Improving Manufacturing Processes Using Simulation Methods

Sławomir Kłos, Justyna Patalas-Maliszewska, Peter Trebuna

Journal of Automation and Control

Improving Manufacturing Processes Using Simulation Methods

Sławomir Kłos1, Justyna Patalas-Maliszewska1, Peter Trebuna2,

1University of Zielona Góra, Faculty of Mechanical Engineering, Licealna 9, 65-417 Zielona Góra, Poland

2Technical University of Kosice, Faculty of Mechanical Engineering

Abstract

Computer simulation is a very important method for studying the efficiency of manufacturing systems. This paper presents the results of simulation research about how buffer space allocated in a flow line and operation times influence the throughput of a manufacturing system. The production line in the study consists of four stages and is based on a real machining manufacturing system of a small production enterprise. Using Tecnomatix Plant Simulation software, a simulation model of the system was created and set of experiments was planned. Simulation experiments were prepared for different capacities of intermediate buffers located between manufacturing resources (CNC machines) and operation times as input parameters, and the throughput per hour and average life span of products as the output parameter. On the basis of the experiments, the impact of the allocation of intermediate buffer capacities on production efficiency is analysed.

Cite this article:

  • Sławomir Kłos, Justyna Patalas-Maliszewska, Peter Trebuna. Improving Manufacturing Processes Using Simulation Methods. Journal of Automation and Control. Vol. 3, No. 3, 2015, pp 83-88. http://pubs.sciepub.com/automation/3/3/9
  • Kłos, Sławomir, Justyna Patalas-Maliszewska, and Peter Trebuna. "Improving Manufacturing Processes Using Simulation Methods." Journal of Automation and Control 3.3 (2015): 83-88.
  • Kłos, S. , Patalas-Maliszewska, J. , & Trebuna, P. (2015). Improving Manufacturing Processes Using Simulation Methods. Journal of Automation and Control, 3(3), 83-88.
  • Kłos, Sławomir, Justyna Patalas-Maliszewska, and Peter Trebuna. "Improving Manufacturing Processes Using Simulation Methods." Journal of Automation and Control 3, no. 3 (2015): 83-88.

Import into BibTeX Import into EndNote Import into RefMan Import into RefWorks

At a glance: Figures

1. Introduction

Computer simulation is a powerful method for designing and analysing manufacturing processes in industry. Today, within a relatively short period of time, it is possible to design or redesign a manufacturing system to implement new production processes using simulation software. On the basis of the layout of a plant and a description of the technology in use, manufacturing resources can be allocated and the performance of the system can be estimated. A simulation model enables us to analyze various alternatives of the manufacturing system configuration and gauge the influence of the different parameters on the efficiency of different manufacturing processes. Advanced tools of computer simulation enable the analysis of the general performance of manufacturing systems, scheduling methods, support of facility layouts, automated material handling, automated guided vehicle systems design, etc. By doing this, we can reduce manufacturing costs and improve the system’s throughput. Typically, however, the building of a simulation model of a manufacturing system is time-consuming and only enables the solution of individual problems. Therefore, it is often the case that a generalization of the results of simulation research is difficult. A very important advantage of a computer simulation is the possibility to create a model of a manufacturing system whose behaviour could be very close to the real system. Modelling the processes of a manufacturing system and production processes is time-consuming and needs expert knowledge which increases the costs of the simulation method. To reduce the modelling time, often some simplifications of a manufacturing system are introduced which do not significantly affect the behaviour of the system. Especially for companies which implement multi-assortment, repetitive production; it is very important to design and organize the production system effectively. Each new project requires the design of a technological process. To analyze the efficiency of a new production process, a model should be created that includes operation and setup times, the number of workers, the capacity of intermediate storage buffers, the availability of machines and tools, etc.

