Keywords

1 Introduction

Flexibility is a very generic term, which can be used to refer to several manufacturing considerations. Sethi and Sethi (1990) claim that flexibility is a complex, multidimensional and difficult concept to define. Beskese et al. (2004) indicate that some terms used for flexibility are the aggregation of others. Moreover, identical terms used by different authors do not always mean the same thing.

Pérez-Pérez et al. (2018) have carried out a remarkable state of the art on production flexibility with more than 330 articles in review and their conclusion is the following: “In recent decades, we have seen considerable literature develop concerned with manufacturing flexibility. Despite the growing interest and the relevant insights of these investigations, previous literature does not clearly bring order and clarity to the academic field that seems to be still fragmented and far from conclusive”. Table 1 illustrates the widely differing definitions used to characterize flexibility in manufacturing systems.

Table 1 Different definitions for flexibility in manufacturing systems
Full size table

In scheduling, flexibility is a widely used concept. For the French Theoretical Scheduling Group and its Applications (Groupe d’Ordonnancement Théorique et ses Applications—GOThA), flexibility represents the degrees of freedom existing in a scheduling system (Pinot 2008). This means that the solution obtained will remain valid if decisions vary within these degrees of freedom. An example is the Flexible Job Shop Problem (FJSP), an extension of the classic job shop scheduling problem which allows an operation to be processed by any machine of a given set. The degree of freedom is then the assignment of the operation to a machine in the set.

For the French standardization organization AFNOR, member of the ISO, flexibility is the ability of a production system to adapt, within a given time, to a variety of products or tasks, i.e. its capacity to adaptation to a wide range of environments (Norme AFNOR X50-310 1991). A flexible system must therefore be able to change in order to adapt to an evolving environment. Accordingly with AFNOR, we consider that flexibility is a response to an ever-changing environment. Thus, flexibility is an external feature. This means that the production system is considered as a black box, operating in an uncertain environment, with the aim of meeting a variable demand. We also do not seek to characterize the flexibility of the system but rather try to define the features that allow the system to be flexible. Thus, we propose the following definition, close to AFNOR’s definition: a flexible production system aims at customer satisfaction by being able to adapt to changes in demand or in the environment, within a defined time frame, while remaining competitive.

This definition will be clarified in the rest of this chapter. The remainder of the chapter is structured as follows: Sect. 2 details the changes drivers within the production system. Section 3 provides explanations about flexibility assessment. In Sect. 4, ways to achieve flexibility are discussed. Sections 5 and 6, respectively, propose a classification and a representation of RMS characteristics. Section 7 concludes this chapter and gives some perspectives.

2 Change Drivers

Flexibility can be summarized as the ability to respond to change. Changes that occur in a production system could either be related to the customer demand or to the production environment. In this section, we will detail each of these change drivers.

2.1 Changes Drivers in Demand

Changes in demand can concern the following aspects.

  • Changes in the product

It consists in creating a new product by modifying features of existing products. The modifications may concern physical features or functionalities of the product.

  • Changes in the manufacturing process

When the customer demands changes in the manufacturing process, we consider it as changes in demand. For example, the customer may request changes in the manufacturing process to improve the mechanical strength of the product.

  • Changes in the lead time

In the case of a make-to-order production, changes in the lead time will have a significant impact on the production system.

  • Changes in the production volume

Production volume has to follow the evolution of the market. For many authors, the production volume is a crucial point of flexibility (Eguia et al. 2017; Koren and Shpitalni 2010).

  • Changes in the quality level

The quality level of a product is often characterized by a nonconformity rate. When the customer demands a higher quality level, i.e. a decrease in the nonconformity rate, it is considered as a change in demand. Chen and Adam (1991) show that the development of flexible production systems has led to an increase in the quality level.

  • Changes in the product price

A reduction in the product price may be demanded by the customer or imposed by the market. However, Van Biesebroeck (2007) indicate that flexibility may increase production costs.

