You are on page 1of 17

484

Int. J. Productivity and Quality Management, Vol. 7, No. 4, 2011

The effect of layout design on productivity: an empirical study Shahrul Kamaruddin* and Sok Yee Khoo
School of Mechanical Engineering, Universiti Sains Malaysia, Engineering Campus, Nibong Tebal 14300, Penang, Malaysia Fax: +604 5941025 E-mail: meshah@eng.usm.my E-mail: khoosokyee@gmail.com *Corresponding author

Zahid A. Khan and Arshad Noor Siddiquee


Faculty of Engineering and Technology, Department of Mechanical Engineering, Jamia Millia Islamia, Central University, Jamia Nagar, New Delhi 110025, India Fax: +91 11 26981259 E-mail: zahid_jmi@yahoo.com E-mail: arshadnsiddiqui@gmail.com
Abstract: This paper presents a comparative study that uses simulation to investigate the performance of different layout designs with respect to different levels of model and head count variability. There are three types of layout designs being examined in this paper: flow line, job shop and cellular layout. These three layouts are evaluated based on three selected performance measures which are throughput time, lateness and labour productivity. These three performance measures are selected because they relate with the overall layout productivity. The simulation model is build by using the WITNESS simulation package and the simulation results are tested in the F-test and Newman-Keuls test. In this study, the radio-cassette player production system has been used as an empirical study. The results show that the effect of headcount variability and model variability do create an impact towards the performance of flow line, job shop and cellular layout. Keywords: layout design; flow line; job shop; cellular layout; simulation; productivity; empirical study. Reference to this paper should be made as follows: Kamaruddin, S., Khoo, S.Y., Khan, Z.A. and Siddiquee, A.N. (2011) The effect of layout design on productivity: an empirical study, Int. J. Productivity and Quality Management, Vol. 7, No. 4, pp.484500. Biographical notes: Shahrul Kamaruddin received the BEng (Hons) degree from the University of Strathclyde, Glasgow, Scotland in 1996, the MSc degree from the University of Birmingham, UK, in 1998 and the PhD from the University of Birmingham in 2003. Currently, he is an Associate Professor with Copyright 2011 Inderscience Enterprises Ltd.

The effect of layout design on productivity


the School Mechanical Engineering (under the manufacturing engineering with management programme), Universiti Sains Malaysia. He has various past experiences with manufacturing industries from heavy to electronics industries, especially in the field of industrial engineering, manufacturing processes and product design. He has more than 20 publications in reputed international and national journals/conferences. His current research interests include simulation and modelling of manufacturing systems, production planning and control, maintenance management and application of artificial intelligence techniques in manufacturing. Sok Yee Khoo received the Bachelors Degree in Manufacturing Engineering from Universiti Sains Malaysia. Zahid A. Khan received the BSc in Mechanical Engineering and the MSc in Industrial and Production Engineering from Aligarh Muslim University, Aligarh, India, in 1986 and 1989, respectively, and the PhD degree from Jamia Millia Islamia (A Central University), New Delhi, India, in 2001 in Mechanical Engineering. He is associated with the Department of Mechanical Engineering, Faculty of Engineering and Technology, Jamia Millia Islamia, New Delhi, India, since 1990. His current research interests include design and process optimisation, application of soft computing techniques, study of the effects of environmental parameters such as noise, vibration, illumination, etc. on humans. He is on the panel of reviewers for Biosystems Engineering. Arshad Noor Siddiquee is an Associate Professor in Production Engineering at Jamia Millia Islamia in New Delhi, India. His major research interest includes welding engineering, machining and optimisation of design and process parameters using the fuzzy modelling. He received his MTech from Indian Institute of Technology (IIT), Delhi (India) and currently pursuing Doctoral Research in the area of Welding from IIT, Delhi.

485

Introduction

A manufacturing system consists of workstations or departments, as well as resources like personnel, material and machinery which must be arranged to form a well-ordered system to maximise benefits. However, it is not an easy task to design best possible layout in order to achieve the desired goal of productivity and profitability, while at the same time ensure safety and satisfaction of workers (Gonzalez-Cruz and Gomez-Senent Matinez, 2010). Ramkumar et al. (2009) discussed the latest scenario in layout design, where manufacturing companies spend a significant amount of resources like time and money for designing and redesigning their facilities. This is because the design of a facility layout has a tremendous effect on the operation of the system and production that it houses. An effective facility layout can actually reduce from 10% to 30% of total operating expenses in manufacturing, annually (Singh and Sharma, 2006). Layout design is basically the arrangement of machines or workstations at production floor to provide smooth movement of resources such as raw materials and workers. An effective layout design is important for good manufacturing of products or delivery of services (Drira et al., 2007). As stated in Hassan (1995), the layout of manufacturing facilities used to be classified as job shop, flow shop and fixed layout. However, the

