Implementing Design For Automatic Assembly

(1)

IN

DEGREE PROJECT MECHANICAL ENGINEERING, SECOND CYCLE, 30 CREDITS

,

STOCKHOLM SWEDEN 2018

Implementing Design For

Automatic Assembly

A recommendation on how to implement and

apply DFAA at Company Y

(2)

Implementing Design For Automatic

Assembly

A recommendation on how to implement and apply DFAA at

Company Y

Filippa von Yxkull

Master of Science Thesis TPRMM: 2018: KTH Production Engineering and Management

Industrial Production

(3)

(4)

Abstract

The need to work with Design for Automatic Assembly (DFAA) has been widely recognized in the literature. However, the implementation of DFAA is not clearly defined. Therefore, the purpose of this master thesis is to investigate and contribute with knowledge of how DFAA should be implemented into an organization, such as Company Y.

Several interviews have been conducted to establish a current state analysis, to receive an understanding of the current problems at Company Y and how to address them. A benchmarking study was conducted, where the three companies Ericsson, Company X and Scania were interviewed. All three companies have successfully implemented DFA and were interviewed with the purpose to obtaining their best practices. The study also included an early implementation of DFAA, where a software based DFA2-method created by Eskilander (2001) was tested on a current product and a new developed design concept at Company Y. Based on this a recommended workflow of the evaluation could be attained.

Based on the empirical gatherings several recommendations of how DFAA should be implemented into the organization could be made. The study highlights that DFAA should be applied as early as possible in the product development process. The DFA2-method should be utilized at product level to facilitate concept selection and at part level to make the products/modules suited for automatic assembly, before the design is “locked” and before a physical prototype is created. The departments that should be working with DFAA includes individuals from production, design quality and purchasing. However, once DFA becomes rooted in the company, more functions in the company’s supply chain will become affected. This means that more functions might need to be included in the work of DFAA. Finally, the study includes a decision model, in which the decisions are based on the measurable values received from the DFA2-method.

(5)

Sammanfattning

Behovet av att arbeta med Design for Automatic Assembly (DFAA) har uppmärksammat i litteraturen. Däremot har implementeringen av DFAA inte blivit tydligt definierat. Syftet med detta examensarbete blir således att undersöka och bidra med kunskap om hur DFAA ska implementeras i en organisation, så som Företag Y.

Flera intervjuer har genomförts för att upprätta en nuvarandeanalys för att få förståelse för de rådande problemen hos Company Y och hur dessa ska hanteras. En benchmarkingstudie genomfördes, där de tre företagen Ericsson, Company X och Scania intervjuades. Alla tre företagen har framgångsrikt implementerat DFA och har intervjuats med syftet att erhålla deras bästa praxis. Studien innefattar även en tidig implementering av DFAA, där en mjukvarubaserad DFA2-metod skapad av Eskilander (2001), har testats på en aktuell produkt och ett nytt utvecklat koncept på Company Y. Baserat på detta kunde ett rekommenderat arbetsflöde av utvärderingen presenteras.

Baserat på empiriska studien kunde flera rekommendationer gällande hur DFAA ska implementeras i en organisation skapas. Studien belyser att DFAA bör tillämpas så tidigt som möjligt i produktutvecklingsprocessen. DFA2-metoden bör utnyttjas på produktnivå för att underlätta konceptvalet och på komponentnivå för att göra produkterna/modulerna lämpade för automatisk montering, detta innan designen är "låst" och innan en fysisk prototyp har konstruerats. Avdelningar som ska arbeta med DFAA inkluderar produktion, designkvalitet och inköp. När DFA blir rotad i företaget kommer dock fler funktioner i företagets supply chain att påverkas. Det innebär att fler funktioner kan behöva inkluderas med arbetet kring DFAA. Slutligen så inkluderar studien en beslutsmodell relaterat till DFAA. Besluten baseras på de mätbara värden från DFA2-metoden.

(6)

Abbreviations

In this section, the abbreviations and glossary for terms commonly used throughout the master thesis will be displayed.

Abbreviations

CE Concurrent Engineering

DFX Design for X

DFA Design For Assembly

DFAA Design For Automatic Assembly

DFMA Design For Manufacture and Assembly

DFA2-index Design For Automatic Assembly Index

CAD Computer-aided Design

BOM Bill Of Material

R&D Research and Development

MFD Modular Functional Deployment

FMEA Failure Modes and Effects Analysis

Glossary

DFA2 The method within the area of DFAA

(9)

Figures & Tables

This section states the figures and tables that were included in the master thesis.

