Feasibility analysis of cobots for automation of tightening stations in motor vehicle engine assembly lines

IN

DEGREE PROJECT VEHICLE ENGINEERING, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2020 ,

Feasibility analysis of cobots for automation of tightening stations in motor vehicle engine assembly lines

GABRIEL TOBLER

Abstract

The trend of increasingly automated tightening stations in the motor vehicle powertrain production industry, is driving the tooling industry to adapt its of- fering. This thesis aims to answer the question if collaborative robots (cobots) are suitable for tightening applications in motor vehicle engine assembly lines.

A collaborative robot is in principle similar to a conventional industrial robotic arm. The part that differentiates it from an industrial robot is its ability to detect and in some cases prevent a collision with a human of close proximity.

In order to answer this question three case studies are conducted where the deployment of collaborative robots is assessed. To specifically answer the case of a motor vehicle engine line, two lines are mapped station by station. Then these stations are recreated with a collaborative robot and a tightening tool and a few applications are tested on an actual motor. The rest is simulated with equations, dictating how much time is needed to complete a sequence of tightenings.

As a result, most stations on a motor vehicle assembly line could in theory be done by using tightening tools together with one or multiple collaborative robots. However, these kinds of lines are often highly automated. For this rea- son large industrial robots and multi-tool fixtures are preferred today. The large volumes of these lines, mostly mitigate the expected financial gains of replacing a human operator with a cobot. In many cases, the cobot would not have the time to complete all bolts of an application on an engine within the give takt time. Multi-cylinder engines (which is the absolute norm for passenger cars) have too many bolts for one cobot and tool to complete. If multiple cobots are to be used, the cost of the station further increases. Furthermore, re-balancing a line simply to accommodate a cobot would also not be beneficial.

Probably a much better application for collaborative robots and tightening tools

are smaller, more operator dependent lines. Here the value of a cobot can be

utilized to a greater extent. These manufacturers have greater product variance

and re-balance their lines more often. As a suggestion for further research, this

segment could be invesigated further.

Sammanfattning

Den allt ¨ okande trenden av mer automatiserade ˚ atdragningsstationer i fordon- sindustrins tillverkningslina av motorer, driver verktygsleverant¨ orer att se ¨ over sin produktportf¨ olj. Detta examensarbete har som syfte att besvara fr˚ agan om kollaborativa robotar (cobots) ¨ ar l¨ ampliga att nyttjas f¨ or ˚ atdragande ap- plikationer i produktionslinor av motorer inom bilindustrin. En kollaborativ robot ¨ ar i sin princip lik en industriell robot arm. Det som skiljer sig ˚ at ¨ ar att en kollaborativ robot har f¨ orm˚ agan att uppt¨ acka och i vissa fall f¨ orebygga en eventuell kollision med en m¨ anniska som befinner sig i n¨ arheten. F¨ or att besvara detta har tre fallstudier utf¨ orts. F¨ or att specifikt kunna besvara fr˚ agan g¨ allande motorproduktionslinor har tv˚ a linor kartlagts station f¨ or station. Sedan har n˚ agra av dessa stationer ˚ aterskapats fysiskt med en motor och en kollab- orativ robot samt ˚ atdragande verktyg. Andra stationer har simulerats med ekvationer som best¨ ammer hur mycket tid som kr¨ avs f¨ or att slutf¨ ora en sekvens av ˚ atdragningar. Resultatet visar att de flesta stationerna i en motorproduk- tionslina inom bilindustrin kan utf¨ oras av ˚ atdragande verktyg monterade p˚ a en eller flera kollaborativa robotar. Men, dessa typer av produktionslinor ¨ ar ofta v¨ aldigt automatiserade. Ofta anv¨ ands stora industriella robotar eller fixturer som inneh˚ aller m˚ anga ˚ atdragande verktyg. De stora volymerna som s˚ adana pro- duktionslinor producerar, d¨ ampar vinsterna av en f¨ orv¨ antad kostnadsbesparing.

Stora och dyra installationer ¨ ar inget problem n¨ ar produktionsoms¨ attningen ¨ ar s˚ a stor. I m˚ anga fall kan en cobot inte ers¨ atta en m¨ anniska eller fixtur inom takttiden p˚ a en motorproduktionslina d¨ ar motorn har flera cylindrar, viket ¨ ar den absoluta normen f¨ or personbilar. Om flera cobotar ska anv¨ andas f¨ or att hinna med takttiden ¨ ar kostnaden snabbt uppe i niv˚ aer som ¨ ar j¨ amf¨ orbara med en mer traditionell installation. Om en cobot skulle appliceras effektivt m˚ aste d˚ a hela produktionslinan balanseras om enbart f¨ or detta, vilket blir ett mycket dyrare projekt.

