Next Article in Journal
Sampling Method Based on Fuzzy Membership for Computing Negative Sample Credibility and Its Applications
Previous Article in Journal
Mobile Mapping Approach to Apply Innovative Approaches for Real Estate Asset Management: A Case Study
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Data-Driven Optimization of CNC Manufacturing Using Simulation and DOE Techniques

1
A. Leon Linton Department of Mechanical, Robotics and Industrial Engineering, College of Engineering, Southfield, MI 48075, USA
2
Engineering Management Département, College of Engineering, Prince Sultan University, Riyadh 11586, Saudi Arabia
*
Author to whom correspondence should be addressed.
Appl. Sci. 2025, 15(14), 7637; https://doi.org/10.3390/app15147637
Submission received: 14 May 2025 / Revised: 25 June 2025 / Accepted: 4 July 2025 / Published: 8 July 2025

Abstract

In the highly competitive manufacturing environment of today, operational success depends on increasing efficiency and cutting waste. The goal of this research is to use Arena simulation software to model a CNC production system to assess and improve system performance. Three different parts are processed by the model, which also includes rework loops in which a portion of faulty products are sent back for further processing. Finding bottlenecks, evaluating important performance metrics such as output, queue lengths, waiting times, and machine utilization, and testing improvement scenarios are the primary goals. Study findings indicate that waiting times were greatly shortened and resource usage was balanced in alternative scenarios, which were accomplished by shifting workloads, line balancing, and modifying inter-arrival durations. The results demonstrate how well simulation can represent and resolve inefficiencies in intricate industrial systems. Manufacturers may optimize manufacturing processes without interfering with ongoing operations thanks to this method, which encourages educated decision-making.

1. Introduction

Computer numerical control (CNC) machining is foundational to modern manufacturing, enabling the production of highly precise and complex parts with minimal human intervention. As global industries move toward smart manufacturing and Industry 4.0 paradigms, the role of CNC systems has expanded beyond precision to include adaptability, integration, and operational efficiency. These shifts compel manufacturers to re-evaluate how CNC operations can be optimized not just at the machine level but across entire production systems to reduce downtime, eliminate waste, and maximize throughput [1,2,3].
In this context, simulation modeling has emerged as a powerful tool to analyze and improve manufacturing processes. This study employs Arena simulation software, v15 to model a CNC production line, focusing on critical performance indicators such as production output, queue behavior, cycle time, machine utilization, and work-in-progress (WIP) inventory. To enhance the model’s realism and practical relevance, a rework loop is integrated, where 10% of Part 1, 15% of Part 2, and 5% of Part 3 are returned to the CNC turning operation for corrective action. This feature simulates quality-related feedback loops commonly found in actual manufacturing environments, adding complexity and reflecting the practical effects of quality control on system performance.

2. Motivation and Research Gap

While numerous studies address CNC optimization, the majority tend to isolate specific machine parameters or rely on idealized scheduling algorithms, offering limited insight into the system-wide operational dynamics of CNC environments. In particular, few studies consider closed-loop production scenarios involving rework, even though such conditions are prevalent in real manufacturing systems and have significant implications for system throughput and resource utilization.
Additionally, current research often overlooks how feedback mechanisms, such as quality-driven reprocessing, affect key performance indicators across interconnected processes. This gap in the literature limits the effectiveness of existing models in supporting data-driven decisions at the managerial level.
This study addresses this gap by proposing an integrated simulation model that captures the interplay between quality control loops and production efficiency in CNC systems. The model offers a holistic view of the manufacturing environment, incorporating both stochastic variability and practical constraints to provide the following:
  • A detailed analysis of bottleneck points in the CNC process;
  • Insights into how rework rates impact resource utilization and cycle times;
  • Simulation-based evaluation of alternative process enhancement strategies.
By focusing on system-level optimization under real-world constraints, this research contributes a valuable framework for performance evaluation and process improvement in CNC machining environments. It ultimately seeks to support data-driven, operational decision-making aimed at enhancing manufacturing resilience and productivity.

