Available online at www.sciencedirect.com
ScienceDirect Procedia CIRP 61 (2017) 481 – 486
The 24th CIRP Conference on Life Cycle Engineering
Simulation-based Assessment of Segmentation and Control Strategies within Multi-variant Productions Steffen Butzera*, Sebastian Schötzb, Andreas Krusea, Anna-Sophie Freytaga, Rolf Steinhilpera,b a
University of Bayreuth, Chair Manufacturing and Remanufacturing Technology, Universitaetsstrasse 9, 95447 Bayreuth, Germany b Fraunhofer Project Group Process Innovation, Universitaetsstrasse 9, 95447 Bayreuth, Germany * Corresponding author. Tel.: +49-921-78516-420; fax: +49-921-78516-105. E-mail address:
[email protected]
Abstract The increasing individualization of products is an enormous challenge for manufacturing companies. An increase of product variants and thus product complexity leads to increased process complexity within productions. An option to face this challenge is the segmentation of productions. Unfortunately, it is hardly possible to assess segmentation and the corresponding control strategies by using common methods, such as value stream mapping. Therefore, a simulation-based approach to assess segmentation and control strategies within multi-variant productions is shown in this paper. The approach is applied for a multi-variant production for medical equipment. Besides the development of the approach, the paper also shows the assessment results for the case study production. © Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ©2017 2017The The Authors. Published by Elsevier B.V. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the scientific committee of the 24th CIRP Conference on Life Cycle Engineering. Peer-review under responsibility of the scientific committee of the 24th CIRP Conference on Life Cycle Engineering
Keywords: Segmentation Strategies; Multi-variant Production; Material Flow Simulation; Lot Size Configuration; Sustainable Production
1. Introduction 2. State of the Scientific Knowledge and Need for Action Increasing globalization of the markets and increasing competition are two of today’s challenges for manufacturing companies. Besides increasing globalization and competition, also the individualization of products is an enormous challenge for manufacturing companies. Thus, the importance of strategic production planning, to increase the efficiency and profitability of productions, increases [1]. Nevertheless, cost efficiency respectively minimization is not the only target dimension, which has to be optimized. This leads to the conflicting goals, to manufacture customized products in the right quality, at the right time and to the right price [1]. An option to face these conflicting goals is the segmentation of productions. Segmentation of productions means the division of product variants and processes into small and transparent units. Thus, the conflicting goals are faced by renunciation of process orientated production structures towards product oriented production structures which are aligned with the customers’ demand [2].
2.1. Production Systems The aims of modern productions are based on the endeavor of companies to save and extend their market success. To do so, manufacturing companies have to face today’s trends and effects [3,4]. For productions, that means unsteady incoming orders, an increasing number of variants as well as the trend towards short dated orders, and at the same time shorter life cycles as well as faster technological progress [5,6]. Due to the variety of internal and external influencing factors, productions are subjected to intense performance and market complexity. To stay competitive and to be able to produce efficient, there is a trend towards holistic and comprehensive production systems [3,4,5]. A production system is described as a production organization, including all concepts, methods and tools. Those components influence the effectivity and efficiency of the production process through their interactions [7]. The advantage of (holistic) production systems is the ability to
2212-8271 © 2017 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 24th CIRP Conference on Life Cycle Engineering doi:10.1016/j.procir.2016.11.261
482
Steffen Butzer et al. / Procedia CIRP 61 (2017) 481 – 486
react targeted and continuously on market changes and technological progress [6,8]. The most noted production system is the Toyota Production System. Topics as wastage reduction, continuous improvement, standardization, teamwork, transparency etc. are the base for many modern production systems. Nowadays, some of the concepts and methods are assigned to Lean Production or Lean Management [4,9,10]. The application of a Lean Production enables the reduction of lead times as well as the increase of flexibility and planning certainty [8,9]. 2.2. Variant Management Variant management is described as the holistic approach to control the variety of variants [11]. For manufacturing companies, mastering the increasing variety of variants is one of the key tasks of the production planning and control (PPC). Wildemann distinguishes between three measures to be taken within variant management, which are: complexity prevention, complexity reduction, and complexity control [8]. Thus, the aim of variant management is the extensive prevention or reduction of complexity and the control of the remaining complexity. 2.3. Methods and Approaches for Production Optimization In this section, methods and approaches to analyze and optimize processes within multi-variant productions are described. Value Stream Methodology The aim of the value stream methodology is to analyze productions regarding wastages and to design lean material and information flows [12]. According to Erlach, the value stream methodology can be subdivided into the value stream analysis and the value stream design [1]. Digital Factory According to the VDI manual 4499, the term Digital Factory is described as the comprehensive network of digital models, methods and analysis tools, which are integrated through a consistent data management. Key element of the Digital Factory is the simulation of production and logistic processes. The aim of the Digital Factory is the continuous analysis, standardization and optimization of productions. 2.4. Production Planning and Control (PPC) Strategies Due to the high number of variants and the volatile customer demand, it is reasonable to control productions flexible and customer oriented. Fig. 1 illustrates the four strategies to control productions, according to Selke [14].