486

S. Kamaruddin et al.

emergence of group technology (GT) has added a new type of layout classification named as cellular layout. In an assembly-based industry, different types of layouts have been used, such as flow line, job shop and cellular layout. On the other hand, most of the manufacturing companies have faced the problem of inflexibility in their production system that subsequently made them unable to fulfil the customer requirements. Therefore, there is a need to investigate the capability of each layout type to adapt with the changes of headcount and number of models. In the present work, layout types that are being investigated are: flow line, job shop and cellular layout. Flow line is a layout that designs according to the sequences of process that need to be performed in order to produce a product. Usually, in a flow line, all jobs are processed by the same set of machines in linear fashion, from the first to last stage and one machine performs all the processing for each stage (Kurz and Askin, 2003). Therefore, each product type will have its own line and it is usually designed to produce a large number of outputs. The major concern in designing a flow line is to evenly distribute the total work of the line into relating workstations so that the bottleneck can be eliminated to improve the layout efficiency. Apart from that, optimal workers distribution plan is also important for the increased flow-line throughput and managed workload (Neubert and Savino, 2009). The latest researches on flow-line layout show the usage of simulated annealing-based approach (Arumugam et al., 2007; Laha and Chakraborty, 2010) and genetic algorithm (Besbes et al., 2010) in optimising the layout. Job shop design on the other hand groups similar activities, such as processes, functions or sub-assemblies into a shop. According to Tay and Ho (2008), this layout design is suitable when there are a wide variety of products but have low production volume. The flexibility of this layout is highest among the three layouts, where different types of products can be produced. However, the job shop is usually inefficient due to higher work in process (WIP) and backtracking. Cellular layout tries to combine the advantages of flow line and job shop. The cellular layout comes from the application of GT whose main idea is to identify and group machines with same contribution in production process (Mahdavi and Mahadevan, 2008). In designing a cellular layout, different machines or workstations to produce products with similar shape or processing sequences will be grouped into a cell. The worker utilisation of cellular layout can be increased because one worker can be assigned to more than one workstation. Therefore, the worker in a cellular layout should be capable in handling different kinds of machines or tasks. Cellular layout is also recognised as an efficient and effective way to improve productivity in a company by minimising material handling cost and minimising load unbalance in production (Hachicha et al., 2008). Based on these three types of layout, an empirical study is conducted in a radio cassette player production system. The aim of this research is to investigate the effect of model variability and headcount variability towards the flow line, job shop and cellular layout for a radiocassette player manufacturing process and select a feasible layout from it. The layouts are simulated and tested using WITNESS software.

Literature review

Layout design is basically the physical arrangement of elements such as workstations or machines in a manufacturing system. The aim of layout design is to provide smooth flow

The effect of layout design on productivity

487

of workers, materials and information through the system. In order to meet this aim, layout design is considered as an important issue in designing any manufacturing system because it involves large amount of investment and any misjudgement at this stage will lead to losses to the company. El-Baz (2004) has stated the importance of facility layout design towards the productivity by saying an effective facility layout design reduces manufacturing lead time, increases the throughput and hence increases overall productivity and efficiency of the plant. Pugazhendhi et al. (2002) have defined flow line, where all jobs require processing on all machines and jobs have unidirectional flow with identical flow pattern. They also define the difference between pure flow line and general flow line. A pure flow line is characterised by the situation, where all the jobs are required to be processed on all machines; whereas general flow line is a case, where some or all jobs have missing operations on some machines; however, the sequence of machine visits is the same and is unidirectional for all jobs. Job shop on the other hand is characterised by a wide variety of products with variable routings and processing times. It is also called as functional layout with universal equipment, where production can take place as per customers specification in small batches. There are two dynamic circumstances that have to be considered; external dynamics related to rush orders, product mix, volumes demanded, etc., and internal dynamics related to machine breakdowns, production rates, operator absenteeism, quality problems and production yields. Land and Gaalman (1996) and Corsten and Gssinger (2004) have defined job shop as a special organisational form of production, where machines carrying out identical or similar operations are combined into one shop. A job shop is always preferred due to the large number of different product types that it can adapt. Production includes relatively small quantities combined with different working cycles and displays relatively strong fluctuation over a certain period of time. These conditions are particularly prevalent in order-driven production of single or small batches. One of the latest researches on job shop is related to the development and analysis of scheduling decision rules for dynamic flexible job shop production system by using simulation (Vinod and Sridharan, 2009). In the experiment conducted, it was found that some operation performance works efficiently on a primary machines but less efficient on other machine. This is modelled as a percentage increase in the processing time when an operation is performed on an alternate machine. The operations are measured based on mean flow time, standard deviation of flow time, mean tardiness, standard deviation of tardiness and also percentage of tardy jobs. The authors proposed six new scheduling rules for job shop layout and the simulation results reveal that the proposed scheduling rules provide better overall performance for various measures when compared with the existing rules. Many manufacturers have introduced modern manufacturing concepts in their manufacturing systems in order to meet customers requirements which fluctuate with time. One of the concepts that is said to have the capability in improving the productivity, quality and reduction in production cost is cellular layout. According to Drolet et al. (1996), cellular manufacturing (CM) is the result of the application of the so-called GT cell, a concept developed by a Russian named Mitrofanow in the 1940s. On the other hand, Molleman et al. (2002) defined cellular layout as the grouping of workers and machines into relatively independent cells, which are responsible for the complete manufacturing of a set of part types.

