Towards a generic design method for reconfigurable manufacturing systems

Journal of Manufacturing Systems 42 (2017) 179–195

Contents lists available at ScienceDirect

Journal of Manufacturing Systems journal homepage: www.elsevier.com/locate/jmansys

Towards a generic design method for reconfigurable manufacturing systems Analysis and synthesis of current design methods and evaluation of supportive tools Ann-Louise Andersen a,∗ , Thomas Ditlev Brunoe a , Kjeld Nielsen a , Carin Rösiö b a b

Department of Mechanical and Manufacturing Engineering, Aalborg University, Fibigerstraede 16, 9220 Aalborg East, Denmark Department of Industrial Engineering and Management, Jönköping University, Gjuterigatan 5, 551 11 Jönköping, Sweden,

a r t i c l e

i n f o

Article history: Received 18 April 2016 Received in revised form 26 September 2016 Accepted 28 November 2016 Keywords: Reconfigurable manufacturing system Reconfigurable manufacturing design Changeability RMS Literature review

a b s t r a c t In today’s global manufacturing environment, changes are inevitable and something that every manufacturer must respond to and take advantage of, whether it is in regards to technology changes, product changes, or changes in the manufacturing processes. The reconfigurable manufacturing system (RMS) meets this challenge through the ability to rapidly and efficiently change capacity and functionality, which is the reason why it has been widely labelled the manufacturing paradigm of the future. However, design of the RMS represents a significant challenge compared to the design of traditional manufacturing systems, as it should be designed for efficient production of multiple variants, as well as multiple product generations over its lifetime. Thus, critical decisions regarding the degree of scalability and convertibility of the system must be considered in the design phase, which affects the abilities to reconfigure the system in accordance with changes during its operating lifetime. However, in current research it is indicated that conventional manufacturing system design methods do not support the design of an RMS and that a systematic RMS design method is lacking, despite the fact that numerous contributions exist. Moreover, there is currently only limited evidence for the breakthrough of reconfigurability in industry. Therefore, the research presented in this paper aims at synthesizing current contributions into a generic method for RMS design. Initially, currently available design methods for RMS are reviewed, in terms of classifying and comparing their content, structure, and scope, which leads to a synthesis of the reviewed methods into a generic design method. In continuation of this, the paper includes a discussion of practical implications related to carrying out the design, including an identification of potential challenges and an assessment of which tools that can be applied to support the design. Conclusively, further areas for research are indicated, which provides valuable knowledge of how to develop and realize the benefits of reconfigurability in industry. © 2016 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

1. Introduction It is widely recognized that changes are inevitable in today’s global market place. Customer demands are becoming more and more dissimilar, the need for customization of product offerings is increasing, and the pressure for rapid new product introductions is growing [1]. In a recent study, it was indicated that the product variety offered to customers has been more than doubled between 1997 and 2012, while product lifecycles have been shortened by

∗ Corresponding author. E-mail addresses: [email protected] (A.-L. Andersen), [email protected] (T.D. Brunoe), [email protected] (K. Nielsen), [email protected] (C. Rösiö).

about 25% [2]. The reasons for this are manifold, e.g. the emergence of different regional market segments, competition among companies to attract customers, and the emergence of new materials and technologies to incorporate in product offerings [3]. Thus, manufacturing companies need to find solutions to efficiently handle fluctuating volumes, customization, and frequent introductions of new variants and generations, in order to remain competitive in the global marketplace. Traditional manufacturing systems, such as dedicated and flexible systems, have major drawbacks in terms of meeting these requirements, as they do not offer adequate responsiveness at a reasonable cost [4]. Dedicated manufacturing lines optimized for one single product and output capacity have in many cases proved inappropriate, as variety, shortening product lifecycles, and demand

http://dx.doi.org/10.1016/j.jmsy.2016.11.006 0278-6125/© 2016 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

180

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

Fig. 1. Publications per year on reconfigurable manufacturing from 1999 to 2015.

changes simply cannot be met with its rigid structure [5]. This is likely to result in the situation where the dedicated lines operate below full capacity and become obsolete rather quickly [6]. On the other end of the continuum, the flexible manufacturing system offers the ability to be converted between the production of different variants, but has in its implementation brought issues of excess flexibility, low production rate, and low return on investments [7,8]. For these reasons, the concept of the reconfigurable manufacturing system (RMS) was introduced in the 90’s, combining the high throughput of traditional dedicated manufacturing lines and the flexibility of the flexible manufacturing systems [5]. The RMS is designed for rapid change through the ability to repeatedly change capacity and functionality in a cost-efficient way, in order to meet different demand situations, in terms of variation in volume as well as in product characteristics [9]. This is enabled through six core characteristics: customization, convertibility, scalability, modularity, integrability, and diagnosability. Customization refers to machine and system flexibility being limited and customized to a specific part or product family, which reduces the traditional trade-off between efficiency and flexibility [4,8]. Convertibility and scalability refer to modifying the capacity and functionality of the system and the machines, which is achieved through modularity and integrability. The last characteristic, diagnosability refers to the ability to read the state of the system and obtain information on which corrections that have to be carried out in order to reach the planned performance, which is particularly important in the ramp-up phase after each reconfiguration. With these characteristics, the RMS is adaptable to changing market conditions, and allows for cost-efficient reuse and prolonged lifetime, which is the reason why it has been widely labelled the manufacturing paradigm of the future [10,11].

1.1. Design for reconfigurability Since the introduction of the RMS concept in the late 90’s, research in the area has increased and broadened notably. The development in number of publications in terms of journal papers, conference proceedings, and books is depicted in Fig. 1, where the annual number of publications is depicted from 1999 where Koren et al. [5] initially coined the RMS concept. The figure includes publication results in English from search terms “reconfigurable” and “manufacturing” or “production” in Scopus, covering relevant subject areas, primarily being engineering, mathematics, decision science, computer science, business, and economics.

These publications on RMS cover multiple research issues and structuring levels of the factory, from the highest level, being the network and the factory, to the lowest structuring level being the workstation and tooling [12]. In general, reconfigurations can be divided in physical and logical types, which is also reflected in the different research issues. Physical reconfigurations involve hard changes in equipment and arrangement of machines, which usually requires changes on lower structuring levels, such as the workstation or the cell. Among dominant research issues on this level is the development of reconfigurable machines covering both tools, fixtures, inspection machines, and material handling systems [13]. These reconfigurable machines have modular structures, which enables quick conversion between different parts within a part family as well as in working speed or volumes [14]. On the other hand, logical reconfigurations involve soft changes such as rerouting and re-planning and are mainly related to research issues on system level or higher levels [12]. On the system level, reconfigurability is achieved by adding, removing, or changing the modules of the system, thereby changing the functionality or the capacity [9]. Research issues on system level include optimal reconfiguration selection [15–17], economic justification models [18,19], and system design methods [17,20–22]. Despite the critical importance of all these research issues, one particularly important concern is the design of the RMS, as this issue precedes all of the remaining and directly affects the system’s abilities to scale capacity and convert functionality. Manufacturing system design can be regarded as the task of mapping requirements into the physical system description [23]. Usually, two approaches to this can be taken; creating entirely new systems adhering to the requirements, or considering already existing systems and modifying these to fit the new requirements. In practice, the first approach is rarely taken, as the existing systems are often taken into consideration when designing new systems, but can be relevant for newly started enterprises or existing enterprises going into entirely new products or technologies [24]. When designing reconfigurable manufacturing, both approaches are relevant, as companies may transition towards reconfigurability from either more dedicated or more flexible system types without completely replacing these. In this regards, reconfigurability can be seen as being placed on a continuum between general flexibility and dedication, which is depicted in Fig. 2. In some cases, it may prove most feasible to rebuild existing dedicated systems, and gradually build increasing reconfigurability by changing some parts of the system.

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

Fig. 2. RMS as an intermediate paradigm between DMS and FMS [9].

However, the task of modifying existing systems does not change the design task considerably. It rather contributes with setting objectives for design, e.g. in terms of the breadth of choices in the design, but essentially the same decisions and activities should be conducted [25]. Thus, transitioning towards reconfigurability by changing either too flexible or too rigid existing systems essentially involves the same design challenges as designing entirely new manufacturing systems; determining the degree of functionality and capacity for the design. Whether it is in regards to developing entirely new reconfigurable systems or incorporating reconfigurability in existing systems, the requirements posed on the RMS may change significantly during its lifetime, which means that design becomes an even more challenging task than in the case of conventional manufacturing systems [21,26]. In order to meet this challenge, it is first of all necessary to apply a long-term view of the manufacturing system, in order to secure the economic feasibility for multiple product generations and market situations [27]. In other words, the system should be designed to be changeable in functionality and capacity. Secondly, the RMS design is closely integrated with the product portfolio, as the system should be able to produce multiple product variants within a product family, but also different generations of the same product family [27]. Finally, there is a clear interrelation between designing the system and planning the system configurations in its lifetime, as the initial design process determines the window of opportunities for the following reconfigurations [13,26]. Previous research indicates that conventional manufacturing system design methods do not support the design of RMS, as they do not consider reconfigurability characteristics and meet the specific challenges indicated above [20,21]. Numerous authors indicate that a systematic design method for RMS is lacking [13,20,28–31]. For instance, Bi et al. [13] present a review of methods for architecture design, configuration design, and control design in an RMS and conclude that a systematic design method is missing. Moreover, in current research, the actual successful implementation of reconfigurable manufacturing is yet to be widely reported. In addition to this, some researchers investigated the attitudes and barriers towards reconfigurability in industry and conclude that various misconceptions and a general lack of knowledge about reconfigurability is present [20,32] and that various critical barriers exist in regards to its implementation [29,33]. Therefore, the importance of the development of a systematic design method for RMS should be emphasized, as it provides practitioners with the necessary knowledge of how to develop and realize the benefits of reconfigurability. 1.2. Research questions The research presented in this paper falls in the category of design methodology, which is a specific aspect of design research

