Journal Search Engine
Search Advanced Search Adode Reader(link)
Download PDF Export Citaion korean bibliography PMC previewer
ISSN : 1598-7248 (Print)
ISSN : 2234-6473 (Online)
Industrial Engineering & Management Systems Vol.19 No.3 pp.644-656

A Combined Simulation-based Taguchi Robust Design Approach for Improved Parameter Design

Abebaw Mulat Ashenafi*, Sisay Geremew
Faculty of Mechanical and Industrial Engineering, Department of Industrial Engineering, Bahir Dar University, Bahirdar, Ethiopia
*Corresponding Author, E-mail:
November 7, 2019 April 24, 2020 June 12, 2020


The modeling and examination of coordinated manufacturing systems have turned out to be progressively imperative since the wide acknowledgment of ideas for various leveled control and processing plant robotization. Even though preceding studies of this area demonstrate that the outcomes of simulation can be used for further investigation of design of experiment, investigation of optimum manufacturing process parameter to reduce the cost of manufacturing for a product did not receive comparable attention. Therefore, this paper deals with the optimization of process parameters in a manufacturing systems to minimize the Unit Manufacturing Cost (UMC) of Un-plasticized Poly Vinyl Chloride (UPVC) pipes using a combined Simulation-Robust Taguchi approach. Thus significant factors along with their optimum combination were found using Taguchi techniques to arrive at the least manufacturing cost. The result of this investigation demonstrates that key parameters for optimum performance of the manufacturing system are feed rate (the entities per arrival) in to the hopper and the extrusion rate. The expected unit cost of manufacturing for the selected pipe model is 239 birr and expected cost of manufacturing reduced by 21.6% because the utilization of the resources increased by 13%, 19% for the operators from the actual performance. With the new perspectives of integrated modeling, the study contributes to the body of knowledge which also opens a new outlook for further studies.



    Many process industries have been giving due attention to compete in the global market by delivering quality products with a minimum cost of manufacturing. The case is not exceptional for plastic industries. Today, with industry so focused on the bottom line, failing to improve the performance have a big impact on profitability (Khan et al., 2014). In many companies, there still exist significance variations in the planned performance and the actual performance in manufacturing plastic pipes (Narasimha and Rejikumar, 2015).This is because most of the actions taken to solve this problems in the factories are not effective enough in increasing the performance (Kerealme et al., 2016) This lower performance in APF occurs due to poor understanding of the processing method, use of Inadequate machines, lack of trained staff, Machine Breakdown, and inappropriate working environments. Subsequently, advances dependent on industrial management are important to show and examine this class of systems (Ikram et al., 2013). Now a day the use of computer simulation have been proposed and executed to tackle the issues of increasing UMC in a manufacturing system (Pfeffer et al., 2008). Since the early progress of models and languages, modeling has developed into a method, which is incredibly valuable as an organization to investigate the model as opposed to this present reality framework, and furthermore to break down the connections between the parameters and result. In addition, there is adaptability in the utilization of simulation and modeling language, the model can be worked as near reality as we need and taken as a basic decision tool (Pichitlamken and Nelson, 2003). Subsequently, it is useful to study, schedule and plan production system utilizing modeling and simulation rather than using complicated mathematical model equations (Piera et al., 2004).


    Kerealme et al. (2016), connected Taguchi robust design approach toward utilizing Taguchi Design; L27 orthogonal array to the extrusion process for creation of UPVC (Unplasticized Poly Vinyl Chloride) funnels. Khajanchee and Jain (2017), connected the Cost of Quality Principle for streamlining of process setting factors for Nylon Sheathing procedure of Optical Fiber Cable production also researched the impact and optimization of eight control factors on material evacuation/removal rate (MRR), surface harshness/hardness and kerf. In Wire Electrical Discharge Machining (WEDM) process for instrument steel D2 by use of Taguchi’s L18 orthogonal array, ANOVA and signal to-noise (S/N) ratio for trial plan also endeavored to advance The Friction particle Stir welding (FSW) process which is a strong state mechanical handling innovation empowering high quality joints in materials already considered with low weldability, for example, the greater part of the aerodynamic aluminum compounds. Simulation-based Taguchi method has been used in many research fields (Mahfouz et al., 2011).