3. Literature Review

A key component of manufacturing system optimization, especially in settings with CNC machining, is the examination of part flow in a production process. Minimal delays, a decrease in work-in-progress (WIP), and optimal machine utilization are all guaranteed by efficient part flow. A crucial tool for examining and visualizing part movements in a variety of systems is discrete event simulation (DES). Due to its versatility in simulating large systems with numerous process stages, decision points, and rework loops, Arena simulation software is one of the most used DES tools. Manufacturers can experiment with various layouts and flow techniques without interfering with real-world production by using simulation, according to [4].
With precise and sequential operations like turning, milling, drilling, and inspection, CNC machining cells are frequently at the center of the production flow in modern manufacturing. Flow modeling becomes more complex when parts undergo various operations because of their dynamic behavior, particularly when rework is included. Accurate simulation of such systems can reveal information about performance measures like throughput, queue length, cycle time, and bottlenecks, according to research by [5,6]. For simulation models to accurately depict system behavior, rework—a major hindrance to smooth component flow—must be included. Rework loops, according to [4], result in longer cycle times, congestion at crucial equipment, and decreased overall efficiency, particularly when machines such as CNC turning are overworked because of the repeated re-entry of defective parts.
Furthermore, process timings can be modeled using suitable probability distributions by statistical tools such as the Arena Input Analyzer, which improves part flow simulation accuracy [7,8,9]. Modeled part behavior is guaranteed to represent actual variability in processing times and inspection findings using statistical fits, such as uniform, beta, or Poisson distributions. Accurate flow analysis also helps in identifying bottlenecks and resource constraints, aligning with Goldratt’s Theory of Constraints, which states that system throughput is determined by the slowest or most burdened process [10,11,12]. As a result, part flow analysis through simulation becomes a crucial method for enhancing production planning, layout design, and quality control strategy decision-making [13,14,15].
Ref. [16] used OptQuest and Arena to search for the optimal supply chain network decisions under three levels of uncertainty. Similarly, Ref. [17] used Arena and OptQuest to determine the best number of workstations in the garment assembly line. By contrast, the authors of [18] demonstrated the possibility of integrating Arena and CPLEX software tools for simulation-based optimization. Evidence from the previous studies shows an integration of Arena and Opt Quest as the most used simulation-based optimization software tool [19]. However, in any design problem, the selection of software to be used for simulation studies is very important and is primarily based on the number of criteria, including the ease of use, animation capability, model development, and input category [20,21,22,23,24,25].

4. Research Objectives and Scope

The primary objective of this research is to develop and analyze a simulation model of a CNC-based production system using Arena simulation software. The study aims to investigate system performance in terms of production efficiency, queue behavior, machine utilization, and the impact of a rework process for defective components. By incorporating realistic production dynamics and quality control feedback loops, the research seeks to provide actionable insights that support process optimization and operational decision-making in CNC manufacturing environments.
In-Scope Activities:
  • Simulation of the CNC production workflow, including CNC turning, CNC milling, drilling, inspection, and rework loops, using Arena software;
  • Evaluation of critical performance metrics such as cycle time, work-in-progress (WIP), machine utilization, and queue dynamics;
  • Implementation of a rework mechanism, where 10% of Part 1, 15% of Part 2, and 5% of Part 3 are returned to the CNC Turning station to reflect quality control processes;
  • Exploration of alternative process configurations and strategies aimed at improving overall production system performance.
Out-of-Scope Activities:
  • Physical implementation or modification of the production system on the factory floor;
  • Detailed cost–benefit analysis or economic feasibility studies for potential machine replacements or upgrades.

5. Process Flowchart

Lathe turning is the first step in a component production process depicted in Figure 1, which is followed by CNC turning, milling, drilling, and inspection. Defective components are returned to CNC for rework, while components that pass inspection go to assembly. While this rework cycle guarantees high-quality output, it lengthens the CNC turning load and queue time.
This procedure is modeled using Arena simulation software, which also assesses important performance indicators like production output, machine usage, waiting time, and queue length. It makes it possible to evaluate enhancements like lowering rework rates, increasing inspection accuracy, or adding CNC machines. Through data-driven tactics to reduce delays, balance machine loads, and boost overall efficiency in manufacturing processes, these simulations assist in identifying bottlenecks and streamlining workflow.
The research methodology employed in the Arena simulation project is depicted in Figure 2. Before describing the issue and gathering pertinent information like processing times and arrival rates, a literature review is conducted to comprehend previous research. A simulation model is created and run in Arena following data analysis. After that, output results are examined to find bottlenecks and assess enhancements. Because of this iterative process, the model may be continuously improved depending on performance metrics such as queue lengths, machine utilization, cycle time, and work-in-progress. A realistic, data-driven simulation for CNC production process optimization is guaranteed with this methodical approach.