181

that covers creating knowledge of methods in design, in terms of structure of thinking and actions, methods and techniques used in the process, and concepts and corresponding terminologies [34]. According to Roozenburg and Eekels [34], there are two fundamental concerns in design methodology; answering what the structure in design is, which is descriptive methodology, and answering how the design process should be carried out in terms of which approaches to apply, which is prescriptive methodology. When considering previous research’s indications of limited support for designing RMS, the limited evidence of industrial application, and the significant increase in publications on reconfigurable manufacturing in the latest years, it is of high importance to approach both descriptive and prescriptive design methodology. This calls for investigating the current state-of-the-art in reconfigurable manufacturing, in terms of identifying design methods and comparing their content, structure, and approach in order to systematically identify a generic design method for RMS. Moreover, identifying and investigating procedures and supportive tools for the design is a critical element, in terms of being able to actually realize the benefits of reconfigurability in industry. Therefore, the following research questions are addressed in the paper: • Which steps and phases make up the structure of the design process for an RMS? • Are current contributions on RMS design opposing or supplementary, and can a generic design method be recognized? • Which supportive tools and procedures could be applied in order to support practitioners in carrying out the RMS design? Addressing these research questions is of both academic and practical relevance. First, in order to aid the transition towards the development of reconfigurability in industry and support practitioners in realizing its benefits, it is a critical step to increase knowledge of how to conduct the design process and which approach to apply, as there are currently multiple answers to this in research. The comparison and synthesis of these current design contributions is therefore an interesting research aspect. Moreover, thoroughly examining current contributions and indicating further needs for research represents a valuable addition to the growing body of literature on RMS. The remainder of the paper is structured as follows: Section 2 presents the method applied for addressing the research questions, including literature retrieval, exclusion, and classification. Section 3 presents the review of design methods for RMS, with focus on classifying them in accordance with their structure and content. In Section 4, a comparative analysis of the reviewed methods is presented, based on their main design elements, design approaches, structures, and differentiating characteristics. In Section 5, a synthesis of the reviewed design methods is identified, which results in a generic design method for RMS. With this synthesized method as the foundation, practical implications, specific procedures, and supportive tools for the design are discussed in Section 6. Conclusively, the findings are summarized and areas for future research are proposed in Section 7. 2. Research method In order to answer the research questions, a thorough and systematic review of current literature is required. Moreover, as the three research questions represent a descriptive and prescriptive approach to examining manufacturing system design, both analysis and synthesis of current literature is involved in the review. Initially, current design methods are analysed and classified in terms of describing their structure and approaches, which forms the foundation for synthesizing their common elements into a

182

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

Table 1 Search Protocol. Filter Type

Description

Results

Inclusion criteria

Search terms: “reconfigurable manufacturing system” AND “design” in title, abstract, or keywords. Sources: Peer-reviewed journals; conference proceedings; and book chapters Language: limited to English Initial exclusion of publications based on screening of abstracts. Exclusion of papers that are not relevant due to treating only one structuring level, primarily tool design, or due to addressing only one design aspect, e.g. optimal configurations, formation of product families for reconfigurations, finding optimal scaling policy, reconfigurable production planning, performance optimization, or machine selection. Second and final exclusion of publications based on screening full publications. Exclusion based on similar criteria as previous step, which excluded publications that only address modelling or logical aspects of reconfigurable manufacturing. Search for additional articles by snowball approach based on previously found articles, e.g. by identifying relevant references and further publications from authors or including publications found in previous searches on the topic. Similar inclusion criteria applied as in the initial and final exclusion. Consolidation of publications from systematic search and snowball approach.

Retrieved from Search: 235 publications

Initial exclusion

Final exclusion

Snowball approach and exclusion

Consolidation

generic design method for reconfigurable manufacturing. This part of the research is based on a classification of design methods into two different types: phase methods and design cycles, which is based on similar distinction in the area of product design [34] and software development [35]. This classification is conducted in order to address the question of whether the current contributions are opposing or supplementing. Moreover, the different procedures or supportive tools that are currently suggested are evaluated and potential future research directions are identified. 2.1. Literature retrieval and exclusion The process of collecting publications to include in the research is carried out by first retrieving publications from searching literature databases, and thereafter excluding publications that are not relevant. In this sense, relevance refers to publications that present a design method or framework for reconfigurable manufacturing. This is built on a general definition of a design method proposed by Roozenburg and Eekels [34], where a design method is defined as a description of a working procedure that in some way guide and support practitioners through the steps of developing a manufacturing system, providing knowledge of the structure of decisions and actions to take, methods and techniques to use, and concepts and terminology to apply. This definition allows a method to have varying degrees of detail and completeness in terms of specific tools and practises to include within each step. However, the common denominator of design methods is that they represent a complete structure with essential steps or elements to conduct when designing the RMS. The retrieval process is based on a primary search in the citation and abstract database Scopus. The applied search consists of a search in title, abstract, and keywords with two blocks being “reconfigurable manufacturing system” AND “design”. Publications that appeared in peer-reviewed journals, conference proceedings, and book chapters were included. The exclusion of papers is conducted in two steps. An initial exclusion is conducted by examining the abstracts of the retrieved publications, and excluding those that are not relevant, as they do not propose a design method or framework adhering to the above definition. The excluded papers were primarily papers that dealt with reconfigurable tooling, papers that dealt solely with one design problem without consideration to more comprehensive design framework or method, e.g. finding optimal configurations, formation of product families for reconfigurations, finding optimal scaling policy, reconfigurable production planning, performance optimization, or machine selection. The ini-

After exclusion: 58 publications

After exclusion: 12 publications

After exclusion: 9 publications

Final sample: 21 publications

tial exclusion served primarily as an initial screening of papers that appeared relevant for further review. The second and final exclusion was based on screening the full publications. The exclusion at this point was once again conducted based on the criteria that the publication should present a design framework or method. Therefore, at this point, publications that treated only the logical design of the RMS were excluded, e.g. designing control systems, as well as publications that solely treated the issue of how to model the RMS through petri net modelling or multi-agent modelling. The publications remaining from the second screening, where subjected to a snowball approach that further revealed publications to include. In addition to these searches, experts in the domain were identified through citation records and reference lists, which provided an additional number of papers to the review. These additional papers were subjected to the same exclusion criteria as the initial body of literature, but an addition of nine papers remained. The results of the literature search are summarized in the search protocol in Table 1. It should be noted, that the final sample of publications for the review not necessarily corresponds to an equal number of design methods, as some design methods are treated in multiple consecutive publications by the same group of authors. Moreover, in terms of addressing the last research question and evaluating supportive tools, a larger number of publications are considered from the initial retrieval, as a main part of these as argued presents parts of the entire design process.

3. Review of design methods for reconfigurable manufacturing Design processes are usually conceptualized either as cyclic problem solving methods or as a sequence of decisions, denoted as phased methods [24,36]. The distinction between these two aspects of design have proven valuable in being applied to distinguishing between methods within areas of product design and software engineering [35]. Table 2 presents a comparison between the two types of design methods based on Roozenburg and Eekels [34] and Hall [36]. In practice, design processes contain both types of design aspects [36], which means that this classification can be applied in order to investigate how current contributions on RMS design differ and whether they are opposing or supplementary. The classification is based on an evaluation of whether the methods merely indicate a cycle of steps that can be applied to solve a specific phase of the design process, in other words explains the logic of steps in design, or indicates a sequence or process of decisions and activi-

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

183

Table 2 Comparison of phased and cyclic design methods. Characteristic

Phased Design Methods

Cyclic Design Methods

Process Constituents

Sequential Phases comprising groups of related activities in-between milestones Can be applied to entire design task Flow of time in design Increased concreteness of design through stages Specification, concept development, preliminary design, detailed design, and implementation

Cyclic Steps in problem solving

Application Essential feature Progression Structure

ties the designer has to conduct, offering something to hold on to in the design process. 3.1. Phased design methods for reconfigurable manufacturing systems In phased methods for design, focus is primarily on the structural aspects related to the design process, where the design is described through different sequential phases that represent groups of related activities that lead to a certain stage of the design [40]. In these methods, the design exists on different abstraction levels, where the phases are increasingly concrete regarding the solution. Such phase methods are well known from product design and conventional production system design [24,25], but the exact terminology, number of phases, and number of intermediate solution levels often vary greatly. Depending on terminology, the structures of phase methods often correspond roughly to phases of design specification, concept development, preliminary design, and detailed design [24]. The approach to RMS design by Rösiö et al. [21,37,38] is based on a design method from conventional system design. It takes point of departure in a stage-gate method for manufacturing system development from Bruch [39] that includes initiation, preparatory design, and detailed design. In addition to this, support for considering reconfigurability is incorporated. An essential part of this is support for how to determine the need for reconfigurability in the system and the link between these needs and the generic RMS characteristics. The design method by Schuh et al. [40] is founded on the distinction between complex and complicated system elements based on predictability and number of elements and interdependencies of the system. The design approach is composed of four steps based on object-oriented design. The first step addresses the identification and classification of change drivers related to the product, volume, and technology. These change drivers are to be classified in regards to how and when the system is required to change, and how predictable they are. The second step focuses on describing the manufacturing system through change profiles that specify how the properties of the system element change with the change drivers. In the third and fourth step, interdependencies are determined between elements, and modules are created with the aim of separating elements that do not change in the same time and for the same reason. Likewise, Heisel and Meitzner [32] propose a sequence of design steps that should be carried out when designing reconfigurable manufacturing systems. The steps range from identifying reconfigurability requirements, time and cost efforts, and reconfigurability extent to the design of modules and their interfaces. Bi et al. [13] define three types of design issues in RMS that should be addressed in a design process. The first is the architecture design, where the system’s components and interactions are determined. The aim of this phase is to define a modular system architecture that can produce as many variants as possible, in order to deal with change and uncertainty. The second design issue is the configuration design, where the configurations for specific tasks are