Figures

FIGURE 1 ILLUSTRATION OF THE STRUCTURE OF THE MASTER THESIS ... 4

FIGURE 2 ILLUSTRATION FROM (BOOTHROYD, DEWHURST AND KNIGHT, 2011) DISPLAYING THE "OVER THE WALL" APPROACH. ... 6

FIGURE 3 ILLUSTRATION FROM (ESKILANDER, 2001) DISPLAYING THE HIERARCHICAL STRUCTURE OF DFX. ... 7

FIGURE 4 TABLE OF THE MOST COMMON COMMERCIAL DFA METHODS (ESKILANDER, 2001) ... 9

FIGURE 5 ILLUSTRATION OF DFMA SOFTWARE, DFA PRODUCT SIMPLIFICATION (DFMA.COM,2018) ... 12

FIGURE 6 GUIDELINES FOR PRODUCT LEVEL AND PART LEVEL (ESKILANDER, 2001) ... 13

FIGURE 7 SNAPSHOT OF THE DFX MODULE PROVIDED BY AVIX (AVIX.EU, 2018) ... 14

FIGURE 8 ILLUSTRATION OF THE VISUAL AIDS PROVIDED BY AVIX (AVIX.EU, 2018) ... 15

FIGURE 9 RESULT OF AN DFA2-‐ANALYIS ON PART LEVEL. ... 15

FIGURE 10 ILLUSTRATION OF TIME SAVINGS ACHIEVED BY IMPLEMENTING DFMA (BOOTHROYD, DEWHURST AND KNIGHT, 2011). . 19

FIGURE 11 FACTORIES WITHIN THE FACTORY (ERICSSON AND ERIXON, 1999) ... 20

FIGURE 12 ILLUSTRATION OF THE STUDY’S RESEARCH PROCESS ... 23

FIGURE 13 ILLUSTRATION OF THE FLOW CHART OF THE LITERATURE SEARCH (COLLIN AND HUSSEY, S. 82, 2014) ... 27

FIGURE 14 SNAPSHOT OF ACTION LIST (ULIN, 2018) ... 33

FIGURE 15 DFA IN ERICSSON’S DESIGN PROCESS ... 34

FIGURE 16 DFA IN THE DESIGN PROCESS ... 35

FIGURE 17 ILLUSTRATION OF THE AP-‐LIST MEASUREMENT ... 36

FIGURE 18 ILLUSTRATION OF THE PRODUCT DEVELOPMENT PROCESS AT COMPANY X ... 40

FIGURE 19 ILLUSTRATION OF A SNAPSHOT OF THE DFA CHECKLIST AT SCANIA (HOLMER 2018) ... 41

FIGURE 20 ILLUSTRATION OF SCANIA’S INTERNAL WIKI (KLINGNELL, 2014). ... 42

FIGURE 21 ILLUSTRATION OF SCANIA’S PRODUCT DEVELOPMENT PROCESS ... 44

FIGURE 23 COMPANY Y’S PRODUCT DEVELOPMENT PROCESS (INTRANET, 2018). ... 47

FIGURE 25 EVALUATION OF CURRENT DESIGN OF THE MECHANICAL COMPONENT ... 53

FIGURE 26 RESULTS OF THE NEW DESIGN CONCEPT ... 54

FIGURE 27 ILLUSTRATION OF THE CRITERIA “WHEN VISION IS NEEDED” ... 55

FIGURE 28 DFAA IN COMPANY Y’S PRODUCT DEVELOPMENT PROCESS ... 59

FIGURE 29 REGRESSION ANALYSIS (SOLME, N.D) ... 60

FIGURE 30 RECOMMENDATION OF A DECISION MODEL REGARDING DFAA ... 61

FIGURE 31 RECOMMENDATION OF A DECISION MODEL REGARDING DFAA ... 66

Tables

TABLE 1 RECORD OF ALL THE INTERVIEWS, MEETINGS AND OTHER ACTIVITIES HELD DURING THE THESIS. ... 24

TABLE 2 ERICSSON’S DESIGN PROCESS ... 33

TABLE 3 LIST OF THE PARTS IN THE CURRENT DESIGN OF THE SUBASSEMBLY OF THE MECHANICAL COMPONENT ... 51

TABLE 4 LIST OF THE ASSEMBLY ORDER OF THE CURRENT DESIGN. ... 52

TABLE 5 LIST OF THE PARTS IN THE NEW DESIGN OF THE SUBASSEMBLY OF THE MECHANICAL COMPONENT. ... 53

(10)

Foreword & Acknowledgements

This master thesis was conducted in collaboration with Company Y. The project is completing my master in Production Engineering and Management at KTH, Royal Institute of Technology. The study was conducted in the spring of 2018, Stockholm, Sweden and constitutes 30 credits. Firstly, I would like to thank my supervisor at KTH, Mauro Onori who provided valuable feedback. Secondly, I would like to thank my supervisor at Company Y and everyone involved at Company Y who contributed with engagement, support and valuable opinions.

I would also like to thank Mikaela von Yxkull and David Norman for the support during my work.

(11)

(12)

1. Introduction 

In this chapter, the background, the problem definition, purpose and research question, study’s expected contribution and the disposition will be presented. Finally, the delimitations of the thesis will be presented.

1.1 Background

Globalisation has over the past 20 years changed the face of manufacturing. Low-cost countries like China were used as offshoring opportunities to lower operational costs and cut supply chain costs. However, with today’s rising labour costs as well as the availability of more efficient automation system, offshoring might come to an end (Manenti, 2014).

In addition to this, there has also been a change in the way consumers purchase their goods, as they hold information about both prices and market trends. Consumers are becoming more impatient and reluctant to long lead-times, demanding highly customized products at a low cost. There is therefore a need for companies to be able to respond quickly to changes in demand and get closer to their customers (Manenti, 2014).

The world’s largest growth trend in the robotics industry happens in Asia with China as the market leader (IFR International Federation of Robotics, 2017). This is due to countries like China are experiencing labour shortage and pressure from low labour countries in Southeast Asia, such as Vietnam, which have resulted in China putting more emphasis on automation. European companies are struggling to keep up and be competitive to gain market share from countries like China (Onori, Sandin and Alsterman, 2012). In addition to this, there is a rapid decline in the workforce in Europe. To be able to strengthen the competitiveness and address these issues companies need to automate their production (Onori and Oliveria, n.d.) If specific conditions are met, the robot system market is expected to rise in Europe (Onori, Sandin and Alsterman, 2012)

(13)

DFA (Design for Assembly) is a methodology that has been used since the 1980s. The method aims to simplify the product structure and reduce the assembly cost (Boothoyd1992). DFAA (Design for Automatic Assembly) is based on DFA, however only focusing on automatic assembly, since working with DFA for manual assembly does not correspond to the product being suited for automatic assembly. Designing a product for automatic assembly allows maximum flexibility as the product can be assembled both manually and automatically (Eskilander, 2001).

1.2 Problem Formulation

Despite the many reasons to invest in automatic assembly solutions, designing for automatic assembly has not received highest priority of the design department at manufacturing companies and it is still common that products are assembled manually (Eskilander 2001). One manufacturing company that is recognising the benefits of automating the assembly process is Company Y. With volumes expected to increase there is a need to increase the output from the assembly shop, with human assembly workers limited to work a few hours per day. In addition to this, the management of Company Y has set a requirement of 90% automation level on the new designed products introduced in production.

Despite this, Company Y is lacking a structured way of making their products more suited for automatic assembly. Currently, the work of creating a competitive product is predetermined by the work of the designer. There is therefore a need to implement a structured method to make the products more suitable for automatic assembly to increase the application of automation in

production. Design For Automatic Assembly (DFAA1) is a systematic approach to determine

how well a product/module is designed for automatic assembly.

1.3 Purpose & Research Question

The purpose of this master thesis is to investigate how Company Y should implement DFAA. This to make the products and components more suitable for automatic assembly and increase the application of automation in production. Based on the purpose, a main research question has been formulated:

v How should Company Y implement DFAA, to increase the possibility to automate the

assembly process?