Fig. 1. Strategies to control productions, according to [14].
Lot Size Configuration Within the lot size configuration, it is defined, how many orders of one variant are pooled to one lot. The optimal lot size is oriented at the customer takt time respectively the order size as well as at the efficient usage of the production capacities. Besides the optimization of the starting lot size, there are also several strategies available, to change lot sizes within the production flow. Due to technical or organizational reasons, it can be useful to pool or separate lots before defined process steps. Furthermore, the parallel processing of lots on several machines can be useful. [14,15] Allocation of Resources Within the allocation of resources, the production processes are allocated to the available resources, according to defined criteria. Criteria to allocate resources are, among others: workload, set-up effort and process time. [14] Definition of Production Sequence The definition of the production sequence can be done according to certain rules, e.g. First In - First Out (FIFO). Other criteria to prioritize orders are the completion date, the shortest or longest total process time or the shortest set-up time. [14,15] Order Release Strategy The strategy to release orders defines when and where the production orders are released. Strategies are, among others, control according to manufacturing order, load oriented order release or control by the Kanban principle. The Kanban principle is a decentralized order release strategy which works according to the pull system. Thus, overproduction is avoided, as production resources produce only if there is a demand. [16] 2.5. Segmentation Strategies The segmentation of products and processes is a common strategy within variant management in order to manage complexity within productions. Especially in case of a high number of variants, segmentation can reduce cycle and set-up times and thus make productions more efficient. The segmentation of productions is done due to product and process specific criteria or due to demand and market oriented criteria. For each of the segments, it is necessary to develop an own production strategy. [8] The segmentation of a production leads to targeted division and aggregation of products (product segmentation) and processes (process segmentation). The segmentation is done based on detailed analysis of the variants spectrum, such as cluster or ABC-analysis. Criteria are, among others, similarities, quantity, customer demand, lot size, delivery times, lead times and set-up times. [2] Segmentations are useful, if the demand of single variants as well as the variation are predictable, intermediate-term and long term [2]. The conflicting goals of the segmentation are to reach a high utilization of resources, despite the division into product and process segments [2].