2.2 Changes Drivers in the Production System Environment

These changes may concern the following aspects of the production system environment.

  • Changes in the supply

Suppliers may change their products or even disappear. Duclos et al. (2003) state that “the flexibility of the supply chain includes flexibility in establishing the relationships with partners”.

  • Changes in the availability of technical or human resources

Resources may be unavailable for many reasons (breakdowns, maintenance, absenteeism, etc.).

  • Changes in the legislative or regulation environment

Companies are subject to legislation and regulation, which vary. These changes may impact the manufacturing process.

  • Adaptation time and readjustment cost

Companies, as well as people, need time to adapt to change. This adaptation time is an important feature of flexibility. The adaptation time must be aligned with the evolution of the market. It should be noted that the adaptation can be gradually moving from a state to another. This may also cause a temporary slowdown or a drop in productivity (Huang et al. 2018). For example, the implementation of new resources can lead to a temporary production shutdown.

The readjustment cost characterizes the financial investment required by the company to adapt to changes (Van Biesebroeck 2007). Obviously, the return on investment (ROI) must be acceptable.

3 How to Assess Flexibility?

As said in the introduction, instead of trying to characterize the flexibility of a system, we seek to define the features that allow the system to be flexible. For this, for each of the change drivers identified, we will try to define the flexibility features of the system, i.e. the features that enable the system to be flexible.

3.1 Flexibility Features of the System in Relation to Demand

  • Changes in the product

This aspect of flexibility consists, for the production system, in absorbing changes in the product features (shape, dimensions, etc.). A first idea would therefore be to characterize the ranges of change of a particular feature, which the system can realize, e.g. a window manufacturer is able to produce openings between 0.5 and 2.5 m.

  • Changes in the manufacturing process

To facilitate changes in the manufacturing process, a catalogue of all the production operations that the company can realize will be useful. However, it will not measure the company’s ability to master new processes.

  • Changes in the lead time

To evaluate the impact of changes in the lead time, the minimum time that the production system is able to achieve, between launch in production and the availability of the product, will be a useful measure.

  • Changes in the production volume

The production volume of a product can be characterized by the maximum volume that the production system is able to produce over a given period of time.

Another important feature to characterize changes in the production volume is the increment with which the system is able to adapt to demand. Figure 1 shows that the smaller this increment is, the better the system will be able to accurately track changes in demand.

Fig. 1
figure 1

Productivity increment

Full size image

  • Changes in the quality level

This can be characterized by the lowest nonconformity rate that the company is able to achieve.

  • Changes in product price

For a product, the ratio of the production cost to the price is an indication of the company’s ability to adapt to a price reduction.

3.2 Flexibility Features of the System in Relation to the Environment

  • Changes in the supply chain

Suppliers defection is a significant risk that a flexible company must address. One way to evaluate the ability to cope with this risk would be to measure the redundancy of suppliers.

  • Availability of resources

Two parameters are important in order to characterize this risk, (i) the failure or absenteeism rate and (ii) the resources overcapacity.

Let Oi an operation, Ni the number of resources capable of achieving operation Oi. On a given period T corresponding to the time required to replace a resource, pi is the processing time of Oi, and pi is considered fixed whatever the resource. Resources overcapacity of Oi can be computed as:

$$S_{i} = N_{i} - E\left( {\frac{{p_{i} }}{T}} \right)$$
(1)

where E refers to the excess integer part.

In other words, Si is, for operation Oi, the number of resources capable of replacing a failed (or absent) resource.

  • Changes in the legislative or regulation environment

A company prepared to respond to the risk of legislative uncertainty will be more flexible. For example, a company that has strong international experience or a legal department will be better equipped to cope with this risk.

3.3 Graph of the System’s Flexibility Features

As shown in Fig. 2, features of flexibility influence each other. For example, increasing product diversity will often increase production costs and decrease the production volume. An increase in the quality level will require the evolution from a statistical control to a systematic control of the products. This increase in controls will lead to increased production costs.

