- Email: [email protected]
A Proposal Simulation Method towards Continuous Improvement in Discrete Manufacturing
Available online at www.sciencedirect.com
ScienceDirect Procedia CIRP 57 (2016) 270 – 275
49th CIRP Conference on Manufacturing Systems (CIRP-CMS 2016)
A Proposal Simulation Method towards Continuous Improvement in Discrete Manufacturing Victor Emmanuel de Oliveira Gomesa,b, Luis Gonzaga Trabassoa* a Technological Institute of Aeronautics, São José dos Campos, Brazil SENAI Institute of Technology in Metal Mechanics, Belo Horizonte, Brazil
b
* Corresponding author. Tel.: +0-000-000-0000 ; fax: +0-000-000-0000. E-mail address: (victor; gonzaga)@ita.br; [email protected]
Abstract In kaizen improvement projects, the stages of analysis, and application of the proposed improvements are often a trial-and-error cycle carried out by direct experimentation. This feature is a major source of uncertainty in resource dimensioning. This paper presents the design and development of a sequence of activities that emphasizes the application of simulation capabilities as a tool to aid the continuous improvement process at discrete manufacturing, in the context of the Lean Manufacturing approach. © Published by Elsevier B.V. This ©2016 2015The TheAuthors. Authors. Published by Elsevier B.V.is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of Scientific committee of the 49th CIRP Conference on Manufacturing Systems (CIRP-CMS 2016). Peer-review under responsibility of the scientific committee of the 49th CIRP Conference on Manufacturing Systems Keywords: kaizen improvement projects; discrete manufacturing; discrete event simulation.
1. Introduction This work seeks to systematize the analysis for layout modifications carried out during kaizen events in discrete manufacturing companies. A method is proposed to execute the process of discrete event simulation (DES) inserted into kaizen activities. Discrete manufacturing companies need often a flexible manufacturing system that can develop quality and time-tomarket according to product demand fluctuations. These requirements imply regular analyses in current production processes to generate improvements related to costs or operational practices, which can result in facility layout modifications. A lot of companies have chosen to apply changes in their shop floor by means of Kaizen events which are characterized, in part, by direct experimentation and trialand-error cycles. Therefore, occasionally there are mistakes to predict the behavior of future states. Traditionally, the use of kaizen has assisted improvements in production systems. However, when it is applied to more complex problems, with greater amount of data to be analyzed, some negative factors are highlighted [1.2]:
x Kaizen teams are not always knowledgeable about the process under study or are not prepared for analysis of complex processes; x Variation must be addressed, both random and structural; x Data must be fully analyzed to help understand the random nature of system behavior; x The interaction between system components must be assessed; x The future state must be validated before it is implemented to minimize or eliminate the period of trial and error adjustments; x Alternatives to the future state must be systematically identified and considered. The mentioned factors are substantially related to the knowledge of the process and precisely to the management and analysis of data associated with the behavior of a production system. This analysis can be done by means of collaborative tools for manufacturing environment simulation, since these tools are attached to the practices applied by the corporation. In this context, the main contribution of this
2212-8271 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the scientific committee of the 49th CIRP Conference on Manufacturing Systems doi:10.1016/j.procir.2016.11.047
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
paper is on the systematic merging of a traditional discrete event simulation method with the kaizen event method. 2. Related work Some of difficulties regarding the use of Lean Manufacturing approach are related to the planning towards the effective implementation [3,4]. For this reason, simulation tools have been used to support systems specifications for manufacturing improvement in order to give greater effectiveness for analysis of resource utilization on production layout modifications. Standridge & Marvel [1] approach the use of simulation in kaizen projects and claim that the future state layout must be validated through simulation before an improvement implementation plan, in order to minimize or eliminate periods of trial and error adjustments. Kumar & Phrommathed [5] have applied discrete events simulation by means of popular software tools to examine a paper sheet cutting operation. The re-designed operation has resulted in setup time reduction and productivity increased. Grimard et al. [6] describe the validation of a future state of a re-designed injector calibration work cell using a deterministic simulation. The simulation results were used to refine initial estimates of throughput and validate worker movement in the cell. The works [1] [5] and [6] have used the method described by Banks et al. [7] and Kelton et al. [8], applied for analysis of future state in production systems. They report a better evaluation and increase in the effectiveness of planning for modifications by the use of DES. However, there is no formalized routine for the integration of processes improvement and simulation activities. Khalil et al [9] proposed a routine for implementations of improvements by means of DES intended to increase the amount of potentials solutions generated towards the future state of a production system. Nevertheless, roles and activities were not assigned to kaizen team members, which have prominent importance in improvement process [10]. Although closely related to this paper, there are some important differences between these approaches and the problem approached herein. These works use simulation tools often under the domain of digital manufacturing experts, but without leverage the company´s collective knowledge, which is so emblematic in the process improvement approaches. A common problem faced by many companies looking to employ simulation tools into manufacturing process is to obtain the information that their users really need. [11]. In this context, the effective use of these tools is related to standardization of procedures of own corporate system. The kaizen process is one of these procedures already applied in many discrete manufacturing corporations and establishes an opportunity of obtaining standard information needed by simulation tools users and, consequently, their joint use can help employers adhere to use of simulation tools. Compared with [1], [5], [6] e [9], a differential approach in this article is the formalization and a detailed description of a method for more accurate diagnoses of future states during kaizen events. The proposed method inserts discrete event simulation routine into kaizen activities.
