Assessing Shop Floor Layouts in the Context of Process Plans

Introduction

 According to Andersen et al., in today's globalized production environment, changes are inevitable, and manufacturers must take advantage of either new technologies, changes in products or manufacturing processes. Paydar & Saidi-Mehrabad point out that for a company to survive, it is very important to meet the needs of customers, who demand products with high quality and lower prices. Continuous improvement of processes and products are key to have loyal customers and corporate survival.

Andrade & Fernandes point out the importance of integrating the stages of the production process through the interoperable integration since its planning. In this context, Mahdavi et al. mention that the use of process plans with pre-established alternatives can be advantageous, since a previously selected alternative resource can be included during the process plan elaboration. The presence of pre-planned alternatives can provide more flexibility in decision making.

In turn, Khilwani et al., Silva & Rentes , Defersha & Hodiya and Sieckmann et al. show that the appropriate choice of the shop floor layout is essential to enable and increase the competitiveness of a company in today's globalized economy.

In the context of a cellular manufacturing system, its advantages are: smaller batch sizes, reduction of work in process, and reduction of flow time. Ferreira & Reaes emphasize that both the reduction of flow and setup times, as well as the flexibility of the part routes, are very important characteristics for the viability of the business. Seeking to combine these two characteristics, the concept of virtual cells arises.

According to Mclean et al., a virtual cell is composed of machines from different departments that are associated logically during a certain time interval. Although physically arranged as in the job shop arrangement, the layout works as a traditional manufacturing cell. A virtual cell is not identifiable by a physical cluster, but by data files and processes within a production controller.

Dakov et al. and Rabbani et al. point out that VCMS (Virtual Cell Manufacturing Systems) were a first attempt at practical use for the concept of virtual cells. They mention as an important advantage of virtual cells over other layouts the almost unlimited adaptability to different production volumes without reducing the efficiency and effectiveness of the system.

Paydar & Saidi-mehrabad cite as an advantage of virtual cells the fact that the cellular arrangement occurs virtually, which is appropriate in the case of significant variations in the demand of parts, not being necessary the physical rearrangement of machines.

Ramachandran & Prasad cite the better performance of the virtual cell layout in different scenarios. It is worth mentioning the swiftness with which the system responds to demand changes.

For the analysis of different types of layouts and their effects on the production systems, different situations can be simulated. For Freitas Filho, simulation allows to understand the dynamics of a system, to analyze and to predict the effects of changes. Watanapa & Wiyaratn emphasize the importance of simulation, since it enables considering different scenarios to analyze the same problem. Mesquita et al. point out that the use of simulation assists in the anticipation of analyzes of real problems found in companies.

In turn, Kleijnen et al. state that design of experiments (DOE) assists the analysis of simulation results. In addition, DOE avoids the trial-and-error technique for finding solutions.

Based on the standard adopted by Mello, who did not consider the type of layout, he attempted to show that the use of alternatives provides advantages to manufacturing systems. Reaes presented a more comprehensive research, concluding that the virtual cellular layout performs better than traditional job shop and cellular layouts. However, Reaes did not consider in detail the design of products, limiting to 50 different types of parts. The parts in the present work were composed randomly, that is, a universe of features was stipulated, from which an infinite number of different products could be generated, trying to stimulate a scenario as close as possible to reality, where the market demands more and more flexibility, variety of product mix and shorter life cycles of products.

With regard to relevance, the results of the virtual cellular layout with a small number of resources, 21 in total, are highlighted, complementing the results found by Reaes, with 50 features, showing the applicability of the virtual cellular layout.

The choice of these three layouts was due to the virtual cellular layout linking the positive characteristics of both the job shop and the cellular layouts.

In this scenario, the objective of this paper is to compare the job shop, virtual cell and traditional cell layouts, identifying for which conditions each layout achieves better performance for different response variables.

