Suggestions for implementation of Statistical Process Control in Lithium-ion battery processes

Suggestions for implementation of Statistical

Process Control in Lithium-ion battery

processes

A field study at Northvolt

Maximilian Trydegård

Magnus Blide

Industrial and Management Engineering, master's level 2020

Luleå University of Technology

Preface

This thesis is the product of the study we have conducted at Northvolt in Västerås and the final work in our master's in quality technology at Luleå Technical University. It has been a rewarding and inspiring time working with quality in the exiting lithium-ion battery production industry at Northvolt (NV). We would first and foremost like to express our gratitude to Lina Emilsson who allowed us to conduct our thesis at NV. We would also like to thank our supervisors at Northvolt, Neda Bigdeli and Taekyoun Kwon. Their commitment to our thesis and their guidance at the facility in Västerås helped us get valuable insight into the operation, this enabled us to nuance the field study and produce results we are proud of. We would also like to thank our supervisors at Luleå Technical University, Francesca Capaci and Jens Aldenlöv. Their support regarding strategy and report structure has been of great value during the work. Our classmates Adam Sahlberg, Phuc Nguyen, Emil Hellström, Malin Malmborg and Hannes Ingre Aronsson have contributed with reviewing drafts of the report at several occasions during the thesis project, which we are deeply grateful for.

As this thesis also is a result of our five years at Luleå Technical University, we would like to thank all involved professors and teachers at the Department of Business Administration, Technology and Social Sciences (ETS).

Abstract

Preliminary table of contents

1. Introduction ... 1

1.1 Background ... 1

1.3 Purpose and Objectives ... 3

1.4 Delimitations ... 3

2. Theoretical framework ... 4

2.1 Statistical process control ... 4

2.2 Success factors in SPC implementation ... 5

2.2.1 Methodologies for Implementation of SPC ... 7

2.3 Choosing of control charts ... 11

2.4 Phase 1 of statistical process control charts ... 12

2.4 Capability analysis ... 13

2.5 Process prioritization ... 13

2.6 Critical to quality characteristics ... 13

2.6.1 Concept and definition ... 13

2.7 KPI based approaches for Quality management ... 14

2.7.1 A KPI-based method for Quality assurance in battery production ... 15

2.7.2 Identification of new cause and effect relations ... 16

2.8 Alarms and nonconformities / Out of control values ... 16

3. Method ... 17 3.1 Overall method ... 18 3.2 Research strategy ... 19 3.3 Literature study ... 19 3.4 Field study ... 20 3.4.1 Data gathering ... 20 3.4.2 Thematic analysis ... 22

3.4.3 Analysis of the quantitative data ... 24

3.4.4 Analysis of knowledge from processes and NV ... 24

4. Field study ... 25

4.2 Quantitative data ... 25

4.3 Process information and structure. ... 27

4.3.1 Measuring procedures ... 27

4.3.2 Parameter characteristics ... 27

4.3.3 Cause and effect relationships ... 28

4.4 Evaluation of measurement systems ... 28

4.5 Quality parameter controls ... 28

4.6 Software ... 28 4.7 Critical-to-quality characteristics ... 28 4.8 Handling of nonconformities ... 29 5. Analysis ... 29 5.1 Thematic analysis ... 29 5.1.1 Theme structure ... 30

5.1.2 Theme content relevance ... 31

5.1.3 Analysis of critical factors rankings ... 33

5.1 Analysis of NV organization ... 33

6. Conclusions & Recommendations ... 35

6.1 Framework for SPC implementation ... 35

6.2 Organizational Supporting activities ... 38

6.3 Technical supporting activities ... 38

7. Discussion ... 39

7.1 Purpose ... 39

7.2 Planning ... 39

7.3 Practical contributions ... 39

7.4 Theoretical contributions ... 40

7.5 Validity and Reliability ... 40

7.6 Future research ... 40

Table of figures

Figure 1, Process overview lithium-ion battery production... 2

Figure 2, Framework for SPC implementation... 8

Figure 3, Roadmap for SPC implementation ... 9

Figure 4, Selection process... 11

Figure 5, KPI system... 16

Figure 6, Flowchart for out of control situations... 17

Figure 7, Overview of the overall method ... 18

Figure 8, Friedman hypothesis test. ... 27

Figure 9, Thematic analysis ... 29

Table of tables

Table 2, Indicies ... 15

Table 3, Search terms ... 19

Table 4, List of respondents ... 22

Table 5, Thematic analysis ... 23

Table 6, Results from ranking of factors ... 26

1. Introduction

The chapter provides a background of the report. First, a description of the problem is presented followed by the purpose and delimitations of the thesis.

1.1 Background

As the world becomes more aware of the challenges regarding climate change the need for sustainable energy solutions increases every day. The rapid changes in the political climate put pressure on many organizations and call for large industry reforms. To reduce the environmental impact, experts on the field agree that renewable sources should replace oil and coal to generate electricity (Panwar, Kaushik & Kothari, 2011).

The general population’s reaction to the climate changing has created a demand for new industries to grow. In recent years alternatives to traditional combustible fuels have gained in popularity. Where the usage of electricity as fuel is seen as a much better alternative. This has led to the growth in the lithium-ion battery (LIB) industry due to the easy application in different fields, mainly in the automobile industry due to the popularisation of electric cars (Kley, Lerch, & Dallinger, 2011). Northvolt (NV) is a company that will bring energy-aware LIBs to the European market, with a vision of controlling the entire supply chain. NV will produce batteries from factories such as the Giga factory NV Ett in Skellefteå, Sweden and NV Zwei in Germany (NV, 2020). At NV Labs in Västerås, Sweden the quality of the processes in NVs LIB mass production is tested to meet the criteria of the customer, ensure the safety of the personnel and the quality of the product. The quality of the final product is affected by the quality of the processes and the materials that the battery cells are made from (Zhang et al., 2000). Failing batteries can result in devastating consequences, including explosions and fire (Wang, Ping, Zhao, Chu, Sun, & Chen, 2012). The lack of quality can lead to safety risk. Therefore, the quality of the product is critically important (Dougthy & Roth, 2012).