Can be repeated in successive phases of design Flow of logic in design Increased knowledge of design through cycles Analysis, synthesis, simulation, evaluation, and decision

selected for the operating periods. This design is carried out within the range of available configurations specified by the modular system architecture. The last design issue is the control design, where process variables are determined in order for the configuration to fulfill its given task satisfactorily. Even though these three types of design issues are suggested mainly as a framework for classifying literature on RMS, they do still represent a structured view on RMS design, including phases of design activities that should be carried out together. The method proposed by Deif and ElMaraghy [41] apply similar terminology. They propose an RMS design architecture containing two modules; a design module and a control module divided into three layers. In the design module, the different design activities are described, while the control module describes the control of the design process at each level based on high-level strategic objectives and limitations. The first layer of the design framework is a market capture layer, where the requirements for capacity and functionality are determined based on different market profiles. The second layer, the system-level reconfiguration layer generates system alternatives or configurations based on the required levels of capacity and functionality. These alternatives are described in terms of hard, soft, and human aspects. Moreover, a selection of the most feasible configuration is also included in this layer through evaluating different performance objectives. The last layer is the component-level reconfiguration layer, where the implementation of the reconfiguration in the components of the system is addressed. Both the design methods proposed by Bi et al. [13] and Deif and ElMaraghy [41] indicate not only different design phases, but also repetition and feedback within the sequential structure. This feedback covers a repetition of selecting configuration for a period and doing the control design or component configuration. This structure essentially shares the same idea as design cycles, which is explained further in the subsequent section. However, the reason why these contributions have not been classified as merely design cycles, is that they do not carry out cycles of solution development and evaluation, but rather select a solution concept and step-wisely increase the level of detail. In product design, two types of phase methods can usually be recognized, which are methods that cover solely the design, and methods that also include the full development process, including phases related to commercialization, production development, and marketing [34]. A similar distinction can be made in the area of production design, where design methods cover solely the design activities, and the development methods cover both strategic planning, design, implementation, and ramp-up [24]. In respect to the design of production systems, phased development methods can be argued as being methods that include phases of design, implementation, and reconfigurations in an integrated way, which can be argued as being clearly represented by Tracht and Hogreve [26]. In their method, focus is explicitly on the interrelation between design and planning of the RMS. Moreover, their work presents strong focus on which decisions that have to be made in both the design and the planning phase of the RMS. Both the design and

184

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

the reconfiguration phases are elaborated through a number of different decision steps to guide the designers and planners of the system. These procedures are a combination of conventional system design steps and new procedures for modular systems, such as clustering operations and defining the degree of modularity. In the operating time of the designed system, reconfigurations are carried out in response to variant changes, capacity changes, and product changes. In the framework, the decision steps for each of the types of configurations are elaborated and differentiated for short, medium, and long-term horizons. In the work by Bryan et al. [42–44], a similar clear distinction is made between initial design of the first generation system, and the continuous design of further generations in the lifetime of the operating system. In their initial work, [43], two phases of design are considered: initial clean sheet phase with design of both the system and the product family and a co-evolution phase with reconfigurations. In the initial design, a profit optimization approach is applied in order to concurrently decide which products to include in a family, which variants to produce, the need of workstations, and assignment of tasks to workstation. In the reconfiguration phase, it is considered how to incorporate new needs in the product family and how to make changes in the assembly line. In later works, the authors further enhance this method by addressing the problem of how to concurrently design product families and assembly systems [44] and how to design assembly system configurations for evolving product families [42]. The common feature in these contributions is that the design is approached through optimization. 3.2. Design cycles for reconfigurable manufacturing systems In a design cycle or a problem solving process, different steps of problem solving are structured in a cycle [45]. Essential characteristics of this type of design structure is that a solution to a problem is tentatively chosen and tried out, which leads to evaluation and corrective actions. Moreover, in order to solve a problem, the cycle most often has to be carried out multiple times, forming a spirallike trial-and-error process of increased knowledge. This design cycle contains the following steps: analysis that translates functional requirements into criteria of specifications for the solution, synthesis in terms of generating provisional design proposals, simulation and evaluation of solutions, and finally a decision of which design proposal to proceed. Francalanza et al. [46] propose a conceptual system design approach based on a generic design cycle described by Roozenburg and Eekels [34]. The design approach contains analysis and determination of system requirements, which is made on basis of product, process, and business needs. Secondly, synthesis is carried out, containing decisions regarding level of changeability, changeability enablers, and specific elements such as layout and machines. Hereafter, it is proposed that simulation and evaluation is carried out before a final decision on the implementation is made. The structure of this method implies that the entire design process is a problem solving cycle that is repeated numerous times, until a satisfactory production system is designed meeting the initially stated requirements. Abdi and Labib [17,47,48] present an RMS design cycle and elaborate its specific elements and procedures. The method is divided into three parts that span from strategic design to tactical design, where are configuration link combines these two design levels. On the strategic level, the RMS design strategy is developed, which refers to deciding on the type of manufacturing system that best fits current and future requirements related to cost, responsiveness, quality, and operator skills. The strategic design level is the first step, as it provides decision support for selecting among generic manufacturing system types, such as dedicated manufacturing, reconfigurable manufacturing, and a hybrid of the two system

types. For this purpose, analytical hierarchy processing (AHP) is applied, so that multiple criteria, multiple actors, and multiple alternatives can be considered simultaneously. The second part of the method is the reconfiguration link, which links the strategic and tactical design of the system [48]. In this part, products are grouped into families and the family to produce in each configuration period is selected. The grouping of products is based on clustering, where operational similarity is the main criteria defined through a similarity coefficient. It is assumed, the RMS is able to produce various different product families through different configurations, which means that the product family to design or configure the manufacturing system for in a given time period has to be selected. This selection is based on AHP, where market requirements, such as customer satisfaction, market share, and sales, and manufacturing requirements, e.g. investment in tools, operation cost, overhead cost, reusability, operator skills, and feasibility, are included as objectives. The selected product family serves as input to the tactical design, which is the third and last element in the design method. In this step, the economic and operational feasibility of the configuration for the selected product family is evaluated and justified through AHP, which leads to detailed design in terms of determining layout, machines, and tools, which is not treated in detail in their work [47]. The three design elements presented by Abdi and Labib [17,47,48] are highly iterative and forms a loop-structure or a continuous design cycle. Al-Zaher et al. [49] also propose a framework that is based on problem solving theory, where analysis, synthesis, and evaluation are essential steps. The framework is designed for a specific case of an automotive framing system and is based on a life-cycle view of the manufacturing system. The framework contains four stages: system analysis, system design, operations and maintenance, lifecycle extension through reconfigurations, and a testing and evaluation phase. This framework can be considered a continuous design framework of problem solving type, as it addresses the design as a problem solving and cyclic activity throughout the life of the manufacturing system. The method proposed by Puik et al. [50–52] applies specifically to the reconfiguration phase of the system’s life, and focuses on the design and implementation of new configurations of the RMS. The motivation of their work is that effective methods for concurrent product design and production engineering is needed, in order to successfully launch new reconfigurations. The framework is a type of optimization cycle, which contains system development, risk assessment, and improvements in a loop structure. The development stage builds on principles from axiomatic design, where decompositions in different domains are carried out [50], whereas risk assessment is built on failure mode effect analysis [52]. The tool, qualitative modelling and analysis of processes (QMAP), contains three parts; high-level data diagrams that describe basic functionalities of the entire process, activity models that describe activities in more detail, and risk analysis highlighting which parts of the new process that possess the highest risk in the implementation. So far, the design methods and frameworks reviewed can in particular be argued as being adaptions of design cycles to RMS design or as depicting the design process as a sort of problem solving activity. Additional design frameworks can also be argued as applying this cyclic analysis and synthesis aspect of design. For instance, the framework by AlGeddawy and ElMaraghy [53] which is proposed for designing changeable manufacturing. They emphasize the need for bi-directional design of product and production systems and apply terminology from axiomatic design [54]. In this sense, the design framework contains a product and manufacturing system element, which is coupled in a loop structure through process and capability matrices. When there is a change in the product, this change is translated into the process matrix, which causes a change in the system if the current capabilities are not sufficient. In

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

turn, these new capabilities represent a new opportunity of product development, which closes the loop structure. The work by Benkamoun et al. [22,55] also applies terminology and approaches from axiomatic design. In their work, the design is viewed as a problem that is solved applying the following logic: requirement analyses, decomposition of problem into functional requirements, design of functional architecture, and design of physical architecture and modules [22]. In further works, the authors consider initial design stages of changeable manufacturing, called conceptual high-level design [55]. They propose an intelligent design environment aiming at assisting designers in coping with the problem of designing for changeability and leveraging on changeability. The design for changeable manufacturing systems contain three interrelated elements that is changeability need analysis, modularity design, and platform design. Thus, focus is explicitly on creation of modules and corresponding interfaces, while leveraging on this modular and changeable design refers to reusing the modules of the system. Axiomatic design can be applied to what usually constitutes design phases: establishing design objectives, generate solutions alternatives, analyze solutions, and select the best one, and implementation [54,56]. It is built on the idea that the solution exists in different domains, and that mapping from requirements to parameters between domains decreases the level of abstraction of the solution. This corresponds roughly to one of the basic ideas in phased design. However, the process of doing this mapping and decomposition is highly iterative, and the evaluation of system alternatives that is based on design axioms, can be argued being related to design as problem solving. Therefore, despite the fact that these two latter design methods focus largely on the logic and rationale of different design steps and clearly exhibit cyclic structures, they do also contain the phased aspect of design, as they indicate some sort of systematic procedure of steps that has varying degrees of abstraction regarding the solution.

4. Comparative analysis of design methods for reconfigurable manufacturing The design frameworks and methods reviewed in the previous section deal specifically with reconfigurable manufacturing systems, but the approaches and terminology vary greatly. One differentiating characteristic that has already been treated is the predominate aspect of design being represented in the frameworks. However, there are multiple other differentiating features present, which is presented in Tables 3 and 4. For instance, the starting and ending points of the design methods vary significantly. The framework by Abdi and Labib [17] initiates with the initial strategic decisions of which type of manufacturing system to design, covering both DMS and RMS. The remaining frameworks contain an implicit assumption that a certain degree of reconfigurability, changeability, or modularity is needed, and that the actual decision to launch modular manufacturing compared to dedicated manufacturing is not conducted directly in the sequence of decisions included in the method. A potential drawback of this is that the actual economic potential and justification of introducing reconfigurable manufacturing is not considered, which do not aid the transition in the manufacturing industry that traditionally is based on conventional manufacturing concepts. This issue of selecting and justifying investment in RMS compared to other types has also been dealt with as a separate research issue in RMS literature, without explicit placement in a larger sequence of design steps [18,19,57]. In regards to the ending point of the frameworks, there are various different alternatives. Some of the frameworks finalize the design process with a creation of modules making up the system

