Effects of step-by-step line balancing in apparel assembly line

Abstract

The mixed work assignment technique was applied to the modular production concept to solve the apparel assembly line balancing problem. Specifically, an algorithm was implemented to generate workstations by first classifying tasks into modules through an analysis of the manufacturing process and assigning grouped tasks in the single task–multiple workers and multiple tasks–single worker assignment methods. Then, worker assignment was sequentially performed considering the skill level of the worker. A simulation of producing 100 men’s shirts was performed. Owing to line balancing, the total production time was considerably lower than that reported in previous studies. The enhanced performance was attributable to the fact that unlike previous studies that limited the apparel assembly line balancing problem to worker assignment, line balancing was performed through both task and worker assignment in this study. The results demonstrated that task assignment must be considered in the design stage in the apparel assembly line balancing problem.

Keywords

Apparel assembly line line balancing apparel assembly line balancing step-by-step line balancing modular production

Introduction

The implementation of smart factories has emerged as a key goal for the Fourth Industrial Revolution.^1–3 Notably, apparel manufacturing is typically labor intensive and involves a technological gap with the other manufacturing industries. Research on implementing smart technologies in apparel manufacturing remains limited.⁴ In this context, the implementation of smart factories is expected to be facilitated by applying concepts demonstrated to be effective in other manufacturing industries to apparel manufacturing to help reduce the technological gap. The modular production concept has been applied in several manufacturing industries such as automobiles, computers, and electronics to alleviate the production complexity.⁵ This concept can help facilitate customization and enhance the resource utilization.⁶ Therefore, the application of the modular production concept to apparel manufacturing has attracted widespread interest.

The core of apparel manufacturing is the assembly line, over which an apparel unit is produced. The production time associated with the assembly line must be decreased to enhance the factory profits. A key technique to reduce the production time is apparel assembly line balancing, which is aimed at avoiding bottlenecks by optimizing the design and operation of the assembly line.^7,8 In general, line balancing aims to optimally distribute assembly tasks among all workstations.⁹ Therefore, assembly line balancing involves designing and operating the assembly line in a manner that ensures the time required for workers in each workstation to complete their assigned tasks is as uniform as possible, thus avoiding bottlenecks. However, conducting assembly line balancing encounters challenges due to variations in task difficulty, task completion times, required machines, process sequence, and worker skill levels that need to be considered.

Apparel assembly line balancing in the design stage involves considerable research on measuring the time required to perform tasks.¹⁰ Specifically, machines suitable for performing each task in the apparel manufacturing process are selected, and the time required to perform each task is measured. A certain amount of time is added to the measured time to obtain the standard allowable minutes (SAM) value for each task. Assembly line balancing is performed by assigning tasks and workers with constraints such as the SAM of tasks, machines, resources, and skill level of workers to ensure that the time consumed by workers to accomplish their tasks is as uniform as possible. Notably, assembly line balancing is an NP-hard problem.^10,11 Apparel assembly line balancing in the operation stage can be performed by applying innovative operating techniques such as lean manufacturing,¹² group technology,¹³ and resource management.¹⁴

In this study, the modular production concept was applied in the design stage of the apparel assembly line. Moreover, a mixed task assignment technique was developed for apparel assembly line balancing: First, tasks are assigned to ensure that the cycle time of each workstation is as uniform as possible. Next, workers are assigned to each workstation considering their skill level. Finally, the effect of step-by-step line balancing on the production time was analyzed.

Related work

The apparel assembly line balancing problem at the design stage can be solved through task and worker assignment techniques.