F¨ ormodligen en mycket b¨ attre applikation f¨ or kollaborativa robotar ¨ ar min-

dre och mer operat¨ orsberoende produktionslinor. H¨ ar kan v¨ ardet med cobots

b¨ attre nyttjas. Dessa tillverkare har st¨ orre produktvarians och m˚ aste balansera

om deras produktionslinjer mer frekvent. Som ett f¨ orslag f¨ or framtida studier

kring kollaborativa robotar i produktionslinor ¨ ar detta ett segment som kan

unders¨ okas ytterligare.

Preface

I would like to thank my engaging and helpful colleagues at Atlac Copco ITBA.

They have been very friendly and supportive throughout the duration of this thesis. I wont name anybody by name here so that I cannot forget anyone. The people from the Nacka office, application centers and customer centers know who they are.

Special thanks to Jonas C Andersson for being my supervisor at Atlas Copco.

Special thanks to Lihui Wang for being my supervisor at KTH.

Special thanks to Jenny Jerrelind for being my examinor at KTH.

Abbreviations

MVI Motor Vehicle Industry

GI General Industry

ISO International Organization for Standardization

Cobot Collaborative Robot

PLC Programmable Logic controller

HMI Human Machine Interface

I/O Input and Output

OEM Original Equipment Manufacturer

MBC Main Bearing Caps

CPS Crank Position Sensor

CR Connecting Rods

ROSR Rear Oil Seal Retainer

Z/T Zero Torque

R/D Rundown

B/O Back out

AI Artificial Intelligence

Contents

1 Introduction 2

1.1 Background and problem description . . . . 2

1.2 Purpose . . . . 5

1.3 Research questions . . . . 6

1.4 Delimitations . . . . 6

1.5 Expected contribution . . . . 6

2 Methodology 8 2.1 Research approach . . . . 8

2.2 Case study . . . . 8

2.3 Data gathering methods . . . . 9

2.3.1 Literature study . . . . 9

2.3.2 Criticism of sources . . . . 9

2.3.3 Interviews . . . . 10

2.4 Data analysis . . . . 11

2.5 Quality of the research . . . . 11

2.5.1 Validity of research . . . . 12

2.5.2 Generalizability of research . . . . 12

2.5.3 Reliability of research . . . . 12

3 Literature review 14 3.1 Asimov’s laws . . . . 14

3.2 Cobots . . . . 14

3.2.1 Safety . . . . 15

3.3 Tightening station types . . . . 19

3.3.1 Manual tightening stations . . . . 19

3.3.2 Semi-automatic stations . . . . 19

3.3.3 Automatic stations . . . . 20

3.3.4 Robot stations . . . . 20

3.3.5 Cobot stations . . . . 21

3.3.6 XYZ-stations . . . . 21

3.4 Torque reaction force and cobots . . . . 21

4 Empirical data 26 4.1 Motor vehicle powertrain assembly . . . . 26

4.1.1 Engine assembly line . . . . 26

4.2 Case study: SYMBIO-TIC, ASSAR, Sweden . . . . 32

4.3 Case study: Engine builder, USA . . . . 33

4.4 Case study: Mirror supplier, USA . . . . 34

5 Models 36 5.1 The collaborative robot: Cobot . . . . 36

5.2 Time to complete a station . . . . 36

6 Results 38

7 Conclusion and discussion 40

8 Suggestions for future research 42

1 Introduction

The trend of automation in the motor vehicle assembly segment drives the need for assembly stations and their tooling equipment. This study aims to asses different types of automatic tightening stations and their fitness to replace semi- automatic tightening stations or manual tightening stations. This section will introduce the study.

1.1 Background and problem description

Technology advancements, such as computing power, in combination with de- mands for highly productive assembly lines and an increase in global operator cost has contributed to an increased need for fully automated assembly stations.

These stations completely remove the operator from the station. Meaning that the process is initiated and completed automatically.

This shift in station type preference brings with it some changing requirements

for station equipment and therefore manufacturing equipment providers. The

station equipment for tightening stations involves a multitude of components

such as tooling equipment, actuators, sensors, PLC, fixtured, safety equipment

and much more. The most critical component for the quality of a joint, is the

tightening tool itself and the process which controls how the bolt is being fas-

tened. For most tooling providers with global footprint and a large product

portfolio, such as AMT, APEX, Atlas Copco, Bosch, CP, DDK, Desoutter and

Stanley, the tightening tools are sorted into one of two categories: hand held

tightening tools and fixtured tightening tools. Of course, these two categories

can have further categorization. However, the differentiation of tools that are

held with an operator by hand and those tools that are mounted in more auto-

mated station is deeply rooted in the industry. They follow different directives

and are mostly driven by different controllers and drive units. Hand held tools

are held and operated by a human operator (Figure 1). Fixtured tools are fixed

with mounting equipment and either controlled automatically by the line PLC

or by the press of a trigger by a human operator. However, they are not carried

around by an operator. They can be fixed in an industrial robot (Figure 2) or

a fixture that is pulled down to the part which is to be tightened (Figure 3).

