Abstract
Existing methods for evaluating manufacturing process chain complexity consider the number of machines, state of machines, number of parts, operation time, and processing sequence of parts. However, such evaluation methods ignore human factors. To consider human factors, human cognitive decision-making process factors are considered in the complexity evaluation of production processes. Accordingly, a new objective evaluation method of the human factor complexity is proposed. In the proposed method, sewing operations are taken as an example, and the human factor complexity is classified into perceived and cognitive complexity. Information entropy is used to measure cognitive complexity according to the type and quantity of sewing workers’ cognitive activities. The results show that various methods have significant differences in the evaluation of the complexity level of the production process chain. Specifically, the calculation results of the proposed evaluation method are much greater than those of other methods. This indicates that human cognitive and perceived complexities account for a large proportion. Therefore, human factor complexity cannot be omitted.
1. Introduction
Currently, the applications of manufacturing system complexity are an active research area. However, complexity has no clear definition [1]. Scholars have defined manufacturing system complexity from different perspectives [2]. Isik [3] argued that complexity may have adverse effects, such as high operating costs, delivery delays, and inventory shortages. Complexity also negatively impacts the attributes of manufacturing systems, such as productivity [4], profit [5], and quality [6]. The research on the complexity of manufacturing systems mainly focuses on processing and assembly. Processing is the process of machining raw materials and semifinished products into target requirements through certain processes and methods. Processing complexity is usually divided into static complexity and dynamic complexity. Static complexity is structural complexity, which is related to the structure and configuration of manufacturing systems. It includes various elements such as people, machines, cache, logistics, and the relationship among them. Dynamic complexity refers to the uncertain factors and system probability in a specific period, such as the adjustment of a plan, change in the order, and task deviation [7]. Frizelle and Woodcock [2] were the first to use information entropy to evaluate manufacturing system complexity. However, there is no in-depth analysis of applicability and effectiveness with respect to complexity. Deshmukh [8] used information entropy to evaluate structural complexity and provided the basis for static and dynamic complexity evaluation. Modrak and Zuzana [9] proposed a method for evaluating static complexity and analysed the complexities of two different manufacturing layouts using the method. This method mainly considered factors such as equipment, parts, and processing sequences, but human factors were ignored. Zhang [10] proposed a method for evaluating static complexity by considering the system’s structure and components based on information entropy. Zhang et al. [11] established the static and dynamic entropy models of a cell manufacturing system to reduce its structural complexity. Most of these studies developed complexity models and evaluation methods by describing the state of a manufacturing system. However, manufacturing systems are complex and dynamic; therefore, a method to accurately describe the actual state of a system still needs further research. Complexity theories are most widely used in assembly. As assembly is the final process of a manufacturing system, assembly workers need to complete assembly tasks within a limited time. The complexity of the assembly process is mainly related to uncertainty, work content, and time pressure [12]. It is also related to the diversity of products or parts under a customised production mode [6, 13]. The complexity caused by product diversity and operation complexity in assembly has been the focus of various studies. Falck et al. [14] proposed the basic criteria of assembly system complexity evaluation from 16 dimensions, such as material, operation instruction, and operation time. However, these basic criteria mainly consider objective factors. He et al. [15] proposed a method for modelling and evaluating the structural complexity, process complexity, and operation complexity of an assembly system, but this method only analyses the simple mixed flow assembly system. Zhu et al. [16] proposed a measurement method of operator selection complexity, mainly considering product diversity and information in the assembly process, and developed a complexity model of a multilevel mixed-model assembly system suitable for serial structures. This mathematical model reveals the propagation mechanism of complexity in multilevel mixed-model assembly systems. Based on the modular arrangement of predetermined time standards (MOD method), Alkan et al. proposed a method to measure the operation complexity of a manual assembly system from three aspects, namely, action effort, operation diversity, and operation scale, and verified the effectiveness of this method through simulation [17]; however, the actual case was not analysed. Zaeh et al. [18] argued that workers’ participation in a task is based on three interrelated factors, namely, time, cognition, and knowledge. However, the knowledge and cognitive factors still need to be further adjusted to ensure that they are applicable to any assembly operation. As the garment industry belongs to the fast fashion industry, it is characterized by multiple varieties and small batches. The garment production is still labor-intensive and the production process is highly complex. Weaving production has the characteristics of processing and assembly, which requires high technical level of sewing workers. Sewing is an important part of the weaving production system. During the sewing process, the sewing worker works synchronously with the machine, and the sewing worker plays a leading role in the production process. In weaving production system, weaving complexity has an important impact on production efficiency and product quality. However, little is known about the effect of weaving complexity on task performance. The research results of manufacturing system complexity based on process or assembly cannot be directly applied to the field of weaving. To address this, in this study, a new production process complexity measurement method with both processing and assembly characteristics is developed to provide theoretical support for improving the production management and decision-making level of garment weaving industry.