488

S. Kamaruddin et al.

On the other hand, Hassan (1995) has reviewed the development of the cellular manufacturing (CM), and furthermore developed a framework in developing the GT layout. In his review, he had mentioned three major steps in developing GT layout, which are formed from part families and machine cells; arrangement of the machines or workstations within each cell and determining the configuration of cells on the facility layout. However, the cellular layout problem is more concerned with the process of determining the best arrangement of machines or workstation within each cell. There are three common ways in workstation arrangements which are single row (flow line), multirows (job shop) and loop layout. The objectives in determining the best arrangement are the minimisation of movement cost and backtracking while maximising the throughput. Furthermore, there are nine steps suggested by Hassan (1995) in his framework for developing the cellular layout. The steps begin with preparing preliminary data, determining the suitability of GT manufacturing, dealing with layout flexibility, identifying significant factors, formation of part families and machines cells, preparing layout data, developing cell layout, developing cell system layout and examining the location of bottleneck machines. Panchalavarapu and Chankong (2005) highlighted the issue of cell formation, which is concerned with identifying part families and machine cells. They noted that cell formation that ignores the integration between part similarities and processing similarities has lead to complex material flow in the system. Therefore, they redefined the cell formation as identifying machine cells, part families and a combination of sub-assemblies so that once the part families are completely processed within a cell they are also assembled within the cell. Subsequently, they proposed a mathematical model to determine assignment of parts, machines and sub-assemblies to manufacturing cells. The mathematical model is based on the similarities of parts, machines and subassembly, which can be computed from partmachine incident matrix and part sub-assembly incident matrix. Furthermore, Panchalavarapu and Chankong (2005) utilised a case study to analyse the production system that they designed by adopting the assembly considerations. The evaluation of the cell design depends on varying the numbers of cell. Comparison between cell design based on GT alone, and with assembly considerations are carried out. Results show that the cell design, which adopted the assembly consideration have higher efficiency and smoother material flow in the system as compared to the cell design based on the GT alone. Cesan and Steudel (2005) highlighted the importance of workers assignment in the CM system, because the productivity of the system is determined by the combination of machine and labour resources. Therefore, the objective is to study labour flexibility in CM systems, particularly in cell implementations allowing intra-cell operators mobility. Thus, the impact on different allocation strategies in the cell will be investigated here. Using a multiple exploratory case study approach, the complexity, actual deployment and potential of labour flexibility are examined in several cells from two manufacturing firms, using labour limited cells with intra-cell mobility. Researchers also focus on the comparative study between different layout types. Li (2003) has proposed to investigate the impact of set-up and processing time variability for job shop under Kanban system. Besides set-up and processing time variability, there are three other factors that are considered in the research, they are shop layout, production flow patterns and the amount of set-up time reduction achievable. Therefore, a functional layout, a cellular layout with unidirectional flow and a cellular layout with backtracking flow are allowed to build to compare the layout performances under set-up

The effect of layout design on productivity

489

and processing time variability. The performance measures in this research are average WIP, average flow time and average set-up to processing time ratio. Besides the functional layout, there are four levels of evaluation on the cellular layout, which involves cellular layout with backtracking allowed, cellular layout with unidirectional flow, one-piece flow and batch flow. The results show that the functional layout is superior to a cellular layout if the set-up time variability is high. However, as the set-up time decreases to a medium or lower level, the performance of a cellular layout (with batch intra-cell flow) becomes comparable with the functional layout. In addition, with a medium set-up time reduction, set-up and processing time does not have a significant effect on the relative shop configurations, since both functional and cellular layout performs similarly. With a large set-up time reduction, a cellular layout is more likely to outperform a functional layout. Only low processing time variability with medium to low set-up time variability renders a cellular layout with one-piece and unidirectional intra-cell flow, which is superior to that with batch intra-cell flow. Furthermore, for a cellular layout with batch intra-cell flow, adopting unidirectional part flows or allowing backtracks does not lead to any significant performance difference for all factor levels investigated. Research on the impact of layout design towards the labour productivity was carried out by Aase et al. (2004). In the research, they have made the comparative study between straight-line assembly line and U-shape assembly line systems in affecting the labour productivity. Result shows that labour productivity has improved significantly under certain conditions when switching from a straight-line assembly line to a U-shape assembly line. This is due to the workers in the U-shape assembly line, who can perform multitasking by moving accordingly in the U-shape configuration, and subsequently increased the percentage of manpower utilisation. However, the results also show that there are cases where the changes of straight-line assembly line to U-shape assembly did not result in improving the labour productivity. This phenomenon shows that precaution steps are necessary before changing a straight-line to a U-shape assembly line for the purpose of labour productivity improvement. Huertas et al. (2007) conducted a comparison between U-shape and flow-through layout in a large capacity warehouse. They estimated and evaluated the operational costs and average picking time of alternative layout for the warehouse, which is a distribution centre with large variety of products. In this research, an analytical model was developed and implemented to measure the average distances between centroids of fixed positions for commodities. The model was used to evaluate two new alternatives of layout and operations of the warehouse. It was found that the option with the layout with docks on long opposite sides of the warehouse, and the operation without a separate picking zone minimises overall operational costs. On the other hand, Farrington and Nazemetz (1999) examined the effect of system configuration, underlying system structure, demand variation and operation time variability on overall system performances. Three types of layouts are selected to be investigated; they are: a cellular layout, a dedicated job shop and a pure job shop. For the demand variation, it involved two levels of demand pattern variability that are low demand pattern variability and high demand variability. Similarly, there are two levels of processing time variability that is chosen in this work. The low processing time variability had an average coefficient of variation of 0.3, while the high processing time variability had an average coefficient of variation of 0.6. The simulation models of each layout type were firstly constructed in SLAM II, and each of the experiments will be