    A combination of simulation based DOE was polished by (Yang, 2009) to quantitatively evaluate the impact of various reasons of the bullwhip effects. In other research (Yang, 2009), utilized Taguchi technique to deal with minimization of simulation runs. As case of service area, (Yang and Chou, 2005) combined Discrete Event Simulation (DES) and design of experiment (DOE) to allocate nurses and consulting rooms to every orthopedist. The considered limitations were four appointment arrangement instructions and three progressions of patients. Tsai (2002) used this combination to clarify managerial issues in integrated systems of manufacturing. Li et al. (2009) used this approach to production network improvement. Yang (2009) applied a simulation based- Taguchi approach to design an engine in an incredible electrical organization. Seeking other potential areas, the suggested method of is beneficial to recognize the preferred strategy of maintenance management. Found the right level of delayed differentiation using the simulationbased Taguchi method. Based on (Pichitlamken and Nelson, 2003) experimental design is a functional technique for process improvement and product development. Among the various techniques in DOE, Taguchi method could be used to determine the influence of process factors on the output response using minimum number of experiment. Taguchi design can be utilized to evaluate a process by considering the incontrollable (noise) parameters (Bettonvil et al., 2009). The Taguchi approaches were initially used in quality engineering (Phadke, 1989). It was argued that quality cannot be achieved economically through inspection and statistical analysis, i.e. the socalled ‘on-line quality control’, because inspection and statistical quality control can never fully compensate for a poor design. In order to achieve desirable product quality by design, Taguchi recommended a three stage process: system design, parameter design and tolerance design (Phadke, 1989). It is seen that Taguchi Methods are completely suitable for the activities of simulation.

    From the above frames of references, a simulation model just goes about as an instrument in analyzing execution. It is basically a trial and error strategy, and does not specifically give clarifications to watched framework practices (Tsai, 2002). The Author have used ARENA to model this class of system in spite of tackling the cost problem. Discrete event simulation tool is also used to model continuous process if the specific case is cost in particular. Jilcha et al. (2015) researched worker and machine performance in manufacturing system using simulation. The simulation is done by using Arena software. The performance measure used is throughput and it is directly related to waiting time; work in process, resource or capacity utilization. Simulation models were built for the continuous (extrusion) and injection workshops separately and finally a model is prepared for the purpose of conducting the computer based simulation.

    The essentiality of ARENA, discrete event simulation (DES), to model the continuous manufacturing process is apart from possession of performance characteristics, since companies have a multi- performance characteristics measure and need to be optimized not according to any method proposed for Quality performance only but according to a method which can particularly address the specific case (cost) with improving the quality of the product or keeping it as it is. It is, hence, important to build techniques to reduce the bare bones of the experimentation and to translate the outcomes. In this paper, the utilization of Taguchi Methods in the work of modeling and simulation is proposed to accomplish the above objective. Preceding studies of this area demonstrate that the outcomes of simulation can be used for further investigation of DOE. In this research, a simulation based Taguchi approach was proposed to simulate and investigate the plastic pipe processing industry. Beside this, to the best of the researcher knowledge this is an initial effort made to figure out the optimal manufacturing process parameters to reduce the cost of manufacturing UPVC pipe applying both computer simulation and Robust Taguchi approach.


    Various steps of this investigation are enlightened in next stage. A schematic diagram of these steps is appeared in Figure 1. It shows how the Taguchi approaches can be coordinated with computer simulation to examine plastic pipe process industry. A point by point clarification of each step is explained as follows:

    • 1. generating an idea, the writing survey and issue articulation.

    • 2. Procedure mapping to give a schematic perspective on the model.

    • 3. Model construction, data collection and model development in ARENA to conclude the model. A point by point clarification of each section is discussed in respective areas.

    • 4. Model verification and validation have been conducted in a different section.

    • 5. A progression of main factors should be determined to perform a DOE. Detailed explanation of each main factor is discussed in each sections.

