Application of Computer Simulation to Evaluate Performance Parameters of the Selective Soldering Process

Abstract

The growing complexity of production systems in the technology sector demands advanced tools to ensure efficiency, flexibility, and cost-effectiveness. This study presents the development of a simulation model for a selective soldering line at a technology manufacturing company in Poland, created during an engineering internship. Using FlexSim 24.2 software, the real production process was replicated, including input/output queues, manual insertion (MI) stations, soldering machines, and quality control points. Special emphasis was placed on implementing dynamic process logic via ProcessFlow, enabling detailed modeling of token flow and system behavior. Through experimentation, various configurations were tested to optimize process time and the number of soldering pallets in circulation. The results revealed that reducing pallets from 12 to 8 maintains process continuity while offering cost savings without impacting performance. An intuitive operator panel was also developed, allowing users to adjust parameters and monitor outcomes in real time. The project demonstrates that simulation not only supports operational decision-making and resource planning but also enhances interdisciplinary communication by visually conveying complex workflows. Ultimately, the study confirms that simulation modeling is a powerful and adaptable approach to production optimization, contributing to long-term strategic improvements and innovation in technologically advanced manufacturing environments.

1. Introduction

Modern manufacturing industries face a range of challenges arising from the dynamic development of technology, global competition, and increasing customer demands for quality, efficiency, and flexibility. Manufacturing companies must continuously adapt to changing market conditions by implementing advanced technologies and improving their processes to ensure high product quality while simultaneously reducing production costs and lead times. In this context, simulation tools play a particularly important role, as they enable comprehensive modeling, analysis, and optimization of production processes, thereby supporting informed strategic and operational decision-making.

Simulation is also a key enabler of smart factory implementations within Industry 4.0 frameworks, supporting connectivity, automation, and intelligent decision-making [1].

Production process simulation is a tool that allows real-world working conditions [2] to be replicated in a virtual environment, enabling the testing of various operating scenarios without interfering with the actual process. The use of advanced software, such as FlexSim, allows for the analysis of material flow, identification of bottlenecks, evaluation of machine and resource performance, and assessment of how changes affect the entire production system [3,4]. In the context of electronics manufacturing where precision and technological complexity are critical, simulation becomes an indispensable element that supports the development and refinement of processes [5].

Moreover, simulation contributes to the design of sustainable manufacturing systems by enabling detailed evaluation of environmental and operational impacts [6].

One of the most demanding stages in the manufacturing of electronic assemblies is the selective soldering process, which requires precise joining of specific points on printed circuit boards. Selective soldering enables the creation of durable and reliable solder joints, but it requires advanced planning and accurate synchronization of resources and operations. Introducing a simulation tool makes it possible to analyze critical aspects of the process, such as cycle time, operation sequence, resource availability, and material flow, thereby improving the efficiency of the entire system.

The aim of this study was to develop a simulation model of the selective soldering process for a technology company in Poland. The work focuses on creating a tool to support the analysis and improvement of this key process while also examining potential execution scenarios. The results of the study serve two main objectives. First, they provide detailed insights into opportunities for improving the efficiency of the selective soldering process at the company, leading to better resource utilization and reduced production lead time. Second, the project aims to develop a universal tool that can be used in future efforts to analyze and optimize other production processes within the company. As such, the simulation model not only supports the company’s current operational needs but also serves as a foundation for its long-term development strategy and adaptation to future technological and market challenges.

Upon reviewing the literature related to the topic under investigation, we observed that simulations are most commonly conducted for specific, case-based scenarios. In such studies, the majority of process parameters are incorporated into the simulation model as they are encountered in real-world conditions. Consequently, our focus was directed toward the development of a universal and adaptable tool. As a result of this effort, a flexible framework was created, capable of handling various process durations and allowing for easy reconfiguration of equipment within the production line.