Fig. 2
figure 2

Graph of flexibility features

Full size image

3.4 Typology of Companies Regarding Flexibility

Regarding their flexibility, we propose to define three families of companies (Fig. 3).

Fig. 3
figure 3

Typology of companies regarding their flexibility

Full size image

The F1 family concerns companies that manufacture few products, very stable, frequently sold in large quantities, i.e. mass production. For example, the company LU, located in the area of Nantes (France), produces a biscuit, the “Petit Beurre”, which has remained unchanged since 1886. Every year, it produces about a billion biscuits, packaged in 41 million packs.

The F2 family corresponds to companies capable of manufacturing a very wide variety of products, without having significant productivity requirements. Additive manufacturing is a good example for this. The product diversity is almost infinite but with very low productivity.

Most of the companies belong to the F3 family. These companies must constantly seek a compromise between product diversity and productivity. Current market trends, which are increasingly volatile, require them to be able to quickly modify their products (diversity) while being profitable (productivity).

4 How to Achieve Flexibility?

This question, which is crucial, cannot be answered generally but family by family.

4.1 Flexibility and Production System Family

Dedicated production systems (DMSs) are production systems adapted to a specific type of product and a fixed production volume. Highly automated, these systems are capable of high productivity, with relatively low production costs. This type of production system is particularly well suited to F1 family companies. On the other hand, DMSs are not flexible at all. Their structure is “fixed”, i.e. very little or no modifiable (Koren et al. 1999).

Development of Information and Communication Technologies (ICT) has led to the emergence of flexible manufacturing systems (FMSs). These production systems are essentially made of numerically controlled machines (Eguia et al. 2017) controlled by programmable logic controllers (Lafou 2016). FMSs have a general flexibility, corresponding to all the functional possibilities offered by their machines. The variations in the characteristics of the products manufactured by FMS are within a range corresponding to these possibilities. The machines have the ability to produce many different products: a program change enables to switch from one type of product to another. Highly developed in the field of machining, additive manufacturing by further increasing the diversity of manufactured products is nowadays very representative of FMS. Yet, the quantities produced are relatively small, while the investment costs are often high. As a result, their profitability is difficult to achieve. FMSs offer flexibility on the types of products manufactured, but they are limited by their cost and production volume (Koren et al. 1999). FMSs are particularly suitable for the F2 family companies.

These two types of the system represent two extrema for the flexibility. DMSs are not flexible but very productive, while FMSs are very flexible but expensive. The current trend is to move towards companies in the F3 family, i.e. to have production systems that adapt to changes, while remaining profitable. Reconfigurable manufacturing system (RMS) is a third family of production systems adapted to the F3 family.

4.2 Towards RMS

Reconfigurable manufacturing systems (RMS) can respond to changes in the market quickly, at an acceptable cost. To achieve this, RMS combines the productivity of DMS with the flexibility of FMS. Their flexibility is controlled: only what is necessary and nothing more. This means that in response to a change in the demand or in the production system environment, the system adapts to respond to this change. If the quantity to be produced doubles, the system will be configured only to double its production volume. If it is a product feature that changes, the system will be configured to take this change into account. And if a machine in the system fails, the system will be configured to continue production despite the failure.

By offering a solution for companies in the F3 family, RMS is often the only alternative for companies that need to combine flexibility and productivity. In the following, we propose to see how RMS will be able to meet this double objective: flexibility and productivity, while remaining profitable.

4.3 Characterization of RMS

Reconfigurability is the ability to quickly and cost-effectively adapt to changes. RMSs are systems whose physical and logical structures, at all levels, can be changed quickly and at a lower cost to adjust production capacity and functionality around a family of products in response to sudden changes in the market (Koren et al. 1999). Changes in physical and/or logical structure are made in order to satisfy changes in demand and in the production system environment. To achieve this, RMSs are endowed with characteristics that allow them to vary their structure from the outset. Each of these characteristics has a role in the ability of RMS to be flexible, productive and profitable. Eight characteristics are listed in the literature: modularity, scalability, convertibility, customization, diagnosticability, integrability, mobility and adaptability (Koren et al. 1999; Mehrabi et al. 2000; Lameche 2018). The first six are the most frequently cited.