Several researchers have defined the apparel assembly line balancing problem as a worker assignment problem. Chan et al.¹⁵ divided the process of producing men’s shirts into several tasks and measured the SAM of each task. Next, they performed apparel assembly line balancing by assigning each task to an optimal single worker considering the data such as the SAM, precedence relationship of tasks, and skill level of workers. However, even when the skill level of the workers is considered, the single task–single worker assignment method cannot effectively solve the line balancing problem because the SAMs of tasks typically have large deviations. Zeng et al.¹⁶ used a genetic algorithm to solve the apparel assembly line balancing problem through worker assignment. They used the single task–multiple workers assignment method, in which single tasks are assigned to multiple workers. Although the line balancing performance was improved by assigning single tasks with long SAMs to multiple workers, it was limited by single tasks with short SAMs. Chen et al.¹⁷ applied the data of Chan et al.¹⁵ to a meta-heuristics algorithm and used it to solve the apparel assembly line balancing problem through workers assignment. The authors applied the multiple tasks–single worker assignment method, in which multiple tasks are grouped and assigned to a single worker. Even in this case, the line balancing performance was limited because the tasks were grouped based only on the branches suggested by Lin,¹⁸ and the SAM values were not considered. Overall, although considerable effort has been devoted to solve the apparel assembly line balancing problem based on worker assignment, the effectiveness of these techniques has been limited. Therefore, before workers are assigned to each workstation, line balancing must be performed based on task assignment, considering the task order and SAM of each task. Also, the single task–multiple workers assignment method and multiple tasks–single worker assignment method must be applied to tasks with large and small SAM values, respectively.

Several researchers have formulated the apparel assembly line balancing problem as a task assignment problem. Kayar and Akalin¹⁹ applied the Hoffman method, known as heuristic line balancing, to assign tasks to each workstation. However, the authors did not consider worker assignment. Bongomin et al.¹⁰ applied a ranked positional weight heuristic method to assign tasks to each workstation. Although line balancing was partially improved, worker assignment was not considered. Moreover, when tasks were grouped on the basis of positional weights, the workflow became complicated because distant tasks were grouped together.

In general, the workflow complexity can be alleviated by applying the modular production concept in the design stage of the assembly line. A few researchers have applied the modular production concept to apparel assembly lines. For example, Lin¹⁸ applied the modular production concept to the problem considered by Chan et al.¹⁵ Each branch was defined as a module in the precedence relationship of tasks. However, the objective was to accelerate the solution process of the genetic algorithm, and apparel assembly line balancing was not considered. Kim and Kim²⁰ applied the modular production concept to the apparel assembly line. However, the design of the production line was implemented at the module level, and worker assignment was not considered. Jung et al.⁸ applied the modular production concept to the apparel assembly line but performed line balancing in the operation stage, with the apparel assembly line considered to be designed by experts.

Overall, none of the existing studies have explored the effect of line balancing by applying the modular production concept in the design stage of an apparel assembly line. Therefore, it is desirable to establish an algorithm for performing line balancing through sequential task and worker assignment.

Methods

In this study, the modular concept was applied in the design stage of an assembly line. A mixed task assignment method was applied, in which the single task–multiple workers and multiple tasks–single worker assignment strategies were simultaneously applied to assign tasks. Additionally, workers were assigned to tasks considering their skill levels. Figure 1 shows the framework of the proposed method.

Figure 1.

Module-based mixed tasks assignment process.

First, tasks were classified into each module through the process analysis of the apparel to be produced. Next, the number of workers and output on the assembly line were determined by comprehensively considering the resources and delivery schedule. After determining the number of workers, the number of workstations was clarified by calculating the task ratio for each module. One worker was assigned to one workstation to perform the assigned tasks. Therefore, the number of workers and number of workstations in the assembly line were identical. Next, task grouping was performed to assign tasks to workstations. In this study, the Toyota Sewing System,¹⁷ derived from the Toyota Production System for the apparel industry, was adopted. The Toyota Sewing System not only requires less floor space but also offers more flexibility and a better working environment. It was designed to enable multiple workers to perform a single task or a single worker to perform multiple tasks based on the workflow. Therefore, the Toyota Sewing System is a flexible system for task distribution, making it suitable for line balancing. Finally, the worker assignment for each workstation was performed considering the skill level of the workers.

Calculation of the number of workstations for each module

The number of workstations for each module was calculated as follows.