Steffen Butzer et al. / Procedia CIRP 61 (2017) 481 – 486
2.6. Material Flow Simulation According to the VDI manual 3633, the term simulation describes the replication of a system including its dynamic behavior in a model with the capability for experiments. The aim is to gather knowledge which is transferable into reality. [17] In general, simulations are useful, if experiments or changes in a real system are too cost or time expensive or not possible at all [17]. The term model is defined as the simplified replication of a planned or existing system including its processes. Deviations between the model and the real system are deliberate to reduce complexity. But, the deviations have to be kept in reasonable dimension, to be able to transfer the knowledge gathered in the model to the real system. [17] Simulation studies are the targeted analysis of the behavior of a model by repeating simulation runs with systematic parameter or structure variation [17]. In terms of PPS, the simulation of material flows is an effective and efficient methodology to assess capacitive and temporal correlations and their stochastic disturbance variables, and thus to assess strategies and optimization methods or existing processes [18]. The adjustment of control variables, e.g. process parameters, enable the comparison of various solution approaches. Thus, the methodology enables to determine the optimal solution without disturbing the running system. [19,20] To simulate production processes respectively material flows and to assess time-varying states, the discrete event simulation has been established [8]. Thus, simulation tools, such as Siemens PLM Plant Simulation, enable the simulation of scenarios, e.g. to identify bottlenecks in productions, and to illustrate the simulation results in diagrams [22,23]. 2.7. Need for Action Due to the variety of specific influencing factors, not all of the before mentioned optimizing strategies are convenient for all manufacturing companies and operations. Unfortunately, there is a lack of knowledge, both in industry and science, when it comes to the assessment of segmentation and control strategies, especially within multivariant productions. It is hardly possible to assess these strategies by using common methods, as value stream mapping, due to the missing possibility of considering dynamic processes and correlations. Therefore, in this paper a simulation-based approach to assess segmentation strategies within multi-variant productions is shown. The approach is applied for a multivariant production of medical equipment. Besides the development of the approach, the paper also shows the assessment results for the case study production. At the end of the day, the results of the paper will support companies to choose the right segmentation and control
483
strategy and thus to produce in a more resource efficient and sustainable way. 3. Research Approach The assessment of segmentation and control strategies was performed on a simulation based approach. This approach enables the consideration of dynamic processes and the targeted assessment of optimization concepts respectively strategies. As case study, a multi-variant production in the medical equipment industry sector was used. Examples for process models to perform simulation studies are quite common in scientific literature [17,21,24,25]. For this paper, the VDI manual 3633 [17] was used as a base for the simulation studies. The VDI process model is roughly structured into the three phases: Preparation, development and evaluation. The three phases are described in more detail in the following. 3.1. Preparation of the Simulation Study Within the preparation phase, the aims of the simulation studies were defined and the data base was generated and analyzed. The aims defined for the simulation studies were to decrease the set-up times and to increase the output of the case study production by implementing several segmentation and control strategies. The segmentation and control strategies were implemented at the bottleneck of the production, because it has the highest cycle time, compared to the other processes, and thus is crucial for the overall lead time. Furthermore, the lead times should be minimized and the utilization of the work stations be maximized. The data were gathered by performing a value stream analysis. Thus, an overview of the material and information flows was generated. In addition to that, further information as production orders, production scheduling etc. had to be analyzed. 3.2. Development of the Simulation Model The development of the simulation model was performed by using the Siemens PLM Software Plant Simulation, version 10.0.2. The data, which were available due to the value stream mapping and the data analyzes, were transformed into a dynamic simulation model. According to Eley, it is important to validate simulation models and the results, to ensure that the behavior of the model corresponds with the behavior of the real factory [25]. Therefore, the model was validated by comparing output data from the real factory with the output data from the simulation model, based on the production data from 2015. The focus of the simulation studies was the bottleneck of the production. Therefore, the bottleneck was identified, based on the value stream map respectively the operator balance chart (OBC). Fig. 2 illustrates the targeted development of the simulation model.
484
Steffen Butzer et al. / Procedia CIRP 61 (2017) 481 – 486
unbalanced utilization rates of the work stations respectively resources. By implementing segmentation strategies, a mixture of the options could be applied. 3.3. Experiments and Analysis Based on the developed and validated simulation model, simulation studies were performed. Simulation studies are based on the variation of input and disturbance variables and the analysis of effects on output variables. The simulation studies are planned and defined in an experimental design. In Plant Simulation, the tool to be used to generate an experimental design is termed experiment administrator. Table 1 illustrates the experimental design performed. As described before, each scenario was assessed for the three defined lot size configurations. Table 1. Experimental design. Strategy
Segmentation Process
Fig. 2. Targeted development of the simulation model.
