000
Roof_Accessories	50	0	49.971	457.224.350	0.000	457.479.370
Roof_Xinsertion	104	104	0.010	45.437	0.000	116.460
Roof_Xtig	134	134	0.259	885.210	0.000	2.231.831
Storage_Structure	134	134	0.126	428.535	0.000	540.141
Structure_Xassembly	134	134	0.137	468.667	0.000	962.199
Warehouse_Base	108	108	0.006	25.042	0.000	119.712
Warehouse_Console	184	134	49.99	124.310.91	121.198	454.590.00
Warehouse_Cutting	128	53	74.866	267.581.01	0.000	457.018.00
Warehouse_Extraction	212	212	0.035	76.160	0.000	135.026
Warehouse_Profile	0	0	0.000	0.000	0.000	0.000
Warehouse_Roof	104	104	0.007	30.619	0.000	82.941

shows, on the other hand, the buffers statistics. These give information on the amount and storage time of material within these buffering systems. The results reveal saturation, with a tendency to reach or approach their maximum capacity. This phenomenon is particularly evident in doors/cover accumulation systems such as “Painting_D”, where the number of inputs significantly exceeds the number of outputs (549 inputs vs. 449 outputs), with a buffer “Avg Size” of 99.359. This indicates significant accumulation, which can cause a slowdown in processing time, as evidenced by the “Max Time” of 454.048 s. Similarly, the racker accumulation systems, “Racker” and “Racker01” show a gap between input and output, accompanied by high maximum processing times (452.893 and 453.277 s, respectively), suggesting that the buffers reach their maximum capacity and negatively affect system efficiency. In addition, the buffers serving the welding task, “Base_Xtig”, while having a balance between “Total In” and “Total Out”, also have a “Max Time” of 2261 s, which could indicate episodes of temporary saturation affecting overall performance.

The numbers indicate that workloads exceed their operational capacity, which prolongs waiting times that require attention to optimize workflow and increase peak load handling capacity.

Therefore, studying the simulation results for one month of work activity, the following is observed:

-

Need for significant setup time to reach maximum capacity: The daily production of 30 lockers, compared to the monthly production of 550 lockers, indicates a significant initial ramp-up period. Comparing the outputs obtained at different time intervals, the production process needs significant setup time to reach maximum capacity.

-

Inadequacy of storage systems compared to production capacity: Buffer saturation, as evidenced by parameters such as the difference between inputs-outputs and a maximum processing time, clearly shows that storage systems are undersized compared to production outputs.

-

Inefficiencies in the production process and bottle necks: The idle and blocked state of the machines shows the inefficiency of the process. In particular, the significant percentage of blocking of the machines preceding the welding and adjusting machines, which instead turn out to be 80% busy in the simulation, identify these as bottlenecks in the process. In fact, a study of the model input data shows that these take about an hour to run, a significant amount of time compared to the rest of the process activities, which are around 20 min.

4.8. What-If Analysis

Discrete-event simulation has so far proven to be an effective tool for assessing the inefficiency of an existing system, reevaluating and designing processes, studying the impact of potential changes, and investigating the complex relationships among system variables [109]. Another great potential that characterizes it and makes it a necessary tool for change in this kind is its ability to develop what-if analysis [110]. This is a data-intensive simulation whose goal is to examine the behavior of a complex system under certain assumptions called scenarios [111]. This makes it possible to examine and delineate the proper sizing of industrial automation installed or to be installed (mobile robots, AVGs, stacker cranes, etc.), particularly in mixed production processes, i.e., where there are operator-managed stages and automation-managed stages. It allows quantifying the space and resources required for inter-operational or decoupling buffers and assessing the proper sizing of operators [112].

The critical issues that came out of the analysis of the simulation results were the starting point for the experimental analysis. The scenarios evaluated were defined by the expert team after careful study of the model with the goal of advancing toward Smart Manufacturing and ensuring an overall improvement in business efficiency and competitiveness. The choice of the proposed solutions over other alternatives was motivated by the need to achieve significant improvements in operational efficiency and production capacity. In particular, the expert team found that augmenting storage systems and implementing advanced automated technologies represented the best trade-off between cost and benefit, offering higher return-on-investment potential than other options examined.