2. Related Studies
Complexity measurement is the basis for managing and controlling complexity. Every manufacturing company should have the most appropriate level of complexity. Before adjusting the level of complexity to an appropriate or ideal level, it is necessary to measure the complexity [19–21]. However, the quantification of complexity is difficult [21]. Brinzer and Schneider [23] classified the measurement methods of manufacturing system complexity into two types: objective and subjective complexity measurements. Objective complexity considers the measurement of the internal factors of a manufacturing system, such as the configuration and structure of the system. Objective complexity is measured using information entropy [2], mathematical modelling [24], and information technology [25]. Subjective complexity measurement considers human factors, such as human perception and cognitive complexity. The methods for measuring subjective complexity are information entropy [26] and questionnaire methods [27], as shown in Figure 1.
Information entropy is frequently used to evaluate manufacturing system complexity. The existing methods for assessing production process chain complexity are as follows.
Method 1 (static complexity evaluation method proposed by Deshmukh et al. [24]): this method is relatively simple. It considers three parameters of a manufacturing system, namely, the number of machines, number of operations, and number of parts. Its formula is as follows [24]:where r, m, and n represent the numbers of machines, operations, and parts, respectively.
Method 2 (static complexity evaluation method proposed by Frizelle and Woodcock [2]): this method uses information entropy to evaluate static complexity. However, it only considers the processing state of machines. When the machine processes different parts, it is regarded to be in different states. Its formula is as follows [2]:where M represents the number of machines, Sj represents the number of possible planned states of the j th machine can be in, and Pij represents the possibility that the j th machine is in state i. In other words, Pij is expressed as the ratio of the machine operation time to the production cycle.
Method 3 (static complexity evaluation method proposed by Zhang [10]): this method also uses information entropy to evaluate the static complexity of a manufacturing system. In this method, the processing and idle states of the machine are considered. However, similar to Method 2, when different machines process different parts, they are regarded to be in different states. Its formula is as follows [10]:where M represents the number of machines, Sj represents the number of possible planned states of the j th machine can be in, and Pij represents the possibility that the j th machine is in state i. Method 4 (static complexity of a manufacturing process chain considering the processing sequence [8]): this method considers the processing sequence of parts based on two assumptions. First, in the manufacturing process chain, machines are usually arranged in series or parallel. According to the two layouts, the probability of the kth part being processed on the j th machine is expressed as Pjk. Second, when the part is processed on a serial machine, the value of Pjk is 1/MS, where MS is the number of serial machines. When the part is processed on a parallel machine, the value of Pjk is 1/MP, where MP is the number of parallel machines. When the part is processed on a parallel machine with a serial/parallel mixed layout, Pjk is expressed as 1/(MSMP). Its formula is as follows [9]:where Pjk expresses the probability that the kth part is processed on the j th machine, according to the processing sequence of the parts. N represents the number of parts processed in the manufacturing process chain, and M represents the number of machines involved in the manufacturing process chain.
A summary of the aforementioned evaluation methods is provided in Table 1.
3. Method for Evaluating Sewing Process Chain Complexity considering Human Factors
Based on Method 4, we evaluate the manufacturing process chain complexity of garment production considering human factors. Because sewing plays a crucial role in garment production, we mainly discuss the method for evaluating sewing production process chain complexity. The evaluation framework is shown in Figure 2.