Source: Claudio Decker Junior, João Carlos Espíndola Ferreira, Elisa Henning, and Carla Roberta Pereira, https://www.scielo.br/scielo.php?script=sci_arttext&pid=S0103-65132019000100202
This work is licensed under a Creative Commons Attribution 4.0 License.

Brief literature survey

This section contains a brief literature review of works that deal with manufacturing layouts, and process planning within different layouts.

Layouts

A layout establishes the physical location of the transforming resources, i.e., the facilities, machines, equipment, operators, etc.. In addition, it establishes how transformed resources (i.e., raw materials, materials, parts, etc.) should flow along operations. According to Silva & Rentes , the physical location has impacts in several aspects on the shop floor, and Khilwani et al. point out that the type of layout should focus on the flow of materials, seeking to increase production efficiency and reduce costs. The provision of transforming resources in a facility directly affects production costs and throughput. An adequate allocation of resources contributes significantly to the increase of efficiency of the operations and reduces transportation costs.

Virtual cells operate virtually as cells from a logical point of view, that is, resources are dedicated to manufacture a family of parts, similarly to traditional cells. At the moment when family of parts is completely manufactured, the virtual cell formation is undone logically, and each individual resource being able to re-group with others constitute a new virtual cell.

Figure 1 shows an example of virtual cell layout, where three virtual cells are temporarily formed to manufacture one production order each, with the geometric figures representing different types of machines present on the shop floor.

Figure 1 An example of a virtual cell layout.

Arkat & Ghahve point out that the temporary virtual arrangement is often adjusted, being changed according to the characteristics of the batches of parts to be manufactured by the production control and available resources.

It is noteworthy that in the literature there are several approaches regarding the layout theme, for which a simplified Systematic Literature Review (SLR) was developed based on the guidelines of Tranfield et al.. Karl et al. cites the importance of systematic literature review as a way to build approaches from a large number of studies, without neglecting relevant research. Table 1 presents the 5 steps that describe the literature review protocol (SLR) from Tranfield et al. and Karl et al..

Table 1 SLR protocol.

Process planning in the context of shop floor layouts

The task of process planning involves the interpretation and conversion of the product design for its manufacture.

According to Ferreira & Wysk, there is often more than one alternative resource to perform an operation, and consequently to manufacture a part. Colosimo et al. point out that, normally, the elaboration of process plans does not consider alternative resources, although Zattar et al. highlight the importance of the results achieved with the use of alternative plans.

Deja & Siemiatkowski point out that the availability of process plans with alternatives is an important factor to integrate product design, process planning and production scheduling.

Fung et al. mention that the capacity of each machine and the operations that it can be perform are well defined. However, some operations can be performed on more than one machine, which results in the dynamic aspect of process planning.

Lv et al. point out that studies of virtual manufacturing cells focus, in addition to layout, algorithms for virtual cell design, process planning, and process planning models.

Arkat & Ghahve identified that most researches assumed that set-up times established in process plans are independent of production sequencing, thus being neglected or included in processing times, which simplifies the analysis but affects the outcome of the solution.

Dakov et al. emphasize the use of virtual cell layout in conjunction with process planning. They developed a theoretical model for the cellular arrangement based on workstations and an algorithm that elaborates the process planning simultaneously.

Morales & Ronconi considered mathematical models to solve the production scheduling problem of the classic job shop model, evaluating six different formulations and solve 45 problems from the literature. They consider the static deterministic case, where the processing time of each operation and the sequence of operations for each task are known and not variant, and all machines and tasks are available at the beginning of processing.

Fuchigami et al. evaluated 12 methods to minimize the total production sequencing time (makespan). The setup times were considered and their emphasis was on flexible flow lines.

There is a gap in the literature in the effective comparison between different types of layouts, in such a way that it can help the shop floor professional to specify the best layout for each scenario. In this context, this study seeks to analyze and compare the performance of three different types of layout through experiments based on simulations under conditions of a make-to-order (MTO) manufacturing system (i.e. a system that produces parts to satisfy orders with specific due dates). The method used in this work is presented in section 3.

