A strategy for improving reliability in assembly processes

(1)

Akademin för Innovation, Design och Teknik

A strategy for improving

reliability in assembly

processes

Bachelor’s thesis

15 hp

Product and process development

Tommy Dahlström

Rapport nr: 1

Handledare, Mälardalens högskola: Paulina Myrelid Examinator: San Giliyana

(2)

ABSTRACT

This report presents an explorative examination of the scientific literature regarding current methods utilized in mixed-model assembly lines and the challenges that are faced. Empirical data from a real mixed-model assembly line was collected and analyzed to explore the applica-tion of the identified methods and the problems that are faced in a realistic situaapplica-tion.

The report consists of a theoretic framework on the topics of assembly systems and topics rel-evant to managing the challenges. In addition a case study is conducted where an assembly system is observed and real production data is collected. Product assembly lines make up one of the final steps in the manufacturing process of a product. By implementing proven methods for process improvements highly effective assembly lines are realized. If an assembly line is experiencing reliability issues the probability of the company’s end-customers being directly affected is considerably increased.

Ensuring reliability in a mixed-model assembly line (MMAL) is an important task with an in-creasing level of difficulty given the current evolution of advanced technology. The resulting increase in product variety and complexity made possible increases the demands on the entire manufacturing process and the assembly line in particular. Added product variety and complex-ity poses an increased challenge from an automation perspective as well due to the limitations in flexibility inherited by industrial robots in comparison to a human operator. Many challenges in manufacturing and assembly processes stem from product development, increased variation between product models, and changes in production systems, due to the resulting increase in the complexity of the production processes (ElMaraghy & ElMaraghy, 2016).

Historically, a vast number of manufacturing companies have seen significant improvements in their product quality and productivity, leading to improved financial results as well as customer satisfaction by implementing process development methods such as Lean production, Six Sigma, or Lean Six Sigma.

Therefore, the purpose of this report is to explore assembly system reliability and complexity management.

The scope of this report is restricted to focus on reliability in assembly systems from a quality perspective. Suggestions for improvements are biased towards Lean Six Sigma methods and complexity management as described in current literature with the consequence of potentially valid methods and solutions being excluded. The theoretical framework of this report is based on selected books and scientific articles on the topics of complexity, assembly systems, process development, and human performance in the context of assembly systems.

The case study project is executed following the DMAIC-process as per the L6S methodology. Lean six sigma methodologies are applied in accordance with the selected literature with the objective of identifying a potential strategy for mitigating the quality deficiencies observed in the company´s assembly processes.

The research presented in this report follows the abductive reasoning concept (Säfsten & Gus-tavsson 2020). Abductive reasoning is initialized by a conclusion, which is followed up by constructing a theory based on literature, and finally analyzing empirical data to validate the constructed theory in the context given by the initial conclusion.

Additionally, a root cause analysis is conducted. The root cause analysis is initiated by con-structing an Ishikawa diagram to categorize and identify possible causes of the identified errors.

(3)

By systematically working through each of the categories the examined problem is look upon from different points of view, making for a comprehensive analysis. In this step the 5-why’s method is utilized to support the deductive reasoning process. The potential root causes are then utilized to construct a list of possible solutions and suggestions for improvement for the exam-ined problem. The improvement suggestions are based on deductive reasoning as well as doc-umented improvements from the literature.

(4)

FOREWORD

I would like to express my sincere appreciation to my devoted supervisor and all the other great people at the studied company for their generosity, time and support. Getting the opportunity to meet and work with you has been a privilege. You made this accomplishment possible, I am forever grateful.

Thank you Beata, Karl, Kenth, Ken, Caroline, Anders, Sarkat, Bogdan, Mathias, Vesa, Nico-laus, Tomas, Paulina, Kenth, Malin.

(5)

The introductory segment of this report describes the explored topics and related challenges. In addition, a problem is identified and utilized to derive the purpose of the report and formu-late the research questions that are explored. The segment is concluded by stating the delimi-tations of the project.

1.1 Background

Product assembly lines make up one of the final steps in the manufacturing process of a product. By implementing proven methods for process improvements highly effective assembly lines are achieved by many organizations. Since the assembly process is located far downstream in the manufacturing process it is sensitive to disturbances caused by any of the preceding seg-ments of the manufacturing process such as casting, welding, cutting, painting, pre-assembly, and even raw materials supply. If an assembly line is experiencing reliability issues the proba-bility of the company’s end-customers being directly affected is considerably increased. This makes the assembly line a high-stakes environment where the demands on productivity, quality, flexibility, and reliability are compounded.

Many challenges in manufacturing and assembly processes stem from product development, increased variation between product models, and changes in production systems, due to the resulting increase in the complexity of the production processes (ElMaraghy & ElMaraghy, 2016). According to Zhu et al. (2008) complexity has been defined as a measure of how product variety complicates the manufacturing process. Modern complex products commonly consist of a significant number of components in combination with the implementation of mechanical and electrical systems together with software and control systems (ElMaraghy et al., 2012). Advances in technology enable a seemingly ever-increasing progression of the products being manufactured around the globe. Consequently, the demands placed upon the manufacturing industry are changing as the market evolves. As shown by Zhu et al., (2008), increased product variety commonly impacts the performance of assembly processes negatively, resulting in qual-ity deficiencies as well as productivqual-ity loss. Given the current developments, manufacturing companies are forced to develop their business at a fast pace, a challenge of its own on top of the obstacles and uncertainties already mentioned.

However, as clarified by Nozhy and Badrous (2011) complex does not necessarily imply com-plicated. A system may contain a vast amount of different components making it complicated, and yet be predictable in operation. Conversely, a seemingly simple system with few compo-nents may be subject to outside factors making it complex without being complicated. As de-scribed by Wang et al., (2011, cited by Nozhy and Badrous, 2011) there are potential upsides to the presence of complexity in production systems as well as in product design. Strategically enhancing complexity instead of attempting to minimize it could possibly result in increased flexibility, enabling the company to conform to a larger variety of customer demands (Nozhy and Badrous, 2011). Such an approach requires effectively controlling the complexity through applying structural complexity management focused on the realization of competitive ad-vantages (Wang et al., 2011, cited by Nozhy and Badrous, 2011).

Historically, a vast amount of manufacturing companies has seen significant improvements in their product quality and productivity, leading to improved financial results as well as customer satisfaction by implementing process development methods such as Lean production, Six

(8)

2

Sigma, or Lean Six Sigma (L6S). Generally, the challenge of specifically managing increasing complexity is not prioritized within said methodologies in spite of displaying great potential. Despite the increasing degree of automation in modern assembly systems, ElMaraghy and ElMaraghy (2016) emphasize the continued importance of human-centered assembly systems due to increased flexibility and the challenge of fully automating complex processes.