(14)

Ø Where in the development process should the early adaption of DFAA be

implemented?

Ø Which departments should be involved and what competence is need? Ø What should the decision model look like?

1.4 Delimitations

The study is delimited to only analyse two DFA methods. However, only one method was chosen to be analysed in practice namely, i.e. the DFA2-method created by Eskilander (2001) and provided by the Company Solme as a software solution. The other DFA method- DFMA by Boothroyd and Dewhurst (2011) was only treated in the literature study.

Furthermore, two designs were chosen to be evaluated using the DFA2-method, the current design subassembly of a mechanical component and a new developed concept of the mechanical component. The DFA method includes a cost analysis that reveals the costs related to the design concepts. However, this thesis will not include any in-depth financial aspects or cost aspects. It is evident that the financial aspects linked to the work of DFAA will affect the decisions making and the overall business model at an organisation. This should be viewed separately.

The study includes an investigation on how other companies have implemented and applied DFA. The investigation is delimited to analysing three manufacturing companies that successfully have implemented DFA, namely Ericsson, Scania and Company X.

1.5 Study’s Expected Contribution

The Expected contribution of the study is on an academic level and a managerial level.

Academic Level: On an academic level this study aims to contribute in the area of DFA.

Several researchers have highlighted the need to work with DFAA in order automate the assembly system, as well as the benefits when implemented (eg. Boothroyd, Dewhurst and Knight, 2011; Eskilander 2001). However, little information could be found on how it should be implemented and the workflow of the methodology. This study aims to contribute with this knowledge.

Managerial Level: Managerial level corresponds to what the findings of the study entails for

(15)

1.6 Disposition

The disposition of the thesis is illustrated in figure 1 below. This to enable the reader to easily develop an understanding of the structure of the thesis. Figure 1 gives an overview of the eight chapters with the related sub-sections.

Figure 1 illustration of the structure of the master thesis

(16)

2. Literature & Theory

In this chapter, the literature and theoretical studies connected to DFA will be presented. First, definitions of important terms and concepts will be given.

2.1 Definitions

The thesis aims to investigate an early implementation of DFAA and how it should be applied. The theory of DFA has been considerably investigated by Boothroyd (1992). Despite the literature being written over twenty years ago, it is still relevant and based on current research about DFA. According to Moultrie and Maier (2014) the field of DFA is relatively small and the absence of academic papers suggest that new information will be difficult to discover. Many principles used in engineering design textbooks were established in the 1960s and the 1970s. The work of DFAA has been extensively investigated by Eskilander (2001) and is therefore referred to throughout the entire thesis. The DFAA method that was applied in this master thesis is based on Eskilander’s DFA2-method. To understand the literature, the definitions of the main concepts will be described, i.e. assembly, concurrent engineering, DFX, DFA and DFAA.

2.1.1 Assembly

Assembly is a part of the production system. According to Nof, Wilhelm and Warnecke (1997) the assembly process is defined as:

“The aggregation of all processes by which various parts and subassemblies are built together to form a complete, geometrically designed assembly or product (such as a machine or an electronic circuit) either by an individual, batch or a continuous process” (Nof, Wilhelm and

Warnecke, 1997, 2).

According to Sanders (2009) the assembly process has an impact on the products lead-time, quality and cost and is therefore considered to be the most vital processes. The assembly process reports for 40% of the manufacturing cost. In addition to this, beyond 70% of the manufacturing costs is set in early conceptual design phases. Assembly should therefore be considered early in the design cycle (Sanders, 2009)

(17)

2.1.2 Concurrent engineering

Concurrent engineering is defined as an approach of working in multi-disciplinary teams as well as working parallel. By doing so, every phase of the product life cycle is co in every stage of the product development (Eskilander 2001). This is to avoid the mistakes associated with the traditional sequential product development, where the different departments are maintained separately (Filippi and Cristofolini, 2010). The traditional serial development can be compared to the “over the wall” approach (see figure 2), where the design is thrown over an imaginary wall to the manufacturing department to handle the issues that occur in production (Boothroyd, Dewhurst and Knight, 2011).

Figure 2 Illustration from (Boothroyd, Dewhurst and Knight, 2011) displaying the "over the Wall" approach.

According to Eskilander (2001) there are two main advantages by working with concurrent engineering. Firstly, problems that would have been discovered later in the development chain, can be identified early. Secondly, if the work is done parallel the total lead-time can be shortened. In addition to this, by working in parallel, feedback from the manufacturing department can be included in the design of the product, which can result in an improved product. However, there is a need for tools to enable visualisation of the relationship between the parameters of design and manufacturing. According Li and Hwang (1992) DFA is a valuable tool for achieving concurrent engineering.

2.1.3 DFX

(18)

Figure 3 Illustration from (Eskilander, 2001) displaying the hierarchical structure of DFX.

2.1.4 DFA

Design for Assembly (DFA) was established in the 1980s and is as relevant today as when it was first recognised. DFA is a process for improving a product design for an easy and cost-effective assembly (Moultrie and Maier, 2014).

There are several benefits of working with DFA and Egan (1997) categorises the potential effects into short-term and long-term. The short-term effects of implementing DFA correspond to the number of reduced parts, reduced assembly time and reduced manufacturing and assembly costs. The long-term effects of implementing DFA correspond to improved quality and that conditions for concurrent engineering are facilitated.

According to Moultrie and Maier (2014) there are two ways DFA can be considered i.e. general DFA heuristics and systematic methods for analysing assemblies. General DFA heuristics correspond to the images which display “good assembly practices” (e.g. all assembly should be done from above) versus “poor assembly practices” (e.g. assembly is done from several directions). The heuristic images are easy to understand, though are difficult to use in a structured fashion, for example in a design review context. Systematic methods for analysing assemblies correspond to the methods that enable designers to identify and improve the design in a structured way (Moultrie and Maier, 2014)

2.1.5 DFAA

Design for Automatic Assembly (DFAA) is based on DFA (see figure 3 above), however only focuses on automatic assembly. Boothroyd and Dewhurst (2011) discuss robot assembly, which is included in automatic assembly. Automatic assembly is defined as any mechanical assembly process that has no human interaction. The issues are similar if the assembly process is flexible automated or hard automated (Eskilander, 2001).