6. Processing Time Data

Table 1 presents the processing times for Part 1 across five components. CNC turning and inspection are the most time-consuming stages, with times ranging from 25 to 30 min and 30 to 35 min, respectively. Lathe operations take between 15 and 20 min, while drilling is the shortest step (4–5 min). The data suggest that Part 1 has the highest overall processing time, especially at CNC turning and inspection, which may cause delays or queuing in the system.
As shown in Table 2, Part 2 exhibits more balanced and moderate processing times across components. Lathe operations vary from 10 to 15 min, and CNC turning ranges from 20 to 25 min. Milling and inspection also show moderate durations (15–25 min), while drilling remains uniform and short (3–4 min). The relatively shorter times compared to Part 1 indicate better flow, though CNC turning and inspection remain key steps to monitor in simulation.
Table 3 shows the most efficient processing for Part 3. Lathe, CNC turning, and milling are consistent, mostly ranging between 12 and 20 min. Drilling is notably the shortest stage (2.5–3.5 min), while inspection ranges from 20 to 22 min. The uniformity and shorter durations across all steps make Part 3 the most optimized among the three, suitable for enhancing throughput in the production system simulation.

7. Input Analyzer

As shown in Figure 3a–c, the Arena Input Analyzer was used to determine the best-fit probability distribution for a given dataset for parts by comparing the square error values of various distributions. A lower square error indicated a better fit to the data. The analysis results are as follows: Based on the square error values, a uniform distribution provided the best fit for the data, having the least square error (0.113) among all the values considered in the distribution (Figure 3a). Therefore, the uniform distribution was recommended for use in the simulation model for an accurate representation of the input data in the same way the input analyzer was used for all the dataset examples shown in (b) and (c) distributions.
As shown in Table 4, the processing time data for different parts and operations were analyzed using the Arena Input Analyzer, selecting the best-fit distributions for simulation modeling. Lathe and CNC turning mainly followed uniform and beta distributions, indicating bounded variability. Milling, applicable to Part 2 and Part 3, showed higher variation when modeled by beta distributions. Drilling exhibited triangular and normal distributions, reflecting defined ranges and symmetric variation. Inspection processes varied, with Poisson and beta distributions capturing both discrete and continuous behavior. Rework steps exhibited triangular and normal distributions, representing short, consistent tasks. These distributions ensure a realistic simulation of each process and part type in the production system.

8. Building the Model

Figure 4 illustrates the Arena simulation model (part entity) used to represent the CNC production system. In the model, the parts arrived at fixed intervals of 30 min, reflecting a constant inter-arrival time. The production line consisted of five main processes: CNC turning, CNC milling, drilling, inspection, and rework loops for defective items. A transfer time of 2 min was incorporated between each station to simulate material handling and movement.
The simulation was executed over a total duration of 80 operational hours (equivalent to 16 h per day across five working days), aligning with a typical full-shift schedule. To ensure robustness and account for variability in system behavior, the model was run over 10 independent replications.
This simulation framework enabled a comprehensive analysis of key performance indicators such as cycle time, work-in-progress (WIP), machine utilization, and bottleneck identification, providing critical insights for evaluating and improving the effectiveness of the production system.
Figure 5 summarizes the output, utilization, and queue performance of the simulation. The system’s overall output of 167 units over the simulation time demonstrated its processing power. Resource 2 and Resource 5 had the highest utilizations (86.34% and 85.27%), suggesting potential bottlenecks. Resource usage varied throughout the stations. Resource 4’s low utilization rate of 12.85%, on the other hand, suggested either underuse or excess capacity. These variations are graphically highlighted in the utilization bar chart. Queues 2 and 5 had a moderate buildup in queue performance, which was consistent with high-utilization resources. While some stations ran successfully, others encountered delays that would have affected throughput, according to the queue chart. All things considered, Figure 5 identifies important areas for development, such as balancing resource loads and cutting down on waiting times to increase system effectiveness.
Table 5 shows the output results of the simulation, providing important new information on the CNC manufacturing process. The inter-arrival time had a mean of 30 and was distributed exponentially. At 86.54% and 85.27%, respectively, Stations 2 and 5 had the greatest usage rates, suggesting possible bottlenecks as shown in Table 6. Queues 2 (1.778) and 5 (0.6878) had correspondingly longer lines and longer waiting times, indicating delays at these locations. However, at 12.85%, Station 4 was noticeably underutilized, suggesting potential inefficiencies. Changing production loads are reflected in the WIP chart, which displays system flow variability. Confidence intervals for runs of n = 10 and n = 25 confirmed the consistency of the results. Enhancing system performance, decreasing delays, and increasing productivity can be achieved by addressing bottlenecks at Stations 2 and 5, whether through capacity expansion, task reallocation, or rework reduction.