185

architecture, some finalize with steps that are related to the actual physical implementation, while others stop the sequence with a decision of whether or not to invest in the design, assuming that a detailed design of the approved solution follows. In combination with the starting point, this highlights that the different reviewed design cycles apply to different phases of the entire design process. Moreover, this characteristic is highly related to the inclusion of the system’s operating lifetime, which refers to including the task of continuously reconfiguring the system. Some of the methods do not include this, as they solely cover the initial design of the system, whereas others include a repetition of some of the design steps or new steps related to the reconfiguration. This also implies a high degree of integration with product design, as the continuous repetition of design steps often occurs in combination with changes in the product portfolio. The integration of the manufacturing system design framework with product design is represented to various degrees in the reviewed frameworks. A high degree of integration is represented by the co-evolution frameworks [43,58], where there is a loop between changes in the product domain and in the process domain. Similar to this, some of the frameworks consider procedures to conduct when either new variants or new products are introduced. Another type of integration with product development is analysis of the product portfolio and its development in the initial identification and assessment of change drivers, which is covered in the great main part of the reviewed methods. Moreover, clustering of products for the design of configurations was included in one method, which can be argued as explicit consideration of the RMS characteristic, customized flexibility, within the design process. This issue of formulating product families based on operation similarities has also been dealt with as a separate research issue in current RMS literature [59,60]. However, it is notable that this explicit decision of which product family or families to customize the design for is only covered to limited extent, leaving this issue as part of the analysis and determination of requirements for the manufacturing system. The last notable difference between the reviewed frameworks is the coverage of manufacturing levels in the design steps, procedures, and decisions. In respect to manufacturing levels, a widely applied terminology is the six structuring levels of the factory described by Wiendahl et al. [61]. This exact terminology is applied by some of the methods that cover design of both higher levels, such as the factory, but also lower levels, such as workstations and tools. Some methods only consider the overall manufacturing system, while leaving the design of lower levels out of the scope. Nevertheless, in doing this, decisions regarding levels, type, and degree of reconfigurability is neglected, which are essential design decisions. Some of the reviewed contributions merely represent conceptual frames that leaves room for future research and applications of various specific procedures and tools throughout the design steps. Others include more comprehensive methods and procedures to guide the practitioners in the design process. In addition to this, the applicability of the frameworks or design methods has not yet been widely addressed or proven. Abdi and Labib [17,48] include a case study and demonstrate the applicability of their three design steps in this case. Deif and ElMaraghy [41] apply their design architecture to an automatic PCB assembly line, and the method by Al-Zaher et al. [49] was developed in an industrial case of automotive framing systems. Thus, literature presents only a few real life case applications, which supports the conclusion of previous researchers that there appears to be a lack of support for the transition of companies from having non-configurable manufacturing systems to developing reconfigurable systems [13]. Moreover, when considering this lack of support, the great variety in design methods is noteworthy, as it may appear as a sign that

186

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

Table 3 Design methods for reconfigurable manufacturing systems with phased structure. Authors

Initial activity

Final activity

Consideration of operating lifetime of system

Integration with product design

Coverage of manufacturing levels

Rösiö et al. [21,38]

Identification of change drivers and need for reconfigurability

Not explicitly considered

Identification of product-related change drivers

Mainly system level

Schuh et al. [40]

Identification of change drivers

Detailed design of system considering reconfigurability characteristics Creation of modules

Not explicitly considered

Systems, subsystems, and elements

Heisel und Meitzner [32]

Definition of requirements for reconfigurability

Definition of characteristics of modules

Not explicitly considered

Bi et al. [13]

Architecture design based on requirements for reconfigurability Defining need for scalability and convertibility

Design of system control in operating phase of system Physical implementation of configuration

Tracht & Hogreve [26]

Identification of requirements for system

Bryan et al. [42–44]

Concurrent design of product families and assembly systems

Compilation of initial design and procedures for reconfigurations when changes occur Reconfigurations of assembly system

Repetition of configuration and control design Repetition of selection and implementation of reconfigurations derived from modular system design Includes reconfiguration phase in design framework

Identification of product-related change drivers Analysis of trend in product development in identification of reconfigurability requirements Not explicitly considered

Deif & ElMaraghy [41]

Includes reconfiguration planning in design method

Identification of requirements for functionality changes

Procedures for product and variant changes included in reconfiguration phase Co-evolution of products and processes

From system to tools

System, cell, and tool level included under architecture design System and its hard, soft, and human components

From system to workstation

Workstations and resources

Table 4 Design methods for reconfigurable manufacturing with cyclic structure. Authors

Initial activity

Final activity

Consideration of operating lifetime of system

Integration with product design

Coverage of manufacturing levels

Francalanza et al. [46]

Identification and analysis of requirements for system Selection and justification of system type

Decision on whether or not to proceed with design proposal

Not explicitly considered

Product range analysed in order to define system requirements

From factory to station

Decision on feasibility of design

RMS design loop

System level

Al-Zaher et al. [49]

Analysis of needs and specification of the system

Puik et al. [50–52]

Functional decomposition and analysis of sublevels

Reconfigurations of initial design including testing/evaluation before implementation Implementation and effect analysis of new configurations

Includes both initial design and life cycle extension through reconfigurations Covers optimization and reconfigurations of the system

Product clustering as reconfiguration link Inclusion of product-related objectives in system selection Characteristics of product considered in analysis

AlGeddawy and ElMaraghy [53]

Identification of requirements for system

Benkamoun et al. [22,55]

Identification of need for changeability

Description of system and product architecture and coupling between Description of architecture and procedure for leveraging on the modular architecture

Abdi & Labib [17,47,48]

there is little agreement on the structure of the design process. This conclusion is consistent with previous research’s indications that a systematic design method for RMS is lacking [13,20,28–31], which is unfortunate in multiple ways. First, as the applied terminology varies significantly, this may lead to inconsistency and lack of clarity about what reconfigurable manufacturing is and how it should be developed. Secondly, there is currently no directions or support

Multiple levels of the system including tooling Basic process modules and their detailed design

Bi-directional changeability loop

Concurrent product design and production development during reconfiguration Co-evolution of products and processes

Leveraging on changeability

Identification of need for changeability

System modules

System covering factory and machines

for deciding on which design method to apply, including which procedures to follow and tools to apply. One way of approaching this issue is to seek uniformity in the reviewed literature, and develop a generic design method for reconfigurable manufacturing including support for deciding which specific approaches and procedures to apply in each step, which is the aim of the following sections.

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

5. Synthesis of design methods for reconfigurable manufacturing Hall [36] and Roozenburg and Eekels [34] argue for combining the problem solving and phased aspects of design, due to the recognition that these two aspects complement rather than oppose each other. Moreover, they argue that often only minor differences exist between variants of the same types of design methods, which are mainly of terminological nature. Following this logic, this section presents a synthesis of the reviewed design methods for reconfigurable manufacturing.

5.1. Synthesis of phased design methods The phased design methods for reconfigurable manufacturing share similarities in regards to distinguishing between phases and in terms of which activities to include in each phase, despite their differences in terminology. Generally, they form a sequence of making a requirement specification, making a system solution, implementing the physical manufacturing system, operating the system, and continuously evaluating its ability to fulfill requirements leading to reconfigurations. Some of the methods, which were argued as being phased design methods, leave out the operation and reconfiguration part, whereas others extend the method to covering the entire life-cycle of the manufacturing system. However, collectively, the phases represented in the methods do not differ significantly for the design part, but differ in coverage of the operation phase and initial planning phase. Thus, a generic phase method for reconfigurable manufacturing can be distilled from the reviewed design methods, which is depicted in Fig. 3. The generic design structure represents a separation of design activities that are somewhat common to the reviewed design frameworks. A common starting point for the design methods is the initial analysis of need for reconfigurability. The terminology varies, e.g. “capturing the market” [40], “identify change drivers” [21,40], or “identify need for scalability” [46], but the essence lies in analysing and specifying the requirements for the manufacturing system that is to be designed. This part of the design process, is often regarded as a part of the preparatory design or background study in conventional system design [24], and can roughly be translated into clarifying the design task as defined by Roozenburg and Eekels [34]. However, there is a slight difference in the inclusion of project and investment planning in the methods, which is covered by the work of Rösiö [21] and Heisel and Meitzner [32]. All of the included contributions form a sequence of activities collected in phases that subsequently increases the level of detail. However, the level of detail of procedures and sub activities within each phase varies greatly, as well as the exact separation of activities. Two notable differences should be mentioned here. The main part of the reviewed methods exhibits a separation of design activities according to what normally constitute conceptual design, concept selection, and detailed design [24]. In this respect, different solution concepts are initially developed and evaluated, where one concept is transferred to detailed design. This is clear in the work by Rösiö et al. [21,38], Deif and ElMaraghy [41], and Bi et al. [13]. However, this process of going from multiple alternative concepts to one detailed is not evident in the remaining works. However, they do to high extent illustrate a distinction between more basic design and further advanced design, such as in the work by Schuh et al. [40], Heisel and Meitzner [32], and Tracht & Hogreve [26]. Examples of basic design activities are defining the level of automation of the system, determining system elements, and determining degree of modularity. Only the work by Bryan et al. [42–44] does not follow this logic, as only two phases are treated covering initial design and reconfigurations.

187

Despite slight differences in exact activities and number of intermediate abstraction levels in the reviewed methods, two phases of basic and detailed design can in general be identified. In this sense, basic design refers to the creation of broad solutions of manufacturing systems that can be further detailed in the subsequent phase. In conventional system design, conceptual design of manufacturing systems refers to activities related to selecting capacity, planning materials and production flow, selecting type of processes, specifying equipment, choosing level of automation, and developing layout, whereas detailed design considers the selected system concept in a more detailed manner [24]. Obviously, these areas should also be considered when designing a reconfigurable manufacturing system. However, not all of these topics are explicitly considered or addressed in the reviewed design methods. Collectively, the reviewed literature suggests that the basic design covers decisions related to degree of modularity and granularity of system elements, degree of reconfigurability to build in the system, degree of automation, clustering of operations, identification of interfaces and dependencies, and assembly method. Conceptual solutions that are modular considering logical, physical, and human aspects of the reconfigurable manufacturing system are designed in this phase. In the detailed or advanced design phase, the design concept or solution that is selected for implementation is further specified and prepared for implementation. The reviewed literature suggests that this advanced design includes defining the details of the modules in terms of their interfaces and required tooling and equipment. The last two phases of the method are implementation and reconfiguration. Usually, conventional design methods finalize with the implementation and run-in. However, in the case of a RMS, continuous reconfigurations and redesigns are needed in order to extend the lifetime of the system through its modular architecture. This reconfiguration phase is often described as being a repetition of some steps in the design framework, e.g. going back to clustering operations [26] or by going back and further detailing the design of some conceptual configurations of the system [41], which translates to repeating steps regarding detailed design. As described, the generic phases identified here roughly cover phases applied for conventional manufacturing system design. Obviously, the content of the phases covers specific activities related to designing reconfigurable manufacturing, and the last phase of reconfigurations is unique to the design method for reconfigurable manufacturing. However, the extent to which the different reviewed contributions guide practitioners in each phase varies greatly, which means that focus generally is on the structure of the design process rather than on how to conduct each step. In this respect, it is relevant to consider the combination of phases and problem solving activities. 5.2. Synthesis of cyclic design methods Common to the problem solving methods is that they represent some sort of cyclic design process that is repeated until the solution is considered satisfactory. However, the design cycles in these contributions cover different elements of the design process. For instance, Francalanza et al. [46] cover the design of initial solutions without concern for continuous design in the operating life of the system. Likewise, Abdi and Labib [17,47,48] and Puik et al. [50–52] cover parts of the lifetime of a manufacturing system, the initial strategic/tactical design and the optimization of the system during reconfigurations respectively. In contrary, the method by Al-Zaher et al. [49] covers the entire lifetime of the manufacturing system, where testing and evaluating is carried out continuously prior to implementing new processes or equipment in the reconfigurations. Thus, it can be concluded that the problem solving cycle in the reviewed design methods can be applied to different phases of the

188

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

Fig. 3. Phases for RMS design.