The perception and cognitive complexities are mainly discussed when considering human factors. The perception and cognitive processes can be regarded as processes of information processing, as shown in Figure 3. The information input is considered a perception process and considers the relationship among products, tools, processes, and work areas. Information processing and information output are classified as cognitive processes. The cognitive process has four cognitive functions based on the second-generation human reliability analysis method: observation, interpretation, planning, and execution. These cognitive functions correspond to 15 cognitive activities. According to the types and quantity of cognitive activities corresponding to sewing operations, information entropy is used to evaluate cognitive complexity. The flow chart of the human factor complexity evaluation is shown in Figure 4.
3.1. Method for Evaluating the Perceived Complexity of Sewing Workers
In the garment production process, information input is classified as a perception process. Sewing workers’ perception of information is influenced by many factors, such as the quantity and diversity of information [26]. The production process mainly involves four types of information: product information (X1), process information (X2), tools and equipment information (X3), and workplace information (X4). For any information variable X, which has n possible values (Y1, Y2, … , Yn), we assume that the information variables have specific relationships among them, such as calling, being-called, self-relation, and no-relation. The relationship can be defined as R = (self-relation, calling, being-called, no-relation) = (1, 1, 1, 0). Rij (i = 1, 2, … , nj; j = 1,2,3,4) represents the relationship between a perceived information variable and other perceived information variables. According to Kong (2018), the formula for calculating the perception complexity of the m th sewing process is expressed as follows [26]:where
3.2. Method for Evaluating the Cognitive Complexity of Sewing Workers
The cognitive process involves information processing and information output based on the process shown in Figure 3. The second-generation human reliability analysis method classifies cognitive functions into four categories: observation, interpretation, planning, and execution. These cognitive functions can be further divided into 15 cognitive activities: coordination, contact, comparison, diagnosis, evaluation, identification, implementation, maintenance, monitoring, observation, planning, recording, adjustment, scanning, and inspection. When any cognitive activity belongs to any cognitive function, it is represented by “√.” The corresponding relationship among them is shown in Table 2.
s is used to indicate work step. The mth sewing process is decomposed into ns work steps. According to the second-generation human reliability analysis method, the type of cognitive activity and cognitive function are judged for each work step, COsaf represents the ath cognitive activity and the fth cognitive function of the sth work step, as shown in Table 3. According to Table 2, when the sth work step is judged as the ath cognitive activity and this cognitive activity belongs to the fth cognitive function, COsaf = 1; otherwise, COsaf = 0. COf represents the sum of the ath cognitive activity and this cognitive activity belongs to the fth cognitive function. Pcof represents probability of the fth cognitive function in the mth sewing process. According to Table 3, cognitive complexity of the mth sewing process is quantified using information entropy, which can be expressed as follows:
3.3. Method for Evaluating Sewing Production Process Chain Complexity
Based on the aforementioned qualified method, the sewing production process chain complexity is defined and quantified as follows:wherenp represents the number of perceived complexities. nc represents the number of cognitive complexities.
3.4. Case Analysis of the Sewing Process Chain Complexity Evaluation
Currently, most garment production enterprises adopt the bundle mode. Owing to the increasing demand for multiple varieties and small batches, the complexity continues to increase. In production workshops, each cut piece is processed according to an arranged order. Sewing workers and machines work synchronously. Particularly, sewing workers play a significant role in ensuring efficiency and quality. Therefore, throughout the evaluation of the sewing production process chain complexity, we must consider the processing sequence of cut pieces and human factors. In this study, we investigate garment enterprise A and adopt the production process of a women’s clothing workshop as an example. We use the evaluation methods mentioned earlier to evaluate the production chain complexity. These evaluation methods are compared and analysed to determine the differences between them. The layout of the women’s clothing workshop is shown in Figure 5.
