ELIMINATING PRODUCT QUANTITY AND QUALITY
LOSSES DUE TO CHANGEOVER OPERATIONS IN THE
FOOD PROCESSING INDUSTRY: A CASE STUDY
AUTHOR
M. Stapelbroek (S3094545/B70705527)
Student MSc Technology and Operations Management (Dual Award)
MASTER THESIS (DUAL AWARD)
MSc Technology & Operations Management (University of Groningen) MSc Operations & Supply Chain Management (Newcastle University)
SUPERVISORS
Dr. O.A. Kilic (First supervisor, University of Groningen) Dr. Y. Yang (First supervisor, Newcastle University)
Prof. Dr. D.P. van Donk (Academic advisor, University of Groningen) 10 December 2018
ABSTRACT
ACKNOWLEDGMENTS
10th of December 2018
I would like to thank my thesis supervisors, Dr. Onur Kilic, Dr. Ying Yang and co-advisor Prof. Dr. Dirk Pieter van Donk, as they were great sources of support during my thesis project. Their endless intellectual support and encouragement were a great relief. The regular feedback of the supervisors and co-advisor served as inspiration for the thesis topic of changeover losses in Food processing industries.
Furthermore, I would like to express my gratitude to case company supervisor Thom Dijkstra as a great source of inspiration in both academic and practical fields. He contributed to my thesis research and helped me to profile myself as well as possible to the case company which resulted in a challenging job at one of the largest food and beverages companies in the world. Besides to Thom Dijkstra, I want to thank all colleagues of the case company for their cooperation and trust in my thesis project.
I would also want to thank Marc Nieboer, my fellow thesis companion and friend, for the inspiring conversations on food processing industry challenges. Finally, I would like to greatly thank my parents, Eddie and Ria Stapelbroek, for their endless support and belief in me, all the way from VMBO to the Double degree master of Operations Management at the University of Groningen and Newcastle.
Martijn Stapelbroek
Table of contents
1. INTRODUCTION ... 1 2. THEORETICAL FRAMEWORK ... 4 2.1 Changeover process ... 4 2.2 Shut-down losses ... 5 2.3 Clean-up losses ... 6 2.4 Setup losses ... 7 2.5 Start-up losses ... 8 3. CASE STUDY ... 10 3.1 Case selection ... 10 3.2 Data collection ... 12 3.3 Data analysis ... 17 4. RESULTS ... 204.1 Changeover operations at the case company ... 20
4.2 Influence factors of changeover losses ... 22
5. DISCUSSION ... 32
6. CONCLUSION ... 36
REFERENCES ... 39
APPENDICES ... 42
Appendix A: Unit of analysis ... 42
Appendix B: Reliability and validity criteria ... 44
Appendix C: Interview and case study protocol ... 45
Appendix D: Coding scheme ... 48
Table of figures
Figure 1. Changeover process in processing industries. ... 4
Figure 2. Outline of the production process of the case company ... 11
Figure 3. Overview of coding steps ... 18
Figure 4. Production output of sub-unit A with a changeover during shift change ... 21
Figure 5. Production output of sub-unit D after high precision changeover ... 24
Figure 6. Production output of sub-unit B with poor quality changeover equipment ... 29
Figure 7. Critical themes in reducing changeover losses ... 35
Table of tables Table 1. Influential factors of changeover losses per phase ... 9
Table 2. Sub-unit characteristics ... 12
Table 3. Overview of observations ... 13
Table 4. Interview questions per influence factor and changeover loss ... 14
Table 5. Overview of interviews... 15
Table 6. Overview of historical data reports ... 16
Table 7. Compliance to SMED-lists ... 23
Table 8. Setup time exceedance with shift change ... 25
Table 9. Overview of the extent of data gathering and reporting by operators ... 27
Table 10. Cross-case analysis: Effects and significance of influence factors on changeover losses .... 30
List of abbreviations
CI
= Continuous improvement
CIP
= Clean-in-place
FPI
= Food processing industry
KPI
= Key performance indicator
SMED
= Single minute exchange of dies
Page 1 of 52
1. INTRODUCTION
Changeovers play a crucial role in food processing industries (FPI), as changing market demands is pushing manufacturers to enrich their product portfolio and demanding more flexible production systems, while at the same time must compete on costs (Gungor & Evans, 2017). Besides being very time consuming, changeovers in FPI result in significant product losses and excessive usage of water and energy (hereafter changeover losses) which can reach up to thousands of euros per changeover (Flapper, Fransoo, & Broekmeulen, 2010; Gungor & Evans, 2017). Since changeover operations can occur multiple times per week per production line, the magnitude of changeover losses is having a significant impact on the production efficiency of FPI manufacturers. Moreover, reducing changeover losses, results in a higher quality of the product and requires less rework activities (Akkerman & van Donk, 2008; Gungor & Evans, 2015). Therefore, as profit margins in the FPI are low and production is capital intensive, reducing changeover losses is an interesting avenue to increase profits and reduce the environmental footprint (Richter & Bokelmann, 2016).
Previous studies on changeovers mainly focused on reducing production time losses by achieving faster changeovers (Claassen, Gerdessen, Hendrix, & van der Vorst, 2016; Copil, Wörbelauer, Meyr, & Tempelmeier, 2017; Kopanos, Puigjaner, & Georgiadis, 2012; Van Elzakker, Zondervan, Raikar, Grossmann, & Bongers, 2012; Van Goubergen & Van Landeghem, 2002). The most widely acknowledged and applied methodology by manufacturers for achieving faster changeovers is the Single Minute Exchange of Dies (SMED) introduced by Dillon and Shingo (1985) in the Japanese Toyota production system. SMED is an improvement programme for changeovers with the objective of achieving faster changeovers by conducting changeover analysis, training and improvement techniques, and design improvement (Gungor & Evans, 2017; R. McIntosh, Culley, Mileham, & Owen, 2000). Manufacturers across industries significantly reduced their setup times with the application of the SMED-methodology. Although, this approach provided some time and cost benefits, changeover losses still have a significant burden on the economic and environmental performance of FPI manufacturers (Gungor & Evans, 2017).
Page 2 of 51
2018; Stefansdottir, Grunow, & Akkerman, 2017). Therefore, it is argued by Gungor and Evans (2017) that the influence factors of shut-down, clean-up, setup and start-up should be jointly considered by manufacturers to reduce the changeover losses. Building on the study of Gungor and Evans (2017), the term influence factors is used for this study, which can be defined as factors or underlying root causes that affect the changeover losses within the manufacturing environment. Hence, to reduce changeover losses, manufacturers must be aware of these influence factors and understand their relationship and effect on the changeover losses. As Gungor and Evans (2017) argue that current literature on the influence factors is underdeveloped, this study identifies the influence factors of changeover losses. Besides identifying the influence factors, an understanding is developed of the effects of the influence factors on changeover losses. Manufacturers that possess the capabilities to effectively manage influence factors of changeover losses can gain significant competitive advantage by producing in an economically efficient and environmentally friendly manner (Richter & Bokelmann, 2016; Stefansdottir et al., 2017).