First, the SAM of each module ( $T_{M_{i}}$ ) was calculated by summing up the SAM values of all tasks in each module ( $S_{t_{j}}$ ) using equation (1):

T_{M_{i}} = \sum S_{t_{j}} (t_{j} \in M_{i})

(1)

Then, the total SAM ( $T_{t o t}$ ) was calculated by summing up the SAM values of all modules using equation (2).

T_{t o t} = \sum T_{M_{i}}

(2)

The number of workstations for each module ( $N_{M_{i}}$ ) was calculated using equation (3).

N_{M_{i}} = \frac{T_{M_{i}}}{T_{t o t}} \times N_{0}

(3)

where, $N_{0}$ = total number of workers

= total number of workstations

$N_{M_{i}}$ was rounded off because the number of workstations must be an integer. In this process, because of rounding errors, the sum of the number of workstations for each module may differ from the total number of workstations. In this case, the number of workstations for each module must be adjusted to minimize the deviation in the cycle time of each module, as shown in equation (4).

{N_{M_{i}}} = m i n (\sum {(S_{t o t} - S_{M_{i}})}^{2})

(4)

The cycle time of each module ( $C_{M_{i}})$ , used as the standard time for the task assignment, was calculated using equation (5).

C_{M_{i}} = \frac{T_{M_{i}}}{N_{M_{i}}}

(5)

The cycle time of the apparel assembly line ( $C_{t o t}$ ) was calculated using equation (6).

C_{t o t} = \frac{T_{t o t}}{N_{0}}

(6)

Step 1: Task assignment for generating workstations in modules

Figure 2 illustrates the algorithm for grouping tasks in modules and assigning them to workstations.

Figure 2.

Process flow of task grouping and workstation formation.

Theoretically, if all workers in the assembly line perform their tasks in the range of 95%–105% of the cycle time, no bottlenecks will occur.¹³ Therefore, the tasks for which the sum of SAMs range from 95% to 105% of the cycle time for each module (according to the task order in each module) are grouped and assigned to a workstation. If the sum of the SAMs exceeds 105% of the cycle time, the last task of the grouped tasks is shared with the next workstation. The workload is shared to ensure that the SAM of this workstation is 100% of the cycle time.

The grouping of tasks such that the SAM of all workstations is 100% of the cycle time may be optimal in terms of line balancing. However, in this configuration, a larger number of tasks are shared among multiple workstations than those when the tasks are grouped such that the cycle time is in the range of 95%–105%. The resulting increase in the complexity of the manufacturing process may not outweigh the benefits of the grouping. The differences in the SAM values across workstations can be adjusted by additional line balancing based on worker assignment.

Step 2: Worker assignment at workstations

The skill level of the workers is considered in the assignment problem as follows.

The adjusted SAM ( $S_{i j}$ ), which is the time required for worker j to perform task i, can be calculated by dividing the SAM of task i ( $S_{i}$ ) by the number representing the skill level ( $L_{i j}$ ) using equation (7).

S_{i j} = \frac{S_{i}}{L_{i j}}

(7)

Therefore, the adjusted SAM of the workstation, which is the time required for worker j to perform the tasks of the workstation, can be calculated by summing the adjusted SAM of each task assigned to the workstation using equation (8).

W_{i j} = \sum S_{i j} (i \in W_{i})

(8)

The problem of assigning workers to workstations considering their skill level is aimed at assigning a worker whose time required to perform the tasks assigned to a workstation is the closest to the cycle time of the apparel assembly line, as shown in Figure 3.

Figure 3.

Process flow of worker assignment.

First, the workstation with the longest SAM is identified because this workstation has the highest probability of bottleneck occurrence. Next, the adjusted SAM for all available workers is calculated by applying the skill level of the workers at this workstation. The adjusted SAM of all available workers is compared, and the worker with the SAM most similar to the cycle time of the apparel assembly line is assigned to this workstation. The assigned worker is removed from the available worker list, and the assigned workstation is removed from the to-be-assigned workstation list. This process is repeated until workers are assigned to all workstations. Consequently, workstations with large SAM values are assigned expert workers for the relevant tasks, and workstations with small SAM values are assigned non-specialist workers for the relevant tasks.