490

S. Kamaruddin et al.

replicated 16 times to gain a sufficient accuracy results. The comparative study between the layouts are focused on eight performance measures, which are average distance moved per order, average job lateness, average time in system, average work in process, average machine utilisation, average number of departmental interactions, mean time to reappearance of part families at machines and average number of open orders in the system. The results show that the cellular layout is always superior to the job shop with respect to distance moved per order, number of departmental interactions and mean time to reappearance of part families at machines; the only exceptions to this being the two cases when the dedicated job shop had equivalent or slightly superior results for the mean time to reappearance of part families at the machines. Therefore, it can be concluded that the cellular layout is preferred as compared to job shop in the light of most of the performance measures. This present work is similar with Farrington and Nazemetz (1999), but the comparative study here is between flow line, job shop and cellular layout under different levels of headcount variability and model variability, which was not discussed in their work. Therefore, this study is conducted to investigate the overall productivity in different layout types under different models and headcount variability.

Methodology

To achieve the objective stated, there are three basic layout types that need to be developed. The layout development starts with hand modelling process, where a conceptual layout will be developed. In order to have a clearer view on how the material flow in the layout, precedence table and precedence diagram, a conceptual layout model was constructed. In the precedence table, each task and standard time involved in producing the product will be listed down. Besides, the table also include the sequence of each task. In addition, a few assumptions have been made to simplify the layout design problems as shown below. 1 2 3 4 5 the distance between workstations is neglected the travel time between workstations is assumed to be constant all the products in the layout are assumed to be in good quality condition with zero defects no buffer exists between two adjacent workstations all the workers in the layout are assumed to be skilled workers.

3.1 Design of layout


Line balancing is a major concern in designing the flow-line configuration. Therefore, a theoretical cycle time and theoretical minimum number of workstations are calculated initially. The formula for theoretical cycle time and the theoretical minimum number of workstation are:

Theoretical cycle time

total time available in a month desired output

(1)

The effect of layout design on productivity

491

Theoretical minimum number of workstations

total processing time cycle time

(2)

After determining the theoretical cycle time and theoretical minimum number of workstations, the task is assigned one at a time to the workstations until the total station time is less than or equal to the cycle time achieved. In the mean time, no other tasks are carried out due to time and sequence restrictions. However, this may not be applicable in the real world and may cause station time far lower than the theoretical cycle time, subsequently causing the workload to be unbalanced in the layout. Therefore, the station time is allowed to exceed the theoretical cycle time, and parallel workstation can be used to reduce the layout bottleneck. In flow line, each worker will only be assigned to one workstation. In addition, the material flow in the flow line is designed as one-piece flow and the travel time between workstations are assumed to be 3 sec. On the other hand, the design of job shop starts with grouping the processing tasks into different shops according to the process, function or sub-assemblies. After determining the number of shop, the bottleneck of the layout is identified. In order to reduce the bottleneck, parallel workstations can be used and subsequently increase the layout efficiency. Here, the material flow is designed to be moved in batches of 5, and the travel time between shops is assumed to be 10 sec. For the design of cellular layout, it starts with group processing tasks into cell. In this study, the cell formation is based on the concept of single row that is similar with flow line, which in turn simplifies the intracellular flow. This concept is chosen because it is suitable for assembly processes. In cellular layout design, one worker can be assigned to more than one workstation. Therefore, the U-shape arrangement within cell is chosen, because it provides easier path for movement between workstations. Subsequently, the worker utilisation for cellular layout can also be increased. Moreover, the issue of balanced workload also need to be taken into consideration while designing the cellular layout to reduce the bottleneck of the layout. The example of conceptual layout for flow line, job shop and cellular layout with 12 workers are shown in Figures 13, respectively.
Figure 1 Conceptual layout for flow line (see online version for colours)

492
Figure 2

S. Kamaruddin et al.
Conceptual layout for job shop (see online version for colours)

Figure 3

Conceptual model of cellular layout (see online version for colours)