Fig. 4. Design cycles for RMS design.

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

development. This is consistent with the idea that problem-solving cycles are and should be conducted in each phase of the design process [36]. A critical element in the design cycles for reconfigurable manufacturing is the criteria applied for making the decision of how to proceed with the provisional design. However, only the work by Abdi and Labib [17,47,48], includes specifically stated evaluation criteria related to economical and operational aspects. In Fig. 4, it is shown how the basic design cycle constitutes the backbone of different types of design methods for reconfigurable manufacturing. Only the reviewed contributions that in particular exhibits the problem solving aspect of design is included here. The two remaining design frameworks are not included in the figure, as the axiomatic design approach as argued not directly can be adapted to the design cycle, but rather shares characteristics with both the phased and cyclic aspect of design.

5.3. Generic method for design of reconfigurable manufacturing systems In the previous sections, it was argued that the phased methods for design and development could be synthesized into a number of generic RMS stages. Thus, a common underlying pattern of the reviewed design methods was identified, which reveals a common structure that is followed in the design and development of reconfigurable manufacturing systems. Moreover, it was argued that the different examples of problem solving cycles in the reviewed design methods cover different phases of the design and development. By combining these two aspects of design, a method that is suited both for project management and for capturing the iterative sub processes created, which is depicted in Fig. 5. In this method, each phase include a cycle of steps, where a decision leads either to the subsequent phase or to iterations of previous steps. Moreover, the entire design process is cyclic in itself, as the reconfiguration phase results in repeating some steps within basic and advanced design. Thus, the design method integrates a lifecycle view of the manufacturing system.

6. Implications for development of reconfigurable manufacturing The synthesized method corresponds roughly to the overall structure of conventional system design methods as described by for instance Bellgran and Säftsten [24] or Wu [25]. However, there are some significant differences, which makes it important to consider the practical implications in regards to conducting each step in the design method, as well as the available tools to assist in the design process. The practical implications of conducting RMS design can be argued as being largely determined by the differences existing between designing conventional manufacturing systems and designing reconfigurable manufacturing systems. These differences increase the complexity of the design process, which is a challenge to the effective implementation of reconfigurability in industry. Overcoming this challenge requires changes in the way the design process is conducted. In this respect, previous research has emphasized different prerequisites for being able to design reconfigurable manufacturing, such as having a life-cycle perspective on production systems, correlating product and production design, having long-term view on investments, and having a holistic perspective on manufacturing systems [21,27,33]. Evidently, these prerequisites indicate capabilities, knowledge, or perspectives that should be present in manufacturing companies, in order to handle the design challenges listed above. However, being able to overcome these challenges also relies heavily on having available supportive tools and procedures in every design phase.

189

6.1. Management and strategic planning phase In the first phase of the design method, which traditionally covers planning and making strategic decisions about the development project, a challenge in RMS design is to justify investments and potential in reconfigurable manufacturing. This activity has not been widely regarded as an activity within the design and development process, even though determining the need for reconfigurability and justifying its investment is one of the first crucial steps in its implementation [32,47]. A major challenge in relation to this is the life-cycle perspective and the uncertain nature of predictions. However, some procedures can be recognized as support for this pre-design step, which are economic justification models, models that compare system alternatives, and more pragmatic procedures to determining RMS potential. An example of such a model can be found in the work by Kuzgunkaya and ElMaraghy [18], where an evaluation model is presented in order to analyze, assess, and justify RMS investments. The model uses fuzzy multi-objective optimization incorporating both financial and strategic objectives, such as flexibility and responsiveness. Likewise, Singh et al. [19] and Abdi and Labib [17] apply fuzzy analytical hierarchy processing logic for the selection of manufacturing systems based on different criteria, such as responsiveness, product cost, operator skill, and convertibility. More pragmatic and case-based approaches to quantifying the potential of reconfigurable manufacturing on different levels have also been presented [6,62]. Evidently, there are significant differences between these suggested approaches and tools, where requirements for data, information, and skills of practitioners should be emphasized. In the economic justification and optimal system selection models, estimating anticipated costs and other quantitative criteria related to each system alternative is an essential task, which might prove difficult and time-consuming in some companies. Moreover, these approaches can be argued as requiring advanced computational skills, which also might be an obstacle in its practical application. In contrary, the less normative methods can be more easily and effortlessly applied, but are unlikely to constitute sufficient foundation for investments in reality. In addition, justifying and evaluating the decision to invest in reconfigurability should be considered in relation to the possibility of gradually changing the functionality and capacity of the system, e.g. by continuously increasing the reconfigurability of the initially designed system or an already existing system. However, this gradual and long-term development of systems or the transition from more dedicated or flexible systems towards reconfigurability is not wellcovered in current literature, which implies that future research should focus on approaching reconfigurability on a less aggregate level and as capability that can be gradually achieved. 6.2. Clarification of design task In the second phase, the approaches applied for completing the task of identifying need for scalability and convertibility are in general vaguely defined, and the reviewed literature provide only limited indications of which specific procedures to apply for doing this. Tracht and Hogreve [26] argue that this task includes considering future system layouts and the expected number of configurations per year, suggesting a combination of conventional design methods such as requirement lists with e.g. creative brainstorming. Likewise, Rösiö [21] proposes scenario development as an applicable method, where different scenarios can be developed based on identified change drivers and their potential outcomes, as well as a set of questions is suggested which can be applied for constructing scenarios. In addition to this, the perspective of production system portfolios has also been set forward to aid the strategic planning of production system development taking account for future changes and reconfigurations of the systems

190

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

Fig. 5. Generic RMS design Method.

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

[63,64]. In this respect, design and development activities are results of strategic portfolio planning, where changes in systems and products are planned iteratively and in the long-term. Likewise, the idea of co-evolution has also been proposed as a way of describing changes and adaptions of products, processes, and manufacturing systems [65]. Thus, it has been suggested that studying the path of evolution of both product features and manufacturing capabilities can be beneficial in terms of determining plausible future evolution trends [58]. The application of cladistics is essential in the co-evolution approach, where the historical development in both products and production are classified to get insight into mechanisms of changes. However, most of the contributions on design of reconfigurable manufacturing assumes that change drivers and need for change in the systems lifetime are identified initially, but the specific approach to this are rarely described, which might represent a critical barrier to the design of reconfigurable manufacturing. 6.3. Basic design In the basic design phase, identification of product families, and deciding on degree, type, and level of reconfigurability were emphasized as challenging elements. As such, the formation of product families have been dealt with for more than a decade in both research and practise, where product families are defined as groups of related products derived from the same product platform [66–68]. The definition of what constitutes a product family can be done in various ways depending on perspective, e.g. marketing or resource perspective [48]. However, as the reconfigurable manufacturing system is built around, product families to reduce the trade-off between flexibility and efficiency, it is essential to design product families concurrently with the production [3]. In this respect, three concerns can be identified in literature; grouping products into product families, designing the corresponding optimal configuration, and adapting the system configurations to evolving product families [43,58]. Nevertheless, literature offers several approaches to dealing with these concerns. Goyal et al. [59] focus on the issue of grouping products or parts into families, in order to design a reconfigurable manufacturing system. In their approach, products are clustered based on an operation similarity coefficient. In the work by Abdi and Labib [48], the concerns of grouping products and then selecting configurations is treated in sequence. They propose a reconfiguration link of the RMS, which contains a definition of product groups or families and a selection of the product family to produce in the configuration. The grouping of products is based on clustering techniques, where operational similarity is the main criteria defined through the similarity coefficient. The selection is based on analytical hierarchy processing where market objectives and manufacturing objectives are considered. A premise in this approach is that the RMS can produce one product family in each configuration and that the manufacturer has to select the optimal configuration for each family, and then continuously select one product family to produce in each operating period. This premise is also present in the work by Xiaobo et al. [16,69–71], who propose a stochastic model containing optimal configuration design and optimal selection, using a mathematical optimization approach to determine the optimal configuration for a family from several feasible configurations. The perspective applied in these contributions differs slightly from the initially proposed RMS concept by Koren [5,72], where a configuration usually corresponds to one or two types of products from the same product family. Yigit et al. [73,74] also apply this perspective, and propose a method for optimizing modular products for the reconfigurable manufacturing system. In this work, the problem of selecting optimal module instances of a product is considered simultaneously to the cost of reconfiguration, assum-