2.1.6 DFA guidelines for automatic and robot assembly

(19)

for manual assembly, machining, high-speed automatic assembly and robot assembly etc. However, in this section only the guidelines for automatic and robot assembly will be described. The problem when automating the assembly process, is to automatically handle the part, rather than inserting operation. Therefore, when considering design for automation the design emphasises on ease for automatic feeding and orienting (Boothroyd, 1992). The design rules for high speed automatic assembly are categorised into rules for the product design and design rules for the parts. These are summarised by Boothroyd, Dewhurst and knight (2011) as follows:

Product design rules

1. Reduce the number of parts.

2. Establish a base, where the assembly is built.

3. Establish features on the base part that enable the part to have a stable position in the horizontal plane.

4. Establish a design where the product can be built up in layers. The parts should be assembled from above and positively located, i.e. no indication to move during the action of horizontal forces.

5. Design parts with chamfers or tapers. This will guide and position the parts correctly. 6. Avoid fastening operations (e.g. screw fastening), since they are both time consuming

and expensive.

Design of a Part rules

1. When parts are placed in bulk, they can tangle with similar parts. Try to avoid holes or slots and projections. This can also be done by making the slots smaller than the projections.

2. Establish symmetry in parts so no orienting devices are needed.

3. When symmetry cannot be established the asymmetrical features should be exaggerated. This is to facilitate orienting.

The design guidelines for robot assembly are similar for those of high-speed automatic assembly and manual assembly. The design rules for robot assembly are summarized by Boothroyd, Dewhurst and Knight, (2011):

1. Minimize the part count.

2. To make the parts self-aligning, use features such as leads, chamfers, lips etc. This is of importance, to ensure consistent and fault-free in part insertions.

3. Parts that are not secured immediately after insertion should be self-locating. This is of importance since a single robot arm cannot hold down unsecured arms and special fixturing is needed.

(20)

6. The need to manipulate a previously assembled part or reorienting partial assembly should be avoided.

7. Parts should be designed so they can be simply handled from a bulk. This is to avoid that the parts to nests or tangle when stored in bulk. Part that should be avoided include, flexible parts, abrasive parts (since they will wear the handling systems), sticky or magnetic parts etc.

8. When parts need to be presented by automatic feeders, make certain that they can be oriented using simple tooling.

9. When parts that needs to be presented by automatic feeders, make certain that they are presented in an orientation, which they can be gripped and inserted without manipulation.

10. Ensure stable resting if parts need to be presented in magazines or part trays, so they can be gripped and inserted without manipulation.

2.2 DFA methods

There are several methods that are used to support product design. The twelve commercially available DFA methods are listed in figure 4 (Eskilander 2001).

Figure 4 Table of the most common commercial DFA methods (Eskilander, 2001)

(21)

2.2.1 Boothroyd’s & Dewhurst’s DFMA Method

Boothroyd and Dewhurst are pioneers in the field of design for manufacture and assembly and created the methodology DFMA. The term “DFMA” is a combination of the terms DFM (Design For Manufacture) and DFA (Design For Assembly). The purpose of the methodology is to reduce the number of assembly operations by reducing the number of parts and to make the assembly operations easier to perform. There are four methods for DFA available from Boothroyd and Dewhurst Incorporated, namely methods for manual, robotic, automatic and printed circuit boards. However, the most commonly used DFA method is for manual assembly (Boothroyd, Dewhurst and Knight, 2011).

A central role of DFA-methodology is to reduce the part count, since the spate part reduction decreases the cost of assembly and the total cost of parts. According to Boothroyd, and Knight, (2011) to determine if the part needs integration, three question should be asked:

1. Does the part move in relation to already assembled parts, during assembly of the product?

2. Should it be of a different material or can it be isolated from parts already assembled?

3. Must the part be separate from already assembled parts, otherwise assembly or

disassembly would be impossible?

If any of the three questions are answered with “yes”, the part/component is validated for its existence and needs assembling. If all the three questions are answered with “no” the part is a candidate for integration (Boothroyd, Dewhurst and Knight, 2011).

The manual DFA-analysis is based on estimated assembly times for handling, insertion and fastening of parts. To estimate the manual assembly times a classification and coding system is used, where factors affecting the handling time and insertion time are taken into consideration. Examples of part features affecting the handling time are the part symmetry, thickness, size, weight and if a part requires two hands for manipulation. Features that influence the time to insert and fasten a part is if it is designed to avoid jamming, its part geometry, if it has obstructed access and if there is restricted vision (Boothroyd, Dewhurst and Knight, 2011) A design that does not correspond to the ideal design is punished with a longer assembly time, as it takes longer to e.g. orient (Eskilander, 2001).

DFA-index

(22)

Formel 1: Assembly efficiency formula (Boothroyd and Dewhurst (2011)

Ema is defined as the assembly index or assembly efficiency.Nmin is corresponds to the minimum

hypothetical number of parts. The sum theoretical minimum of parts corresponds to the ideal number of parts in a product, i.e. possible integration of parts according to the three questions

stated in the text above. ta refers to f the basic assembly time for one part. Finally, tma refers to

the sum of total estimated assembly time, i.e. the time it takes to assemble the product, when the part presents no assembly difficulties (Boothroyd, Dewhurst and Knight, 2011).

High speed assembly and robot assembly

As aforementioned, it is possible to analyse the products and component for ease for automatic assembly, i.e. high-speed assembly or robotic assembly. The difference between the two types of assemblies is that high-speed assembly requires less flexibility as well as a lower equipment cost (Eskilander, 2001).

The analysis for automatic assembly resembles the analysis for manual assembly except that the analysis estimates cost for orientation and handling instead of estimated assembly times. The result of the evaluation is an average cycle time (Boothroyd, Dewhurst and Knight, 2011).

Software solution DFMA Inc.

(23)

Figure 5 Illustration of DFMA Software, DFA product Simplification (dfma.com, n.d.).

2.2.2 DFA2, Design For Automatic Assembly method

DFA2 is a method within the domain of DFAA. The DFA2 method can be applied in early developments of products, since it does not require a prototype. The main objective with the method is to achieve a design that is as non-complex as possible, which yields the simplest assembly process. However, the functional requirements still should be fulfilled (Eskilander, 2001)

Design rules

DFA2 consists of several structured design rules or guidelines. The purpose with the design rules is to provide the user with information to design the product in respect to automatic assembly. This enables the user to focus on one assembly problem at a time as well as ensure that no information is overlooked (Eskilander, 2001).