3. Simulation Process Simulation is an experimental process which uses a detailed model of a real system to determine responses to changes caused in its structure, environments and boundary [12]. A simulation analysis changes according to the type of system analyzed and may be continuous or discrete. Discrete event simulation is suitable for problems in which variables change in discrete times and by discrete steps. On the other hand, continuous simulation is suitable for systems in which the variables can change continuously [8]. The automotive industries have increasingly used the simulation as a prominent decision support tool. Most makes use of discrete-event simulation (DES) to model manufacture systems and analyze issues related to factory layout, process flow, material handling systems, capacity planning, utilization of manpower, investment in new equipment, production and logistics scheduling [14]. The development of a computer program is just one of the many activities of a simulation process. For this to be successful, other activities should be followed. This set of activities or process is known in the literature as simulation methodology or lifecycle of a simulation model [7.8]. 4. Simulation aided continuous improvement - MAPS The term "kaizen event" is used to indicate a limited time period where are realized identification and implementation of improvements [15]. In a typical Kaizen Blitz project, a crossfunctional multilevel team of 6 to 12 members works intensely, 12 to 14 hours a day, to rapidly develop, test, and refine solutions to problems and leave a new process in place in just a few days. [10]. The method developed in this work, called MAPS Melhoria Auxiliada por Simulacão (Simulation Aided Improvement), has routines of a simulation process inserted in Kaizen activities, with the purpose of increasing the level of knowledge about the stages of Kaizen event process and improve the decision making for modifications of factory layout. The method MAPS contemplates the approach of continuous improvement when it is considering that at one point in the Future State becomes the Current State and emphasizes the application of simulation capabilities in the step to implement improvements, which traditionally occurs on the third day of the event kaizen . The MAPS considers that the corporation has already defined the sector in which the project for improvement application will be made. It is also considered that the kaizen team is already formed and contains a simulation analyst, who is responsible for setting the measurement team and the model validation team within the team kaizen. The MAPS consists of four steps: x x x x
Step 1 – Define the Project; Step 2 – Current-State Analysis; Step 3 – Computational Modeling; Step 4 – Future-State Analysis.
271
272
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
by simulation analyst. The models developed are validated by the model validation team, still during the activity. x Identify opportunities for improvement: The kaizen team is in charge of activities to identify opportunities for improvement in the productive system under analysis. Each waste is described, filmed (when necessary), documented and presented to the kaizen team. After analyzing the waste and make suggestions for improvement, a plan of action shall be drawn. x Collect operational data: Measurements are made by the team of measurement. Probabilistic data collection (time of supply, cycle times of machining operations, among others) is supervised by the simulation analyst. He is also responsible for the processing of data collected (cronoanalysis results) and statistical inference. 4.3. Computational Modeling
Fig. 1. MAPS (Simulation Aided Improvement).
4.1. Define the project The first step is to define the project. Occur, necessarily, the following activities: Description of the analyzed system; Setting dates and procedures for application design improvement; Definition of the goals of the simulation; Presentation of project performance indicators; Setting the Model validation team and measurement team. The system description is taken by kaizen team leader, and should contain characteristics, as productive resources, provision of resources, number of operators, shifts etc. Two members from kaizen team (necessarily, people who know the production system under analysis) form the model validation team along with the simulation analyst. Besides, the measurement team is de-fined, in order to acquire data from the shop floor. 4.2. Current-State Analysis In this step the kaizen team goes to the shop-floor (Gemba kaizen) to map the process and identify waste. This step guides the current state mapping and provides data to the conceptual modeling. It has three different activities, which are described below: x Process Mapping: The simulation analyst is responsible for acquire information about the process, such as production sequence, area, distances and relationship between operations. This information, which represents a description for the current state, is represented by means of flowchart and the manufacturing floor plan. This mapping also assists in obtaining the conceptual model, developed
After validate and verify the conceptual model, the simulation analyst elaborates the computational model by means of simulation software. Operational data collected are analyzed through specific tools (e.g. DataFit, Minitab, Excel) or other available simulation packages. Afterwards the application of computer modeling software, the simulation is performed and the results are validated by the validation team models. The validation of the computational model is a continuous process, which follows the entire life cycle of the improvement project. 4.4. Analyze the Future State This step uses the information gathered in the previous steps to simulate experimental models and, with the kaizen team, set the template for the Future State of the productive system analyzed. It consists of the following activities. x Conduct Experiments: This activity is characterized by the development of computational models that meet the requirements for the future state model of the production system. From the information obtained in previous stages or in corporation history, if any, the model validation team defines models (as needed) to be analyzed with the kaizen team during a workshop. x Perform Workshop: This activity is characterized by meetings organized by kaizen team for analysis and improvement proposals, which will be implemented in the simulation environment. The moderator of this workshop should be part of the model validation team. Basically, this workshop activities are developed in the following sequence: Introduction, analysis of the current situation, development of possible solutions; documentation. The models suggested are presented by the model validation team, containing a comparison of simulation results between the model of the Current State and the models suggested for the Future State. Subsequently, possible
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
complementary solutions are developed by kaizen team, and if there is new information from the improvement actions (e.g. reducing the operations cycle time), these are documented to be implemented in the computational model chosen for the Future State. x Define Future State: After the final definition of the model for the Future State, the changes suggested in the workshop are implemented in the computational model, as well as the new information from the improvement actions. Finally, the simulation results of the Future State are compared with the results for the Current State. Furthermore, the performance indicators are analyzed and presented.
x Modification of bench arrangement. The results on the machines usage in the current state and future state (F-S) were presented in terms of working, waiting and blocked rates, and are illustrated in Figure 2. A significant change stands out in the working rate allocated to OP20, which influenced the behavior of the subsequent operations.