Page 3 of 51
study provided valuable insights how the major burden of intensive cleanings can be postponed and therefore reduced. Third, the results of this study confirmed the arguments of previous studies that more emphasis must be placed on the start-up phase as this phase often causes changeover losses ten time greater than losses associated with the setup phase. It is therefore a must for manufacturers to consider the start-up phase as part of the changeover process to reduce changeover losses. Finally, this study adds two additional influence factors to current literature on changeover losses. Building on all the influence factors, this study provided an improvement framework applicable for FPI manufacturers with three critical themes to reduce changeover losses. To achieve the goals stated above, the following research question is developed:
“What are the influence factors of production quantity and quality losses in the food processing industries during changeovers and how should these factors be managed?”
Page 4 of 51
2. THEORETICAL FRAMEWORK
This section provides the theoretical background of the discussion centred around changeover losses in the FPI. The first subsection describes why changeover operations are inevitable for FPI manufacturers. Furthermore, an explanation of the changeover process and its associated phases from theory is given. After discussing the changeover process and the phases, the following four subsections discuss the influence factors described in theory per changeover phase.
2.1 Changeover process
Changeover operations are an inevitable aspect of the FPI manufacturing environment. Changeover operations can range from replacing some machine equipment to performing intensive cleanings or both at the same time (Gungor & Evans, 2017). Changeovers are inevitable as there is a continuously changing and diverse market demand which requires FPI manufacturers to enrich their product portfolio and being able to be flexible in switching within this product portfolio. To stay competitive, manufacturers require flexible production systems that are able to changeover between product batches as fast and cost efficient possible. Besides considering the complexity of the problem, current academic literature discussing changeover operations and associated phases have not been consistent in using terms and definitions, as changeovers and setups have been used interchangeably in literature. Moreover, neither any degree of precision is used to define the shut-down and start-up phase within changeover operations, as these are sometimes referred to ramp-up, run-up, and run-downs (Stefansdottir et al., 2017; Tsarouhas, 2013; Van Elzakker et al., 2012). To clarify the terminology used in this study, a graphical representation of the changeover process is given in Figure 1.
Setup Changeover Production rate (Units) Time (hours) Target Shut-down losses Setup los ses St art-up losses
Successor batch
Cl ean-up losses
Page 5 of 51
In this study, a changeover is defined as “the complete process of shut-down, clean-up, setup and starting-up from the predecessor batch to the point of meeting the target production quantity and quality rate of the successor batch” (McIntosh, Culley, Mileham, & Owen, 2000: 6). This definition is the most appropriate, as all losses associated with a changeover operation from shutting down the process to reaching target are considered rather than only focusing on the setup phase. Changeover losses can be categorised to product quantity or quality losses. Product quantity losses include the quantity of products that are not made compared to target production due to a slower or no production rate during the start-up (Gungor & Evans, 2017). Product quality losses include all non-conforming products produced during the shut-down or start-up phase of the changeover (Culley et al., 2003). To conduct a thorough research of the influence factors in the four phases of the changeover process, a literature review is performed on product quantity and quality losses in shut-downs, clean-ups, setups and start-ups within changeover operations.
2.2 Shut-down losses
Production line shut-downs are a typical production policy in the process industry to guarantee high quality of the final products and comply with hygienic standards (Kopanos et al., 2012). In general, shut-downs can be divided into two categories, scheduled and unscheduled shut-downs. First, scheduled shut-downs are planned activities on process equipment due to meetings, breaks, weekends or maintenance (Tsarouhas, 2007; van Wezel, 2001). Second, unscheduled shut-downs may occur due to defects and failures (Muchiri & Pintelon, 2008). However, due to the focus on changeover losses, this research is limited to scheduled shut-down losses. During a scheduled shut-down, the last quantities of the production batch are transformed through the production system before the setup process for the new product takes place (Mileham et al., 2004). The shut-down phase entails slower production and eventually stoppage, which can result in production losses and quantities of products that do not meet the quality specifications (Raak, Symmank, Zahn, Aschemann-Witzel, & Rohm, 2017).
Page 6 of 51
factor found in literature is the learning process of operators, as it is argued by Van Goubergen and Van Landeghem (2002) that the learning process of operators during changeover operations plays a key role in reducing losses during shut-downs and subsequent changeover phases over time. The fifth influence factor is product recovery. During the shut-down there can be a significant amount of material left in the pipes and tanks (Fryer & Asteriadou, 2009; Gungor & Evans, 2017). These leftovers may still be saleable in which case it should be recovered or considered as waste. Although, in both cases the bulk of these product leftovers should be removed to perform an efficient cleaning, as Goode et al. (2013) argues that product recovery has a significant effect on the overall cleaning resources and time needed. Finally, another important aspect is capturing and managing the acquired knowledge by operators during changeover operations. Therefore, the sixth crucial influence factor is data gathering and reporting from operators to the management and vice versa, as the operators are the key source of information on implications during changeover operations (Fan, Phinney, & Heldman, 2015; Goode et al., 2013). Adequate data reporting and sharing among works and managers can contribute to better problem solving and lower shut-down losses (Gungor & Evans, 2017).
2.3 Clean-up losses
Clean-up losses also contributes for a large part to the changeover losses within the process industry (Fan et al., 2018). Due to the use of agricultural raw materials which are mainly perishable, the production lines have to be cleaned regularly to avoid micro-organisms from multiplying and affecting the quality of the end-product (Fan et al., 2015; Flapper et al., 2010). Besides perishability and yield of agricultural raw materials, processing industries are characterised by a large variety of products recipes. Hence, to assure the standard quality of each product, cleanings are required to prevent mixing raw material leftovers from the preceding with the successor batch (Kılıç, 2011).
Page 7 of 51
et al., 2018). Finally, guidance and procedures have found to be critical in cleaning operations, as it is argued that operators regularly overuse the cleaning system (Goode et al., 2013). Therefore, not only the (over)design of cleaning systems, but also the operator skills and capabilities for controlling the system contribute to clean-up losses.
2.4 Setup losses
After the production shut-down and sometimes during the clean-up, the actual setup takes place. This involves removing old tooling and process equipment and replacing it with the required equipment for the next production batch (Allahverdi & Soroush, 2008). According to the SMED methodology, both setups and clean-ups can be defined as internal time, as these activities occur when production is halted (Claassen et al., 2016; Dillon & Shingo, 1985).