1.1. Literature overview

Simulation has been successfully implemented in a lot of research related to manufacturing system design and operation. Computer simulation offers very effective tools for visualizing, understating, and analyzing the dynamics of manufacturing systems [1]. Due to its complexity and importance, the buffer allocation problem has been studied widely and numerous publications are available in the literature [2]. Stanley and Kim presented results of simulation experiments made for buffer allocations in closed serial-production lines [3]. For the line, a single buffer space is the room and the associated material handling equipment that is needed to store a single job that is a work-in-process, and buffer allocation is the specific placement of a limited number of buffer spaces in a production line. The authors demonstrated a buffer allocation decomposition result for closed production lines, and also provided evidence that optimal buffer allocations in closed lines are less sensitive to bottleneck severity than in open production lines. The placement of buffers in a production line is an old and well-studied problem in industrial engineering research. Vergara and Kim proposed a new buffer placement method for serial production lines [4]. The method is very efficient and uses information generated in a production line simulation whose conceptual representation of job flow and workstation interaction can be described with a network which aims to place buffers in order to maximize throughput. They compared the results of the new method against a method for buffer placement based on a genetic algorithm. Yamashita and Altiok [5] proposed an algorithm for minimizing the total buffer allocation for a desired throughput in production lines with phase-type processing times. They implemented a dynamic programming algorithm that uses a decomposition method to approximate the system throughput at every stage. Gurkan used a simulation‐based optimization method to find optimal buffer allocations in tandem production lines where machines are subject to random breakdowns and repairs, and the product is fluid‐type [6]. He explored some of the functional properties of the throughput of such systems and derived recursive expressions to compute one-sided directional derivatives of throughput, from a single simulation run. Shi and Gershwin presented an effective algorithm for maximizing profits through buffer size optimization for production lines [7]. They considered both buffer space cost and average inventory cost with distinct cost coefficients for different buffers. To solve the problem, a corresponding unconstrained problem was introduced and a nonlinear programming approach was adopted. Qudeiri et al. used a genetic algorithm for studying the design of serial - parallel production line [8]. They tried to find the nearest optimal design of a serial parallel production line that maximized production efficiency by optimizing buffer size between each pair of work stations, machine numbers in each of the work stations and machine types. Nahas et al. formulated a new optimal design problem of a parallel production line, where parallel machines and in-process buffers are included to achieve a greater production rate [9]. The main objective was to maximize the production rate subject to a total cost constraint. Nourelfath et al. formulated a new problem of the optimal design of a series production line system, and developed an efficient heuristic approach to solve it [10]. The problem was solved by developing and demonstrating a problem‐specific system algorithm. Fernandes and Carmo-Silva [11] presented a simulation study of the role of sequence-dependent set-up times in decision making at the order release level of a workload controlled make-to-order flow-shop. They indicated that the local strategy, which has been traditionally adopted in practice and in most of the studies dealing with sequence-dependent set-up times, does not always give the best results. Matta [12] presented mathematical programming representations for the simulation-based optimization of buffer allocation in flow lines.

1.2. Problem specification

In this paper, a simulation method is used to analyse the impact of intermediate buffer capacities and lot sizes on the throughput of a production line. The buffer allocation problem is an NP-hard combinatorial optimization problem which deals with finding optimal buffer sizes to be allocated into buffer areas in a production line [13, 14]. In general, the buffer allocation problem is classified into three categories according to its objective function [13]:

1. Maximize the throughput rate of the production line with fixed amount of buffer sizes.

2. Minimize the total buffer size to achieve the desired throughput rate of the production line.

3. Minimize the average amount of work-in-process in the production line.

The presented paper takes into account the combination of the first two problems. The main problem considered in the paper can be formulated as follows: Given a production line with a determined number of manufacturing resources, operation times and set-up times. How does an intermediate buffer capacity and sequence of operation times affect the throughput of the production line? Using a simulation method, the best relation between allocated buffer capacity and throughput for several variants of operation times is examined. On the basis of the proposed simulation experiments, the principles by which solutions are reached can be analyzed and evaluated. In the next chapter, the simulation model of the production system is described and assumptions for simulation experiments are formulated. The third chapter contains the results of the simulation research and an analysis of the behaviour of the manufacturing system. In the last chapter, conclusions and directions for further study are presented.