2. Materials and Methods

This chapter presents an overview of computer simulation in manufacturing processes, highlights the significance of selective soldering, and provides justification for selecting FlexSim as the simulation tool for the process.

2.1. Simulation

Simulation is a way of recreating a real-world system using a model, which allows for experiments and analysis to be carried out. Such a model can reflect both the physical environment and the logical rules that govern how the system works. Simulation goes beyond simple imitation in that it also includes building the model, designing experiments, preparing input data, and analyzing the results [7].

This approach helps in testing systems, understanding their behavior, generating new ideas, and improving existing processes. Depending on how the system is modeled and how it changes over time, simulations can be grouped into different types [8]. These include the following:

• Continuous simulation models systems with continuous state changes over time, often using differential equations. It is commonly used in dynamic system analysis and feedback modeling [9,10,11,12].
• Agent-based simulation (ABS) models systems as collections of autonomous agents whose interactions lead to emergent behavior. ABS is suited for decentralized, complex systems [13,14].
• Monte Carlo simulation relies on random sampling to analyze systems under uncertainty, supporting risk assessment in fields such as manufacturing and supply chains [8,15,16].
• Discrete-event simulation (DES) is a modeling approach used for systems where changes in state occur at specific, discrete points in time referred to as events. These changes are typically triggered by occurrences such as the start or completion of operations, the arrival of materials, or changes in resource availability. Because of this, DES is especially suitable for modeling production systems, logistics operations, and queuing environments.
• Simulation is widely applied across diverse industries, including chemical process design for sustainable fuel production, where it supports efficient plant layout and process optimization [17].

The structure of DES relies on a simulation clock and an event list, allowing the model to progress step-by-step from one event to the next. This makes it possible to accurately represent dynamic changes in the system without the need to track continuous state fluctuations. Furthermore, the appropriate granularity of activity recognition significantly influences the accuracy and applicability of simulation models, as demonstrated in heavy civil engineering contexts [18]. DES can be implemented using different methods, such as traditional event list models or agent-based event handling. Recent studies emphasize the integration of simulation optimization approaches with adaptive decision support systems to meet the dynamic requirements of Industry 5.0 environments, enabling flexible and resilient production scheduling [19]. Thanks to its high flexibility and precision, DES has become a widely used tool in industrial engineering. It supports process optimization, identification of bottlenecks, and evaluation of various production scenarios [7,8].

Recent studies also highlight the integration of simulation with optimization techniques to enhance production system performance, flexibility, and adaptability to changing conditions [20].

For instance, Lewicki [21] applied DES to simulate production line downtimes, while Pawlak et al. [22] used it to compare planned and simulated production outcomes, both confirmed by real-world data. DES is particularly effective in environments where activities are performed in stages and transitions occur irregularly over time. Typical examples include manufacturing lines, computer networks, and mining operations [23,24]. In this study, discrete-event simulation was the core method used to model the selective soldering process. This type of soldering, commonly found in electronics manufacturing, consists of a series of discrete steps: manual component insertion, transport of pallets, soldering by a robotic nozzle, and quality control touch-ups. Each of these operations marks a distinct event in the production timeline, making DES an ideal modeling framework.

Using DES enabled a detailed and realistic representation of material flow, processing times, queue behavior, and resource utilization throughout the selective soldering line. It also allowed us to test different scenarios such as varying input rates or machine availability without interfering with actual production. As such, this approach provided valuable insights into process efficiency, potential improvements, and system behavior under varying conditions [25].

A wide range of algorithms is now available for simulation optimization, supporting complex manufacturing decisions under uncertainty [26].

2.2. Selective Soldering

Selective soldering is an advanced technology used in electronics assembly that enables the precise soldering of specific points on printed circuit boards (PCBs). This process is particularly useful for devices with complex designs, where soldering all connections simultaneously, as in wave soldering, could damage sensitive components or lead to defects. Selective soldering allows for localized soldering without affecting adjacent elements.

