1. Introduction
The company under study specializes in producing garment products. The main products of the company include men vestons, jackets, shirts, and trousers. According to statistics, the average on-time delivery percentage of the company in the past is 85%. It is low and does not meet the expectation of the company leaders. Through analysis, the cause of the problem is due to a long production time, low productivity, and a high defect rate.
This paper uses Lean Six Sigma tools to solve the problem of low on-time delivery through reducing the production time, increasing productivity, and improving the defect rate. The research is carried out on a product with the highest demand on a line of the shirt garment production area.
2. Literature review and research methodology
Lean Six Sigma is the fastest way to improve time and quality; therefore, we can minimize operation costs, minimize capital investment, increase value, and increase customer satisfaction. Lean Six Sigma is an integrated approach of Lean Manufacturing and Six Sigma. If we use Lean Manufacturing alone, we cannot control the process and solve quality problems in the best way. If we use only Six Sigma, we cannot minimize the time, cost, and capital investment. Simultaneous use of Lean and Six Sigma tools helps us achieve high quality, high-speed production, and low cost with breakthrough improvements in a short time. Lean Six Sigma gives faster results than either Six Sigma or Lean Manufacturing.
The research methodology is based on Lean Six Sigma theory, with the platform of DMAIC procedure (1), including five steps:
(1). Define
(2). Measure
(3). Analyze
(4). Improve
(5). Control.
The define step defines the problem, objective, and scope of the study. In this step, the objective is defined qualitatively. The measurement step collects data, maps the current state value stream, identifies current performance indexes, and sets objectives quantitatively. The analysis step finds out the root causes of the problem and then identifies solutions and tools to solve the problems. The improvement step implements solutions to improve the process in order to achieve the objectives. The control step controls the improved process to maintain improved results.
The tools used in the steps of DMAIC procedure include value stream management, work design, SMED, line balancing, Kanban, FIFO, autonomous maintenance, visual management, design of experiments (DOE), and control charts. All tools used can be found in references (1) and (2), along with articles (3) and (4).
3. Define
The problem that the company encounters is a low percentage of on-time delivery. The average on-time delivery percentage of the company is 85% and needs to be improved. Through analysis, the cause of the problem is due to a long production lead time, low productivity, a high cycle time, or a high defect per unit rate. The research is carried out on a product family with the highest demand (5, 6). The SIPOC diagram of the product family is as in Figure 1.
Figure 1. SIPOC diagram.
This paper focuses on the process of back finish with six stations as in Table 1.
Table 1. Stations in back finish process.
The defined stage is summarized in the project charter as in Table 2.
Table 2. Project charter.
4. Measure
The collected data in the back finish process, including number of workers n, change over time (COT), cycle time (CT), lead time (LT), and defect per unit (DPU), are shown in Table 3.
Table 3. Collected data of back finish process.
Currently, the company works 6 days/week, 1 shift/day, 8 h/shift. The average break time is 40 min/day. The average COT is 65 min/day. The available producing time is as follows:
with the demand of 1500 pcs/day. The talk time TT is calculated as follows:
Work in process inventory WIP and time in process TIP are calculated and shown in Table 4.
Table 4. Data of WIP and TIP.
With the data above, the current state map of the process is as in Figure 2.
Figure 2. Current value stream mapping.
From the map, the current indexes, their values, and performance assessment are as in Table 5.
Table 5. Current indexes.
The objectives of the research are set as in Table 6.
Table 6. Objectives of research.
5. Analyze
5.1. Analyze the long-time problem
The long-time problems are demonstrated by long LT and long CT, exceeding the TT. The balance chart is as in Figure 3.
Figure 3. Current balance chart.
The process is unbalanced, leading to the increase of WIP inventory, causing long LT. Besides, CT of the process is 25 s, higher than TT, which is 15 s; the current production rate does not meet the demand rate. The causes of the problem are analyzed through the fish bone diagram as in Figure 4.
Figure 4. The fish bone diagram shows the causes of long-time problems.
From the diagram, the causes, solutions, and solving tools are demonstrated in Table 7.
Table 7. Solutions and tools for problems.
5.2. Analyze the high-defect problem
The causes of high-DPU problems are analyzed through the fish bone diagram as in Figure 5.
Figure 5. The fish bone diagram shows the causes of defect problems.
The Pareto charts for the causes are as in Figure 6.
Figure 6. The Pareto chart analyzes the causes of high-defect problems.
The fish bone diagram for the skip stitch shoulder effect is as in Figure 7.
Figure 7. The fish bone diagram shows the causes of skip stitch shoulder defects.
The fish bone diagram for the skip stitch yoke effect is as in Figure 8.
Figure 8. The fish bone diagram shows the causes of skip stitch yoke defects.
From the fish bone diagrams, the causes, solutions, and solving tools for skip stitch shoulder and yoke defects are presented in Table 8.
