Master Thesis Exploring Integration Problems of Process Planning and Scheduling and the Potential of a Digital Twin Model for Firms with a Flexible Manufacturing System

(1)

1

Master Thesis

Exploring Integration Problems of Process Planning and

Scheduling and the Potential of a Digital Twin Model for

Firms with a Flexible Manufacturing System

Anand Super (s2761378)

MSc Supply Chain Management

University of Groningen

18/03/2019

(2)

2

Abstract

In today’s highly competitive market, being able to produce high variety/low volume series with a Flexible Manufacturing System (FMS) can create many opportunities for a firm to stay competitive. In practise however, traditional production planning & control (PP&C) systems are not able to handle the increased complexity of scheduling in a dynamic production environment. Furthermore, different FMS scheduling optimisation methods have been proposed in literature, but most are too complex or too inefficient to be used in practise. To effectively deal with this environment and the complexity of the FMS, firms are often using a three-level decision-making hierarchy to make planning and scheduling decisions related to the FMS. Process planning and scheduling decisions are often made at each level separately and sequentially, which could result in a lack of integration between the decisions made. This paper explores what aspects of a PP&C system affect the degree of IPPS, and what the effect of a lack of IPPS has on the work floor. Additionally, the potential of a Digital Twin Model to improve IPPS is researched. A multi-case study is performed at two Dutch sheet metal working firms to analyse the PP&C system and the integration of an FMS in this system. The results indicate that while PP&C system does allow for some degree of IPPS for the FMS specifically, the overall IPPS is still low. This results in some problems on the work floor like the FMS producing too much and production starting too late. Furthermore, to improve IPPS, the Digital Twin Model has the potential to facilitate the information exchange between process planning and scheduling.

(3)

3

Preface

The last couple of months of working on this project has been challenging but rewarding work. It has been a great learning experience and I am proud to present the finished version, which you are reading now.

But first, I would like to thank several people for helping me making this thesis possible. First, I would like to thank dr. J.A.C. Bokhorst, my thesis supervisor, who provided excellent feedback and helped steer this project in the right direction. I would also like to thankdr. ir. T. Bortolotti for providing feedback on my proposal.

(6)

6

Abbreviations

FMS - Flexible Manufacturing System

IPPS – Integration of Process Planning and Scheduling PP&C – Production Planning & Control

(7)

7

1. Introduction

Ever since the early 1970s, Flexible Manufacturing Systems (FMS) have been an important subject in manufacturing related research because of its great potential for offering both customized and cost-effective manufacturing (Karsk, 2002). Researchers’ interest in FMSs was particularly high during the 1980s and early 1990s, but as time went on, this interest slowly went down. But recently, interest in FMSs started to gain the attention of researchers again, and this may be due to technical advances that can solve some FMS related issues.

An FMS can be defined as an integrated, computer-controlled complex of automated material handling machines and numerically controlled machine tools that can process a variety of part types simultaneously (Stecke, 1983). The largest application of these systems is in small batch production where its efficiency is getting near to the mass production efficiency (Kostal & Velisek, 2010).

However, in practise it is often difficult for firms to justify the high initial investment required to purchase an FMS due to the risk of operational problems when it is not properly designed or managed. Since an FMS is often used as a shared resource by multiple value streams, which adds more complexity to the planning of production, traditional design and management methods may not apply. Mismanagement of an FMS can lead to production hold ups which can also affect other value streams (Banaszak & Krogh, 1990). A value stream can be described by all actions (both value-added and non-value added) required to produce a product, starting from raw materials and ending when the product is received by the customer (Rother & Shook, 2003). Raj, Shankar, & Suhaib (2007) discussed several of these issues encountered by scholars and practitioners regarding the planning, design and implementation of an FMS. The authors concluded that currently there is a large gap in the literature between the proposed approaches/algorithms for different components of an FMS and real-life complexities.

(8)

8 Also, even though there has been much research over the years on maximizing FMS performance, these methods often do not take into account both up- and downstream processes of the value chain. Considering that the FMS is often part of multiple value chains, an optimal solution to maximize the utilization of the FMS may not be optimal in terms of firm performance. Customer satisfaction is also an important factor for the success of a firm. Maximizing production effectiveness to meet customer expectations, while also trying to maximise efficiency to reduce costs, can lead to conflicting objectives (Okongwu et al., 2012).

To deal with the added complexity of the FMS and the dynamic environment in which firms using this machine generally operate in, the FMS decision-making hierarchy in the PP&C system are often structured is three levels (Slomp, 1993). However, the decisions made at each level of this hierarchy are often made separately and sequentially, which can result in a relatively static system which is not able to quickly respond to disturbances. The separate and sequential decision-making can result in a lack of in integration of process planning and scheduling (IPPS) decisions, which can have a negative effect on the overall work floor performance (Lihong & Shengping, 2012). It is therefore important for firms to improve the degree of IPPS to get the most out of the FMS. One way of doing this is by increasing the information exchange between the different levels of the decision-making hierarchy (Phanden et al., 2013).

With the emergence of industry 4.0 technology, new opportunities arise to improve the information exchange between the different levels of the decision-making hierarchy. The Digital Twin Model applies several industry 4.0 technologies to create a digital representation of a firm’s system using historic, real-time, and behaviour data which allows for example for autonomous production decisions to be made (Negri et al., 2017). While research on this topic is still very limited, especially in the context of an FMS, it has the potential of providing a substantial competitive advantage to a firm.

Based on this, the following research question is formulated: What factors of the PP&C

system causes a lack of IPPS and what is the potential of a Digital Twin Model to improve IPPS?

In this paper two firms are analysed; both are Dutch sheet metal working firms who use at least one FMS in their production process. Different data collecting methods will be used, but mainly semi-structured interviews to collect most of the necessary data.

(9)

9 by the discussion in the fifth section. Conclusions and limitations can be found in the final section.

2. Theoretical Background

An FMS consists of two parts, the physical parts and the control part (Ruiz et al., 2009). The physical parts are composed of the physical resources, and the control part is used to determine how the physical parts are organized and optimized (Ruiz et al., 2009). Generally, the main goal for an FMS is to achieve both flexibility and productivity while meeting the fluctuating demands of today’s competitive market (Kim et al., 2012). This flexibility includes being able to vary on quantities and varieties of part types to produce (Raj et al., 2007). However, successfully meeting these objectives is highly dependent on how the FMS is controlled.