5. Case Studies To check the feasibility, applicability, usability and limitations of MAPS, it was applied to corporations operating in the automotive manufacturing industry, using kaizen events to implement improvements in their processes. Computational models were produced using Tecnomatix Plant Simulation TM software and analyses of occupation of production resources were performed in a DataFit from the same application. The Case Study-A emphasizes the comparison between the results of the improvement modifications obtained through MAPS, with results obtained in a kaizen process without this aid. The Case Study-B shows a layout modification performed in the same company in a similar manufacturing cell, and highlights the use of historical data and best practices of kaizen simulated by MAPS. 5.1. Case Study A The system is a manufacturing cell (MC) that monthly produces 7152 auto parts machined in pairs (right and left) by six machines, and transported by five operators with the help of treadmills. There are three work shifts daily, totaling 15 operators per day (figure XX). The model validation and measurement teams were formed by the kaizen leader and the specific goals for the kaizen event were defined as:
Fig. 2. Production resources usage rates for current and future states (F-S).
The possibility of an operator removal from the manufacturing cell, previously found in kaizen, was ratified by the simulation results. It is observed that, in Figure 6, the transport time of operator 3 in operation 30 has decreased. This occurred due to the approach of benches and treadmills. As a result of an operator removal in the new layout, there was an increase in the waiting and blocked time rate for operations 40/50, inspection and cleaning, besides an increase in the transporting time rate of operator 4. At the end of the future state simulation, the productive capacity of the manufacturing cell increased by approximately 30%. The production efficiency of the operators also increased, as shown in Figure 3.
x Free up one operator for other activities; x Improve the productive capacity by 30 %. The schedule established by the model validation team lasted ten days between the phases Set Design and Analyze the Future State, forecasted in MAPS. After the steps of process mapping, data collection and identification of opportunities for improvement, a computational model was developed from the Current State, and then validated by the models validation team. The future state models were analyzed during the workshop referred in MAPS, and the changes made in the current state model of the MC, as shown in Figure 4, were: x Modification of material flow direction; x Modification of operations 20 and 30 sequence;
Fig. 3. Rates of parts manufactured monthly (current and future states).
Traditionally, at the end of the fifth day of kaizen events an analysis of benefits is performed to compare the current state to the estimated future state. This analysis is based on indicators of system performance and verifies the influence of the proposed changes. Furthermore, it serves to justify investment of resources. However, treatment of the time variable for the analysis of the MC’s resource behavior influences directly in the estimated results.
273
274
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
When using mean times for machine processing and operator handling, variations derived from randomness in the system behavior are not considered. Consequently, the results of the estimates for the MC’s future state may present significant differences. This means that it can estimate gains that are not real. This case study illustrates this type of occurrence, and confirms that the results obtained by computer simulation are closer to the behavior of the real system than those obtained by calculations in spreadsheets. The Table 1 details this aspect. The initial data of kaizen event, corresponding to the current state of the manufacturing cell, and also data of parts production for the future state are presented. The data called conventional were obtained by calculations in spreadsheets, the simulated data were obtained through the simulation process, using the method proposed in this paper. Table 1. Comparison of estimated capacity increase for future state. Production (parts/ month)
Amount of operators (H) x (3 shifts)
Productive capacity (parts/H month)
Productive capacity increase
Initial data
7152
15
476
-
Conventional
8333
12
694
46%
Simulated
7405
12
617
30%
It is observed that the results estimated by the conventional method, without simulation, indicate an increase in production capacity by 46% for the future state. If compared with the results estimated by MAPS, with the help of simulation, the increase in production capacity is expected to be 30%. Errors in the assessment of the future state can lead to misguided investments. Whereas the simulation minimizes these errors, the great difference in the results of this comparison shows the limitation of evaluation in decision making for changes in kaizen processes, when simulation tools are not used. For this case study, the difference in results can be considered punctual, as this work was carried out only for one manufacturing cell. However, the greater the complexity of the real system, in other words, the larger the number of components of the production system (manufacturing cells) and their interactions, the larger the errors caused by the random behavior of their resources (machines, operators, conveyors) will be, both in simulation processes and estimation by spreadsheets. Because it is a process of continuous improvement, restrictions and waste will be identified over time to increase the productive efficiency of the MC. The computational models used for this MC will serve as references for new changes in the same MC and for similar MCs, and the data acquired will remain in the database of simulated models. Thus, the next kaizen events may contain much more information, resulting in greater efficiency in the improvement process and faster kaizen event.
5.2. Case Study B The case study B was conducted in a manufacturing cell (MC) presented in Figure 4 which produces two types of similar automotive parts (Part type-A and Part type- B), machined in pairs. Parts type-B are produced by machines on the left, as represented in figure 8, while type-A parts are produced by the machines on the right. Both sides present the OP10 (Operation 10), OP20, OP30, OP40/50, OP60 and OP70. Just OP80, visual inspection, it is common to both. The MC has a total of 10 operators. The schedule established by the model validation team lasted 20 days and the specific objectives of the kaizen event were defined as: x Provide 30% of manpower for other activities; x Provide a machining center (OP20/30) to another sector of the company; x Increase production capacity by 40%.
Fig. 4. case study B – Current state layout.