1.2 Problem statement

Ensuring reliability in a mixed-model assembly line (MMAL) is an important task with an in-creasing level of difficulty given the evolution of advanced technology in the current time. The resulting increase in product variety and complexity increases the demands on the entire man-ufacturing process and the assembly line in particular. Added product variety and complexity poses an increased challenge from an automation perspective due to the limitations in flexibility by industrial robots in comparison to a human operator. To secure that high quality demands are met, additional quality control processes may be introduced within the manufacturing sys-tem, further increasing the complexity of a given station (Alkan et al. 2018). A significant amount of the newly emerging challenges in assembly systems are in many cases only manage-able by human operators, thus potentially increasing the workload to an unsustainmanage-able degree. Therefore, the purpose of this report is to explore assembly system reliability and complexity management.

1.3 Research questions

How can Lean Six Sigma methods complement complexity management in assembly pro-cesses?

How can assembly system reliability be improved in Lean Six Sigma companies?

1.4 Delimitations

The scope of this report is restricted to focus on reliability in assembly systems from a quality perspective. Suggestions for improvements are biased towards Lean Six Sigma methods and complexity management as described in current literature with the consequence of potentially valid methods and solutions being excluded. An additional limitation is the lack of participation in the studied company’s continuous quality improvement operations due to restrictions in ac-cordance with the local Covid-19 mitigation protocols. Finally, to ensure compliance with cor-porate secrecy policies some company specific information regarding production volumes and disturbance frequencies has been excluded from this report.

(9)

3 2 METHODOLOGY

The following segment describes the methods that was implemented when conducting the re-search presented in this report.

2.1 Research methodology

This report consists of a theoretic framework on the topics of assembly systems and topics relevant to managing the challenges that are present for an assembly system. In addition a case study is conducted where an assembly system is observed and real production data is collected. The research presented in this report follows the abductive reasoning concept (Säfsten & Gus-tavsson 2020). Abductive reasoning is initialized by a conclusion, which is followed up by constructing a theory based on literature, and finally analyzing empirical data to validate the constructed theory in the context given by the initial conclusion. As described by Säfsten and Gustavsson (2020) the concept of triangulation is implemented by utilizing several different databases for data collection to ensure and improve the validity of the theoretical framework presented in this report.

Peer-reviewed literature is prioritized to ensure that the presented research maintains a high level of quality and validity. The choice of databases for data collection is based on the require-ments for access to peer-reviewed articles that are relevant to the explored topics.

In addition to the chosen databases Scopus, IIEEE Xplore, and ScienceDirect, secondary sources located by utilization of the snowballing technique stand for a major contribution to the presented theoretical framework. The specific application of the snowballing technique in this report refers to further exploring referenced sources from the collected literature, thus allowing for a deeper understanding and more detailed descriptions of the explored topics.

The studied company’s chosen philosophy for process improvement is the L6S-methodology. Given that the company is operating according to a well implemented L6S-methodology, the project follows the same concept to ensure that the suggested strategy is appropriate.

2.2 Data collection

The initial conclusion is acquired from the studied company’s observation regarding assembly related quality defects. Data collection for theory construction is performed by utilizing scien-tific data bases and selected books relevant to the topics. The empirical data originates from the conduction of an exploratory case study which contributes to creating a broader understanding of the topic and its means of expression in actual real situations (Säfsten & Gustavsson, 2020). To achieve the objective of formulating a compatible strategy for improving reliability in the company’s assembly processes, literature on the topic of lean six sigma is explored and included in the theoretic framework. In preparation for the project literature on the topic of assembly process improvement was explored where connections between complexity and the occurrence of quality and productivity issues in assembly systems. Additionally, interest in the topic of the effect of human performance in assembly systems was expressed by the company during a meeting early in the project.

(10)

4

The activities performed within the case study includes one semi-structured interview with the production planner of the assembly line chosen for the project. Additionally, observations are made in the company´s assembly processes in addition to collecting and analyzing the com-pany´s production data. The case study project is initiated by a comprehensive data collection and data analysis with the aim of achieving a holistic overview of the current state.

The focus of the interview was planned variation in the assembly line and the factors that are taken into consideration in production planning. The objective of the interview is to identify possible causes of increased complexity based on the literature presented in the theoretical framework in this report. The interview duration was approximately 20 minutes and the semi-structured format was chosen to ensure relevance for the topic of complexity in the assembly system.

Observations in the assembly processes was made under the guidance of the assembly line’s production manager with the purpose of increased understanding of the current state, the as-sembly line’s layout, the methods utilized for ensuring quality and productivity, the level of automation, and the nature of the tasks performed in the assembly line was observed. In addi-tion, a process map is created for illustrations purposes, leading to improved understanding and ensuring validity of the conclusions made in the initial data analysis.

Initial conclusions from the data collection and analysis are verified through observations and communication with the company’s operations quality engineer during in-person- and online meetings in addition to email communication. Following the initial analysis of the data through categorizing, and visualizing the data by utilizing graphs in Excel, appropriate statistical infer-ence methods are used to verify the conclusions. The statistical tests are performed to evaluate the probability of the observations made in the graphs are due to natural variability.

2.3 Data analysis

The collected production data originates from the assembly line’s disturbance log, the internal production audit’s error record, and the errors observed by the testing department. All collected error data was acquired in the form of Excel work sheets. Following the sorting and categori-zation of the collected data an initial analysis with focus on observable pattern recognition and recurring errors is conducted.

Lean six sigma methodologies such as Pareto charts, histograms, the 5-why’s method, Ishikawa diagrams, and statistical inference are applied in accordance with the selected literature with the objective of identifying a potential strategy for mitigating the quality deficiencies observed in the company´s assembly processes.

Following the completed data collection, a qualitative data analysis is performed following the iterative three sub-step method described by Säfsten and Gustavsson (2020). Initial data display is executed through the implementation of Pareto charts, the data reduction phase includes sort-ing and categorizsort-ing data in preparation for the final phase consistsort-ing of maksort-ing conclusions and verification (Säfsten & Gustavsson, 2020). For the categorization of the data the company’s list of defined error categories is utilized as a guiding tool for increased consistency.