191

ing that a particular configuration of the system can be used to produce a particular instance of the product family. Bryan et. al [42–44] also applied optimization, but focus on solving the task of forming product families and designing the assembly system simultaneously. Co-evolution is essential in this model, as first and future generations of the system are designed by joint consideration of assembly system and product family formation through mathematical optimization. Focusing specifically on procedures and tools to handle the challenge of designing a manufacturing system based on product families, reveals several different research perspectives and procedures. First, there is currently different views on how much variety the RMS is able to handle, whether it is several product families or one product family, representing different underlying premises of the contributions. However, there is limited support for actually making the decision in practice of how much variety to cover in the system, in order to effectively trade off the objectives of flexibility and productivity. Moreover, the brief review of approaches in the design phase reveals that current contributions differ in terms of focus, in the exact tools applied, and in whether the concerns of forming families and selecting configurations are treated in sequence or concurrently. In particular, the latter can be emphasized, as it also shows that there is not one single answer to how this stage of design is integrated with other design stages, such as product architecture design and utilization of the operating RMS. From a practical viewpoint, it may thus be difficult to select the appropriate approach to apply given the circumstances of the case. However, it can be argued that applying these tools in practise, requires a high degree of knowledge of reconfigurability and ability to collect and process rather large amounts of data and information. The second challenge emphasized in this stage of design is related to determining degree, type, and level of reconfigurability. These three decisions are highly related and can be argued as being rather case-specific and hard to generalize, which may be the reason why they are rarely treated explicitly in the reviewed literature. The first issue, determining the level of reconfigurability, refers to the system’s ability to scale capacity and change functionality. Koren [9,75] has shown how different configurations are associated with different degrees of scalability and convertibility. Thus, an essential part of this design step includes having performance metrics related to differences in reconfigurable design options. Such metrics have been presented by e.g. Maler-Speredelozzi and Koren [76] or Ko et al. [77]. The second issue, deciding on the type of reconfigurability, is closely connected to the level in the system on which reconfigurability is implemented [78]. Reconfigurability may cover the ability to scale capacity, convert between product variants, and adapt to new product generations. Commonly applied concepts in regards to this is the changeability classes and enablers proposed by ElMaraghy and Wiendahl [78]. In addition, Terkaj et al. [79] present an ontology on flexibility aiming at providing a standardized method to analyse flexibility. Thus, it appears that the concept and theoretical foundation for describing and classifying types and levels of reconfigurability and flexibility are rather well established in research. However, the practical issue of translating requirements into the needed level and type of reconfigurability is another concern. Axiomatic design has been suggested for this in some contributions [53,55], whereas Terkaj et al. [79] propose an approach that utilizes the flexibility ontology to decide on physical components of the system. Rösiö [21,37,38] considers this issue, by relating the different RMS characteristics to identified change drivers in order to decide on the type of reconfigurability to design into the system. Nevertheless, in the reviewed design methods, the issue of making decisions in this area are generally vaguely described, and only a few case studies show how actual decisions could be conducted with the suggested approaches.

192

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

6.4. Advanced design In the advanced design phase, challenges in regards to the detailed design of modules in terms of their logical and physical aspects were emphasised. Schuh et al. [40] propose object-oriented design for this, where objects or modules are created based on a separation of complex and complicated system elements. Likewise, Benkamoun et al. [55] propose different types of module and interface types according to changeability strategies, which can be applied to determine the modularity of the system simultaneously to designing the production platform. Often this is regarded as the architecture of the manufacturing system or as a way of doing platform-based development. In this approach, the RMS is considered constituting a platform that can be developed in relation to the product platform, where a platform is defined as a set of subsystems and interfaces that forms a common structure from which a variation of systems or products can be developed [80]. When applying this approach, the centre of attention is the achievement of commonality and control of interfaces, where different modelling approaches can be used, e.g. function-means modelling [80] and product family master plans extended for manufacturing [81,82]. However, the challenge of co-developing production and products through platforms appears to be a remaining challenge that goes far beyond the issue of modelling the platforms. It is also in the advanced design phase, that detailed decisions and specification of the physical and logical aspects of the manufacturing system are conducted. Decisions in this area are closely linked with decisions in the initial basic design regarding type, level, and degree of reconfigurability needed by the system. Often reconfigurability in hardware is enabled by reconfigurable machines that are able to be quickly converted between varieties within product families [83]. Usually, reconfigurable machines cover machine tools, fixture systems, assembly machines, inspection machines, and material handling systems, which all are essential to the RMS paradigm, as they provide customized flexibility and ability to reconfigure on equipment level through combinations of basic and auxiliary modules [14]. Some guidelines exist in this area, for instance the design principles for reconfigurable machines proposed by Katz [14]. However, even with current research contributions in this area, their effective implementation is limited, as they are currently not broadly available and still in development [61,83,84]. The logical aspects of the system cover decisions in the area of the process planning system, and system for production planning and control. Process planning refers to planning and defining the steps needed for executing a process, which should be able to handle both short-term and long-term changes. An example of a way of dealing with this is proposed by Azab and ElMaraghy [85], where a mathematical model and optimal solution algorithm is proposed for reconfiguring process plans. Moreover, in terms of production planning and control, it has been highlighted that manufacturer’s logistical planning systems should be designed to become adaptable to reconfigurations in manufacturing [61]. A related issue that appears to receive attention in research is the ability to model the reconfigurable manufacturing system, in order to be able to analyse, simulate, and verify the system before reconfigurations [86–88]. Nevertheless, designing and deploying such approaches in industry can be argued as being related to a high degree of complexity, e.g. in terms of having available knowledge bases and rules for planning and being able to apply accurate and advanced mathematical models [61]. 6.5. Reconfiguration phase The last phase of the design method contains the challenge of selecting among alternative configurations during operating time, and deciding when and how to reconfigure the manufacturing

system during its lifetime. As argued, this step often contains a repetition of steps from the design phase, e.g. in terms of adding a new product to the product families, designing additional modules, or simply selecting a new configuration from the number of feasible choices. Development of optimization models for this has been widely suggested [15,71,89]. In addition to this, Youssef and ElMaraghy [90] apply a reconfiguration smoothness metric in this reconfiguration process, in order to evaluate the required cost, time, and effort associated with a reconfiguration. Their work emphasized the need for additional practical considerations in the planning of a reconfiguration. Another critical issue in the reconfiguration is the ramp-up time, which defines the time from a given reconfiguration is started to the point where production reaches its planned output in volume, variety, and quality [5]. Examples of approaches to improve this ramp-up process are human error predictions related to quality [91] and the application of stream-of-variation analysis. Moreover, the already discussed approach proposed by Puik et al. [50–52] could also be applied as an engineering tool for improving the reconfiguration process. In general, however, research on ramp-up is currently primarily focused on classifying generic challenges and their impact [92,93] and has not been considered explicitly from a RMS perspective. 6.6. Suggested procedures and applicable tools In Table 5, a summary of the different design challenges, suggested procedures, and applicable tools discussed in the previous sections are presented. In this respect, procedures are defined as specific courses of actions to approach the indicated challenge. The explicit ways of conducting these procedures are regarded as the applied tools, which indicate technical skills and requirements that the design team should possess. The list of suggested procedures and applicable tools are a combination of findings from the reviewed literature, additional publications from the initial retrieval described in Section 2, and findings from previous research in the area. 7. Conclusion and future research The reconfigurable manufacturing system has been widely recognized as being the manufacturing paradigm of the future, due to its ability to rapidly and efficiently change capacity and functionality in order to meet demand and product changes. However, when designing reconfigurable manufacturing systems, practitioners face a challenging task compared to designing dedicated manufacturing systems. Therefore, the aim of the research presented in this paper was to identify a design method for reconfigurable manufacturing systems, determine if current contributions are opposing or supplementary, and investigate practical implications of the design, including which supportive procedures and tools that can be applied in the design process. The applied research method covers a systematic literature review, classification of identified contributions, and analysis and synthesis of the findings. Among the main findings of the research is that it may appear as if there is little agreement on the structure of the design process, which however is largely due to varying terminology, level of detail, and whether a phased or cyclic perspective on design is applied. In fact, it was argued that a common underlying pattern of the reviewed design methods could be identified, where the phased methods were synthesized into a shared phase structure and the problem solving cycles were argued as covering these different phases. By combining these two aspects of design, a synthesized and more generic design method for reconfigurable manufacturing was identified. Finally, the practical implications

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

193

Table 5 Suggested procedures and applicable tools for RMS design. Design Phase

RMS design challenges

Suggested procedures

Applicable tools

Management and planning

Preparation of investment request including justification for investment in reconfigurable systems that are more expensive that dedicated equipment. Quantification of the potential in reconfigurable manufacturing compared to conventional manufacturing systems. Having a long-term view on manufacturing systems and products

Economic evaluation and justification [18]

Mathematical optimization [18]

Optimal system selection [17,19]

Analytical hierarchy processing [17,19]

RMS potential assessment [6,62]

Basic calculations [6,62]

Identification of scalability and convertibility requirements

Scenario development [21]

Brainstorming [26]

Integrated product and production system portfolios [63] Co-evolution approach [58]

Questionnaire on change drivers [21] Cladistics [66]

Grouping products into families [48,59] Decide on optimal configurations [16,48,69] Co-evolution approach [42–44]

Clustering techniques [48,59]

Clarification of design task

Basic design

Identification and selection of product families for the design of the manufacturing system

Decisions regarding degree of reconfigurability, type of reconfigurability, and level of implementation

Advanced design

Decisions regarding system modules and the granularity of these

Design of reconfigurable of hard and soft aspects of the system including control system

Reconfiguration

Selecting among alternative configurations during operation and deciding when and how to reconfigure the manufacturing system during its lifetime Enabling and securing quick and efficient ramp-up after reconfigurations

and challenges related to this design method were discussed and available supportive approaches and tools for the design process were identified, which reveals areas for further research.

