1
FLOW IMPROVEMENT IN OPERATIONS
MANAGEMENT
2
Master Thesis
DDM Operations management
Sjoerd Mulder
Nieuwe Boteringestraat 7
9712PE Groningen
Student number:
University of Groningen: S2227088
Newcastle University: 170709282
Supervisors:
3
Abstract: Many improvements proposed by the field of Operations Management relate to the control of flows. A new approach has been developed to diagnose the inhibitors of flow in production processes, being different kinds of inventory. Inventory buffers are used to cope with different variability issues in the production process. A better flow is created if inventory levels are reduced. First, this design science research validates the new diagnosis tool by showing that it is capable of identifying different types of inventory and that it generates consistent results. Second, it expands it by giving an overview of the most common improvement measures in operations management and analyses which inventory types they address. The analysis shows that some inventory types cannot be reduced with methods described in literature. Furthermore, managers still need a method to test to ensure they are implementing the appropriate measure. The new diagnosis tool appeared to be able to check if improvement measures are a good fit with the issues they are aiming to resolve.
4 CONTENTS
Preface ... 6
1. Introduction ... 7
2. Theoretical Framework ... 9
2.1 Root cause identification framework ... 9
2.2 Framework analysis ... 10
3. Methodology ... 11
4. Part one: Validating the new approach ... 14
4.1 Production process analysis ... 14
4.2 Approach analysis ... 17
5. Part two, section one: Analysis of common improvement measures in operations management . 19 5.1 Common improvement measures ... 19
5.1.1 Small lot size ... 19
5.1.2 Set up time reduction ... 19
5.1.3 Pull Production ... 20
5.1.4 Cell Layout ... 21
5.1.5 Poka-Yoke ... 21
5.1.6 Just In Time delivery ... 21
5.1.7 Total Preventive Maintenance ... 22
5.2 Theory of Swift, Even flow ... 22
5.3 Analysis of variability improvement measures in operations management literature ... 23
6. Part two, section two: Improvement measures analysis ... 25
6.1 Basic end date spreading ... 25
6.2 Order release rules ... 26
6.3 SMED project ... 27
6.4 new approach as improvement measure analysis tool ... 28
7. Discussion ... 29
8. Conclusion ... 30
5
References ... 32
Appendix ... 34
Appendix A: interview protocol ... 34
Appendix B: Translation interview protocol ... 36
6 PREFACE
Before you lies the master thesis “Flow improvement in Operations Management - from root cause identification to flow control”. A research conducted to provide better methods to control the flow in an organisation. It has been written to complete the dual award in operations management from the University of Groningen and Newcastle University. I conducted this research from August to December 2018.
This project was undertaken at the request of Broekema in Veendam. The research was difficult and many changes were made along the way. Fortunately, people involved in the company, and particularly Gerda Bos and Coen den Hertog, as well as my supervisors from both institutions, dr. Martin Land and Prof. Chris Hicks, were always available to assist me and willing to answer my questions.
I would like to thank my supervisors for their excellent guidance and support during this process. I would like to thank the company for giving me all the support and facilities I needed to conduct my research. I also would like to thank all the foremen and other employees from Broekema for answering my questions and their openness.
I would like to thank Jasper Dijkhoff and Ruben Meijer for their help during the process. They were always prepared to give me feedback and were willing to debate the topics present in this thesis. I want to thank Leah Hamilton for helping me finalizing the thesis.
Lastly I would like to thank all my friends and family. If I ever lost interest, you kept me motivated. Special thanks to my parents for always supporting me throughout my entire studies.
I hope you enjoy your reading. Sjoerd Mulder
7 1. INTRODUCTION
A central concept in the field of operations management is flow. When materials are processed, they will progress through a series of activities where they are transformed. Between activities products may dwell for some time in inventories. These waiting times do not add value to the product. The materials moving through the process, nor the assets performing the activities may be fully utilized (Slack, Brandon-Jones and Johnston, 2016). When material waits in inventories to be used in the production, the flow of the organisation does not reach its full potential. Flow has been a predominant factor of most operations management theories but companies still struggle to realize flow improvements (Land et al., 2018). To face issues created by flow control problems, the root cause of these problems needs to be identified and addressed.
8
This study will use design science as a methodology as part of a larger design science research since the problem is raised by operation managers. Design science is driven by field problems, knowledge is developed by engaging with real-life operations management problems (van Aken, Chandrasekaran and Halman, 2016). In the past the company subject to this research implemented improvement measures which were not successful. They were not a fit with the root cause they wanted to address. A tool would help managers to check if the measure is a good fit to avoid implementing wrong and costly improvement measures. The field driven problem in this thesis is therefore the need for a better method to cope with flow control issues. The research questions this thesis answers is:
How can managers improve flow by addressing flow control issues? To answer this question, three sub question will be studied successively:
(1) Is the research approach developed by Land et al (2018) a good method to identify the root causes of inventory?
(2) Which improvement measures are currently known to reduce variability in the field of operations management?
9 2. THEORETICAL FRAMEWORK
This thesis will expand the research done by Land et al (2018) by validating and improving the approach. In this section the background of the new approach will be discussed.
2.1 ROOT CAUSE IDENTIFICATION FRAMEWORK
Operations management involves in what way organizations design, improve and deliver products (Slack, Brandon-Jones and Johnston, 2011). Many production companies have flow control problems where the production process does not reach its full potential. To improve flow organisations should know what the root causes are that inhibit the flow. For managers and researchers who want to improve flow it is essential that they know why inventories exist (Land et al., 2018).
There are many studies done to reduce variability issues. However, none of them focussed on the identification of the different types of inventory. The root cause of the problems needs to be known to address it and have an effective measure in place. Land et al. (2018) solved this problem by developing a comprehensive framework to understand why flow items wait in inventories. They found that each type of inventory relates to a core source of variability. These root causes can be found by following a systematic procedure where the following questions are asked:
1. For which process input (missing input) are the flow items in this inventory waiting? 2. Which input creates the main source of variability that causes missing inputs? 3. What form does this variability take?