Description of the proposed method

This section contains a description of the proposed method for the evaluation and comparison of three different manufacturing layouts using simulation and design of experiments.

Design of experiments

According to Sanchez, experimental design employs factors, which are independent variables, to test the impact on the outputs of the experiments, which are the response variables. As stated by Montgomery, this technique is useful to verify if a given factor influences, or not, the response of a system.

For Gomes and Costa, an experimental problem can be supported by the design of experiments and the statistical analysis of the data. There are several types of data collection strategies for DOE, and one of them is the complete factorial design.

Table 5 shows the control factors used in this paper.

Table 5 Control factors used in the proposed method.

Simulation model

With regard to the simulation model, this task was divided into two stages. In the first stage, the production orders of the batches of the parts are carried out, in which the following activities are performed: generation of the parts’ features, specification of the quantity of duplicate features, specification of the batch size, specification of the demand type, generation of the machining sequence, and calculation of the machining times. Chin emphasizes the importance of batch size analysis, evaluating different strategies of batch sizes applied to virtual cells, seeking to reduce manufacturing times.

In the second stage, the system is simulated using the Arena software. The results obtained are stored in a spreadsheet for later statistical analysis. The flowchart shown in Figure 2 illustrates the stages and activities of the proposed method.

Figure 2 Stages and activities for performing the simulation.

Application of the proposed method

Three different systems were modeled to evaluate the response variables according to the control factors. For all three models, a make to order (MTO) production system was considered. A system for the job shop layout was modeled, another for the traditional cellular layout and a third for the virtual cell layout.

In the first stage of the simulation model, the MS Excel random tool was used to generate batches of pieces, and all the pieces were randomly generated from the 40 possible features. The process plans were generated from commercial data of machining times to execute each feature, assuming that the machines of the model are able to execute these features. Table 6 presents the time of execution each feature used in the model.

Table 6 Time of execution each feature.

Results

This section is subdivided into two steps. In the first step the results of the analysis of variance (ANOVA) for each type of layout are obtained, comparing each control factor with each of the response variables. Then, in the second stage the layout is analyzed as a control factor of the experiment, and it is verified for which response variables the layout is significant.

Results for each layout

In order to facilitate the understanding of the results, a summary is presented in Table 8 , which shows the relationship between each control factor and the response variables for each of the three layouts. It should be mentioned that the analysis of statistical significance was done by using ANOVA. The "s" was used to represent the statistical significance between the control factor and the response variable, while "ns" illustrates the non-existence of statistical significance. In both cases the level of significance was 0.05.

Table 8 Comparison between control factors and response variables for each layout.

Comparison between layouts

In this second stage of results analysis the type of layout was included as a control factor with three levels: job shop, traditional cell, and virtual cell. It was verified how the response variables behaved in relation to the type of layout. This analysis of statistical significance was based on the value of P using ANOVA. It is noticed in Table 9 that the type of layout has statistical significance ("s") for all response variables to the significance level of 0.05.

Table 9 Comparison between control factors and response variables.

Figure 7 Main effects plot of the final time in the layouts.

The cells highlighted in Table 10 shows the three factors in which the confidence intervals do not intersect between the virtual cell and the traditional cellular layouts.

Table 10 Confidence intervals for the three layouts.

Figure 8 Main effects plot of the layouts for throughput.

Figure 9 shows how the layouts behaved in relation to the average work in process. In this response variable, the job shop and virtual cell layouts again present better performance, and the virtual cell layout was still better than the job shop. The virtual cell layout was better in approximately 27% in relation to the traditional cell layout, while for the job shop layout the superiority was 10%. This shows that the virtual cell layout can be located between job shop and traditional cell layouts, having advantages of both. However, observing the confidence interval (confidence level of 95%) ( Table 6 ), statistically it cannot be stated that these values are different.

Figure 9 Main effects plot of the average WIP in the layouts.