Prod. sequence
Resources
no
no
no
Scenario Random
The bottleneck was developed in more detail as the other work stations. That led to time and cost savings within the simulation based approach. In the next step, the processing logic of the control strategies had to be implemented at the bottleneck. That was done with the scripting language SimTalk. The three segmentation strategies lot size configuration, allocation of resources and definition of production sequence were analyzed in detail. The order release strategy was not analyzed, due to the before mentioned targeted model development. The assessment of order release strategies would require the development of a model of the whole production in detail. The lot size configuration was assessed by setting two different lot sizes and the combination of both (different lot sizes for different process segments). The conflicting goals of the lot size configuration are, that a lot size above the customer demand leads to increasing costs for storage, whereas a lot size below the customer demand leads to increasing costs for production. By implementing segmentation strategies, a mixture of the two options could be applied. The allocation of resources was analyzed by comparing the fixed allocation of resources against the flexible respectively random allocation of resources. The conflicting goals of resource allocation are, that a fixed allocation allows the reduction of set-up times, whereas a random allocation is more flexible. By implementing segmentation strategies, a mixture of the options could be applied. The definition of production sequence was analyzed by comparing the set-up time oriented definition of the production sequence against a FIFO oriented definition of the production sequence. The conflicting goals of the definition of production sequence are, that the set-up times can be reduced by the set-up time oriented definition of the production sequence, whereas the capacity respectively the utilization rate of the resources is not considered. Thus, this can lead to
Allocation of resources
no
Control
Product
Random definition of machine with free capacity. no
no
no
yes
yes
no
Defined machine per variant.
Definition of production sequence
Definition of machines based on the set-up time.
Product segmentation 2
Segmentation based on the number of variants.
Product segmentation 1
Segmentation based on the customer demand.
Product segmentation combination
no
yes
yes
yes
no
yes
yes
yes
no
yes
yes
yes
no
no
Segmentation based on the number of variants and the customer demand.
The output variables considered for this paper are the output of the production, the set-up times and the utilization rate of the bottleneck work station. Due to the dynamic of the simulation model, it was necessary to perform several (20) simulation runs for each simulation study. The influence of the stochastic input parameters is illustrated in charts including confidence intervals. In the last step, the results of the simulation studies were analyzed. 4. Results and Appraisal of the Simulation Study In this section, the results of the case study based simulation study and the appraisal of the case study based simulation study are described. 4.1. Results of the Simulation Study Following, the results of the simulation based assessment of segmentation and control strategies are described.
Steffen Butzer et al. / Procedia CIRP 61 (2017) 481 – 486
Fig. 3 illustrates the results of the first simulation study. The segmentation and control strategies were assessed regarding the number of products, which were produced in 220 working days exclusive 30 days of settling time. The three lines illustrate the results for the three lot size strategies which had been implemented (lot size 12, 4 and a combination of both: lot size 12 for variants with a high customer demand and lot size 4 for variants with a low customer demand).
485
Fig. 4 illustrates the simulation results for the combined lot size strategy including confidence intervals. It can be seen that especially the strategies allocation of resources and product segmentation 1, which are oriented at the customer demand, had strong variations. The reason for that is the dependence on stochastic variables, such as variant sequence etc. Fig. 5 illustrates the different time slices depending on the strategy applied. On the one hand, the set-up times could be reduced, on the other hand, the working times were also reduced when some of the segmentation and control strategies were applied. The reason for that are the already described conflicting goals of the segmentation and control strategies. Segmentation and control strategies are based on forecast values. The variation of these values can lead to increasing waiting times, especially in case of allocated resources.
Fig. 3. Number of products depending on the segmentation and control strategy.
It can be seen that some of the strategies implemented lead to a lower output compared to the random strategy. The reason for that are the lower utilization rates of the work stations. Especially the allocation of resources cannot keep up with the random definition of machines. Also the reduction of the lot size does not increase the output. The reason for that is the relatively high time slice for set-ups. Therefore, not all of the implemented segmentation and control strategies were useful within the case study production.
Fig. 5. Time sliced depending on the segmentation and control strategy.