As shown in Figure 6, the machines found in the women’s clothing sewing workshop mainly include the following types:(a)Sewing machines: they are the most important machines in garment production, as shown in Figures 6(a) and 6(b)(b)Instruments for ironing: they are used for flat ironing, shaping, and other operations, as shown in Figure 6(c)(c)Special equipment: they are used to complete special sewing operations, such as sewing buttonholes, attaching sleeves, and pressing, as shown in Figures 6(d)–6(h)
Group A, which processes windbreakers (18SSF-1), is used as an example for evaluating the production process complexity. The cut pieces of women’s windbreakers are shown in Table 4, where J represents the machine labels. The machines are arranged in series: J1, J2, …, J18 (J1–J7 represent the sewing machines or ironing equipment, and J8–J18 represent the special machines).
3.5. Evaluation Results of Method 1
According to Method 1, in the women’s clothing sewing workshop A, with m = 96, n = 28, and r = 18, the sewing production process chain complexity is calculated using (1) as follows:
3.6. Evaluation Results of Method 2
The windbreaker (18SSF-1) production in the women’s clothing workshop A is used as an example, and the standard operation time for sewing the windbreaker is shown in Table 5.
According to Method 2, the sewing production process chain complexity is calculated using (2) as follows:
3.7. Evaluation Results of Method 3
Method 3 considers the processing and idle states of a machine. The sewing production process chain complexity can be calculated using (3) as follows:
3.8. Evaluation Results of Method 4
The sewing process is similar to the assembly process, and the cut pieces can be regarded as individual parts. Accordingly, the method proposed by Modrak and Zuzana [9] for evaluating manufacturing process chain complexity is used to present the process chain diagram of cut pieces using sewing equipment, as shown in Figure 7. Cut pieces are indicated by “ ,” sewing equipment is indicated by “ ,” and the processing route of cut pieces is indicated by “ .” The cut piece is indicated by a solid line in the sewing equipment processing and by a dotted line in special machine processing. Taking 18SSF-1 as an example, its cut pieces comprise 28 pieces, including eight sleeve cut pieces (X1–X8), six pocket cut pieces (D1–D6), four front cut pieces (Q1–Q4), two rear cut pieces (H1 and H2), four collar cut pieces (L1–L4), two hanging cut pieces (G1 and G2), and two belt cut pieces (Y1 and Y2). The processing of each cut piece needs to be completed using J1–J18 sewing equipment. We draw the process chain of each cut piece in each sewing equipment according to the cutting process and processing sequence. Taking the sleeve cut piece X1 of 18SSF-1 as an example, according to its process requirements, the processing of X1 needs to be performed on the sewing equipment J2, J1, J6, J12, J10, J3, and J5. The processing sequence is J2 ⟶ J1 ⟶ J6 ⟶ J12 ⟶ J10 ⟶ J3 ⟶ J5 ⟶ J10. The process chain of cut piece X1 is drawn according to the processing sequence. Then, the process chains of all cut pieces are drawn using the same method. Consequently, the manufacturing process chain of 18SSF-1 is formed.
The example shown in Figure 7 for cut piece X1 has the following processing order: J2 ⟶ J1 ⟶ J6 ⟶ J12 ⟶ J10 ⟶ J3 ⟶ J5 ⟶ J10. These machines are all arranged in series, and the probability of each machine being used is equal, equal to 1/8 (e.g., Pj1 = 1/8). Therefore, the process static complexity of cut piece X1 is calculated using (4) as follows:
The production process static complexity of the other cut pieces is calculated similarly. The calculation results are shown in Table 6.
Based on (4), the production process chain static complexity for group A is calculated as follows:
4. Results of Evaluating Production Process Chain Complexity considering Human Factors
4.1. Perception Complexity Evaluation
The information entropy method is used to evaluate the perception complexity. The production characteristics and basic attributes of the resources required by sewing workers to complete the sleeve setting process are described in Table 7.
The four product information variables in the information acquisition phase of the setting of the sleeve operation are as follows: X1 = (three product variables: semifinished clothes, sleeves, and thread), X2 = (four process variables: alignment, suturing, thread removal, and inspection), X3 = (three tooling variables: sewing machine, scissors, and mannequins), and X4 = (two workplace variables: workbench and mannequins’ area).