An analysis of these characteristics led us to propose a classification as shown in Fig. 4. Yet, customization does not appear on this classification, while reconfiguration strategy has been added. This is discussed in more detail in the next sections.

Fig. 4
figure 4

Classification of RMS characteristics

Full size image

5 Classification of RMS Characteristics

RMS characteristics occur at different levels of reconfigurability. Some make it easier to detect a need for reconfigurability, others facilitate rapid and efficient reconfiguration and others define the objectives of the reconfigurable system.

We therefore propose to classify the RMS characteristics regarding three aspects of RMS: (i) the objectives of RMS, (ii) the functioning of RMS and (iii) the design of RMS.

5.1 Characteristics Relating to RMS Objectives

Convertibility is the ability to change the functionality of a system to meet new production requirements (Maganha et al. 2018). It can concern machines, the overall structure of the system or the production control system. In the case of machines convertibility, this may correspond to a change of machining axes, a change of tool or a change of fixing. Applied to machines, this feature expands their set of functions, e.g. change of tool. Convertibility has to be controlled by the product diversity. For example, a machine capable of milling and drilling can only realize products that need either milling or drilling. If the evolution of the system requires new production processes, convertibility can lead to a new system structure by adding or removing workstations that provide opportunities for new products. Convertibility can also be achieved by modifying the production control system.

Customization refers to the system’s ability to produce, within a product family, a product with features specific to a particular customer. Customization is one of the causes of changes in the product. Within a product family, customization is controlled by the catalogue of possible variants (Koren and Shpitalni 2010). RMS must be able to produce each customized product, without necessarily change drastically its configuration. For this, the RMS configuration can be customized regarding the ranges of the product family’s characteristics. The RMS customization can be done by using multiple tools on the same machine or by dedicated machines (e.g. a LASER marking station to place the customer’s logo). Yet, customization can be seen as a motivating target of convertibility.

Scalability is the ability of a system to adjust its production volume. This can involve adding or removing technical or human resources. In that case, the adjustment has to be done with the objective to balance the load of each resource. The production volume’s adjustment may also necessitate optimizing the system transport. Indeed, paths between resources have also a great influence on the lead time. The scalability of a system is characterized by the number of resources added and/or removed to expand the overall production capacity, the time required to make these changes and the time spent in transporting products between resources.

Adaptability refers to the system’s ability to be responsive to changes in production volume and product characteristics (Maganha et al. 2018). Adaptability is therefore a combination of scalability and convertibility. The decision to manufacture a new product (convertibility) may depend on the volume of production envisaged (scalability). A company that is able to reconfigure its production system to manufacture a new product at the volume demanded by the market while remaining profitable is an adaptable company. Adaptability can be characterized the minimum volume for which the system can be reconfigured.

5.2 Characteristics Related to the Operation of RMS

Numerous problems can occur during the operation of a production system: sudden disruptions, e.g. machine failure, or progressive disruption, e.g. a rate of nonconformity that increases slowly. The causes of the problems that arise must be quickly identified and an appropriate strategy implemented. We must therefore answer two questions: where and how to act. This requires a system diagnostic in order to decide whether to reconfigure if necessary. Hence, another characteristic of the operation of RMS appears: the reconfiguration strategy. Indeed, according to diagnosis, this characteristic allows to quickly address the problem.