One way to reach high product quality is to control the processes via statistical process control (SPC) (Montgomery, 2013). Sousa, Rodrigues & Nunes (2017) aimed to gain an understanding of the process in pre-production to prepare for mass-production of LIBs. The authors observed mean and range values in production and created control charts, and via analysis of the charts, they could list critical factors to consider in mass production. The processes seemed to be in control and ready for mass production. However, after several months of production, numerous customer complaints forced the team to conduct further investigation. The number of defect products implies that the process was unable to produce within the specification limits. Individuals moving range control charts were produced and a significant change in mean and variability of the process was detected. Root cause analysis led to a solution to the problem but the important conclusion Sousa et. al (2017) took from this case study was that factors considered not critical in test production could eventually turn out critical in mass production. The importance of the continued use of SPC becomes evident in a case like this.

process step by step from preparing the active materials that go in the cell, to the filling and formation of the cell, see Fig 1. The same process steps are performed for both the anode and the cathode, which make up the battery cell, with some variation due to the desired characteristics of the two parts. The scope of this thesis concerned the first two steps in the process. Slurry preparation, which is referred to as Slurry Mixing in this report, is about preparing, and mixing the active materials of the battery cell. For the anode, this includes mixing of the active graphite, conductive carbon and binder solution. For the cathode, this includes mixing of NMC-oxide conductive carbon and binder solution.

Coating & drying, referred to as Coating in this report, is the process of coating the foil (current collectors) with the slurry (Smekens et. al, 2016). This process step follows Slurry Mixing. The slurry is coated onto the current collectors, which are made of aluminium (Cathode) and copper (Anode). The thickness of the coated slurry depends on the electrodes maximum available capacity (MAC). The coated foil is then dried in an oven to dry the solvent. For the anode, the solvent is water. For the cathode, the solvent is NMP (N-Methyl-2-pyrrolidone).

Figure 1, Process overview LIB production – inspired by Smekens et. al (2016)

1.2 Problem description

young age of all processes. NV does not have a clear plan for the implementation of SPC into their company and their quality system. Some initial driving forces towards implementation can be found but no clear plan. (NV, 2020).

1.3 Purpose and Objectives

The LIB process at NV is complex and has many criteria that need to be met, especially in terms of quality. Implementation of SPC is associated with many challenges, although the usage of SPC is valuable to ensure viable quality to the customer as well as ensuring that the final product is safe to produce. The purpose or main objective of this thesis is to investigate:

“How can SCP be implemented in LIB production processes”

The purpose will be achieved by analysing data from the field study, which includes analysis of qualitative and quantitative data compared with relevant theory on the matter. The field study will include further investigation regarding quality parameters, current sampling and measurement plans for gathering data, observing possibilities implementing control charts. To structure the work of the field study the following sub-objectives were developed.

1. Understand the processes and the opportunities related to SPC at NV 2. Collect data using interviews and observations at NV

3. Analyse the data and give recommendations for future quality routines when implementing SPC.

1.4 Delimitations

2. Theoretical framework

The theoretical framework presents practical applications of statistical tools and SPC such as Shewhart diagrams and capability analysis. The focus of this section is on studies that implement these tools. Previously made studies showcasing important success factors and quality parameters in the production of LIB is also presented.

2.1 Statistical process control

Since the 1980s the manufacturing industry has gone through a revolution in connection to quality work, this in parallel with the development of new manufacturing methods and technologies. This revolution shows a shift in focus, from the detection of defects to the prevention of instability through a reduction in process variation. W.E Deming popularized this statistically built technology originally developed by Walter Shewhart, this method is called SPC. SPC is a collection of powerful problem-solving tools to achieve process stability and improved capability, through a reduction of process variability (Lim, Antony & Arshed, 2016). SPC is a valuable method to understand process behaviour and to help operators and managers who work in production to make real-time decisions. The systematic use of SPC leads to simpler ways to identify sources of variation and out-of-control situations. Effective implementation will also ensure mechanisms to eliminate unusual sources of variation and allows these actions of eliminating unusual variation to be a part of the work that engineers, managers, and operators perform daily (Elg, Olsson &, Dahlgaard, 2008).

Rantamäki, Tianen and Kässi (2013) describe an important part of the successful usage of statistical tools and statistical thinking. Statistical thinking is a philosophy of learning and action that relies on the following fundamental principles: All work is done in a connecting system,

variations exist in all processes and the understanding of reduction in variation is the key to success (Rantamäki et al., 2013). Statistical thinking gives an understanding of how statistical tools

can be used to improve processes. The base for SPC is in the partition of usual sources of variation and special causes for variation. Usual causes for variation will always be present in a process. An analysis in connection to usual cases in a stable process will not lead to improvements. A special cause variation will lead to an unpredictable output of the process. Control charts can help distinguish these two types of variations. Rantamäki et al. (2013) stresses that SPC is not a program or software that can be bought and implemented without a technique that includes the usage of problem-solving tools, such as, control charts, pareto diagrams, and cause and effect diagrams. Organisational processes are also important for the success of SPC and should, therefore, be treated with an implementation. Without supporting processes that drive improvements, individual tools have a short lifetime (Rantamäki et al., 2013).

often fails due to the low knowledge and understanding of the techniques and implementations of SPC (Lim & Anthony, 2013).

Montgomery (2013) and Gafaar & Keats (1992) agrees with Lim et al. (2016), they all describe that SPC can be summarized as a collection of powerful problem-solving tools for achieving better process stability and improved capability by the method of reducing variability. Montgomery (2013) lists the seven major tools used in SPC as follows:

1. Histogram or stem-and-leaf plot 2. Check sheet

3. Pareto chart

4. Cause-and-effect diagram 5. Defect concentration diagram 6. Scatter diagram

7. Control chart

These tools are an important part of SPC, but they can only be described as the technical aspects of what SPC brings. Successful implementation of SPC will bring many other aspects and create an environment which helps all individuals seek continuous improvement in quality and productivity (Montgomery, 2013).

2.2 Success factors in SPC implementation

SPC can be a valuable technique for understanding processes and to make real-time decisions accordingly. If this method is used systematically it can be used to detect and reduce variation as well as detecting out of control situations (Elg, Olsson, & Dahlgaard, 2008). Many factors will affect the implementation of SPC and these factors can be called Critical success factors (CSF). If there is a better control regarding the factors there is a higher chance for SPC implementation success (Elg et al., 2008). These factors can both be technical and organizational, both should be considered when implementing SPC (Elg et al., 2008). Management commitment has shown to be a critical factor for success (Antony, Balbontin & Taner (2000); Rohani, Yosuf & Mohamad (2010); Rungasany, Antony & Gosh (2002)). Other factors have also been shown to be critical with the implementation of SPC such as selection of quality characteristics (Antony et al., (2000); Rohani et al., (2010); Rungasany et al., (2002); Elg et al., (2008)) and it is a good practise to choose characteristics that can be measured precisely, accurately and with stability (Antony et al., 2000). The success factors found in literature is illustrated in the table below, similarities are evident from different research and many of the factors are reoccurring.