(11)

5

Next, a root cause analysis is conducted. The root cause analysis (RCA) is initiated by con-structing an Ishikawa diagram to categorize and identify possible causes of the identified errors. This categorization serves as a tool for breaking the problem down into smaller components for a more focused analysis. Traditionally the Ishikawa diagram is made up of six categories; man-agement, man, method, measurement, machine, and material. By systematically working through each of the categories the examined problem is look upon from six different points of view, making for a comprehensive analysis. In this step the 5-why’s method is utilized to sup-port the deductive reasoning process. Identifying a broader range of possible causes of the prob-lem in question increases the likelihood of locating the true root-cause is magnified. The poten-tial root-causes are then utilized to construct a list of possible solutions and suggestions for improvement for the examined problem. The improvement suggestions are based on deductive reasoning as well as documented improvements from the literature.

2.4 DMAIC

The application of the DMAIC-process in this report is utilized to enable a structured examina-tion and analysis of the explored topic. The Define-phase is focused on problem formulaexamina-tion based on communication with the studied company in conjunction with exploring relevant lit-erature. The Measure-phase includes data collection at the studied company and constructing the theoretical framework. In the Analyze-phase the collected data is visualized, consolidated, and analyzed to enable comparisons between theory and empirical data. The Improve-phase consists of identifying similarities and deviations between the empirical data and the theoretical framework which is the utilized to construct science-based suggestions for improvement. In the concluding Control-phase a strategy for improved reliability in assembly systems is formulated.

(12)

6 3 THEORETICAL FRAMEWORK

The theoretical framework of this report is based on selected books and scientific articles on the topics of complexity, assembly systems, process development, and human performance in the context of assembly systems. The objective of the theoretical framework is to consolidate relevant literature to support empirical data analysis and enable a discussion on the topic of assembly system reliability in the presence of complexity and its potential benefits, drawbacks, and consequences.

3.1 Assembly systems

This segments introduces the concept of an assembly system, the challenges that are faced, and some common methods for overcoming said challenges.

3.1.1 Definition of assembly systems

An assembly line is a manufacturing process consisting of sequentially attaching individual components to form a unified finished product (Thomopoulos, 2014). For maximum efficiency the assembly time is divided as evenly as possible between each sequential work station. As described by ElMaraghy and ElMaraghy (2016), an assembly system is made up of assembly work stations, inventory buffers, and transportation systems. Commonly utilized key perfor-mance indicators (KPI) for assembly systems are the traditional metrics of cost, quality, lead time, productivity, and overall equipment efficiency (Miqueo, et al. 2020).

3.1.2 Mixed-model assembly lines

There are various different types of assembly systems such as; single-model assembly, mixed-model assembly (MMAL), and batch assembly (Thomopoulos, 2014). As their respective names implies, single-model assembly lines operate by producing only one model on the line whereas the batch assembly system will handle different models but with a pre-determined quantity of each model at a time. The common practice of handling several different models in the same assembly line without changeovers is referred to as mixed-model assembly lines (ElMaraghy & ElMaraghy 2016).

According to Zhu, et al. (2008), mixed-model assembly lines contribute to the ability to handle increased variety, most commonly in the form of a flow line layout. The method is widely used, especially in automobile assembly (Zhu, et al., 2008). Thomopoulos (2014) describes how MMALs can be implemented for accumulating inventory or in a make-to-order fashion. When used for inventory re-stocking purposes, the assembly line is producing a mix of different mod-els, but each individual unit of a given model is identical.

Conversely, implementing a MMAL for make-to-order configuration generally results in a sig-nificant increase in variation between each individual unit due to the specifications varying for different customers. Because of increasing product diversity resulting in shortened product life cycles and reduced production volumes for a given product, the challenge of maintaining a sufficient level of flexibility while quality and productivity is ensured is a central component to assembly system performance (ElMaraghy & ElMaraghy 2016).

(13)

7

In a given assembly work station the operator may be required to perform a multitude of differ-ent choices apart from just picking the appropriate compondiffer-ents for the product in question. These choices can include selecting the appropriate fixture and settings, tool selection and its accessories such as bit-type, and also making sure that the right procedure is performed accord-ing to the specific product (Zhu, et al. 2008)

According to ElMaraghy and ElMaraghy (2012) design-for-assembly is a method that has been proven to be successful in decreasing assembly costs by taking the assembly of a product into consideration already on the manufacturing stage. Simplifying the assembly process by imple-menting common tools and fixtures for several different product variations is a method that can reduce the complexity of an assembly system.

According to Miqueo et al. (2020) MMALs are a requirement for handling mass customization in an economically sustainable way due to its astonishing flexibility. Optimization of MMALs is commonly performed by ensuring the most efficient solution for assembly work station se-quencing and balancing to enable a constant flow without unnecessary product queueing and waiting time due to differences in lead time between stations in the assembly line. In addition, analyzing the complexity of material-handling and manufacturing complexity can be useful for balancing the workload in the assembly line (Miqueo, et al. 2020)

Reliability of an assembly system can be defined as the consistency of which the desired output is achieved from a given input. From a quality perspective this would refer to the number of defects in relation to the production volume. According to Quick (2019) the long-term objective should ideally be 0 defects. According to the findings by Nozhy and Badrous (2011), managing the complexity of the assembly system is a proven way of improving its efficiency and cost effectiveness.

3.2 Process development

This segment presents common philosophies for improving business processes and the methods and tools associated with each philosophy.

3.2.1 Quality

According to Bergman and Klefsjö (2012) quality is a products ability to fulfil, and ideally surpass the customer’s needs and expectations. The essence of this definition of quality is the realization that to obtain loyal and satisfied customers merely meeting the customer demand may not be good enough Sometimes, the customer may need to be pleasantly surprised due to the acquired product exceeding the expectations.

3.2.2 Quality assurance

Quality assurance refers to activities that are carried out before the actual production takes place with the purpose of avoiding defects by mitigating errors beforehand (Bergman & Klefsjö 2012).

Opposed to quality assurance is the concept of quality control where activities with the purpose of detecting defective products are performed after the production process has already been

(14)

8

completed (Bergman & Klefsjö 2012). Quality assurance in assembly processes include meth-ods such as tolerance design and statistical analysis (ElMaraghy & ElMaraghy 2016).

In addition to the costs for repairing or replacing the product, complaint handling, and additional shipping, defective products being delivered to the customers can considerably damage the cus-tomers’ level of satisfaction. Potentially leading to further detrimental effects such as loss of future sales, customer compensation costs, and damaged reputation. Two common philosophies for managing as well as improving quality are Six Sigma and Total Quality Management (Kra-jewski, et al. 2019).

3.2.3 Total quality management (TQM)

As described by Krajewski, et al. (2019). There are three central elements to the philosophy of TQM; customer satisfaction, employee involvement, and continuous improvements. The three elements serve the purpose of achieving a high level of quality and process performance. In TQM, both the satisfaction of internal and external customers is prioritized by working with five customer related dimensions of quality. These dimensions are, product specification fulfil-ment, perceived value in relation to cost, functionality, customer support, and perceived service quality (Krajewski, et al. 2019).