Upon completion of the data collection activity, the current state's computational model is designed to be subsequently validated by the model validation team. Figure 10 shows a screenshot of this model. He assisted to visual verification of bottlenecks and its effects on the capacity, and served as a reference for the elaboration of the model of the future state. The proposed changes by MAPS workshop were modeled and simulated after the current state model validation. These changes are listed and illustrated in Figure 11, according to the numerical order shown below: 1. 2. 3. 4.
Remove a machining center of OP20 and OP30; Replace machine OP10 (type-A); Install mats on the side to reduce risk and beats; Modify machine placement of OP40 and OP50 to decrease movement of operators and allow an operator to occupy more than one workstation; 5. Insert the OP60 (inspection 1) and the OP70 (inspection 1) at OP80.
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
A comparison between the current state and future state is presented in Table 2. At the end of kaizen, their goals were met. An observation made six months after modification layout indicated a decrease in the rate of scrap parts by 60%. Regarding the method MAPS, using data from computational models performed for the case study A helped in the development of algorithms used in computational models of the case study B. This reduced the time for development of computational models. Table 2. Productive capacity comparison between current state and future state.
Productive capacity (parts/H month)
Amount of operators (H)
Product Type
Current State
Future State
Gains
Type-A
417
701
67%
Type-B
436
552
26%
Type-A+ Type-B
853
1253
46%
Type-A+ Type-B
11
7
36%
6. Conclusions This work achieved the objective to systematize the application of simulation into continuous improvement process on the shop floor, using a method that formalizes simulation activities in conjunction with the kaizen tasks. The case studies highlighted the importance of the use of simulation techniques applied to support the continuous improvement process. The main positive factor of method MAPS is to increase the level of knowledge of the behavior of the analyzed system. The use of probabilistic data considers the random behavior of the real system and can avoid gross errors in estimating earnings improvements. The comparison of the results of kaizen with deterministic and probabilistic data, shown in case study 1, illustrates this situation. It is emphasized that historical data and computational models from MAPS can be used in future analyses. This fact minimize the time to develop new similar computational models to the same company. The access to information is the main limitation of MAPS. When the simulation analyst is not familiar with the real system analyzed, there is more time spent with the modeling process and the possibility of errors in this activity. Access to historical data on the operation of systems is an important factor for the time taken to validation activity and computational modeling. If the mapping of all the company's processes becomes a common practice, along with the creation of a data base (with
input and output data) for computational models, simulation processes to aid in modifying layouts will be faster and more efficient, and consequently improvement projects will be more effective. Acknowledgements The authors thank FAPEMIG (Minas Gerais State Foundation for Research Development) for the financial support. References [1] STANDRIDGE, C.R.; MARVEL, J.H. 2006. Why Lean Needs Simulation. In: Winter Simulation Conference, Piscataway, Institute of Electrical and Electronics Engineers, p. 1907-1913.2006. [2] OLIVEIRA, C. S. Aplicação de Técnicas de Simulação em Projetos de Manufatura Enxuta. Estudos Tecnológicos - Vol. 4, n° 3: 204-217. 2008. [3] WOMACK, J.P.; JONES, D.T. A mentalidade enxuta nas empresas. 3ª ed., Rio de Janeiro, Campus, 408 p. 2004. [4] Ibrahim A. Rawabdeh, (2005),"A model for the assessment of waste in job shop environments", International Journal of Operations & Production Management, Vol. 25 Iss: 8 pp. 800 – 822. [5] KUMAR, S., PHROMMATHED, P. Improving a manufacturing process by mapping and simulation of critical operations. Journal of Manufacturing Technology Management, 17(1), 104-132. 2006. [6] GRIMARD, C.; MARVEL, J.H.; STANDRIDGE, C.R. Validation of The Re-Design of a Manufacturing Work Cell Using Simulation. In: Winter Simulation Conference, Piscatawa, p.1386-1391. 2005. [7] BANKS, J.; CARSON, J.; NELSON, B. Discrete-event system simulation. New Jersey: Prentice Hall, 1996. [8] KELTON, W. D., SADOWSKI, R. P., SADOWSKI, D. A.Simulation with ARENA. 3 Ed, New York, McGraw-Hill Companies Inc. 2004. [9] KHALIL, R., KANG, P., STOCKTON, D. “Integration of Discrete Event Simulation with an Automated Problem Identification”, IMECS– (International Multi-Conference of Engineers and Computer Scientists), Hong Kong. 2010. [10] MIKA, G.L. Kaizen Event Implementation Manual, Kaizen Sensei, Wake Forest, NC. 2002. [11] DE CARLI, P.C.; DELAMARO, F.C.; SALOMON, V.A.P. Identificação e Priorização dos Fatores Críticos de Sucesso na Implantação de Fábrica Digital, Revista Produção. São Paulo. 2010. [12] HARREL, Charles R.; MOTT, Jack R. A.; BATEMAN, Robert E. BOWDEN, Royce G. GOGG, Thomas J. Simulação: Otimizando sistemas. 2. ed. São Paulo, SP: Instituto IMAM. 134 p. 2002. [13] BERTRAND, J. W. M.; FRANSOO, J. C. Modeling and simulation: operations management research methodologies using quantitative modeling. International Journal of Operations & Production Management, v. 22, n. 2, p. 241-264, 2002. [14] KUEHN, W. Digital factory: integration of simulation enhancing the product and production process towards operative control and optimization. International Journal of Simulation, v. 7, n. 7, p. 27-29, 2006. [15] SÁNCHEZ, A. M., PÉREZ, M. "Lean indicators and manufacturing strategies", International Journal of Operations & Production Management, Vol. 21 Iss: 11, pp.1433 – 1452, 2001.