Table 1, Success factors for successful SPC implementation

Success factors for a successful SPC implementation

Antony , Balbontin and Taner (2000)

Rohani, Yosuf and Mohamad (2010)

Rungasamy Antony and Gosh (2002)

Control charts X X X

Documentation and update of knowledge of processes X Cultural change X X X Identification of critical characteristics X X X Deployment X

SPC training and education X X X

Process prioritisation and definition X X X Measurement system evaluation X X X Management commitment X X X Use of SPC software X X Use of SPC facilitators X X

Use of pilot study X X X

Teamwork X X X

The authors Rungasam, Antony and Ghosh (2002) researched the most important factors when implementing SPC using surveys, where the order of 12 critical success factors was ranked as follows: (1) management commitment, (2) teamwork, (3) identification of critical quality

characteristics, (4) control charts, (5) documentation and update of knowledge of processes, (6) measurement system evaluation, (7) process prioritisation and definition, (8) cultural change, (9) SPC training and education, (10) use of pilot study, (11) use of SPC software, (12) use of SPC facilitators. Elg et al (2008) researched the same subject, their results are similar to Rungasam et

al., (2002) were the factors: management commitment, general and specific training, identification

of processes to work with, choice of parameters and control chart techniques to be studied, and project team, were the most prominent.

2.2.1 Methodologies for Implementation of SPC

Rungasam et al., (2002) identified real benefits with the application of SPC. They found that the most important and prominent benefits were: Reduction of waste, improved in-process control,

improved packaging efficiency, improved process analysis and monitoring, improved communication, improved involvement of people. Schippers and Trip (1997) corroborate this and

describes further benefits such as Financial benefits, better communication with customers and

suppliers and that the organization will be statistical-oriented leading to decisions made on data.

Antony and Taners (2003) developed this framework by determining the essential ingredients that will make an application of SPC successful. Four essential areas were determined and according to Antony and Taner (2003) they should be in focus, the essential areas are management issues, engineering skills, statistical skills and teamwork skills.

The areas can be described as follows:

1. Management issues include total company commitment, resources for training and education, creating an environment that is responsive for actions on the process/systems 2. Engineering skills include the understanding of SPC and the benefits that it gives, action

taken for out-of-control situations, prioritizations of processes

3. Statistical skills include statistical stability and capability, selection of appropriate control charts, interpretation of the control charts and out-of-control situations.

4. Teamwork skills include the formation of process action teams for out-of-control situations, company-wide understanding of SPC, its benefits and rewards.

Antony and Taner (2003) developed a framework according to the formerly mentioned areas above and a critical analysis of literature. This can be seen in Fig.2 below

Figure 2, Framework for SPC implementation, inspired by Antony and Taner (2003)

Using the framework further development resulted in Antony and Taner (2003) creating a roadmap which is a structured way helps the implementation of SPC, this roadmap is based on the critical success factors for SPC implementation. Antony and Taner (2003) did this due to the realization that companies and academia often focus purely on control charts and not on the leadership and implementation of SPC. Antony and Taners (2003) solution is showcased in a structured way in

Fig 3 that follows below.

STEP ACTIONS

1 Understand the importance of SPC

2 Get appreciation from top management

3 Prolong training programmes in all parts of the company

4 Identify a pilot project and estimate potential costs and savings possibilities

5 Choosing an implementation team

6 Evaluation of measurement systems

7 Choosing of suitable quality characteristics related to the processes

8 Choosing of appropriate control charts

9 Develop a process action team for out of control situations

10 Document and update process knowledge

11 Evaluate the SPC usage

Figure 3, Roadmap for SPC implementation

Schippers and Tripp (1997) describe a similar implementation process as Antony and Taner (2003) which is divided into four phases. These phases are as follows:

• Phase 1: Awareness. • Phase 2: Pilot projects.

• Phase 3: Integral implementation in production. • Phase 4: Total quality.

of financial benefits, better communication and involvement with operators. After the initial meeting, the steering committee should convene to determine what processes to handle in the pilot project (Shippers & Tripp, 1997).

In phase 2 the steering committee should assign teams called “process action teams” or PATs to work on process decided in step 1. The committee should work together to bring the process under control using the steps mentioned below. The overall goal of the process is for a team to implement these 10 steps and bring the process which is examined into control. The steps are 1. Process

description, 2. Cause and effect analyses, 3. Risk analysis, 4. Improvements, 5. Define measurements, 6. R&R study, 7. Control charts, 8. Out of control action plan. 9.Process capability study, 10. Certification. The steps are then grouped into 4 parts where (Shippers & Tripp, 1997):

● Steps 1, 2 and 3: Describes the process and search for weak points

● Steps 4 and 5: Search for improvements for weak points and to implement them ● 6,7 and 8: To define measurements- and control loops to control process

● Steps 9 and 10: To assess the performance and arrange for continuous improvement.

Phase 3 aims to integrate SPC into further processes using more PATs. This stage can stretch over a much longer period, from about 1.5 years to 2.5 years, it depends on the amount and complexity of the processes (Shippers & Tripp, 1997).

In phase 4 Shippers and Tripp (1997) describe that all the processes are under control, the PATs should be dismissed and transformed into process improvement teams. Their job is then to ensure the control of the processes, tackling problems, searching for opportunities for continuous improvement. In this phase, the SPC should be broadened into other parts of the company. Phase 3 does have an impact on other areas such as development, purchasing, customer etc but in phase 4, the SPC approach should be actively extended. Further, SPC is based on prevention instead of detection, therefore, it is a logical step to start using SPC to reduce variation in developing products and processes development (Shippers & Tripp, 1997).

2.3 Choosing of control charts

Many processes will benefit from the usage of SPC. The usage of control charts is an important part of implementing SPC but some specific steps should be thought of before considering control charts (Rungasam, Antony, & Ghosh 2002). Montgomery (2013) describes these steps in a summarized matter like the following:

1. Determining which process characteristics to control

2. Determining where the charts should be implemented in the process 3. Choosing the proper type of control charts.

4. Taking actions to improve processes as the result of SPC control chart analysis 5. Selecting data-collection systems and computer software

Anjard (1995) points out that the selection of proper SPC charts is essential for the use and implementation of SPC. Variable control charts are designed to control product or process parameters which are to be measured on continuous measurement scales such as pounds, inches, miles etc. For the continuous measurement data, the primary charts used are the X-bar, R and the Individual moving range charts. For attribute data, there are p-charts, c-charts, np-charts and u-charts (Anjard, 1995). The selection process is represented in the decision tree in Figure 4. The basic data to be charted is represented by variables or attributes, further decisions are followed in each decision point in the chart.