Employee involvement facilitates change in the company’s business culture by clearly defining the internal customer of each employee in the organization to illustrate how each employee contributes to ultimately fulfilling the needs by the external customers.

Working with the concept of internal customers improves cross-functional coordination within the organization and holds each employee responsible for ensuring quality throughout the or-ganization with the concept of “quality at the source”. Also, the addition of self-managed teams further empowers the employees and encourage increased involvement.

Continuous improvements refer to making marginal modifications wherever possible with the purpose of them adding up to a significant improvement over time. Utilizing the increased em-ployee involvement through the engagement of each emem-ployee in the problem solving process related to their individual work tasks is a powerful method of achieving comprehensive im-provements within the organization.

Commonly used problem solving tools of TQM are statistical process control (SPC) and the plan-do-check-act (PDCA) cycle, the objective is often to achieve a reduction of non-value adding activities.

Figure 1, the PDCA-cycle, inspired by Krajewski et al. (2019)

Act Plan

Do Check

(15)

9 3.2.4 Total Quality Control (TQC)

According to Process Improvement Japan (2021) TQC is a continuous process with the objec-tive of ensuring, improving, and standardizing a high level of quality. The TQC process follows the PDCA-cycle by defining goals and standard procedures in the Plan-phase. The Do-phase consists of employee training and implementation of the standardized procedures. In the Check-phase inspection of the products is performed. The concluding Action-Check-phase includes taking appropriate measures based in the outcome of the preceding quality inspection.

3.2.5 Six sigma

Krajeweski, et al. (2019) describes Six Sigma as a philosophy with the purpose of evaluating and improving process performance. Six Sigma is based upon fulfilling the customer’s expec-tations by implementing data, and statistical analysis to continually improve the business and manufacturing processes with the goal of achieving the desired output with low variability (Kra-jewski, et al., 2019). Six Sigma originates from Motorola Corporation’s adaptation of the con-cept of TQM in an attempt to counteract the quality issues that was experienced at the time (Arnheiter & Maleyeff 2005).

The aim of Six Sigma is to achieve a level of near perfection with a maximum of 3, 4 defects per million opportunities (Taghizadegan, 2013).

Six Sigma methodologies has been implemented with great success in several different manu-facturing industries, especially manufacturers of complex products have found the concept very helpful in consistently improving quality and reliability. Some examples of these industries are consumer electronics, automobile, aerospace, and the computer industry (Arnheiter & Maleyeff 2005). Also, the healthcare industry has benefitted from Six Sigma implementation despite the lack of particularly complex products due to the inherent risk of a defective product causing physical harm, thus raising importance of a very high level of quality (Arnheiter & Maleyeff 2005).

A central concept of Six Sigma is the 5-step improvement model Define-Measure-Analyze-Improve-Control (DMAIC)-process that is utilized to systematically improve the business pro-cesses.

Implementation of the DMAIC-process starts with defining the problem and its scope in the Define-phase, making sure that the problem is specific and not too broad to enable measuring and management of the improvement process.

In the Measure-phase specific performance metrics are identified and data is collected to eval-uate the current state in the process and enable the following data analysis.

In the Analyzing-phase, the collected data is sorted, categorized, structured, and visualized to allow for an accurate root cause analysis. Commonly implemented tools in the analyzing pro-cess are Pareto charts, histograms, cause-and-effect diagrams, and statistical inference methods. In the Improve-phase possible ideas for improvement are identified and evaluated based on the context of the current situation to ensure applicability and that the benefits outweigh the costs of implementing the suggested improvement. The concluding Control-phase serves the purpose of making sure that the implemented improvements are maintained so that the new and im-proved state is kept as a new standard. (Taghizadegan, 2006; Krajewski, et al. 2019).

(16)

10

Figure 2, the DMAIC-process, inspired by Krajewski, et al. (2019)

3.2.6 Lean production

According to Liker (2004), lean production is a comprehensive application of the Toyota Pro-duction System within a company. Just-in-time proPro-duction is a central concept being pioneered within Lean production (Arnheiter & Maleyeff 2005).

Process improvement in lean production is focused on reducing the total lead time by improving process flow and eliminating non-value adding activities (Liker 2004). Non-value adding ac-tivities are referred to as waste and has been divided into the eight different categories of; over production, over processing, waiting time, transports, unnecessary movement, inventory, de-fects, and underutilization of employees (Krajewski, et al., 2019). Waste reduction is achieved by identifying the value adding activities throughout the production process in a value stream map, implementing a pull based constant flow production system using the concept of Heijunka to level out the production load (Krajewski, et al., 2019). Consequently, the cycle times are synchronized to match the customer demand, also referred to as Takt time (Krajewski, et al., 2019). Finally the company keeps striving for excellence through continuous improvements referred to as Kaizen (Liker 2004). Other frequently utilized concepts of Lean production are SMED, 5S, standardization, total productive maintenance, and Poka-yoke (Krajewski, et al., 2019; Arnheiter & Maleyeff 2005). The Lean production methodologies are commonly ob-served to enable drastic reductions in manufacturing lead times, allowing the manufacturer to produce smaller batches made specifically to customer order (Arnheiter & Maleyeff 2005).

3.2.7 Lean Six Sigma (L6S)

The business strategy of Lean Six Sigma provides enhanced results in terms of both financial return as well as customer satisfaction by achieving increased process performance (Snee, 2010). According to Arnheiter and Maleyeff (2005), companies that choose to operate along the lines of either Lean production or Six Sigma methodology are exposed to a risk of reaching a point of diminishing returns. However, combining the two concepts is a viable solution, the data driven nature of Six Sigma in combination with the streamlining focused Lean production concept have potential for further enhancing quality and productivity measures while avoiding stagnation (Arnheiter & Maleyeff 2005).

(17)

11

According to Taghizadegan (2006), L6S is a customer focused, data driven approach to process improvement. In addition to improving manufacturing and business processes, L6S emphasizes improvements within the scope of the employee’s business management capabilities to also improve financial goal fulfilment (Taghizadegan, 2006). Process improvement in L6S is fo-cused on quality assurance by detecting root causes early to prevent errors from occurring (Taghizadegan, 2006).

Snee (2010) recommends two different strategies for approaching improvements and problem solving based on the L6S methodology: the top-down approach, or the bottom-up approach. The top down approach is initialized with a business goal as a starting point and utilizes goal breakdown to identify the required process improvements, whereas a bottom-up approach starts off by identifying process problems that are decreasing the process performance. As further described by Snee (2010) Six Sigma methods are commonly implemented for complex prob-lems with no evident solution.