The relationship matrix can be obtained from the information relationship diagram shown in Figure 8. Table 8 lists the relationships between the different information variables and the perception complexity results of the product information variables (X1). The method for evaluating the perception complexity of the other production information variables (X2, X3, and X4) is the same as that of X1. The perception complexity of the mth sewing process is obtained using equation (5)–(7) as follows:
4.2. Cognitive Complexity Evaluation
Take setting in sleeve process as an example, the operation of setting in sleeve is decomposed into 23 work steps. According to the second-generation human reliability analysis method, the type of cognitive activity and cognitive function are judged for each work step. The 23 work steps are analysed; when the work step belongs to cognitive activity, it is represented by “ √ ”; the results as shown in Tables 9 and 10.
According to Tables 2, 3, 9, and 10, the cognitive complexity for setting the sleeve is then calculated by (8); the results are shown in Table 11.
The sewing process of the women’s windbreaker involves 96 processes. The same method as the one for setting the sleeves is adopted using (5) and (6). The perception complexity and cognitive complexity of each sewing process are calculated. The calculation results are summarised in Table 12.
According to equations (10)–(13),
The evaluation results show that Method 1 considers the machines, parts, and operations in the production process. Method 2 considers the machines and operations. Method 3 considers machines and parts and operations. Method 4 considers machines, operations, and processing sequence of parts. And method 5 (proposed in this study) incorporates the human factors based on Method 4, as shown in Table 13. With the addition of the human factors, the complexity of the production process chain increases significantly. This indicates that human cognitive and perceived complexities account for a large proportion. Therefore, human factor complexity cannot be omitted.
5. Conclusions, Discussion, and Future Studies
5.1. Conclusions and Discussion
Sewing operations are highly dependent on the workers. During the sewing process, cognition and perception complexities have an impact on weaving efficiency and garment quality. In this study, complexity theory is applied to garment production. A new evaluation method of the production process chain complexity considering human factors is proposed. Based on the quantitative results of human factor complexity shown in Table 12, changes in perceived complexity and cognitive complexity are not necessarily synchronous. That is, for any process, if the perceived complexity is high, then the cognitive complexity is not necessarily high, and vice versa. However, for processes with relatively complex sewing processes and difficult operations, such as setting in sleeves, the perceived complexity and cognitive complexity are relatively high. Meanwhile, if the operation time of a sewing process is long, its perceptual complexity and cognitive complexity are not necessarily high. For example, the operation time of ironing bag covers is 150 s, and its perceptual complexity is 2, whereas the operation time of cufflink folding is 36 s and its perceived complexity is 2.584. Based on the comparative analysis results in Table 13, the complexity of the production and manufacturing process chains significantly increases when human factors are considered. The proposed evaluation method is very useful for the complexity evaluation of the production process chain, especially for manual manufacturing systems. Furthermore, factors in the manufacturing process, such as machines, parts, operation, and human factors, are all considered in the proposed method. Human factors are particularly described in detail.
5.2. Future Studies
From this analysis, the proposed evaluation method provides theoretical support and evaluation tools for garment shops. Although this study did not investigate how complexity affects weaving system efficiency and product quality, the proposed method provides an algorithm tool for further research in this field. Vidal and Hernandez [28] argued that it is necessary to reduce complexity in manufacturing systems. In future studies, we shall focus on reducing the complexity of sewing production chains using mathematical modelling, simulations, and other methods to improve production efficiency and reduce the defect rate. [22].
Data Availability
The data that support the findings of this study are available from (third party) (W&F Bird Group co.,Ltd.). Restrictions apply to the availability of these data, which were used under license for this study. Data are available from http://user52060.nz86.com/.
Conflicts of Interest
The authors do not have any conflicts of interest to declare regarding the publication of this paper.
Acknowledgments
The authors also would like to thank Jilin W&F Bird Group Co., Ltd., for allowing the use of the research sites. This study was supported by the Scientific and Technological Development Plan of Jilin Provincial Science and Technology Department (Grant ID 2020122355JC).