Page 8 of 51
2.5 Start-up losses
The start-up is the final phase of a changeover procedure and is sometimes referred to the ramp-up phase. The start-ramp-up process has received quite some research attention in the automotive and microelectronic industry (Matta et al., 2008; Surbier et al., 2014). It is argued that the start-up is the most costly phase of a changeover, as product losses during start-ups can be up to ten times greater than that associated with the set-up itself (McIntosh et al., 2000; Mileham et al., 2004). During the start-up, machines experience continuous fine-tuning which result in turning off and on processors. Therefore, while a start-up can be defined as the time to reach the standard quality and output rate, it is often unclear when a start-up has ended, as machines are turned on and off within this changeover phase (Stefansdottir et al., 2017).
Page 9 of 51
To conclude this section, an overview of the influence factors per changeover phase is illustrated in Table 1. In this table, the discussed influence factors are in the rows and the shut-down, clean-up, setup and start-up processes represent the columns. Several influence factors of changeover losses have been identified in literature, however there is only one study to date that jointly considers the effects of the influence factors on the different phases in reducing changeover losses. Consequently, due to this lack of research, it remains unclear for FPI manufacturers how to manage the joint effects of the influence factors across changeover phases effectively (Gungor & Evans, 2017).
Table 1. Influential factors of changeover losses per phase Changeover loss / Influence
factor Shut-down losses Clean-up losses Setup losses Start-up losses Publication
Quality of preparation x x x x (Gungor & Evans, 2015; Utne et al., 2012)
Operator capabilities x x x x (Gungor & Evans, 2017; R McIntosh et al., 2007; Mileham et al., 2004)
Product recovery process x x (Fan et al., 2018; Fryer & Asteriadou, 2009; Gungor & Evans, 2017)
Quality of changeover equipment x x (Gungor & Evans, 2015; Taylor, Flapper, Fransoo, & Broekmeulen, 2010)
Learning process x x x x (Gungor & Evans, 2015, 2017; Matta et al., 2008)
Data gathering and reporting x x x x (Bulent Dal, Phil Tugwell, 2014; Gungor & Evans, 2015, 2017)
Out-of-date SOPs x x x x (Culley et al., 2003; Goode et al., 2013; Gungor & Evans, 2017)
Precision of changeover activities x x x (Braglia, Frosolini, & Gallo, 2017; McIntosh et al., 2007)
Overdesign clean-in-place x (Fan et al., 2015; Fryer & Asteriadou, 2009)
Cleaning parameters x x (Compton et al., 2018; Fan et al., 2015, 2018)
Page 10 of 51
3. CASE STUDY
The aim is to explore the influence factors and their effects on changeover losses and develop a framework that supports FPI manufacturers to reduce changeover losses by effectively manage these influence factors. A case study approach seems appropriate as influence factors of changeover losses are still in its exploratory phase. Hence, the inductive approach of a case study approach is preferred over other research approaches. Moreover, Yin (2009) argues that a case study lends particularly well in studies which aim to answer a ‘How’ research question. Voss, Tsikriktsis, & Frohlich (2002) stress that research with an exploratory nature requires the researcher to be able to study the phenomenon in great depth in its natural setting through observations of the actual changeover practices. More specifically, there is chosen for an embedded case study in which data is acquired by comparing four production lines within the production plant, because of three reasons. First, an embedded case study enables the researcher to explore the influence factors and the phenomenon changeover losses in great depth. Moreover, the researcher is able to explore and compare the significance of the phenomenon across multiple sub-units. This results in a broader exploration of the research question, while at the same time a more robust, generalizable and testable theory is developed (Eisenhardt and Graebner, 2007). Second, the ability to understand influence factors and changeover losses in FPI is limited, as most research so far focused on reducing changeover times by achieving faster changeover (Gungor & Evans, 2017). Therefore, an embedded case study provides the opportunity to explore the influence factors and changeover losses in greater depth (Eisenhardt, 1989). Third, an embedded case study lends perfectly for observations of the increase or decrease of changeover losses in multiple sub-units, rather than collect this data from one unit or second-source data via managers perceptions (Meredith & Vineyard, 1993).
3.1 Case selection
Page 11 of 51
certain influence factors of changeover losses exist. Furthermore, exploring and understanding the influence factors of changeover losses can help other manufacturers in managing and reducing changeover losses.
Depalletiser Filler Capper Labeller Tray packer Kitchen Bottles Label materials Bottle caps Palletiser End product
Figure 2. Outline of the production process of the case company
Third, because the production plant of the case company encountered the changeover losses for some time, historical production and maintenance reports are available which can quantify the losses. Moreover, priori possible counter-measures have already been identified by the case company. Therefore, more emphasis can be placed on the development of the framework for managing the negative impacts. The production starts with the workstation that processes pallets packed with empty bottles and the line finishes with full pallets of bottles filled with the end product. A more detailed description and process flow of the production line can be found in appendix A.
Page 12 of 51
needs to be adjusted and replaced to produce the successor batch. Likewise, for the product recipe, if the successor batch recipe had a completely different composition of ingredients, an intensive cleaning was required. As shown in Table 2, Sub-unit C and D entail a high complex changeover. Sub-unit C is high complex, as the packaging of the predecessor and successor batch have a large deviation. Sub-unit D is also high complex with a long changeover duration. Besides changing and adjusting machine equipment, an intensive cleaning was required to ensure that the successor batch is conform to quality standards. Sub-unit A and B are less complex, as these changeovers do not include an intensive cleaning program nor have very different packaging dimensions. However, sub-unit B was classified as medium, because the workstations at this production line are rather old and therefore equipment is hard to replace and adjust. The experience of the operators was assessed with the managers by using the personnel planning and take an average of the experience in the particular shift team.
Table 2. Sub-unit characteristics
3.2 Data collection
According to Hyer et al. (1999), both the reliability and validity of the case study findings are improved by using multiple measures drawn from different data sources. Data triangulation is ensured by verification of the collected data through cross verification of the multiple sources (Ketokivi & Choi, 2014). As the case company provides the opportunity to collect both qualitative and quantitative data, multiple data sources are used. Data sources for this research include observations, semi-structured interviews and historical operational data and documentation (production performance reports, failure data, and real-time machine performance). The data collection is set up in three stages. There is chosen to collect data in three stages, because data collected in earlier phases can serve as relevant input in the next data collection phase. After each data collection stage, a meeting was arranged with plant manager, Continuous Improvement (CI) manager and production manager to ensure internal validity in each of the data collection stages. An overview of all the validity and reliability criteria is shown in appendix B.