Figure 4, Selection process, inspired by Anjard (1995)

Antony and Taner (2003) describes the process of constructing the control charts with the following steps:

• Establish a suitable and responsive environment for action.

• Define the process where SPC must be applied and its link the other processes both upstream and downstream.

• Determine the quality characteristic (or process parameters) which needs to be measured, monitored, and managed.

• Define the measurement system and determine whether the current measurement system is capable to do its intended job.

• Understand the type of data and select a suitable control chart for the process.

The control charts are used to:

1. Detect special causes of variation in real time to adjust the process accordingly.

2. Identify patterns in the variation of the process which can act as early indicators of an upcoming defect.

3. Provide statistical limits, which defines the natural tolerance of the process and are used to stabilize the process.

Most statistical control charts are simply a plotted average of a small sample that provides information on the stability of the process. The control limits can and are often represented by plus or minus 3 sigma limits of the normal curve, representing 97.73% of the data. If plotted (averages) exceed the control limit, the process may and is probably out of control (Anjard, 1995).

2.4 Phase 1 of statistical process control charts

2.4 Capability analysis

Statistical techniques can be useful throughout the product cycle. Statistical techniques include development activities before manufacturing, quantifying process variability, analysing the variability in relation to product requirements or specification. Montgomery (2013) describe these activities as process capability analysis. Montgomery (2013) points out that a process capability study usually measures parameters or critical-to-quality characteristics on the product, not the process itself. By controlling the process parameters and the sampling of the product one can draw conclusions regarding the stability of the process over time (Montgomery, 2013).

Capability ratios is a convenient and simple way to measure process capability (Montgomery, 2013). The ratios Cp and Cpk are widely used in industry and managers often require capability analyses and reports from their processes. To perform a capability analysis the process that is going to be analysed needs to be in statistical control (Somerville & Montgomery, 1996).

2.5 Process prioritization

Normally in production, a product is produced through several processes and subprocesses which all together contribute to the final product. In reality, it is not practical to implement SPC plantwide in the first instance due to cost and time constraints (Antony & Taner, 2003). The best way to take care of this problem is to prioritise processes using their technical and statistical criticality. Xie, Goh and Xie, (1995) means that technical criticality refers to how the process is relevant to the quality of the final product and production process. Whereas statistical criticality is connected to the stability and capability of the process. Tools such as Quality function deployment (QFD) and Analytical Hierarchy Process (AHP) can be used but will still be based on personal judgement and subjective thoughts (Xie, Goh & Xie, 1995).

2.6 Critical to quality characteristics

In order to satisfy the customer, a product needs to possess corresponding quality characteristics (QCs), which are the inherent properties of the product. To effectively measure the product quality the QCs should be a polymorphous object satisfying requirements coming from customers, enterprises, and engineers in different perspectives (Tang & Chang, 2009). To understand the relationship between product quality and QCs Tang and Chang (2009) developed a theory in connection to this. QCs consists of three layers: 1. Quality dimensionalities (QDs), which fulfil requirements from customers, 2. Quality essentials (QEs), which fulfil marketing and competitive requirements from the enterprises, 3. Quality attributes (QAs), which fulfil the design requirements of engineers. The QCs meet quality requirements such as functionality, reliability, safety,

durability, maintainability, economic ability, environment ability and aesthetic properties of

product from customers (Tang & Chang, 2009).

2.6.1 Concept and definition

specification limits or other factors related to the product or service (Tang & Chang, 2009). Tang and Chang (2009) define the concepts of CTQs based on the design for six sigma and the focusing on “vital few” and not “trivial many” and is as follows: “CTQs are a few QCs that determine and

affect product quality, and CTQs are an aggregation of QDs, QEs and QAs of the product. In design, the CTQs correspond with the performances as well as the specification of a product or features of its assemblies, parts or processes, whose variation or absence from target can draw forth unacceptable impact on the functionality, cost or safety of a product. Further, the authors

describe that the variation or absence of CTQs will influence the satisfaction, safety, reliability, conformance to regulations, consistency, function, and performance of a product (Tang & Chang, 2009).

2.7 KPI based approaches for Quality management

LIBs face a major challenge in industry-wide implementation due to the complexity of the production system. LIBs have high-quality requirements in terms of target figures in connection to capacity and safety but are also characterized by a high degree of complexity which is a direct cause of the number of process steps. Each process step has a lot of input and output variables. Input variables are material properties, process parameters or disturbances. Output variables are intermediate product characteristics that may serve as input for subsequent processes. A LIBs process can contain more than 600 variables that will create a large amount of cause-effect relationships (CERs) and experts consider around 75% of the CERs to be critical in terms of product quality (Kornas et al., 2019). Kornas et al. (2019) identified that commonly used approaches such as DOEs and FMEAs to identify CERs did not apply well to the complex production of LIB. The authors instead identified that multivariate approaches are promising for the investigation of interlinked CERs and therefore developed an approach that uses quality-related KPIs in particular on multivariate process capability indices (Kornas et al., 2019).

Korans et al. (2019) describe that KPIs are often used to support decisions and to monitor, control and coordinate businesses. Quality indicators can be quality rate, scrap rate, defect rate, yields, sigma levels etc. KPIs are effective for a general production review but cannot be applied to identify interlinked root causes or detailed problem properties (Kornas et al., 2019). However, a widely used KPI for process analysis is the process capability indices. Several different indices can be applied both for univariate analyses and to multivariate approaches (Kornas et al., 2019). The most used univariate indices are known as process capability Cp and critical process capability Cpk. These indices can be extended by also considering the loss of quality when variation moves from a predefined target value. This leads to the indices Cpm and Cpmk (Kornas et al., 2019).

Table 2, Indicies

Index Formula Explanation

Cpk Compares the process variance with the

spread of the upper and lower specification limits, also considering the position within the specification limits.

Cpmk This index does also consider that the

quality loss increases as the variation moves from a pre-defined target value.

MCpk A multivariate process capability index.

Which contains all correlations or covariances of the regarded process characteristics. Applicable to bundle numerous characteristics from different process steps.