275
ScienceDirect Procedia CIRP 57 (2016) 270 – 275
49th CIRP Conference on Manufacturing Systems (CIRP-CMS 2016)
A Proposal Simulation Method towards Continuous Improvement in Discrete Manufacturing Victor Emmanuel de Oliveira Gomesa,b, Luis Gonzaga Trabassoa* a Technological Institute of Aeronautics, São José dos Campos, Brazil SENAI Institute of Technology in Metal Mechanics, Belo Horizonte, Brazil
b
* Corresponding author. Tel.: +0-000-000-0000 ; fax: +0-000-000-0000. E-mail address: (victor; gonzaga)@ita.br; [email protected]
Abstract In kaizen improvement projects, the stages of analysis, and application of the proposed improvements are often a trial-and-error cycle carried out by direct experimentation. This feature is a major source of uncertainty in resource dimensioning. This paper presents the design and development of a sequence of activities that emphasizes the application of simulation capabilities as a tool to aid the continuous improvement process at discrete manufacturing, in the context of the Lean Manufacturing approach. © Published by Elsevier B.V. This ©2016 2015The TheAuthors. Authors. Published by Elsevier B.V.is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of Scientific committee of the 49th CIRP Conference on Manufacturing Systems (CIRP-CMS 2016). Peer-review under responsibility of the scientific committee of the 49th CIRP Conference on Manufacturing Systems Keywords: kaizen improvement projects; discrete manufacturing; discrete event simulation.
1. Introduction This work seeks to systematize the analysis for layout modifications carried out during kaizen events in discrete manufacturing companies. A method is proposed to execute the process of discrete event simulation (DES) inserted into kaizen activities. Discrete manufacturing companies need often a flexible manufacturing system that can develop quality and time-tomarket according to product demand fluctuations. These requirements imply regular analyses in current production processes to generate improvements related to costs or operational practices, which can result in facility layout modifications. A lot of companies have chosen to apply changes in their shop floor by means of Kaizen events which are characterized, in part, by direct experimentation and trialand-error cycles. Therefore, occasionally there are mistakes to predict the behavior of future states. Traditionally, the use of kaizen has assisted improvements in production systems. However, when it is applied to more complex problems, with greater amount of data to be analyzed, some negative factors are highlighted [1.2]:
x Kaizen teams are not always knowledgeable about the process under study or are not prepared for analysis of complex processes; x Variation must be addressed, both random and structural; x Data must be fully analyzed to help understand the random nature of system behavior; x The interaction between system components must be assessed; x The future state must be validated before it is implemented to minimize or eliminate the period of trial and error adjustments; x Alternatives to the future state must be systematically identified and considered. The mentioned factors are substantially related to the knowledge of the process and precisely to the management and analysis of data associated with the behavior of a production system. This analysis can be done by means of collaborative tools for manufacturing environment simulation, since these tools are attached to the practices applied by the corporation. In this context, the main contribution of this
2212-8271 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the scientific committee of the 49th CIRP Conference on Manufacturing Systems doi:10.1016/j.procir.2016.11.047
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
paper is on the systematic merging of a traditional discrete event simulation method with the kaizen event method. 2. Related work Some of difficulties regarding the use of Lean Manufacturing approach are related to the planning towards the effective implementation [3,4]. For this reason, simulation tools have been used to support systems specifications for manufacturing improvement in order to give greater effectiveness for analysis of resource utilization on production layout modifications. Standridge & Marvel [1] approach the use of simulation in kaizen projects and claim that the future state layout must be validated through simulation before an improvement implementation plan, in order to minimize or eliminate periods of trial and error adjustments. Kumar & Phrommathed [5] have applied discrete events simulation by means of popular software tools to examine a paper sheet cutting operation. The re-designed operation has resulted in setup time reduction and productivity increased. Grimard et al. [6] describe the validation of a future state of a re-designed injector calibration work cell using a deterministic simulation. The simulation results were used to refine initial estimates of throughput and validate worker movement in the cell. The works [1] [5] and [6] have used the method described by Banks et al. [7] and Kelton et al. [8], applied for analysis of future state in production systems. They report a better evaluation and increase in the effectiveness of planning for modifications by the use of DES. However, there is no formalized routine for the integration of processes improvement and simulation activities. Khalil et al [9] proposed a routine for implementations of improvements by means of DES intended to increase the amount of potentials solutions generated towards the future state of a production system. Nevertheless, roles and activities were not assigned to kaizen team members, which have prominent importance in improvement process [10]. Although closely related to this paper, there are some important differences between these approaches and the problem approached herein. These works use simulation tools often under the domain of digital manufacturing experts, but without leverage the company´s collective knowledge, which is so emblematic in the process improvement approaches. A common problem faced by many companies looking to employ simulation tools into manufacturing process is to obtain the information that their users really need. [11]. In this context, the effective use of these tools is related to standardization of procedures of own corporate system. The kaizen process is one of these procedures already applied in many discrete manufacturing corporations and establishes an opportunity of obtaining standard information needed by simulation tools users and, consequently, their joint use can help employers adhere to use of simulation tools. Compared with [1], [5], [6] e [9], a differential approach in this article is the formalization and a detailed description of a method for more accurate diagnoses of future states during kaizen events. The proposed method inserts discrete event simulation routine into kaizen activities.