Meanwhile, Lean production methods are more appropriate for eliminating wasteful activities. In the analysis of opportunities for improvement by Snee (2010), information and material flow between the process steps are identified as a common source of performance deficiencies. While such a situation may seem to call for exclusive implementation of Lean methods, the most ef-fective strategy is to simultaneously apply Lean and Six Sigma methods for a more comprehen-sive solution to the identified problem. This strategy addresses the possibility of an observed flow related problem actually originating from poor process performance or vice versa. Taghi-zadegan (2013) describes Lean Six Sigma (L6S) as a proactive method for maximizing produc-tivity and achieving the right quality, at the right cost, at the right time.

Snee (2010) identify two primary pitfalls of deploying L6S, management systems for the over-arching program, and management of the improvement projects. The former includes insuffi-cient leadership support, failure to utilize in-house expertise, and focusing too much on training instead of improvement. The latter pitfall includes projects being insufficiently connected to business goals, assigning the wrong people for the projects, too many people in the project group, and the project lasting tool long combined with the individuals in the group not being given sufficient time to work on the project. Ideally, improvement projects should not last longer than six months and the project groups should consist of 4-6 people given sufficient time to complete the project and ensure effectiveness.

Snee (2010) further suggests that training is combined with the improvement projects for en-gagement from the participants and more cost-effective training. Having the training directly result in real improvements also adds more value in the short-term compared to separating training activities from the improvement projects.

Arnheiter and Maleyeff (2005) presents a strategy for how an organization can capitalize on the combined benefits of Lean and Six Sigma. By incorporating an overarching philosophy with the objective of ensuring global optimization across the organization and formulating a process for decision-making based on generating value for the customers, the essence of Lean is cap-tured. In addition, including the Six Sigma concepts of ensuring that the decisions are based on data-driven methods to ensure long-term accuracy, which when paired with the introduction of a structured training plan for all employees as first seen in TQM also improves employee en-gagement. Finally, implementing methods for minimizing variation in the business processes improves quality and reliability. Quick (2019) describes several strategies for applying the L6S concepts in different situations. For mitigation of quality defects a traditional PDCA-inspired systematic problem solving approach is recommended. In the event of excessive variability within a process Quick (2019) advises for a data driven statistical process control approach.

(18)

12 3.2.8 Automation

The tasks performed in an assembly line are often of a monotonous nature, and thus fatigue-inducing from a physical- as well as mental perspective (Thomopoulos, 2014). These traits make assembly lines viable for being automated by implementing industrial robots and convey-ors in the system. In the short term, the investment required for automating an assembly process or even an entire assembly line could seem deterrent.

However, choosing to go the route of automation could prove to be a superior alternative in the long-term despite the costs related to hardware and software purchase, installation, and contin-uous maintenance due to the improvements in quality, productivity and reliability.

According to Fast-Berglund and Mattsson (2017), automation can be divided into two separate categories: physical automation and cognitive automation. Physical automation refers to the assembly process operator being fully or partially replaced by an industrial robot or other tech-nology-based resources in the completion of a given task. Physical automation ranges from no automation in a fully manual process, via semi-automated processes where human operators are still involved to a degree, and finally fully automated processes where all activities are per-formed without the need for human intervention.

Cognitive automation is a software based method for automating decision making within an assembly process with the objective of eliminating assembly related errors while improving the operator’s work environment (Fast-Berglund, et al., 2013). Targeted activities of cognitive au-tomation include situational assessment and monitoring in addition to error management. Ac-cording to Fast-Berglund and Stahre (2013) there is a need for cognitive automation to achieve assembly system that are both sustainable and flexible. Some examples of cognitive automation strategies are placing components for assembly in the order they are to be used, utilizing mobile information carriers to ensure accessibility of work instructions, implementing product fixtures for more convenient assembly, and pick-by-light systems that aids the operator in choosing the next component or procedure without the need for additional reasoning.

Identifying the best suited automation solution can be a challenge due to the trade-offs between the four factors of; production volume, batch sizes, product diversity, and flexibility. It is com-mon to design the automation solution so that one of the previously mentioned factors are op-timized (Fast-Berglund & Mattsson 2017). However, despite the increasing degree of automa-tion in modern assembly systems, ElMaraghy and ElMaraghy (2016) emphasize the continued importance of human-centered assembly systems due to increased flexibility and the difficult challenge of fully automating complex processes.

3.2.9 Industry 4.0

Industry 4.0 is commonly referred to as the fourth industrial revolution. The driving forces behind Industry 4.0 is the rapid development of new technology enabling innovations like Ar-tificial intelligence, Internet of Things (IoT), Big data analytics, Augmented reality/Virtual re-ality, and Additive manufacturing. Miqueo, et al. (2020) identifies Big data, IoT, Cyber Physi-cal systems, vision systems, and AR as particularly useful in providing additional support for assembly operators.

(19)

13

According to Miqueo, et al. (2020) the Lean production methodology has positive synergy ef-fects with Industry 4.0 due to the ability of new technology to enhance Lean methods. Imple-mentation of Industry 4.0 technology has the potential for allowing a shift in the main function of assembly operators to favor decision making and information management over engaging directly in traditional manual labor. As commonly found in successful Lean implementing or-ganizations, a cross-trained workforce provides the necessary flexibility to enable the shift to-wards the more dynamic operation made possible by Industry 4.0 technology.

ElMaraghy and ElMaraghy (2016) describes AR-technology as a cost effective and valid tool for assembly system utilization. The technology’s ability to enable the user to view the real environment with the addition of virtual objects is useful both in training of new employees as well as in the execution of difficult work tasks where additional guidance can improve the end result.

3.3 Complexity

This segment describes the concept of complexity, its effects on assembly systems and opera-tors, and methods for measuring, managing, and reducing complexity.

3.3.1 Definition of complexity

According to Samy and ElMaraghy (2012) there is no universal definition for complexity. How-ever, it is stated that complexity depends on the two dimensions of time and uncertainty. The uncertainty stems from lack of information and/or the unpredictability of the process in question while the time factor is based on decisions and time-dependent operations.

Modern engineering systems are increasing in complexity, particularly larger systems can reach a significant level of complexity compared to modular systems.

An observed benefit of a higher level of complexity within a system is the ability to perform an increased amount of different functions, however the increased versatility comes with the trade-offs in the form of negative effects on quality and productivity seen in both operations as well as management (Samy & ElMaraghy 2012).