Table 8. Solutions and tools for skip stitch shoulder and yoke.
The fish bone diagram for the broken stitch shoulder defects is as in Figure 9.
Figure 9. The fish bone diagram shows the causes of broken stitch shoulder defects.
From the fish bone diagrams, the causes, solutions, and solving tools for broken stitch shoulder defects are presented in Table 9.
Table 9. Solutions and tools for causes of defects.
6. Improve
6.1. Quick change over SMED
Change over time needs to be reduced to lift up TT. Based on (2), SMED is used by the following steps.
Step 1: Identify time elements.
Elements and their time are collected, as in Table 10.
Table 10. Time of elements.
Step 2: Separate external elements
The internal and external elements are separated as in Table 11.
Table 11. Separate external elements.
Step 3: Convert internal elements to external elements.
The internal elements are converted to external elements as in Table 12.
Table 12. Convert internal elements to external.
Step 4: Streamline remaining elements
The remaining elements are streamlined as in Table 13.
Table 13. Streamline remaining elements.
After using SMED, COT, APT, and TT change as in Table 14.
Table 14. Comparison of parameters before and after applying SMED.
The balance chart of the process after using SMED is as in Figure 10.
Figure 10. Balance chart after applying SMED.
The process CT is 25 s, still higher than TT. Work design needs to be used to reduce the CT of each station to meet the TT.
6.2. Work design
Collecting the working elements of each station and their processing time is as in Table 15.
Table 15. Collected data of working elements.
By analyzing and removing wasted motion, reallocating the work load of the right and left hands, designing support tools, and rearranging station layout, the results are that the CT reduced while the number of workers remained unchanged, as shown in Table 16.
Table 16. Working elements after the improvement.
The balance chart after using work design is as in Figure 11.
Figure 11. Balance chart after applying work design.
The process CT is now 19 s, which still exceeds the TT, 16.8 s. Line balancing is used to reallocate work load of each station to meet demand.
6.3. Line balancing
Applying the line balancing model in (3), we reallocate the work load and worker for the stations with constraints of the element order and the goal of meeting TT. The results are as in Table 17.
Table 17. Working elements data.
The balance chart after using line balancing is as in Figure 12.
Figure 12. Balance chart after applying line balancing.
The process CT now is 16.5 s, which meets the TT, 16.77 s. On the other hand, the number of workers has reduced to 8, the number of stations has reduced to 5, and the line balancing efficiency has increased to 92.12%. The process is more balanced, but we still have to control WIP inventory.
6.4. WIP inventory control
With reference to (1), in order to control WIP, Kanban systems would be placed between stations 2 and 3 and between stations 4 and 5. FIFO lanes would be placed between stations 1 and 2 and between stations 3 and 4.
Applying Kanban models in (1), the numbers of Kanban cards, N, are calculated, as in Table 18, with the demand D of 1500 products, a lot size Q of 5, and α = 0.1. The formula of N is as below:
Table 18. Calculation of Kanban card quantity.
According to models in (4), the FIFO lane sizes, S, are calculated as in Table 19.
Table 19. Calculation of FIFO lane size.
6.5. Autonomous maintenance
In order to solve the causes of machine failures, machines have been restored to their basic conditions. AM activities are then established by the following steps.
Step 1: A standard checklist for routine cleaning, inspection, and lubrication activities is established as in Table 20.
Table 20. Standard checklist for routine cleaning, inspection, and lubrication.
Step 2: Establish needle and thread inspection standards as in Table 21.
Table 21. Needle and thread inspection standards.
Step 3: Design checking notes for cleaning, inspection, and lubrication as in Table 22.
Table 22. Checking note for cleaning, inspection, and lubrication.
6.6. Visual control
In order to solve needle-size-not-appropriate problems, a visual tool has been used by matching size needles with colors. Also, we put them in the same color boxes. Depending on needle sizes, the storage capacity of each box is determined as in Table 23.
Table 23. Needle sizes and storage capacity of each box.
6.7. Thread storage system design
A low-quality thread is one of the most important causes which lead to broken stitch problems. By using 5-Why analysis, the root cause is identified, that is, the storage time of the thread. The quality of long-time-storage threads has been reduced. To solve the problem, a thread storage system has been designed to determine the storage time as in Figure 13. This system contains three types of cards:
Figure 13. Thread storage system.
– Time card: Cards indicate the storage date of the material.
– Empty card: Cards attached on empty batches.
– Emergency red card: This is the only card that is attached on the longest storage time batch, which is priority for using.
The operational process is as follows:
– Import materials are stored in the empty batch (with empty cards).
– Then, empty cards are replaced with time cards with storage dates of the batch.
– Emergency cards with the longest storage time batch.
– In case of stocking out batches (with emergency cards), they are replaced with empty cards.