3. Simulation Process Simulation is an experimental process which uses a detailed model of a real system to determine responses to changes caused in its structure, environments and boundary [12]. A simulation analysis changes according to the type of system analyzed and may be continuous or discrete. Discrete event simulation is suitable for problems in which variables change in discrete times and by discrete steps. On the other hand, continuous simulation is suitable for systems in which the variables can change continuously [8]. The automotive industries have increasingly used the simulation as a prominent decision support tool. Most makes use of discrete-event simulation (DES) to model manufacture systems and analyze issues related to factory layout, process flow, material handling systems, capacity planning, utilization of manpower, investment in new equipment, production and logistics scheduling [14]. The development of a computer program is just one of the many activities of a simulation process. For this to be successful, other activities should be followed. This set of activities or process is known in the literature as simulation methodology or lifecycle of a simulation model [7.8]. 4. Simulation aided continuous improvement - MAPS The term "kaizen event" is used to indicate a limited time period where are realized identification and implementation of improvements [15]. In a typical Kaizen Blitz project, a crossfunctional multilevel team of 6 to 12 members works intensely, 12 to 14 hours a day, to rapidly develop, test, and refine solutions to problems and leave a new process in place in just a few days. [10]. The method developed in this work, called MAPS Melhoria Auxiliada por Simulacão (Simulation Aided Improvement), has routines of a simulation process inserted in Kaizen activities, with the purpose of increasing the level of knowledge about the stages of Kaizen event process and improve the decision making for modifications of factory layout. The method MAPS contemplates the approach of continuous improvement when it is considering that at one point in the Future State becomes the Current State and emphasizes the application of simulation capabilities in the step to implement improvements, which traditionally occurs on the third day of the event kaizen . The MAPS considers that the corporation has already defined the sector in which the project for improvement application will be made. It is also considered that the kaizen team is already formed and contains a simulation analyst, who is responsible for setting the measurement team and the model validation team within the team kaizen. The MAPS consists of four steps: x x x x
Step 1 – Define the Project; Step 2 – Current-State Analysis; Step 3 – Computational Modeling; Step 4 – Future-State Analysis.
271
272
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
by simulation analyst. The models developed are validated by the model validation team, still during the activity. x Identify opportunities for improvement: The kaizen team is in charge of activities to identify opportunities for improvement in the productive system under analysis. Each waste is described, filmed (when necessary), documented and presented to the kaizen team. After analyzing the waste and make suggestions for improvement, a plan of action shall be drawn. x Collect operational data: Measurements are made by the team of measurement. Probabilistic data collection (time of supply, cycle times of machining operations, among others) is supervised by the simulation analyst. He is also responsible for the processing of data collected (cronoanalysis results) and statistical inference. 4.3. Computational Modeling
Fig. 1. MAPS (Simulation Aided Improvement).
4.1. Define the project The first step is to define the project. Occur, necessarily, the following activities: Description of the analyzed system; Setting dates and procedures for application design improvement; Definition of the goals of the simulation; Presentation of project performance indicators; Setting the Model validation team and measurement team. The system description is taken by kaizen team leader, and should contain characteristics, as productive resources, provision of resources, number of operators, shifts etc. Two members from kaizen team (necessarily, people who know the production system under analysis) form the model validation team along with the simulation analyst. Besides, the measurement team is de-fined, in order to acquire data from the shop floor. 4.2. Current-State Analysis In this step the kaizen team goes to the shop-floor (Gemba kaizen) to map the process and identify waste. This step guides the current state mapping and provides data to the conceptual modeling. It has three different activities, which are described below: x Process Mapping: The simulation analyst is responsible for acquire information about the process, such as production sequence, area, distances and relationship between operations. This information, which represents a description for the current state, is represented by means of flowchart and the manufacturing floor plan. This mapping also assists in obtaining the conceptual model, developed
After validate and verify the conceptual model, the simulation analyst elaborates the computational model by means of simulation software. Operational data collected are analyzed through specific tools (e.g. DataFit, Minitab, Excel) or other available simulation packages. Afterwards the application of computer modeling software, the simulation is performed and the results are validated by the validation team models. The validation of the computational model is a continuous process, which follows the entire life cycle of the improvement project. 4.4. Analyze the Future State This step uses the information gathered in the previous steps to simulate experimental models and, with the kaizen team, set the template for the Future State of the productive system analyzed. It consists of the following activities. x Conduct Experiments: This activity is characterized by the development of computational models that meet the requirements for the future state model of the production system. From the information obtained in previous stages or in corporation history, if any, the model validation team defines models (as needed) to be analyzed with the kaizen team during a workshop. x Perform Workshop: This activity is characterized by meetings organized by kaizen team for analysis and improvement proposals, which will be implemented in the simulation environment. The moderator of this workshop should be part of the model validation team. Basically, this workshop activities are developed in the following sequence: Introduction, analysis of the current situation, development of possible solutions; documentation. The models suggested are presented by the model validation team, containing a comparison of simulation results between the model of the Current State and the models suggested for the Future State. Subsequently, possible
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
complementary solutions are developed by kaizen team, and if there is new information from the improvement actions (e.g. reducing the operations cycle time), these are documented to be implemented in the computational model chosen for the Future State. x Define Future State: After the final definition of the model for the Future State, the changes suggested in the workshop are implemented in the computational model, as well as the new information from the improvement actions. Finally, the simulation results of the Future State are compared with the results for the Current State. Furthermore, the performance indicators are analyzed and presented.
x Modification of bench arrangement. The results on the machines usage in the current state and future state (F-S) were presented in terms of working, waiting and blocked rates, and are illustrated in Figure 2. A significant change stands out in the working rate allocated to OP20, which influenced the behavior of the subsequent operations.