9. Improvements

9.1. IMP#1: Add Parallel CNC Turning Machine

This was implemented to split the load from the heavily utilized CNC turning process.
Impact: It reduces waiting time and queue size, enhancing throughput without a significant increase in WIP.

9.2. IMP#2: Optimize Inspection Step

Alternate 1: Add an additional inspection station.
Alternate 2: Reduce inspection time if feasible.
Impact: This will decrease accumulation at the final stage and shorten the overall cycle time.

9.3. IMP#3: Increase Inter-Arrival Time

Test with EXPO (35) and (40)
Impact: The system is near capacity. A slight slowdown may stabilize queues.

10. Updated Flowchart

The updated flowchart shows the CNC production process in Figure 6. Accepted parts proceed, while rejected ones go back for rework. This closed-loop system ensures quality but increases the workload at CNC turning. The re-inspection loop emphasizes quality control, leading to longer queues and higher utilization at CNC turning, as shown by simulation results. This structure helps identify bottlenecks and optimize flow. Adding rework to the model allows the Arena simulation software to test how quality issues affect WIP, cycle time, and machine utilization, aiding data-driven improvements.

11. Results of Improvements

Table 7 shows the system’s performance under three improvement strategies, and the baseline (Expo 30) is compared in the table. There is a trade-off between system modifications and throughput, as evidenced by the output being highest in the baseline (167) and lowest in Improvement #3 (124). Waiting and line times dropped dramatically with improvements, indicating improved flow, particularly in Queues 2 and 5. Resource utilization, especially Utilization 2 and 5, decreased in Improvements #2 and #3, indicating fewer bottlenecks. Utilization 2 was still high in all cases, however, suggesting that this is a crucial issue that needs attention. The confidence intervals suggested consistent output trends across replications. All things considered, Improvement #1 provided a well-rounded strategy with reasonable output, shorter wait times, and better resource allocation without seeing sharp declines in efficiency.
Figure 7a–d present 95% confidence intervals derived from 10 Arena simulation runs, illustrating the variability and consistency of key performance metrics. Narrow confidence intervals suggest stable outcomes with minimal variability, thereby confirming the reliability of the model. This bolsters the validity of the conclusions drawn from the simulation data across repeated replications.
Chart 1 bar chart shows the average wait times for the existing, Improvement 1, Improvement 2, and Improvement 3 scenarios. The existing system had the longest queue at 1.75. Improvement 1 drastically reduced the queue to nearly zero. Improvement 2 lowered it to about 1.55. Improvement 3 moderately reduced it to 0.45. Improvement 1 was the most effective in reducing Queue 2, followed by Improvement 3, while Improvement 2 slightly improved over the existing system. Thus, Improvement 1 is the recommended option.
Chart 2, titled “Utilization 2,” presents the utilization rates across four scenarios: existing, Improvement 1, Improvement 2, and Improvement 3. The existing system showed the highest utilization at 87%. In contrast, Improvement 1 significantly reduced utilization to 40%, indicating potential underutilization. Improvement 2 maintained a high utilization rate of 78%, comparable to the existing scenario. Improvement 3 achieved a moderate utilization of 67%. While Improvement 1 effectively reduced queue length, it did so at the cost of low resource utilization. In comparison, Improvement 2 offered a more balanced outcome, combining queue reduction with sustained high utilization.