Results and discussion

Data

Experiments were conducted using the data of Chan et al.,¹⁵ which have been used in several studies related to apparel assembly line balancing. These data correspond to the manufacturing of men’s shirts, which involves 41 tasks. Specifically, the data consist of the SAM of the tasks, precedence relationship of the tasks, and skill level of workers for the tasks. Table 1 lists the specific tasks and SAM values, defined as the time required to perform each task 100 times sequentially. The task order was defined considering the precedence relationship (Figure 4(a)) between tasks. Table 2 summarizes several skill levels of the 41 workers for all tasks, expressed as a number between 0 and 1.5, with 0 meaning no skill and 1.5 meaning fully skilled. And skill levels for each task were generated randomly. The machine type was not considered in line balancing because information regarding the machine used for the tasks is not available.

Table 1.

Tasks and SAM* of men’s shirt (unit: min).¹⁵

No.	Name	SAM	No.	Name	SAM
1	Spot fuse collar fall	41	22	Set yoke label	20
2	Top fuse collar fall	30	23	Set yoke	34
3	Sew collar stay pocket	25	24	Join shoulder	55
4	Runstitch collar fall	48	25	Set sleeve under placket	47
5	Trim, turn, and press collar fall	30	26	Set sleeve top placket	67
6	Topstitch collar fall	65	27	Finish sleeve placket	44
7	Hem collar band	28	28	Sew sleeve placket buttonhole	36
8	Attach collar band	46	29	Sew sleeve placket button	65
9	Turn and press collar band	57	30	Set sleeve	58
10	Topstitch collar band	56	31	Topstitch armhole	64
11	Sew collar band buttonhole	32	32	Join side seam	23
12	Sew collar band button	41	33	Hem bottom	32
13	Set center front placket	37	34	Hem cuff	28
14	Hem right front edge	19	35	Runstitch cuff	57
15	Trim neckline	28	36	Turn and press cuff	49
16	Sew center front buttonhole	30	37	Topstitch cuff	50
17	Sew right front button	29	38	Sew cuff buttonhole	55
18	Hem pocket mouth	47	39	Sew cuff button	32
19	Crease pocket	66	40	Set cuff	43
20	Set pocket	54	41	Set and close collar	48
21	Sew yoke pleats	33

SAM: standard allowable minutes: time required to perform 100 grouped tasks.

Figure 4.

Heuristic module classification: (a) precedence relationship of tasks¹⁵ and (b) module classification.

Table 2.

Arbitrary skill levels of the workers for each task.¹⁵.