Hu, et al. (2008) defines complexity as the average uncertainty in a random process of handling product variety. Complexity is further described mathematically as an entropy function referred to as 𝐻 .

𝐻 𝑝 , 𝑝 , … . , 𝑝 = −𝐶 𝑝 log 𝑝 Complexity entropy function as defined by Hu, et al. (2008).

Rodriguez-Toro et al. (2002) highlights the importance of accurately defining complexity as a necessary prerequisite for effective complexity management which is an important aspect in ensuring assembly process performance and desired quality of the end product. Nozhy and Badrous (2011) divides system complexity into two parts, static structure and dynamic com-plexity. Product variety, system structure, and relationships between segments of the system makes up the static structure. Dynamic complexity refers to unpredictability of the systems operational behavior over time. A system can display a combination of lower respectively

(20)

14

higher levels of both static and dynamic complexity; where a system with a high level of both static and dynamic complexity is considered extremely complex.

3.3.2 Consequences of increased complexity

According to Zhu, et al. (2008), increased product variety has been observed to have a signifi-cant negative impact on the performance regarding productivity as well as quality of mixed-model assembly lines in the automobile industry. Nozhy and Badrous (2011) describes the as-sembly system design as well as operator performance in the presence of increased product variety as possible root causes of said impact. Similarly, according to findings by ElMaraghy and ElMaraghy (2016), the added complexity caused by increases in product variety has been shown to increase production costs and is considered a major industrial challenge in current time.

According to findings by Falck, et al. (2012), assembly complexity displayed a significant cor-relation with increased assembly lead time and costs. In addition, a significant cor-relationship be-tween complexity and ergonomics in assembly processes was observed. An increased ergo-nomic load could possibly increase the perceived complexity in an assembly process. Falck, et al. (2012) further describes how an increased level of physical load in manual assembly pro-cesses results in increased quality errors, implying a clear relationship between ergonomics and assembly related errors that affect the quality of the products produced.

Further findings by Falck, et al. (2012) conclude that complexity of assembly processes display a significant correlation with increased costs but not with the occurrence of errors, as opposed to ergonomics that did correlate significantly with error occurrence. Even though the observed relationship between complexity and ergonomics was not possible to specify in detail, assembly lead time did not show signs of being a factor in failure occurrence.

3.3.3 Complexity management

Nozhy and Badrous (2011) describes three different approaches for managing complexity; com-plexity reduction through simplifying structures, comcom-plexity prevention by implementing meth-ods for assessing complexity in development phases, and complexity control for mitigating the remaining complexity that cannot be reduced. In addition, Nozhy and Badrous (2011) describes the process of product development as a segment where complexity management is crucial. Reducing complexity has been proven to lead to reduced direct costs as well as indirect costs. According to Samy and ElMaraghy (2012) acknowledging the compromises being made re-garding performance, cost, and complexity in the design process of an assembly system is of utmost priority. Due to the effects of increased complexity on the aspects of assembly system performance, product quality, and process reliability, mitigating complexity is of significant economic importance (Samy & ElMaraghy 2012).

However, according to ElMaraghy and ElMaraghy (2012) there are trade-offs to consider re-garding specifically product complexity. The objective should not be to reduce product com-plexity at all costs since the comcom-plexity of a product may contribute significantly to the value of the product from the customers’ perspective. Offering a customizable product with a rela-tively broad range of functionality adds complexity while increasing its relevance for an in-creased customer base.

(21)

15

Mattsson, et al. (2014) describes how managing complexity can be done by either removing, simplifying, avoiding, or preventing complexity. However, removing complexity in final as-sembly it is often not a feasible option. In such situations, trying to simplify the process to achieve the greatest possible reduction of the complexity is a viable option. If possible, complex assembly solutions should be avoided to avoid long assembly times and increased costs (Falck, et al. 2012).

Many of the tasks performed in manual assembly processes are commonly difficult to control and monitor for quality deficiencies, often an error is not discovered until a later stage of the assembly system. An exception is the assembly of bolted joint connections where controlling the tightening torque and turning angle of the bolts are enabled by the tools used in the assembly work station (Fast-Berglund, et al. 2013).

According to the findings by Mattsson, et al. (2016b), variety of the assigned tasks was deter-mined to be the most common cause of complexity in assembly processes. More specifically, most contributions to complexity were connected to ergonomics and to managing different product variations (Mattsson, et al., 2016b).

As demonstrated by Zhu, et al. (2008), a linear relationships was identified between assembly lead time for an examined wiring assembly task and the information content of the task. An increase in information content was also observed to lead to an increase in the frequency of error occurrence.

In regards of assembly system complexity, the first station in the assembly line potentially has the maximal influence on the others; while the last station has no influence on the complexity of the other stations). As such, the complexity of the assembly system has the potential to prop-agate as it travels downstream through the assembly line. As a result, operations at the later assembly stations can experience increased complexity caused by previous stations in the as-sembly line.

A strategy for ensuring performance and finding root causes presented by Zhu, et al. (2008) is to identify the stations within an assembly line where the incoming complexity is at a relatively high level and apply measures for mitigating errors at those stations. Further, by identifying the stations with a higher level of outgoing complexity the root cause can be located to guide efforts in mitigating complexity.

3.3.4 Measuring complexity

A convenient method for evaluating the complexity of an assembly station using objective data is to add up the number of different product variants and the number of components being handled within the work station (Mattsson, et al. 2012).

Operator choice complexity (OCC) is a measure of complexity based on the recurring choices that the operator of an assembly process is required to make, choices include both product va-riety and process information such as work instructions (Zhu, et al., 2008). OCC is an indirect measure of human performance in making choices, such as selecting parts, tools, fixtures, and assembly procedures in a mixed-model, manual assembly environment (Zhu et al., 2007). Fast-Berglund, et al. (2013) observed a positive correlation between OCC and the occurrence of assembly errors. Meaning that if OCC is increased, assembly error occurrence is also pro-portionally increased.

(22)

16

The choice process defined by Zhu, et al. (2008) is made up of a choice sequence with respect to time, enabling mathematical modelling of the OCC as a discrete time, discrete state stochastic process. According to findings by Zhu, et al. (2007), the probability of an operator to make a mistake is decreasing the more certain the operator is about the choices in the coming tasks, as a result, the perceived complexity is reduced. To aid an operator in being more certain about the choices that has to be made cognitive automation can be implemented (Fast-Berglund & Stahre 2013)

CompleXity Index (CXI) is a questionnaire based tool for measuring complexity in assembly systems that enables quantifying the level of complexity within an assembly work station (Mattsson, et al. 2016b). According to Mattsson, et al. (2016b), CXI can provide opportunities to identify potential improvement areas as well as bottlenecks which is a helpful attribute in complexity management. A seemingly convenient strategy for evaluating complexity recom-mended by Mattsson, et al. (2012) is to start with a quantitative analysis due to its measurability, and then use the results to identify where to implement a deeper analysis with CXI. When com-paring the CXI-results with objective data based on product variety and quantity of components used to validate the method, Mattsson, et al. (2012) observed a correlation between the investi-gated variables.