Stage 1: Observations
An observation entails that the researcher operates as a participant observer during a changeover process at one of the four production lines at the production site of the case company. Operators, engineers and shift leaders are one of the richest sources of information on influence factors and changeover losses, as these employees directly experience these phenomena. Therefore, as a participant
Characteristic Sub-unit A Sub-unit B Sub-unit C Sub-unit D
Changeover complexity Low Medium High High
Changeover duration (min.) 60 90 120 300
Page 13 of 51
of the actual changeover practice, the researcher had the opportunity to respond and ask questions about events that occurred during the changeover operation. In total eleven observations are conducted by the researcher to compare the acquired knowledge and understanding of the changeover process in FPI, influence factors and changeover losses from theory with practice (Akkermans & Vos, 2003). An overview of the observations per sub-unit is shown in Table 3.
Table 3. Overview of observations
Observation Sub-unit Duration (min.) Date
1 A 90 09/07/2018 2 A 120 13/07/2018 3 A 60 24/07/2018 4 B 120 06/07/2018 5 B 140 17/07/2018 6 B 120 27/07/2018 7 C 210 11/07/2018 8 C 160 19/07/2018 9 C 120 04/08/2018 10 D 320 30/07/2018 11 D 300 03/08/2018
From the observations it became clear that the changeover operations discussed in literature is not in line with the changeover practice at the case company. Moreover, additional influence factors of changeover losses have been found in practice that are not mentioned in literature yet. These findings are confirmed during the intermediate meeting with the experienced managers. Hence, the key challenge raised from the first data collection stage was to align the interview protocol with the industry practice of the production site of the case company. The interview protocol must align with industry practice to collect as much data, while at the same time keeping the protocol generalizable for theory. To achieve this, findings from observations were included in the design of the interview protocol.
Stage 2: Semi-structured interviews
Page 14 of 51
Numbers indicate which interview question belong to what factor and phase
participant for the more complex and critical questions. The critical questions on the influence factors and changeover losses were arranged from least difficult to complex and detailed questions. This was achieved by starting the interview with a question about changeover using a visual graph of the production speed which is familiar to all the interviewees. By building up complexity of the interview questions, confidence was created during the interview questions to ensure that the interviewee felt comfortable in sharing his knowledge, opinion and experience. The interview questions were of an expansive nature to allow the participant to take the question into several directions. The final step in the development of the interview protocol was a pilot test with the plant manager. During this pilot test, the clarity of the preface, interview structure and questions were assessed. From this pilot test, two additional sentences have been added to the preface and several questions have been restructured. Although, the interview questions are written in English, the actual questions are asked in Dutch to eliminate misinterpretation during the interview due to a possible lack of linguistic skills. Table 4 provides an overview of the interview questions per influence factor and changeover loss. As shown, shut-down losses and its influence factors are neglected in the interview protocol. Based on findings from the first data collection stage it was decided not to include this phase and its influence factors. Detailed reasoning for the final list of influence factors and changeover losses can be found in the results.
Table 4. Interview questions per influence factor and changeover loss Changeover impact /
Influence factor Setup losses Clean-up losses Start-up losses
Out-of-date SMED-lists Q8, Q11 Q8, Q11
Precision of changeover
activities Q12, Q15 Q12, Q15 Q12, Q15
Changeover during shift
change Q13 Q13 Q13 Learning process Q10, Q12, Q22, Q23 Q10, Q12, Q22, Q23 Q10, Q12, Q22, Q23 Operator capabilities Q6, Q7, Q8, Q20 Q6, Q7, Q8, Q20 Q6, Q7, Q8, Q20
Data gathering and reporting Q2, Q3, Q21 Q2, Q3, Q21
Shift team composition Q4, Q5, Q6, Q17 Q4, Q5, Q6, Q17 Q4, Q5, Q6, Q17 Quality of changeover
equipment Q9, Q18 Q9, Q18
Surface contamination Q16
face-to-Page 15 of 51
face at the case company in The Netherlands. The interviews are conducted by one researcher. The researcher was leading the conversation with the informant and asking the questions, while taking notes and observing the quality of the interview (Eisenhardt, 1989). During the interviews, the managers, operators and shift leaders were asked to provide a detailed view on how they experienced the changeover losses during the clean-up, setup and start-up process of the specific sub-units (Yin, 2009). The unique opportunity to collect data across varying hierarchical levels of the organisation resulted in having a rich qualitative dataset to build a more robust, generalizable and testable theory (Eisenhardt and Graebner, 2007). An overview of the interviewees is given in Table 5. To ensure reliability of the data collected, questions were sent to the informants beforehand to prevent indistinctness at the beginning of the interview. Moreover, After each interview was transcribed, the transcribed document was sent back to the informants for verification and possible revisions (Yin, 2009). To ensure internal validity, the interviews are recorded by using an audio recording application and were all transcribed within the next 24 hours after the interview took place (Emans, 2002). Any information recorded that might harm the privacy of the informants or the case company was excluded from the data, because complete anonymity and confidentiality was a condition of the informants in participating in this research. Each interview concluded in an informal discussion with the informants which contributed to the context for the analysis.
Table 5. Overview of interviews
Tactical level Experience (years)
Plant manager 4
CI manager 31
Production manager 3
Maintenance manager 31
Sub-unit Operational level Experience (years)
Page 16 of 51
Stage 3: Historical data
Besides collecting qualitative data through interviews, the case company released historical data sets for analysis, see Table 6. This historical data includes production reports and maintenance reports in Microsoft Excel which contains rich information on failures, breakdowns and downtime with associated reasoning. Especially, events before, during and after changeovers contained valuable information to quantify the qualitative results. The provided historical quantitative data of failures, idle time, line restraints and unscheduled downtimes is used to increase the validity and reliability of the subjective evidence collected from the interviews on the influence factors of changeover losses (Ketokivi & Choi, 2014). Data triangulation is ensured by contrasting the quantitative data with the qualitative data in writing the results.
Production reports. The production reports of the four filling and packaging lines are made available by the case company. A production report is an Excel-file generated via the ERP-system and contains data of the last twelve production months which includes around 300 in-production days of 24 hours. The production reports of the production lines hold on average 68,000 rows where each row contains unique data of a specific event such as failures, idle, line restraints and unscheduled downtime. The unique events in the production reports are automatically created by the filling machine per production line. After an event is created the operator can add a detailed explanation of the event.