Stages of the Selective Soldering Process [27]:

• Flux application: Flux is applied to selected points on the printed circuit board (PCB) to remove oxides from the soldered surfaces and improve wettability. The application is performed using precisely controlled nozzles to avoid contamination in other areas of the board.
• Preheating: The board undergoes controlled preheating to reduce the risk of thermal stress. Preheating prepares the PCB for the actual soldering process while ensuring that its components are not damaged.
• Soldering: The soldering process is carried out using soldering nozzles that apply molten solder alloy to the selected points on the board. The use of a protective atmosphere, typically nitrogen, minimizes solder oxidation, resulting in high-quality joints. The process is highly precise, making it suitable for working with complex and densely populated PCBs.
• Cooling: After soldering is completed, the board is cooled in a controlled manner, allowing the solder to solidify quickly and the joints to stabilize [28].

Applications of Selective Soldering:

Selective soldering is primarily used for boards with mixed assembly of THT (Through-Hole Technology) and SMT (Surface Mount Technology) components, where it is necessary to protect sensitive elements. This technology is also employed in the production of devices that require high precision and reliability, such as medical, telecommunications, and automotive equipment [29].

Advantages of Selective Soldering:

• Precision: This technology enables soldering of only selected points, helping to avoid defects in other areas of the PCB.
• Component protection: The process minimizes the risk of thermal damage to sensitive components on the board.
• Flexibility: Selective soldering is well-suited for complex board layouts and various types of components.
• Joint quality: A protective atmosphere and precise control of process parameters ensure high-quality solder joints.
• Material savings: Accurate dosing of flux and solder reduces material consumption, thereby lowering production costs [30].

Recent research demonstrates that integrating AI-based optimization algorithms with simulation models can significantly improve the efficiency and quality of selective soldering processes, enabling more robust and adaptive manufacturing systems [31].

Limitations of Selective Soldering:

• Longer process time: Selective soldering is slower compared to wave soldering, as only specific points are soldered.
• Initial costs: Implementing the technology requires the purchase of advanced soldering equipment and training for operators.

In summary, selective soldering is precise and versatile but slower and more expensive than other soldering techniques. Nevertheless, it proves especially valuable in projects with a high level of complexity.

2.3. Factors Influencing the Selection of a Simulation Tool

Selecting the right simulation software is a critical step in effectively modeling and analyzing manufacturing systems. The choice should be guided not only by general functionality but also by how well the tool aligns with the specific nature of the modeled process in this case, the selective soldering process. A range of simulation platforms is available on the market, including general-purpose environments such as AnyLogic, SIMUL8, SIMIO, and FlexSim, each offering different strengths. However, not all of them provide the same level of specialization and user friendliness when applied to complex industrial processes.

AnyLogic is a flexible simulation platform that supports multiple modeling approaches, including discrete-event, agent-based, and system dynamics. It is well suited for hybrid models and large-scale strategic simulations, particularly in research and multidisciplinary projects. However, its general-purpose nature often requires more extensive customization to accurately model detailed industrial operations.

SIMUL8 is known for its simplicity and ease of use. It is commonly applied in service systems and healthcare process modeling. While its intuitive interface is accessible to beginners, SIMUL8 may be limited in terms of advanced features, detailed 3D visualization, and integration with external engineering tools, making it less suitable for complex manufacturing processes.

SIMIO combines object-oriented modeling with a visual process design environment. It offers strong support for scheduling and real-time simulation, and it is particularly useful for capacity planning and operational decision support. However, its learning curve and license structure can be challenging for smaller or highly specialized projects.

Among these tools, FlexSim stands out as one of the most advanced simulation environments, specifically developed for manufacturing and logistics systems. It offers a unique combination of usability, modeling depth, and industry-relevant features that make it especially effective in real-world production applications [32,33].