5. Case Studies To check the feasibility, applicability, usability and limitations of MAPS, it was applied to corporations operating in the automotive manufacturing industry, using kaizen events to implement improvements in their processes. Computational models were produced using Tecnomatix Plant Simulation TM software and analyses of occupation of production resources were performed in a DataFit from the same application. The Case Study-A emphasizes the comparison between the results of the improvement modifications obtained through MAPS, with results obtained in a kaizen process without this aid. The Case Study-B shows a layout modification performed in the same company in a similar manufacturing cell, and highlights the use of historical data and best practices of kaizen simulated by MAPS. 5.1. Case Study A The system is a manufacturing cell (MC) that monthly produces 7152 auto parts machined in pairs (right and left) by six machines, and transported by five operators with the help of treadmills. There are three work shifts daily, totaling 15 operators per day (figure XX). The model validation and measurement teams were formed by the kaizen leader and the specific goals for the kaizen event were defined as:
Fig. 2. Production resources usage rates for current and future states (F-S).
The possibility of an operator removal from the manufacturing cell, previously found in kaizen, was ratified by the simulation results. It is observed that, in Figure 6, the transport time of operator 3 in operation 30 has decreased. This occurred due to the approach of benches and treadmills. As a result of an operator removal in the new layout, there was an increase in the waiting and blocked time rate for operations 40/50, inspection and cleaning, besides an increase in the transporting time rate of operator 4. At the end of the future state simulation, the productive capacity of the manufacturing cell increased by approximately 30%. The production efficiency of the operators also increased, as shown in Figure 3.
x Free up one operator for other activities; x Improve the productive capacity by 30 %. The schedule established by the model validation team lasted ten days between the phases Set Design and Analyze the Future State, forecasted in MAPS. After the steps of process mapping, data collection and identification of opportunities for improvement, a computational model was developed from the Current State, and then validated by the models validation team. The future state models were analyzed during the workshop referred in MAPS, and the changes made in the current state model of the MC, as shown in Figure 4, were: x Modification of material flow direction; x Modification of operations 20 and 30 sequence;
Fig. 3. Rates of parts manufactured monthly (current and future states).
Traditionally, at the end of the fifth day of kaizen events an analysis of benefits is performed to compare the current state to the estimated future state. This analysis is based on indicators of system performance and verifies the influence of the proposed changes. Furthermore, it serves to justify investment of resources. However, treatment of the time variable for the analysis of the MC’s resource behavior influences directly in the estimated results.
273
274
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
When using mean times for machine processing and operator handling, variations derived from randomness in the system behavior are not considered. Consequently, the results of the estimates for the MC’s future state may present significant differences. This means that it can estimate gains that are not real. This case study illustrates this type of occurrence, and confirms that the results obtained by computer simulation are closer to the behavior of the real system than those obtained by calculations in spreadsheets. The Table 1 details this aspect. The initial data of kaizen event, corresponding to the current state of the manufacturing cell, and also data of parts production for the future state are presented. The data called conventional were obtained by calculations in spreadsheets, the simulated data were obtained through the simulation process, using the method proposed in this paper. Table 1. Comparison of estimated capacity increase for future state. Production (parts/ month)
Amount of operators (H) x (3 shifts)
Productive capacity (parts/H month)
Productive capacity increase
Initial data
7152
15
476
-
Conventional
8333
12
694
46%
Simulated
7405
12
617
30%
It is observed that the results estimated by the conventional method, without simulation, indicate an increase in production capacity by 46% for the future state. If compared with the results estimated by MAPS, with the help of simulation, the increase in production capacity is expected to be 30%. Errors in the assessment of the future state can lead to misguided investments. Whereas the simulation minimizes these errors, the great difference in the results of this comparison shows the limitation of evaluation in decision making for changes in kaizen processes, when simulation tools are not used. For this case study, the difference in results can be considered punctual, as this work was carried out only for one manufacturing cell. However, the greater the complexity of the real system, in other words, the larger the number of components of the production system (manufacturing cells) and their interactions, the larger the errors caused by the random behavior of their resources (machines, operators, conveyors) will be, both in simulation processes and estimation by spreadsheets. Because it is a process of continuous improvement, restrictions and waste will be identified over time to increase the productive efficiency of the MC. The computational models used for this MC will serve as references for new changes in the same MC and for similar MCs, and the data acquired will remain in the database of simulated models. Thus, the next kaizen events may contain much more information, resulting in greater efficiency in the improvement process and faster kaizen event.
5.2. Case Study B The case study B was conducted in a manufacturing cell (MC) presented in Figure 4 which produces two types of similar automotive parts (Part type-A and Part type- B), machined in pairs. Parts type-B are produced by machines on the left, as represented in figure 8, while type-A parts are produced by the machines on the right. Both sides present the OP10 (Operation 10), OP20, OP30, OP40/50, OP60 and OP70. Just OP80, visual inspection, it is common to both. The MC has a total of 10 operators. The schedule established by the model validation team lasted 20 days and the specific objectives of the kaizen event were defined as: x Provide 30% of manpower for other activities; x Provide a machining center (OP20/30) to another sector of the company; x Increase production capacity by 40%.
Fig. 4. case study B – Current state layout.
Upon completion of the data collection activity, the current state's computational model is designed to be subsequently validated by the model validation team. Figure 10 shows a screenshot of this model. He assisted to visual verification of bottlenecks and its effects on the capacity, and served as a reference for the elaboration of the model of the future state. The proposed changes by MAPS workshop were modeled and simulated after the current state model validation. These changes are listed and illustrated in Figure 11, according to the numerical order shown below: 1. 2. 3. 4.