Figure 1: Hand held station. Operator holding and controlling the tool [1]

Figure 2: Robot station. Completely automated robotic cell with tightening tools mounted on robot head

3

Figure 3: Automatic station. Fully automated cell with tools tools in fixture that is moved up and down with actuators. Controller by the line PLC.

These two categorizations of tooling equipment carry with them certain tooling characteristics that are contradicting to each other. These characteristics are important background information in order to understand the problem descrip- tion. Hand held tools are preferably light because the operator should be able to carry them in an ergonomic manner. Fixtured tools are preferably durable because they are difficult to replace since they are mounted in quite complex fixtures or robots. This also means that they become heavy due to the require- ments for durability and robustness on the materials used for all components.

This means that light weight and durability are two attributes that contradict each other and it is not possible optimize a tool for both.

Reaction forces are a similar situation. They are preferably minimized on hand held tools since operators cannot handle reaction forces equivalent or higher to 10 - 12 Nm [2]. When working with tools on a daily basis, higher exposure of vibrations and high torque will increase the risk of developing white finger syn- drome. If the reaction forces exceed the 10 - 12 Nm threshold, a hand held tool will need an external device, such as a reaction arm or bar to handle the reaction force. More on reaction forces will be discussed in Section 3.5. Because of the limitations provided by humans, hand held tools use more ergonomic tightening strategies and components. Fixtured stations are not negatively impacted by the reaction force that they are subjected to. Here it is preferred to use the most accurate strategies irregardless of reaction force. The fixture or robot have no problem handling these forces continuously for a long time.

Another attribute is the human machine interface (HMI). These are buttons and indicators on the tool that are aiding the interaction with an operator.

They are used for feedback on the operation performed, triggering the process or changing between clockwise (CW) operation and counter clockwise (CCW) operation. Fixtured tools are not in need of any HMI and to add this to the tool is thus unnecessary and cost driving. On top of that it introduces a point of failure.

Table 1 summarizes the differences in attribute preference for hand held and

fixtured tools and how every attribute that is important to prioritize for one tool type, is secondary for the other.

Table 1: Attributes preference for hand held and fixtured tools

Attribute Hand held Fixtured

Low weight Yes No

Increased durability No Yes

Low reaction force Yes No

HMI Yes No

There are station types that are seeing increased adoption among motor vehicle manufacturers. They combine a tool provided by a tool manufacturer with a collaborative robot to replace an operator or otherwise more expensive solution.

The big question here is what tool should be used together with these cobots.

They are not human, meaning they lack the need for any HMI, while simultane- ously needing more logic and I/O to communicate with other equipment. They cannot carry the same weight as an industrial robot and need to be relatively light (under 10 kg). They also need to provide a low reaction force compared to direct driven fixtured tools. This all means that the tools out on the market today are not optimized for operations with cobots. In an optimal world, the tooling manufacturers would simply design a new kind of tool which is opti- mized to use with such small robotic arms. This situation has caused the need for an industrial tool provider to conduct an investigation regarding this kind of tool. Only if the combination of cobots and tools add enough value for the manufacturing company (i.e. their customers) and sees adoption possibilities in many applications, the development of such a new tool can be justified. This could highlight a need for tooling manufactures to produce a tool that is neither optimized for fixtured use or hand held use, rather for something inbetween.

The hope is that these station types are cheaper to purchase and integrate than traditional semi-automatic, automatic or robot station. Furthermore the aim is to benchmark this kind of integration to other alternatives such as Cartesian station (also known as XYZ stations). All of these station types are explained in more detail in Section 3.3.

1.2 Purpose

With the background in mind, the purpose of this thesis is to understand imple- mentations of collaborative tightening cells and aid the evaluation of develop- ment of a tool which is optimized for collaborative robots. In order to evaluate this, two topics need to be evaluated:

• Understand the limitations and possibilities of these more cost effective alternatives to pull-down stations and robot stations

and to

• Evaluate their effectiveness on existing engine assembly lines within the motor vehicle industry

5

1.3 Research questions

The title of this thesis, Feasibility analysis of cobots for automation of tightening stations in motor vehicle engine assembly lines, aims to understand in what situations (if any) and why, cobots can be implemented as automated tightening stations. To make the purpose more tangible, two research questions can be formulated:

• RQ1: What is the current state of automation when it comes to tightening stations including cobots?