3.2 Simulation modelling


After the conceptual layouts are designed, it will be converted into simulation layout model by using WITNESS simulation package. Simulation is an integral part of any effective layout planning and design in order to produce more realistic layout (Aleisa and Lin, 2005). Basically, the simulation modelling process involves coding and linking. Coding involves creating the required elements by defining, displaying and detailing the icon, while linking includes linking each element with rules or actions. Each of the simulation layout models is evaluated by the verification and validation process. In verification, the detail of each element, the link between each elements and any programme involved in the layout is inspected to make sure that it resembles the conceptual model that is being planned during the hand modelling stage. On the other hand, a validation of simulation model is to determine whether the elements in the simulation model represent the corresponding real world with sufficient accuracy. The approach used in this work is to compare the model results with the calculation results from the simulation run.

The effect of layout design on productivity

493

3.3 Design of experiment


As stated earlier, the objective of this study is to investigate the effect of model variability and headcount variability of different layout configurations on its performance measures. Therefore, the layout performance measures are gained by running the simulation models based on two experiments. The first experiment is concerned with the headcount variability, where initial layout with 12 workers is reduced to 11 workers and to 10 workers. Besides, each layout type with each level of headcount variability will also be tested on three levels of model variability, which are low, medium and high variety. The purpose of this experiment is to investigate the changes in layout performance when the number of workers is reduced to 11 workers and to 10 workers. The second experiment is concerned with the model variability. In this experiment, the number of workers in each layout is set at 11 workers and all three types of layout are tested on three levels of model variability, which are low, medium and high variety. In addition, each level of model variety is divided into three levels of treatments. For example, in the low variety, it involved 57 models of changes. Therefore, this experiment will investigate the effect of different levels of model changes to the layout performances. In running both experiments, there are also few assumptions being made which are as follows:

replication of simulation runs in each experiment is done five times the number of workers in all three types of layouts in each comparative study is same the desired volume is set at 9,000 units the due dates is set at 17 days processing time for each model types is assumed to be constant, but the set-up time for each model is different the simulation layouts performed at starting condition where no WIP is in the layout.

3.4 Performance measures


Several dependent variables that can be observed from the simulation runs are selected so that a comparison between layouts can be made. The performance measures are chosen according to its indication, relatively to the layout productivity. The selected performance measures are listed as follows:
Throughput time: the throughput time is defined as the time needed to produce one unit of product. Therefore, a lower value of throughput time will indicate better layout productivity. The equation used to calculate the throughput time is shown in Equation (3).

Throughput time

flow time(minutes) output(units)

(3)

494

S. Kamaruddin et al. Lateness: lateness is defined as the difference between the job completion time and the due date time. If the answer is positive, it is considered as lateness and otherwise it is the earliness. Lateness is measured to test the capability of the layout to adapt with the desired production volume. The equation used in calculating the lateness is shown in Equation (4).

Lateness due date time (min) completion time (min)


(4)

Labour productivity: labour productivity is a measurement on the utilisation of workers in certain layout. It is defined as the number of unit products being produced by a worker per hour in this case. Therefore, labour productivity shows how productive a worker is in a layout. The equation in calculating the labour productivity is shown in Equation (5).

Labour productivity

[desired output (units) 60(min)] [no. of worker flow time (min)]

(5)

Results and discussion

After the simulation runs of five replications, the data of three performance measures are obtained. The mean value of each performance measures are tabulated into Table 2, which illustrates the simulation results from Experiment 1, while Table 3 illustrates the simulation results from Experiment 2. In addition, the simulation results also have been taken as an input to the F-test to determine whether the experimental factor has an effect on the layout performances. For those cases where the F-value is slightly over the critical F-value, NewmanKeuls test is carried out to determine which pair of comparison treatments caused the significant impact. The F-test in both the experiments shows that the effect of headcount variability and model variability do cause a significant impact on the layout performance except for the job shop at low variety in the second experiment. This phenomenon shows that the job shop is more capable in handling the flexibility and is not easily getting affected by the changes. In Experiment 2, there are some cases where the F-value is only slightly greater than the critical value at 99% level of confidence. For that reason, the NewmanKeuls test has been carried for those cases. The results of NewmanKeuls have been shown in Table 1.
Table 1 Flow line Job shop Cellular layout NewmanKeuls test results in Experiment 2 Low variety Medium variety High variety Low variety Medium variety High variety 57 models 1014 models 1519 models 57 models 1519 models 1517 models 1519 models

The effect of layout design on productivity


Table 2 Experiment results for Experiment 1
Throughput time(min/unit) 1.1761 1.7015 1.1617 1.1839 1.7038 1.1662 1.1883 1.7071 1.1768 1.0623 1.5176 1.0271 1.0705 1.5184 1.0298 1.0761 1.5208 1.0319 1.0165 1.1484 0.9500 1.0226