    • 6. Last advance is to decide an appropriate Taguchi design. The Taguchi model was built, and some confirmation runs confirmed the originated results.


    A pipe production line situated in Ethiopia has been utilized as the case investigation of this research. In spite of the fact that, the situation has been broadly studied about, it is as yet worth examining since past work didn't (1) take this unpredictable factors related with cost (2) consider combined characteristics, like, the interactions among the rate of extrusion, feed rate (the entities per arrival). In this work, five factors have been studied and the model deals with more decision points simultaneously. The production system layout is shown in Figure 2 and the processes are briefly described as follows: PVC pipes are manufactured by extrusion of raw material PVC, and generally follow the same steps of typical pipe extrusion operations: Feeding of raw material pellets / powder into the PVC twin screw extruder, Melting and heating in multiple extruder zones, Extruding through a die to shape into a pipe, Cooling of the shaped pipe and Cutting of PVC pipes to the desired length


    The pipe manufacturing plant design is displayed in Figure 2. This scheme is applied to plot the whole procedure of model. The activity stream and whole procedure are appeared in numbers.


    This section talks about the various procedures of model development. It incorporates the description of process, model assumption, information gathering and model construction. These subsections are then trailed by model verification and validation

    6.1 Discerption of Processes

    In general, there are 5 primary workshops in Amhara pipe factory. Raw materials in the form of garden hose and master batch are supplied by foreign countries like Asia and in household like Elsewady Cables Ethiopia and coordinated to raw material store for storing and quality checking.

    They will be sent to two working process continuous and injection process and turn on heat treatment and take the softened plastic with screw thread at that point constrained into shape (mold). As shown in Figure 3 the bundle will be set up by hauloof. After the item is done it will be checked for quality and afterward packaging will be done. At the end; the item will be kept in finishing store. The process is continuous (extrusion) process flow which starts at raw material accepting, and balling unit machine as shown by Figure 3.

    6.2 Model Assumption

    In APF, there are three shifts for each day. The actual working time for whole system is 8:00 working hours per day. Since it isn’t normal that an individual will work without certain interruptions, the workers may set aside effort for their own personal needs.

    Assumptions are taken for modeling and simulation of performance variation at time to rest, machine failure, and power off.

    The model for one shift are investigated, modeled and analyzed

    In the examination only polyvinyl chloride (UPVC) pipe line is chosen.

    Raw material is expected as a constant input, no interruption of activity is happened. The manufacturer working time is 8:00 working hours per shift. Distribution is fitted to input analyzer; it assesses the distribution parameter and computes various proportions of the information.

    6.3 Model Input

    Distribution is fitted to input analyzer; it evaluates the distribution’s parameter and calculates a number of measures of the data. In order to select which type of distribution is used, the author has compared the square error of each distribution. The larger the square error value, the further away the fitted distribution is from the actual data (Jilcha et al., 2015). Therefore the Table 1 fit all summary orders for the distribution from smallest to largest square error. From this it can be seen that triangular is the one with smallest square error and thus it is selected. The simulation model input data’s in Table 1 analyzed using ARENA input analyzer.

    6.4 Implementation

    As indicated by the methodology referenced above, evaluation and optimization of the UPVC pipe manufacturing process can be examined as pursues. Since the goal of this investigation is to obtain the best performance by planning and scheduling the operational parameters of the system, the required points are: what are the system performance measures and in what manner should the operational parameters be coordinated? The answer for these questions can be found by analyzing the system very well (Tsai, 2002).

    The operational parameters are:

    • Extrusion rate-(A)

    • Set-up time-(B)

    • Capacity of the loading station-(C)

    • Feed Rate (entities from the hoppers to the extruder) (D)


    • Maintenance breakdown-(E)

    • While the system performance measures are:

    • Time in system- (T)

    • Queues in loading station-(N1)

    • Utilization of workers and machines- (U1 and U2)

    • Logic of Flow

    Logic flow describes the way by which the entity acts during its journey in the simulation model. It was easy to observe the route the entity follows during the model building stage. The animation part of the Arena was very helpful in ensuring that everything works as desired especially after commencing modification on the model. Many developments to the model were done before the authors reached the final model shown in figure that has the same layout of the real factory. All modifications and continuous verification were aimed to mimic the onsite reality and to mimic the behavior of pipe in the production line within the factory limitations. These entities pass through different inspection stations and when finished they leave the model (get disposed) at the end of the run, Arena provides us with a report containing the data for the performance measures. The animation feature provided by ARENA is involved to check the process of simulation and to assist in model verification. The snapshot of animation is shown in Figure 4.