The decision to use FlexSim in this study was based on both its proven track record in industrial use cases and its rich feature set. A number of recent studies confirm the practical value of FlexSim in optimizing complex production systems:

• In a pharmaceutical production study, FlexSim was used to model tablet manufacturing under various shift systems. The simulation enabled campaign time reductions of 50–53% and labor savings of 9–14%, without affecting actual operations [34].
• In a filling machine optimization project, FlexSim helped implement a mathematical model that allowed simultaneous multi-flavor yogurt filling, significantly reducing process time and increasing flexibility [35].
• Another study applied FlexSim to identify system bottlenecks and evaluate production layout redesign. The simulation revealed critical inefficiencies and supported improvements in workforce utilization [21]. Additionally, a Polish case study demonstrated that FlexSim effectively models complex production logistics in 3D and facilitates optimization of throughput and resource allocation [36].
• In the logistics domain, FlexSim was used to optimize material replenishment processes. The integration of lean manufacturing principles led to a measurable increase in operational efficiency [37].

Additionally, simulation has been successfully applied to optimize PCB manufacturing processes, where detailed models of production lines enabled better decision-making in layout design and process planning [33].

Beyond its empirical success, FlexSim offers a number of technical advantages that make it particularly suitable for industrial process modeling:

• Seamless data integration from Excel and CAD programs (e.g., AutoCAD 2024, SketchUp), allowing rapid updates and compatibility with existing engineering documentation.
• Built-in SQL support, enabling complex queries and efficient management of simulation-related data within the model.
• A visual Process Flow panel that facilitates the creation of simulation logic using prebuilt or custom blocks ideal for intuitive, flexible model development.
• A broad 3D object library and realistic textures that help create visually engaging models, even for non-technical stakeholders.
• High modeling flexibility through a drag-and-drop interface, enabling rapid prototyping and scenario testing in complex environments [38].

Thanks to these capabilities, FlexSim not only accelerates model development but also ensures that simulation results are understandable and actionable for decision-makers across technical and managerial roles.

Given the complexity and event-driven nature of the selective soldering process, which involves multiple precise stages and time sensitive operations, FlexSim provides the most appropriate simulation environment. Its discrete-event engine is capable of accurately capturing machine cycles, buffer delays, and operator interaction factors essential for improving process reliability and performance.

In summary, while various simulation tools offer different benefits, FlexSim clearly emerges as the most effective and comprehensive solution for modeling discrete, event-driven production systems. Its application in this study is fully justified by both the technical demands of the selective soldering process and the broader requirements of modern industrial simulation [21,22].

3. Results

This chapter presents the process of constructing a simulation model aimed at optimizing the selective soldering process. In addition to the process parameters and their interrelationships, the development of visual elements representing the process is also described.

3.1. Constructing the Simulation

In the first stage of model development, the necessary 3D objects were transferred from the FlexSim library into the workspace (Figure 1). These objects were then visually customized and parameterized to accurately reflect the real-world process (Figure 2). Input and output buffers were appropriately adapted, assigned new names corresponding to their functions, and visually adjusted to resemble actual warehouse carts, while retaining their default functional settings.

The MI (Material Insertion) workstations were designed in a modular way, allowing for flexible adaptation of the model to a variable number of operators and easy management of the line layout.

Each workstation includes an operator, a roller conveyor, a processor responsible for operation time, and a decision point supporting flow control.

The selective soldering station and the inspection station were also built using the “Processor” object, whose appearance was modified using 3D models from the SketchUp platform to better reflect the characteristics of real equipment. An operator was also added to the inspection station; however, due to their transport-related role, the operator was not assigned to the structure of the “plane” object.

In summary, all 3D objects in the model were visually and functionally adapted to reflect real process conditions and enable effective simulation.