Remove a machining center of OP20 and OP30; Replace machine OP10 (type-A); Install mats on the side to reduce risk and beats; Modify machine placement of OP40 and OP50 to decrease movement of operators and allow an operator to occupy more than one workstation; 5. Insert the OP60 (inspection 1) and the OP70 (inspection 1) at OP80.
Victor Emmanuel de Oliveira Gomes and Luis Gonzaga Trabasso / Procedia CIRP 57 (2016) 270 – 275
A comparison between the current state and future state is presented in Table 2. At the end of kaizen, their goals were met. An observation made six months after modification layout indicated a decrease in the rate of scrap parts by 60%. Regarding the method MAPS, using data from computational models performed for the case study A helped in the development of algorithms used in computational models of the case study B. This reduced the time for development of computational models. Table 2. Productive capacity comparison between current state and future state.
Productive capacity (parts/H month)
Amount of operators (H)
Product Type
Current State
Future State
Gains
Type-A
417
701
67%
Type-B
436
552
26%
Type-A+ Type-B
853
1253
46%
Type-A+ Type-B
11
7
36%
6. Conclusions This work achieved the objective to systematize the application of simulation into continuous improvement process on the shop floor, using a method that formalizes simulation activities in conjunction with the kaizen tasks. The case studies highlighted the importance of the use of simulation techniques applied to support the continuous improvement process. The main positive factor of method MAPS is to increase the level of knowledge of the behavior of the analyzed system. The use of probabilistic data considers the random behavior of the real system and can avoid gross errors in estimating earnings improvements. The comparison of the results of kaizen with deterministic and probabilistic data, shown in case study 1, illustrates this situation. It is emphasized that historical data and computational models from MAPS can be used in future analyses. This fact minimize the time to develop new similar computational models to the same company. The access to information is the main limitation of MAPS. When the simulation analyst is not familiar with the real system analyzed, there is more time spent with the modeling process and the possibility of errors in this activity. Access to historical data on the operation of systems is an important factor for the time taken to validation activity and computational modeling. If the mapping of all the company's processes becomes a common practice, along with the creation of a data base (with
input and output data) for computational models, simulation processes to aid in modifying layouts will be faster and more efficient, and consequently improvement projects will be more effective. Acknowledgements The authors thank FAPEMIG (Minas Gerais State Foundation for Research Development) for the financial support. References [1] STANDRIDGE, C.R.; MARVEL, J.H. 2006. Why Lean Needs Simulation. In: Winter Simulation Conference, Piscataway, Institute of Electrical and Electronics Engineers, p. 1907-1913.2006. [2] OLIVEIRA, C. S. Aplicação de Técnicas de Simulação em Projetos de Manufatura Enxuta. Estudos Tecnológicos - Vol. 4, n° 3: 204-217. 2008. [3] WOMACK, J.P.; JONES, D.T. A mentalidade enxuta nas empresas. 3ª ed., Rio de Janeiro, Campus, 408 p. 2004. [4] Ibrahim A. Rawabdeh, (2005),"A model for the assessment of waste in job shop environments", International Journal of Operations & Production Management, Vol. 25 Iss: 8 pp. 800 – 822. [5] KUMAR, S., PHROMMATHED, P. Improving a manufacturing process by mapping and simulation of critical operations. Journal of Manufacturing Technology Management, 17(1), 104-132. 2006. [6] GRIMARD, C.; MARVEL, J.H.; STANDRIDGE, C.R. Validation of The Re-Design of a Manufacturing Work Cell Using Simulation. In: Winter Simulation Conference, Piscatawa, p.1386-1391. 2005. [7] BANKS, J.; CARSON, J.; NELSON, B. Discrete-event system simulation. New Jersey: Prentice Hall, 1996. [8] KELTON, W. D., SADOWSKI, R. P., SADOWSKI, D. A.Simulation with ARENA. 3 Ed, New York, McGraw-Hill Companies Inc. 2004. [9] KHALIL, R., KANG, P., STOCKTON, D. “Integration of Discrete Event Simulation with an Automated Problem Identification”, IMECS– (International Multi-Conference of Engineers and Computer Scientists), Hong Kong. 2010. [10] MIKA, G.L. Kaizen Event Implementation Manual, Kaizen Sensei, Wake Forest, NC. 2002. [11] DE CARLI, P.C.; DELAMARO, F.C.; SALOMON, V.A.P. Identificação e Priorização dos Fatores Críticos de Sucesso na Implantação de Fábrica Digital, Revista Produção. São Paulo. 2010. [12] HARREL, Charles R.; MOTT, Jack R. A.; BATEMAN, Robert E. BOWDEN, Royce G. GOGG, Thomas J. Simulação: Otimizando sistemas. 2. ed. São Paulo, SP: Instituto IMAM. 134 p. 2002. [13] BERTRAND, J. W. M.; FRANSOO, J. C. Modeling and simulation: operations management research methodologies using quantitative modeling. International Journal of Operations & Production Management, v. 22, n. 2, p. 241-264, 2002. [14] KUEHN, W. Digital factory: integration of simulation enhancing the product and production process towards operative control and optimization. International Journal of Simulation, v. 7, n. 7, p. 27-29, 2006. [15] SÁNCHEZ, A. M., PÉREZ, M. "Lean indicators and manufacturing strategies", International Journal of Operations & Production Management, Vol. 21 Iss: 11, pp.1433 – 1452, 2001.
275