To prove the robustness of the applied strategies, a second simulation study with a different set of variations respectively different numbers of products per variant was performed. Fig. 6 illustrates the results of this simulation study. It can be seen that the strategic segmentation and application of control strategies is only target-aimed if the customer demand is stable. If the customer demand is unstable, increasing waiting times are arising.
Fig. 4. Number of products depending on the segmentation and control strategy including confidence intervals. Fig. 6. Robustness of the segmentation and control strategies.
486
Steffen Butzer et al. / Procedia CIRP 61 (2017) 481 – 486
4.2. Appraisal of the Case Study based Simulation Study It could be proved that the material flow can be optimized by the application of segmentation and control strategies. Especially the bottleneck within multi-variant productions with increased effort for set-ups can be optimized. Thus, the utilization rate can be increased and the cycle times can be reduced. Nevertheless, the case study based simulation study also showed that the strategic application of segmentation and control strategies require certain basic conditions. The demand of the single variants as well as the variation has to be predictable. If not, increasing waiting times can occur. Besides that, it has to be mentioned that the unique characteristic of the case study production inhibited “better” results of the segmentation and control strategies. The creation of the variants were done at the bottleneck station. Furthermore, the bottleneck station is the first process step within the production. Thus, the variety of variants could not be avoided respectively managed by the application of segmentation and control strategies. Therefore, the potential of improvement of the segmentation and control strategies could not be proved by the case study based simulation study entirely. 4. Conclusion and Outlook The paper showed a simulation-based assessment of segmentation and control strategies within multi-variant productions. The main result of the paper is the successfully performed simulation based approach to assess segmentation and control strategies within multi-variant productions. Furthermore, it was shown that the application of segmentation and control strategies respectively their combination can reduce set-up times, increase the utilization rate of work stations, and increase the output of productions. In future research, the segmentation and control strategies will be applied and assessed in more case study productions. Thus, the research results, both, regarding the simulation based simulation approach and regarding the segmentation and control strategies, will be strengthened and validated. References [1] Erlach K. Wertstromdesign. Der Weg zur schlanken Fabrik. 2nd. Edition. Berlin, Heidelberg: Springer-Verlag; 2010. [2] Grundig C-G. Fabrikplanung. Planungssystematik, Methoden, Anwendungen. 4th Edition. München: Carl Hanser Verlag; 2013. [3] Abele E, Reinhart G. Zukunft der Produktion. Herausforderungen, Forschungsfelder, Chancen. München: Carl Hanser Verlag; 2011. [4] Spath D. Produktionsarbeit der Zukunft – Industrie 4.0. FraunhoferInstitut IAO. Stuttgart: Fraunhofer-Verlag; 2013. [5] Bauernhansl T. Die Vierte Industrielle Revolution – Der Weg in ein wertschaffendes Produktionsparadigma. In: Bauernhansl T, ten Hompel M, Vogel-Heuser B. editors. Industrie 4.0 in Produktion, Automatisierung und Logistik. Anwendung, Technologie, Migration. Wiesbaden: Springer Vieweg; 2014. p. 5-36. [6] Westkämper E, Zahn E. Wandlungsfähige Produktionsunternehmen. Das Stuttgarter Unternehmensmodell. 1st Edition. Berlin, Heidelberg: Springer; 2009.