3.2. Configuration of Flow Elements (PCBs and Soldering Carriers/Pallets)

In the “FlowItem Bin” tab in FlexSim, it is possible to edit existing objects representing basic shapes that simulate flow elements in the process. This can be observed in the case of the simulated soldering pallet (Figure 3, left side). Alternatively, as shown in the soldering machine example, it is possible to upload a prepared skp file, as demonstrated with the PCB model (Figure 3, right side).

3.3. Variable Table and Import of Process Data

After placing the objects within the model space, key variables controlling the simulation were defined. The main objective was to determine the minimum number of soldering pallets required in the initial buffer so that with a given number of MI stations and process times the production process would run smoothly and without interruptions. The number of pallets after which the operator stops inspection and returns them to the beginning of the process was also taken into account, as it affects both efficiency and flow continuity.

The variable table (

Table 1. Input data for the model.

Variable	Value	Unit
ModelTime	0	(hours)
NumOfStations	2	(pcs)
NumOfCarriers	12	(pcs)
BatchOutCarriers	4	(pcs)
ConveyorLength	1	(m)

Source: Table developed based on in situ observations.

) includes parameter names, their values, and units of measurement. The most important variables include the following:

• ModelTime—processing time of the entire planned batch;
• NumOfStations—number of MI workstations;
• NumOfCarriers—number of pallets in circulation;
• BatchOutCarriers—number of pallets after which the operator moves them to the buffer;
• ConveyorLength—length of roller conveyors.

The process time table (

Table 2. Process times for selected products from the selective soldering line.

Assembly	MI 1	MI 2	Mach	Touch Up	Sequence	Batch_Size	BatchPerCarrier
227091-500	276.92	276.92	171	276.92	7	33	4
227109-502	600	600	228	600	1	40	4
227463-500	327.27	327.27	285	240	3	50	5
227529-000	180	180	178.6	180	4	42	2
227581-201	190	190	263.07	190	6	60	3
227581-201KW	190	190	263.07	190	2	36	3
30325600-I	150	150	600	600	5	52	4
30325610-I	150	150	190	190	0	0	0
DTR0000239370	228	228	244.28	190	8	64	5
DTR0000240739	123.5	123.5	131.67	131.67	0	0	0

Source: Table developed based on in situ observations and extracted process data.

) contains data for individual types of PCBs, including the following:

• Product identifier (Assembly) [string value];
• Operation times at MI stations, the soldering machine, and inspection (MI 1, MI 2, Mach, Touch Up) [seconds];
• Processing sequence (Sequence) [seconds];
• Batch size (Batch_Size) and number of products per pallet (BatchPerCarrier) [integer value].

After importing the tables containing process times and model variables, values were assigned to the “Parameters” and “PerformanceMeasures” tabs. In the “Parameters” tab, the number of soldering stations (“NumOfStations”) and the number of pallets after which the operator moves them to the initial buffer (“BatchOutCarriers”) were defined (Figure 4). By linking these variables, it becomes possible to use the Experimenter tool in FlexSim at a later stage. Meanwhile, the “PerformanceMeasures” tab defines the optimization criterion in this case, “ModelTime”, which represents the total processing time for all PCB units (Figure 5).

3.4. Navigation Panel for the Model

The next step was to create a navigation panel for the model, aimed at facilitating the adjustment of the soldering line to the current needs of the plant’s engineering team.

The panel includes three buttons that trigger embedded scripts, as well as two sliders that allow stepwise changes in the values displayed to their left (Figure 6).

The control panel overview begins with a description of the sliders, which are linked to the corresponding table values. These sliders are directly associated with the variables representing the number of workstations (“NumOfStations”) and the length of the roller conveyors (“ConveyorLength”). Their use enables the user to intuitively and incrementally adjust these parameters—by 1 unit for the number of stations and by 10 cm for conveyor length. As a result, the model interface becomes clear and user-friendly. Additionally, the value display labels are also linked to the appropriate table entries, ensuring data consistency and real-time updates.