7.1. Future research directions Assessing available supportive approaches and procedures for carrying out the design of reconfigurable manufacturing is considered highly relevant in order to aid the transition towards reconfigurability in industry. However, the assessment in the previous section reveals numerous areas for future research, both in regards to limitations of tools to apply throughout the steps, lack of supportive tools, and support for how to select approach and tool to apply. The first research issue that has been identified throughout the research presented in this paper is related to the suggested procedures and availability of tools to apply in the design process. With focus being specifically on potentially challenging aspects of designing reconfigurable systems rather than dedicated systems, a

Apply reconfigurability metrics to decide on system alternatives [76,77] Match system requirements with reconfigurability forms [21,37,79]

AHP [48] Mathematical optimisation [43,69,73,74] Axiomatic design [53,55]

Linkage between drivers and RMS characteristics [21,37,38] Flexibility ontology [79]

Separation of complicated and complex system elements [40] Define interface and module requirements [55] Design production modules and platforms [80] Apply reconfigurable machines [14] Design adaptable process and production planning systems [61,85] Design control system [86–88]

Object-oriented design [40]

Optimal configuration selection [15,71,89] Assessment of reconfiguration smoothness [90] Quality predictions [91] Reduction of variation Concurrent redesign of product and process [50–52]

Mathematical optimization [15,71,89] Sensitivity analysis [90]

Function-means modelling [80] PMFP [81,82] Modelling techniques [86–88] Mathematical optimization [85]

Human error modelling [91] Stream of variation methodology Product/process development cycle [50–52]

major issue was identified in terms of how to identify and express need for change in the system’s lifetime. It has been argued that this issue has been merely overlooked in research, while potentially being one of the most critical and complicated tasks in reality. Therefore, as this element has proved to be a foundation for all further design decisions, it is highly relevant that practically applicable procedures are developed for this and tested in industrial setting, e.g. in terms of how to express the need for reconfigurability through metrics related to volume, variety, and their uncertainties. Moreover, another interesting area for further research is to connect these predictions of required changes to essential decisions on how much variety to cover in the system design, which type and degree of reconfigurability to focus on, and on which levels to implement it. The review of design procedures and tools highlights that there is currently lack of research on how to actually solve design issues in practise, as only a limited number of case studies or case applications are available. Some areas, such as the design of reconfigurable process and production planning and design of reconfigurable

194

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

machines still present significant research challenges, and may therefore be difficult to implement and test in industry. However, it has been emphasized throughout this research, that most available tools represent rather advanced decision support, which may prove difficult to apply in manufacturing companies where there is general uncertainty about what reconfigurability actually is [20,32]. Moreover, in regards to this it is often assumes that large amounts of data and information is readily available and that knowledge bases and procedures for planning are rather advanced in companies, which might not always be the case. Thus, supporting the transition towards reconfigurability in industry should be a future research concern. The last issue that has been emphasized in this research, is related to how to conduct the design process in practice in terms of selecting which approach and supportive tools to apply. In most of the design phases, there are multiple options indicated in current research on reconfigurable manufacturing of how to approach parts of the design process. Some of the options merely indicate different ways of approaching a design issue, but there is currently no considerations or indications of when these different tools are appropriate. It is reasonable to argue, that the choice of approach would differ between manufacturing companies, as the requirements for being able to apply them differ. Some characteristics of companies that could impact this choice could be the ability to apply advanced computational methods and modelling, level of sophistication of current planning methods, willingness and knowledge related reconfigurability, and the size and type of cross-functional development team. However, further exploring differences between how different company types develop reconfigurable manufacturing would be an interesting area for research.

References [1] Gilmore JH, Pine BJ. Markets of one: creating customer-unique value through mass customization. Boston: Harvard Business Review Press; 2000. [2] Roland Berger Strategy Consultants GmbH. Mastering product complexity. Online. Roland Berger Strategy Consultants GmbH; 2012. [3] ElMaraghy H, Schuh G, ElMaraghy W, Piller F, Schönsleben P, Tseng M, et al. Product variety management. CIRP Ann Manuf Technol 2013;62:629–52. [4] Koren Y. General RMS characteristics. Comparison with dedicated and flexible systems. In: ElMaraghy HA, editor. Reconfigurable manufacturing systems and transformable factories. Springer; 2006. p. 27–45. [5] Koren Y, Heisel U, Jovane F, Moriwaki T, Pritschow G, Ulsoy G, et al. Reconfigurable manufacturing systems. CIRP Ann Manuf Technol 1999;48:527–40. [6] Andersen A, Brunoe TD, Nielsen K. Investigating the potential in reconfigurable manufacturing: a case-study from danish industry. In: Umed S, Nakano M, Mizuyama H, Hibino N, Kiritsis D, von Cieminski G, editors. Advances in production management systems: innovative production management towards sustainable growth. Springer; 2015. p. 274–82. [7] Zhang G, Liu R, Gong L, Huang Q. An analytical comparison on cost and performance among DMS, AMS, FMS and RMS. In: Dashcenko AI, editor. Reconfigurable manufacturing systems and transformable factories. Springer; 2006. p. 659–73. [8] Mehrabi MG, Ulsoy AG, Koren Y, Heytler P. Trends and perspectives in flexible and reconfigurable manufacturing systems. J Intell Manuf 2002;13:135–46. [9] Koren Y. The global manufacturing revolution: product-process-business integration and reconfigurable systems. John Wiley & Sons; 2010. [10] Mehrabi MG, Ulsoy AG, Koren Y. Reconfigurable manufacturing systems and their enabling technologies. Int J Manuf Technol Manage 2000;1:114–31. [11] Mehrabi MG, Ulsoy AG, Koren Y. Reconfigurable manufacturing systems: key to future manufacturing. J Intell Manuf 2000;11:403–19. [12] Andersen A, Brunoe TD, Nielsen K. Reconfigurable manufacturing on multiple levels: literature review and research directions. In: Umed S, Nakano M, Mizuyama H, Hibino N, Kiritsis D, von Cieminski G, editors. Advances in production management systems: innovative production management towards sustainable growth. Springer; 2015. p. 266–73. [13] Bi Z, Lang S, Shen W, Wang L. Reconfigurable manufacturing systems: the state of the art. Int J Prod Res 2008;46:967–92. [14] Katz R. Design principles of reconfigurable machines. Int J Adv Manuf Technol 2007;34:430–9. [15] Youssef AM, ElMaraghy HA. Optimal configuration selection for reconfigurable manufacturing systems. Int J Flex Manuf Syst 2007;19:67–106. [16] Xiaobo Z, Jiancai W, Zhenbi L. A stochastic model of a reconfigurable manufacturing system Part 1: a framework. Int J Prod Res 2000;38:2273–85.

[17] Abdi MR, Labib AW. A design strategy for reconfigurable manufacturing systems (RMSs) using analytical hierarchical process (AHP): a case study. Int J Prod Res 2003;41:2273–99. [18] Kuzgunkaya O, ElMaraghy H. Economic and strategic perspectives on investing in RMS and FMS. Int J Flex Manuf Syst 2007;19:217–46. [19] Singh R, Khilwani N, Tiwari M. Justification for the selection of a reconfigurable manufacturing system: a fuzzy analytical hierarchy based approach. Int J Prod Res 2007;45:3165–90. [20] Rösiö C, Säfsten K. Reconfigurable production system design-theoretical and practical challenges. J Manuf Technol Manage 2013;24:998–1018. [21] Rösiö C. Supporting the design of reconfigurable production systems; 2012. [22] Benkamoun N, ElMaraghy W, Huyet A, Kouiss K. Architecture framework for manufacturing system design. Procedia CIRP 2014;17:88–93. [23] Chryssolouris G. Manufacturing systems: theory and practice. Springer Science & Business Media; 2013. [24] Bellgran M, Säfsten K. Production development: design and operation of production systems. London: Springer-Verlag; 2009. [25] Wu B. Manufacturing systems design and analysis. Springer Science & Business Media; 2012. [26] Tracht K, Hogreve S. Decision making during design and reconfiguration of modular assembly lines. In: ElMaraghy H, editor. Enabling manufacturing competitiveness and economic sustainability. Springer; 2012. p. 105–10. [27] Rösiö C, Jackson M. Enable changeability in manufacturing systems by adopting a life cycle perspective. In: Zäh M, editor. Proceedings of the 3rd international conference on changeable, agile, reconfigurable and virtual production. 2009. p. 612–21. [28] ElMaraghy HA. Flexible and reconfigurable manufacturing systems paradigms. Int J Flex Manuf Syst 2005;17:261–76. [29] Malhotra V, Raj T, Arora A. Evaluation of barriers affecting reconfigurable manufacturing systems with graph theory and matrix approach. Mater Manuf Processes 2012;27:88–94. [30] Malhotra V, Raj T, Arora A. Reconfigurable manufacturing system: an overview. Int J Mach Intell 2009;1. [31] Malhotra V, Raj T, Arora A. Excellent techniques of manufacturing systems: RMS and FMS. Int J Eng Sci Technol 2010;2:137–42. [32] Heisel U, Meitzner M. Progress in reconfigurable manufacturing systems. In: Dashcenko AI, editor. Reconfigurable manufacturing systems and transformable factories. Springer; 2006. p. 47–62. [33] Andersen A, Nielsen K, Brunoe TD. Prerequisites and barriers for the development of reconfigurable manufacturing systems for high speed ramp-up. Procedia CIRP 2016;51:7–12. [34] Roozenburg NF, Eekels J. Product design: fundamentals and methods. Chichester: Wiley; 1995. [35] Sommerville I. Software engineering. Pearson Education; 2011. [36] Hall AD. Three-dimensional morphology of systems engineering. Springer; 1969. [37] Rösiö C. Considering reconfigurability characteristics in production system design. In: ElMaraghy H, editor. Enabling manufacturing competitiveness and economic sustainability. Springer; 2012. p. 57–62. [38] Rösiö C, Hedsttöm R. Linking production strategy to production system specification—a case study. In: Rosén B, editor. Proceedings of the international 3’rd Swedish production symposium. The Swedish Production Academy; 2009. p. 12–8. [39] Bruch J. Management of design information in the production system design process; 2012. [40] Schuh G, Lenders M, Nussbaum C, Kupke D. Design for changeability. In: ElMaraghy HA, editor. Changeable and reconfigurable manufacturing systems. Springer; 2009. p. 251–66. [41] Deif AM, ElMaraghy WH. A systematic design approach for reconfigurable manufacturing systems. In: ElMaraghy HA, ElMaraghy WH, editors. Advances in Design. Springer; 2006. p. 219–28. [42] Bryan A, Hu SJ, Koren Y. Assembly system reconfiguration planning. J Manuf Sci Eng 2013;135:041005. [43] Bryan A, Ko J, Hu S, Koren Y. Co-evolution of product families and assembly systems. CIRP Ann Manuf Technol 2007;56:41–4. [44] Bryan A, Wang H, Abell J. Concurrent design of product families and reconfigurable assembly systems. J Mech Des 2013;135:051001. [45] Hart CW. Doing a literature review. Sage Publications; 1998. [46] Francalanza E, Borg J, Constantinescu C. Deriving a systematic approach to changeable manufacturing system design. Procedia CIRP 2014;17:166–71. [47] Abdi MR, Labib AW. Feasibility study of the tactical design justification for reconfigurable manufacturing systems using the fuzzy analytical hierarchical process. Int J Prod Res 2004;42:3055–76. [48] Abdi M, Labib A. Grouping and selecting products: the design key of reconfigurable manufacturing systems (RMSs). Int J Prod Res 2004;42:521–46. [49] Al-Zaher A, ElMaraghy W, Pasek Z. RMS design methodology for automotive framing systems BIW. J Manuf Syst 2013;32:436–48. [50] Puik E, Gielen P, Telgen D, van Moergestel L, Ceglarek D. A generic systems engineering method for concurrent development of products and manufacturing equipment. In: Ratchev S, editor. Precision assembly technologies and systems. Berlin, Heidelberg: Springer; 2014. p. 139–46. [51] Puik E, Telgen D, van Moergeste L, Ceglarek D. Qualitative product/process modelling for reconfigurable manufacturing systems; 2013. p. 214–8. [52] Puik E, Telgen D, van Moergestel L, Ceglarek D. Structured analysis of reconfigurable manufacturing systems. In: Azevedo A, editor. Advances in