2.7.1 A KPI-based method for Quality assurance in battery production

Kornas et al. (2019) have developed a KPI system that will try and connect already existing analytical methods and the complex process chain of LIB production. This to later show how CERs can be identified in a production ramp-up as well as tracking problems in the process chain.

The system is based on a hierarchical structure as shown in Figure 5. Product and process characteristics are aggregated in 4 layers to identify root causes of quality deviations and CERs. The first level represents physically measured parameters such as material properties, disturbances, or process parameters. As per the characteristics of the measurements a univariate Cpk is calculated for these. The second level represents the intermediate product characteristics, this is done by aggregating the input values from the first level into an MCpk. An example to this would be that the thickness of the cell body is represented by an MCpk. If the intermediate product characteristics are measured directly, they will be represented by Cpmk. The third level represents the product characteristics of a LIB cell, such as the capacity, the coulombic efficiency or the weight. These are also evaluated using MCpk and Cpmk. The top level is represented by the main KPI also represented by a MCpk, a KPI which aggregates the product characteristics from the third level. The approach using MCpk was originally developed using normally distributed data. It is not always certain that the dataset used is normally distributed, therefore this needs to be checked, and if not normally distributed, it needs to be transformed before entering the MCpk calculation. Preferably using the box-cox or root transformation (Kornas et al., 2019).

Figure 6, KPI system, inspired by Kornas et al., (2019)

2.7.2 Identification of new cause and effect relations

By solely monitoring the process no new knowledge of the process will be gained, a further benefit of the KPI-system is the identification of new CERs. This can be done by comparing the values of MCpk and the Cpmk.

In summary Kornas et al. (2019) state that uncertainties are among the top impediments during the ramp-up of LIB production. Further, Kornas et al. (2019) introduced multivariate KPI-based method can be applied to identify CERs in the production of LIBs. It also shows high potential in other areas, to monitor a complex process chain and to help identify problems by using a target-oriented analysis path. Kornas et al. (2019) describe further benefits of the KPI system such as it provides a condensed and structured illustration of the production chain. It reduces the control and monitoring effort and helps to limit the scope of a DoE. Therefore, the KPI can be used as an overall quality assurance tool (Kornas et al., 2019).

2.8 Alarms and nonconformities / Out of control values

process. In other words, an action plan is needed when discovering that a process exhibits special causes of variation (Antony & Taner, 2003). The flowchart below, figure 6, showcase Antony and Taner (2003) process of handling out of control situations.

.

Figure 6, Flowchart for out of control situations, inspired by Antony and Taner, 2003

3. Method

3.1 Overall method

The overall method of the thesis work is shown below in figure 7, the initial purpose of the report was to implement SPC into the processes which were present at NV labs. As the project progressed COVID-19 created obstacles for the thesis work and an alteration to the purpose was stipulated. Where the initial purpose was demanding an actual implementation of SPC and later changed to understand and investigate the best way to implement SPC into NVs LIB production process. During the thesis work, process knowledge and informal interviews were held throughout the timeframe of the thesis. A study on SPC implementation and connecting topics were conducted to build a solid foundation for the field study. The field study acted to objectively observe the LIB processes, understanding the opportunities at NV regarding SPC implementation, including digital systems and quality-related aspects. Moreover, semi-structured interviews with persons from different backgrounds and positions were conducted. Furthermore, a quantitative method for determining the most important factors for successful implementation at NV was developed. This data was gathered in connection to the interviews, which is described further in section 3.4 Data

gathering The data collected in the field study formed the foundation for the analysis which is

further described in section 3.4.2 thematic analysis and 3.4.3 analysis of quantitative data. Lastly, recommendations on the implementation of SPC were given and a discussion regarding the results of the thesis was held.

3.2 Research strategy

The approach of the thesis can be described as descriptive, the thesis aim was to achieve new insight into the process for LIB production regarding SPC implementation. Marincola (2007) describes the descriptive approach of studies suitable when clear hypothesis cannot be stipulated and tested, but when there is a need for relevant information on a specific subject. Moreover, Grimes & Schulz (2002) claim that descriptive studies emphasise features of new events and can be used to plan recourses and monitor trends. The LIB process is complex and new at NV and in the northern parts of the world new. This approach was chosen to provide useful information as a foundation for future implementation of SPC at NV.

3.3 Literature study

A literature study was conducted to learn what has been done on the field. Already existing data and literature is according to Kothari (2004) categorized as secondary data. Secondary data can be in the form of published or unpublished data, during this study published data was gathered for the literature study. Published data came from articles and reports published in journals but also some information from books published by well-known and cited authors. Kothari (2004) explains the problems in using secondary data where the data might be unsuitable or inadequate in the context of the research question the report wants to determine and handle. Therefore, the focus of the data gathering was on primary data gathered at NV. Google Scholar and Scopus were used in the search for scientific articles. Well-renowned scientific journals with articles no older than 15 years were prioritised. However, exceptions were made for particularly interesting, well-cited articles and relevant books. A summary of the most used search terms is showcased below in table 2.

Table 3, Search terms

SEARCH TERM GOOGLE SCHOLAR (RESULTS) SCOPUS (RESULTS) “STATISTICAL PROCESS

CONTROL ADVANTAGES”

26 500 2 017

“STATISTICAL PROCESS CONTROL SUCCESS FACTORS”

5 010 270

“LITHIUM ION BATTERY PRODUCTION”

912 2 490

“SPC IMPLEMENTATION” 920 750

3.4 Field study

The following section the method for the data gathering of the field study is described.

3.4.1 Data gathering

Both primary and secondary data was gathered, where the most focus was on collecting primary data. Kothari (2004) explains primary data as data collected afresh and for the first time, specifically for the collector’s purpose. Secondary data is described as data that already has been collected by someone else for a different purpose. Secondary data was gathered through informal interviews and observations made at the organization. The observations included insights in the digital system used in NV, information regarding the process of LIB as well as knowledge on data and the quality of that data in the organization. The data in connection to SPC implementation was gathered with Antony and Taner (2003) roadmap as a foundation, this roadmap was chosen due to its great alignment with other studies regarding SPC implementation. Furthermore, the development of Antony and Taners (2003) roadmap was done using other authors framework discussion their strengths and weaknesses. Because of these reasons, the roadmap deemed to be of adequate quality.