495

Exp 1

H/C variability 10

Model variability L

Layout FL JS CL

Lateness(min) 1405.3416 6133.3606 1275.2867 1475.5037 6154.2449 1315.4760 1515.2230 6183.4891 1410.8838 380.7430 4490.2842 64.0975 454.6200 4498.0109 88.2189 505.2550 4507.5438 106.7193 -31.7983 1156.0749 -630.1992 23.3803

Labour productivity (unit/worker/hour) 5.1014 3.5264 5.1649 5.0667 3.5215 5.1451 5.0490 3.5148 5.0987 5.1346 3.5911 5.3105 5.0953 3.5890 5.2967 5.0686 3.5865 5.2861 4.9190 4.3537 5.2633 4.8895

FL JS CL

FL JS CL

11

FL JS CL

FL JS CL

FL JS CL

12

FL JS CL

FL

Note: FL flow line; JS job shop; CL cellular layout; L low; M medium; H high.

From Table 3, it is observed that there are only six cases, where the NewmanKeuls test is needed. For example, in the flow line at low variety, the layout performances are only affected when the number of models change from 5 models to 7 models. In another words, the changes of models from 5 models to 6 models would not cause any significant impact on the layout performances.

496
Table 3 Exp 2

S. Kamaruddin et al.
Experiment results for Experiment 2 No. of models 5 models Layout FL JS CL 6 models FL JS CL 7 models FL JS CL M 10 models FL JS CL 13 models FL JS CL 14 models FL JS CL H 15 models FL JS CL 17 models FL JS CL 19 models FL JS CL Throughput time(min/unit) 1.0625 1.5186 1.0271 1.0635 1.5188 1.0276 1.0646 1.5192 1.0284 1.0705 1.5197 1.0298 1.0732 1.5201 1.0310 1.0749 1.5207 1.0312 1.0761 1.5208 1.0319 1.0801 1.5215 1.0337 1.0825 1.5222 1.0343 Lateness(min) 382.6556 4487.1889 64.0975 391.4959 4489.4757 68.5029 401.8221 4493.0432 75.6824 454.6201 4494.1219 88.2189 478.4711 4500.9895 99.1049 494.3404 4506.3867 101.0076 505.2547 4507.5438 106.7193 541.1605 4513.8053 125.7047 562.8033 4515.8453 128.6440 Labour productivity (unit/worker/hour) 5.1336 3.5919 5.3105 5.1289 3.5913 5.3080 5.1233 3.5899 5.3039 5.0953 3.5893 5.2967 5.0827 3.5883 5.2905 5.0744 3.5868 5.2894 5.0686 3.5865 5.2861 5.0499 3.5849 5.2765 5.0387 3.5844 5.2737

Model variability L

Note: FL flow line; JS job shop; CL cellular layout; L low; M medium; H high.

From both experiments, the observation shows that at any layout type the throughput time and lateness is increased when the numbers of models are increased. This is because when more changes of model involved in a production system, more time has been spent at the set-up which causes the time needed to produce a product increased. Subsequently, the lateness will also be increased due to the value of lateness, which is proportional with the throughput time. On the other hand, the productivity decreased when the numbers of model changes are increased. This is caused by the frequent changes of orders that results in spending more time on the non-productive work for the worker. Furthermore, the throughput time and lateness have been increased when the numbers of workers are decreased. The reason for this phenomenon is the decrement of workers in the layout has

The effect of layout design on productivity

497

subsequently reduced the total work force in the layout. Therefore, the workload for each worker has been increased, and the time needed to produce each product is more. However, there is a different trend of labour productivity when the number of workers is decreased in the layout. For the job shop, the decrement of workers has decreased the labour productivity. This may be due to the unbalanced scenario in the layout, where lesser number of workers in the layout will cause higher bottlenecks with blockage. Thus, some of the workers may stay in the non-productive status or idle status while waiting for jobs being processed at the successor station. For the flow line and cellular layout, the labour productivity has been decreased from 12 workers to 11 workers and increased from 11 workers to 10 workers. This phenomenon shows that the flow line and cellular layout are better balanced design at 12 workers and 10 workers as compared with the 11 workers layout. From the F-test, it shows that there is significant effect on the selected performance measures by changing the experimental factor which is the headcount variability and model variability. Therefore, it can be concluded that we can measure the performance of each layout by considering the selected performance measures. Considering the results from Tables 2 and 3, the cellular layout was always superior to the flow line and job shop based on the three selected performance measures. Here, the comparative study will be discussed on each performance measure. Cellular layout has the lowest throughput time at all levels of headcount variability and model variability. The trend is followed by flow line and job shop configuration. Essentially, the throughput time in each layout is affected by the even distribution of workload, travelling time between workstations and types of material flow. In the cellular layout, the well-balanced workload for worker that assigns one worker to more than one workstations to increase the worker utilisation; the shorter travelling time between workstations due to the workstations are closely arranged, and material moves in batch has caused the time needed to produce one product to be shorter in the cellular layout. In contrast, job shop has the highest throughput time, and this may be due to the unbalanced scenario in the layout that caused the longer awaiting time at the workstation when blockage occurred. As mentioned earlier, lateness is proportional with the throughput time. This is because the lateness in completing the available orders are affected by the time needed in producing a product, which is the throughput time. Therefore, when the throughput time is high, the lateness will also be high. As a result, the cellular layout has the lowest lateness value among the three layout types. Besides, the cellular layout with 12 workers also shows negative value at all levels of model variability. On the other hand, the flow line with 12 workers is only capable in finishing the orders at low variety in time, where the job shop is unable to finish the orders at any level of headcount variability and model variability. These phenomena show that cellular layout design in this study is capable in handling different levels of model variability with small quantities, while flow shop is capable in producing high volume with low model changes. Although job shop is said to be capable in handling the flexibility, but in this case where no variable routings are involved, the advantages of job shop has been concealed and only the inefficient job shop appeared. With respect to the labour productivity, cellular layout has once again achieved the highest value followed by the flow line and job shop. This is due to the U-shape of workstations arrangement that has facilitated the workers to move between workstations and allowed the worker to be assigned to more than one workstation. As a result, the