Maintenance report. Besides quantitative production reports, the engineering department made available all the maintenance jobs performed last year to the four packaging and filling lines. All the maintenance jobs conducted at the four packaging and filling lines are documented in text in an Excel-file. The maintenance report of the last year comprises in total 2354 jobs where again each row holds a unique maintenance event performed at one of the four packaging and filling lines. Each maintenance job is initiated by an operator or engineer, which can elaborate on the issue and performed action. All the maintenance jobs documented and performed in the last 24 hours are communicated with shift leaders, product manager and Quality managers during daily production meetings.
Table 6. Overview of historical data reports
Sub-unit A Sub-unit B Sub-unit C Sub-unit D
Time period 07/2017 - 07/2018
Production days 300 288 282 230
Page 17 of 51
3.3 Data analysis
To achieve the aim to support FPI manufacturers in improving their product quality and quantity during changeover losses, the case data is analysed by conducting a within-case and cross-case analysis. The purpose of the within-case is to gain familiarity and understanding of the influence factors in each sub-unit. After the within-case was conducted, a cross-case analysis was performed to enhance the generalisability. This was achieved by forcing the researcher to look beyond the influence factors per sub-unit and juxtaposing the findings per individual sub-unit with the objective to identify whether the influence factors appeared in the other sub-units.
Within-case analysis
The qualitative data was analysed by following three coding steps suggested by Miles and Huberman (1994): data reduction, data display and conclusion. First-order coding refers to the process of “selecting, focusing, simplifying, abstracting, and transforming the data that appear in written up field notes or transcriptions” (Miles and Huberman, 1994: 10). After data reduction, second-order coding provides higher order categories that emerge from the data discovered during the process of data reduction. Finally, the third-order coding identified patterns by juxtaposing the influence factors from literature and the observations (Barratt, Choi, and Li 2011). Throughout this study, the tools Atlas.ti and Microsoft Excel supported the data analysis process.
First order coding. All data is reduced to quotes sentences and paragraphs that are relevant for this research. The data collected from the interviews are reduced to the quotes that hold the richest and most relevant information. To provide an example of how quotes are coded and reduced; when an informant said: “Standardised moulds in the technical department do not fit the equipment of the
production line” this quote was reduced to “Misfit of standardised moulds with production line”.
Second order coding. The first-order codes are examined and coded into descriptive code categories by applying a priori coding. This implies that the categories for second-order coding are established prior to the analysis and based upon the influence factors derived from literature and observations, which are presented in a coding scheme (appendix D). First, priori coding was applied by assigning the influence factors from literature to the reduced data resulted from the first-order coding. If a quote for example included: “lack of skills” or “inexperienced operator”, the sentence was assigned to the second-order code operator capabilities. Quotes that refer to “training”, “learning”, or “education”, are assigned to the influence factor learning process. Besides the influence factors from literature, influence factors found during observations have been applied for the second-order coding process. When a quote mentioned “incomplete SOPs”, “incorrect SMED” or “faulty parameter standard”, the quote was assigned to out-of-date SMED-list. In the cases that quotes contained “coordination”, “fixed teams” or “freelance operator”, the sentence was assigned to shift team
Page 18 of 51
Third-order coding. To execute the third-order coding, the influence factors from the second-order coding are further linked to the shut-down, clean-up, setup and start-up losses. When a code showed “lack of changeover capabilities due to in experienced crew”, it was assigned to knowledge. In the case codes included “training”, “SOPs”, “changeover during shift-change”, it was assigned to
method. By using third-order codes, patterns are sought out to investigate connections between the
influence factors in reducing productivity losses in the four phases of the changeover operation. When a causal relationship between influence factors was found, it was interpreted that it influenced the productivity losses during the changeover phases. For example, when a code mentioned “less time pressure on operators during changeover operations improves the quality of the learning process between operators”, it was recognized as Knowledge management. Figure 3 demonstrates how the researcher progressed from data reduction (first-order codes), to descriptive codes (second-order categories), and eventually to third-order themes.
First-order coding Data reduction Data extracted Data reduced Data filtering Second-order coding
Categories (Priori coding)
Coding scheme (appendix X) Influence factors from literature
Influence factors from practice
Third-order coding
Critical themes
Pattern matching productivity losses across changeover phases
Input variables for the coding process
Figure 3. Overview of coding steps
Cross-case analysis
Page 19 of 51
Page 20 of 51
4. RESULTS
In this section the findings are presented. The first subsection is devoted to results collected through observations during changeover operations across the four sub-units at the production site of the case company. In the second subsection it is described what influence factors are found present in each of the sub-units for causing changeover losses. After the influence factors in the sub-units are described independently, patterns and the significance of each influence factor on changeover losses are described in a cross-case analysis.
4.1 Changeover operations at the case company
The first stage of data collection included observations where the researcher functioned as a participant observer during multiple changeover operations at the four sub-units. The aim of the observations was to get familiar with the changeover process and its phases in practice. The eleven observations within the four sub-units revealed crucial information that had consequences on the second and third data collection stage of the influence factors and their effect on changeover losses. The findings described below are considered in the development of the interview protocol and historical data analysis.
Shut-down process. It is found at all four sub-units of the case company that the changeover operation does not include a shut-down phase. The case company is able to stop the production process without having a shut-down phase, as the product flow through the piping can be halted immediately when required. As the shut-down phase of the changeover operation is not present, the researcher was not able to observe nor question the operators about the influencing factors of shut-down losses. Thus, the shut-down phase and the influence factors product recovery and quality of preparation from literature are not considered in the data analysis of the semi-structured interviews and historical data.
Clean-up process. While literature suggests that the clean-up often plays an important role in the changeover operation, observations and questioning of operators shown that this is in most instances not the case at the four units of the case company. Only in one of the eleven observations at sub-unit D, the participant experienced an intensive CIP program during the changeover operation. Through multiple discussions with the production manager and plant manager it has been verified that CIP programs only occur at sub-unit D. It only occurs at this sub-unit, as this is a sauce filling and packaging line with a diverse range of mild to spicy recipes. Despite this wide variety in recipes, the case company is able to schedule the product recipes in such an effective manner that CIP programs can be largely postponed towards a non-productive day in the weekend.
Page 21 of 51
Figure 4. Production output of sub-unit A with a changeover during shift change equipment was cleaned externally from the production line. Therefore, the operator was able to immediately start replacing and adjustment machine settings after stopping the production process instead of performing cleaning activities first. On the other hand, in all sub-units, the operator had to foam and disinfect machine equipment at all places where the raw product is in open areas after the setup phase. The influence factors cleaning parameters, product recovery and overdesign CIP are not considered in further analysis, as these are not found to be present at the case company due to efficient planning of product recipes.