12. Design of Experiment

The regression equation in uncoded units is as follows:
Output = 461.0 − 8.200 X1 − 224.0 X2 + 6.000 X3 + 5.600 X1*X2 − 0.4000 X1*X3 + 17.00 X2*X3 − 0.4000 X1*X2*X3
where,
X1: Arrival time
X2: Capacity of the CNC turning process
X3: Capacity in the inspection station
Table 8 illustrate the high and low values of the above mentioned factors in the study, while Table 9 shows the results of the interaction of all variables at all levels.
The Pareto chart in Figure 8 shows that none of the factors or interactions were statistically significant at the 0.05 level, as no effect exceeded the reference line of 33.88. However, the interaction between factors A and B (AB) and the main effects of B and A showed relatively larger influences on the output, with AB being the highest. Factor C and its interactions (AC, BC, and ABC) had minimal effects. For practical improvement, focus should be placed on factors A, B, and C, and their interaction.
Figure 9 shows the plot of the output’s main effects, illustrating the variation in the mean output with varying quantities of variables X1, X2, and X3. As each element increased from its lowest to its highest level, the graph shows that output fell. After X1 and X3, factor X2 had the sharpest slope, indicating that it had the greatest effect on production. For example, when X1 was at 35, X2 was at level 1, and X3 was at level 1, the greatest mean output was recorded. Setting the process at the lowest levels of all three factors, we found that X2 will particularly optimize production.
Figure 10 illustrates the design of experiments (DOE). This interaction plot illustrates how three parameters (X1, X2, and X3) affect the results. The non-parallel and intersecting lines suggest that the X1, X2 interaction is significant, indicating that the level of X2 influences the effect of X1. The relatively parallel lines in X1, X3, and X2*X3 graphs suggest minimal interaction between these factor pairings. The notable interaction between X1 and X2 indicates that they should be analyzed together rather than separately. This figure assists in identifying important element combinations for optimizing process output.
Optimal factor levels from the regression equation:
X1 = 37.22, X2 = 4.21, and X3 = 7.37

13. Conclusions

Long wait times, queue accumulation, and unequal resource use were among the inefficiencies in the CNC manufacturing process that were effectively found by the component flow analysis utilizing the Arena simulation software. This study assessed how three important factors—arrival time (X1), capacity of CNC turning (X2), and inspection frequency (X3)—affect system performance using design of experiments (DOE). According to the simulation results, throughput, idle time, and resource balance all significantly improved when these variables were optimized. Arrival time and inspection frequency were found to have the most effect on output, according to regression analysis and Pareto charts. All things considered, Arena is a useful tool for simulating and refining intricate manufacturing systems, allowing for data-driven choices that boost output, cut waste, and simplify part flow in the CNC manufacturing setting.
The findings indicate that the simulation and design of experiments (DOE) approach applied in this study can be effectively extended to other multi-stage manufacturing systems, particularly those involving rework loops. Considering that rework rates in certain industries range from 5% to 15% of total production, analyzing rework handling and its impact on overall system performance proves valuable for assessing the broader applicability of this method. This approach demonstrates significant potential for enhancing manufacturing efficiency across various operational contexts.
Furthermore, incorporating additional performance indicators, such as energy consumption (which can account for up to 20% of operational costs in some industries) or unplanned downtime (which can range from 5% to 20% of operating time), into future simulation models is a logical next step. Addressing inefficiencies during non-productive periods or in inefficient processes could lead to substantial cost savings. Therefore, adopting a more comprehensive evaluation framework that includes these factors is crucial for driving deeper improvements in manufacturing performance.

Author Contributions

Conceptualization, A.A. (Ahad Ali) and V.S.; methodology, A.A. (Ahad Ali); software, V.S.; validation, A.A. (Abdelhakim Abdelhadi) and A.A. (Ahmad Alkhaleefah); formal analysis, V.S.; investigation, A.A. (Ahad Ali); resources, A.A. (Ahad Ali); data curation, A.A.; writing—original draft preparation, V.S.; writ-ing—review and editing, A.A. (Abdelhakim Abdelhadi); visualization, A.A. (Ahmad Alkhaleefah); funding acquisition, A.A. (Ahmad Alkhaleefah). All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data sharing is not applicable to this article.