3.3.5 Complexity countermeasures

Mattsson, et al. (2014) presents a possible strategy for supporting the operator by managing the complexity of an assembly process is to ensure that the information and instructions provided are effectively simplified to minimize the need for strenuous thinking and reasoning, thus sup-porting the preferred intuitive thinking processes of the operators. The simplification of the information in the assembly process for improved understanding and ease of interpretation was shown to result in increased performance due to a reduction of assembly related errors. Matts-son, et al. (2016b) identified information delivery and ergonomics as common areas with po-tential for improvement based on feedback from CXI questionnaires.

According to further findings by Mattsson, et al. (2014) improvement was successfully achieved in an assembly process by ensuring that the components were consistently placed in a logical, standardized sequence, while avoiding similar parts being placed too close to one another. Additional cognitive support could be provided by implementing realistic pictures and highlighting differences between similar components. To cognitively support the operators of the assembly process, the implementation of signs can enable rule-based behavior which could results in quicker processing with a reduced demand for increased effort and time spent search-ing for information.

In a later study, Mattsson (2018) describes how it is uncommon for assembly work stations to be designed with the intention to support the operator, in most cases the work station is designed with an emphasis on technology. Several benefits can be obtained by improving the cognitive- and physical support the operator with increased usability such as improvements in productiv-ity, product qualproductiv-ity, employee satisfaction, and enthusiasm. Fast-Berglund, et al. (2013) also recognizes a need for increased cognitive support in assembly work stations where errors due to mistakes by the operator like incorrectly installed components and missing components are occurring.

(23)

17

Key components for enhancing cognitive support developed by Mattsson, et al. (2014) are uti-lizing a consistent information layout and sequence, making differences between product vari-ations clearly visible using highlights, and implementing realistic illustrvari-ations. If possible, fur-ther cognitive support can be provided by reducing the information quantity and displaying the three next oncoming product types for the operators. ElMaraghy and ElMaraghy (2016) also identifies AR-technology as a valuable, cost efficient tool in assembly system design. Keller et al. (2020) conclude that assembly processes involving a high degree of complexity can be sta-bilized through the introduction of digital aids. This method was observed to be particularly effective in cases with less experienced operators or more complex processes. According to Fast-Berglund, et al. (2013) the addition of smart tools and smart instructions could reduce the perceived complexity in work stations where complexity is high. However, Li, et al. (2018) observed an initial deterioration of quality when implementing a software based solution for information delivery in assembly work stations, the observed effect was thought to be a conse-quence of the operators’ inexperience with the particular device.

Further findings by Li, et al. (2018) include a comparison between text-only work instructions and similar information containing pictures for illustration, the illustrative instructions de-creased cognitive load and frustration among the operators. Text-only instructions had an op-posite effect by increasing the perceived cognitive workload, possibly due to an increased de-mand for uninterrupted concentration to achieve a correct interpretation. An identified benefit of work instructions printed on paper compared to a digital information source is the possibility of physically rearranging the documents to allow for an improved overview. However, the fea-sibility of this method decreases as the number of product variations increases due to the in-creasing challenge of managing printed documents in larger quantities

As described by Nozhy and Badrous, (2011) there are potential upsides to the presence of com-plexity in production systems as well as in product design. Strategically enhancing comcom-plexity instead of attempting to minimize it could possibly result in increased flexibility, enabling the company to conform to different customer demands. Such an approach requires effectively con-trolling the complexity through applying structural complexity management focused on the re-alization of competitive advantages. To ensure a sufficient ability to handle increasing com-plexity due to changes and reconfigurations, quality assurance and error safety are identified as critical factors (ElMaraghy & ElMaraghy 2016).

It could be possible to prevent complexity from spreading through the assembly system by implementing the principle of delayed differentiation which refers to assigning tasks that con-tain a greater number of different variations to a station as far downstream in the assembly line as possible (Zhu et al. 2008). In a prior study, Zhu, et al. (2007) presents a solution for mini-mizing complexity in manufacturing systems by assembly sequence planning. The measure of complexity utilized is OCC and it is evaluated and optimized by creating a simulation model of an assembly system. The optimization problem is solved through a transformation into a net-work flow model, allowing an optimal solution to be extracted by implementing dynamic pro-gramming methods. However, the applicability of the simulated model is questioned by Zhu, et al. (2007) due to restrictive assumptions affecting the models accuracy.

(24)

18 3.3.6 Complexity and human performance

The following segment is a brief overview of human performance in the context of completing complex assignments and how it can be improved for increased performance. In addition, the effects of complexity on operator performance is explored in this segment.

According to observations made by Ericsson, et al. (1993), the addition of detailed instructions in combination with developing new procedures facilitated improvements for individuals ap-pearing to have stalled out on a suboptimal level of performance. Especially in the presence of increasing complexity of the given tasks, explicit instructions of the optimal method in combi-nation with supervision. Allowing for individualization is a powerful strategy for stimulating performance improvements. A key factor is a well thought out sequence of increasing the task complexity by implementing individualized supervision above group instruction when feasible. In addition, Ericsson, et al. (1993) identifies a lack of deliberate practice in regards to work activities past the introduction stage. Deliberate practice refers to in a structured activity ex-ploring and overcoming weaknesses that are prohibiting desired performance within the given task for the individual in question.

Given the repetitive nature and requirement for effortful concentration, deliberate practice may not provide much enjoyment for its participants, thus requiring the trained individuals to be sufficiently motivated to practice for the sake of achieving the improvements that are desired. According to Mattsson, et al. (2016) there is an identified research gap regarding the sustaina-bility of the working conditions of operators in assembly systems. Thus, implying a degree of uncertainty concerning the long-term effects on both health and performance. Mattsson, et al. (2016) further concludes that manufacturing companies need to prioritize employee well-being and consideration for the emotional state of the employees. Negative emotions has been ob-served to affect operator performance, therefore it is crucial to mitigate dissatisfaction and bore-dom across the company’s work force to ensure long-term performance.