Koren and Shpitalni (2010) define diagnosticability as the ability to automatically read the system status to detect the causes of failures so that they can be quickly corrected. This feature allows failures to be located and processed in such a way that the system is not penalized. To achieve this, it is necessary to adopt a reconfiguration strategy specific to the situation. Depending on the causes of failure observed, the system may be slightly or deeply modified. We introduce here the notion of the level of reconfiguration, what Ketfi (2005) and Kanso (2010) called reconfiguration category:

  • Level 0 consists in acting without stopping the system. This level of reconfiguration does not require any changes in the system structure. At this level, the system can be corrected during operation, or during a very short downtime, e.g. a sensor that no longer works or a pallet stuck on a conveyor. Level 0 of the reconfiguration is actually a correction of the system, without having to modifying it. It is a matter of bypassing the system’s failure to keep it running, e.g. in the case where two machines perform the same function if one fails the other takes over. It is a corrective reconfiguration, which is done with minimal effort.

  • Level 1 consists of modifying the system to meet changes in demand or in the system environment. This level of reconfiguration requires changes in the functionalities of the system: either by adding components already available or by repositioning/reassigning components. This reconfiguration involves changes in the system structure, e.g. a machine added or removed from the system or a truck moved from one area to another. The reconfiguration will be done when the system is down. In fact, it is an evolution of the system, since it moves from one operating state to another. It is an evolutionary reconfiguration, which requires more effort than level 0. The reconfiguration effort can be determined by the cost and time required to complete a reconfiguration (Huang et al. 2018).

  • Level 2 also consists of modifying the system, but with a greater effort. For example, you can add a component, specially designed for reconfiguration, to the system, e.g. a trapezoidal pallet holder that does not exist in our resource store.

Diagnosticability is used to determine the level of reconfiguration. It is also used to determine whether the reconfiguration concerns the machine or the system level. By measuring the reconfiguration effort, diagnosticability allows different reconfigurations to be evaluated. These evaluations can be done using a simulation tool. Diagnosticability is the characteristic allowing the triggering of a reconfiguration. Indeed, it gives, at any time, information about the state of the system, allowing a diagnostic to be set and reconfiguration strategies to be evaluated.

5.3 Characteristics Related to the Design of RMS

Modularity, integrability and mobility are characteristics related to the design of RMS. They allow the deployment of the configuration strategy chosen at the end of the diagnostic phase.

Modularity describes the use of standard and interchangeable units to satisfy a variety of functions (Lameche et al. 2017). In a modular system components are completely separated from the system so that the addition, replacement and/or modification of a module (component) are possible (Benderbal 2018). This characteristic of RMS enables changes in machines functions or changes in the production volume. Modularity can be controlled by the number of independent modules available in the system.

The choice of the granularity of the modules during the design phase of the RMS is very important. If the modules are complex, there will be few. In that case, the increment will be important during a reconfiguration but the number of possible module layouts relatively low. On the other hand, modules may be simple, with limited functionalities, but in large numbers. This will increase the number of possible layouts of the modules. But as there will be more interfaces between the modules, the reconfiguration of the system will be more complex. We must find a balance between the size of the modules and the number of interfaces.

Integrability is the ability to integrate modules with each other. This is possible through interfaces common to both parties (Koren et al. 1999). The interfaces can be physical or logical in order to facilitate communication between the different elements (Lameche 2018). Standard interfaces facilitate module interchangeability and reduce reconfiguration time and costs. For example, for a robot with a removable and standard tip, it will be easier to put a two-finger clamp in place of a four-finger clamp if these clamps have interfaces compatible with those of the robot tip. Yet, if the interfaces are not standardized, new grippers will have to be manufactured, which will lead to delays and additional costs for the reconfiguration. Integrability is particularly important in the case of a level 2 reconfiguration where a new module must be designed. The correct definition of interfaces will facilitate the design and integration of this new module with existing modules.

Lameche (2018) defines mobility as the ability to move products through the system. It is based on the use of transport resources (conveyors, robots, AGVs, etc.). It can be characterized by the number of products transported per time unit or the number of possible paths from a point to another. A system reconfiguration can consist in modifying the path of a product, either to save time or to add an operation in the manufacture process of a product. This route definition can be done by using the elements already present in the system or by adding new transport resources, e.g. adding an AGV to reduce transport time. In the case of programmable transport resources, e.g. AGV or robots, it is necessary to have a logical integration between the elements already present in the system and those added. For example, the control system of a new AGV must be compatible with the one already used in the system.