498

S. Kamaruddin et al.

workload for each worker in the cellular layout is well balanced and increases the labour productivity. After comparing the three performance measures for the three types of layout configurations under different levels of headcount variability and model variability, the cellular layout with 12 workers is selected as the feasible layout that is capable to finish the available orders on time. Besides, cellular layout with 11 workers can be taken as another alternative in selecting the feasible layout in producing 9,000 units of products. This is because the cellular layout with 11 workers is only delayed by 2 hr in completing the jobs at high variety, and this situation can be easily overcome by adding two hours of overtime. However, the cellular layout with 11 workers has met its limit in producing the total volume of 9,000 units and may not be capable in handling a higher product demand, while the cellular layout with 12 workers still has space for increment in product demand. Therefore, it can be concluded that cellular layout is well performing as compared with flow line and job shop. In addition, cellular layout with 12 workers and 11 workers are two alternatives that can be chosen in order to run a production with different model variability. In comparison to the theoretical contributions of the empirical study, the research has provided new insights from the managerial perspective. It has been identified that by understanding the emerging requirements in relation to the demand, variation of model and workers has enabled the management to deal with the challenges involving the effect of layout design. The nature of broader view is specifically preferred in relation to the kind of challenges faced by a company operating in a low volume and high-mix environment. As the findings exemplify, it is not enough for the management to concentrate on developing and improving the layout without prior understanding of the impact towards the overall productivity. Moreover, by understanding and incorporating various factors that have been highlighted in the findings, it can provide significance value in unravelling the advantages of adopting various layouts in assembling various products. This is particularly crucial in the current scenario, where the nature of the global markets is more focus due to homogenisation of supply and demand.

Conclusion

The purpose of this work is to determine the effect of two experimental factors (headcount variability and model variability) towards different types of layout design. Eventually, there are three levels of each experiment which consists of low, medium and high parameters. Besides, three types of performance measures, which are the simulation results collected from the simulation runs have been chosen to provide a basis for comparing the layout performance. These performance measures are chosen, because it can represent the productivity of the layout indirectly. The performance measures are throughput time, lateness and labour productivity. In order to test the experimental factor on the layout model designed initially, two experiments have been planned. Experiment 1 intended to investigate the effect of headcount variability, while Experiment 2 was set to investigate the model variability at constant headcount. From the F-test, it can be concluded that the effect of headcount variability and model variability do have an impact towards the layout performance, except for the job shop with 11 people at low variety. Finally, the feasible layout design for radiocassette player

The effect of layout design on productivity

499

production system will be chosen among the three layout designs based on the performance measures. Results show that the cellular layout with 12 people is capable in handling high-volume and high-model variability as required in the cassetteradio player production system and cellular layout with 11 workers is another alternative in this empirical study, where it can achieve the desired demand by adding two hours of overtime. Some of the effects of dependent variables in this research are not clear due to the levels of treatments that are considered to be too narrow. Therefore, the future work can test the simulation layouts with levels of treatments that have larger difference. For example, in Experiment 2, the levels of treatments for model variability can be set as 4 mix models, 8 mix models and 12 mix models in the low variety level. By doing this, the effect of different mix models at a model variability level can be observed clearly. Besides, the job shop simulation layouts can also be tested at products in different routings to find out its level of flexibility in handling this kind of situation.

Acknowledgements
The authors would like to acknowledge the reviewers of this paper for their suggestions and recommendations in improving the manuscript and also the research grant provided by Universiti Sains Malaysia and Ministry of Science and Technology (MOSTI) for funding this study.