Acknowledgments

The Authors would like to thank Prince Sultan University in Saudi Arabia for their financial support of this research.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Groover, M.P. Fundamentals of Modern Manufacturing: Materials, Processes, and Systems, 7th ed.; Wiley: Hoboken, NJ, USA, 2020. [Google Scholar]
  2. Banks, J.; Carson, J.S.; Nelson, B.L.; Nicol, D.M. Discrete-Event System Simulation, 5th ed.; Pearson: London, UK, 2010. [Google Scholar]
  3. Kelton, W.D.; Sadowski, R.P.; Swets, N.B. Simulation with Arena, 6th ed.; McGraw-Hill: New York, NY, USA, 2015. [Google Scholar]
  4. Choi, H.; Yu, S.; Lee, D.; Noh, S.D.; Ji, S.; Kim, H.; Yoon, H.; Kwon, M.; Han, J. Optimization of the Factory Layout and Production Flow Using Production-Simulation-Based Reinforcement Learning. Machines 2024, 12, 390. [Google Scholar] [CrossRef]
  5. Jain, S. Modeling and analysis using simulation. In Industrial Engineering Handbook; McGraw-Hill: New York, NY, USA, 1991. [Google Scholar]
  6. Law, A.M.; Kelton, W.D. Simulation Modeling and Analysis, 4th ed.; McGraw-Hill: New York, NY, USA, 2007. [Google Scholar]
  7. Ali, M.; Kumar, M. A conceptual Framework of flexibility focused integrated Manufacturing system. In Proceedings of the National Conference on Recent Advance in Mechanical Engineering, CAET, Etawa, India, 17 October 2007. [Google Scholar]
  8. Browne, J.D.; Dubois, K.R.; Sethi, S.P.; Stecke, K.E. Classification of flexible manufacturing systems. FMS Mag. 1984, 114–117. [Google Scholar]
  9. Beach, R.; Muhlemann, A.P.; Price, D.H.R.; Paterson, A.; Sharp, J.A. A review of manufacturing flexibility. Eur. J. Oper. Res. 2000, 122, 41–57. [Google Scholar] [CrossRef]
  10. Barad, M.; Sapir, D.E. Flexibility in logistic systems modeling and performance evaluation. Int. J. Prod. Econ. 2003, 85, 155–170. [Google Scholar] [CrossRef]
  11. Chan, F.T.S.; Bhagwat, R.; Wadhwa, S. Flexibility performance: Taguchi’s method study of physical system and operating control parameters of FMS. Robot. Comput.-Integr. Manuf. 2007, 23, 25–37. [Google Scholar] [CrossRef]
  12. Caprihan, R.; Wadhwa, S. Impact of routing flexibility on the performance of an FMS—A Simulation Study. Int. J. Flex. Manuf. Syst. 1997, 9, 273–298. [Google Scholar] [CrossRef]
  13. Garg Amik Deshmukh, S.G. Maintenance management: Literature review and directions. J. Qual. Maint. Eng. 2006, 12, 205–238. [Google Scholar] [CrossRef]
  14. Gerwin, D. Manufacturing flexibility: A strategic perspective. Manag. Sci. 1993, 39, 395–410. [Google Scholar] [CrossRef]
  15. Karsiti, M.N.; Cruz, B.J., Jr.; Mulligan, J.H., Jr. Simulation studies of multilevel dynamic job shop scheduling using heuristic dispatching rules. J. Manuf. Syst. 1992, 11, 346–358. [Google Scholar] [CrossRef]
  16. Bongomin, O.; Mwasiagi, J.I.; Nganyi, E.O.; Nibikora, I. Improvement of garment assembly line efficiency using line balancing technique. Eng. Rep. 2020, 2, e12157. [Google Scholar] [CrossRef]
  17. Jerbi, A.; Ammar, A.; Krid, M. Performance optimization of a flexible manufacturing system using simulation: The Taguchi method versus OptQuest in Simul. Trans. Soc. Model. Simul. Int. 2019, 95, 1085–1096. [Google Scholar] [CrossRef]
  18. Chiadamrong, N.; Piyathanavong, V. Optimal design of supply chain network under uncertainty environment using hybrid analytical and simulation modeling approach. J. Ind. Eng. Int. 2017, 13, 465–478. [Google Scholar] [CrossRef]
  19. Elnaggar, G. Effect of Operator Skill Level on Assembly Line Balancing in Apparel Manufacturing: A Multi-Objective Simulation Optimization Approach. In Proceedings of the International Conference on Industrial Engineering and Operations Management, Bangkok, Thailand, 5–7 March 2019. [Google Scholar]
  20. Borodin, V.; Bourtembourg, J.; Hnaien, F.; Labadie, N. COTS software integration for simulation optimization coupling: Case of ARENA and CPLEX products. Int. J. Model. Simul. 2019, 178–189. [Google Scholar] [CrossRef]