5.4 Relationship Between the Characteristics of RMS

As shown above, there are three groups of RMS characteristics: characteristics related to the objectives of the reconfiguration, characteristics related to the reconfiguration decision and characteristics related to RMS design. These different characteristics interact with each other (Fig. 5). Thus,

  1. (i)

    a good diagnostic of the system will lead to a good choice of reconfiguration strategy, which will reduce the reconfiguration effort,

  2. (ii)

    the more independent the modules, the more it will be possible to define new functions for the system and, consequently, more products to manufacture.

Fig. 5
figure 5

Graph of reconfigurability characteristics

Full size image

Modularity promotes convertibility and customization. Modularity is facilitated by the presence of standardized interfaces, i.e. integrability.

6 Proposition of a Representation of the Characteristics of RMS

In order to visualize the characteristics of RMS, we propose to use an  eight-dimensional Kiviat diagram as shown in Fig. 6.

Fig. 6
figure 6

Kiviat diagram of RMS characteristics

Full size image

The difficulty remains the metric(s) used to characterize each of these eight characteristics. Many propositions are made in the literature. Sethi and Sethi (1990) have proposed an assessment of the different aspects of flexibility. About scalability, Deif and El Maraghy (2007) propose a model for capacity scalability in RMS, based on a system dynamics approach. Concerning convertibility, Eguia et al. (2017) propose an approach based on data envelopment analysis (DEA). They suggest the use of an MMF DEA model (multiple modes of functioning) to evaluate the effectiveness, especially for the convertibility of RMS. Other authors propose approaches based on multicriteria decision to synthesize the characteristics of RMS: Maier-Speredelozzi and Gu (2002) proposal is based on AHP, Lateef-Ur-Rehman and Rehman (2013) proposal on PROMOTHEE.

All these works prove that it is possible to quantify the eight characteristics of RMS. It is then to be possible to accurately represent the reconfigurability of a production system. Figure 7 presents how the Kiviat diagram could be used to compare different manufacturing systems. For FMS (Fig. 7a), convertibility is highly developed, but other aspects of reconfigurability are not very present. For DMS (Fig. 7b), none of the characteristics of the reconfigurability is developed. Figure 7c shows an example of a reconfigurable system. Finally, Fig. 7d shows an example of a good reconfigurable system but with gaps for the diagnosticability and reconfiguration strategy. These two axes are important axes of future research, but significant advances in Information and Communication Technologies (ICT) concerning these two axes should allow the development of RMS.

Fig. 7
figure 7

Examples of  Kiviat diagrams for manufacturing systems

Full size image

7 Conclusions

This chapter has proposed a new definition for flexibility. The proposed definition enriches what already existed, and its analysis led to a better comprehension of elements that influence flexibility. However, some of the change drivers mentioned in this chapter are difficult to quantify.

Depending on their productivity and their flexibility, three types of production system can be defined: dedicated manufacturing systems (DMSs) which are very productive but not flexible, flexible manufacturing systems (FMSs), which (as the name implies) are very flexible but not very productive and reconfigurable manufacturing systems (RMSs) which are a good compromise to achieve both flexibility and productivity. RMSs have features that allow them to adjust their capacity and functionality exactly when and where they are needed.

From the literature, we have listed eight characteristics for RMS. We have enriched this list by proposing a new characteristic for RMS: reconfiguration strategy. All these characteristics have been discussed, and we have classified them into three families. Finally, we have proposed an approach based on Kiviat diagram to compare RMS.

Thanks to the new characteristic, i.e. reconfiguration strategy, we have identified three different levels of reconfiguration. In the future work, we will propose, for each level, a model for reconfigurability analysis. Thus, the choice of configurations will be made based on an analysis that is specific to a reconfiguration level, which will certainly facilitate decision-making for the configuration.