10 2.2 FRAMEWORK ANALYSIS
The new approach from Land et al (2018) is designed as a diagnosis tool. With this new diagnosis tool managers can find the root causes for certain inventory points. However, it is not clear what managers can do to resolve these issues and how these root causes can be taken away. Common literatures focuses on variability issues but it is not specified which root cause they address. It is not known if a manager identifies, for example, an inventory being Capacity Anticipation Induced Congestion, which measures he needs to take to resolve the problem. Even if there is a pretty good understanding on the root cause of the problem, companies do not know if the measures created are a good fit for these specific problems. This research aims to create a method which allows managers to create a better flow control. First, it wants to validate the work done by Land et al (2018) and check if it is indeed a good diagnosis tool which will produce consistent results. This will be important because managers need to know what the root cause of the problem is in order to take it away effectively. For example, inventory can be accumulated at a certain point. A simple solution would be to raise the capacity of the machine that has to process the items that accumulate at this point. But if the source of the problem lies with demand or flow item problems, a larger machine capacity will not resolve the problem.
11 3. METHODOLOGY
This thesis is part of a larger design science research cycle as described by e.g. van Aken (2004). The connection between knowledge and practice in design science lies in the creation of useful things that are established by scientific knowledge (Wieringa, 2009). Design science started with the work of Simon (1996) who introduced the science of design. It aimed at improving actual situations. Later Hevner et al. (2004) created a methodological framework for design science with the emphasis on creating new and innovative artefacts. Design science tries to find solutions for problems faced by industry and translates these solutions to more generic designs. These problems can be described as practical problems. To tackle practical problems and evaluating these solutions, the goals of the stakeholders need to be addressed (Wieringa, 2009).
Since the goals of the stakeholders change over time the design science method involves testing the design through a largely iterative process. A design science cycle contains the following steps: Problem investigation, treatment design, design validation, treatment implementation and implementation evaluation (Wieringa, 2013). The design cycle is not a chronological framework. Designers move back and forward through phases and they often redesign previous phases. Having a grounded argument for the design is not enough, it needs to be evaluated repeatedly (Hevner, 2007). As a consequence, this study needs to redo some phases done by Land et al. (2018) to evaluate the approach properly.
This research consists of two phases. The first phase will be used to validate the new approach to contribute to the design science cycle. The second phase will aim to contribute to literature by expanding the new approach and make it more than a diagnosis tool.
12
Second, it will be compared to the original research to check if the new approach generates consistent results. This research is done at the same company as the initial research done by Land et al. (2018) to determine if it generates consistent results when the analysis is redone by another researcher at another point in time.
The second phase of this research will aim to expand the new approach created by Land et al. (2018) by analysing the most common improvement measures in the field of operations management that focusses on the reduction of variability in the production process. Furthermore, it will be tested if and how this new approach can be used to check if the flow improvement measures are a good fit with the problem they are aiming to resolve. First, an overview of the most common flow improvement literature that focuses on variability issues will be given and it will be analysed which missing input, main source of variability and which form of variability the measure addresses. This will be linked to the root causes of inventory that can be identified by the new approach.
Second, improvement measures that are implemented or proposed by the company subject to this research will be analysed using the new approach. It will be tested if and how the new diagnosis tool can be used to analyse flow improvement measures. The data gathered by the validation of the new approach will be used together with semi structured interviews performed with the production manager and the planner of the company. The diagnostic analysis will be performed by analysing each step in the production process where inventory could be aggregated together with the production manager. The semi structured interviews are conducted to determine how these measures are aimed to reduce certain inventory levels. It will be analysed if these measures are the right fit for the root causes that are aimed to be reduced by checking if they are indeed addressing the right missing input, main source of variability and form of variability, as identified during the validation of the new approach. These interviews will be conducted in Dutch. The Dutch version of the interview guide of the semi structured interview can be found in appendix A. The English translation can be found in appendix B. An overview of the structure of this research is shown in table 3.1.
Table 3.1: Research structure
Part Subject
Part 1 (Chapter 4) Validation of diagnosis tool created by Land et al. (2018)
Part 2, section 1 (Chapter 5) Analysis of common improvement measures and linking them to the inventory types of the new approach created by Land et al (2018)
13
Based on the above study setup, this study can be positioned more precisely within the field of design science. Gregor & Hevner (2013) created a design science research knowledge contribution framework. This research would be classified as an “improvement” project. Many companies face flow control issues for many years. This problem can thus be specified as known and mature. The goal of improvement projects is to create better solutions to known problems (Gregor and Hevner, 2013). The known problems are the following: The missing of a good diagnosis tool to find the root causes of inventory and, finding and checking appropriate improvement measures once these root cause is known. With the completion of this thesis managers and researchers are able to do both.
14 4. PART ONE: VALIDATING THE NEW APPROACH
This section aims to validate the new approach. The validation of the new approach will be conducted at the same company as the first research was executed by Land et al. (2018). The new approach can be considered validated if the approach is indeed capable of identifying the root causes of the inventory present at a production process and if the new approach generates consistent results. The research is conducted at the same company as the initial research to validate that the approach does indeed generates consistent results. At the end of this section the results of both tests will be discussed.
To perform the analysis, a detailed flow chart of the process will be used. Together with the production manager, the process steps are analysed and the reasons for the different types of inventory will be given until it is clear what the missing input, main source of variability and form of variability of the different inventory points are. When these are known, inventory points can be labelled according to the labels shown in figure 2.1. A flowchart of the process researched is shown in figure 4.1.
Figure 4.1: Detailed flow chart
4.1 PRODUCTION PROCESS ANALYSIS
15
Some of the steel rods can only be delivered by one supplier. The lead times for these steel rods are uncertain due to the production process of the supplier. These inventory items can be labelled as Supply Safety Stock.
There are two forms in which the steel arrives at the facility. It can either be delivered in rods or in coils. Coils need to be cut completely into rods since changing coils in between orders is very time consuming and is therefore avoided. Since it is preferred that steel coils need to be cut until the coil is completely used, the company combines orders of the same type of steel. Some of the inventory at point a can thus be classified as Batch Waiting Time.
The belts have a shorter lead time compared to the rods. The decoupling point before punching is not a customer order decoupling point since the punching process is operating at full capacity the entire time. The flow items will be coupled to customer orders after the belts are punched. Since the belts arrive in large orders inventory point h is classified as Supply Cycle Stock.
The majority of the belts that the company receives are belts that need to be punched. There are 2 types of machines present at the facility. One type can process belts that are not punched yet and the others can only process pre-punched belts. Punching belts to feed the second type of machine is done by a puncher that works at full capacity and cannot cope with fluctuations in demand. The customer order decoupling point is at inventory point i because the items at this point are waiting to be coupled to an order. Once the order is received these belts are cut and processed. There are many varieties in the types of belt the company can deliver. Due to the fluctuation in orders for each type of belt the company uses Demand Safety Stock at costumer order decoupling point i. Furthermore, since the punching process works on full capacity and cannot cope with fluctuations in demand, additional inventory is kept at point h. When demand is low, the punching machine will process more orders to cope with a peak in demand which the machine is incapable to handle. This type of inventory is called Demand Anticipation Stock. When cutting belts (process number 9), different orders are combined. This will reduce the setup time of the process and is highly preferred by operators to changeovers in different kind of belts. Therefore some of the inventory at point i is not waiting for demand but is waiting until they can be processed in a batch. This inventory is identified as Batch Waiting Time.
16
However, Batched Supply Induced Congestion is not the primary cause of inventory at point c (prior to pressing). Pressing is a process which is considered a large bottleneck due to its set up time. Capacity at these machines is limited. Material will wait until it can be processed and can, therefore, be labelled as Batch Transformation Induced Congestion. Sometimes rods need to be processed by multiple presses and will have an even longer lead time.
Steps 4 and 5, hardening and discharging, are optional processes. First, the rods move into an oven in a continuous flow. After receiving the heat treatment the rods will be discharged in another, large oven. This process is time-consuming. The discharging oven has a large capacity and the rods are stacked upon a large rack prior to the discharging treatment. The first oven has a large setup time. It takes a long time to set the oven to the desired temperature, which is expensive. Due to these high set up costs in combination with the overcapacity of the oven, it is decided to let it run for only a couple of days a week. Therefore, inventory at point d is considered to be Batch Transformation Induced Congestion. As mentioned before, the output of the pressing process is also batched. Since the hardening oven works in a continuous flow and the supply of flow items is batched some of the inventory is identified as Batched Supply Induced Congestion. The discharging oven will wait until it can process enough rods to reach full capacity. Therefore, inventory at point e can be classified as Batch Waiting Time.
After pressing or discharging, rods will continue to the riveting department, be cast or a loader is added to the rods. There are many different types of loaders of which a customer can choose from to be added to the final product. Adding the loaders to the rods can take a lot of time. Demand for the different kind of loaders is fluctuating and hard to predict. Inventory point g is therefore classified as Demand Uncertainty Induced Congestion. Demand for a casted rod is also fluctuating. Inventory point f can consequently be labelled as Demand Uncertainty Induced Congestion as well.
After cutting, belts can follow two paths. If they get a hinge joint they will first receive rivets and afterwards the hinge is attached to the belt. If the customer requires one of the other joints (e.g. an endless belt) the joint is made before rivets are inserted. At inventory point j, belts are waiting to be processed by the machines that will insert the rivets. The units will stay relatively short at this inventory point. Inserting rivets is a short process and can be done quickly but the belts will mostly arrive in batches since multiple belts are cut at once. Inventory point j is labelled Batched Supply Induced Congestion. It is hard to predict when a customer requires a certain joint. The joint department cannot always cope with peaks in demand for certain types of joint (e.g. an endless joint) and inventory will then be accumulated at inventory point k. Inventory at this point is therefore labelled as Demand Uncertainty Induced Congestion.
17
which are finished, several semi-finished products are waiting until they can be assembled. Each department optimizes their own schedules and the delivery to the riveting department is not synchronized. These inventory points can be classified as Supply Uncertainty and Batched Supply Induced Assembly Waiting Time for the unplanned and planned schedules respectively.
4.2 APPROACH ANALYSIS
The aim of phase one is to validate if the new approach is able to identify different inventory points and to check if the new approach generates consistent results. Using the three questions of the framework, researchers and managers were able to label all the inventory points investigated. There was not a single inventory point of which the root cause could not be identified. It can be concluded that the new approach is able to identify ten different inventory points in this production process when the questions provided are used.
The second validation concerned the results of the new approach. The labels of different inventory points of the research done by Land et al (2018) compared to this research can be found in table 4.1 where the inventory points Identified by Land et al (2018) are marked with an L and the inventory points identified by this research are labelled with an M. Please note that the research of Land et al (2018) did not include the inventory points g and k.
18
19
5. PART TWO, SECTION ONE: ANALYSIS OF COMMON IMPROVEMENT MEASURES IN OPERATIONS MANAG EMENT
This chapter consists of two parts. First the most common literature that proposes improvement measures in the field of operations management that focuses on variability issues will be described. Second, it will be analysed if these measures can be linked to the root causes of inventory they aim to address.
5.1 COMMON IMPROVEMENT MEASURES
There are several solutions to improve the flow in an organization that addresses variability issues. The majority of measures come from the principles that are linked to lean manufacturing. These solutions often assume certain flow inhibitors and then apply methods to reduce or get rid of the negative effects caused by the inhibitor. This section gives an overview of the most common and accepted flow improvement methods. It will first give a short description of the measure and will than analyse how it will reduce a certain kind of variability.
5.1.1 SMALL LOT SIZE
Many factors are influenced by the size of the batches in which a company produces its products. Cost, quality, lead time and the flexibility of the operation are all influenced by lot sizing. Large lots were favoured many years in industry to avoid large set up and order costs. The drawback of large lot size is that the lead times are longer and the flexibility of the production process is reduced. To be more lean and to create a better flow small lot sizes are favourable. Buffers, however, still need to be present to cope with any variability present in the operations (Nicholas, 2011). With a smaller lot size, the transfer batches (items that go from one department to another) can be reduced. The supply of items could be more smooth and congestions are less likely to appear since there will be a reducing in the batch sizes that arrive at a machine which reduces batched supply. However, it will probably create more set ups. This could have an impact on the capacity of the machines and can generate a congestion at the inventory point that is located before the production process which is going to produce in smaller lots.
5.1.2 SET UP TIME REDUCTION
20
There are many ways to reduce set up times but the Single Minute Exchange of Dies (SMED) methodology created by Shingo (1985) is the most common. This four step method is aimed to reduce or eliminate the time and need for set ups. When Setup time reduction is realized the entire production process will probably be less complex. Planning could become easier since the process gained flexibility. It could reduce the batched supply of items through the organization and create more capacity for the machines. Furthermore, due to the reduced time needed for set ups, batches probably have less waiting time until they are processed. This could reduce the batched waiting time. Furthermore, since complexity of the planning could be reduced, the assembly of different items can be synchronized more easily.
5.1.3 PULL PRODUCTION
Realising short throughput times could be achieved by optimizing material control. The flow of material on the shop floor can be regulated by the authorisation of the start of a job, creating a priority list for jobs, releasing new material to the shop floor and initiation succeeding activities (e.g. Transport) (Riezebos, 2010). Pull production systems are types of material control systems. They control which material is allowed to flow through different stages in the production process. A pull system is able to react to changes and problems in the production process. When one of the production processes faces problems the upstream processes do not receive any signals to produce (Nicholas, 2011).
In literature, there are two main pull production control methods. One is Kanban, which uses authorization cards which states if products can be produced or moved and what kind of material is needed, its destination and source (Nicholas, 2011). Another pull production method is POLCA (Paired-cell Overlapping Loops of Cards with Authorization). POLCA aims to make the principals that work in a Kanban system applicable to a make to order production facility (Riezebos, 2010). POLCA was first described by Suri (1998) and is a quick response manufacturing technique to control the material flow. This method is best to be used in the context of a cellular organisation with a high level of material requirements planning. The systems tells a cell when that cell is authorised to work.
21
5.1.4 CELL LAYOUT
For operations that are considered being a relatively high variety production process, a change towards a cell layout could result in an improved flow. A cell layout is a layout where the machines required to produce different parts of the operation are clustered. A product can move through different cells until it is considered a finished product. Cell layouts try to bring some order in complex flow operations (Slack, Brandon-Jones and Johnston, 2016). In a cellular layout, autonomous groups (consisting of workers and machines) work on a set of part types (Bokhorst, Slomp and Gaalman, 2004).
It is possible to create a fast throughput of items and create more flexibility with a cell layout. However it can be costly to implement it could require investments in additional equipment since some machines are used in multiple processes (Slack, Brandon-Jones and Johnston, 2016). Cellular manufacturing aims to move a high variety of items through the process as quickly as possible while reducing waste. It could be capable of a reduction in the transportation of supply items and the size of transfer batches.
5.1.5 POKA-YOKE
Poka-yoke is a method which focuses on the design of products and operations. It is a concept of fail-saving the process and trying to eliminate human error. Human mistakes are to some extent inevitable and the poka-yoke method focusses on preventing human error becoming defects. Poka-yokes are simple systems or devices that are implemented into a process to avoid mistakes (Slack, Brandon-Jones and Johnston, 2016). Poka-yokes are important in pull production systems where, because of small inventories, stoppages anywhere interrupt the entire process (Nicholas, 2011). Poka-yoke methods can create a higher quality of products. Prevention of defects in the process before they appear is the best way to reduce defects (Dudek-Burlikowska and Szewieczek, 2009). It is a measure that focusses on a first-time right principle. Uncertainty whether products with the desired quality will be supplied by other departments could be reduced. When there are less defects and less measures needed for quality checks the flow could be improved. It could also be used to fail proof set ups. When it is used as a set up time reduction measure it could affect the same root causes as the SMED methodology described in section 5.1.2.
5.1.6 JUST IN TIME DELIVERY
22
inventories in the entire factory it is mainly focused on the arrival of products. With a good functioning JIT method the variability in product arrival time will be reduced. When these parts are used in assembly, the supply of flow items in assembly will be less uncertain. When a company uses the JIT method the batches it receives will be the same size as the items that they need, eliminating batched inventory problems.
5.1.7 TOTAL PREVENTIVE MAINTENANCE
Total Preventive/Productive Maintenance (TPM) is a system that focusses on the maintenance and improvement of machines in an organisations. TPM can be defined as: “a system that is designed to maximize equipment effectiveness (improving overall efficiency) by establishing a comprehensive productive-maintenance system covering the entire life of the equipment, spanning all equipment-related fields (planning, use, maintenance, etc.)” (Tsuchiya, 1992, p.4). TPM provides a company-wide approach to maintenance management. It can be separated in and long-term elements. The short-term elements will focus on maintenance programs for the production and maintenance department while the long-term elements will focus on the design of new equipment and the elimination of downtime. Long-term elements will involve many areas of the organisation (McKone, Schroeder and Cue, 2001). TPM is designed to prevent stoppage, defect and speed losses as well as speed reduction caused by several forms of failures by improved manufacturing methods and equipment maintenance (Chan et al., 2005). With TPM in place it is less likely to have a breakdown or a failure. The machines should have the ability to operate according to their capacity. This could, among other things, reduce the need to buffer for the chance of a breakdown.
5.2 THEORY OF SWIFT, EVEN FLOW
As seen in the previous section, lean provides several improvement measures that focusses on reducing variability. There are other theories that also focusses on certain type of variability issues. Most of the measures linked to these theories could have the same effect as some of the lean principles. However, these theories could be more applicable than lean and thus be a good alternative to improve the flow in the organisation.
23
products since materials wait to be produced. The swiftness can be measured in throughput time of materials (Schmenner and Swink, 1998).
To create a more even flow, the variability of the demand in production and the process operations steps need to be reduced. It is better to produce with a level production plan than with production plans that have irregular quantities or due dates. The greater the variability in the process, the less productive it is. Quality management is important to the Swift, Even Flow theory. It helps to lower the variability in the products and will avoid bottlenecks (Schmenner and Swink, 1998). This theory could be more applicable if the problems are mainly caused by demand variability issues since that is the main focus of this theory. However, most of the improvement measures will have the same effect as the improvement measures described in section 5.1.
5.3 ANALYSIS OF VARIABILITY IMPROVEMENT MEASURES IN
OPERATIONS MANAGEMENT LITERATURE
Improvement measures described in section 5.1 are all created to contribute to a better flow in the organisation. Each of these measures is aiming to reduce a different kind of variability in the production process. An overview of the most common improvement measures and the relation to the new approach can be found in table 5.1. This overview is made by analysing the different methods according to the questions that need to be asked to identify the root causes of inventory. For example, Total preventive maintenance focusses on the reduction of uncertainties in the production process by making the processes and equipment more reliable. This influence the arrival of different Flow Items and Capacity as missing inputs. The main source of variability is the supply of Flow Items and the Capacity. It makes both more reliable. The form is Uncertainty so the inventory types the measure influences are: Supply Uncertainty Induced Congestion, Supply Uncertainty Induced Assembly Waiting Time, Capacity Uncertainty Induced Congestion and Capacity Uncertainty Induced Assembly Waiting Time. An overview of all the different inventory types and their abbreviation can be found in appendix C. In figure 5.1 the framework of Land et al. (2018) is used to give an overview of which inventory points are addressed by the common flow improvement measures. The inventory points addressed by the common literature are highlighted. The common improvement measures will give generic solutions for companies facing one of these issues. It gives a clear overview of possibilities, but these measures need to be customized in order to be implemented.
24
Table 5.1: Overview of the common improvement measures in relation to the root causes they aim to address
Improvement
measure Missing Input
Main source
of variability Form of variability
Affected root cause
Small lot size Capacity, Flow Items Supply of Flow Items, Capacity Batched BSIC, BSIAWT, BTIC, BTIAWT, BWT Setup time reduction Capacity, Flow Items Supply of Flow Items, Capacity Uncertainty (assembly items), Predicted Fluctuation (assembly items), Batched SUIAWT, SAIAWT, BSIC, BSIAWT, BTIC, BTIAWT, BWT Pull production Capacity, Flow Items Demand, Supply of Flow Items, Capacity Uncertainty, Predicted Fluctuation, DUIC, DUIAWT, DAIC, DAIAWT, BDIC, BDIAWT, SUIC, SUIAWT, SAIC, SAIAWT, CUIC, CUIAWT, CAIC, CAIAWT,
Cellular layout Flow Items (different type) Supply of Flow Items, Capacity Uncertainty, Predicted fluctuation SUIC, SUIAWT, SAIC, SAIAWT
Poka-Yoke Capacity, Flow Items
Supply of Flow Items,
Capacity
Uncertainty SUIC, SUIAWT, CUIC, CUIAWT
JIT delivery Demand, Flow Items Supply of Flow Items Uncertainty, Predicted Fluctuation, Batched SSS, SUIAWT, SAS, SAIAWT, SCS, BSIAWT Total Preventive Maintenance Capacity, Flow Items (different type) Supply of Flow Items, Capacity
25
6. PART TWO, SECTION TWO: IMPROVEMENT MEASURES ANALYSIS
Tailor made solutions cannot be found in literature. Furthermore, as is seen in section 5, not all inventories can be covered by the most common flow improvement measures. Some of the problems also arise from the complexity of the specific production situation in a company. It would be preferable for managers if they are able to analyse other options to improve the flow in the organization to avoid the implementation of wrong and possibly expensive measures. This section aims to analyse if the new diagnosis approach created by Land et al. (2018) described in section 2.1 can be used, next to the diagnosis tool, to check if improvement measures proposed by the company are a good fit with the problem that they are trying to resolve.
As mentioned in section 4 the main issue that the company subject to this research faces concerns Supply Uncertainty and Batched Supply Induced Assembly Waiting Time. These problems are assembly problems at inventory point l (figure 4.1), right before riveting. At this point most of the inventory is accumulated. To reduce the inventory levels at this point, the company introduced a number of improvement measures. First, these measures will be analysed by using the new approach. Second, it will be discussed if the new approach provides is an appropriate tool to analyse improvement measures. These improvement measures are proposed by the company based on the knowledge of the earlier diagnosis done in Land et al. (2018). In this section, first, the improvement measures and the aim they have according to the company are explained. Using the new approach, the measure will be analysed. To be more precise, there will be an assessment to determine if it addresses the correct missing input, source of variability and form of variability, as it has been identified as the root cause of the inventory at the considered point. If there is a match it can be concluded that the measure fits with the problem the company tries to resolve or reduce.
6.1 BASIC END DATE SPREADING
26
expects to have more changeovers. This can result in a larger overall set up time and less capacity. However, it could also result in smaller batches that are sent to the next department. The main source of variability the company addressed is the supply of flow items. By making a smoother planning, the supply of the items to the successive station is expected to be more even. The forms of variability that are addressed are uncertain and batched supply. The uncertainty is expected to be less since the orders will be produced on a daily basis and are planned more precise. The riveting department is expected to have a better forecast of arriving items. The batched supply of flow items will be reduced since there will be less orders combined and the production will be carried out closer to its predetermined schedule. Currently orders are planned by using a backwards scheduling method starting from the assembly process. This provides deadlines for the production of the individual items to be assembled. Individual items are given an end date. This end date is the last possible date to get an item delivered on time to the costumer, accounting for assembly lead time. The individual items are scheduled according to capacity of the machines and the planned lead time of the individual production processes. Items are not interchangeable due to the high variety of the products the company produces.
The different items to be assembled, e.g. rods and belts, are not synchronized except for the originally planned end dates. Due to capacity restrictions items can be scheduled earlier than they are needed, earlier than their end date, either within the rod department or the belt department. This can result in long wait times until the items needed for assembly arrive at the riveting department. Furthermore, operators are allowed to combine orders to reduce the total set up time, which is another issue at the company. This results in a further desynchronization of the different assembly items and thus a large amount of work in progress. Sometimes orders at one department are rushed to deliver them “on time” when other items of the same order still need to start their production process. These issues are expected to reduce with this new measure.
When the missing input, source of variability and form that this measure is addressing is analysed, it can be determined that it could influence Supply uncertainty induced Congestion, Supply Uncertainty Induces Assembly Waiting Time, Batched Supply Induced Congestion and/or Batched Supply Induced Assembly Waiting Time where this measure is implemented. The aim was to reduce Supply Uncertainty and Batched supply induced Assembly Waiting Time at inventory point l ( figure 4.1). Therefore, it can be concluded that this measure will have an effect on the root causes of inventory point l.
6.2 ORDER RELEASE RULES
27
demand. The belt department however, is very flexible. The set up times are negligible and the lead times are significantly shorter than the lead times for rods.
The operation of the rods department will be leading for the belts department in the new situation. The end date for the items for both departments will not change but the order release to the operators will. Belts will only be released for production when the first step of the production process for rods, cutting, is finished. Only when the Rod Department signals that rods for an order are ready, operators can see the corresponding belts appear in their work list. Since errors still exist in the production process and operators are allowed to combine orders, not every order is produced according to schedule. This measure is expected to create a better synchronization between the rods and belts. Items laying idle at the riveting department ready to be assembled are expected to be reduced. This measure aims to reduce the Supply Uncertainty Induced Assembly Waiting Time when items of different departments need to be assembled and when it is difficult to produce according to schedule.
The missing input that this measure could address are the flow items since it focusses on an order release for the different items. This measure concerns the supply of the different flow items to the riveting department. It could reduce the time gap of the different items that are send to the riveting department. This department could have less uncertainty of the arrival of the different items. The production of rods and belts are linked to each other so when one of the items arrive the other one is expected soon to follow. This could reduce the Supply Uncertainty Induced Assembly Waiting Time. It can be concluded that this measure is a fit with the problem that is aimed to be resolved.
6.3 SMED PROJECT
A SMED analysis is being conducted for one of the presses. This project is still ongoing, but some small easy fixes are in place. The SMED project will be continued by the company in the near future. Due to the basic end date spreading, operators at the presses will have less room to combine orders. Without setup time reduction, this could create problems for the pressing department since they need to set up the presses more often. Inventory at point c (figure 4.1), Batch Transformation Induced Congestion, could be increased due to this measure. The SMED project aims to reduce the negative side-effects of the basic end date spreading and to create more flexibility for the pressing department. This SMED project is seen as a necessity by the company when the organisation faces a peak in demand. The aim is to reduce the Batch Transformation Induced Congestion at inventory point c as well as the Batched Supply Induced Assembly Waiting Time at point i since the batches that arrive and leave this station will be reduced in size.
28 6.4 NEW APPROACH AS IMPROVEMENT MEASURE ANALYSIS TOOL
As seen in analysis of the different improvement measures the new approach could be capable of performing a check to see if the measure is a fit with the company. This test is however quite basic and only shows if modifications to an common improvement measure and custom made improvement measures addresses the correct type of inventory. It does not show what the effect (impact) will be and what possible negative side effect are to processes outside of the scope of the measure.
29 7. DISCUSSION
This study focussed on flow improvement measures in Operation Management. This field was chosen because flow control issues play a predominant role and managers continually struggle to achieve an optimal flow. The researched validated the new approach of Land et al. (2018) as a good diagnosis tool. This was done by repeating the research as done in the original paper at the same company. We answered the three questions, which form the core of the diagnosis approach, for each of the inventory points present at the company. Based on this, all root causes of inventories could be found. The repetition of the diagnosis of the original paper was executed to check whether the method generates consistent results and whether these results do not dependent on the researcher that executes the diagnosis. Although there were some differences in the studies, the method still provides consistent results. These differences were most probably the cause of a shifting root cause of the inventory points over time. The initial method developed by Land et al. (2018) found the root causes by identifying the variability issues faced at each of the inventory points. To determine which measures could reduce variability in the field of operations management, common improvement measures were analysed. The majority of these measures are linked to the theory of lean. These measures were chosen since lean management provides a set of tools which reduces the variability in production processes. Other theories, like the theory of Swift, Even Flow, will have measures that are similar to the measures linked to lean. The measures analysed could therefore be determined as representative for the majority of flow improvement measures. These topics appeared to be chosen well since they address the greater part of the inventory points. However, other, less generic, methods could be studied as well. Especially for the root causes created by (customer) demand as being the missing input. For example, forecasting could reduce Demand Safety Stock since it could probably provide more demand certainty.
30 8. CONCLUSION
This theses aimed to answer the question: “How can managers improve flow by addressing flow control issues?”. This is done by 1) validating the approach developed by Land et al. (2018), 2) to improve it by determining which improvement measures are most common to reduce variability in the field of operations management and 3) to determine how the new approach could be used to analyse flow improvement measures.
The first phase of the research validated the approach by Land et al. (2018) by redoing the initial diagnosis to determine which variability issues were the root cause for specific inventory points at a medium sized capital goods manufacturer. The approach is capable of identifying ten different kinds of inventory types and it generated results consistent with an earlier diagnosis that used the same approach in the same case. The approach generates the best results if there are more iterations of the research performed at different time intervals.
The second phase of this study aimed at finding solutions for the problem that managers still may not know what to do to improve flow, once the different types of inventory points are identified. First, the most common improvement measures were analysed. The improvement measures analysed were: Small lot size, Setup time reduction, Pull production, Cellular layout, Poka-Yoke, JIT delivery and Total preventive maintenance. It was found that inventory of multiple types can be reduced by using one or several of these methods. A table is provided that reveals which method may help to reduce each inventory type. With the help of this table managers could determine which method they can costumize and implement to create a better flow by adressing variability issues.
Not all inventory types can be improved with the methods existing in common literature, or are a suitable option for company specific problems. Managers still need the ability to test if the measures they are proposing are a good fit with the problems they are aiming to resolve. Based on the diagnosis provided, the production manager in the case company, initiated a number of measures to improve flow. The method created by Land et al. (2018) appeared to be able to determine if there is a fit with the proposed improvement measure and the type of inventory which is aimed to be reduced. This could be done by testing if the correct missing input, main source of variability and form of variability is addressed by the measure.
31 8.1 LIMITATIONS AND FUTURE RESEARCH
This research focussed on variability reduction in operations management. Operation management is not the only sector which faces difficulties with flow control. This diagnosis could probably also be used in a service or health care environment, if flow control is an issue. This research did not include these fields and new research could be conducted to analyse if this new approach is also a good diagnosis tool for these fields. If the new approach appears to be a good diagnosis tool as well, common improvement measures in these field could be analysed to determine which measures could improve each type of inventory.
As shown, not all inventory types could be addressed with a common improvement measure yet. More research need to be conducted to these inventory types and how these can be addressed. Studies could focus on these types to create new generic measures or methods which are able to reduce the levels of these inventories.
32 REFERENCES
Aghazadeh, S. M. (2004) ‘Does manufacturing need to make JIT delivery work?’, Management Research News, 27(1–2), pp. 27–42.
van Aken, J. E. (2004) ‘Management research on the basis of the design paradigm: The quest for field-tested and grounded technological rules’, Journal of Management Studies, 41(2),
van Aken, J., Chandrasekaran, A. and Halman, J. (2016) ‘Conducting and publishing design science research: Inaugural essay of the design science department of the Journal of Operations Management’, Journal of Operations Management. Elsevier B.V., 47–48, pp. 1–8.
Berk, J. (2010) Cost Reduction and Optimization for Manufacturing and Industrial Companies, Cost Reduction and Optimization for Manufacturing and Industrial Companies. Wiley.
Bokhorst, J. A. C., Slomp, J. and Gaalman, G. J. C. (2004) ‘Assignment Flexibility in a Cellular Manufacturing System - Machine Pooling versus Labor Chaining’, in, pp. 1050–1057.
Chan, F. T. S., Lau, H. C. W., Ip, R. W. L., Chan, H. K. and Kong, S. (2005) ‘Implementation of total productive maintenance: A case study’, International Journal of Production Economics, 95(1), pp. 71– 94.
Dudek-Burlikowska, M. and Szewieczek, D. (2009) ‘The Poka-Yoke method as an improving quality tool of operations in the process’, Journal of Achievements in Materials and Manufacturing Engineering, 36(1), pp. 95–102.
Gregor, S. and Hevner, A. R. (2013) ‘Positioning and presenting design science research for maximum impact’, MIS Quarterly, 37(2), pp. 337–355.
Hevner, A. R. (2007) ‘A Three Cycle View of Design Science Research’, Scandinavian Journal of Information Systems, 19(2), pp. 87–92.
Land, M., Thürer, M., Stevenson, M. and Fredendall, L. (2018) Flow improvement - A Design Science Approach.
McKone, K. E., Schroeder, R. G. and Cue, K. O. (2001) ‘The impact of total productive maintenance practices on manufacturing performance’, Journal of Operations Management, 19, pp. 39–58.
Monden, Y. (2011) Toyota production system, An integrated approach to Just-In-Time, Journal of Experimental Psychology: General. Taylor & Francis Group.
33
Olhager, J. (2003) ‘Strategic positioning of the order penetration point’, International Journal of Production Economics, 85, pp. 319–329.
Riezebos, J. (2010) ‘Design of POLCA material control systems’, International Journal of Production Research, 48(5), pp. 1455–1477.
Schmenner, R. W. and Swink, M. L. (1998) ‘On theory in operations management’, Journal of Operations Management, 17(1), pp. 97–113.
Shah, R. and Ward, P. T. (2007) ‘Defining and developing measures of lean production’, Journal of Operations Management, 25(4), pp. 785–805.
Shingo, S. (1985) A Revolution in Manufacturing: The SMED System. Productivity Press Cambridge. Simon, H. A. (1996) The sciences of the artificial. MIT Press.
Slack, N., Brandon-Jones, A. and Johnston, R. (2011) Essentials of Operations Management. Pearson. Slack, N., Brandon-Jones, A. and Johnston, R. (2016) Operations Management. eight edit. Pearson. Suri, R. (1998) Quick response manufacturing: a companywide approach to reducing leadtimes. Portland, OR: Productivity Press.
Tsuchiya, S. (1992) Quality Maintenance: Zero Defects Through Equipment Maintenance. Productivity Press.
34 APPENDIX
APPENDIX A: INTERVIEW PROTOCOL
INTERVIEUW PROTOCOL
Intervieuw Broekema
Sjoerd Mulder – Rijksuniversiteit Groningen – Newcastle University Introductie
Flow in een organisatie is een belangrijk onderwerp voor operations managers. Dit onderzoek richt zich op problemen die spelen met betrekking tot de flow in het productie proces. Door middel van een nieuwe methode kunnen de oorzaken van verschillende voorraden geïdentificeerd worden. Het doel van dit interview is om te kijken wat de impact is van de verschillende verbeter maatregelen aan de hand van de nieuw ontwikkelde methode.
Aan de hand van interviews willen we er achter komen wat het doel van de verbeter maatregel is en hoe die bewerkstelligd denkt te worden. Dit zullen we doen door het productieproces door te lopen met de geïnterviewde voor elke maatregel apart. De verschillende voorraden waarop de maatregel impact op zal hebben willen we zo in kaart brengen. Hierbij wordt de focus gelegd op de productie en assemblage van riemen en spijlen die gebruikt worden voor de producten van Broekema.
Zijn er tot nu toe nog vragen?
Ik heb een aantal vragen voorbereid maar voel je vooral vrij om, los van de vragen, dingen toe te voegen als je die belangrijk vindt. Is het goed als dit interview opgenomen wordt voor onderzoeksdoeleinden? (Bij een positief antwoord zal het interview vanaf hier opgenomen worden)
35
Hierbij zal er gekeken worden op welke voorraadpunten en welke dimensies (missende input, hoofdoorzaak voor de variabiliteit of de vorm van variabiliteit) de maatregel impact heeft. De vragen per blok zullen hetzelfde zijn.
Blok 1: Basic end date spreiding
Wat is het doel van deze maatregel?
Welke voorraden denken jullie hiermee te verminderen?
Hoe denken jullie dat de voorraden hierdoor verminderd worden?
Zijn er neveneffecten van deze maatregel te verwachten in andere delen van het proces?
Blok 2: Order release rules
Wat is het doel van deze maatregel?
Welke voorraden denken jullie hiermee te verminderen?
Hoe denken jullie dat de voorraden hierdoor verminderd worden?
Zijn er neveneffecten van deze maatregel te verwachten in andere delen van het proces?
Blok 3: SMED project
Wat is het doel van deze maatregel?
Welke voorraden denken jullie hiermee te verminderen?
Hoe denken jullie dat de voorraden hierdoor verminderd worden?
36 APPENDIX B: TRANSLAT ION INTERVIEW PROTOC OL
INTERVIEWPROTOCOL
Interview Broekema
Sjoerd Mulder – Rijksuniversiteit Groningen – Newcastle University Introduction
Flow is an important issue for operations managers. This research focuses on problems regarding flow in the production process. By means of a new method the root causes of different inventories can be identified. The aim of this interview is to analyze the impact of the various improvement measures with the newly developed method.
Based on interviews, we want to find out what the goal of the improvement measure is and how you think it can be achieved. This will be done by analyzing the production process step by step with the interviewee for each measure individually. The different inventory points on which the measure is expected to have an impact on will be identified. The focus of this interview will be on the production of belts and rods that are used for the conveyor belts of Broekema.
Are there any questions so far?
I have prepared a number of questions but feel free to add things, if you think it is important, apart from the questions. Do you mind if this interview is recorded for research purposes? (With a positive answer the interview will be recorded from here)
At each block the following flow chart will be discussed:
37 Blok 1: Basic end date spreiding
What is the goal of this measure?
Which inventory points do you think the measure can reduce? How do you think the measure can influence the inventory points?
Are side effects of this measure to be expected in other parts of the process?
Blok 2: Order release rules
What is the goal of this measure?
Which inventory points do you think the measure can reduce? How do you think the measure can influence the inventory points?
Are side effects of this measure to be expected in other parts of the process?
Blok 3: SMED project
What is the goal of this measure?
Which inventory points do you think the measure can reduce? How do you think the measure can influence the inventory points?
38 APPENDIX C: INVENTORY LABEL ABBREVIATIONS
Demand Safety Stock: DSS
Demand Uncertainty Induced Congestion: DUIC
Demand Uncertainty Induced Assembly Waiting Time: DUIAWT Demand Anticipation Stock: DAS
Demand Anticipation Induced Congestion: DAIC
Demand Anticipation Induced Assembly Waiting Time: DAIAWT Demand Cycle Stock: DCS
Batched Demand Induced Congestion: BDIC
Batched Demand Induced Assembly Waiting Time: BDIAWT Supply Safety Stock: SSS
Supply Uncertainty Induced Congestion: SUIC
Supply Uncertainty Induced Assembly Waiting Time: SUIAWT Supply Anticipation Stock: SAS
Supply Anticipation Induced Congestion: SAIC
Supply Anticipation Induced Assembly Waiting Time: SAIAWT Supply Cycle Stock: SCS
Batched Supply Induced Congestion: BSIC
Batched Supply Induced Assembly Waiting Time: BSIAWT Capacity Safety Stock: CSS
Capacity Uncertainty Induced Congestion: CUIC
Capacity Uncertainty Induced Assembly Waiting Time: CUIAWT Capacity Anticipation Stock: CAS
Capacity Anticipation Induced Congestion: CAIC
Capacity Anticipation Induced Assembly Waiting Time: CAIAWT Capacity Cycle Stock: CCS
Batched Capacity Induced Congestion: BCIC