  21. Prajapat, N.; Tiwari, A. A review of assembly optimisation applications using discrete event simulation. Int. J. Comput. Integr. Manuf. 2017, 30, 215–228. [Google Scholar] [CrossRef]
  22. Hosamo, H.; Mazzetto, S. Integrating Knowledge Graphs and Digital Twins for Heritage Building Conservation. Buildings 2025, 15, 16. [Google Scholar] [CrossRef]
  23. Mazzetto, S. Interdisciplinary Perspectives on Agent-Based Modeling in the Architecture, Engineering, and Construction Industry: A Comprehensive Review. Buildings 2024, 14, 3480. [Google Scholar] [CrossRef]
  24. Abdelrahim, M.; Almakhles, D. Dynamic Event-Triggered Control for Sensor–Controller–Actuator Networked Control Systems. J. Sens. Actuator Netw. 2024, 13, 61. [Google Scholar] [CrossRef]
  25. Amudhavalli, P.; Zahira, R.; Umashankar, S.; Fernando, X.N. A Smart Approach to Electric Vehicle Optimization via IoT-Enabled Recommender Systems. Technologies 2024, 12, 137. [Google Scholar] [CrossRef]
Figure 1. Process flowchart.
Figure 1. Process flowchart.
Applsci 15 07637 g001
Figure 2. Research methodology.
Figure 2. Research methodology.
Applsci 15 07637 g002
Figure 3. (a) Uniform distribution using the Arena Input Analyzer; (b) normal distribution using the Arena Input Analyzer; (c) beta distribution using the Arena Input Analyzer.
Figure 3. (a) Uniform distribution using the Arena Input Analyzer; (b) normal distribution using the Arena Input Analyzer; (c) beta distribution using the Arena Input Analyzer.
Applsci 15 07637 g003
Figure 4. Arena simulation model.
Figure 4. Arena simulation model.
Applsci 15 07637 g004
Figure 5. Output results from the Arena simulation.
Figure 5. Output results from the Arena simulation.
Applsci 15 07637 g005aApplsci 15 07637 g005b
Figure 6. Updated process flowchart.
Figure 6. Updated process flowchart.
Applsci 15 07637 g006
Figure 7. (a) The 95% confidence intervals (n = 10) in inter-arrival time Expo (30); (b) 95% confidence intervals n = 10) in Improvement #1; (c) 95% confidence intervals (n = 10) in Improvement #2; (d) 95% confidence intervals (n = 10) in Improvement #3.
Figure 7. (a) The 95% confidence intervals (n = 10) in inter-arrival time Expo (30); (b) 95% confidence intervals n = 10) in Improvement #1; (c) 95% confidence intervals (n = 10) in Improvement #2; (d) 95% confidence intervals (n = 10) in Improvement #3.
Applsci 15 07637 g007aApplsci 15 07637 g007b
Chart 1. The average wait times improvements for the three scenarios.
Chart 1. The average wait times improvements for the three scenarios.
Applsci 15 07637 ch001
Chart 2. The utilization rates across four scenarios (including the existing one).
Chart 2. The utilization rates across four scenarios (including the existing one).
Applsci 15 07637 ch002
Figure 8. Interaction factors significance at 0.05 level.
Figure 8. Interaction factors significance at 0.05 level.
Applsci 15 07637 g008
Figure 9. Output’s main effects results.
Figure 9. Output’s main effects results.
Applsci 15 07637 g009
Figure 10. Interaction output results.
Figure 10. Interaction output results.
Applsci 15 07637 g010
Table 1. Part 1 processing times (mins).
Table 1. Part 1 processing times (mins).
Process StepComponent 1Component 2Component 3Component 4Component 5
Lathe1518161820
CNC Turning3027292725
Drilling45545
Inspection3032353430
Table 2. Part 2 processing times (mins).
Table 2. Part 2 processing times (mins).
Process StepComponent 1Component 2Component 3Component 4Component 5
Lathe1510121315
CNC Turning2025222320
Milling2015182019
Drilling33444
Inspection2222252225
Table 3. Part 3 processing times (mins).
Table 3. Part 3 processing times (mins).
Process StepComponent 1Component 2Component 3Component 4Component 5
Lathe1817181818
CNC Turning2018202018
Milling1212151515