Increased accumulation capacity ensures more efficient management of inventory and materials, reducing downtime and improving production flexibility. This places the company in a position of sustainable competitive advantage, allowing it to adapt quickly to changes in demand and maintain a high level of customer service. At the same time, the introduction of automated systems is seen as crucial to increasing productivity and improving quality. These offer higher accuracy and consistency than the manual processes typical of the process at hand, reducing the margin for human error, leading to a reduction in production defects and an improvement in the overall quality of finished products.

In parallel with a market survey of automated welding and adjustment tools, two scenarios were evaluated:

• SCENARIO 1 (S1): 30% increase in accumulation systems, implementation of an automated welding robot with a 12 min run time and an adjustment system with a 15 min run time.
• SCENARIO 2 (S2): There was a 40% increase in accumulation systems, implementation of an automated welding robot with an execution time of 12 min and an adjustment system with an execution time of 15 min.

The two proposed solutions were both simulated for one working month. The results showed greater benefits in the second case, thus with a greater increase in storage systems. This is evidenced by the results of shown in

Table 10. Comparison of simulation results between optimized and non-optimized model.

		Process “as Is”	Scenario 1	Scenario 2
% Blocked	Preparation_Roof	45%	41%	25%
	Laser_Cutting	56%	41%	23%
	Inserts	57%	45%	28%
	Fold	58%	46%	27%
	Extraction	58%	49%	26%
% Idle	Online_Testing	43%	36%	33%
	Insertion_of_componets	45%	39%	31%
	Testing	43%	33%	32%
	Testing3	45%	35%	28%
	Storage	45%	41%	29%
	Packaging	50%	44%	37%
	Assembly	50%	46%	36%
Number of cabinets manufactured		130	158	220

, where the performance indices used in the study were used to compare the “as is” process with the two proposed scenarios. S2 experiences a 70% increase in monthly production compared to the unoptimized process conditions due to a reduction in the state of machine blocking before welding and setting by 30–40% and a reduction in the idle state of the entire process by 20–30%.

The analysis conducted by the experts, together with the management team and corporate ownership, confirmed the validity of the simulation results through a rigorous verification process. This increased confidence in the accuracy of the models, which were found to be representative of complex real-world industrial systems. Based on this solid evidence, the decision was made to proceed with the practical implementation of the proposed solutions. This approach not only identified concrete and feasible improvement scenarios but also ensured that the solutions were tailored to the specific needs and operational dynamics of the company. The effective integration of advanced simulations into business decision-making established a solid foundation for successful implementation, supporting the company on its path to continuous improvement and increased competitiveness in the dynamic environment of Industry 4.0.

At present, the automated welding robot, shown in Figure 10, is being put into operation along with the high-tech adjustment systems, while the expansion of storage systems is being technically studied.

Related research to this study was conducted by [113], who studied the integration of Industry 4.0 principles in an SME specializing in the production of ambulance structures. The company faced inefficiencies due to product customization, which slowed production. Digital transformation included the adoption of an ERP system, Lean philosophy, and a reorganization of operations. The simulation tested the changes, predicting an increase in weekly production from 5.83 to 6.83 plants and a 19.34% reduction in cycle time, thus improving overall efficiency.

Instead, in the study by [114], simulation provided a measurable view of potential improvements in adopting Lean principles in an electronics assembly system. Using Witness software v.14, DES allowed testing of various scenarios to optimize operations. Results showed a 55% reduction in part holding time, model changeover time from 11 to 3 min, and a 70% decrease in inventory. Warehouse space was also reduced by 37%, demonstrating the effectiveness of simulation for more informed business decisions.

In the study by [115], simulation optimized the assembly line of an industrial gearbox by reducing the “makespan” by 94%, from 2.70 to 0.16 h. Using DES, it was possible to model and analyze the entire production flow, identifying bottlenecks and improving resource management and operational efficiency. The simulation enabled the generation of realistic production data, which was monitored in real time via the IoT platform. These data provided a detailed view of performance, allowing KPIs such as daily throughput and OEE to be calculated.