A key element of the interface is the “Build Model” button, which is responsible for automatically generating the assembly line model based on the entered data. Pressing this button creates the specified number of manual insertion (MI) workstations, builds all necessary directional connections for pallet flow, and adjusts the positions of other model components to fit the current configuration. The logic of the button’s operation was implemented in FlexScript (Figure 7).

The script consists of four main sections. The first part of the code is responsible for assigning object variables to the key components present in the model. In the following lines, a “TaskExecuter” object is also created, in this case representing the inspection operator. Next, control variable values, namely the number of MI workstations and the conveyor length, are assigned based on the previously defined parameter table. At this stage, group variables are also defined to facilitate later management of model elements, including their removal.

The next section of the script is based on two main control structures. The first is a conditional statement that checks whether the initial buffer has any active directional connections, which prevents overwriting the existing configuration. The second structure is an iteration loop, where the number of repetitions corresponds to the required number of MI stations. Within the loop, a copy of the template workstation object is created, and its internal components are extracted. Each new instance is assigned a unique name and added to the appropriate groups. Its position and the length of its associated roller conveyors are also defined.

In the final stages of the script, all necessary directional connections are created, and the positions of remaining objects are adjusted to ensure the full functionality of the line. Lastly, all decision points previously added for testing and flow control in earlier versions of the model are removed from the simulation.

3.5. Generation of Simulation Flow Elements

The first step involved creating logic that generates digital representations of the previously imported quantities of PCBs and soldering pallets used in the process (Figure 8). A block was also parameterized to assign labels to the token, representing all essential elements of the model. These labels greatly simplify the development of simulation logic by providing clear references to relevant components.

Analyzing this section of the “ProcessFlow” from the top, we can first observe a gate named “secure block”, which safeguards the model startup against inconsistencies such as an incorrect number of constructed MI workstations compared to the data from the process table. In the event of such a mismatch, the token generated in the “source” block is redirected to a “STOP” location, where the entire process is terminated.

Further down, the “actual products list” block remains visible, with its script shown in Figure 9. This script generates a derived table (Figure 10) based on the one previously imported from Excel (Table 1) and uses SQL syntax to sequence it according to the values specified in the “Sequence” column (Table 1).

Moving forward, two consecutive “Run Sub Flow” blocks were created (Figure 11), which in FlexSim serve the function of building iterative loops.

In this part of the logic, based on the previously created and sorted process data table and the model variable table, the appropriate quantities of PCBs and soldering pallets are generated, along with corresponding labels, at the initial queues. Figure 12 presents the labels added to the PCBs, which will facilitate the creation of code for reading the correct processing times at specific workstations.

3.6. Placing PCBs onto Soldering Pallets

In the following section, logic was created to place PCBs onto soldering pallets (Figure 13). Starting again from the top, the “Assign Dynamic Labels” block (Figure 14) sets all necessary numerical labels, which are later used in the “Run Sub Flow” blocks to define the number of iterations for specific actions assigned to individual products.

Moving further into the first nested Sub Flow loop, there is a “Wait for Event” block that waits for the arrival of a pallet at the location shown in Figure 15, then passes the token to the second nested loop, where the following actions are performed sequentially:

• “Load” and “Unload”, which are executed by the MI operator closest to the input queues. These are purely visual aspects of the model and therefore do not require any processing time. They are performed instantly and serve to represent the number of PCBs moving on the soldering pallets.
• “Set boards location”, responsible for properly positioning the boards at equal distances depending on the required quantity.

The next step is to verify whether the next batch of PCBs in the input buffer matches the quantity that, according to the process table, the operator should place on the soldering pallet. For example, if there are two boards in the queue but the batch size assigned to that product is 4, the token enters the “Fill unful batch” block. In this block, the number of iterations equals the amount that can actually be processed, after which the token returns to the sequence that arranges the boards on the soldering pallet. This is implemented by a script, a fragment of which is shown in Figure 16.