(24)

Figure 6 Guidelines for product level and part level (Eskilander, 2001)

Criteria

The method also includes a qualitative evaluation criteria. The evaluation criteria are based on a levelled points scheme, where:

• 9 points are rewarded if it is the best solution from an automatic assembly perspective

• 3 points are rewarded if it is an acceptable solution. However, the solution is not completely satisfactory from an automatic assembly perspective.

• 1 point is rewarded if the solution is unwanted from an automatic assembly perspective.

The possible hazards of choosing a solution that is not perfect in respect to automatic assembly is visualized using the levelled point scheme, with the three levels 9,3 and 1 (Eskilander 2001). At the end of the evaluation the points rewarded to the product/components are summarized. The total score for the evaluated product is divided by the maximum ideal score and multiplied by a factor of 100, which is illustrated in the formula 2. Hence, the DFA2-index tells the user how close the product is to an ideal solution (Eskilander 2001).

(25)

In addition to this, the DFA2-method supports a cost analysis, which is based on an activity analysis (Eskilander 2001).

Software solution AviX DFX-Module

AviX is mainly a video analysis based software created by the company Solme. AviX enables analysis of the manual assembly processes, by combining video analysis with time and motion studies. Currently the software offers nine modules, AviX Mehod, AviX Balance, AviX Resource Balance, AviX FMEA, AviX Smed, AviX DFX, AviX Ergo, Execution Optimizer and Shop Floor viewer (Avix.eu, n.d.).

DFX is a collective name for the three methodologies, DFS (Design For Service), DFM (Design For Manufacture) and DFA (Design for Assembly), where the DFA method is based on Eskilander’s DFA2-method. AviX DFX offers standard templates, however also provides the possibility to supplement the system by offering the ability to customize the templates. This means that new aspects (also known as guidelines) could be created or updated as well as redundant aspects in the method could be removed.

(26)

The software provides visual aids, such as videos and images, which is illustrated in the snapshot in figure 8.

Figure 8 Illustration of the visual aids provided by AviX (AviX.eu, n.d.)

All information and comments during the DFA-analysis could be captured in AviX. This is of importance when issues with the design are recognized. According to Solme (2018) it is advisory to document the cause and what actions that need to be taken. In addition to this, the problem/action could be assigned a priority level from low, high and intermediate.

The result of the DFA-analysis is a DFX report, where all the evaluated parts and subassemblies are displayed (see figure 9). From this, each part and subassembly receives a DFA2-index, which is used to calculate the aggregated DFA2-index for the product/module. In addition to this, an assembly time is estimated for each component as well as an aggregated assembly time for the entire product/module. Finally, there is a possibility to calculate the purchase cost for the product/module.

Figure 9 Result of an DFA2-analyis on part level.

(27)

Application of DFA2

DFA2 could be used to evaluate an early design development or used to evaluate an already existing product/module. In the early design of a new product/module it is preferable that it is modularised using the MFD-method. The analysis should be initiated by the guidelines for the entire product/module. The next step is to evaluate each component, using the guidelines that correspond to the part level (Eskilander, 2001).

When using the method for an already existing product, the first step is to analyse the structure of the object, which determines the assembly sequence. The assembly sequence is initiated with the base object, on which all remaining parts are assembled. The object can be broken down into subassemblies, if the assembly occurs without disruptions. The object is analysed on a product level and then part level (Eskilander 2001).

2.3 Implementation of DFA

There is no “correct approach” of implementing DFA. However, DFA can be facilitated by using a demonstrator, until the DFA tools and techniques become a standard approach when introducing new products (Eskilander, 2001). The following section concludes the important aspects when implementing DFA into the organization.

2.3.1 Company mission, vision and objectives

The company needs a mission to distinguish itself from other companies on the market. The mission incudes the company’s future state- “the vision”. The vision is the intended objectives, that are supposed to serve as a guide for the internal decision making. A well-defined vision can motivate employees. The vison can be broken down into short and long term goals. The

goals should be defined as SMART2_{(Hallin and Karrbom Gustavsson, 2015).}

According to Synnes and Welo (2015) to be able to make successful decisions concerning automation, they should be associated with the organizations long term goals. This since both over-automation and under-automation can have a negative effect of the company’s competitiveness (Synnes and Welo, 2015). Furthermore Frohm (2008) concludes that “The

main reason that an automation project ends in failure is unrealistic or undefined objectives” (Frohm, p 65, 2008).

2.3.2 Implementation strategies

According to Bruke and Carlson (1987) the implementation of DFA should be done as following: 1. Accommodate a DFA overview to senior management

(28)

3. Define objectives, for example reduce costs. 4. Select a pilot program

5. Select a test case

6. Establish the individual members and the structure of the team. 7. Establish coordination of the training.

8. Initiate the first workshop.

9. Maintain and keep having meetings as needed.

According to Dean and Susman (1989) there are four approaches that can be used for structuring an organisation for “manufactural design” namely:

1. Manufacturing sign-off

The manufacturing sign-off refers to the veto power that is given to the manufacturing department over the design department. With other words, the product cannot be released until the manufacturing department gives their permission. The main advantages with this approach is that a product with a low reducibility is unlikely to reach production. The main disadvantage of this approach is that there is no base for creative interchange between the two functions.

2. The integrator

The integrator corresponds to the individuals that work with the designers with the producibility problems and operates as liaisons to the manufacturing team. This means that the integrator’s job is to maintain the manufacturing and the design standpoints in balance. In other words, if an integrator supports manufacturing too heavily he/she will lose credibility with the designers and vice versa. A drawback with this approach is that it’s too difficult to find an individual to act as an integrator, since the education for manufacturing and design requires two degree programs. In addition to this, using an integrator can result in the so called “guru syndrome” i.e. if only the integrator works with producibility then no one else will. The organization will become very dependent on one or a few individuals working with ease for assembly. Finally, the approach does not facilitate concurrent engineering.