Task no.	Worker no.
Task no.	1	2	3	4	5	6	7	8	9	. . .	33	34	35	36	37	38	39	40	41
1	–	0.215	0.14	0.382	0.19	1.494	1.047	0.849	0.89	. . .	0.696	1.173	0.475	0.441	1.194	0.967	0.907	0.564	0.788
2	0.047	0.751	0.943	0.549	1.48	0.945	1.256	1.344	1.203	. . .	0.724	0.438	0.275	0.173	1.053	0.293	1.143	0.138	0.81
3	1.292	0.033	0.42	1.256	0.459	1.037	1.128	0.033	1.384	. . .	0.644	0.69	1.44	0.22	1.497	1.115	1.017	0.979	1.361
4	0.304	0.889	1.316	0.472	0.082	0.811	1.353	0.384	0.767	. . .	1.269	1	0.56	1.057	0.698	1.498	1.082	0.486	0.601
5	0.409	0.014	0.242	0.472	0.153	0.146	0.28	1.327	0.99	. . .	1.067	0.134	1.333	0.892	1.124	1.23	0.934	0.062	0.594
6	1.007	1.162	0.41	0.985	1.3	1.121	1.163	1.027	1.499	. . .	0.044	0.381	0.018	0.422	0.365	1.019	0.967	1.129	0.641
7	0.478	0.976	0.816	1.038	0.832	0.223	0.401	0.048	1.154	. . .	0.306	0.111	0.134	1.315	0.06	0.216	0.02	1.283	0.258
8	0.243	1.156	1.441	1.354	1.046	0.349	0.443	0.237	1.074	. . .	0.653	0.16	0.939	1.147	0.943	0.885	1.24	1.433	0.989
9	0.558	1.062	0.262	0.478	0.151	1.32	1.054	0.233	1.341	. . .	1.23	1.375	0.879	0.861	1.106	0.183	0.59	1.39	0.521
10	0.639	0.836	0.236	0.718	1.003	1.071	0.685	0.619	1.492	. . .	0.364	0.571	0.945	1.049	0.08	1.479	0.292	0.033	0.554
⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞	⁞
31	0.723	1.483	0.512	1.022	1.211	0.912	0.469	0.07	0.944	. . .	0.13	1.34	0.948	0.06	0.456	0.05	0.776	1.284	0.978
32	0.224	1.161	1.228	1.391	0.881	0.386	0.593	0.088	0.765	. . .	1.459	1.435	0.953	1.023	1.262	0.993	0.465	1.478	1.108
33	1.312	1.124	1.226	0.416	0.165	1.045	0.822	0.152	1.453	. . .	0.849	0.209	1.309	1.274	0.652	0.625	0.515	0.373	1.293
34	0.431	0.297	0.212	1.051	0.935	0.837	0.391	1.205	0.152	. . .	0.823	0.649	0.029	0.767	0.875	0.226	0.332	1.04	0.239
35	1.159	0.215	0.14	1.102	1.392	1.318	0.227	1.432	1.485	. . .	1.106	0.368	0.297	0.52	1.01	0.648	0.202	0.152	0.21
36	1.465	1.37	0.101	0.75	0.811	0.541	0.439	0.173	0.036	. . .	0.489	0.771	1.07	0.584	0.548	0.225	1.461	1.477	0.835
37	0.739	1.176	0.09	0.996	0.489	0.546	0.115	0.32	0.925	. . .	0.122	0.837	1.022	0.547	1.156	1.274	1.046	0.712	0.984
38	1.332	0.884	0.291	0.237	0.672	0.933	1.058	1.145	0.534	. . .	0.13	1.259	0.167	0.781	0.676	1.349	1.314	1.273	1.106
39	1.241	1.297	1.009	0.345	0.34	1.17	0.495	1.47	0.766	. . .	0.559	0.947	0.977	0.406	0.044	0.247	0.724	0.528	1.279
40	0.03	1.034	0.341	1.42	0.277	0.686	1.184	0.746	0.969	. . .	1.299	0.284	1.133	1.405	0.963	0.102	1.274	0.555	0.225
41	0.212	0.432	0.856	0.547	1.036	0.308	0.983	0.048	1.215	. . .	1.201	0.817	1.192	0.729	1.003	0.368	0.712	1.234	0.577

Module-based mixed task assignment

Tasks were classified into modules considering the precedence relationship (Figure 4(b)). Tasks #1–12, #13–20, #21–23, and #25–29, and #34–39 corresponded to the collar, front, back, sleeve, and cuff parts, respectively. Because the back part corresponded to only a few tasks, the tasks for the front and back parts were bundled and set as a torso module. Modules were defined for the other parts. Tasks #24, 30–33, 40, and 41, which corresponded to the combination of each part to complete the shirt, were assigned to a join module.

In previous studies, the number of workers in the assembly line was set as 41 or 13, and the output of shirts was set as 100. Therefore, to ensure a fair comparison with the results of the previous studies, the number of workers in the basic apparel assembly line was set as 13 and the output of shirts was set as 100 in this study. The assignment of 41 workers in the single task–single worker assignment method was not attempted. Instead of that, an experiment was conducted in which 41 workers performed tasks in three apparel assembly lines, by designing two additional assembly lines consisting of 14 workers each (from the remaining 28 workers).