After loading the boards onto the soldering pallet, the token enters the “Boards to process” gate (Figure 17), which checks whether there are still PCBs in the initial buffer to be processed. If so, the token exits through the output labeled “C” and returns to the board placement logic (Figure 13). If not, it proceeds sequentially to the “End Model Time” and “STOP” blocks.

The first block is responsible for recording the time required to process all products into the “ModelTime” row in Table 1. The second block stops the simulation model run (Figure 18).

3.7. Calibration of Process Data Reading at Process Stations

The final stage in the model calibration process was the development of a script that automatically adjusts processing times for selected products at all stations included in the model (Figure 19).

In lines 9 and 10 of the script, two key pieces of information are retrieved: the “Sequence” label assigned to the PCB, which corresponds to the row in the process data table (Table 1), and the station name, which is also the name of the column from which the appropriate processing time value should be obtained. This approach allows the model to dynamically adjust process parameters, ensuring an accurate representation of real production conditions.

4. Discussion

Due to the proper parameterization of 3D objects aligned with the flow logic reflecting the real process, and the use of tools for measuring the soldering line’s performance, the model enables the use of a tab dedicated to testing various scenarios. This functionality allows for the analysis of the most efficient process solutions with respect to predefined objective functions (Figure 20), which optimize parameters such as “ModelTime” (processing time) and “NumOfCarriers” (number of pallets in circulation).

Figure 21 presents the point solutions that are the most rational for a given production plan.

For the coefficients defining the number of soldering pallets relative to the number of processed pallets after which the inspection operator should move them to the initial buffer, the values are as follows:

• 11–6;
• 8–3;
• 7–6.

All these values correspond to a production plan scenario with an estimated duration ranging from 10 to 14 h. However, this range is too broad to allow for a reliable and precise evaluation of the efficiency of the individual options.

To obtain a more precise set of solutions, we used the “Experimenter” tab, which allows manual definition of scenarios for analysis. The values I tested are presented in Figure 22 as raw data tables and charts. The scenario analysis was conducted with three repetitions to assess potential variations in the results. However, due to the definitively specified process times derived from the input data table and the absence of stochastic elements in the process, these repetitions proved unnecessary.

As the analysis shows, the tested scenarios for configurations 6–3, 6–4, and 8–6 significantly deviate from the other results. Most of the results, marked with boxes in Figure 22, reach the same value of 10.69 h. Among all the options considered, the most rational solution appears to be the scenario requiring the smallest number of soldering pallets. This scenario is marked with a blue box in Figure 22 and corresponds to the 8–4 configuration.

The final result of the analysis conducted for the batch of products from Table 1 is a reduction in the number of soldering pallets from the previous 12 to 8 (Figure 23). The simulation confirmed that this change will not negatively impact process efficiency. This means that the same number of products will be processed using a reduced, thoroughly tested number of soldering pallets in the model, as well as the threshold at which the inspection operator returns the pallets to the initial queue.

After selecting the most relevant solutions for a given production plan, it is possible to perform a comprehensive interpretation of the corresponding statistics, including resource utilization and line efficiency (Figure 23).

5. Conclusions

The objective of the project was to develop a simulation model of the selective soldering line to optimize production processes at a technology company in Poland, where I had the pleasure of spending time during my internship. The work involved replicating the real production process by analyzing key technological and operational elements while utilizing FlexSim software for modeling and simulation.

Based on the conducted production process analysis, a detailed representation of each stage was created, including the functions of input and output queues, Manual Insertion (MI) workstations, the selective soldering machine, and the touch-up station. The simulation model accounted for elements such as buffering, transport of soldering pallets, and quality control at every stage of the process. An important aspect was the implementation of the “ProcessFlow” logic, which accurately reflected token flow dynamics within the model and mirrored the real process behavior.