The primary data in this thesis was qualitative and was collected through interviews with key people in the organization. The interviews were semi-structured and were conducted in person at NV Labs, in Västerås. Dicicco-Bloom and Crabtree (2006) describe some of the benefits of individual, semi-structured interviews for qualitative data gathering. The One-on-one approach allows the dialogue more space and the respondent is given a better opportunity to express his or her thoughts more freely than in a situation of a group interview or a focus group. In contrast to surveys, semi-structured interviews allow open questions which allow the dialogue to be personal and, in some cases, deeper and more honest (Dicicco-Bloom & Crabtree, 2006). A questionnaire was created with the most important and relevant findings from the literature study in mind. The questionnaire was structured in three sections.

Section 1

The first section included questions regarding general information about the respondent. This section also included open questions about their knowledge and experience of organizational change and the implementation of new routines. The respondent's knowledge of SPC was also discussed early in the first section.

Section 2

The second section was more detailed and the focus in this section was to dig deeper into the respondent's thoughts on going from test production to mass production. Many of the persons interviewed had previous experience of quality work. However, the interest in this section was to put this knowledge into the context of LIB production.

Section 3

also included an exercise for quantitative data were the respondents were given 12 plastic cards with different organizational ingredients on each card. The respondents were also given a whiteboard pen and ware asked to rate the different aspects in order of importance when implementing SPC on a large scale, going from test-production to mass production. The 12 organizational ingredients or aspects were derived from the literature on successful implementation. Earlier studies that have made claims on what is most important when implementing SPC, the aim here was to discover how the theories work in the context of LIB production. Two additional respondents took part in the factor ranking exercise, making it eight respondents in total.

Secondary data was collected through research of relevant industry-specific literature, this data acted as guidance for analysis and further knowledge of the processes that the primary data during the project was collected from.

Respondents

Table 4, List of respondents

RESPONDENT ROLE AT NV EXPERIENCE

A1 Process Quality Expert Over 10 years of experience in quality work at Samsung

A2 Head of Quality Background from the pharmaceutical industry. Experience in Sex Sigma, DOEs and process capability.

A3 Analytical chemist PhD in battery materials (crystal structure and electronic properties)

A4 Manufacturing Technician, team leader

Many years of experience as an operator in the paper and pulp industry, 9 months at NV.

A5 Data Analytics Quality Engineer PhD in solar energy technology. Responsible for the data quality and data analytics applied to quality at NV.

A6 Master Thesis Student (Quality team)

Theoretical knowledge from the university. Working with MSA at NV.

3.4.2 Thematic analysis

A thematic analysis was done following Braunus and Clarkes (2006) six phases for analysis of qualitative data, in this case from semi-structured interviews. In the table 4, each step along with Braunus and Clarkes (2006) description of the activities linked to each step is presented.

Table 5, Thematic analysis

PHASE ACTIVITIES 1 Familiarizing yourself with your

data

Transcribing data (if necessary), reading and re-reading the data, noting down initial ideas.

2 Generating initial codes Coding interesting features of the data in a systematic fashion across the entire data set, collating data relevant to each code.

3 Searching for themes Collating codes into potential themes, gathering all data relevant to each potential theme.

4 Reviewing themes Checking if the themes work in relation to the coded extracts (Level 1) and the entire data set (Level 2), generating a thematic ‘map’ of the analysis.

5 Defining and naming themes Ongoing analysis to refine the specifics of each theme, and the overall story the analysis tells, generating clear definitions and names for each theme.

6 Producing the report The final opportunity for analysis. Selection of vivid, compelling quotes, final analysis of the selected extracts, relating back of the analysis to the research question and literature, producing a scholarly report of the analysis.

In phase one, the raw material from the interviews was examined multiple times to create an understanding of the central ideas and aspects of the respondent’s answers. Moreover, to create an overall picture of the material. In this phase, the material was also transcribed into written form. Braunus and Clarkes (2006) stress that when working with data such as recorded interviews or similar, transcribing is necessary for thematic analysis. During this phase, initial ideas were written down in preparation for phase two.

In phase two relevant quotes and sentences were identified and summarised in an orderly manner to develop codes. Codes represent a basic element or information of the raw data that interests the analyst (Boyatzis, 1998). Internal discussions led to which quotes and sentences were considered most critical for the thesis. The interpretive part of the analysis starts in phase three (Braunus and Clarkes, 2006). The search for themes among the codes with similar underlying aspects was initiated. The focus at this stage was to understand how different codes were linked. A thematic map was created to get a better overview of codes and potential themes.

Clarkes (2006) claim that this helps to understand if the potential themes reflect the core meaning of the data set as a whole. The thematic map needed to be an accurate representation of the data.

Defining and refining of the themes was done in phase five. The aim of this phase was to work with the themes to deeper understand the essence of each theme. Braunus and Clarkes (2006) stress that a detailed analysis of each theme needs to be written in order to gain this understanding, paraphrasing is not enough. This made it easier to ensure that the connection to the objective of the thesis was strong enough. Individual themes were refined both in relation to the other themes but also in relation to the overall essence of the data and the aim of the thesis itself. Braunus and Clarkes (2006) describe that a good way to test if the themes are refined enough is to try to describe each theme, its content and aim, in a couple of sentences. If this is possible, no further work is needed regarding refinement (Braunus & Clarkes, 2006).

In the final phase, the summarization of the refined material was done, Braunus and Clarkes (2006) stress that the goal in this phase is to accurately present the story the data is telling in a compelling way that makes an argument in relation to the literature. The most important extracts were selected to support the themes as evidence, in preparation for the analytical comparison with the theories suggested in the literature.

3.4.3 Analysis of the quantitative data

The gathered quantitative data was extracted from the ranking exercise. The participants ranked 12 factors for SPC implementation from 1-12, this was done with eight people in total. Siegel (1957) declares nonparametric statistical tests as tests that do not make numerous or stringent assumption about the population. This leads to them often being called “ranking tests” or “order tests”. Siegel (1957) also states that many of these tests use their data as ranks of observations. Because the quantitative data used in the thesis project has an orderly characteristic a nonparametric statistical test was being used. This method was created by Friedman (1937) and is now well known as the Friedman’s two-way analysis of variance by ranks (Pereira, Afonso & Medeiros, 2015). The Friedman’s test determines whether the rank totals for each factor will differ significantly from the values which would be expected by chance (Pereira et al, 2015). This method stipulates a null hypothesis (H0): That there is no difference in the respondent's ranks and the (H1): There is a difference in the respondent's ranks. The null hypothesis is rejected at the alpha level of significance, which was chosen to 0.05. The test was carried out using the statistical software package Minitab. If the p-value shown in the Minitab analysis is below 0.05 the null hypothesis can be rejected. Further Minitab shows the median value of each factor, where the lowest indicates what was most important for the respondents.

3.4.4 Analysis of knowledge from processes and NV

Further information and knowledge collected in connection to the implementation of SPC were also analysed. This was not done according to an established scientific analysis method. Rather it was merely compared to the literature to understand how in the best manner SPC could be implemented into NV and their processes.

3.5 Validity and reliability of the data

During the semi-structured interviews, biased answers were avoided by avoiding asking follow-up questions that would lead the interviewee to give such an answer. Further, the usage of the master thesis students as moderators created an environment of neutrality, as the students were not part of the company as an employee. To further strengthen the trust, a promise on anonymous representation in the report was stated to all respondents, to ensure that all answers would be as genuine as possible. All respondents were chosen carefully in order to be representative of the title and segment that the respondents later would represent. Although some relationships between the moderators and the interviewees were initiated before the interview, which may have led to biased answers due to those relationships. The quantitative data was gathered inside the organisation of NV, meaning that the data will only be representative for the organisation. This data should not be considered representative for the battery production industry, but rather a footprint of the values of the employees at NV. To investigate if the gathered answers differ from answers that could be expected by chance a Friedman test was conducted. The questions asked during the interviews was always the same to ensure the reliability of the results, this was also true for the factor ranking test. To strengthen the reliability further the factor ranking test could have been redone by the same participants sometime after the initial test.

4. Field study

The field study shows the results of the data gathering. This data comes from observations at the organization as well as results from the semi-structured interviews. The information presented in sections 4.3 - 4.8 was gathered throughout the thesis project with informal interviews, observations and studies on internal documents at NV.

4.1 Interviews

The interviews that were conducted with all respondents lasted between 30-60 minutes. All interviews were then transcribed to create a solid foundation for later analysis. All the interviews were held face to face, due to the Covid-19 situation some of the interviews were postponed but could in later stages of the thesis project be finalized.

4.2 Quantitative data

Table 6, Results from ranking of factors

FACTOR A1 A2 A3 A4 A5 A6 A7 A8 AVERAGE SCORE MANAGEMENT COMMITMENT 1 1 5 12 3 1 1 2 3,429 PROCESS PRIORITIZATION AND DEFINITION 2 4 11 6 4 4 4 3 5 TEAMWORK 5 8 10 9 1 3 2 4 5,429 IDENTIFICATION OF CRITICAL QUALITY CHARACTERISTICS 6 5 1 8 6 5 7 5 5,429

USE OF PILOT STUDY

3 3 4 4 10 11 5 6 5,714 SPC TRAINING AND EDUCATION 10 2 2 2 7 7 10 1 5,714 DOCUMENTATION AND UPDATE OF KNOWLEDGE OF PROCESS 4 7 12 1 9 6 6 8 6,429 CULTURAL CHANGE 11 9 9 11 2 2 3 7 6,714 MEASUREMENT SYSTEM EVALUATION 9 6 7 7 5 10 8 10 7,429 USE OF SPC SOFTWARE 8 12 8 3 8 12 9 9 8,571 CONTROL CHARTS 7 10 3 10 12 9 12 11 9 USE OF SPC FACILITATORS 12 11 6 5 11 8 11 12 9,143

Figure 8, Friedman hypothesis test.

4.3 Process information and structure.

Information regarding the process was gathered throughout the field study. Initially, the thought of the field study was to implement SPC into the processes of Coating and Slurry mixing. Although complications in connection to Covid-19 hindered further progress regarding this. Plans for measurements and control charts for each process step can be seen in appendix 2. These results will not be used as they were initially intended but rather to further answer the current stated purpose. The process structure of LIB is unique and is a mix of continuous processes, assembly of parts and batch production. This unique setup of production will create a lot of cause and effect relationships in production. An example of that relationship is shown in the SIPOC diagram (Cannot be displayed due to confidentiality) where the slurry is the output from slurry mixing, and then becomes an input in coating. Further down in the process the output from coating (coated foil) will become input in later stages of cell assembly.

4.3.1 Measuring procedures

The measuring techniques that are set up in the factory can be categorized into three types and are showcased in table 6 below.

Table 7, Measuring procedures

MEASUREMENTS EXPLANATION TECHNIQUES TYPE INLINE AUTOMATIC

MEASURING

Inline automatic digital measurements which happen continuously.

Digital Continuous

INLINE OPERATOR MEASURING

Samples are taken from production and measured directly in the same environment as production occurs.

Operators (Manual) Samples

OFFLINE MEASURING Samples are taken in production and transferred to a lab where measurements are taken. Quality control – laboratory (Manual) Samples

4.3.2 Parameter characteristics

4.3.3 Cause and effect relationships

Cause and effect relationships have been developed to some degree with FMEAs and PFMEAs which described the risk of certain quality characteristics. Although the correlation between different quality characteristics has not been established further.

4.4 Evaluation of measurement systems

The measurements systems are currently being evaluated in the entire factory. Procedures such as Gauge R&R, linearity and bias evaluation. These activities are ongoing and not yet finished.

4.5 Quality parameter controls

Firstly, NV is currently not setting up control charts to check the variation and the statistical control of the processes. NV are however capturing data that would allow the set up for control charts. As described in section 4.1.1, NV are using three types of measuring techniques that capture data of the product and process. Control plans have also been set up for all processes in the LIB production. This include an initial plan for control charts setup and sampling strategy. Although it is not complete for all product or process parameters in none of the process steps. A full control plan cannot be displayed due to confidentiality, although the structure showed in Appendix 2 shows a general outline of what is included in the control plan.

4.6 Software

The values which are measured continuously are automatically saved and displayed using a software called Grafana. Grafana for NV acts to capture data in real-time, while also saving that data. This is then used for problem solving and traceability. A picture that visualizes this could be displayed here but could not be displayed due to confidentiality.

Other software's developed by Microsoft are also used, for example, Excel, Word and PowerPoint. Data is saved and stored using SharePoint, which is a document cloud service that allows people to share their documents with specific teams or persons. The values measured by operators and the QC-lab is still manually inserted into Excel.

4.7 Critical-to-quality characteristics

4.8 Handling of nonconformities

The handling of nonconformities is carried out using a software called IPA where all nonconformities are saved. All non-conformities are followed up by actions, where the persons in charge of a specific area will be notified. Actions are often 8D-projects, if the problem is considered larger, or other bettering actions.