    6.5 Model Validation

    The model was executed in the ARENA simulation programming. The animation feature provided by ARENA is included to check the procedure of simulation and to aid model verification (Matta, 2008;Melouk et al., 2013). The snapshot of animation is appeared in Fig. Contrasting the throughput of the model outcome, and the actual production rate and the simulation model production rates respectively; it is seen that the actual daily production rates ranges from 150-200 pipes/day. The total daily production of the virtual model is 201/day.

    It is found that the Production rate for the virtual mode approximates the rates for the factory. Hence, the simulation model is validated (Yang et al., 2007) the deviation between the virtual model result of a single

    PVC pipe manufacturing cost and the actual cost of a product should be minimal to say the model represent the reality. Hence a customized equation is developed to identify the actual manufacturing cost of a particular production pipe as shown in equation 1.

    The customized Mathematical equation for unit cost of manufacturing pipe model (OD = 110mm; Nominal pressure = 4 bar; Thickness = 2.2 - 2.7 mm; Length = 6m) can be calculated as

    U M C = D M C + 0. 123 ( D L + F O H ) + 0. 123 ( D L + F O H ) + 0.0 6 ( D L + F O H ) + 0. 258 ( D L + F O H ) + 0. 148 ( D L + F O H ) + 0. 123 ( D L + F O H ) + 0.0 37 ( D L C + F O H ) + 0.0 ( D L + F O H ) + 0.0 37 ( D L + F O H )

    where DMC, DL and FOH are expenses of direct material, direct labour and Factory (manufacturing plant) overhead (Birr) respectively. Mi, Bi, Si, Ei, Di, Ci, Cui, Dri and Sti are costs for mixing, blending, sucking, extrusion, dyeing, cooling, cutting, deco ring and storing. For the calculation of these cost function, the following information are utilized: The average cost of producing a single pipe is 291birr as determined from the manufacturing plants information. This cost is separated to all the processes using an approximate weigh tages value. Mi, Bi, Si, Ei, Di, Ci, Cui, Dri and Sti are 0.123, 0.123, 0.06, 0.258, 0.148, 0.123, 0.037, 0.06 and 0.037. The objective is to determine the optimum combination of Operation parameter values to obtain a minimum unit manufacturing cost.


    The authors define a system as an entity with input variables and output variables (Xu et al., 2010). The input factors are rate of extrusion, set up time, capacity of the loading station, feed rate and Maintainace breakdown with its respective range and the response factor is the manufacturing cost. The setting of operational parameters for each input factor with their respective level is shown in Table 2.

    7.1 Selection of Design

    An enormous number of trials must be completed when the number of process parameters increase (Montevechi et al., 2006). In spite of tackling this problem, the Taguchi strategy a special design of orthogonal array to consider the whole parameter space with just few exami-nations. Five UPVC pipe processing parameters are considered as controlling variables. Every parameter has three levels - to be specific low, medium and high, indicated by 2, 1 and 3 respectively. As per the Taguchi technique, if five parameters and having 3 levels for every parameter L27orthogonal array should to be utilized for the experimentation.

    Three levels or settings are chosen for every parameter and the levels are dismantled adequately far apart so as to cover a maximum region by the levels. For this case, three levels are utilized to permit Identification. The low value of the parameter variable is set in Level 2 and the high level of the parameter variable is arranged in Level 3, while the center worth is set to Level 1. These are recorded in Table 3.