Simulation was key to identifying bottlenecks and testing improved solutions in the study conducted by [116] on a flexible manufacturing system for bearing rings. Using Tecnomatix Plant Simulation, the entire process was modeled, enabling optimization of station loading and reduction in downtime. The main bottleneck was identified in the grinding station, suggesting the addition of a parallel station to improve production capacity. Before optimization, the stations were only 25% utilized on average, with workload dispersion ranging from 21.20% to 100%. As a result of the simulation, the efficiency of the stations increased with an operating capacity of 90–100%.

How to improve the production efficiency of an industrial brewing plant was studied by [117]. The simulation identified points of process improvement in the fermentation and conditioning stages. Through what-if analyses, different operational scenarios, including the addition of tanks, were tested to increase production capacity. As a result of this solution, production capacity increased by 750% from 4.67 million to 39.76 million liters per year of filtered beer. In addition, the cycle time per batch was reduced from 8.57 days to about 24 h.

In the study by [118], simulation was used to study and optimize the production process of a mattress factory that showed several inefficiencies regarding the company’s ability to meet growing market demand. The simulated model showed uneven resource utilization: some workstations were overloaded, operating well over 100 percent of their capacity, while others were underutilized, with a utilization rate of 22 percent. The best simulated solution yielded significant results. The average production time per order was reduced from 32.09 h to 28.8 h resulting in reduced operator overload and increased overall production efficiency.

Simulation was key to improving material flow in the study conducted by [119] on an industrial material handling system. Using a digital twin-based on physical simulation, the entire material transport process was modeled, allowing optimization of parameters such as speed and acceleration in order to prevent physical disturbances, such as parts tipping over. The main problem was identified as the difficulty in maintaining the stability of materials during transport at higher speeds. Through simulation, it was possible to increase the transport speed from 5000 to 16,000 steps/s², reducing transport time by 27%. Before optimization, transport times were 2.02 s, which decreased to 1.48 s after simulation, thus improving productivity and reducing the risk of disturbance during operation.

Compared to the existing literature, this study is distinguished by its integrated methodological approach and its concrete application in a real manufacturing SME context. Unlike previously documented DES applications, which often focus on single operational improvements or simulated assessments in controlled environments, this work proposes a structured and scalable framework that combines digital modeling, what-if analysis, empirical validation, and decision support. By integrating simulation with process analysis and actual implementation of solutions, it bridges the gap between theory and practice, making an original contribution to industrial research in Industry 4.0. In addition, the proposed approach explicitly considers the role of the human factor and interaction with digital technologies, an aspect often overlooked in previous studies.

5. Future Developments

The current historical moment is characterized by an ever-faster changing demand and frequent and sudden supply shocks [120]. The increasingly customized nature of supply, coupled with other important requirements affecting all companies (both the more structured ones and small and medium-sized enterprises), the need to contain costs, reduce delivery times and promote greater flexibility, mean that manufacturing companies must become a dynamic reality, able to easily reconvert according to production needs and product specifications [121]. The set of technologies that respond to this need for flexibility and efficiency today is included among the enabling technologies of Industry 4.0 [122]. Industrial IoT makes industrial processes more efficient and interconnected [123]. A manufacturing plant is only truly modern if it can accurately capture and analyze data from smart devices that communicate with each other [124].

The importance of data collection culture cannot be underestimated in the context of Industry 4.0. The ability to capture, analyze, and use data from smart devices is essential to improve operational efficiency and support informed decisions [125]. This data-driven approach is crucial to establishing a Smart Factory that is agile and responsive to market dynamics. In this context, modelers take on a key role. Their close collaboration with manufacturers and decision makers facilitates the development of models that accurately reflect real-world systems [126]. This synergy makes it possible to design realistic and viable improvement scenarios, optimizing production processes and ensuring the feasibility of proposed solutions [127].

The integration of an advanced data management culture with interdisciplinary cooperation between modelers, manufacturers and decision makers is a crucial pillar for the success of manufacturing enterprises in the dynamic environment of Industry 4.0 [128].

To develop a manufacturing enterprise in this context, managers need to consider all aspects necessary to create the foundation for a true Smart Factory, starting with implementing the enabling technologies of the Industry 4.0 paradigm [129]. In more detail, we intend to continue this study by analyzing the following

Table 11. Technologies to be integrated with simulation for the development of smart manufacturing.