In the next sub-sections, the characteristics of an FMS are discussed first (2.1). This is followed by challenges associated with FMSs (2.2). The third sub-section discusses the potential of more technologically advanced PP&C systems (2.3). And finally, the conceptual model is discussed (2.4).

2.1 Characteristics of an FMS

The commonly used definition often describes an FMS as an integrated, computer-controlled complex of automated material handling machines and numerically computer-controlled machine tools that can process a variety of part types simultaneously (Stecke, 1983).

Because an FMS is able to use a wide range of tools which can quickly be changed, it has high product flexibility, meaning that it can economically produce a wide set of different products. Besides this type of flexibility, depending on the configuration of the FMS, there are other types of flexibility. Browne et al. (1984) argued that an FMS is not just called flexible when it just produces a wide variety of parts on a fixed operation line, but instead described eight different types of flexibility which an ideal FMS would possess.

(10)

10

FMSs generally require a high initial capital investment because the systems commonly consist out of expensive resources, like for example CNC machines (Koren & Shpitalni, 2010). Because of the flexibility of an FMS, firms often share the FMS with multiple value streams to maximize its utilization to make an FMS economically viable. A value stream can be defined as all the actions required (both value adding and non-value adding) to bring a product through the production flow, from raw materials into the arms of the customer (Rother & Shook, 2003). Within this production flow, there is not only a movement of materials through the factory, but also the flow of information and both are important when discussing value streams (Rother & Shook, 2003).

The collection of the different value adding activities of each value stream is how the firm generates its revenue. It is therefore important that value streams do not impede each other when multiple streams go through the same FMS simultaneously. However, literature of FMS optimization is mainly focused on solving problems directly related to the FMS, while value chain differences are not accounted for. Due to the complexity of an FMS, only a select number of problems are minimized in the PP&C system, which will be discussed in the next section.

2.2 Challenges associated with FMSs

While theoretically, an FMS should provide the benefits of more manufacturing flexibility and efficiency, in practise this is not always the case. A variety of operational problems have been identified which affect the performance of the FMS. According to Tiwari & Vidyarthi (2000), these problems can be divided in to pre-release and post-release decisions. The FMS planning problem is the pre-release decision, where the pre-arrangement of parts and tools before the FMS production process begins are considered. Post-release decisions are FMS scheduling problems, which deal with routing and sequencing of parts when the FMS is operational (Tiwari & Vidyarthi, 2000).

(11)

11 2.2.1 FMS PP&C system

Slomp (1993) provided a general description of a decision hierarchy of a production control system of an FMS, and the three decision levels are: assignment level, off-line level, and on-line level.

As described by Slomp (1993), at the assignment level realizable throughput times and realistic workloads for the FMS are generated. The off-line level receives the orders from the assignment level, and the responsibility of the off-line level includes scheduling a good fit between the important characteristics of an FMS and the received orders. Decisions at this level include for example the batching of certain orders and the assignment of operations and tools to workstations. Activities of the on-line level are based on the information from the off-line level and the actual status of the FMS. Decision include for example when orders are released to an FMS and deciding on the sequence of competing activities which have to be performed by an FMS.

However, this type of relatively static and centralized scheduling system does have some drawbacks in certain environments. Even though this system may provide a globally better schedule in an environment where there are not many disturbances, it has shown to be inefficient in a highly dynamic environment with many disturbances (Ouelhadj & Petrovic, 2009), in particular at the on-line level. These disturbances typically are real-time events which can make the schedule inefficient or insufficient and can be divided into real-time events related to resources, and events related to jobs (Ouelhadj & Petrovic, 2009). An FMS is generally implemented in a highly dynamic manufacturing environment, and this type of static shop floor control and scheduling is therefore most likely not as efficient and effective as it could be.

Additionally, according to Raj et al. (2007), an FMS environment adds further complexity and difficulties for scheduling, due to the usage of versatile machines which are able to perform different operations resulting in many routing options. Current scheduling methods struggle to find both reliable and practical schedules that can deal with this type of environment.

(12)

12 2.2.2 Scheduling Challenges

As mentioned before, production scheduling needs to deal with problems related to for example routing and the sequencing of parts (Tiwari & Vidyarthi, 2000). The following problems are identified (Stecke, 1983, Tiwari & Vidyarthi, 2000):

• Part type selection • Machine grouping

• Determination of product ratio • Batching of part types

• Allocation of pallets and fixtures

• Allocation of operations and tools among machines (loading problem)

Stecke (1983) suggests that these problems can be solved sequentially, or alternatively, candidate solutions can be generated repeatedly to find a suitable solution. While many techniques have been proposed for controlling and integrating the different components of the FMS, most of these controls are found to be too complex in practical FMS implementations (Raj et al., 2007). These problems are connected by common restrictions like available machine types and tool magazine size, but most researchers treat the previous mentioned problems separately for the sake of simplicity (Chan & Swarnkar, 2006). A wide array of solutions for each of the previously mentioned problems have been researched in past literature. Solutions range from non-linear mixed integer programming models to advanced algorithms (Kim et al., 2012).

These mathematical problems are usually not solved by directly attempting to maximize production, instead surrogate objectives are formulated (Stecke, 1983). For example, in solving the loading problem to maximize throughput for an FMS, minimizing the FMS imbalance is commonly used as a major criterion for optimization of the loading problem (e.g. Tiwa Ri et al., 1997).

(13)

13 (Okongwu et al., 2012). However, maximizing production effectiveness to meet customer expectations while also trying to maximise efficiency to reduce costs, can lead to conflicting objectives (Okongwu et al., 2012).

It is therefore important to align the firm objectives with FMS related scheduling and operational decisions. However, current scheduling methods are likely not able to effectively deal with the added complexity of considering the whole system, and often do not take into account customer due date performance. Additional challenges are that machines in SMEs typically include a wide range of machines, ranging from very old to more advanced machines (Solvang et al., 2012). This can cause machine/software compatibility issues and result in no inter-machine communication where a machine is operated in a standalone mode (Solvang et al., 2012). It is therefore difficult to integrate the FMS with other systems that are operating in the same firm (Grieco et al., 2001).

2.2.3 IPPS Problem of Firms With an FMS

(14)

14 done by non-expert planners and cause an ineffective collaboration between process planning and scheduling. (Bensmaine et al., 2014).

As discussed in 2.2.2: due to the complexities of the FMS and the dynamic environment in which these machines operate, it can be difficult to make an accurate production schedule. This results in these firms often using a three-level decision-making hierarchy (which was discussed in 2.2.1) to perform the process planning and scheduling function for the FMS. Because these activities are performed separately and sequentially in three different levels instead of two, it potentially decreases the IPPS even more which can result in problems like a decrease in flexibility of the PP&C system and a decrease in the ability to benefit of the flexibility which an FMS is able to offer to adequately respond to disturbances on the work floor. It is therefore in particular important for firms with FMSs to improve IPPS. In literature, an approach to increase IPPS is to increase the information exchange between the process planning and scheduling departments (Phanden et al., 2013). More advanced PP&C systems which allow for this increased information exchange will be discussed next, which could improve IPPS.

2.3 Advanced PP&C Systems

Production Planning and Control Systems are a crucial tool for a firm to meet the increasing demand of customers in the current competitive manufacturing climate (Stevenson et al., 2005). Functions of a PP&C system include for example the planning of material requirements, capacity planning, the scheduling of jobs, and choosing the right PP&C systems to perform these critical functions is therefore a crucial strategic decision (Stevenson et al., 2005). As suggested earlier, PP&C systems which are used by firms with FMSs may lack IPPS, which can result is several problems. Two examples of PP&C systems which utilize more advanced technologies and have the potential of benefiting a firm with an FMS are described next.

2.3.1 Dynamic Scheduling System

(15)

15

interacting with each other through the definition of appropriate rules in a given environment (Barbati et al. 2012). These agents are autonomous and are attached to physical or functional manufacturing entities in the facility, where each agent is responsible for their own local schedule and to cooperate with other agents to achieve a global optimal and robust schedule (Ouelhadj & Petrovic, 2009). A multi-agent system is recognized to be an effective tool to increase IPPS because it allows to perform process planning and production scheduling simultaneously (Phanden, Jain, & Verma, 2011). Additionally, multi-agent technology is argued to address IPPS problems by implementing dynamic production control geared to handle variabilities in incoming quantities, like machine breakdowns or materials delays (Lee et al., 2008).

However, as noted by Barbati et al. (2012), agent-based approaches do not always provide the best quality of solutions. The authors argue that agent-based solutions allow for dividing problems into sub-problems, but they often lack a global view of the state of the system, which is necessary to find a good solution. Additionally, the communication and negotiation between agents is not very efficient and can take a long time, especially for problems of a larger scale (Lihong & Shengping, 2012).

2.3.2 Digital Twin Model

Related to the multi-agent concept, the Digital Twin Model has the potential of providing a global view of the state of the system in addition of dividing planning related problems into sub-problems. With the emergence of industry 4.0, cyber-physical-systems present the opportunity for multi-agent systems to become smarter and make more accurate decisions with the use of for example, Internet of Things, big data, and cloud computing (Wang et al., 2016) and increases IPPS. Smart objects for example can be reconfigured dynamically to achieve high flexibility, while big data can be used for giving global feedback and coordination to result in high efficiency (Wang et al., 2016). One of the main concepts of cyber-physical-systems is the Digital Twin Model. This model can be defined as a:

“… virtual representation of a production system that is able to run on different simulation

(16)

16

features to forecast and optimize the behavior of the production system at each life cycle phase in real time” (Negri et al., 2017, p.946).

In contrast with the traditional multi-agent models, a Digital Twin Model allows for decisions to be made not only based on real-time data of the physical system, but it can also consider the system’s historical data and predict the future state of the system using AI (Negri et al., 2017). This allows for more accurate decision to be made.

Within the definition of the Digital Twin, Kritzinger et al. (2018) proposed three sub-categories: Digital Model, Digital Shadow, and the Digital Twin. According to these authors, these terms are often used synonymously, but the given definitions differ in the level of integration between the physical object and the digital counterpart. Some digital representations are modelled manually and not directly connected to any physical objects like an FMS, while other models are fully integrated with real-time data exchange. This difference in automation of the data flow is illustrated in figure 2.1. The remainder of this paper specifically makes this distinction in data flows when discussing the Digital Twin Model.

Figure 2.1 Data Flow between the digital and physical objects (Adapted from Kritzinger et al., 2018)

(17)

17 2.4 Conceptual Framework

Figure 2.2, Conceptual Framework

To answer the research question “What factors of the PP&C system causes a lack of IPPS and

what is the potential of a Digital Twin Model to improve IPPS?”, the following conceptual

model is developed (Figure 2.2).

Based on the theoretical review, different aspects of a firm’s PP&C system have been identified which can influence the degree of IPPS for a firm with an FMS:

Scheduling Decision Making Hierarchy: The way the decision-making hierarchy is structured

can potentially influence the degree of IPPS integration. For example, more information sharing between the process planning and scheduling functions in the decision-making hierarchy is likely to increase the IPPS of an FMS (Phanden et al., 2013).

Consideration of FMS Characteristics: At what levels of the decision-making hierarchy FMS

(18)

18 lead to the need for many adjustments at the lower level of the decision-making hierarchy and can negatively affect the degree of IPPS.

FMS Optimization Method: The type of optimization method used, and who is responsible

for FMS optimization may result in conflicting objectives between the production planning and scheduling function of an PP&C system. If overall firm objectives are not adequately accounted for in the FMS optimization method, it could have a negative effect on the IPPS.

Consideration of Value Stream Characteristics: The FMS is often part of multiple value

streams as discussed in 2.1. Whether or not the different value streams are considered at the different levels of the decision-making hierarchy could affect IPPS. If for example at the higher level of the decision-making hierarchy, the firm perspective is taken and the objective is to maximize the most important value streams, while at lower levels this is not considered and the focus in only on optimization of the FMS, it can create conflicting objectives in the planning and scheduling decisions.

FMS Flexibility: An FMS has the ability to provide several types of flexibility as discussed in

2.1. But to fully benefit from these types of flexibility, like routing flexibility, it has to be accounted for at all levels of the decision-making hierarchy. Otherwise, if for example changes are made in the routing for the FMS, it may affect other machines as well and that may require production planning and scheduling adjustments. If the PP&C system is not able to account for these changes, it can indicate that there is a lack of IPPS.

The lack of IPPS can result in several problem on the work floor as discussed in 2.2.3, such as an inflexible PP&C system. This can potentially create inefficiencies on the work floor and ultimately affect firm performance. In addition to identifying problems caused by a lack of IPPS with the current PP&C system that firms with FMSs use, the potential of a Digital Twin Model is researched to improve the IPPS.

3. Methodology

(19)

19

can be achieved through observing the actual practice (Meredith, 1998). A multi-case increases the generalizability of the results. Therefore, a multi-case study as a research method is the appropriate method for this paper.

The research can be divided into two parts. In the first part, a multi-case study is performed at two firms, exploring how the PP&C system is structured and how the FMS is integrated in this system. Additionally, an attempt is made to generalize these findings. The second part of the research is more conceptual. Based on the results of the first part of the research and the literature review of the Digital Twin Model, suggestions are made on the potential of a Digital Twin Model. An attempt to validate the findings was done at one case.

3.1 Case Selection

The units of analysis of this paper are small and medium sheet metal working firms using FMS(s) in their production process. These types of firms are relevant to this paper, because they generally operate in a complex high variety, low volume production environment where production planning and scheduling for an FMS is likely to be challenging. To increase the generalizability of the results, two cases were selected. To be able to predict similar results for both cases, literal replication was used for the case selection. Both cases where selected because of these following characteristics: the firm has at least one FMS, is a SME, and operates in a similar industry and environment as the other case. This resulted in the selection of two Dutch sheet metal working firms, one small-sized firm with one FMS, and a medium-side firm with four different FMSs. See 3.5 for a more extensive description of these firms.

3.2 Data collection

(20)

20

To validate these findings, triangulation of different data sources is used. The other methods include field observations, analysis of internal documents, and informal instructed interviews. Field observations includes noteworthy observations of how the FMSs are being operated, as well as observations of problems experienced on the work floor. Internal documents included ERP-output and customer performance data to analyse value stream and firm performance. The additional informal interviews allowed for finding relevant questions for interviews, but also to validate results of the semi-structured interviews.

Each firm was visited for five days to collect data. During each site visit, a database for each firm of all collected raw data was created. This database included all noteworthy field observations that were observed at this firm, notes from informal unstructured interviews, internal documents, and electronically recordings of the semi-structed interviews, and the transcripts of these recordings. This database is available on request. This information was condensed in a summary report for each firm (Appendix I & II), and the firms were given the opportunity to make comments or changes on this report.

Additionally, Firm B was visited again to discuss the findings of the summary report and to validate the concept of a Digital Twin Model.

Figure 3.1 Conducted Semi-structured Interviews

3.3 Data analysis

For the first part of the research, the PP&C system of each firm was described and analysed based on the data which was collected. This was followed by a cross-case analysis in an attempt to generalize findings and make conclusions. In the second part of the research, based on the problems described in the first part of the research, the potential of a Digital Twin Model was analysed. The concept of a Digital Twin Model was discussed with one firm in an attempt to validate to concept.

(21)

21 3.4 Case Description

3.4.1 Firm A

Firm A is a small sheet metal working firm with around 25 employees and a yearly turnover of around €2.5 million. Production ranges from producing single pieces to small series, varying from small parts to larger constructions. Their competitive focus is on offering a high level of service by being able to produce a wide range of products on a relatively short notice and delivering the order on time. The main production activities include laser cutting, bending metal parts with 2 press brake machines, and welding. The flexible manufacturing system in this case is the laser cutting machine. This machine is called the Mazak HyperGear HG510 which is able to cut steel sheets with a maximum size of 3000 x 1500 mm. This machine can be described by the classification of different FMS subtypes by MacCarthy & Liu (1993) as a single flexible machine. It is a single CNC machine with computer-controlled tool changing capabilities, however material handling is done manually by an operator.

3.4.2 Firm B

(22)

22

Figure 3.2 Machine Characteristics

These machines produce during the manned shift of 8 hours but are also able to run during the unmanned night as long as there are enough materials in the magazine, room for the finished products, enough tools in the magazine, and no problems like breakdowns.

4. Results

First, a within case analysis will be performed for the two cases, where the different concepts of the conceptual model will be discussed to indicate what factors cause a lack of IPPS. This is followed by a cross-case analysis to attempt to generalize the findings. After that, the potential of a Digital Twin Model will be explored to improve IPPS.

4.1 Case Analysis Firm A

4.1.1 Scheduling decision-making hierarchy

As can be seen in Appendix I, the scheduling decisions-making hierarchy in the PP&C system at this firm can be divided into three levels. At the first level, the production planner is responsible for production planning and also, to some degree, for production scheduling. The planner is responsible for accepting new orders, scheduling the order in the ERP-system, and indicating which order gets priority, and negotiating a delivery date with the customer. All machines are often scheduled to operate at maximum capacity. However, the production schedule is not fixed, it is often adjusted by the production planner to account for new orders, changes in priorities, changes on the work floor, and changes in customer due dates.

At the second level, the programmer of the FMS is responsible for production planning and scheduling, to some degree, for the FMS specifically by determining what orders will be batched and cut from the same metal sheet and when it should be released to the FMS. A

Machine Characteristics Punching machine

(Trumatic 5000R)

Punch Laser Combination Machine (Trumatic 7000)

CO2 Laser machine (Bystronic Byspeed)

Fiber Laser machine (Trulaser 5030 Fiber)

Automatic sheet loading and unloading

system T T T T

Automatically separates parts from the

metal sheet T T

Quickly able to switch between the laser

and punching attachment T

Limited tool magazine size T T

(23)

23 batch of orders with parts off a certain metal thickness are executed by the operator twice a week and these batches include orders that have to start production within three weeks. However, when there is some excess capacity available on the work floor, the production planner sometimes decides to plan some orders forward in such a way that these could be included in these batches as well. The operator of the FMS is responsible for producing the batches and has the freedom to choose the sequence of production to prevent a changeover. Overall, the way how the decision-making hierarchy is structured allows for only some integration between production planning and scheduling. While the production planner is responsible for both production planning and scheduling, the schedule is often changed at the lower levels of the decision-making hierarchy to fit the production schedule with the actual status of the work floor. Because these planning related decisions are made separately and sequentially in the decision-making hierarchy, it negatively affects the integration between production planning and scheduling.

4.1.2 Consideration of FMS characteristics

As described at 3.4.1, the CO2 laser cutting machine is relatively simple for an FMS. Tools can be changed automatically and there is no tool magazine size limitation. Furthermore, material handling is also done manually by the operator.

Nevertheless, the production planner only considers the available capacity of the FMS when scheduling. The FMS programmer does take FMS characteristics into account when deciding which orders can be batched based on the type of materials and the tools needed. The FMS operator mainly considers tool limitations when sequencing the production of batches of orders in such a way that changeovers are minimized.

(24)

24

4.1.3 FMS Optimization Methods

The FMS was optimized by the programmer in terms of efficiency by batching as many orders as possible on the same metal sheet. The intend was to limit the amount of metal sheet changes the operator had to perform. Additionally, the operator of the FMS has the freedom to choose the sequence in which the batches are produced to potentially reduce the number of tool changes required.

FMS optimization is therefore mainly done at the lower levels of the decision-making hierarchy. However, these optimization related decisions are not incorporated into the production schedule which was generated by the production planner. Therefore, because there is lack of communication between the different levels on what FMS optimisation related decisions are made, it negatively affects IPPS.

4.1.4 Consideration of value stream characteristics

Each customer order is unique to that customer. Therefore, each customer can be seen as a separate value stream. Customer specific value stream characteristics are accounted for by the production planner. As stated by the production planner: “We have an Excel sheet

with our top 10 best customers, A-customers, B-customers, C-customers. We already try during order entry to give the right customer the right priority. Customers who bring a lot of orders and revenue during the year, we do not push aside for customers who only order once or twice in a year, they just have to wait longer if needed.”

The different levels of priority of each order is a factor when the FMS programmer and operators decide which orders to batch or produce first. However, their main objective is still on maximizing the efficiency of the FMS. This can result in conflicting objectives for planning related decisions between the production planner and programmer/operators, and the plan made by the production planner may not fully align with the actual schedule which is executed on the FMS, which negatively affects the degree of IPPS.

4.1.5 FMS Flexibility

(25)

25 minutes.

While the FMS has some product flexibility, the way how to PP&C system is structured does not allow to leverage this flexibility. The production planner does not consider FMS flexibility, the generated production schedule assumes that each order requires a changeover. Additionally, the programmer and operator of the FMS are primarily focused at maximizing the efficiency of the FMS by producing as many similar orders in a sequence to limit changeovers. However, considering that the FMS is usually not the bottleneck, this is not always needed, and excess capacity is available to benefit from the product flexibility of the FMS. A higher degree of IPPS would potentially allow to better balance efficient use of the FMS while also benefiting from the product flexibility of the FMS.

4.1.6 Degree of IPPS

Due to the complexity of the FMS and the dynamic environment in which this firm operates, many FMS production schedule related decisions are made by the FMS programmer and operator that aim to maximize the efficiency of the FMS. Even though the production planner is responsible for making the production planning and schedule, many FMS related planning and scheduling decisions are made at the lower levels of the decision-making hierarchy that are only focused on the FMS itself.

Between the production planner, programmer, and FMS operators there are differences in the factors that they consider when making planning or scheduling decisions. The production planner mainly looks at customer value stream characteristics when indicating which orders should be produced first. The programmer and operator do factor in the importance of certain customers to some degree when making schedule and planning related decisions but are mainly focused on efficiently using the FMS given the FMS limitations. Decisions made at each level of the decision-making hierarchy are generally separate and sequential, which results in a relatively low degree of IPPS.

4.1.7 Work Floor Performance

(26)

26 an average due date performance of 80%. However, this 80% does not include the number of times the customer had to be called to renegotiate the promised due date, which happened occasionally. The main objective of the programmer and operator is to maximize the utilization of the FMS. But even though the FMS is optimized based on efficiency, it is often not even the bottleneck in the production process. As stated by the production planner, the TIG-welding stations are often the bottleneck and sometimes even the press brake machines, depending on the type of orders that are accepted. Considering that this firm’s strategy is focused on short lead times and high due date performance, the objective of maximizing the efficiency of the FMS by the programmer and operator may not be in line with the overall firm objective.

Additionally, maximizing the efficiency of the FMS while it is not the bottleneck creates more problems on the work floor. The overcapacity of the FMS can be observed on the work floor where a large WIP inventory can be found at the buffer of the press brake machines. Due to the batching of orders, production has sometimes already started on the FMS several weeks in advance, and subsequently remain waiting several weeks in the buffer before the next production steps starts.

4.2 Case Analysis Firm B

(27)

27 prevent changeovers. This is especially important for the punching FMSs, where tool and magazine limitations have more impact on FMS efficiency. For a more detailed description of the scheduling decision-making hierarchy, see Appendix IIb.

The way how the decision-making hierarchy is structured allows for some degree of IPPS for the FMS. However, the overall IPPS is still low. The main reason for the lack of integration is the limited coordination between the production planner and programmers. Although the production planner provides a general sequence in which orders have to be produced, the actual decision of which orders are batched on what FMSs and when the batches are released are mainly made by the programmers. Therefore, the programmers are partially responsible for both production planning and scheduling. While this method allows for efficient use of the FMSs, it often resulted in orders being released too late for production. This problem was especially noticeable when there were not enough similar orders to batch. The programmers had to make the decision when to release the batches anyway, even though it may not be as efficient for the FMS. Even between the two programmers there were differences in the general rules they used when to release the batches. The programmer of the punching and punching/laser FMS states that the order should be released on the scheduled starting date but is willing to wait another day. The other programmer is willing to wait “a couple of days” after production was supposed to start. This results in orders already being late even before they have started production, and the following production processes often do not have the available capacity to make up for this delay. The production planner than has to make daily planning adjustments on the work floor in an attempt to minimize these delays.

4.2.2 Consideration of FMS Characteristics

(28)

28 Therefore, while these FMS characteristics have a large impact on the actual performance of the FMSs, they are not accounted for during production planning. This results in a production plan that is relatively inaccurate, which negatively affects IPPS.

4.2.3 FMS Optimization Method

The FMS programmers are responsible for batching as many orders as possible to make sure that the machines are always running and operate as efficiently as possible. Additionally, the focus of the operators is mainly on optimizing the punching machines by sequencing orders based on tool limitation, such that the number of tool magazine changes are minimized.

While the production planner indicated that the efficient use of the production resources is an important objective, meeting customer due dates is also increasingly important. However, because the objective of the FMS programmers and operators are mainly focused on the efficient use of the FMSs, there are sometimes conflicting objectives between production planning and production scheduling. The conflicting objective are especially noticeable when the programmers have to decide when to release a sub-optimal batch of orders. The programmers are often inclined to wait a couple days longer in the hope of new orders to make a batch more efficient. But this often results in orders starting production too late, and the customer due date being missed. Considering that the FMSs in this case usually are not the bottleneck in the production process, reduces the need to only maximize efficiency. These conflicting objectives negatively affect IPPS.

4.2.4 Consideration of Value Stream Characteristics

The production planner indicated that he informally does differentiate between customers based on how much revenue they generate (A-, B-, C-customers), and this difference in customer importance in the production schedule was confirmed during informal talks with several other employees, suggesting that important customers generally get more priority in the list of orders which is generated during production planning.