Figure 10 illustrates the behavior of the type of layout with respect to the response variable referred to as average resource utilization. Once again, the job shop and virtual cell layouts outperformed the traditional cell layout. The virtual cell layout had a resource utilization 20% higher than the traditional cell layout. Observing the confidence interval (confidence level equal to 95%) ( Table 6 ), it can be stated that the values between the virtual cell and traditional cell systems are different.

Figure 10 Main effects plot of the resource use in the layouts.

Figures 11 and 12 show the behavior of the layouts with respect to the waiting total time and lead time variables, respectively. It is noticed a very similar behavior for both cases, where again the virtual cell and job shop layouts performed better.

Figure 11 Main effects plot of the waiting time in the layouts.

Figure 12 Main effects plot of the lead time in the layouts.

With regard to the waiting time, the performance of the virtual cell layout was again better than the others, about 10% in the average with respect to the job shop layout, and far better than the traditional cell layout (about 50%). Observing the confidence interval (confidence level of 95%) ( Table 6 ), it can be stated that the values between the virtual cell and traditional cell layouts are different.

Regarding lead time, on average the virtual cell layout had a good result, approximately 7% shorter than the job shop, and 50% shorter than the traditional cell layout. Observing the confidence interval ( Table 6 ) (confidence level equal to 95%), it can be affirmed that the values between the virtual cell and traditional cell layouts are different. In summary, Table 11 presents the performance of each layout for the response variables, always having the virtual cellular layout as the reference of the comparison. The minus sign indicates a worse result when compared to the virtual cell, and the plus signal indicates a better result compared to the virtual cell.

Table 11 Summary of comparison percentages.

Conclusion

The aim of this research was to compare the virtual cell layout with other types of layout. It has a performance similar to the job shop layout, but it is better than the traditional cell layout. In addition, a greater number of control factors and response variables were considered compared with previous studies such as Mello and Reaes. This greater amount of response variables provided a broader analysis of the three layouts.

In general, such different results found in the case of the traditional cell layout can be understood by the wide variety of parts. Similar results were found by Ramachandran & Prasad . All parts in each batch were generated from 40 pre-established features, and Excel's random function resulted in high batch variability, leading to a poor performance of the traditional cell layout.

A significant factor for the poorer performance of the traditional cell layout was the parts supplier. By leaving the supplier working exclusively in the cell, there was an imbalance of this operation, that is, while in a cell the supplier worked uninterruptedly, while the supplier became idle in the cell for cylindrical parts. This behavior influenced the processing times, and reduced throughput and resource utilization.

Another issue to be considered is the continuous flow, which, in the case considered in this paper, did not take place as in the traditional cell layout, since the production was based on batches of parts and not one piece flow. The setup times were the same for all three layouts, and the only difference occurred in entry of the batch in the cells, when the time was ten times shorter for the traditional cell layout when compared to the virtual cell and job shop layouts.

Moreover, as Ramachandran & Prasad suggest, the transition from the job shop layout to the virtual cell layout is easier when compared with the costs involved in the case of the transition to the traditional cell layout.

The objective of showing a knowledge gap with respect to virtual cells signals the importance of this topic, and the possibilities that have not yet been investigated about it in manufacturing companies. As a suggestion of future work, the amount of levels of some factors can be increased, which would improve the focus on the influence of certain factors on the response variables. Another suggestion would be the exclusion of one or more factors, seeking to focus on factors that can be directly controlled in the manufacturing system without depending on the interference of the customers. For example, factors such as features per part are customer requirements, and can hardly be influenced by the system.

With regard the papers limitations: the use of a process plan with alternatives, it is suggested to include an additional level for this factor, or two different penalties, in order to investigate the gains that are pointed out in the literature. It is suggested to address the issue of operator flexibility, which was implemented in a simplified way in this research. A study should be carried out relating the results found in this paper with the equipment costs, which may originate from obsolescence, maintenance, upgrade and replacement.