3. Cross-functional teams

Cross-functional teams means that different departments work collaboratively throughout the product development process. At the minimum, this involves individuals from the manufacturing department and from the design department. The main advantage with this approach is that the different functions in the organisation can influence the design before release and that concurrent engineering is facilitated. The disadvantage with this approach is that designers feel prohibited, due to the cross functional team demand of the design becomes too unrealistic and the designer’s creativity become undermined. The approach requires that the individuals that are involved in these cross-functional teams receive broader knowledge within producibility, as there is no longer a

(29)

4. The product-process design department

The product-process design department corresponds to creating a single department that is responsible for both the product and the process. The greatest structural change is created through this approach, which also can cause the greatest resistance from the employees. However, the approach supports concurrent engineering and there is mutual education through a day-to-day contact. There are a series of variations to this approach namely:

• A senior manager that controls both the product and the process, however each function have separate units.

• A single department including product and process engineers, with a manager that is responsible for both the two functions.

• Individuals that are responsible for both functions, i.e. product-process engineers, which are composed into one department.

According to Ahm and Fabricious (1990) it is important that the management is involved and understand the practises of a DFM/DFA project, i.e. that it requires more resources to be assigned in the beginning of the developing phase.

2.3.3 Implications of implementing DFA

The implementation of DFA is prone to be more successful, if there are extensive databases available and cross-functional communication skills already set in the organization. The databases with information of manufacturing and design can be “bought” from consultant firms. However, the organizational structures needed to facilitate cross functional communication skills cannot be “bought”. Boothroyd, Dewhurst and Knight (2011) state several falsely claimed reasons for not implementing the DFA. Four of them are described below:

Not enough time

(30)

Figure 10 Illustration of time savings achieved by implementing DFMA (Boothroyd, Dewhurst and Knight, 2011).

Not invented here

Difficulties can be experienced when new methods or techniques are suggested to designers. Especially when the decision of implementing DFA comes from the top management, who has the wish to reap the success resulting from DFA. This can result in resistance from the designers.

Ugly baby syndrome

Conducting a DFA-analysis often involves an outside group analysing a design, trying to find improvements to make the design more suited for assembly. However, this can create great resistance from the designer, as telling a designer their design needs improvement can be compared to telling a mother her baby is ugly.

We have been doing it for years

“We have been doing it for years” relates to the fact that a company already uses a producibility procedure to make parts more suitable to manufacture. However, these are often done in the end of the design process where the vital decisions which affect the manufacturing costs already have been made.

2.4 Impact DFA has on the future of manufacturing

According to Onori (2002) product design can be seen as an integral step in assembly system development. To raise the potential of automation, DFA plays a central role. DFA decreases the obstacles such as assembly complexity and incompatibility and industrial robots will be easier to integrate. In this section, the impact DFA has of the future of manufacturing will be described.

2.4.1 Modularity in respect to DFA

(31)

allocated to a specific assembly system. This enables small factories within the factory to be built. Parts of the product assortment can therefore be automated (Ericsson and Erixon, 1999). As pointed out by Eskilander (2001), if module area two assembles a module that is applied in numerous products through the product assortment, a flexible assembly system might be economically feasible. This is because the cost is shared among many module variants (Eskilander, 2001). However, with an automatic assembly system in the area (factory) the products need to be suited for automatic assembly, i.e. suited for the assembly process.

Figure 11 Factories within the factory (Ericsson and Erixon, 1999)

According to Eskilander (2018) modularity facilitates the work of DFA and DFAA. Applying DFA on single products is not sufficient as it only leads to a specific module variant being optimal, whilst the other module variants within the product assortment being overlooked. This can lead to the risk of sub-optimization, if the product architecture has not been defined beforehand (Ericsson and Erixon, 1999).

Through modularization, the number of product variants can be controlled. The same modules and components can be applied in more than of product family. This enables the possibility to have different product families with the same base materials or operation sequences, which can be used when designing a new product or production system (Eskilander, 2001). With other words, modularisation contributes to a high configurability and lowers the part count. According to von Yxkull (2018) this makes the work of DFAA become more efficient and not as cumbersome.

A modular method that includes DFA is the Modular Function Deployment method (MFD). The method consists of five steps and is used to define modules i.e.:

(32)

4. Evaluate the concepts

5. Improve the modules by applying DFA.

2.4.2 Evolvable assembly systems

To maintain a competitive assembly system there are two conditions that need to be met. Firstly, the assembly system needs to be able to respond quickly to the changing conditions. Secondly, the assembly activities need to be performed in an efficient way. The former, could be measured as the time it takes to reconfigure the system, to able to deal with a new product situation. The latter, could be measured as the function of the optimal assembly sequence. For an assembly system to remain ideal for the next production scenario, the system must be re-configured and re-planned. This is both time consuming and costly, as it requires the production to stop. Due to the frequent changes of customer needs, it is seldom successful to run the production to be ideal. With other words, there is a trade-off between production optimality and system responsiveness and it needs to be balance between process optimality and the ability to adapt to new requirements (Akillioglu, Neves and Onori, 2010).

Evolvable Assembly Systems (EAS) is a next generation assembly system that focuses on both agility and mass customization. EAS is based on reconfigurable modular concepts that allow continuous system evolution (Akillioglu, Neves and Onori, 2010).

(33)

3. Method

In this chapter, the methodology of the study will be discussed. This includes describing the research design & process, the primary sources, the secondary sources and the quality of the analysis. Finally, the ethics taken into consideration while conducting the thesis will be presented.

3.1 Research Design & process

The study was conducted at Company Y, with the purpose to investigate the conditions to implement DFAA into the organization. The master thesis was limited to 20 weeks and in the initial phase of the thesis, considerable amount of time was spent at Company Y (see Gantt-chart in appendix B). This has led to a good understanding of how Company Y currently works with making their product more suitable for automatic assembly. However, most empirics were gathered from the benchmarked companies Ericsson, Company X and Scania. The method is defined as a qualitative method with qualitative interviews and observations.

The thesis initiated with formulating a non-trivial research problem. Based on the problematization, a purpose and research questions could be formulated. The purpose of the thesis is classified as an exploratory research, since the aim of the study is to investigate patterns rather than to test a hypothesis (Collis and Hussey, 2014). In addition to this, the study has not previously been explored to any greater extent (Blomkvist and Hallin, 2015). An inductive approach was selected, where the problematization, purpose and research questions were continuously reviewed (Blomkvist and Hallin, 2015)

The research process was initiated with a pre-study to formulate the problem definition, followed by a literature review. After this the empirics of the study were gathered, which consisted of interviews, observations and an early implementation of DFA2 (workshops). In the analysis, the literature study and the empirical study were linked to be able to draw the conclusions. The research process is illustrated in figure 12.