Drilling3.12.53.33.03.5
Inspection2022202220
Table 4. Processing data for all parts using the input analyzer.
Table 4. Processing data for all parts using the input analyzer.
Process StepPart 1 (min)Part 2 (min)Part 3 (min)
LatheUNIF (14.5, 20.5)9.5 + 6 * BETA (0.551, 0.394)POIS (23.2)
CNC turningUNIF (24.5, 30.5)19.5 + 6 * BETA (0.394, 0.551)17.5 + 3 * BETA (0.477, 0.365)
MillingN/A14.5 + 6 * BETA (1.17, 0.745)11.5 + 4 * BETA (0.886, 0.718)
DrillingTRIA (3.5, 4.8, 5.5)TRIA (2.5, 3.8, 4.5)NORM (3.08, 0.337)
InspectionPOIS (32.2)POIS (23.2)19.5 + 3 * BETA (0.365, 0.477)
ReworkTRIA (2.5, 3.2, 4.5)TRIA (1.5, 2.2, 3.5)NORM (3, 0.632)
Table 5. Output analysis.
Table 5. Output analysis.
ResultsInter-Arrival Time Expo (30)
Output167
Total time of parts 1, 2, and 32.87, 3.15, 3.12
Number in parts 1, 2, and 352, 57, 65
Number out parts 1, 2, and 348, 55, and 64
Queue 10.603
Queue 21.778
Queue 30.0013
Queue 40.0
Queue 50.6878
Waiting time 10.2172
Waiting time 20.4919
Waiting time 30.000898
Waiting time 40.0
Waiting time 50.3275
Utilization 167%
Utilization 286.54%
Utilization 339.13%
Utilization 412.85%
Utilization 585.27%
WIPApplsci 15 07637 i001
95%Confidence interval (n = 10)Applsci 15 07637 i002
95%Confidence interval (n = 25)Applsci 15 07637 i003
Table 6. The possibility of botelnecks.
Table 6. The possibility of botelnecks.
BottlenecksProcessUtilizationQueue
PrimaryCNC Turning86.54%1.778
SecondaryInspection85.27%0.6873
Table 7. System performance improvement.
Table 7. System performance improvement.
ResultsInter-Arrival Time Expo (30)Improvement #1Improvement #2Improvement #3 Expo (40)
output167147148124
Total time of parts 1, 2, and 32.87, 3.15, 3.122.54, 2.29, 2.262.47, 2.91, 3.02.31, 2.12, 2.39
Number in parts 1, 2, and 352, 57, 6559, 48, 4955, 54, 4244, 45, 39
Number out parts 1, 2, and 348, 55, and 6455, 45, 4754, 53, 4142, 44, 38
Queue 10.6030.420.630.29
Queue 21.7780.01451.53580.45
Queue 30.00130.011770.00010.0
Queue 40.00.0030.00.0004
Queue 50.68780.55020.00.25
Waiting time 10.21720.15960.25400.13
Waiting time 20.49190.004680.50350.17
Waiting time 30.0008980.0097080.00010.0
Waiting time 40.00.001630.00.003
Waiting time 50.32750.550.00.15
Utilization 167%60%57%50%
Utilization 286.54%40%77%65%
Utilization 339.13%32%31%28%
Utilization 412.85%12%11%9%
Utilization 585.27%81%39%67%
Table 8. High and low values of the factors in the study.
Table 8. High and low values of the factors in the study.
LowHigh
Arrival timeX13540
Capacity of the CNC turning process X212
Capacity in the inspection station X312
Table 9. Results values at all level of interaction of the factors in the study.
Table 9. Results values at all level of interaction of the factors in the study.
X1X2X3OutputHigh QueueHigh Utilization
135111410.7577%
240111240.459667%
335211160.3862%
440211250.595671%
535121360.8872%
640121150.3861%
735221140.256446%
840221170.477648%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Sevella, V.; Ali, A.; Abdelhadi, A.; Alkhaleefah, A. Data-Driven Optimization of CNC Manufacturing Using Simulation and DOE Techniques. Appl. Sci. 2025, 15, 7637. https://doi.org/10.3390/app15147637

AMA Style

Sevella V, Ali A, Abdelhadi A, Alkhaleefah A. Data-Driven Optimization of CNC Manufacturing Using Simulation and DOE Techniques. Applied Sciences. 2025; 15(14):7637. https://doi.org/10.3390/app15147637

Chicago/Turabian Style

Sevella, Vijay, Ahad Ali, Abdelhakim Abdelhadi, and Ahmad Alkhaleefah. 2025. "Data-Driven Optimization of CNC Manufacturing Using Simulation and DOE Techniques" Applied Sciences 15, no. 14: 7637. https://doi.org/10.3390/app15147637

APA Style

Sevella, V., Ali, A., Abdelhadi, A., & Alkhaleefah, A. (2025). Data-Driven Optimization of CNC Manufacturing Using Simulation and DOE Techniques. Applied Sciences, 15(14), 7637. https://doi.org/10.3390/app15147637

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop