Revisiting MRP nervousness
A case study on variability and buffering in MRP-controlled manufacturing environments
Friederike Nette
A thesis submitted for the degrees of
M.Sc. Technology & Operations Management (EBM028030) Faculty of Economics and Business
at University of Groningen
M.Sc. Operations & Supply Chain Management (NBS8399) Newcastle University Business School
at Newcastle University
First supervisor: Dr. Martin J. Land Second supervisor: Prof. Christian Hicks
December 10, 2018
Abstract
This thesis studied MRP nervousness which typically results from variability in MRP- controlled production systems and analysed how variability and inventory can be created by managerial and planning-related decisions. Both in today’s theory and in today’s practice it becomes obvious that the sources of MRP nervousness are not sufficiently researched at the moment. One the one hand, because no research has been conducted in this specific field in the last two decades. On the other hand, because many companies still struggle to successfully protect their MRP against nervousness. An inventory analysis framework that was proposed in the recent literature, however, offers a new perspective on variability. In this case study, the framework was applied in a Chinese large-scale manufacturing company to (i) analyse how managerial and planning-related decisions can create variability and inventory in production systems in order to extend the current analysis of the sources of MRP nervousness and (ii) to validate and extend the framework. Although no new managerial and planning-related sources of MRP nervousness itself could be revealed in this study, an amplifying mechanism was found which significantly increased the variability in the system and resulted from the design of its planning approach and procedure. Moreover, this study showed that inventory can result from managerial decisions, whereby it does not necessarily have to be directly attributable to a source of variability, as it is currently assumed in the framework. The study hence provides important insights in the relationship between variability and inventory in MRP-controlled production systems, which indicate directions for future research on MRP nervousness as well as for the future development of the inventory analysis framework.
Keywords: MRP nervousness, inventory, buffering, variability
Acknowledgement
First of all, I would like to thank my first thesis supervisor Dr. Martin Land of the Faculty of Economics and Business at the University of Groningen for his guidance, feedback and involvement in this thesis. He always had an open ear whenever I had a question about my research and writing. Additionally, I would like to thank him for making it possible for me to conduct my research in China and for visiting me personally during my stay. Secondly, I would like to thank Prof. Matthias Thürer of the Faculty of Operations Management at the Jinan University in Zhuhai for his support, inspiration and involvement. A special word of thanks is moreover due to Prof. Thürer for receiving me in China, taking great personal care of me during my stay and making it an unforgettable experience for me. Further, I would like to thank my second thesis supervisor Prof. Christian Hicks of the Newcastle University Business School at the Newcastle University for his valuable comments on my research. I would also like to thank everyone involved at the case company and at the Jinan University for their time and efforts.
Without them this thesis would not have been possible. Finally, I want to express my gratitude to my parents for providing me with continuous support and encouragement throughout my years of study and the process of researching and writing this thesis and to my friends Viktoria Szabo and Helena Mollá García for being there whenever I needed them.
Göttingen, December 2018 Friederike Nette
Table of Contents
List of figures ... IX List of tables ... X List of abbreviations ... XI
1. Introduction ... 1
2. Theoretical background ... 4
2.1 The MRP approach ... 4
2.2 Impact of uncertainty on MRP ... 6
2.3 Impact of decision-related fluctuations on MRP ... 8
2.4 Strategies to protect against variability in MRP ... 8
2.5 New developments in variability analysis ... 10
2.6 Research question ... 11
3. Methodology ... 13
3.1 Research method ... 13
3.2 Case selection ... 14
3.3 Data collection and analysis ... 14
4. Case introduction ... 16
4.1 The case company and the production process ... 16
4.2 The production process in detail ... 17
5. Results ... 19
5.1 Analysis of the planning approach and procedure ... 19
5.2 Inventory analysis ... 21
5.3 Variability analysis ... 24
6. Conclusion ... 30
6.1 Research findings ... 30
6.2 Theoretical and practical implications ... 31
6.3 Limitations and future research ... 32
7. Epilogue ... 34
7.1 Evaluation of current variability reduction techniques ... 34
7.2 Proposal of improvement approaches ... 35
8. References ... 39
Appendix A – Inventory typology ... 41
Appendix B – Research overview ... 42
Appendix C – Layout of the mould injection production ... 44
Appendix D – Production planning data table ... 45
Appendix E – Instability in the production planning ... 47
List of figures
Figure 2.1: The MRP procedure (adapted from Murthy and Ma, 1991) ... 5
Figure 2.2: The inventory analysis framework (Land et. al, 2018) ... 10
Figure 4.1: Simplified production process of the AC production ... 16
Figure 4.2: Process map of the AC production ... 18
Figure 5.1: Planning procedure in the mould injection planning department ... 20
Figure 5.2: Extract from the production planning data table for material no. 10333011 ... 26
Figure 5.3: Planning procedure of the overall AC production ... 27
Figure 5.4: Daily production planning for material no. 10333011 ... 28
Figure 5.5: Current production situation - Inventory type 3 ... 28
Figure 5.6: Current production situation – Inventory type 5 ... 29
Figure 7.1: Optimized planning procedure – Inventory type 2 ... 36
Figure 7.2: Improved production situation with Heijunka – Inventory type 3 ... 37
Figure 7.3: Improved production situation with Heijunka – Inventory type 5 ... 38
List of tables
Table 2.1: Steps of the MRP procedure ... 5
Table 2.2: Categories of Uncertainty in MRP Systems (Source: Whybark and Williams, 1976) ... 6
Table 2.3: Sources of uncertainty impacting MRP systems ... 7
Table 2.4: Strategies to protect MRP against variability ... 9
Table 4.1: Columns in the requirements plan of the mould injection plant ... 21
Table 5.1: Inventory analysis - Inventory type 1 ... 22
Table 5.2: Inventory analysis - Inventory type 2 ... 22
Table 5.3: Inventory analysis - Inventory type 3 ... 23
Table 5.4: Inventory analysis - Inventory type 4 ... 23
Table 5.5: Inventory analysis - Inventory type 5 ... 24
List of abbreviations
AC air conditioner
APICS American Production and Inventory Control Society BOM bill of materials
CNC computer numerical controlled CODP customer order decoupling point ERP enterprise resource planning MPS master production schedule MRP material requirements planning MRP II manufacturing requirements planning MTO make-to-order
PPC production planning and control
1. Introduction
Most production systems involve some kind of assembly, which entails that product parts or components are produced separately before being merged in a final operation (Cransberg et al., 2016; Hopp and Spearman, 2011). Because the demand for the product parts or components in the earlier stages of the production typically depends on the externally-driven demand for the end items, a different planning approach is required in multi-stage systems – so-called materials requirement planning (MRP) (Gelders and Van Wassenhove, 1981; Guide and Srivastava, 2000). Successfully implemented, MRP can significantly improve the inventory and customer service levels of the production system (Ho, 1989; Whybark and Williams, 1976). An often cited obstacle to the successful implementation of MRP, however, is MRP nervousness, or system nervousness, which arises when changes at one stage of the system propagate to the other stages of the system and make frequent rescheduling necessary (Blackburn et al., 1986, 1985; Kadipasaoglu and Sridharan, 1995). Implementing changes in MRP is necessary because the three process inputs that need to be synchronized in a production process – supply of flow items, demand and capacity – are subject to variability or, more precisely, uncertain or decision-related fluctuations, which have their roots in the internal and external environment of the system (Hopp and Spearman, 2011; Land et al., 2018). The resulting “instability in planned orders” (Blackburn et al., 1986, p. 413) at all stages of the MRP increases both process lead times and the amount of safety stocks required in the system and leads to an ultimately decreased system performance (Blackburn et al., 1986; Buzacott and Shanthikumar, 1994).
The relevance of MRP nervousness for today’s businesses becomes obvious in both theory and practice. A variety of studies is devoted to the topic of MRP nervousness in the literature (see e.g., Murthy and Ma, 1991; Dolgui and Prodhon, 2007). However, while earlier studies focused on identifying different sources of MRP nervousness and on developing strategies to protect against it, the research on sources of MRP nervousness has notably decreased in the recent time. Instead, research in this area has turned to programming approaches and simulation studies, which aim to advance the applicability of MRP in different production environments (Li and Disney, 2017). Considering the “growing frequency and magnitude of changes in the technology and managerial methods” (Voss et al., 2016, p. 166) that is characterizing today’s business environment. However, it must be recognized that the operating environment of MRP systems has significantly changed in the last two decades (Reschke and Schuh, 2017; Voss et al., 2002). There might hence be new opportunities for variability to arise in the internal and
are not yet covered in the current literature. Since no explicit research on the sources of MRP nervousness has been conducted in the last two decades, there is an obvious gap in the current literature and a need to revisit this topic. This need is also supported by what can be observed in the current practice. In 2016, for example, the American Production and Inventory Control Society (APICS) devoted a comprehensive webinar to the topic of MRP nervousness, which should enable participants to stabilize their MRP (APICS, 2016). This also indicates that many companies still suffer from nervous MRP systems in practice. Finally, also the case company of this research experiences a nervous MRP and struggles to effectively identify the root causes of its nervousness, which underlines that the analysis related to this topic that is provided in the current literature is insufficient.
A new perspective on variability, which is typically the main cause of MRP nervousness, is offered by an inventory analysis framework that was proposed in the more recent literature by Land et. al (2018). The framework is based on a typology of 28 different inventory types and uses these types to diagnose the underlying sources of variability that disturb the process flow in the system. A comprehensive introduction of the framework will be given in Section 2.5.
Since the inventory analysis framework considers variability in much greater detail than it was done by earlier scholars, it might also shed new light on the sources of MRP nervousness and is thus expected to improve its current analysis. At the same time, the research could help to validate and extend the proposed framework, as it does only consider uncertainty, batching and predicted fluctuations as possible forms of variability at the moment. By applying the framework in an MRP-controlled manufacturing environment – which is known to be extremely vulnerable to the impacts of variability (Guide and Srivastava, 2000) – it is expected that especially planning-related sources of variability and MRP nervousness could be revealed in this study, which extend the current analysis in this field. The purpose of this research is therefore to investigate how managerial and planning-related decisions can create a source of variability and inventory in production systems by applying the inventory analysis framework of Land et al. (2018) in an MRP-controlled production system.
The research is carried out in the form of an exploratory case study in a Chinese large-scale manufacturing company. The contribution of the research is twofold; a theoretical contribution is made by reviewing and extending the sources of MRP nervousness that are currently provided in the literature and by validating and extending the proposed inventory analysis framework of Land et al. (2018); a practical contribution is made by providing general design solutions for the encountered MRP problems and by demonstrating the applicability of the
framework in a large-scale production facility where the inventory is generally too large to be observed on the shop floor.
The remainder of this thesis is organised as follows. Section 2 starts with a review of relevant literature on MRP and variability to derive the research question. Section 3, afterwards, describes the research methodology, before Section 4 introduces the case company. The results of the research are presented in Section 5 where different sources of variability that exist in and contribute to MRP nervousness in the production system of the case company are investigated, discussed and linked to managerial and planning-related decisions. Section 6 summarizes the findings of this research and addresses its implications and limitations. To conclude, Section 7 outlines different improvement approaches for the identified inventories and sources of variability in the production system of the case company.
2. Theoretical background
Section 2 first reviews relevant literature on MRP and variability. Therefore, the general MRP approach is described in Section 2.1 and the impact of different forms of variability (uncertain and decision-related fluctuations) on MRP is explained in Section 2.2 and 2.3. Section 2.4 provides an overview of strategies to protect against MRP nervousness, before Section 2.5 presents the inventory analysis framework of Land et al. (2018). Based on this, the research question is derived in Section 2.6.
2.1 The MRP approach
Despite being developed in the 1970s MRP still forms an important backbone for most production planning and control (PPC) approaches in today's practice (Guide and Srivastava, 2000; Hopp and Spearman, 2011). As indicated by its name, the purpose of the traditional MRP (Orlicky, 1975) is to plan material requirements. The specific target of MRP is thereby on planning purchase and work orders (Murthy and Ma, 1991) “to satisfy material requirements generated by external demand” (Hopp and Spearman, 2011, p. 110). Commonly known is that MRP is especially suitable for PPC in multi-stage or assembly systems where the demand for components or product parts depends on the market-driven demand for end items (Gelders and Van Wassenhove, 1981; Guide and Srivastava, 2000). Since MRP takes into account this distinction between dependent and independent demand as well as the available parts inventory, it has a significant potential to improve the customer service and inventory levels of the system to enhance its overall performance (Ho, 1989; Whybark and Williams, 1976).
The traditional MRP later developed into manufacturing resource planning (MRP II) and various versions of enterprise resource planning (ERP) to include further manufacturing activities and business functions (Hopp and Spearman, 2011; Yeung et al., 1998). However, since the underlying logic of the approach remained unchanged, only the basics of MRP will be covered in the following of this research (Koh et al., 2002). A complete discussion of MRP can be found in Orlicky (1975) and Vollmann et al. (1997).
Figure 2.1 shows the three inputs of MRP – the master production schedule (MPS), bills of materials (BOM) and inventory records – and how they come together in the MRP procedure to achieve the desired outputs (Murthy and Ma, 1991).
Figure 2.1: The MRP procedure (adapted from Murthy and Ma, 1991)
As can be seen in the figure, the main driver of the MRP procedure is the MPS, which is set up on the base of existing customer orders and demand forecasts and states when and in which quantity each end item is required (Hopp and Spearman, 2011; Murthy and Ma, 1991). On the contrary, the BOM of each end item reflects its product structure and contains information, such as the amount of each part type that is required for the assembly and the corresponding production lead times. The inventory records, finally, provide the available on-hand inventory and scheduled receipts for each product part and thus require frequent updating. Table 2.1 explains each step of the procedure that has to be followed according to the MRP logic in more detail (Hopp and Spearman, 2011; Murthy and Ma, 1991).
Table 2.1: Steps of the MRP procedure
1 The demand information for, most commonly, each end item per time period within the planning horizon is obtained from the MPS.
2 The BOM of each end item is used to ‘explode’ the end item demand into requirements for parts, which are then aggregated to gross requirements for each part per time period.
3 The inventory record of each part is used to subtract the respective on-hand inventories and scheduled receipts from the gross requirements to derive the net requirements.
4 The net requirements are offset in time to determine the due dates of the parts production. Based on this, recommendations for purchase and work orders can be derived, capacity requirements can be determined and, if necessary, rescheduling of the MPS can be requested.
Even though the MRP logic would exclude the use of inventory buffers in theory, different sources of variability exist in the MRP environment, which make the use of buffers necessary in practice (Guide and Srivastava, 2000; Hopp and Spearman, 2011). Commonly known from OM (Operations Management) theory, is, however, that holding inventory is expensive. A main concern when managing production systems is hence to minimize the required inventory and underlying variability in the system, while at the same time avoiding stockouts and maintaining high customer service levels (Dolgui and Prodhon, 2007; Hopp and Spearman, 2011).
Reducing the variability in MRP-controlled production environments becomes even more important considering that the effect of the variability in the MRP system is usually intensified due to the integrative nature of the planning approach. More precisely, if changes are implemented at one stage of the MRP, they automatically propagate to the other stages of the MRP, which causes a general instability in the planning or – in other words – a nervous system (Blackburn et al., 1986, 1985; Kadipasaoglu and Sridharan, 1995). In the literature, MRP nervousness is also described as a result of a combination of uncertainty and planning-related decisions, which refers to the fact that changes in MRP often have to be implemented due to uncertainty, but the degree, to which the variability is propagated throughout the system is determined by lot-sizing rules or other planning-related decisions (Blackburn et al., 1986, 1985; Hopp and Spearman, 2011). Both the reduction of uncertainty-related variability and the reduction of decision-related variability is hence important for the successful management of MRP-controlled production systems. Both forms of variability are reviewed hereafter.
2.2 Impact of uncertainty on MRP
Uncertainty or uncertain fluctuations result from unpredicted events that happen in the internal and external environment of the MRP and cannot be controlled and planned (Hopp and Spearman, 2011; Land et al., 2018). A variety of studies has examined uncertainties and their impacts on MRP (Guide and Srivastava, 2000).
As one of the first, Whybark and Williams (1976) defined four categories of uncertainty in MRP by distinguishing between the type and source of uncertainty that is affecting the system.
Thereafter, it can occur with regard to either the timing or the quantity of the gross requirements for product parts at the demand-side or the scheduled receipts of orders at the supply-side, as shown in Table 2.2.
Table 2.2: Categories of Uncertainty in MRP Systems (Source: Whybark and Williams, 1976) Sources
Demand Supply
Types Timing Requirements shift from one period to another
Orders not received when scheduled Quantity Requirements for more or less
than planned Orders received for more or less than planned
Another categorization was used by Ho (1989) who studied the impacts of operating environments on MRP nervousness, thereby expressing the operating environment with two different kinds of variables; while environmental variables reflect factors that lie outside the
production system, operating variables refer to factors that are rooted inside the system.
Similarly, Koh et al. (2002) classified the sources of MRP nervousness into input uncertainty, including supply and demand uncertainty caused by factors outside the system, and process uncertainty, covering supply and demand uncertainty resulting from factors inside the system.
In an effort to identify specific sources of MRP nervousness, a comprehensive literature review was carried out as part of this research. Table 3.2 summarizes the identified sources that are relevant for this research and classifies them into external and internal sources of uncertainty, as it has been proposed by Koh et al. (2002).
Table 2.3: Sources of uncertainty impacting MRP systems External sources
External demand uncertainty
Random variation in the demand for end items (in terms of timing or quantity) resulting from changes in customer orders or forecasts; leading to changes in the MPS (Whybark and Williams, 1976)
Supply uncertainty
Random variation in the supply of raw materials and product parts (in terms of timing or quantity) resulting from changes in vendor deliveries; leading to changes in the scheduled receipts (Ho, 1989; Murthy and Ma, 1991)
Internal sources
Product structure Random variation in the demand for raw materials or product parts resulting from changes in the product structure; leading to changes in the BOM (Murthy and Ma, 1991)
Process yield uncertainty
Random variation in the process yield (in terms of timing or quantity) resulting from various sources, such as lead time, quality variability and failure of production equipment (explained hereafter); leading to changes in the scheduled receipts (Guide and Srivastava, 2000; Mula et al., 2006)
Lead time uncertainty
Random variation in the processing time (Dolgui and Prodhon, 2007; Mula et al., 2006)
Quality uncertainty
Random variation in the product quality, e.g. due to scrap losses (Guide and Srivastava, 2000; Mula et al., 2006)
Failure of production equipment
Random variation in the equipment performance, e.g. due to machine breakdowns or tooling problems (Kadipasaoglu and Sridharan, 1995; Koh et al., 2002; Mula et al., 2006)
2.3 Impact of decision-related fluctuations on MRP
Different from uncertain fluctuations, decision-related fluctuations result from managerial and planning-related decisions that are taken with regard to the production system. They can hence be controlled and planned (Hopp and Spearman, 2011).
Even though uncertain and decision-related fluctuations require fundamentally different management approaches as they fundamentally differ in their controllability, their impact on MRP is party similar. More precisely, also managerial and planning-related decisions cause variations in the quantity and timing of the demand or supply, which have been pointed out as categories of uncertainty in MRP in the previous section. A typical example of a decision- related fluctuation is supply variability, which occurs when orders are grouped together instead of being ordered lot for lot to save ordering cost (Dolgui and Prodhon, 2007). These so-called batching decisions, are further elaborated by Cransberg et al. (2016) who described different forms of batching (sequential batching, simultaneous batching) as means to reduce set-up costs as well as nesting as a technique used to minimize material wastage by combining different work orders. The variability that is introduced by this kind of decisions can be controlled and thus reduced by optimizing the respective planning decisions (Hopp and Spearman, 2011).
However, while currently a variety of lot-sizing rules exist to determine optimal order or production quantities for each time period (Ho, 1989), Dolgui and Prodhon (2007) pointed out that finding rules which are optimal in general and for all levels of production systems is difficult. This also explains the large amount of studies that have been devoted to the development of lot-sizing algorithms in the context of MRP-controlled manufacturing systems (see Mula et. al, 2006).
2.4 Strategies to protect against variability in MRP
Various strategies to protect MRP systems against variability and nervousness have been proposed in the literature. A detailed review on MRP under uncertainty and strategies to cope with different forms of uncertainty in MRP can be found in Mula et al. (2006) and Dolgui and Prodhon (2007). Table 3.3, however, focuses on the relevant strategies for this research and provides explanations as well as references for each strategy. Important to note is that the table classifies the strategies into variability reduction techniques and buffering techniques for the purpose of this research due to their different operating approaches in the system; while variability reduction techniques seek to decrease the variability that has to be dealt with in the system by modifying planning parameters, such as lot-sizing rules and the planning horizon,
buffering techniques accept the existence of variability and focus on reducing the impact of variability in the system instead. This distinction is crucial for the following of this research since buffering techniques imply that excess resources (inventory, capacity, time) are held within the system, which is not the case in variability reduction techniques (Hopp and Spearman, 2011; Koh et al., 2002).
Table 2.4: Strategies to protect MRP against variability Buffering techniques
Safety stock Inventory buffers are placed in the production system to protect against variability in supply and demand.
Whybark & Williams (1976) Blackburn et al. (1986)
Buzacott & Shanthikumar (1994) Vollmann et al. (2005)
van Kampen et al. (2010) Safety lead time Orders are released earlier than planned, so that
end items are available before their actual due date to protect against variability in supply and demand.
Murthy & Ma (1991) Whybark & Williams (1976) Buzacott & Shanthikumar (1994) van Kampen et al. (2010) Hedging The gross requirements in the MPS are
increased to protect against variability in demand.
Murthy & Ma (1991)
Overplanning The production exceeds the quantities stated in the MPS to protect against variability in demand.
Murthy & Ma (1991) Mula et al. (2006)
Yield factor The BOM is modified to incorporate a yield factor. The yield factor specifies the quantities of process inputs required to produce a specific output for each stage of the production system to protect against process yield variability.
Murthy & Ma (1991) Mula et al. (2006)
Variability reduction techniques Schedule
freezing The production schedule is fixed within the planning horizon to avoid changes in the MPS due to new orders that are placed in the planning horizon.
Blackburn et al. (1986) Sridharan & Berry (1990) Zhao & Lee (1993) Vollmann et al. (2005) Forecasting
beyond the planning horizon
The demand forecast covers more periods than included in the planning horizon to avoid that the MPS is changed due to orders which are likely to be placed near to the end of the planning horizon.
Carlson et al. (1982)
Blackburn et al. (1986) Vollmann et al. (2005)
Lot-sizing rules The order quantity to be produced in each time period is determined with algorithms.
Ho (1989) Blackburn et al. (1985)
2.5 New developments in variability analysis
The previous section showed how early MRP literature identified sources of variability and revealed how inventories could be used as buffers to protect against variability. A new perspective on variability is offered by an inventory analysis framework that was recently proposed by Land et al. (2018). The framework analyses each inventory in a production process to identify the underlying source of variability in the system. It is summarized in Figure 2.2 and will be briefly explained in the following of this section.
Figure 2.2: The inventory analysis framework (Land et. al, 2018)
As already mentioned in the introduction, three process inputs are usually required for a process step to commence; demand, flow items and capacity (see Figure 2.2). As soon as one of these inputs is missing, flow items cannot be processed and pile up in inventory before the respective process step. By distinguishing all possible scenarios, in which a process input is missing due to different sources and forms of variability, Land et al. (2018) derived a typology of 28 different inventory types, which can be found in the Appendix (Appendix A). Because each type indicates a different source of variability in the system, which requires different means to reduce it, the framework allows a detailed analysis of the root causes of inventory and lays the foundation for the development of context-specific solutions. Since the framework thereby distinguishes between three different forms of variability – uncertainty, batching and predicted fluctuations – it provides a much more granular view on variability than the traditional OM literature (Land et al., 2018).
2.6 Research question
As has become clear so far, it is necessary to reduce the variability in the production system to decrease the MRP nervousness and the amount of buffering inventory required in the system (Dolgui and Prodhon, 2007; Hopp and Spearman, 2011). Although a significant body of research exists on different sources of variability in MRP and strategies to protect against nervousness in MRP, it seems that there is still a gap in the current research. On the one hand, no research on the sources of MRP nervousness has been conducted in the last two decades, while the operating environment of MRP systems has changed with the increasing globalisation and digitalisation (Reschke and Schuh, 2017; Voss et al., 2002). On the other hand, the practice shows that many companies still struggle to protect their MRP against nervousness (see Section 1). Together, this indicates that the sources of MRP nervousness are not sufficiently researched at the moment and that there is a need to revisit this topic.
The inventory analysis framework of Land et al. (2018) introduces a new perspective on variability, which might also shed new light on the sources of MRP nervousness. As explained in the previous section, it distinguishes 28 different inventory types by looking at the three different process inputs and forms of variability that can affect the inputs – namely uncertainty, batching and predicted fluctuations. The current literature on MRP nervousness does only cover the basic sources of uncertainty, e.g. external demand and quality uncertainty (see Table 2.3), and of decision-related fluctuations (see Section 2.3). These do state the main source of the variability that is causing the variability, but do not allow any direct conclusion to be drawn about the affected process inputs. Moreover, the current literature on MRP nervousness does not explicitly distinguish between predicted fluctuations and batching. While batching is, indeed, acknowledged as a factor which can create variability in production systems, predicted fluctuations are generally neglected in the MRP literature. Reviewing the sources of MRP nervousness from a more detailed and systematic perspective might therefore provide new insights in this field. Finally, Table 3.3 also showed that currently a plethora of different buffering techniques exists in the literature. Finding a significant difference between the buffering techniques and their operating mechanism in the system, however, is difficult as all of them imply that additional resources are provided to buffer against variability in the system.
Since the framework allows a detailed analysis of the relation between variability and buffers in production systems, it could also provide new insights in this direction. Contrarily, the framework lacks a direct way to recognize managerial and planning-related decisions as a source of variability at the moment.
By applying the framework in an MRP-controlled manufacturing environment – which is known to be extremely vulnerable to the impacts of variability and thus a suitable environment to conduct this research (Guide and Srivastava, 2000) – in this research, it is expected that especially new managerial and planning-related sources of variability and MRP nervousness could be revealed. These would extend the current research on the sources of MRP nervousness. Moreover, it is expected that the inventory analysis framework could be validated and extended in this study since it limits its consideration of variability to uncertainty, batching and predicted fluctuations at the moment but does not yet cover a planning-related dimension.
In this research the framework is therefore applied to answer the following research question:
How do managerial and planning-related decisions create a source of variability and inventory in MRP-controlled production systems?
This question has been addressed in the early MRP literature, as it was shown in this section.
However, the aim of this research is to assess whether applying the new inventory analysis framework will provide any new insights with regard to the question. Due to the explorative nature of the question, a single case study was performed, which will be addressed in the next section.
3. Methodology
Having derived the research question, Section 3 now presents the research methodology.
Therefore, Section 3.1 describes the research method, before Section 3.2 explains the case selection and Section 3.3 outlines the data collection and analysis.
3.1 Research method
As mentioned before, an explorative case study was chosen due to the explorative nature of the research question. The basic idea of case research is to produce generalizable knowledge, which adds to existing or the development of new theory, by investigating problems in practice (Yin, 2011). As pointed out by Voss et. al (2016), the particular strength of case research is that it enables researchers to study a phenomenon in its natural setting, which allows “the questions of why, what and how to be answered with a relatively full understanding of the nature and complexities of the complete phenomenon” (Voss et al., 2016). Using a case study for this research was hence beneficial, as it allowed to investigate and analyse the various sources of variability and inventories in the production system of the case company, while at the same time to study and link them to their environment. Since case research typically draws conclusion from multiple sources of evidence, different data and data collection techniques could be used within this study to investigate the sources of variability and to establish links to the different inventory types (Eisenhardt, 1989). This was particularly important because the focus of this research was on managerial and planning-related decisions, which required an understanding of both the objective planning approach and procedures in the production system of the case company and its subjective execution through the production managers and planners in the case company. More detailed information on the data collection and analysis, however, will be given in Section 3.3.
According to Karlsson (2016) two criteria determine the quality of case research next to the relevance of its findings and the rigour of its execution; reliability, which refers to the replicability of the case study and the objectivity of its results, and validity, which is defined by the extent to which its findings are correct and generalizable. Since it was decided to limit this study to a single case to go into more depth during the research, there may be limitations to the generalizability of its findings, which will be addressed in Section 6.3 (Voss et al., 2016;
Yin, 2011). A more detailed assessment of the research quality criteria follows in Section 3.3.
3.2 Case selection
The selection of the case company for this research was based on three criteria. First, it was important to choose a manufacturing company, which uses an MRP system for its PPC.
Second, the production process of the company should involve multiple stages or an assembly operation to ensure a certain degree of complexity in the production planning and to increase the probability that MRP nervousness exists. Third, the company should produce discrete and measurable product outcomes, while the production process had to be structured in various process steps, so that different types of inventory could arise within the system (Land et al., 2018). While each criterion was selected due to the underlying research question of this study, the primary aim of criterion 3 was to ensure the applicability of the inventory analysis framework that should be validated and extended in this study.
The selected case company fulfils all of the above stated criteria as it is a large-scale producer of electric appliances, which uses an ERP system to plan and control its production system.
The headquarters of the case company as well as all relevant production plants and warehouses for this research are located in China. While the company produces a range of different products, this study focused exclusively on the air conditioner (AC) production, whose process characteristics and complexity made it a suitable case for this research. More precisely, the unit of analysis of this research was defined to be the mould injection production – which is one of the four production lines involved in the AC production, as will be further explained in Section 4.1 – and its internal planning procedure and production processes. A detailed introduction of the case company and the unit of analysis follows in Section 4.
3.3 Data collection and analysis
The main data collection techniques used in this research were observations and open face-to- face interviews. In total, the data collection and analysis covered a period of nine weeks, whereby the data was collected and analysed in three phases:
Phase 1: The aim of phase 1 was to get an understanding of the production system of the case company and the operational problems it currently faces. Observation studies were performed on the shop floor of each production line and in the different warehouses. In addition, open face-to-face interviews were conducted with production managers and shop floor workers who were considered to be “key informants” with in-depth knowledge in their area (Voss et al., 2016). The results were used to map the production system of the AC production and to identify
possible inventory points. Furthermore, they helped to focus the data collection for the subsequent phases of the data collection and analysis.
Phase 2: The aim of phase 2 was then to get an understanding of the planning approach and procedures that are applied by the case company within the AC production. Therefore, production planners in the planning department of the mould injection production were observed using a “work shadowing” technique. Observations and findings were discussed both during and after the work shadowing, which allowed an immediate clarification of open questions as well as a further contextual analysis and investigation of specific fields of interest.
Phase 3: The aim of phase 3 was finally to analyse the inventory in the production system of the case company to link it to its sources of variability and the underlying managerial and planning-related decisions. Therefore, further qualitative data was collected through open discussions with production planners, but also quantitative data were collected from the ERP system of the case company. The data comprised 60 production planning data tables from the mould injection planning department, which stated the daily production planning for around 1,000 different product parts for the months June and July 2018. The data were systematically reviewed, structured and analysed to confirm and extend the results from all phases of the data collection and analysis.
Studying both the planning approach and procedures and the different inventories and sources of variability in the case company allowed to link all the different concepts. Moreover, using multiple sources of evidence allowed to study the production and planning system of the case company from different perspectives and helped to increase the research validity through triangulation (Eisenhardt, 1989; Yin, 2011). Instability in the production planning, for example, was reported in interviews, could be observed during the work shadowing and finally confirmed in the ERP data. To further ensure validity and reliability of the research, a protocol was prepared for each company visit, which summarizes the main information that were obtained during the visit, as it is suggested by Yin (2011). To ensure reliability of the research, each protocol was translated and sent to the respective interviewees in the case company to allow them to review and revise the information (Yin, 2011). A detailed overview of the research can be found in Appendix B.
4. Case introduction
Section 4 introduces the case company and unit of analysis of this research. Section 4.1 hence outlines the general approach and structure of the AC production process of the case company, whereas Section 4.2 provides a more detailed overview of the production process (outcome of phase 1).
4.1 The case company and the production process
In total, the case company produces 5 different categories of ACs, involving over 100 different products. The production process can slightly differ for the different categories and products and might be adapted to specific customer orders, as will be elaborated in Section 4.2.
Figure 4.1 shows the overall structure of the production process that consists of four production lines – the mould injection production, the condenser production, the sheet metal production and the control production – which feed into a central assembly line. The sheet metal production thereby covers both a production process in the sheet metal plant as well as a subsequent painting process in a different plant and is thus split in the figure. After being produced in one of the four production lines, all parts are first transported to an external warehouse where they are stored until they are called off for the final assembly.
Figure 4.1: Simplified production process of the AC production
Each of the four production lines can be characterized as a disconnected flow line since (i) the production occurs on a limited number of routings though the process and (ii) the movement between the individual process steps is not fully automated so that inventory can build up between the system (Hayes and Wheelwright, 1979). The same observation holds true for the assembly operation.
4.2 The production process in detail
To gain a better understanding of the AC production process, each production line is briefly explained in the following. An overview of the complete process can be found in the process map in Figure 4.2, which also indicates the possible inventory points in the system. The isolation parts of the ACs are externally sourced and hence neglected in the figure.
§ Mould injection:
The mould injection production is responsible for producing the plastic covers of the ACs and is spread over several buildings and machines (see Appendix C). First, plastic granulate is injected in different injection moulding machines, which then produce and output different plastic covers for different product categories (1). Another type of machine afterwards prints the company logo on the covers (2), before they are packed into boxes and transported to an external warehouse (3) to wait for the assembly (26). For the product category of Vitrine ACs the production process slightly differs as an additional step is required, in which doors are attached to the cover after arrival in the warehouse (4).
§ Condenser:
The condenser production starts with cutting pipes (5) and punching aluminium sheets (6). The pipes and sheets are then assembled and widened (7), before the pipes are soldered together (8). Also these parts are packed (9) and transported to an external warehouse to wait for the assembly. However, before the parts can go into the assembly four further production steps are required; since the connections of the pipes are still missing, the parts have to be bent (10) and soldered together with the connections (11), while the solderings, moreover, have to be checked for tightness with an air (12) and helium test (13). After completion of these four steps, also the condenser parts can be packed (14) and transported to the assembly (26).
§ Sheet metal:
The sheet metal production is responsible for producing the metal cases of the ACs. In the first step, the metal sheets, which are delivered as raw material, are cut into smaller sheets (15).
Three types of machines are then available for stamping the smaller sheets. The first type of computer numerical controlled (CNC) machines stamps small metal parts out of the sheets (16), the second type of CNC machines cuts small and round metal parts out of the sheets (17) and the third type of machines stamps metal parts with different shapes out of the larger metal sheets, thereby using different moulds for stamping (18). The metal parts are bent together in a subsequent step (19) and connected by either welding (20), spot welding (21) or riveting (22), before they are packed and transported to a warehouse (23). Some parts, however, skip the
process step 19 or 19 to 23, respectively, and go directly into the warehouse, as depicted in Figure 4.2. Before the product parts from the sheet metal production go into the assembly (26), they have to undergo a powder coating process in the painting plant (24).
§ Control:
The control production produces electronic controls of the ACs, such as displays and embedded control systems. Since a detailed analysis of the production process in the control plant was not possible in the context of this research, the respective process is depicted as a black box in Figure 4.2 (25).
Finally, the product parts of all different production lines are called off according to the requirements that are stated in the central assembly plan and are merged in a final assembly operation (26).
Figure 4.2: Process map of the AC production
5. Results
Section 5 now presents the results of this research. Section 5.1 starts with explaining the general planning approach of the AC production as well as the related planning procedure in the mould injection planning department (outcome of phase 2). Section 5.2 then outlines the inventory analysis that was performed to identify the existing sources of variability in the production system, before Section 5.3 finally links all the different concepts (outcomes of phase 3).
5.1 Analysis of the planning approach and procedure
Similar to the structure of the production process (see Figure 4.1), the planning process of the AC production is spread over different planning departments, more precisely, over one planning department for each production line and a central assembly planning department. The overall planning process starts in the assembly planning department, which sets up a central assembly plan based on a 1-month forecast that it receives from the sales department of the case company. Thereafter, the central assembly plan is communicated to each line planning department, which sets up its individual parts production plan based on the information provided in the central assembly plan. Because the parts production does generally not commence until the requirements for the parts from the assembly department is known, the production system of the case company shows similarities to a make-to-order (MTO) environment, as it is described by Olhager (2003). In the following of this research, the term
‘order’ is therefore used to refer to orders from the assembly plan or the assembly line, respectively, instead of external customer orders. While the assembly planning department acts as a tact giver and coordinator between the different line planning departments, it is important to note at this juncture that there are no official channels available for direct communication between the line planning departments itself.
In the mould injection plant, which is in the focus of this research, the planning department consists of 6 production planners. Each part type or material to be produced in the mould injection production has a planner assigned to it. The assignment of the part types to the planners, however, is independent from the order, which may lead to the case of several planners being responsible for one order. The case company uses an ERP system and its integrated classical MRP procedure to plan its purchase and work orders. The MRP procedure is thereby enhanced by a capacity checking mechanism, as depicted in Figure 5.1. Each step of the procedure is described in the following.
Figure 5.1 Planning procedure in the mould injection planning department
(1) Get gross requirements from ERP system
The daily planning procedure of a production planner in the mould injection planning department starts with the export of an Excel file from the ERP system of the case company.
The Excel file provides the planner with information about when and in which quantity each part type is required by assembly, covering a planning horizon of 10 days. The due date of the part types hereby corresponds to the planned start date of the assembly. The data file further contains information about the current inventory availability, the assigned planner of each part as well as an aggregate (or daily report) of the orders, which sums up orders for the same part types. For example, if there are three orders that require 200 pieces of part type A on day X the aggregate for part type A on day X is 600. This aggregate for each part type is filtered by the planner and extracted into a new Excel sheet, using a Pivot function to transpose the data. If required, the inventory availability can be cross-checked with more detailed inventory data that can be downloaded from the ERP system by the planner, using a VLOOKUP-function for matching the extracted values. The planner also has access to the central assembly plan in the ERP system for further cross-checking.
(2) Calculate net requirements (based on on-hand inventory)
After extraction of all aggregates, the planner uses an IF-function to calculate the net requirements. For example, if the gross requirements on three subsequent days are 20, 200 and 15, while the available inventory is 225, the net requirements become 0, 0, 10 on these days.
(3) Check capacity and request change of central assembly plan
As soon as the net requirements for each part type are calculated and known, the capacity requirements for each day can be determined and a capacity check can be performed (3). The
available capacity is thereby known by the planner from experience or can be checked in the ERP system. Currently, a total of 150 machines is available in the mould injection plant, whereby 30 machines produce only one part type and run continuously, 60 machines change the part type every two to three days, and 60 machines change the part type on a daily basis.
There are two possible outcomes of the capacity check; either the capacity is insufficient or the capacity is sufficient to produce the orders as stated in the central assembly plan. In the case of insufficient capacity, the planner needs to request an adjustment of the central assembly plan by sending a change request to the assembly planning department. The change request is communicated via mail, whereby the planner in the mould injection plant receives a confirmation as soon as the mail is opened. This confirmation is crucial to ensure that all the change requests are considered and initiated in time. In the case of sufficient capacity, the resulting planning sheet is used as the final parts production plan for execution. A schematic illustration of the planning sheet is given in Table 5.1. A complete overview of the sheet can be found in the Appendix (Appendix D).
Table 5.1: Columns in the requirements plan of the mould injection plant Gross
requirements On- hand
Net requirements
Net requirements per day (number of parts)
1 2 3 4 5 6 7 8 9 10
Important to note at this juncture is that step (1) and (2) of the planning procedure in the mould injection department are completely automated (data is retrieved from the ERP system and transformed with Excel functions), so that no variability can be introduced by the planners during these steps. Also, the BOM is stored and already considered in the extracted data file.
The only manual intervention of the planners, which offers them an opportunity to introduce variability in the system, are hence the capacity check and the subsequent change request.
5.2 Inventory analysis
To investigate the different sources of variability that exist in the AC production system of the case company, the inventory analysis framework of Land et al. (2018) was applied. A general challenge thereby was that the amount of inventory in the production system of the case company was generally too large to be observed on the shop floor. More precisely, the inventory could be physically observed but not quantified or directly attributed to its causes on the shop floor. Quantitative data was, therefore, used in this phase to get an understanding of the amount of the different inventory types in the system and open discussions were conducted to attribute them to their underlying causes.
In total, five inventory types were found to be present in the AC production system of the case company. According to the inventory typology of Land et al. (2018) (see Appendix A), these can be characterized as follows:
§ Inventory type 1:
The first inventory could be attributed to a company policy to have all parts completed 24 hours before they are required in the assembly. With this policy the company aims to buffer against supply uncertainty from the production lines to protect the assembly line from starvation.
Table 5.2: Inventory analysis - Inventory type 1
Type Supply Uncertainty Induced Congestion
Missing input Capacity
Source of variability Supply Appearance of variability Uncertainty
Inventory type 1 hence consists of flow items that are assigned to an order from assembly but have to wait for assembly capacity to become available. The underlying source of variability which causes the assembly capacity not to be available is the uncertainty in the supply from the production lines (see Table 5.2).
§ Inventory type 2:
The second observed inventory occurs because the production planners produce more parts than required in the assembly to buffer against demand uncertainty from the assembly. This behaviour is typically influenced by subjective factors, such as the experience and risk aversion of individual planners. Highly risk averse planners, for example, are more likely to overproduce to decrease the risk of starving the assembly line than planners with a lower risk aversion.
Table 5.3: Inventory analysis - Inventory type 2
Type Demand Safety Stock
Missing input Demand
Source of variability Demand Appearance of variability Uncertainty
Inventory type 2 can thus be classified as flow items, which are not yet assigned to an assembly order, so that demand depicts the missing input. The main source of variability is the uncertainty in the demand from the assembly line (see Table 5.3).
§ Inventory type 3:
The third inventory was present because the ordered amount of parts exceeded the available production capacity of the respective process step. As a consequence, the parts production has