5. Analysis

In the following chapter, the analysis of the field study is presented.

5.1 Thematic analysis

From the collected interview data, different patterns and trends could be observed and stated. These patterns were then translated into codes to able to group correlated codes into themes as seen in Figure 10. The thematic map is presented below. The orange shapes represent the codes and the green shapes a deeper level of underlying codes. The blue shapes represent the themes that categorise the codes. Solid lines connect the codes with the themes and the dotted lines show links between codes or links between themes.

5.1.1 Theme structure

A description of each theme and how they are connected is presented below

Communication

This theme was formed mainly from codes built by extracts linked to people who work close to the product, operators, and engineers in production. Their views on SPC implementation were heavily affected by how the day-to-day operation works. Involvement of the operators dominated the underlying aspects of this theme. This included both views on keeping the operation transparent to create a better understanding of the people working on the floor, but also to include operators in decisions. The importance of communication between operators, engineers and managers in production was also evident. The connection from this theme to organizational structure was clear. A lot of challenges with communication can be linked to the structure of the organization.

Resources

The lack or shortage of recourses was the foundation of this theme. Both managers and operators expressed in similar ways that the workload is heavy, and people have too much on their hands. The complexity of the production was also an important aspect of this theme, the complex process requires complex equipment. The equipment includes instruments that have been difficult to calibrate and set up properly for quality checks. This part of the theme is connected to some of the underlying codes in the theme initial planning and structuring, the planning of the production and the process design.

Initial planning and structuring

In this theme, the ambition of automation well thought out process design in relation to SPC became evident. The content of the theme was mainly the importance of great attention and focus in the early stages of an SPC implementation. Many of the experienced respondents stressed the importance of solid pilot projects to lay the foundation for future actions. The test facility NV labs are not as efficient and well organized as the ambition for NV Ett, the design of the process layout will be important in mass production.

Organizational structure

This was the most supported theme, it reflects the entire data set most of the themes. NV is a startup of great and rapidly growing size. The challenges regarding organizational structure seem to have roots in many aspects concerning SPC implementation. Some of the challenges with the size of the organization and a flat management style has already become a lot better, this concerns reporting structure and hierarchy approach. However, who is responsible for different task and processes and what each role means are important, especially when implementing new routines or systems on a large scale. This theme has a clear connection to the theme communication, a wellstructured organization is dependent on good communication. This theme is also connected to experience, like two other themes, training and education and critical parameters. How lessons learned from the test facility NV labs will work as valuable experience in NV Ett was frequently mentioned in all interviews and it is clear that the data points in that direction.

Training and education

This theme was shaped by extracts both from respondents working close to the product and respondents with management roles. The operator’s point of view was grounded in a desire for a better understanding of the whole picture. It was expressed in different ways that an operator would understand quality work better if the right education about correlations in the processes was given. If the operators understand what causes the quality parameters values to change, they would not only have a clearer incentive for quality focus but would also be able to act on alarming values. The other part of this theme concerns education in LIB production, it is complex with a lot of inputs. Several respondents described that it is difficult to implement new ways of working in processes you do not fully understand. This theme was also linked to Experience, the ambition is to learn from the operation at NV labs and take that knowledge to NV Ett.

Critical parameters

In this theme, which is connected to initial planning and structuring, The importance of parameter identification became evident. Moreover, the need for great attention for this early in the process of implementing SPC became clear. The same reasoning concerned the identification of specifications and critical-to-quality characteristics. This theme is also connected to experience. The identified quality parameters at NV labs will be the parameters considered at NV Ett when production starts. The content of this theme indicated that the ambition, in this case, is that the knowledge of critical parameters will be good enough to copy the control plan and implement it in NV Ett when going to mass production.

Experiences

A lot of the themes connect to this theme. In phase four of the thematic analysis, in the search for coherent patterns, the task described as level two concerns coherent patterns in relation to the entire data set. The content of this theme reflects the majority of the content of the data set. The essence of this theme seems to be the value of lessons learned, it can be found in many other themes or underlying aspects of the themes. However, the clearest connections are to the themes Critical

parameters, Training and education and Organizational structure.

5.1.2 Theme content relevance

The data strongly indicates that an early focus on identifying important characteristics of the LIB process is needed for a full-scale implementation of SPC. There is a lot of support in the literature for focusing on understanding the purpose and tools of SPC early on in the process of an implementation, which also seems to be the case at NV. However, the need for understanding the actual process of LIB production became evident when reviewing the data. Not too surprisingly, since the process is complex. The people working with the materials need to understand how their work affects other parts of the production. These views seem to be found in all levels of the organization. Considering the size of the planned production at NV Ett, this must be viewed as critical. The following quotes are examples of relevant view expressed on this matter during the interviews.

A3:

“We should let people learn the interfaces between processes, everybody should be aware of how process steps are connected”

A4:

“provide time for the operators to learn, because none of us are experts in any way.”

As NV matures the challenges regarding organizational structure will eventually decrease. It seems like a lot of progress already has been achieved on this matter. However, since communication appears to be both critical to getting the operation to run smoothly and a basic qualification for SPC implementation, roles and responsibilities need to be crystal clear when moving to mass production at NV Ett. A clear organizational structure is a necessary condition for efficient communication. Moreover, NV is growing fast and the organization has more than 60 nationalities represented among the employees. The communication needs to be efficient in order to implement SPC on a large scale. The structure of the organization is also important for out-of-control situations, clear roles and responsibilities is a condition for handling alarms in production properly, everyone needs to know who owns the different processes and is responsible for action in such situations. This is also connected to more streamlined communication, to know whom to contact in such situations is critical for a sophisticated quality system. The following quotes are examples of relevant view expressed on this matter during the interviews.

A1:

“Organization structure-wise, just in my opinion, we need to have a more clear responsibilities in each team”

“We have grey areas, when something happens in production it is unclear what team will handle the problem”

A3:

“We need to clarify responsibilities for each team, based on these responsibilities we can then structure the organization”

A4:

“communication, that’s the key. According to me. If you don’t involve the right people or the wrong people in the discussion. You need a broad view and you need a broad range of people

The different challenges related to resources was clear during this study, and a lot of it can be explained by the characteristics of the LIB process. The complexity of the production creates a need for complex instruments and a need for sophisticated process design. This will be of the utmost importance when considering SPC at NV Ett. Regarding quality checks, the challenges seem to have its roots in the difficulties of setting up, calibrating and operating the measuring equipment properly. The following quotes are examples of relevant view expressed on this matter during the interviews.