    Changing several operational parameters simultaneously may have interactive effects on the performance which can affect the optimum solution (Unal, 1993). If the case, event that the impact of one parameter on the result relies upon the setting of another parameter, it is said that interaction exists in these two parameters (Phadke, 1989).For this case, four set of two-parameter interaction that might be important were chosen to be studied, i.e. A and C, B and C, C and D, C and E. The UPVC pipe process parameters significant control factors in every response and the chosen levels as well as their corresponding values are taken for the experimentation are outlined under Table 3.

    After alternative level of parameters and possible interactions between parameters have been characterized, the authors have to choose an orthogonal array for conducting the experimental trial (Yang and Chou, 2005). The decision of Orthogonal Array size relies upon the absolute degrees of freedom (DOF) required for the parameters and their interactions (Sankaran et al., 2015). The number of DOF related with every parameter is one less than the number of levels found for that parameter, though the number of DOF related with a parameter interaction is the result of the DOF of every individual parameter. One more DOF is related with the overall mean regardless of the number of control parameters to be studied. Thus, the total number of DOF for this case is 27. In order for an array to be a viable choice, the number of rows must at least be equal to the total DOF required for the case study (Phadke, 1989). Consequently, a L27 array is chosen. After the OA is determined, the authors have assigned parameters and the parameter interactions in suitable column of the OA by utilizing linear graph, which associate to the OA (Phadke, 1989). In this manner, the experimental trial can be conducted as in Table 4.


    After examining impacts of those variables, improved combination of factors are recommended to minimize the manufacturing cost of the UPVC pipes. Furthermore, consequently propose improved combination of factors.

    8.1 Determining Optimal Levels of the Control Parameters

    For the analysis and discussion of experimental results, there are several techniques, which have been utilized (Phadke, 1989). One common approach, ‘Average Value,’ is utilized in this investigation. The average cost for every parameter for all levels are determined and showed in Table 4 using the simulation model output by varying the input variables with the estimated number of replication. Here the sensitivity is computed by taking the difference between the biggest and smallest cost for the parameter (Unal, 1993). These results are separate impacts of every parameter and commonly called ‘main effects’ (Phadke, 1989). From Table, we see that parameter A demonstrates the greatest sensitivity and D is the next most significant.

    8.2 Main Effect Response

    This shows that parameter A and parameter D are significant operational parameters for this manufacturing system. The main effects are likewise indicated graphically in Figure 5. Charting the parameter impacts can give more insight at a glance (Unal, 1993) to estimate parameter interaction effects interaction tables have been utilized (Phadke, 1989).

    Here on Table 6 C-L1denotes parameter C at Level 1, with same nomenclature for the others. The information appeared in the interaction response tables is determined from Table 6. For instance, the information under CL1and A-L1in Table 6 is the average of the result of experimental trial 1-3 where parameter A was at level 1and, simultaneously, parameter C was at level 1. The data in different places of the table are resolved utilizing a similar method (Montevechi et al., 2006).

    This table is additionally appeared in graphical form in Figure 6 and 7. If the lines on the interaction plots are nonparallel, interaction happen and if the lines cross, strong interaction happen between parameters (Phadke, 1989). It tends to be seen that Figure 6 (D and E) shows no interaction because the lines are nearly parallel, while the other show case of strong interaction between parameters. Because the interaction response for the parameters are either no occurrence or not obvi-ous, the ideal levels for the operational parameters would now be able to be chosen by picking the level with the most minimal relative cost from the main ef-fect response table. The optimum levels for the opera-tion parameters are parameter A at level 1, B at level 3, C at level 2, D at level 1, and E at level 2.Since the inter-im between two levels is very large, it is feasible to refine the optimal value of the operational parameters by re-ducing the parameter ranges or increasing the number of levels of every parameter.

    8.3 Selected Interactions Plot

    Besides, the value of parameters whose optimum level inside the acceptable region is fairly certain or whose cost effect sensitivity is insignificant can be fixed and disposed of , e.g. Table 5, we note the sensitivity of parameter E is 16 % and of that of B, the following most minimal value. Consequently, the number of ex-perimental trial do not need to be increased but the re-fined result can be derived (Unal, 1993).The new data for the level of operational parameters are recorded in Table 7. The optimum value determined above is ordi-narily set in Level 2 and smaller interval used to add to it for high level (Level 3) or to subtract from it for low level (Level 1).