2. Simulation model of the manufacturing system

The model of the automated production line was prepared on the basis of a real example of a manufacturing system dedicated to metal tooling in an automotive company. The model and simulation experiments are implemented using Tecnomatix PLM simulation software. The studied manufacturing system includes four technological operations: cutting, turning, milling and grinding. The manufacturing process is divided into four stages by technology and between each two stages an intermediate buffer is allocated. Every stage of the manufacturing system encompasses a determined number of manufacturing resources (three CNC machines in every stage). The simulation model of the manufacturing system was prepared with Tecnomatic Plant Simulation Software version 11.0.0 and is presented in Figure 1.

Figure 1. Simulation model of the manufacturing system

The production is divided into four batch sizes of products (A,B,C,D) and realized cyclically. It was assumed that the efficiency of manufacturing resources is approximately 95%. The model of the production system was prepared on the basis of a real manufacturing company. The operation times are based on a lognormal distribution. A lognormal distribution is a continuous distribution in which a random number has a natural logarithm that corresponds to a normal distribution. The realizations are non-negative real numbers. The density of the lognormal distribution Lognormal(σ,μ) is calculated as follows [15]:

(1)

where σ and μ are respectively mean and standard deviations and are defined as follows:

(2)
(3)

The maximum of the density function is defined as:

(4)

The example of the density function of lognormal distribution Lognormal(3,2) is presented in the Figure 2 [15].

Figure 2. Density function of lognormal distribution – lognormal(3,2)

The variants of operation times are presented in the Table 1.

Table 1. The matrix of variants of operation times

The set-up times are defined in a set-up matrix (see Table 2). The numbers presented in the set-up matrix refer to the set-up time of changing the production batch (for example a batch change from product A to product B takes 15 minutes of set-up time).

The sequence of lot size is presented in the Table 3.

The main variable in the simulation experiments was the intermediate buffer capacity. The proposed values of different combinations of intermediate buffers capacities that define the simulation experiments are presented in Table 4. The combination of the buffer capacities was chosen arbitrarily on the basis of author’s experiences.

Table 4. The matrix of intermediate buffer capacities

In addition to the system throughput per hour, the average life span of products in the manufacturing system was analyzed. The average life span shows us how long the products stay in the system and enables us to identify the level of work-in-process. In the next chapter, the results of the simulation experiments are presented.

3. The outcomes of computer simulation experiments

4. Ease of Use

On the basis of the described simulation model, a set of simulation experiments for four variants of operation times, as presented in Table 1, were performed. The results of the experiments performed for Variant 1 of operation times are presented in Figure 3 a and Figure 3 b.

Figure 3. a) The throughput per hour for Variant 1, b) The average life span for Variant 1

The minimum values of throughput per hour in the manufacturing system were obtained in experiments 1 and 11 (lower then 14 pieces per hour). Interestingly, the minimum average life span in the system were obtained in experiments 1 and 10. In experiment 1, the total sum of the intermediate buffer capacity is smallest and equal to 4 (single capacity of the buffers). In experiments 10 and 11, the buffer capacities increase or decrease from 1 to 4, respectively. Very good results of throughput were obtained in experiment 4, and the last four experiments in which the buffer capacities were significantly increased.

The results of the experiments performed for Variant 2 of operation times are presented in Figure 4 a and Figure 4 b.

Figure 4. a) The throughput per hour for Variant 2, b) The average life span for Variant 2

The minimum value of throughput per hour in the manufacturing system was obtained in experiments 1, 7 and 11 (respectively 14.0; 15.6 and 15.5 products per hour). The minimum average life span in the system was obtained in experiments 1 and 11. Very good results of system throughput by relatively small capacities of intermediate buffers were obtained in experiments 4, 6 and 9.

Figure 5. a) The throughput per hour for Variant 3, b) The average life span for Variant 3
Figure 6. a) The throughput per hour for Variant 4, b) The average life span for Variant 4

The results of the experiments performed for Variant 3 of operation times are presented in Figure 5 a and b. The minimum values of throughput per hour were obtained in experiments 1 and 11; and the lowest level of average life span was obtained in experiments 1, 2, 10 and 11. Very high values of throughput for relatively small buffer capacities were obtained in experiments 4 and 8. Similar to the last case, the average life span increased very significantly for the last three experiments when the total buffer capacity was increased.