Due to the modular design of the model and advanced parameterization, a tool was created that enables the analysis of various production scenarios. Using the “Experimenter” tab in FlexSim, alternative process layouts and configurations were tested to identify the most rational solutions. In particular, the process time (ModelTime) and the number of pallets in circulation (NumOfCarriers) were optimized.

At the final stage of the work, an intuitive navigation panel was developed, allowing engineers to easily operate the model, adjust process parameters, and monitor production line performance. Equipped with visualizations and statistics, this panel provides full insight into the model’s operation and facilitates decision-making related to process configuration.

A key outcome of the project was identifying the potential to reduce the number of soldering pallets in the process from 12 to 8 for specific quantities of PCBs to be soldered. Simulations confirmed that this reduction does not negatively affect process efficiency, and production proceeds without downtime. Such optimization allows for resource reduction while maintaining the required efficiency. The average cost of one soldering pallet is approximately USD 2000, which for the implemented reduction translates into savings of about USD 8000. While this may not be a dramatically high saving at this stage, considering the universality of the model, which can adapt to soldering lines with any requirements, and the fact that several such lines operate within the company’s facility, avoiding unnecessary costs over time can bring significant financial benefits, as well as reduce inaccuracies related to poor estimation of pallet quantities on specific production lines. Additionally, the careful visual representation of the model may serve in the future as a communication tool between engineers and human resources. By clearly and visually presenting the process and collaboration points, the model facilitates understanding and effective information exchange between teams with diverse competencies and specializations.

A crucial phase of the project was the validation of the simulation model against real-world data collected from the production line during the internship. This process ensured that the model’s logic, timing, and resource flows faithfully reflected the behavior of the actual system. The validation procedure involved a comparative analysis between the model outputs and operational data derived from machine logs, manual observations, and production reports. Key performance indicators such as total process time, station-level utilization, average PCB cycle time, and pallet rotation frequency were measured.

The outcome of the validation was a very high conformity score of 94.7%, which strongly indicates that the model accurately reproduces real-life production dynamics. The minor deviations observed (5.3%) were mainly associated with human variability during manual insertion tasks, short-term machine disturbances, and micro-delays not modeled in the simulation to preserve model universality.

To further reinforce the credibility of the model, scenario-based tests and sensitivity analyses were conducted. Known disturbances, such as temporary workstation idleness or shifts in the timing of pallet returns, were introduced in both the real process and the model. The system’s reaction in terms of buffer buildup, decreased throughput, and reallocation of resources showed high consistency, confirming the model’s predictive capability.

In addition to numerical comparisons, visual validation was employed. Engineers familiar with the production process were invited to review the animated simulation. Thanks to FlexSim’s detailed 3D modeling, the flow of materials, worker actions, and machine operations could be visually confirmed as accurate. The qualitative feedback from these experts supported the conclusion that the model realistically and reliably represented the selective soldering process.

With such a robust validation foundation, the model became not only an analytical tool but also a reliable forecasting and decision support environment. It can now be used to assess the consequences of layout changes, pallet number variations, and process sequencing, without disrupting the actual production line.

The summary of benefits presented in this work clearly indicates that simulation models are indispensable tools in production process optimization. They enable detailed replication of real systems in a virtual environment, allowing for in-depth analysis and testing of various scenarios without risking disruption of ongoing operations. Through simulation, bottlenecks can be precisely identified, material flows improved, resources efficiently utilized, and the best operational strategies selected, thereby supporting the development and competitiveness of the enterprise.

FlexSim stands out as an advanced discrete-event simulation tool offering a wide range of features that support process modeling and analysis. Its ability to import data from Excel, AutoCAD, and SketchUp significantly accelerates model building and integration with existing documentation. Built-in SQL support simplifies data management, while the Process Flow panel allows intuitive creation of process logic using both prebuilt blocks and custom solutions.

In summary, simulation is the key to understanding and improving complex processes, enabling informed decision-making within a safe virtual environment, where risk becomes a learning tool and optimization forms the foundation of innovation.