(29)

29 exactly. Some A-customers are still getting high priority during production scheduling, even with the programmers, but there is less pressure to differentiate between the slightly less important customers. This indicates that differences in the importance of customer value streams is not equally accounted for in the different levels of the decision-making hierarchy which can result in conflicting objectives, which negatively affects IPPS.

At this firm, there is a relatively wide difference in terms of flexibility of each FMS. There is some degree of routing flexibility, but not every order can be produced on every FMS. For example, the punching FMSs can only process orders made from a thicker material. Furthermore, not every FMS has the same degree of product flexibility, it is less economical for the punching FMSs to do a changeover when the tools in the magazine have to be changed. It usually takes around 30 minutes for the operator to change tools in the magazine, while the lasers have all tools already available and can quickly change when needed.

During production planning by the production planner, FMS flexibility is barely considered. Optimally utilizing FMS flexibility is mainly considered the responsibility of the programmers. The programmers benefit from routing flexibility by assigning orders to FMSs that have more available capacity. However, once an order is programmed and batched for a certain FMS, the routing often stays fixed even when there are disturbances on the work floor. Because the FMS programmers are partly responsible for the production planning as well as the production schedule of the FMSs, there is some degree of IPPS. However, production planning and scheduling are still sequential activities. Once an order is programmed for a certain FMS, the routing is often not changed, and the benefit of routing flexibility is limited.

(30)

30 are generally made separately and sequentially. While the production planner spends a lot of time on the work floor observing the actual status of orders and making adjustment to the schedule if needed, there is still relatively little coordination with the planning related decision made by programmers. While the FMSs are operating relatively efficiently due to the relatively high degree of IPPS of the FMSs, because the FMSs are often not the bottleneck in the production process, the lack of IPPS on the work floor still creates problems which will be discussed next.

4.2.7 Work Floor performance

(31)

31 4.3 Cross-case Analysis

Similar to Slomp’s (1993) general description of a decision hierarchy of a production control system of an FMS, at both cases the scheduling decision-making hierarchy of the FMS can be divided into three levels. Because of the dynamic firm environment and the complexity of the FMS, many production planning and scheduling decisions related to the FMS are made at the lower levels of the hierarchy, more specifically by the programmers and operators of the FMS.

Between the cases there was a difference in the complexity and number of FMSs on the work floor. For case B this resulted in a more complex decision-making hierarchy where more responsibilities of production planning and scheduling for the FMSs were transferred to the programmers compared to case A. This indicates that an increased complexity of FMSs results in more planning and scheduling responsibilities for the programmers. While this transfer of responsibilities resulted in some degree of IPPS for the FMS specifically, the firms overall IPPS was still relatively low because of the lack of coordination between the programmers and the production planner.

4.3.2 Consideration of FMS characteristics

(32)

32 planner, adjustments have to be made by both the programmer and operator of the FMS. Because these decisions are often not coordinated, it negatively affects IPPS.

4.3.3 FMS Optimization Method

At both firms, the FMSs are optimized on efficiency by the programmers and operators. However, in contrary to the literature where the FMS is often described as a bottleneck machine, this was often not the case at both firms. This results in conflicting objectives at the different levels of the decision-making hierarchy. The main objective of the production planner at both cases is to meet customer due dates cost effectively. The main objective of the programmer and operators is however to maximize FMS efficiency. When the FMS is not the bottleneck, the focus of production planning and scheduling should not be on the FMS but on the new bottleneck instead. In particular, the way how programmers batch orders to maximize the efficiency of the FMSs has a large effect on firm performance. At firm A, the batching rules caused orders often being released too soon to the work floor, while at firm B it caused orders to be released too late. Therefore, the optimization method of the FMS is not always in line with the overall firm objective, and this differences in objectives negatively affects IPPS.

4.3.4 Consideration of Value stream Characteristics

Similar rules to account for value stream characteristics in the PP&C system are used by both firms, where the production planner gives a certain level of priority to each order based on the importance of that customer. However, when the orders are batched and sequenced for production by the programmer or operator, unless an order has top priority, the differences in importance of each order is less important. Instead, optimization of the FMS is still the main priority of the programmers and operators, even though meeting customer due dates of especially the large customers has become increasingly important. This indicates that there are conflicting objectives between the production planning and scheduling actions performed by the production planner and the programmer and operator of the FMS, which decreases the overall IPPS.

(33)

33 because orders at these firms come in irregularly and the exact manufacturing time is often not known it advance, the capacity of each machine on the work floor can change. While the sequence in which batches are produced on a certain FMS can be adjusted, the type of orders in these batches and on what machine cannot, even though this could shift workload to another machine with available capacity. This may be due to the way how the decision-making hierarchy is structured. Production planning related decisions are generally made separately and sequentially at the different levels, which results in a relative inflexible decision-making structure and negatively affects IPPS.

Both firms attempted to increase the IPPS of the FMSs by given more planning responsibilities to the FMS programmers and operators, especially at firm B where the FMSs were more complex. While this integration allowed to efficiently use the FMS by optimizing the FMS given its limitations, FMS schedule decisions lacked the overall consideration of what is optimal for the firm, such as what is optimal for the rest of the production process and what customer value streams should be optimized. This was especially important for both firms, because the FMSs were often not even the bottleneck. This resulted in conflicting objectives at the different levels of the decision-making hierarchy. The production planner has the goal of meeting firm objectives, such as meeting customer due dates and optimally using all firm resources, while the FMS programmers and operators are mainly focused on efficiently using the FMS.

Furthermore, even though there was some degree of FMS IPPS, both firms did not take full advantage of the flexibility which an FMS is able to offer. Especially at firm B where there was routing flexibility, but once an order was batched the routing often did not change. This is because planning related decisions are often still made separately and sequentially in the decision-making hierarchy, and this structure is not able to adequately respond to changes on the work floor. Overall, this indicates that only partial IPPS of the FMS is not enough when the FMS is often not the bottleneck.

4.3.7 Work Floor Performance

(34)

34 it is not properly coordinated with the rest of the production processes. This resulted in for example producing too much WIP or orders not starting production on the right time. Secondly, the lack of IPPS resulted in conflicting objectives, which caused firm objectives of meeting customer due dates not being met. While the production planners attempted to give important customers priority during production planning, the lack of IPPS and conflicting objectives did not prevent customer due date problems of important customers.

4.4 Potential of a Digital Twin Model

The results in 4.1-4.3 indicates that the way how the FMS is managed in the PP&C system results in some degree of IPPS for the FMS itself. However, because some planning and scheduling responsibilities are transferred from the production planner to the FMS programmer, the overall firm’s IPPS is still low which results in several problems on the work floor. To answer the second part of the research question, the potential of a Digital Twin will be researched.

As discussed in 2.3.2, a Digital Twin Model allows with the use of different technologies to make a digital representation of the work floor which is continuously updated with real-time data. This can be used for a variety of purposes, but more specifically in this paper, it has the potential to improve the overall IPPS of the firms discussed in this paper. One way of doing this is by shifting some planning and scheduling decisions from the FMS operators and programmers to the Digital Twin Model and/or the production planner. These types of decisions showed to have a lot of effect on work floor performance, especially because the FMS was not the bottleneck in the production process. In particular, the decisions of what orders to batch and the decisions when these batches were released to the FMS had a large impact. Allowing the production planner to be more in control of these batching decision would allow the possibility to find a better balance between FMS optimization and meeting firm objectives such as customer due dates. A Digital Twin Model could assist in several ways, depending on how advanced the model is.

(35)

35 the different levels of the decision-making hierarchy, which resulted in too large batch sizes. To improve IPPS, a Digital Twin Model could be used by as a tool for the production planner to calculate the optimal batch sizes given the status and limitation of the other machines. Additionally, the model could continuously update the production schedule when new orders come in and priorities of customers change.

Similar for Firm B, the way how orders were batched and released to the FMS created problems on the work floor. But at this firm, the problem was that programmers were not able to adequately balance maximizing the efficiency of the FMS while still meeting the internal due dates, which resulted in orders regularly starting production too late and the inability of meeting customer due dates. Therefore, the responsibility of batching and releasing orders could be shifted to the Digital Twin Model to improve IPPS and find a global optimum that finds the balance between efficiency and due date performance given the firms characteristics and preferences. Furthermore, because the model is able to consciously adjust the production schedule to react to disturbances on the work floor, it also has the potential to continuously adjust what orders are batched and for which FMS. This would allow this firm to take more advantage of the routing flexibility which their FMSs are able to offer. However, the ability of benefiting of routing flexibility is limited even with a Digital Twin Model, because once an order is programmed for a certain FMS, changing the FMS in the schedule may require reprogramming.

Therefore, the way how a Digital Twin Model can improve IPPS is largely dependent on how advanced the model is and how many planning and scheduling responsibility can be done autonomously and continuously by the model itself. However, even with a relatively simple model benefits could be achieved. For the cases in this paper specifically, giving the model, or the production planner using the model as tool the responsibility of determining what orders are batched and when they are released could already improve IPPS by preventing conflicting objectives and finding an overall optimal schedule.

4.5 Validation

(36)

36 in the actual status of machines and orders on the work floor. Less interest was expressed for a model which would be able to make a production schedule autonomously, but this may be because of the complexity of the model, it may not be seen as a feasible solution yet.

5. Discussion

The aim of this paper was to find what factors of the PP&C system cause a lack of IPPS and what effect this had on the work floor. Results are to some extent in line with the literature. The decision-making hierarchy for the FMS is structured in three levels as described by Slomp (1993). Additionally, the results are in line with the paper of Raj (2007) who stated that the proposed FMS optimization methods in literature are often too complex and therefore often not used in practice. In the discussed cases in this paper, the programmers who were responsible for the optimization of the FMS did also not use advanced methods. They often did not even use any pre-specified rules on how orders should be batched to maximize efficiency. Furthermore, the effects of a lack of IPPS are in line with the research of Lihong & Shengping (2012). Both firms experienced conflicting objectives between the different levels of the decision-making hierarchy which resulted in problems on the work floor.

However, not in line with the literature is the way an FMS is often described as a bottleneck machine. This was usually not the case at both firms in this paper. This may have a large impact on how the FMS should be optimized and whether or not this results in conflicting objectives.

(37)

37 which was discussed in 2.3.2 may provide similar benefits and could be interesting for further research.

5.1 Limitations

There are several limitations of this paper that have to be considered that may affect the generalizability of this paper. First, the sample size is only limited to two relatively similar firms from the same industry, size, and region. Additionally, the FMSs that were researched were relatively simple and were often not the bottleneck of the production process. More complex FMSs may be controlled differently, as well as FMSs that are the bottleneck. Furthermore, in terms of method, an attempt was made by the author to limited interview bias by conducting interviews using interview protocols. But the other methods were largely unstructured, potentially resulting in having some interpretive bias. Therefore, more research may be needed to increase the generalizability of the results.

6. Conclusions

In summary, with the ability to produce a wide variety of products at near mass manufacturing efficiency, an FMS may be worth its expensive price tag. However, if the complex machine is not properly managed, its potential may be underutilized. The way how the PP&C system is structed affects how the FMS is managed and optimized, but it also affects the degree of IPPS. A lack of IPPS can cause problems like a lack of flexibility in production planning and an unbalance of workload on the different machines due to conflicting objectives. An argued way of improving the degree of IPPS is by increasing the information exchange between the production planning and scheduling department, and with the rise of industry 4.0 technologies, a Digital Twin Model could potentially facilitate this exchange. To research this potential, several aspects of the PP&C system that affect the degree of IPPS had be identified. These aspects that affect the degree of IPPS are researched as well as the potential of a Digital Twin Model to improve IPPS. Research was conducted via a multi-case study at two SME Dutch sheet metal working firms who use FMSs in their production process. The primary method for data collection was semi-structured interviews, but other methods like field observations were used as well.

(38)

38 because the programmers are able to optimize the FMS given its machine limitation. However, this still results in an overall low IPPS, because there is relatively little coordination between the production planner and the programmer to ensure that the FMS optimization decisions made by the programmer are in line with the rest of the machines. Additionally, these FMS optimization decisions do not always take into account important customer value streams, even though this is also important for firm performance. This resulted in various problems on the work floor, such as the FMSs creating too much WIP inventory, and conflicting objectives between the production planner and programmer in balancing efficiency of the FMS and meeting firm objectives like meeting customer due dates.

The analysis of the potential of a Digital Twin Model indicated that there are several ways in which this model could improve the degree of IPPS, and it is dependent on how advanced the model is designed to be. A more advance model would allow to transfer more planning and scheduling decisions made by the programmer to the production planner and/or to the model itself which would be able to adjust or make the production schedule autonomously. In particular, the analysis of the two firms indicated that the decisions made by the programmers related to the batching of orders and when to release them to the FMSs had a large effect on work floor performance. If the batching decisions are made by the Digital Twin Model or by the production planner using the model as a decision-making tool to find a better balance between FMS efficiency and meeting overall firm objectives, IPPS is likely to be improved.