The results of the experiments performed for Variant 4 of operation times are presented in Figure 6 a and b. The local minimum of system throughput were obtained in experiments 1, 5, 7 and 11. The maximum values of the throughput were obtained in experiments 4, 9 and analogically as in the former cases for the last three experiments. The average life span is minimum in experiments 1, 2, 4 and 10.

The results of the simulation research show that, generally, the throughput of the system increases with increasing buffer capacities, but together with an associated increase in the average life span of products (work-in-process). To find the best compromise between throughput and average life span; flow index θ is proposed. To calculate the index, the value of the throughput of the system is divided by the average life span.

(5)

where Τ and Λ are throughput and average life span, respectively.

The results for the experiments are presented in Table 5. In the table the best two results for each variants are shown. The greater the value of the index, the better the compromise between the throughput and average life span that can be found. The best values of the flow indexes were obtained in Variant 2 of the operation times. For experiments 4 and 10, the best values of the index are obtained.

The chart presented in Figure 7 show the values of the flow indexes for all variants of operation times.

Figure 7. The values of flow indexes for the different variants of operation times

The increase in the intermediate buffer capacities results in an increased average life span and work-in-process. The buffer capacity costs money (place, work-in-process, etc.). To comply with the capacity of the intermediate buffers in the analysis of the efficiency of the investigated manufacturing system, the total buffer productivity index is proposed.

(6)

where Τ and Β are throughput and total buffer capacity, respectively.

In Figure 8, the total buffer productivity indexes for all variants are presented. There are no significant differences among the values of buffer productivity indexes measured for the various variants. The greatest value in the index was obtained in experiment 1, and it resulted from the small value of the total buffer capacity (single capacity of intermediate buffers). The next best relation between the throughput and buffer capacity for all variants was obtained in experiments 10 and 11.

Figure 8. The values of total buffer productivity indexes for the different variants of operation times

5. Conclusions

An analysis of the data collected during the simulation experiments shows us that for each variant of the operation times, in the last four experiments (12, 13, 14 and 15) in which all buffer capacities are equal or greater than 10, the throughput of the manufacturing systems achieved the best results (more than 22 products per hour). For the smallest total capacity of intermediate buffers, the greatest value of throughput was reached in experiments 4, 6, 8 and 9. For flow index, experiments 4, 6 and 10 provided very good results, and for buffer productivity index: experiments 1, 10 and 11. For the manufacturing system, which was investigated in this study, the following general conclusions can be formulated:

1. The intermediate buffer capacity has a significant impact on the throughput and average life span of the system.

2. The proper allocation of buffer capacities provides better values of throughput or average life span of the system.

3. It is possible to find a satisfactory compromise between the throughput and average life span of the system, maximizing the proposed flow index.

4. It is possible to find a satisfactory compromise between the throughput and costs of buffer capacities, maximizing the proposed total buffer productivity index.

Further research will encompass other structures of the manufacturing system and the impact of employees on the throughput of the manufacturing system.

Acknowledgement

This article was created by implementation of the grant projects KEGA 004TUKE-4/2013 Intensification of modeling in teaching II. and III. degree in the field of study 5.2.52 Industrial Engineering and VEGA no. 1/0102/11 Experimental methods and modeling techniques in-house manufacturing and non-manufacturing processes

References

[1]  A. Seleima, A. Azaba, T. AlGeddawy, Simulation Methods for Changeable Manufacturing, 45th CIRP Conference on Manufacturing Systems 2012, Procedia CIRP 3, 2012, pp. 179-184.
In article      
 
[2]  D. R. Staley, D. S. Kim, Experimental results for the allocation of buffers in closed serial production lines, International Journal of Production Economics, Vol. 137 (2012), pp. 284-291
In article      View Article
 
[3]  H. A. Vergara, D. S. Kim, A new method for the placement of buffers in serial production lines, International Journal of Production Research, 2009, 47(16), pp. 4437-4456.
In article      View Article
 
[4]  H. Yamashita, T. Altiok, Buffer capacity allocation for a desired throughput in production lines. IIE Transactions 1998; Vol. 30, pp. 883-91.
In article      View Article
 