(34)

Figure 12 displays the study’s research process

3.2 Primary Sources

The primary sources used in this thesis are, interviews, observations, an early implementation of the DFA2-method in the form of workshops and written materials published by Company Y.

3.2.1 Pre-study

In the beginning of the study a research proposal was offered by the author to Company Y. Thereafter, an informal meeting was held to discuss the problem formulation and a supervisor was assigned. However, only a brief problem explanation was given and an extensive pre-study was conducted. Therefore, initial interviews were conducted both to explore the problem and formulate a current state-analysis. In addition to this a literature review was conducted to define the scope and focus of the thesis.

3.2.2 Interviews

The interviews that were conducted during the thesis were of a semi-structured fashion, meaning that the pre-determined questions were created in advance, see interview guide in appendix A. During the interviews the questions were asked in a vis-á-vis approach, to support flexibility (Blomkvist and Hallin, 2015).

(35)

email. This was done to avoid misinterpretations of what had been said and for the interviewee to approve the information, so that the confidentiality agreements were kept.

The interviews made at Company Y were conducted to formulate the purpose and the current state analysis. The interviewees were represented from different departments. This was done, to receive different perspectives on the issues that were investigated. This is of importance according to Elisenhardt & Grabner (2007) to avoid the findings being subjective.

The interviews at Company Y was conducted with one interviewee at a time. This was done to avoid the answers being affected by another interviewee (Collis & Hussey, 2014). The interviews were recorded with permission of the interviewee, so the information could be transcribed. In parallel to the recordings, notes were taken throughout the interviews. This was done to be able to follow up on what the informant said during the interview (Blomkvist and Hallin, 2015). All the interviews, meetings and other activities conducted in this thesis are presented in table 1 below.

Table 1 Record of all the interviews, meetings and other activities held during the thesis.

Interviews

Title

Date

Place

Description Duration

Design engineer 24 January

2018 Company Y Semi-structured interview 50 min Automation Consultant 25 January 2018 Company Y Semi-structured interview 1h 10 min Global Production Engineering Manager 31 January 2018 Company Y Semi-structured interview 50 min

Project leader 31 January

2018 Company Y Semi-structured interview 50 min Production technician manager 12 February 2018 Company Y Semi-structured interview 40 min PhD-Stephan Eskilander 2 February 2018 Teleconference Semi-structured interview 1 h Alex von Yxkull 18 April

(36)

Producibility engineer at Ericsson 15 March 2018 Teleconference Semi-structured interview 50 min Global production engineers at Scania 21 March 2018 Scania Södertälje Semi-structured interview 2 h 4 May 2018 Teleconference Semi-structured

interview 45 min New Product Industrialisation engineer at Company X 2 February 2018 Teleconference Semi-structured interview 1h 20 min 13 March 2018 Teleconference Semi-structured interview 45 min

Meetings & Other Activities

R&D 25 January 2018 Company Y Initiation of Master thesis meeting 1 h Manager R&D Processes 25 January 2018 Company Y Explanation of Company Y’s PD-process 1 h 23 April 2018 Company Y 30 min Solme DFA2

presentation 5 April 2018 Company Y DFA2 review 1 h R&D and production 9 March

2018 Company Y

Automation

workshop 3 h R&D, production and

purchasing 25 April 2018 Company Y Automation workshops 3 h Solme 2 February 2018 Teleconference Learning to operate AviX DFX module 45 min 3.2.3 Observations

The observations made during the thesis were made to be able to formulate the current-state analysis. Therefore, design reviews and automation workshops were attended where the author acted as an “participating observant”, i.e. interacting and asking questions to the person that was observed. The observation methodology was applied since the questions asked were of exploratory nature (Blomkvist and Hallin, 2015).

3.2.4 Early implementation of DFA2

(37)

the software based DFA2-method. The workshops were conducted with individuals from different functions in the organizations, i.e. the service department, the production department and the design department. The designs to be analysed were selected by the assigned supervisor at Company Y. The designs that were analysed included the current design of a mechanical component and a new developed design concept. The new developed design was based on the current design.

Before the initial workshop was conducted, the current assembly process of the mechanical component was analysed in production. The BOM-list was added manually into AviX and CAD-files of the subassembly were displayed. In addition to this, a short introduction to DFA was presented. This included explanation of what DFA is, the purpose of DFA, advantages of DFA and a review of the methodology DFA2, where a bicycle bell was examined.

At the workshops, both the current design of the mechanical component and a new developed design concept was evaluated using the DFA2-method. During the workshops, the author acted as a moderator and explained and interpreted the guidelines offered by the software AviX (DFX-module). Notes were taken during the workshop. After the initial workshop was concluded, a short survey was sent out via email, with questions corresponding to how the method was experienced.

The choice of the DFA2-method provided by AviX, was based on observations and interviews made in the pre-study at Company Y. The DFA2 method is a well-established method and over 18 companies participated in the development of the method (Eskilander 2011). The method provides Company Y with a structured way of making their products more suited for automatic assembly. The software AviX is already used in preparation at Company Y. This meant that the information gathered in preparation could be linked to together. All data can be collected at the same place, which avoids the risk of data becoming obsolete.

3.3 Secondary Sources

(38)

Figure 13 Illustration of the flow chart of the literature search (Collin and Hussey, s. 82, 2014)

The literature mainly includes books, journals (both digital and paper), conference papers as well as student projects. The search engines utilised in this thesis was KTHB Primo through the KTH library and Google’s search engine for scientific articles, Google Scholar. The key words used to find the relevant literatures were, DFAA, DFA2, DFA, Design for automation, Product

Design, DFX, Modularization.

In addition to this, several documents published by Company Y were provided. The document included information that was used for the current state analysis.

3.4 Quality of Analysis

The quality of the scientific work was closely linked to the reliability and validity of the analysis. Therefore, the two characteristics needed to be scrutinised to be able to ensure the quality of the study.

3.4.1 Reliability

The reliability of the thesis corresponds to the precision and accuracy of the measurement (Collins and Hussey, 2014) or as (Blomkvist and Hallin, 2015) concludes it, studying the right thing. Furthermore, reliability refers to the repeatability of the study, answering the question- if the study was to be repeated a second time, would it produce the same result?