A.-L. Andersen et al. / Journal of Manufacturing Systems 42 (2017) 179–195

[53]

[54] [55]

[56] [57]

[58] [59]

[60]

[61]

[62]

[63] [64]

[65]

[66] [67] [68]

[69] [70] [71] [72] [73]

[74] [75] [76] [77] [78]

[79]

[80]

Sustainable and competitive manufacturing systems. Springer; 2013. p. 147–57. AlGeddawy T, ElMaraghy HA. Changeability effect on manufacturing systems design. In: ElMaraghy H, editor. Changeable and reconfigurable manufacturing systems. Springer; 2009. p. 267–83. Suh NP. Axiomatic design theory for systems. Res Eng Des 1998;10:189–209. Benkamoun N, Kouiss K, Huyet A. An Intelligent design environment for changeability management-application to manufacturing systems. In: Weber C, Husung S, Cantamessa M, Cascini G, Marjanovic D, Montagna F, editors. Proceedings of the 20th international conference on engineering design. 2015. p. 209–18. Yang K, Zhang H. A comparison of TRIZ and axiomatic design. TRIZ J 2000;8. Amico M, Asl F, Pasek Z, Perrone G. Real options: an application to RMS investment evaluation. In: Dashcenko AI, editor. Reconfigurable manufacturing systems and transformable factories. Springer; 2006. p. 675–93. AlGeddawy T, ElMaraghy H. A model for co-evolution in manufacturing based on biological analogy. Int J Prod Res 2011;49:4415–35. Goyal KK, Jain P, Jain M. A comprehensive approach to operation sequence similarity based part family formation in the reconfigurable manufacturing system. Int J Prod Res 2013;51:1762–76. Galan R, Racero J, Eguia I, Garcia J. A systematic approach for product families formation in Reconfigurable Manufacturing Systems. Robot Comput Integr Manuf 2007;23:489–502. Wiendahl H, ElMaraghy HA, Nyhuis P, Zäh M, Wiendahl H, Duffie N, et al. Changeable manufacturing-classification, design and operation. CIRP Ann Manuf Technol 2007;56:783–809. Bejlegaard M, Brunoe TD, Bossen J, Andersen A, Nielsen K. Reconfigurable manufacturing potential in small and medium enterprises with low volume and high variety: pre-design evaluation of RMS. Procedia CIRP 2016;51:32–7. Bruch J, Bellgran M. Integrated portfolio planning of products and production systems. J Manuf Technol Manage 2014;25:155–74. Andersen A, Rösiö C, Bruch J, Jackson M. Reconfigurable Manufacturing—an enabler for a production system portfolio approach. Procedia CIRP 2016;52:139–44. Tolio T, Ceglarek D, ElMaraghy H, Fischer A, Hu S, Laperrière L, et al. SPECIES—co-evolution of products, processes and production systems. CIRP Ann Manuf Technol 2010;59:672–93. Ulrich KT, Eppinger SD. Product design and development. New York: McGraw-Hill; 2004. Jiao J, Simpson TW, Siddique Z. Product family design and platform-based product development: a state-of-the-art review. J Intell Manuf 2007;18:5–29. Simpson TW, Siddique Z, Jiao JR. Platform-based product family development. In: Simpson TW, Siddique Z, Jiao JR, editors. Product platform and product family design. Springer; 2006. p. 1–15. Xiaobo Z, Wang J, Luo Z. A stochastic model of a reconfigurable manufacturing system Part 2: Optimal configurations. Int J Prod Res 2000;38:2829–42. Xiaobo Z, Wang J, Luo Z. A stochastic model of a reconfigurable manufacturing system-part 4: Performance measure. Int J Prod Res 2001;39:1113–26. Zhao X, Wang J, Luo Z. A stochastic model of a reconfigurable manufacturing system-part 3: Optimal selection policy. Int J Prod Res 2001;39:747–58. Koren Y, Shpitalni M. Design of reconfigurable manufacturing systems. J Manuf Syst 2010;29:130–41. Yigit AS, Allahverdi A. Optimal selection of module instances for modular products in reconfigurable manufacturing systems. Int J Prod Res 2003;41:4063–74. Yigit AS, Ulsoy AG, Allahverdi A. Optimizing modular product design for reconfigurable manufacturing. J Intell Manuf 2002;13:309–16. Koren Y. The rapid responsiveness of RMS. Int J Prod Res 2013;51:6817–27. Maler-Speredelozzi V, Koren Y, Hu S. Convertibility measures for manufacturing systems. CIRP Ann Manuf Technol 2003;52:367–70. Ko J, Hu SJ, Huang T. Reusability assessment for manufacturing systems. CIRP Ann Manuf Technol 2005;54:113–6. ElMaraghy HA, Wiendahl HP. Changeability—an introduction. In: ElMaraghy HA, editor. Changeable and reconfigurable manufacturing systems. London: Springer; 2009. p. 3–24. Terkaj W, Tolio T, Valente A. Focused flexibility in production systems. In: ElMaraghy HA, editor. Changeable and reconfigurable manufacturing systems. Springer; 2009. p. 47–66. Michaelis MT, Johannesson H. From dedicated to platform-based co-development of products and manufacturing systems. In: ElMaraghy HA, editor. Enabling manufacturing competitiveness and economic sustainability. New York, USA: Springer; 2011. p. 196–202.

195

[81] Mortensen NH, Yu B, Skovgaard HJ, Harlou U. Conceptual modeling of product families in configuration projects; 2000. p. 68–73. [82] Kvist M. Product family assessment; 2010. [83] Bi Z, Lang SY, Verner M, Orban P. Development of reconfigurable machines. Int J Adv Manuf Technol 2008;39:1227–51. [84] Chaube A, Benyoucef L, Tiwari MK. An adapted NSGA-2 algorithm based dynamic process plan generation for a reconfigurable manufacturing system. J Intell Manuf 2012;23:1141–55. [85] Azab A, ElMaraghy H. Mathematical modeling for reconfigurable process planning. CIRP Ann Manuf Technol 2007;56:467–72. [86] Kahloul L, Djouani K, Chaoui A. Formal study of reconfigurable manufacturing systems: a high level petri nets based approach. In: Marik V, Lastra J, Skobelev P, editors. Industrial applications of holonic and multi-agent systems. Springer; 2013. p. 106–17. [87] Dai X, Li J, Meng Z. Hierarchical Petri net modelling of reconfigurable manufacturing systems with improved net rewriting systems. Int J Comput Integr Manuf 2009;22:158–77. [88] Kahloul L, Chaoui A, Djouani K. Modeling and analysis of reconfigurable systems using flexible Petri nets. In: Liu J, Peled A, Wang B, Wnag F, editors. Theoretical aspects of software engineering. IEEE Computer Society; 2010. p. 107–16. [89] Goyal KK, Jain P, Jain M. Optimal configuration selection for reconfigurable manufacturing system using NSGA II and TOPSIS. Int J Prod Res 2012;50:4175–91. [90] Youssef AM, ElMaraghy HA. Assessment of manufacturing systems reconfiguration smoothness. Int J Adv Manuf Technol 2006;30:174–93. [91] ElMaraghy W, Nada O, ElMaraghy HA. Quality prediction for reconfigurable manufacturing systems via human error modelling. Int J Comput Integr Manuf 2008;21:584–98. [92] Surbier L, Alpan G, Blanco E. A comparative study on production ramp-up: state-of-the-art and new challenges. Prod Plann Control 2014;25:1264–86. [93] Almgren H. Pilot production and manufacturing start-up: the case of Volvo S80. Int J Prod Res 2000;38:4577–88. Ann-Louise Andersen (M.Sc.) is since 2014 a Ph.D. fellow at the Department of Mechanical and Manufacturing Engineering, Aalborg University in Denmark. She earned a M.Sc. degree at Aalborg University in operations and supply chain management in 2014. Her research area is modular and reconfigurable manufacturing systems for high-speed ramp-up. Her work is affiliated with a national research initiative on how to strengthen manufacturing in Denmark through collaborations between research and industry. Thomas Ditlev Brunoe (M.Sc., Ph.D.) is associate professor at the Department of Mechanical and Manufacturing Engineering, Aalborg University in Denmark. He holds a Ph.D. in product configuration systems from Aalborg University 2008. He is currently working on mass customization, modular and reconfigurable production and is involved in a national initiative on strengthening the Danish manufacturing sector (MADE). Before that, he has six years of industrial experience working on product configuration, product modelling, and product lifecycle management. Prior publications by him have appeared in International Journal of Mass Customization, International Journal of Production Research, and International Journal of Flexible Manufacturing Systems. Kjeld Nielsen (M.Sc., Ph.D.) is working as associate professor at the Department of Mechanical and Manufacturing Engineering, Aalborg University in Denmark. He earned a Ph.D. degree in 2013 at Aalborg University in the area of mass customization assessment and measurements for industrial application. Kjeld Nielsen has since 2009 been a full time academia at Aalborg University and before then he had 25 years of experience as practitioner, as entrepreneur, manager, and business angel in privately own ICT companies. Kjeld has during this gained expert knowledge, skills and competences in project management theoretically as well as practically. Carin Rösiö (M.Sc., Ph.D.) is working as a senior lecturer and researcher at School of Engineering, Jönköping University in Sweden. She earned a Ph.D. degree in Innovation and Design from Mälardalen University. She is currently teaching i.a. production system development and inquiry methodology. Carin has also been working at Sandvik Hard Materials and spent one year as Research Trainee at Luleå University of Technology. Her current research interest includes production system design with a focus on reconfigurability, sustainability and the ability to handle change. Her research also comprises production system localization and mobile production systems.