[5]  G. Gurkan, Simulation optimization of buffer allocations in production lines with unreliable machines. Annals of Operations Research 2000, Vol. 93, pp. 177-216.
In article      View Article
 
[6]  C. Shi, S.B. Gershwin, An efficient buffer design algorithm for production line profit maximization. International Journal of Production Economics 2009 Vol. 122, pp. 725-40.
In article      View Article
 
[7]  J.A. Qudeiri, H. Yamamoto, R. Ramli, A. Jamalim, Genetic algorithm for buffer size and work station capacity in serial-parallel production lines, Artificial Life and Robotics 2008, Vol.12, pp. 102-6.
In article      View Article
 
[8]  N. Nahas, D. Ait-Kadi, M. Nourelfath, Selecting machines and buffers in unreliable series-parallel production lines. International Journal of Produc- tion Research 2009;Vol. 47(14), pp. 3741-74. [32].
In article      View Article
 
[9]  M. Nourelfath, N. Nahas, D. Ait-Kadi, Optimal design of series production lines with unreliable machines and finite buffers. Journal of Quality in Mainte- nance Engineering 2005, Vol. 11(2), pp. 121-38.
In article      View Article
 
[10]  N. O. Fernandes, S. Carmo-Silva, Order release in a workload controlled flow-shop with sequence-dependent set-up times, International Journal of Production Research 2011, Vol. 49(8), pp. 2443-54.
In article      View Article
 
[11]  A. Matta, Simulation optimization with mathematical programming repre-sentation of discrete event systems. In: Proceedings of the 2008 winter simulation conference, 2008, pp. 1393-400.
In article      View Article
 
[12]  M.G. Huang, P.L. Chang, Y.C. Chou, Buffer allocation in flow-shop-type production systems with general arrival and service patterns. Computers & Operations Research 2002, Vol.29 pp. 103-21.
In article      View Article
 
[13]  MacGregor Smith J, Cruz FRB. The buffer allocation problem for general finite buffer queuing networks. IIE Transactions, 2005, Vol. 37(4), pp. 343-65.
In article      View Article
 
[14]  L. Demir, S.Tunalı, D. T. Eliiyi, A. Løkketangen, Two approaches for solving the buffer allocation problem in unreliable production lines, Computers & Operations Research, Vol. 40, 2013, pp. 2556-2563.
In article      View Article
 
[15]  Tecnomatix Plant Simulation version 11.0.0 on-line documentation.
In article      
 
[16]  Trebuňa, P., Straka, M., Rosová, A., Malindžáková, M.: Petri nets as a tool for production streamlining in plastics processing, Przemysl chemiczny, 2015, 94, No. 9, 1000-1003.
In article      
 
[17]  Malindžáková, M., Straka, M., Rosová, A., Kanuchová, M., Trebuňa, P.: Modeling the process for incineration of municipal waste, Przemysl chemiczny, 2015, 94, No. 8, 1000-1004.
In article      
 
[18]  Straka, M., Trebuňa, P., Straková, D., Kliment, M.: Computer Simulation as means of urban traffic elements design, Theoretical and Empirical Researches in Urban Management, Vol. 10, Issue 4, 2015, p. 40-53.
In article      
 
[19]  Kollár, P., Nikitin, Y., Straka, M.: The determination of the shelf mass in the universal shelving stacker by measuring the frequency converter torque generating current of the main drive, Manufacturing technology, Vol. 15, No. 3, p. 363-366.
In article      
 
[20]  Malindžák, D., Straka, M., Helo, P., Takala, J.: The methodology for the logistics system simulation model design, 2010, Metalurgija, Vol. 49, No. 4 (2010), p. 348-352.
In article      
 
[21]  Straka, M.: System of distribution logistics of enterprise Alfa, a.s., Acta Montanistica Slovaca, Vol. 15, Special issue 1, 2010, p. 34-43.
In article      
 
  • CiteULikeCiteULike
  • MendeleyMendeley
  • StumbleUponStumbleUpon
  • Add to DeliciousDelicious
  • FacebookFacebook
  • TwitterTwitter
  • LinkedInLinkedIn