Setup process. It was observed that the duration of a changeover operation can range from an hour to four hours or even longer. As operators work in shifts of eight hours it appeared in four of the eleven observations that the changeover included a shift change. In all four observations a significant amount of product quantity- and quality losses was found due to extended setup and start-up times and incorrect machine adjustments, see Figure 4 for the effect of this influence factor on the changeover losses in sub-unit A. Although the influence factor of changeovers with a shift change is not discussed in literature, it is found to play a significant role in causing changeover losses. Therefore, the influence factor changeover during shift change is included in further stages of data collection and analysis.
Start-up process. The largest amount of changeover losses is observed in the start-up process. The start-up process is a realistic reflection of the precision carried out to all activities in the previous changeover phases. Moreover, it is found that particularly in this phase, a well-skilled operator is distinguished from an inexperienced and less-skilled operator. A far lower amount of changeover losses was observed in the sub-units with a majority of skilled and experienced operators during the start-up. The lower amount of changeover losses was explained by the production manager by the fact that these operators functioned as a team, knew each other’s responsibilities and helped each other whenever possible. Therefore, shift team composition is a crucial influence factor of changeover losses. While
P rod u ct io n o u tp ut (uni ts ) Unscheduled downtime
Planned idle time: Changeover Line restraint
Page 22 of 51
theory discusses influence factors as operator capabilities and learning process, the significant importance of shift team composition is not well covered in literature yet. This influence factor and its impact on changeover losses is discussed in the next sections, as it plays a crucial role in reducing changeover losses.
4.2 Influence factors of changeover losses
This subsection describes the influence factors found in each sub-unit separately through the analysis of semi-structured interviews and historical data reports. A coding tree was developed to reduce the rich amount of qualitative gathered from the semi-structured interviews. The coding tree contributed in effectively reducing the data from transcripts to map relationships between the influence factors and changeover losses. The third order coding process and juxtaposing helped to acquire final themes for the influence factors. The coding tree in appendix E provides a clear overview of the mapping process from extensive qualitative data to final critical themes.
Out-of-date SMED-lists. Changeovers are a human centric operation and therefore offering and securing correct SOPs is vital to achieve efficient changeover operations. Although, it was found that the case company provides SOPs for setup and clean-up operations of the machines, it was found that the majority of the shop floor personnel does not use these standards. This finding was confirmed, because all interviewees argued that an operator is not able to perform a setup nor a clean-up by following the provided SMED-list. The operator of sub-unit B stated:
“During the changeover, I tried to setup the machine by following the SMED-list and I came to the shocking conclusion that 14 of the 28 parameter values on this list were incorrect and two parameters were even not mentioned”
Page 23 of 51 Table 7. Compliance to SMED-lists
Sub-unit Number of parameters
Number of parameters not according to SMED
Number of parameters Missing Incorrect standards (%) A 13 3 1 31% B 28 14 2 57% C 52 19 3 42% D 38 12 2 14%
Average across sub-units 42%
Precision of changeover activities. Building on the argument of non-compliant operators towards SOPs, the precision of the performed activities during a setup, clean-up and start-up plays also a dominant role in reducing changeover losses. It was found that the chaotic and incomplete actions of operators of sub-unit A and B during the setup and clean-up resulted in major consequences during the start-up period. The operator was required to halt the production process various times, as the incorrect setup of machines resulted in surface contamination and non-conforming products. The surface contamination appeared from the filling machine and labelling machine along the conveyer belts and -walls. The operator was forced to clean the conveyer equipment and perform fine-tuning adjustments, before running production again. There are several reasons that can justify an operator’s lack of attention to precision during the changeover phases. First, it is found that the learning process to perform a changeover properly is insufficient, as operators try to imitate actions of its mentor without possessing the underlying reasoning. Hence, the operator lacks the complete underlying understanding of why certain parameters or equipment must be changed or replaced. The CI manager stated:
“Operators are taught to imitate actions of their mentor during the introduction training of performing a changeover. There is totally no consciousness or knowledge what actually happens to the machine by changing certain parameters.”
This statement was confirmed during the semi-structured interviews with shift leaders of sub-unit A and D. Secondly, inexperienced operators do not possess the required skills and capabilities to perform an appropriate changeover without using SOPs. Despite the provided SOPs by the case company, an operator still need to make their own decisions and choices which in most cases directly contributes to changeover losses.
Page 24 of 51
correctly adjusting and replacing machine equipment. The high degree of precision paid to the changeover activities paid off, because only a couple minor unscheduled downtimes occurred at the production line. The greatest downtime was a line restraint due to insufficient flow of raw materials which is not in the hand of the filling and packaging line itself. Besides following SOPs and time pressure, the shift leader of sub-unit D noted the importance of intrinsic motivation:
“Intrinsic motivation is essential in reducing product losses due to imprecise actions of operators. Knowledge is power, but their enthusiasm triggers other operators to become pro-active and work according the standards as well.”
Changeover during shift change. All interviewees agreed that changeovers during a shift change result in a significant amount of production losses. It was noted by operators A, B, and D, shift leader B and C and all managers that every operator has its own changeover working routine. This uniformity of working procedures has a negative impact on the transfer of current setup operation from one shift to another. Since there is not a uniform structure of setup activities, it is unclear for the next shift where to proceed setup activities. It takes a lot of time for the operator to find out where the previous shift halted the setup activities and whether the previous setup activities have been performed properly. Particularly in sub-unit A, it was noted that the transfer of the changeover operation was not present. As a result, the setup phase of the changeover took already 40 minutes longer than the planned 60 minutes, as the operator had to check whether all parameters and changeover equipment was set according to standards. Although, the operator was aware that the previous shift executed a large part of the setup and clean-up, the operator insisted to check and in three instances even changed parameters and equipment set by the previous operator. According to the maintenance manager there are mainly three reasons for this: no confidence in the skills of the previous operator, ethical reasons, and perhaps the most important one is that a standardized changeover sequence is missing. It is argued by operators, shift leaders and managers that the lack of a uniform changeover structure that is followed by rigor can
Unscheduled downtime Planned idle time: Changeover Line restraint
Production target Actual production rate
Runtime: Production P rod u ct io n o u tp ut (uni ts )
Page 25 of 51
extend setup durations up to 25 percent. Moreover, despite the longer setup period it remains unsure whether the setup has been carried out with precision. This influence factor is quantified by using last year’s quantitative production reports of the four filling and packaging lines. The results of Table 8 are calculated by comparing the exceedance of setup durations of a setup that takes place during a shift change and a setup without a shift change. As shown in this table, setups that occurred during a shift change at the filling and packaging lines took on average 30 percent longer. Hence, in the case of a setup of 120 minutes, the actual setup time was 156 minutes.
Table 8. Setup time exceedance with shift change Sub-unit Setup time exceedance with
shift change
Setup time exceedance without
shift change Difference
A 25% -2% 27%
B 36% 4% 32%
C 27% -5% 32%
D 30% -1% 30%
Average exceedance 30%
Learning process. Our interviewees confirmed that the influence factor learning process is crucial in reducing changeover losses. It is argued by the interviewees that the current training method of operators is inappropriate to sufficiently learn operators how to perform a setup, clean-up and start-up. Hence, this method lacks structure, uniformity, quality and a feedback loop. The structure of the training is inappropriate, as the training is interpret as placing a new employee next to a more experienced operator. Moreover, it was argued by the shift leaders of sub-unit B and C that often mentors assigned to an operator are not experienced enough. As mentioned by the CI manager:
“The extent of operator learning cannot exceed the current knowledge gained by the mentor that is providing the training. Hence, a mentor who has only accumulated 70 percent of his training, can never transfer more than this 70 percent to the inexperienced operator.”
Besides the structure, the uniformity in the training process is found important. Currently, training to new operators is given by multiple operators which all have their own way of working. Thus, the new operators will learn a diverse set of working methods from different mentors. This results in a confusing training program for the operator and eventually in poor performance at the time the operator starts working on its own. The operator of case A stated:
“Nothing is more annoying than learning from five different instructors. Every instructor has his own method and preferences, so therefore the operator in training will never learn a basic method.”
Page 26 of 51
perform a changeover operation properly. Because of not having an assessment model, there a feedback loop within the training program is missing to improve the training. The ill organised training method within the case company result in insufficient knowledge transition from the mentor to the operator in training. Furthermore, it was found through interviews with operators and shift leaders of sub-unit A and C that the high number of inexperienced operators resulted in distraction of the experienced operators. The shift leader of sub-unit A stated:
“Major product losses were caused by distraction, as an experienced operator had to guide two inexperienced operators during the setup, while at the same time performing a high complex changeover at its own workstation.”
Operator capabilities. In relation to the learning process, capabilities of an operator are overly discussed in all interviews with operators, shift leaders and managers. Important capabilities of operators in reducing changeover losses are the experience and technical knowledge within changeover operations. Moreover, an operator must possess the ability to find and recommend improvement measures along the way of preforming a changeover. However, it is not only the capabilities of an operator that plays a crucial role in reducing changeover losses. It was argued by shift leaders of sub-unit A and C that operators have sufficient skills but sometimes lack the willingness to report measures for improvement. This observation is supported by the maintenance manager:
“Operators need to ask themselves why settings are changed and what caused this deviation from standard. However, operators lack awareness or willingness to use their reasoning to find root-causes.”
Likewise, the capabilities of operators, it is confirmed by all interviewees that the number of experienced technicians is insufficient for the diversity and complexity of the production lines. The lack of trained and skilled technicians can result in major changeover losses, as it influences the quality of preventive maintenance of changeover equipment and the duration of downtimes. Although, the diversity among the production lines asks for specialised personnel, technicians are not allocated to certain production areas. Consequently, the technicians often have a broad general knowledge, but lack specialism to quickly identify and solve problems during the setup, clean-up or start-up.
Page 27 of 51
“If no detailed data is collected and reported besides the automatic reporting system, none of the specific issues will be solved. The technical department cannot solve what they are not aware of. As operators, we have built in a kind of laziness that we depend too much on the system instead of analysing problems ourselves. Consequently, there is a lack of bottom-up information gathering and sharing to shift leaders, technical department and managers.”
It is found that a probable cause of the insufficient data gathering, and reporting is the lack of feedback provided towards the shop floor in case of issue reporting. The production manager argued that the current lack of feedback provided to operators on problem solving results in demotivated operators. The operator does not feel the need to write detailed emails or reports on issues faced during a changeover. In contrast, the operator of sub-unit A and B noted that operators are not pro-active enough in reporting issues, as they perceived the technical department and management as unresponsive to issues. An analysis on the historical production reports was performed to quantify the significance of the insufficient data gathering and reporting at the case company. This analysis was performed by comparing the total amount of breakdown events with the unclassified breakdown events. It is found that in the case of a breakdown at one of the four sub-units, on average sixty-one percent of the time no additional information is gathered nor reported by the operators, see Table 9.
Table 9. Overview of the extent of data gathering and reporting by operators
Sub-unit A Sub-unit B Sub-unit C Sub-unit D
Total events 77120 49877 89133 55880
Breakdown events 10895 7980 14261 8941
Unclassified breakdown events 7437 3831 10411 5096
Classified breakdown events 3458 4150 3851 3845
Insufficient reporting (%) 68% 48% 73% 57%
Besides issue gathering and reporting, there is found a lack in communication among departments during changeover operations. It was noted by the shift leader in sub-unit C that there is often miscommunication between the production and logistic departments during the start-up phase. It was found that the logistic department was not ready for the new production run, because the shift leader was not able to reach the shift leader of the logistic department. The shift leader had to walk himself towards the department and found that the logistic department was not ready yet to stack the trays with products in the right sequence. Hence, the production line was halted for 10 minutes as the logistic department had to setup the machine to the new batch.
Page 28 of 51
of the four sub-units there was insufficient supervision of a shift leader due to poor personnel planning. Consequently, experienced operators were forced to coach inexperienced operators besides their own tasks and responsibilities during the start-up. The shift leader of sub-unit C noted:
“Every time that there is one experienced operator with two temporary employees at the production line, we experience continuous issues during the changeover operation.”
Moreover, not every operator is able to supervise inexperienced operators when there is no shift leader. Especially in teams with multiple temporary or unskilled operators, a lack of supervision results in excessive product losses. It was found in sub-unit D that two operators went for a break, because there was a malfunction during the start-up of the production line, while it is expected from the operators to assist the technicians in analysing and solving the problem. The lack of assistance of the operator resulted in longer downtimes. Besides the supervision it is found that a fixed shift team composition enhanced the synergy and morale within the team. Shift leaders argued that fixed teams resulted in more rotation of operators among workstations. This enables to train people a broader set of skills and creates understanding and awareness of each other’s difficulties during the setup of the workstations. Furthermore, it is argued that fixed teams contribute in evaluating team performance more effectively. Evaluation on pre-arranged KPIs per team is only possible if there are fixed teams, otherwise continuous evaluation and improvement are not realistic. As mentioned by the CI manager and plant manager:
“Compare it to the pit stop crew of the Formula 1. It is a fixed team of people who continuously train on improving and accelerating the replacement of formula car parts. This also applies to a changeover of a production line, the team must be completely one, otherwise performance will not improve.”