Technologies	Opportunities and Challenges	Main Reference	Application
IoT and Data analysis	The integration of simulation approaches with IoT devices enables the collection of real-time data from sensors and machines in an industrial context. These data are subsequently incorporated into a simulation model, with the aim of improving its accuracy and representativeness.	[130]	A comparison analysis, considering a production process as a case study, between a traditional DES model modeled on historical data and a digital DES model connected in real time to process activities through IoT systems. The results showed that the DES model with real-time data provides more accurate predictions of future performance, predicting significant changes where the correlation between input and output is not immediately apparent.
AI and Machine Learning	The combination of simulation approaches with AI algorithms and machine learning constitutes an advanced decision-making system. AI is used to analyze simulation results and suggest optimal changes in processes to meet ecological-environmental objectives. In parallel, machine learning models are continuously trained on simulation results and historical data to optimize process parameters and anticipate potential environmental impacts.	[131]	A data-driven framework to explore the vast design space of Additive Metal Manufacturing (AMM). The methodology provides the basis for predictively quantifying process-structure links using Machine Learning (ML) models. The application of the methodology effectively identified process-structure links in the AMM, including the scaling of microstructure data for rapid identification of process parameter combinations.
Life Cycle Assessment	The fusion of simulation approaches with Life Cycle Assessment (LCA) tools enable manufacturers to conduct an analysis of the environmental impact of their products throughout their lifetime. By simulating the production process and incorporating data from the life cycle assessment, industrial players can identify opportunities for eco-design, selection of alternative materials and development of strategies for recycling and proper disposal at the end of the product life cycle.	[132]	A sustainability assessment framework that combines life cycle analysis and DES to consider the perceived impact of stochastic processes and dynamic behavior in production systems, while also integrating social considerations. A study on the production of a metal component in the aerospace manufacturing sector is proposed. The research results highlight the usefulness of integrating LCA with tools that can capture and reveal the impact of the dynamic and stochastic behavior of production systems.
Renewable Energy Integration	Through the integration of simulation approaches with renewable energy forecasting and management tools, manufacturers can maximize the efficiency of their production schedules in accordance with the availability of clean energy sources. This synergy ensures that energy-intensive processes are planned in synchrony with periods of maximum renewable energy generation, reducing dependency on non-renewable resources and promoting environmental sustainability.	[133]	Implementing a Lean and Green (L&G) approach to assess the applicability of Lean Manufacturing (LM) tools and studying how Lean affects ecological performance in a real business case located in Brazil that manufactures PVC pipes and fittings. DES modeling is used to analyze L&G scenarios and performance. The results of the study emphasize the importance of integration with a DES model to test different scenarios, visualize results considering different levels of Kanban, and study which situation to apply, based on reality, particularly the number of acceptable configurations.
Digital Twin	The digital twin (DT) concept, which is a virtual replica of a physical object, can be synergistically integrated with simulation approaches to enable real-time monitoring and control of production processes. Such integration enables manufacturers to leverage digital twins to simulate the impact of process changes and respond promptly to changes in production parameters. This ensures increased operational efficiency and maintains a constant focus on ecological considerations.	[134]	A simulation model based on data and agents within a digital DT framework. The proposed approach is applied to a real semiconductor manufacturing system. Research results show that DTs go beyond traditional simulation with the help of real-time synchronization through industrial IoT technologies. Specifically, simulation supports off-line experimentation and planning, while DTs offer synchronous execution and modification. DTs help to understand “what can happen” and “what is happening” by allowing management methodologies to be thematically defined.

.

The integration of simulation software (such as Witness software) with other modern instruments and methodologies is expected to open up a world of possibilities for manufacturers. There are several benefits that the managers and decision-making teams can have from the application of the various KET 4.0 and other state-of-the-art systems and methodologies, as shown in Figure 11.

Integration with advanced technologies such as IoT, artificial intelligence and data analytics enables a more comprehensive and dynamic analysis of the production system. Manufacturers can better understand the interactions between different components, identify hidden inefficiencies and make more informed decisions aimed at both digital improvement and environmental sustainability. Data collected in real time keep digital models aligned with operating conditions, improving the accuracy of simulations and enabling the development of strategies consistent with the current state of processes and the surrounding environment.