(39)

39

References

Avventuroso, G., Silvestri, M., & Pedrazzoli, P. (2017). A networked production system to implement virtual enterprise and product lifecycle information loops. IFAC-PapersOnLine, 50(1), 7964-7969. Banaszak, Z. A., & Krogh, B. H. (1990). Deadlock avoidance in flexible manufacturing systems with concurrently competing process flows. IEEE Transactions on robotics and automation, 6(6), 724-734. Barbati, M., Bruno, G., & Genovese, A. (2012). Applications of agent-based models for optimization problems: A literature review. Expert Systems with Applications, 39(5), 6020-6028.

Bensmaine, A., Dahane, M., & Benyoucef, L. (2014). A new heuristic for integrated process planning and scheduling in reconfigurable manufacturing systems. International Journal of Production Research, 52(12), 3583-3594.

Browne, J., Dubois, D., Rathmill, K., Sethi, S. P., & Stecke, K. E. (1984). Classification of flexible manufacturing systems. The FMS magazine, 2(2), 114-117.

Chan, F. T., & Swarnkar, R. (2006). Ant colony optimization approach to a fuzzy goal programming model for a machine tool selection and operation allocation problem in an FMS. Robotics and

Computer-Integrated Manufacturing, 22(4), 353-362.

Cowling, P., & Johansson, M. (2002). Using real time information for effective dynamic scheduling. European journal of operational research, 139(2), 230-244.

Gerwin, D. (1993). Manufacturing flexibility: a strategic perspective. Management science, 39(4), 395-410.

Grieco, A., Semeraro, Q., & Tolio, T. (2001). A review of different approaches to the FMS loading problem. International Journal of Flexible Manufacturing Systems, 13(4), 361-384.

Karlsson, C. (Ed.). (2016). Research methods for operations management. Routledge.

Karsak, E. E. (2002). Distance-based fuzzy MCDM approach for evaluating flexible manufacturing

system alternatives. International Journal of Production Research, 40(13), 3167-3181.

Kim, H. W., Yu, J. M., Kim, J. S., Doh, H. H., Lee, D. H., & Nam, S. H. (2012). Loading algorithms for flexible manufacturing systems with partially grouped unrelated machines and additional tooling constraints. The International Journal of Advanced Manufacturing Technology, 58(5-8), 683-691. Koren, Y., & Shpitalni, M. (2010). Design of reconfigurable manufacturing systems. Journal of

manufacturing systems, 29(4), 130-141.

Kostal, P., & Velisek, K. (2010). Flexible manufacturing system. World Academy of Science,

Engineering and Technology, 53, 825-829.

Kumar, N. S., & Sridharan, R. (2009). Simulation modelling and analysis of part and tool flow control decisions in a flexible manufacturing system. Robotics and Computer-Integrated Manufacturing, 25(4-5), 829-838.

Kritzinger, W., Karner, M., Traar, G., Henjes, J., & Sihn, W. (2018). Digital Twin in manufacturing: A categorical literature review and classification. IFAC-PapersOnLine, 51(11), 1016-1022.

Lee, J. H., & Kim, C. O. (2008). Multi-agent systems applications in manufacturing systems and supply chain management: a review paper. International Journal of Production Research, 46(1), 233-265.

(40)

40 MacCarthy, B. L., & Liu, J. (1993). A new classification scheme for flexible manufacturing systems. The

International Journal of Production Research, 31(2), 299-309.

Meredith, J. (1998). Building operations management theory through case and field research. Journal

of operations management, 16(4), 441-454.

Negri, E., Fumagalli, L., & Macchi, M. (2017). A Review of the Roles of Digital Twin in CPS-based Production Systems. Procedia Manufacturing, 11, 939-948.

Okongwu, U., Lauras, M., Dupont, L., & Humez, V. (2012). A decision support system for optimising the order fulfilment process. Production Planning & Control, 23(8), 581-598.

Ouelhadj, D., & Petrovic, S. (2009). A survey of dynamic scheduling in manufacturing systems. Journal

of scheduling, 12(4), 417.

Phanden, R. K., Jain, A., & Verma, R. (2011). Integration of process planning and scheduling: a state-of-the-art review. International Journal of Computer Integrated Manufacturing, 24(6), 517-534.

Phanden, R. K., Jain, A., & Verma, R. (2013). An approach for integration of process planning and scheduling. International journal of computer integrated manufacturing, 26(4), 284-302.

Raj, T., Shankar, R., & Suhaib, M. (2007). A review of some issues and identification of some barriers in the implementation of FMS. International Journal of Flexible Manufacturing Systems, 19(1), 1-40. Rother, M., & Shook, J. (2003). Learning to see: value stream mapping to add value and eliminate

muda. Lean Enterprise Institute.

Ruiz, M. C., Cazorla, D., Cuartero, F., & Macia, H. (2009). Improving performance in flexible manufacturing systems. The Journal of Logic and Algebraic Programming, 78(4), 260-273.

Slomp, J., 1993. Production control for flexible manufacturing systems: an application oriented approach.

Solvang, B., Sziebig, G., & Korondi, P. (2012). Shop-floor architecture for effective human-machine and inter-machine interaction. Acta Polytechnica Hungarica, 9(1), 183-201.

Stecke, K. E. (1983). Formulation and solution of nonlinear integer production planning problems for flexible manufacturing systems. Management Science, 29(3), 273-288.

Stevenson*, M., Hendry, L. C., & Kingsman, B. G. (2005). A review of production planning and control: the applicability of key concepts to the make-to-order industry. International journal of production

research, 43(5), 869-898.

Tiwa Ri, M. K., Rika, B. H., Rthi, N. V., Jaggi, P., & Mukhopadhyay, S. K. (1997). A heuristic solution approach to the machine loading problem of an FMS and its Petri net model. International Journal of

Production Research, 35(8), 2269-2284.

Tiwari, M. K., & Vidyarthi, N. K. (2000). Solving machine loading problems in a flexible manufacturing system using a genetic algorithm based heuristic approach. International journal of production research,

38(14), 3357-3384.

Wang, S., Wan, J., Zhang, D., Li, D., & Zhang, C. (2016). Towards smart factory for industry 4.0: a self-organized multi-agent system with big data based feedback and coordination. Computer