6.8. Design of experiments
With reference to (1), DOE is implemented in the Yoke Station to minimize skip stitch errors and in the Shoulder Station to minimize skip stitch and broken stitch errors.
6.8.1. Yoke station
The input variables are needle eye stock, rotary hook distance D, and amount of taken thread A. The output variable is the skip defect rate DR. The variables are shown in Figure 14.
Figure 14. Input and output variables for DOE.
The levels of factors are defined as in Table 24.
Table 24. Levels of factors for DOE.
With the number of repetitions of 3, the number of experiments is 36. The data are collected. From the collected data, the ANOVA is shown as in Table 25.
Table 25. Result of ANOVA of Yoke Station.
Through ANOVA analysis, both factors and their interaction have affected defect rate. The contour plot of defect rate DR according to the amount of taken thread A and the distance of needle eye stock and rotary hook D is shown in Figure 15.
Figure 15. Contour plot of defect rate DR.
From the plot, the distance of needle eye stock and rotary hook D should be between 1.6 mm and 2.1 mm, and the amount of thread taken A should be set up from the level of 3.5 to 4.5. The value of the distance of 2 mm and the amount of thread taken of level are chosen.
6.8.2. Shoulder station
The input variables are needle eye stock, rotary hook distance D, amount of taken thread A, and thread tension T. The output variable is the skip defect rate DR. The variables are shown in Figure 16.
Figure 16. Input and output variables for DOE.
The levels of input factors are defined as in Table 26.
Table 26. Levels of input factors for DOE.
With the number of repetitions of 2, the number of replicas is 36. The data are collected. From the collected data, the ANOVA is shown as in Table 27.
Table 27. Result of ANOVA of Shoulder Station.
Through ANOVA analysis, both factors and their 2-factor interaction have affected the defect rate. The interaction plot is shown in Figure 17.
Figure 17. Display of the interaction plot.
From the plot, the lowest defect rate would be attained with a distance of 2 mm, a tension level of 3, and a thread taken level of 3.
The results of applying DOE are shown in Table 28.
Table 28. Results of applying DOE.
6.9. Future state map
After applying Lean Six Sigma tools, the future state map is as in Figure 18.
Figure 18. Future state map.
From the map, the system indexes are presented as in Table 29.
Table 29. Index of the system.
A comparison of the performance indexes of the current state map CSM and the future state map FSM is shown in Table 30.
Table 30. Comparing the performance indexes of CSM and FSM.
The result shows that all the performance indexes have been improved.
7. Control
In order to maintain improved results, the control charts of important system parameters, LT, CT, and DPU, would be developed. With the current data, the control charts are constructed and operated. The LT control chart is as in Figure 19.
Figure 19. LT control chart.
The CT control chart is as in Figure 20.
Figure 20. CT control chart.
The DPU control chart is as in Figure 21.
Figure 21. DPU control chart.
8. Conclusion
The research has implemented Lean Six Sigma to improve a garment production process. The research methodology is based on the platform of DMAIC procedure, including five steps: define, measure, analyze, improve, and control. The tools used in the steps of DMAIC procedure include CED, Pareto diagram, value stream management, work design, SMED, line balancing, Kanban, FIFO, autonomous maintenance, visual management, DOE, and control charts.
After applying Lean Six Sigma tools, the company has reduced the production lead time by 89.21% from 279 to 30.1 min, reduced the production cycle time by 36% from 25 to 16 s, reduced the process defect rate by 37.45% from 14.9 to 9.32%, and then improved the on-time delivery rate. The results show that the research has met the objectives with the advantages of following strict and scientific methodology and using many tools; among them are strong tools like DOE.
The research still had some disadvantages. The number of collected samples used in DOE is small. The control phase has not built up standard operating processes for controlling the whole process. The improvements have not been implemented to verify the effectiveness of research. These restrictions would guide the way for future research.
Author contributions
PNN is the thesis advisor of PH, PTN, and QD. PN has developed the models for the thesis. PH, PTN, and QD have collected and analyzed the data and run the models. PNN has composed the article based on the thesis. All authors contributed to the article and approved the submitted version.
Acknowledgments
We extend our heartfelt appreciation to everyone who has contributed to the completion of this research article, especially our families, Ho Chi Minh City University of Technology (HCMUT), and the scientific community, for their invaluable support.
Conflict of interest
The research is conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
References
2. Bajpai JD. SMED (Single-Minute Exchange of Die) methodology in Garment manufacturing industry: Case study in reducing Style Change Over Time. Bangalore: National Institute of Fashion Technology (2014).
5. Phong NN, Phuc NH, Thanh VV, Sang NV, Hai NV, Hoa NV. Applying lean six sigma to improve productivity and quality of production systems – A case study in the Kia Cabin Welding Line, Model K2700ii & K3000s. J Sci Technol Dev. (2015) 18.