This technology integration enables multidimensional optimization, balancing operational efficiency and ecological impact. Manufacturers can thus simultaneously pursue goals such as increasing productivity, reducing cycle times, and decreasing energy consumption and carbon emissions. The use of artificial intelligence, machine learning and simulation also enables the development of predictive models capable of anticipating critical issues and opportunities, enabling proactive interventions in resource management and strategic planning.

Constantly updating simulation data supports continuous adaptation of business strategies, optimizing processes in response to changes in context. Simulating alternative scenarios prior to implementation enables assessment of the environmental and operational impact of different solutions, reducing risk and improving the quality of decisions.

In addition to short-term operational benefits, the integrated adoption of discrete-event simulation and digital twin enables the pursuit of long-term sustainable benefits. These tools support more efficient and rational use of resources, contributing to the systematic reduction in waste, the design of more resilient production systems, and large-scale energy optimization. In particular, the digital twin enables continuous life-cycle monitoring of production assets, fostering predictive maintenance practices, waste reduction, and improved operating conditions. Looking forward, these approaches enable industrial strategies geared toward environmental sustainability and compliance with increasingly stringent green regulations, integrating economic goals with ecological ones and enhancing business competitiveness in the long term.

However, in an increasingly automated and data-driven manufacturing ecosystem, the human role assumes central importance. The evolution toward intelligent production models requires a rethinking of skills: operators can no longer be limited to executive tasks, but must be able to interact with digital systems, interpret data and contribute consciously to decision-making. In this context, reskilling and upskilling initiatives are key to transforming traditional skills into digital skills that can effectively dialog with automation and simulation.

Decision-making at the managerial level is also evolving: decision makers need to be able to interpret simulation results, assess the operational and environmental implications of hypothesized strategies, and take timely action. Artificial intelligence must be integrated as decision support, without replacing human judgment. In fact, the management of intelligent systems requires continuous supervision that can recognize when it is necessary to adapt to complex or unexpected contexts. In such situations, human intuition, experience and critical capacity remain indispensable elements.

6. Conclusions

Digital transformation is a systemic business strategy, applicable to any industry to address structural inefficiencies and generate new opportunities through digital technologies. In manufacturing, it involves the adoption of enabling tools to increase revenues, reduce costs, improve quality and enhance operational flexibility while pursuing industrial sustainability. The extreme variety of available solutions is a competitive advantage, but it also introduces significant management complexity, especially for small and medium-sized enterprises. In this context, this research proposes a methodological framework supported by discrete event simulation (DES) to rigorously guide the digital transition in real manufacturing environments.

• Theoretical contribution

From a theoretical perspective, the study contributes to the advancement of the literature on Industry 4.0 and Operational Research by introducing an integrated approach between digital simulation, statistical analysis of process data, and empirical validation. The proposed model is not limited to the optimization of physical flows, but includes decision-making elements and human dynamics, highlighting the relevance of human-in-the-loop design even in highly automated contexts. The contribution is in the vein of hybrid methodologies that combine simulation, decision science and digital transformation.

• Enterprise contribution

At the application level, the research provides a concrete case of using simulation to support the digital transformation journey in a manufacturing SME. The modeling enabled the identification of bottlenecks, critical operational issues and optimization margins, returning simulated alternative scenarios to support strategic decisions with low operational impact. The proposed approach is transferable and scalable to similar manufacturing contexts.

• Limitations of the study

The main limitation of the study lies in its application to a single industrial case and the adoption of a single simulation environment. Although the software used has high discrete modeling capabilities, alternative tools such as FlexSim, Simul8, or AnyLogic, which offer specific capabilities such as multi-method simulation (discrete-event, agent-based, system dynamics), advanced support for cyber-physical systems, or integration with real-time data, were not considered. Comparative evaluation could offer greater generalizability and methodological robustness.

• Future Perspectives

Development perspectives include the integration of simulation with emerging technologies such as digital twin, artificial intelligence, and IoT in order to build adaptive models based on real data. Expansion of the framework to multi-paradigm systems and integration with environmental and social metrics (e.g., Life Cycle Assessment) represent further research directions, with the goal of strengthening the sustainability and resilience of digitized production systems.