(39)

not decided in advance what was going to be discussed (Blomkvist and Hallin, 2015). These types of interviews experienced lower reliability than the semi-structured interviews.

The observations conducted and workshops conducted at Company Y have been done in close collaboration with the employees. This improved the outcome of the thesis, however the reliability of the study became lower. This is since it was difficult to replicate the informal conversations. In addition to this, the learnings of the topic DFAA have increased among the participants since the initial workshop, which could affect the result of the workshops conducted at Company Y.

3.4.2 Validity

The validity of the research corresponds to the degree to which a test measures what the researcher intends it to measure and if the result relates to the phenomena under the study (Collins and Hussey, 2014) or measuring the right thing, to put it simply (Blomkvist and Hallin, 2015).

The study was mainly based on interviews and observations. Using semi-structured interviews increased the validity of study, since the interview’s guides ensured that the “right thing” was being studied.

Three companies were benchmarked, to gain different perspectives of how other companies work with DFA. All three companies have successfully implemented DFA, and were therefore of interest. Each interviewee was considered knowledgeable, since they obtained managerial roles within the field of DFA. The interviewees are therefore argued to be valid.

The literature and theory used achieve validity according to Blomkvist and Hallin, (2015), i.e. was presented in the theory and then used in the analysis. This was done to correspond to the purpose and research questions. During the literature research the author has had a critical view of the information gathered. Most of the literature used was published for more than 15 years ago, however is still referred to in present literature and is therefore considered to be valid.

3.5 Ethical considerations

The ethical considerations taken during the study were based on what is written in the Swedish Research Council’s paper. The paper describes four principles requirements of ethnics that need to be met, i.e. the information requirement, the consent requirement, the confidentiality requirement and the good use requirement (Blomkvist and Hallin, 2015).

(40)

interviewee accepted to be interviewed, which entails the second requirement (the consent requirement).

(41)

4. Benchmarking

In this chapter, Ericsson’s, Scania’s and Company X’s way of working with DFA is going to be described. This includes reasons for working with DFA, choice of method and departments working with DFA.

Three manufacturing companies have been chosen for the Benchmark-analysis namely, Ericsson, Company X and Scania. The benchmark was conducted by interviewing industry experts in the field of DFA. The purpose of the benchmarking was to identify best practices and investigate how the firms have implemented and applied DFA.

The choice of the firms to benchmark was based on three main reasons. Firstly, the companies are in the manufacturing business, are of approximately the same size and have similar processes. Secondly, the companies have successfully implemented DFA into their organization and can be viewed as leaders in the area. In addition to this, Ericsson and Company X have implemented the DFA2-metholody provided by AviX, a DFA-method that has been studied in this thesis. Finally, all three companies have experienced similar problems as Company Y is now facing regarding the implementation of DFA.

4.1 Ericsson

Ericsson is a Swedish telecommunication and networking company. The company is one of the largest manufacturers of equipment for telecommunication. Ericsson offers a portfolio of network services, digital services, managed services and Emerging Business, which is driven by 5G and IoT platforms. The company has in the time of writing 100735 employees (Ericsson.com, 2018).

The information in this section is mainly based on several interviews with Anders Ulin (2018). Ulin (2018) is currently working with making base stations more suitable for assembly at Ericsson - a network of base stations enabling mobile phones and other mobile devices to function, through transmitting radio waves (Ericsson.com, 2018).

4.1.1 Reasons for working with DFA

(42)

• Receive feedback of previous experience from running production • Identify assembly problems early in design

In addition to this, DFA as a tool enables an objective way to confront R&D regarding weakness with the design, without the designer feeling criticized.

4.1.2 Choice of DFA method

Ericsson is currently working with the software AviX and utilizes the DFX-module provided by the company Solme. The reason for this is that Ericsson has been working with the software for preparation in production and the DFX-module was licenced due to practical reasons. The DFX-module has been customized to suit Ericsson, where several guidelines/aspects have been removed, updated and created. Before Ericsson implemented the AviX-DFX module, they used Boothroyd’s and Dewhurst’s DFMA Inc for manual assembly.

Advantages with using AviX DFX-module

According to Ulin (2018) the advantages with utilizing AviX’s DFX module is that the same tool is used for preparation in production as in the DFA evaluation sessions. Aforementioned, AviX enables visualisation of the production system in form a video clip. These video clips can be displayed during the DFA-analysis, which increases the understanding of how the products/components currently are assembled. This reduces the need for explaining the current assembly process and the reasons why products need to be designed to simplify assembly. In addition to this, the tool can be customized to the company with questions/guidelines that suit their production and their automation requirements. This in comparison to Boothroyd and Dewhurst DFMA inc, which is more standardised and developed more towards the automobile industry. Finally, as aforementioned, AviX provides an assembly time, which can be transferred

into cost of assembly.

Disadvantages with AviX DFX-module

According to Ulin (2018) the main disadvantage with AviX DFX is that the DFA-index is subjective. Depending on how the members in DFA-team argue for their own specific standpoint, the output can vary. Furthermore, the index can be manipulated during DFA-analysis to establish a better relationship with the design department. The DFA-index will thereby score higher than it should if the team was principled.

4.1.3 Departments working with DFA

Implementing Design For Automatic Assembly

Implementing Design For

Automatic Assembly

A recommendation on how to implement and

apply DFAA at Company Y

Implementing Design For Automatic

Assembly

A recommendation on how to implement and apply DFAA at

Company Y

Abstract

Sammanfattning

Table of contents

Abbreviations

Abbreviations

Glossary

Figures & Tables

Figures

Tables

Foreword & Acknowledgements

1. Introduction

1.1 Background

1.2 Problem Formulation

1.3 Purpose & Research Question

1.4 Delimitations

1.5 Study’s Expected Contribution

1.6 Disposition

2. Literature & Theory

2.1 Definitions

2.2 DFA methods

2.3 Implementation of DFA

2.4 Impact DFA has on the future of manufacturing

3. Method

3.1 Research Design & process

3.2 Primary Sources

Interviews

Title

Date

Place

Description Duration

Meetings & Other Activities

3.3 Secondary Sources

3.4 Quality of Analysis

3.5 Ethical considerations

4. Benchmarking

4.1 Ericsson

1. Introduction 