    The detailed results for refined experiment are also listed in Table below.

    8.4 Determining Optimal Levels of the Control Parameters

    Following the same procedure as above, the refined main effect responses are shown in Table below. The optimum parameter levels identified are given as: Pa-rameter A at level 1, B at level 3, C at level 2, and D at level 1.

    The optimum level of the parameters could be re-fined further by a similar mechanism. However, for the case of this investigation, the above outcomes are viewed as adequate, because the cost sensitivity is not minimal. Note that the proposed optimum level need not relate to one of the raw of experimental trial (Phad-ke, 1989).

    The determined optimum settings of operational parameters are recorded in Table 11.

    Using these optimum settings to re-run simulation, the predicted system performance are listed in

    8.5 Comparison of Actual (without Adjustment) and Predicted System Performance

    The optimality of the system is verified as the re-sult on Table 12 implied a significant reduction in varia-tions of the important performance characteristics (cost) of Amhara Pipe Factory.


    In this investigation, as the pipe industrial facility system contains five primary operational parameters, each at three levels, the understanding of the total im-pact of every parameter and every possible interaction requires 125 simulation runs. Thus, the application of Taguchi Methods in the simulation has been examined. The outcomes demonstrate that these techniques can be helpful in decreasing the number of experimental trial (only 36 simulation runs have been utilized) ex-pected to decide the best working parameters. The result of this investigation demonstrates that key parameters for optimum performance of the manufacturing system are feed rate (the entities per arrival) in to the hopper and the extrusion rate. The expected unit cost of manu-facturing for the selected pipe model is 239 birr and expected cost of manufacturing reduced by 21.6% because the utilization of the re-sources increased by 13%, 19% for the operators from the actual performance.

    As a part of the general result of this investigation, Taguchi technique plays a vital role in any organiza-tions (regardless of whether it is manufacturing tangible output or service rendering) to fulfill their customer needs, by manufacturing quality item or delivering qual-ity services by decreasing superfluous costs either prior to production or after the product was delivered to cus-tomers.


    This work was supported by Amhara Pipe Factory and Bahir Dar technology institute. The authors would like to tanks them for their valuable collaboration.



    Methodology of the research.


    System process mapping of UPVC pipe manufacturing (Jilcha et al., 2015).


    Model translation and construction through ARENA.


    The snapshot of animation.


    The plot of the main effect response.


    The plot of the interaction response for A and C with B and C.


    The plot of the interaction response for for C and D with D and E.


    PVC pipe process time distribution in continuous (extrusion)

    The optimum settings of operational parameters

    Process parameters and levels

    Designation of Control Factors for conducting experiments and Experimental results found in each runs