Step 1: Task assignment

Table 3 summarizes the results of generating workstations for each module, grouping tasks, and assigning them to workstations using the module-based mixed task assignment technique. Four, three, two, two, and two workstations were generated in the collar, torso, sleeve, cuff, and join modules, respectively. The notation can be understood as follows: Grouped tasks 1-2-3-4(60) on workstation 1 meant that tasks #1, #2, #3, and #4 are assigned to workstation 1, and task #4 is a shared task with workstation 2. Moreover, 4(60) indicates that 60 out of 100 executions of shared task #4 are assigned to workstation 1. The number in [ ] indicates the cycle time for each module, which is the standard for grouping tasks in the module. The SAM is the time required to perform 100 grouped tasks assigned to each workstation. The results highlighted that the proposed algorithm successfully assigned tasks such that the SAMs of all workstations were in the range of 95%–105% of the cycle time for each module.

Table 3.

Workstation generation in modules.

Module	Workstation	Grouped tasks	SAM* (min)	$S A M / C T$ (%)
Collar [124.8 min]**	1	1-2-3-4(60)***	124.8	100.0
	2	4(40)-5-6-7(38)	124.8	100.1
	3	7(62)-8-9	120.4	96.5
	4	10-11-12	129.0	103.4
Torso [132.3 min]	5	13-14-15-16-17(63)	132.3	100.0
	6	17(37)-18-19-20(16)	132.4	100.0
	7	20(84)-21-22-23	132.4	100.0
Sleeve [129.5 min]	8	25-26-27(35)	129.4	99.9
Sleeve [129.5 min]	9	27(65)-28-29	129.6	100.1
Cuff [135.5 min]	10	34-35-36	134.0	98.9
Cuff [135.5 min]	11	37-38-39	137.0	101.1
Join [161.5 min]	12	24-30-31(76)	161.6	100.1
	13	31(24)-32-33-40-41	161.4	99.9

SAM: standard allowable minutes: time required to accomplish 100 grouped tasks.

The number in [ ] indicates the cycle time (CT) for each module.

***

The number in ( ) indicates the workload of each worker in the shared work.

Step 2: Worker assignment

Worker assignment was performed by applying the skill level of workers to each generated workstation. Workers with the smallest difference between the adjusted SAM and cycle time of the apparel assembly line were assigned sequentially and in order from workstation 12 with the longest SAM to 13 > 11 > 10 > 6 > 7 > 5 > 9 > 8 > 4 > 2 > 1 > 3 (Table 4).

Table 4.

Workstation formation.

Module	Workstation	Worker no.	Adjusted SAM* (min)
Collar	1	26	132.6
	2	17	132.6
	3	32	146.3
	4	6	128.8
Torso	5	22	137.5
	6	13	143.8
	7	30	137.2
Sleeve	8	23	133.8
Sleeve	9	28	136.8
Cuff	10	11	134.6
Cuff	11	1	134.7
Join	12	40	131.7
	13	35	143.0

SAM: standard allowable minutes: measured time to accomplish 100 grouped tasks.

Effects of step-by-step line balancing

The results were compared with those of previous studies to evaluate the effect of line balancing using the mixed task assignment technique. Because the existing studies used different performance indicators, the statistics and total production time were compared. First, the statistical metrics were compared with those of the existing studies involving 13 workstations (Table 5). In general, line balancing is aimed at ensuring that the processing time of each workstation is as uniform as possible. Therefore, a smaller difference between the maximum and minimum workstation processing time values and a smaller standard deviation corresponds to superior line balancing. The results of line balancing before and after the worker assignment were superior to those obtained in the previous studies. The inferior performance of the existing methods was attributable to the fact that the workstations were generated, and workers were assigned by simply grouping the tasks for each branch of the precedence relationship without considering task assignment. Task assignment was noted to benefit line balancing: The line balancing realized by mixed task assignment method at the task assignment stage was superior to that in the existing study. The mixed task assignment method improved the effect of line balancing based on worker assignment because the difference between the maximum and minimum values and standard deviation decreased after the worker assignment.

Table 5.

Statistics of line balancing methods for 13 workstations.

Statistics	Chen et al.¹⁷*	This study
Statistics	Chen et al.¹⁷*	Task assignment	Task + worker assignment
Mean (min)	605.0	134.5	136.4
Min (min)	91.5	120.4	128.8
Max (min)	2769.6	161.6	146.3
SD	755.6	12.2	5.0

The best results are indicated.