[7] Schuh G, Gierth A. Grundlagen der Produktionsplanung und –steuerung. Aachner PPS-Modell. In: Schuh G. editor. Produktionsplanung und – steuerung. Grundlagen, Gestaltung und Konzepte. 3rd Edition. Berlin: Springer; 2006. p.11-26 [8] Wildemann H. Entwicklungslinien der Produktionssysteme in der Automobilindustrie. In: Göpfert I, Braun D, Schulz M. editors. Automobillogistik. Stand und Zukunftstrends. 2nd Edition. Wiesbaden: Springer Gabler; 2013. [9] Boppert J, Durchholz J, Günthner W A. Lean Logistics im Wandel – neue Aufgaben, Partner und Rahmenbedingungen. In: Günthner W A, Boppert J. editors. Lean Logistics. Methodisches Vorgehen und praktische Anwendung in der Automobilindustrie. Berlin, Heidelberg: Springer Vieweg; 2013. p. 27-34. [10] Jodlbauer H. Produktionsoptimierung. Wertschaffende sowie kundenorientierte Planung und Steuerung. 2nd Edition. Wien, New York, Heidelberg: Springer; 2008. [11] ROI Management Consulting AG. Variantenmanagement. In: ROI Consulting AG. editors. Management http://www.logistiklexikon.de/lexikon/. München. 2014. [12] Becker T. Prozesse in Produktion und Supply Chain optimieren. 2nd Edition. Berlin, Heidelberg: Springer-Verlag. 2008. [13] VDI Verein Deutscher Ingenieure. Richtlinie 4499: Digitale Fabrik – Grundlagen. Berlin: Beuth Verlag; 2008. [14] Selke C. Entwicklung von Methoden zur automatischen Simulationsmodellgenerierung. In: Zäh M, Reinhart G. editors. Forschungsberichte iwb, Band 193. München: Herbert Utz Verlag; 2005. [15] Kiener S, Meier-Scheubeck N, Obermaier R, Weiß M. ProduktionsManagament. Grundlagen der Produktionsplanung und –steuerung. 10th Edition. München: Oldenbourg Wissenschaftsverlag; 2012. [16] Garrel J, Seidel H, Schenk M. Flexibilisierung der Produktion – Maßnahmen und Status-Quo. In: Schlick C M, Moser K, Schenk M. Flexible Produktionskapazität innovativ managen. editors. Handlungsempfehlungen für die flexible Gestaltung von Produktionssystemen in kleinen und mittleren Unternehmen. Berlin, Heidelberg: Springer Vieweg; 2014. p. 81-126. [17] VDI Verein Deutscher Ingenieure: Richtline 3633: Simulation von Logistik-, Materialfluss- und Produktionssystemen – Begriffe. Berlin: Beuth Verlag. Berlin; 2010. [18] Gierth A, Schmidt C. Zeitdynamische Simulation in der Produktion. In: Schuh G. editor. Produktionsplanung und –steuerung. Grundlagen, Gestaltung und Konzepte. 3rd. Edition. Berlin: Springer; 2006. p. 646681. [19] März L, Wilfied K. Kopplung von Simulation und Optimierung. In: März L, Krug W, Rose O, Weigert G. Editors. Simulation und Optimierung in Produktion und Logistik. Praxisorientierter Leitfaden mit Fallbeispielen. Heidelberg: Springer (VDI-Buch); 2011. p. 41-46. [20] März L, Weigert G. Simulationsgestützte Optimierung. In: März L, Krug W, Rose O, Weigert G. editors. Simulation und Optimierung in Produktion und Logistik. Praxisorientierter Leitfaden mit Fallbeispielen. Heidelberg. Springer (VDI-Buch); 2011. p. 3-12. [21] Wenzel S, Weiß M, Collisi-Böhmer S, Pitsch H, Rose O. Qualitätskriterien für die Simulation in Produktion und Logistik. Planung und Durchführung von Simulaitonsstudien. Berlin: Springer; 2008. [22] Rose O, März L. Simulation. In: März L, Krug W, Rose O, Weigert G. editors. Simulation und Optimierung in Produktion und Logistik. Praxisorientierter Leitfaden mit Fallbeispielen. Heidelberg: Springer (VDI-Buch); 2011. p. 13-20. [23] Siemens Industry Software GmbH & Co. KG. Plant Simulation. http://www.plm.automation.siemens.com/de_de/products/tecnomatix/pla nt_design/plant_simulation.shtml. Köln, 2014 [24] Bangsow S. Praxishandbuch Plant Simulation und SimTalk. Anwendung und Programmierung in über 150 Beispiel-Modellen. München. Carl Hanser Verlag; 2011. [25] Eley M. Simulation in der Logistik. Einführung in die Erstellung ergebnisdiskreter Modelle unter Verwendung des Werkzeuges “Plant Simulation”. Berlin, Heidelberg: Springer; 2012.