However, Mattsson, et al. (2016) additionally points out the influence of individual differences on how a certain task is received by the operator. While increased variety and complexity of the work tasks may be perceived as an enjoyable challenge by one individual, providing moti-vation and energy. The same task may seem overly demanding and stress-inducing for a differ-ent individual that would be more comfortable and effective in a scenario with less uncertainty. According to Kahneman, (2003), the cognitive processes of the recipient of instructions regard-ing any given task must be taken into consideration to avoid unnecessary cognitive load. A consequence of increasing the cognitive load is a decrease in performance in the tasks at hand. According to Kahneman (2003) work instructions should take the operators cognitive processes into consideration to avoid unnecessary cognitive load and decreased operator performance. According to Fast-Berglund, et al. (2013) there is a linear positive relationship between operator choice complexity and the occurrence of assembly errors. As a result, any increase in complex-ity is expected to also increase error occurrence. Additionally, differences in skill level as well as the individual needs and preferences of the different operators should be considered when implementing changes to an assembly process (Mattsson, et al. 2016b).

If work instructions and information is displayed in an excessively comprehensive way, errors may occur as a consequence, particularly in situations of higher stress or when the operator is inexperienced (Fast-Berglund, et al., 2013). A possible solution identified by Fast-Berglund

(25)

19

and Stahre (2013) is the implementation of cognitive support to filter information when reading work instructions, thus decreasing the risk of the operator making a mistake caused by misin-terpreting the instructions.

According to the findings by Mattsson, et al. (2016) there is a weak relationship between oper-ator performance and the operoper-ator’s emotional state. Specifically, if an operoper-ator is experiencing negative emotions such as boredom or stress it could have a detrimental effect on performance in the task at hand. Similarly, Mattsson, et al. (2014) describes how the demands imposed on the process operator’s cognitive ability in an assembly process are mainly of an intuitive nature, as opposed to the slower and more thoughtful actions connected to reasoning. This is due to the fast and automatic actions required to execute the different tasks within an assembly process. An exception is at the occurrence of an uncommon product variation in the process, requiring the operator to search and process information to ensure correct execution of the assigned task. Since the additional difficulty of uncommon products still requires fast execution, a greater level of effort is demanded from the operator (Mattsson, et al., 2014). Zhu, et al. (2007) also noticed a connection between complexity and a measure of cognitive performance, thus imply-ing that reduction of assembly system complexity has the potential for inducimply-ing human perfor-mance improvements.

According to Mattsson, et al. (2016b) the number of different product variations is a possible obstacle for the operator’s ability to concentrate and notice the differences due to limitations to the working memory. The energy expenditure connected to high-effort concentration makes it difficult to sustain over time, especially for monotonous task with many small details in com-bination with strict time and quality requirements.

(26)

20 4 EMPIRICAL FINDINGS

The empirical findings include both qualitative as well as quantitative data. The qualitative data is based on the observations made in the company’s assembly lines, the conducted interview, and information acquired during meetings and conversations with the company’s employees. The collected production data originates from the assembly line’s disturbance log, the internal production audit’s error record, and the errors observed by the testing department. In addition documents describing the quality assurance and control activities together with the assembly line’s work instructions have been studied. All collected production data was acquired in the form of Excel work sheets. An overview of the sources and quantities of the production data is presented in table 1.

Initially the detected errors was examined over a five-year period to provide a larger sample size which could magnify potential differences between different departments. In the later stages focus was changed to the most recent year of production to ensure relevancy and enable a deeper analysis of the current situation within the scope of this report.

Table 1, empirical data overview

Data source Start date End date

Production audit 2016-01-01 2020-12-31

Assembly line disturbance log 2016-01-01 2020-12-31 Assembly line disturbance log 2020-01-01 2020-12-31

Testing department record 2020-01-01 2020-12-31

4.1 Case study: company description

The studied company is a leading manufacturer of industrial machinery and technology based solutions located in northern Europe with a globally distributed clientele. The company is at the forefront of the worldwide technological development and has a pervading focus on customer satisfaction. Utilizing its production facility containing multiple mixed-model assembly lines tailored to the many different product variations, the main function of the operations is to as-semble pre-fabricated components and modules into complete products ready for delivery to the customers.

Through its continuous application of the chosen Lean Six Sigma strategy, the company has achieved a significant level of advancement in its production and assembly system. The assem-bly lines are pull-based flow lines, meaning that the production pace is levelled out by imple-menting a Takt time, giving each product a set lead time for each step throughout the process. Pull-based refers to every assembled product being assigned to an existing customer order which negates the risk of overproduction since there is not any products being assembled with the purpose of maintaining an inventory.

As expected in a L6S company, employee training is of significant priority and there are mul-tiple Lean Six Sigma Black belts and Green belts present within the organization. To ensure

(27)

21

smooth operations and minimize disturbances in the assembly lines, an additional group of em-ployees, referred to as Andon staff are constantly available for in-process problem solving re-quiring additional resources.

4.2 Assembly system reliability

According to the company´s initial analysis, approximately 80% of the quality deficiencies de-tected in the internal production audit are related to errors in the assembly processes. Despite rigorous inspections and testing routines both within the assembly system and in the following testing process, defects are still being left undetected.

The errors are detected on the assembly lines, in the randomized production audit, and in the mandatory testing process. Historically, increases in errors have been observed in conjunction with the introduction of new employees in the assembly processes and also the announcement of lay-offs of contract workers. In addition, uncommon product models have been observed to be a challenge for inexperienced employees since the procedures can deviate significantly from the most common models.

4.3 Process description, Assembly line A.

The observed assembly line is made up of a mixture of semi-automated and human-centered assembly work stations. For certain critical tasks industrial robots are implemented to ensure the desired level of quality and productivity. The automated assembly cells combine turning angle monitoring with tightening torque measurements to ensure that the assembly is within the defined tolerance limits. Fault detection with automatic halt of production is utilized in the au-tomated processes, eliminating the risk of incorrectly assembled products passing through un-detected.

Statistics are kept for all bolted joint connections performed in the automated processes to en-able traceability in the event of customer complaints. The automated work stations are still requiring some human input for material handling, component setup in the assembly fixtures, and problem solving in the event of an error. In the processes a Total quality control (TQC) checklist is filled out following the completion of the tasks in each station.

The tasks performed by the operators within the assembly line is made up of the following procedures:

- Material handling,

- Monitoring of machines and set-up of components for the industrial robots - Manual and automated assembly of bolted joint connections

- Assembly and fitting of electrical wiring - Electric motor installation

- Gearboxes assembly and installation

- Inspecting and testing functionality of installed electrical motors and other mechanical components

- Fulfilling the tasks specified in TQC checklist to ensure that all critical components are within specification

A strategy for improving reliability in assembly processes