Next, the total production time was compared with those of the existing studies in which all 41 workers were assigned. To conduct an experiment with 41 workers, apparel assembly line 1 with 13 workers and 2 apparel assembly lines (2 and 3) with 14 workers each were designed (Table 6). The outputs of apparel assembly lines 1, 2, and 3 were set as 38, 34, and 28, respectively, considering the production capacity of each apparel assembly line.

Table 6.

Two apparel assembly line with 28 workers.

Module	Work station	Assembly line 2		Assembly line 3
Module	Work station	Grouped tasks	Worker no.	Grouped tasks	Worker no.
Collar	1	1-2-3-4(20)*	9	1-2-3-4(17)	33
	2	4(14)-5-6-7(13)	16	4(11)-5-6-7(11)	27
	3	7(21)-8-9	36	7(17)-8-9	24
	4	10-11-12	14	10-11-12	12
Torso	5	13-14-15-16-17(21)	19	13-14-15-16-17(18)	25
	6	17(13)-18-19-20(5)	4	17(10)-18-19-20(4)	7
	7	20(29)-21-22-23	20	20(24)-21-22-23	29
Sleeve	8	25-26-27(35)	2	25-26-27(10)	8
Sleeve	9	27(65)-28-29	10	27(18)-28-29	38
Cuff	10	34-35-36	5	34-35-36	37
Cuff	11	37-38-39	41	37-38-39	31
Join	12	24-30	3	24-30	15
	13	31-32-33(22)	34	31-32-33(18)	21
	14	33-(12)-40-41	39	33(10)-40-41	18

The number in ( ) indicates the workload of each worker in the shared work.

Table 7 compares the simulation results of the three apparel assembly lines with the reported results. Production in lines 1–3 required 63.1, 63.5, and 67.8 min, respectively. Thus, the total production time for 100 men’s shirts was 67.8 min. Despite the presence of one more worker and smaller output, the time for lines 2 and 3 was higher than that for 1 because the workers were assigned to each workstation in the following decreasing order: assembly line 1 > 2 > 3.

Table 7.

Total production time for different line balancing methods.

	Chan et al.¹⁵	Chen et al.¹⁷*		This study
	Chan et al.¹⁵	Chen et al.¹⁷*		Task assignment	Task + worker assignment
Number of workers	41	13	41	13	13	41
Total production time (min)	474.6	-	1282.4	170.0	155.1	67.8

The total production time is replaced with the total adjusted SAM because its value is not available, and the best results are indicated.

Because Chen et al. did not specify the total production time when assigning 41 workers, it was replaced with the total adjusted SAM. When 13 workers were assigned, the total adjusted SAM was not specified, potentially because Chen et al. realized worker assignment through the multiple task–single worker assignment method.

The total production time in this study was 85.7% lower than that reported by Chan et al. when 41 workers were assigned. The dramatic reduction in the total production time is attributable to the fact that the existing studies limited the apparel assembly line balancing problem to worker assignment and did not consider task assignment. In contrast, in this study, both task and worker assignment were sequentially performed. The total production time before worker assignment in the apparel assembly line with 13 workstations was 170.0 min, already 64.2% lower than that reported by Chen et al. with 41 workers. After worker assignment, the total production time reduced by 8.8% to 155.1 min. These results highlighted that that the beneficial effect of line balancing by task assignment was considerably greater than that based on worker assignment.

Conclusions

The modular production concept was applied to solve the apparel assembly line balancing problem. A mixed task assignment method was developed, in which two-step line balancing is performed to assign grouped tasks to each workstation by simultaneously implementing the single task–multiple workers assignment method and multiple tasks–single worker assignment method. Moreover, workers were assigned to each workstation by considering their skill levels.