Quality of changeover equipment. It is found that the quality of the changeover equipment result in significant product losses during changeovers. Although, the technical department uses standardised moulds for maintaining the changeover equipment, it was experienced in sub-units B and C that changeover equipment was incorrectly maintained by the technical department. Figure 6 shows the effect of the influence factor on the changeover losses at sub-unit B. It was argued by the maintenance manager and operators and shift leaders of sub-unit B, C and D that the standardised moulds at the maintenance department often do not match the standards of the machines at the production lines. In line with the previous arguments, the shift leader of sub-unit B noted:
Page 29 of 51
Furthermore, the quality of changeover equipment is affected by the wear of machine chains and gears. The wear of the machine equipment results in continuously changing parameters, so that SMED-lists are no longer correct. The crux is that there are currently no calibration points on the machine equipment which make it impossible to set the machines to standard after maintenance. Therefore, original parameter values of this equipment before maintenance are not correct anymore. This results in out-of-date SMED-lists, as these are often not updated after maintenance activities.
Surface contamination. Although contamination of production in literature is mainly discussed on pipelines and tanks, this influence factor has been experienced differently in practice. In sub-unit C it was found that contamination of the conveyer belts caused excessive product quantity losses during changeover operations. The decision of the case company to only clean the filling machine during a changeover results regularly in contamination issues along the conveyer belts of the production line. Moreover, it was found that operators are often not aware of the contamination and therefore start changing parameters instead of cleaning the specific place in the case of production issues. This was mentioned by all the managers and shift leaders. Shift leader C noted:
“An operator is in direct contact with surface contamination during a changeover. Therefore, the operator must choose to clean it or keep the contaminated spot in mind in the event an error occurs, so that the operator is not immediately going to change parameter values of the machine.”
In contrast, surface contamination was found less significant at sub-unit D. This filling and packaging line uses self-adhesive labelling materials and therefore eliminates the opportunity of surface contamination from adhesives. The only drawback of the self-adhesive labelling system was the large amount of dust that is released which can block the sensors for activating the labelling machine.
P rod u ct io n o u tp ut (uni ts ) Unscheduled downtime Planned idle time: Changeover Line restraint
Production target Actual production rate Runtime: Production
Page 30 of 51
To conclude, the significance of each influence factor was determined across all the four sub-units, as illustrated in Table 10. The significance was determined by juxtaposing the findings per individual case to identify patterns of influence factors across the cases. The extent to which patterns of influence factors emerged in the multiple sub-units determined the significance per influence factor on changeover losses. In the case, an influence factor caused only minor changeover losses in one or two sub-units it was classified as low. Subsequently, if an influence factor caused changeover losses in the majority of the sub-units, it was classified as medium. Finally, in the case an influence factor caused major changeover losses across all sub-units it was classified as high.
Table 10. Cross-case analysis: Effects and significance of influence factors on changeover losses
Influence factors of changeover losses
Data sources used per influence factor
Sub-unit A Sub-unit B Sub-unit C Sub-unit D Overall
Significance Out-of-date SMED-list Observations Interviews Production report Maintenance report • 31% of the standards provided are incorrect
• 57% of the standards provided are incorrect • No SMED-checklist
• 42% of the standards provided are incorrect • No SMED-checklist
• 14% of the standards provided are incorrect
High Precision of changeover activities Observations Interviews Production report
• Halted production line multiple times during start-up phase
• Imprecise changeovers due to lack of skills
• Minor losses due to high precision of activities Low Changeover during shift change Observations Interviews Production report • 27% longer setup times (16 minutes) • Product quantity lost:
8000 (output 30000 hour)
• No information transfer between shifts
• 32% longer setup times (29 minutes) • Product quantity lost:
10633 (output 22000/hour)
• 32% longer setup times (39 minutes) • Product quantity lost:
13000 (output 20000/hour)
• Double checking and setting of parameters
• 30% longer setup times (90 minutes)
• Product quantity lost:
15000 (output 10000/hour)
Page 31 of 51
Influence factors of changeover losses
Data sources used per influence factor
Sub-unit A Sub-unit B Sub-unit C Sub-unit D Overall
Significance
Learning process Interviews
• Lack of uniformity in training procedure
• Multiple mentors per operator in training
• Major losses due to distraction of the experienced operator during a changeover
• Operator does not follow basic working method, as different methods are taught High Operator capabilities Observations Interviews • Technicians lack specialism to quickly identify root causes
• Operators require basic technical knowledge to solve minor issues
• Willingness and intrinsic motivation are found crucial for improvement Medium Data gathering and reporting Interviews Production report Maintenance report • 68% production issues are not reported • Poor follow-up issues
• 48% production issues are not reported
• 73% production issues are not reported
• 57% production issues
are not reported High
Shift team composition
Observations Interviews
• No fixed shift teams • No operator rotation
• Experienced operators are forced to supervise
• 45 minutes downtime
due to supervision Medium
Quality of changeover equipment Observations Interviews Production report • Continuously out-of-date SMED-lists due to changing parameters caused by maintenance • Faulty changeover equipment resulted in 170 minutes downtime 62333 product losses • Machine standards changed after maintenance • Mismatch maintenance moulds with machine
High Surface contamination Observations Interviews • Operator changed parameters instead of clean contamination • Surface contamination was argued to have a low impact on changeover losses
Page 32 of 51
5. DISCUSSION
Manufacturers in FPI must no longer consider changeover operations as non-value-added activities, as changeovers are an important aspect of manufacturing operations and greatly improve production output (Gungor & Evans, 2017). Across the sub-units, influence factors and their significant impact on changeover losses were systematically probed. In total nine influence factors are explored with varying effects on changeover losses. While the influence factors and their significance found may be case specific, the findings are still relevant and applicable for FPI manufacturers, as it improves the current knowledge of FPI manufacturers to reduce changeover losses. The explored influence factors and their significance are classified into three critical themes which are widely discussed in literature and applicable to FPI manufacturers: knowledge management, working methods and organisational design.
Knowledge management is a critical theme to reduce changeover losses, as FPI manufacturers must ensure that shop floor personnel possess and disseminate their capabilities and expertise for improving best practices of changeover operations (Gungor & Evans, 2017). FPI manufacturers deal with knowledge management, as it entails the process of development, storage and dissemination of expertise among shop floor workers to support and improve its performance (Wiig, 1997). Gungor and Evans (2015) also argue the importance of knowledge management in reducing changeover losses, as they argue that the value of knowledge gained through experiences on the shop floor is often underestimated and therefore not effectively used to improve best practices.
Shift team composition is found to play a crucial role in improving knowledge management. FPI