References
Aase, G.R., Olson, J.R. and Schniederjans, M.J. (2004) U-shaped assembly line layouts and their impact on labor productivity: an experimental study, European Journal of Operational Research, Vol. 156, pp.698711. Aleisa, E.E. and Lin, L. (2005) For effective facilities planning: layout optimization then simulation, or vice versa? Proceedings of the 2005 Winter Simulation Conferences, pp.13811385. Arumugam, C., Raja, K. and Selladurai, V. (2007) Agility in two-stage hybrid flow shop parallel machine scheduling through simulated annealing, International Journal of Services and Operations Management, Vol. 3, No. 3, pp.332354. Besbes, W., Teghem, J. and Loukil, T. (2010) Scheduling hybrid flow shop problem with non-fixed availability constraints, European Journal of Industrial Engineering, Vol. 4, No. 4, pp.413433. Cesan, V.I. and Steudel, H.J. (2005) A study of labor assignment flexibility in cellular manufacturing systems, Computers and Industrial Engineering, Vol. 48, No. 3, pp.571591. Corsten, H. and Gssinger, R. (2004) Opportunistic coordination of operations in job shop production, Technovation, Vol. 24, No. 3, pp.219234. Drira, A., Pierreval, H. and Hajri-Gabouj, S. (2007) Facility layout problems: a survey, Annual Reviews in Control, Vol. 13, pp.255267. Drolet, J., Abdulnour, G. and Rheault, M. (1996) The cellular manufacturing evolution, Computers and Industrial Engineering, Vol. 31, Nos. 12, pp.139142. El-Baz, M.A. (2004) A genetic algorithm for facility layout problems of different manufacturing environments, Computers and Industrial Engineering, Vol. 47, Nos. 23, pp.233246. Farrington, P.A. and Nazemetz, J.W. (1998) Evaluation of the performance domain of cellular and functional layouts, Computers and Industrial Engineering, Vol. 34, No. 1, pp.91101.

500

S. Kamaruddin et al.

Gonzalez-Cruz, M.C. and Gomez-Senent Matinez, E. (2010) An entropy-based algorithm to solve the facility layout design problem, Robotics and Computer-Integrated Manufacturing, doi:10.1016/j.rcim.2010.06.015. Hachicha, W., Masmoudi, F. and Haddar, M. (2008) Combining axiotic design and designed experiments for cellular manufacturing systems design framework, International Journal of Agile Systems and Management, Vol. 3, Nos. 3/4, pp.306319. Hassan, M.M.D. (1995) Layout design in group technology, International Journal of Production Economics, Vol. 38, Nos. 23, pp.173188. Huertas, J.I., Ramirez, J.D. and Salazar, F.T. (2007) Layout evaluation of large capacity warehouse, Facilities, Vol. 25, Nos. 7/8, pp.259270. Kurz, M.E. and Askin, R.G. (2003) Comparing scheduling rules for flexible flow lines, International Journal of Production Economics, Vol. 85, pp.371388. Land, M. and Gaalman, G. (1996) Workload control concepts in job shops A critical assessment, International Journal of Production Economic, Vols. 4647, pp.535548. Laha, M. and Chakraborty, U.K. (2010) Minimising total flow time in flow shop scheduling using a simulated annealing-based approach, International Journal of Automation and Control, Vol. 4, No. 4, pp.359379. Li, J.W. (2003) Improving the performance of job shop manufacturing with demand-pull production control by reducing set-up/processing time variability, International Journal of Production Economic, Vol. 84, No. 3, pp.255270. Mahdavi, I. and Mahadevan, B. (2008) CLASS: an algorithm for cellular manufacturing system and layout design using sequence data, Robotics and Computer-Integrated Manufacturing, Vol. 24, pp.488497. Molleman, E., Slomp, J. and Rofeles, S. (2002) The evolution of a cellular manufacturing system a longitudinal case study, Int. J. Production Economics, Vol. 75, pp.305322. Neubert, G. and Savino, M.M. (2009) Flow shop operator scheduling trough constraint satisfaction and constraint optimisation techniques, International Journal of Productivity and Quality Management, Vol. 4, Nos. 5/6, pp.549568. Panchalavarapu, P.R. and Chankong, V. (2005) Design of cellular manufacturing systems with assembly considerations, Computers and Industrial Engineering, Vol. 48, No. 3, pp.449469. Pugazhendhi, S., Thiagarajan, S., Rajendran, C. and Anantharaman, N. (2002) Performance enhancement by using non-permutation schedules in flowline-based manufacturing systems, Computers and Industrial Engineering, Vol. 44, No. 1, pp.133157. Ramkumar, A.S., Ponnambalam, S.G. and Jawahar, N. (2009) A new iterated fact local research heuristic for solving QAP formulation in facility layout design, Robotics and ComputerIntegrated Manufacturing, Vol. 25, pp.620629. Singh, S.P. and Sharma, R.R.K. (2006) A review of different approaches to the facility layout problems, International Journal of Advanced Manufacturing Technology, Vol. 30, pp.25433. Tay, J.C. and Ho, N.B. (2008) Evolving dispatching rules using genetic programming for solving multi-objective flexible job-shop problems, Computer and Industrial Engineering, Vol. 54, pp.453473. Vinod, V. and Sridharan, R. (2009) Development and analysis of scheduling decision rules for a dynamic flexible job shop production system: a simulation study, International Journal of Business Performance Measurement, Vol. 11, Nos. 1/2, pp.4371.

You might also like