The total production time required by 41 workers to produce 100 men’s shirts was 85.7% lower than that reported previously. Even when only 31.7% of the workers were assigned to the apparel assembly line, the total production time reduced by 67.3%. The reason for this drastic reduction in the total production time was that in this study, line balancing was performed by sequentially implementing task and worker assignment, unlike previous studies that limited the apparel assembly line balancing problem to worker assignment and did not consider task assignment. Given that the effect of line balancing by task assignment is considerably greater than that based on worker assignment, the apparel assembly line balancing problem in the design stage must not be limited to worker assignment, and task assignment must be simultaneously considered.

The results also highlighted that module-based task assignment reduces the complexity of workflow in the apparel assembly line because it prevents the grouping of unrelated tasks, and it is also for the design and operation of the apparel assembly line because the impact is limited to the module level even if the apparel assembly line is changed.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Sungmin Kim

References

Kagermann

Wahlster

Helbig

. Recommendations for implementing the strategic initiative industrie 4.0. Report, acatech, DE, April 2013.

Osterrieder

Budde

Friedli

The smart factory as a key construct of industry 4.0: a systematic literature review. Int J Prod Econ 2020; 221: 107476.

Sinha

Roy

. Reviewing cyber-physical system as a part of smart factory in industry 4.0. IEEE Eng Manag Rev 2020; 48: 103–117.

Kim

Ahn

Kang

, et al. A systematic review on smart manufacturing in the garment industry. Fashion Text Res J 2020; 22: 660–675.

Miguel

PAC

. Modularity in product development: a literature review towards a research agenda. Prod Manage Dev 2005; 3: 165–174.

Miller

Elgård

Defining modules, modularity and modularization. In: Design for integration in manufacturing proceedings of the 13th IPS research seminar, Fuglsoe, Denmark, 1998, pp.1–19. Fuglsoe: Aalborg University.

Senem

K-B

. Assembly line balancing in garment productionby simulation. In: Waldemar

(ed.) Assembly line - theory and practice. Rijeka: IntechOpen, 2011: 67–82.

Jung

W-K

Kim

Park

Y-C

, et al. Real-time data-driven discrete-event simulation for garment production lines. Prod Plan Control 2022; 33: 480–491.

Boysen

Fliedner

Scholl

A classification of assembly line balancing problems. Eur J Oper Res 2007; 183(2): 674–693.

10.

Bongomin

Mwasiagi

Nganyi

, et al. Improvement of garment assembly line efficiency using line balancing technique. Eng Rep 2020; 2: 1–18.

11.

Chen

C-C

L-H

, et al. Assembly line balancing in garment industry. Expert Syst Appl 2012; 39: 10073–10081.

12.

Kumar

Mohan

Mohanasundaram

KM.

Lean tool implementation in the garment industry. Fibres Text East Eur 2019; 27: 19–23.

13.

Cui

Yan

The optimization design of uniform’s hanging production line. Int J Clothing Sci Technol 2015; 27: 370–389.

14.

Fatima

Tufail

Improving efficiency of apparel manufacturing through the principles of resource management. Clothing Text Res J 2023; 41: 225–235.

15.

Chan

KCC

Hui

PCL

Yeung

, et al. Handling the assembly line balancing problem in the clothing industry using a genetic algorithm. Int J Clothing Sci Technol 1998; 10: 21–37.

16.

Zeng

Wong

W-K

Leung

SY-S

. An operator allocation optimization model for balancing control of the hybrid assembly lines using Pareto utility discrete differential evolution algorithm. Comput Oper Res 2012; 39: 1145–1159.

17.

Chen

GY-H

Chen

P-S

Dang

J-F

, et al. Applying meta-heuristics algorithm to solve assembly line balancing problem with labor skill level in garment industry. Int J Comput Intell Syst 2021; 14: 1438–1450.

18.

Lin

The single-row machine layout problem in apparel manufacturing by hierarchical order-based genetic algorithm. Int J Clothing Sci Technol 2008; 20: 258–270.

19.

Kayar

Akalin

Comparing heuristic and simulation methods applied to the apparel assembly line balancing problem. Fibres Text East Eur 2016; 24: 131–138.

20.

Kim

Development of a modular garment assembly line simulator. Int J Clothing Sci Technol 2020; 32: 645–659.