    Response table for means

    Interaction response for Parameter A and C

    Refined operational parameters and their level

    Refined experiment trials and results

    Refined main effect response table for means

    The optimum settings of operational parameters

    The predicted system performance

    Before and after optimization comparison


    1. Bettonvil, B. , del Castillo, E. , and Kleijnena, J. P. C. (2009), Statistical testing of optimality conditions in multiresponse simulation-based optimization, European Journal of Operational Research, 199(2), 448- 458.
    2. Ikram, A. , Mufti Nadeem, A. , Saleem, M. Q. , and Khan, A. R. (2013), Parametric optimization for surface roughness, Kerf and MRR in wire electrical discharge machining (WEDM) using Taguchi design of experiment, Journal of Mechanical Science and Technology, 27(7), Pakistan, 2133- 2141.
    3. Jilcha, K. , Berhan, E. , and Sherif, H. (2015), Workers and machine performance modeling in manufacturing system, Computer Science Systems Biology, 8(4), 185-190.
    4. Kerealme, S. , Srirangarajalu, N. , and Asmare, A. (2016), Parameter optimization of extrusión machine producing UPVC pipes using Taguchi method: A case of Amhara pipe factory, International Journal of Engineering Research & Technology (IJERT), 5(1), 65-75.
    5. Khajanchee, A. and Jain, P. (2017), Optimization of nylon extrusion process parameters for optical fiber cable applying matrix experiment using Taguchi method, International Journal of Engineering Science and Computing, 7(4), 6578-6586.
    6. Khan, G. R. , Dalu, R. S. , and Gadekar, S. S. (2014), Defects in extrusion process and their Impact on product quality, International Journal of Mechanical Engineering and Robotics Research, 3(3), 187-194. .
    7. Mahfouz, A. , Shea, J. , and Arisha, A. (2011), Simulation based optimization model for the lean assessment in SME: A case study, Proceedings of the 2011 Winter Simulation Conference, Phoenix, AZ, USA.
    8. Matta, A. (2008), Simulation optimization with mathematical programming representation of discrete event systems, Proceedings of the 2008 Winter Simulation Conference, Miami, FL, USA, 1393-1400.
    9. Melouk, S. H. , Freeman, N. K. , Miller, D. , and Dunning, M. (2013), Simulation optimization-based decision support tool for steel manufacturing, International Journal of Production Economics, 141(1), 269-276.
    10. Montevechi, J. A. , De Almeida Filho, R. G. , and Medeiros, A. L. (2006), Application of factorial designs for reducing factors in optimization via discrete-event simulation, Proceedings of the 2006 Winter Simulation Conference, Monterey, CA, USA, 1977-1984.
    11. Narasimha, R. M. and Rejikumar, R. (2015), Plastic pipe defects minimization, International Journal of innovative Research & Development, 2(5), 1337-1351.
    12. Pfeffer, M. , Oechsner, R. , Pfitzner, L. , Ryssel, H. , Ocker, B. , and Verdonck, P. (2008), Performance optimization of semiconductor manufacturing equipment by the application of discrete event simulation, Proceedings of the 28th Computers and Information in Engineering Conference, Parts A and B, Brooklyn, New York, NY, 505-513.
    13. Phadke, M. S. (1989), Quality Engineering Using Robust Design, Prentice-Hall, Inc. NJ.
    14. Pichitlamken, J. and Nelson, B. L. (2003), A combined procedure for optimization via simulation, ACM Transactions on Modeling and Computer Simulation, 13(2), 155-179.
    15. Piera, M. À. , Narciso, M. , Guasch, A. , and Riera, D. (2004), Optimization of logistic and manufacturing systems through simulation: A colored Petri netbased methodology, Simulation, 80(3), 121-129.
    16. Sankaran, S. V. , Segar, V. , Subramanian, B. , and Krishna, A. (2015), Optimization of manufacturing process Parameters using arena simulation and Taguchi method, International Journal of Mechanical Engineering & Robotics Research IJMERR, 4(1), 435-444.
    17. Syberfeldt, A. and Lidberg, S. (2012), Real-world simulation- based manufacturing optimization using cuckoo search, Proceedings of the 2012 Winter Simulation Conference, Berlin, Germany, 1-12.
    18. Tsai, C. (2002), Evaluation and optimization of integrated manufacturing system operations using Taguch’s experiment design in computer simulation, Article in Computers & Industrial Engineering, 43(3), 591-604.
    19. Xu, J. , Barry, L. , and Hong, J. E. F. F. L. (2010), Industrial strength COMPASS: A comprehensive algorithm and software for optimization via simulation, ACM Transactions on Modeling and Computer Simulation, 20, 1-29.
    20. Yang, T. (2009), an evolutionary simulation-optimization approach in solving parallel-machine scheduling problems: A case study, Computers & Industrial Engineering, 56(3), 1126-1136.
    21. Yang, T. and Chou, C. P. (2005), Solving a multi response simulation optimization problem with discrete variables using a multiple-attribute decisionmaking method, Mathematics and Computers in Simulation, 68(1), 9-21.
    22. Yang, T. , Fu, H. P. , and Yang, K. Y. (2007), An evolutionary- simulation approach for the optimization of multi-constant work-in-process strategy: A case study, International Journal of Production Economics, 107(1), 104-114.