• RQ2: Can a collaborative robot, together with a tightening tool be used to assemble an engine in an motor vehicle assembly plant?

These two questions provide an analysis of the cobot and tightening landscape and can in an effort outside of this thesis aid the evaluation if a tightening tool which is optimized for collaborative stations would be a good development project to pursue.

1.4 Delimitations

In order to reach a sufficient conclusion withing the time of this project, certain delimitations have been made which are stated below. Cobots can be used for a variety of task of which tightening is just one. Other applications for these robotic arms are not considered. This will aid to focus on just tightening ap- plications and go more in depth on those. When evaluating the effectiveness, interviews are conducted with people working in the automotive manufacturing industry. Their views could be different from other industries.

The sole focus of feasibility of implementation will be on engine assembly lines, specifically of high volume lines. Other assembly lines are very different in the way their are built up, takt time and which stations are used.

1.5 Expected contribution

This thesis expects to give an investigative analysis of the tightening automation and collaborative robots landscape. This should provide a good introduction for anyone interested in the topic with many qualitative insights from the industry.

The company where this thesis is conducted should be able to come closer in

the decision if a tool specific to cobots is worth developing.

2 Methodology

In this chapter, the research approach and design is presented. Data gathering, case studies and quality are discussed.

2.1 Research approach

Blomkvist and Hallin [3] recommend students who are pursuing their degree project to use an exploratory research approach if the subject of a study is a fairly new phenomenon. This is especially recommended when questions such as how and why are to be answered. Due to the nature of this thesis, an ex- ploratory research approach is used. This will aid to a good balance between empirical data gathering out in the industry and theoretical gathering from re- ports and books.

Books, academic articles and corporate reports can be used in a deductive method, where they are used to formulate ideas which can later be tested in the study. They can also be used in a inductive method, where research is used to understand empirical findings. The inductive method has less risk to utilize theories and selected research which later shows not to be relevant for the topic [3].

Thus, this exploratory inductive method is chosen for this research. This means data gathering of relevant theory and literature are reviewed continuously [3].

As a result, literature study, data gathering and analysis are always conducted simultaneously and continuously during the project. Theory and empirical data are aligned more efficiently this way (see Figure 4 below).

Figure 4: Exploratory research approach, [3]

2.2 Case study

A case study can be a great qualitative method to enable greater variety of data sources. When trying to find many different perspectives for a project this can be a good method [4]. Case studies, if carried out well, enable deeper insight and allow for qualitative data gathering that is relevant to the topic without solely relying on theory [3] [4]. According to Denscomble, these findings will further be strengthened within an organization by use of triangulation [4].

In order to not only rely on findings from books and articles, case studies are

conducted during this thesis. This way, previous findings can be discussed and

strengthened to provide an as accurate picture as possible to the stakeholders

responsible for this investigation.

2.3 Data gathering methods

The following sections describe what data gathering methods are used in this thesis and why they were chosen.

2.3.1 Literature study

In order to gather existing knowledge and acknowledge this knowledge, a liter- ature study is conducted. This includes topics related to collaborative robots and tightening manufacturing. But is in principle not excluding adjacent fields such as industrial robotics, automotive powertrain production and final assem- bly production. This theory found in literature can help to provide a picture of how these technologies should be implemented and why. Furthermore, by un- derstanding what has been conducted previously, this research can be tailored to support future research.

Literature has primarily been provided by search services Primo and Scopus which are provided by the library. Here, a variety journals and other publica- tions can be found.

2.3.2 Criticism of sources

The literature study consists of academic journals, articles, book and internal documents from the industrial tool provider and other contacts out in the au- tomotive assembly industry. Many factors are to be considered when assessing the credibility of any given source. Literature was assessed based on its publica- tion and credibility. But mostly with use of common sense and a continuously critical view on presented data.

Academic articles construct the base of the literature study. If articles are pub- lished in an academic journal, they are subjected to peer review [4]. This can at least help to indicate credibility. Although even peer reviews are far from a guarantee of quality. There are five factors according to Denscombe, which aid in the evaluation of credibility. Journals which focus on a particular area of re- search can be more credible withing that area. The credibility of the journal also depends on the credibility of its authors and age of the journal. Furthermore, status of the journal in the research community together with the thoroughness and quality of sources used in an article, all help to determine the quality and credibility of an article [4].

Books are provided by the library and were recommended by KTH or people who were in some way part of this research during the project.

Internal documents of companies working with integration of cobots or within the production environment of components used within the automotive industry were used frequently. They provide good quality and real insight into produc- tion environment. The analysis of this data was often aided by the professionals who have created these internal documents.

Other articles and information gathered from websites is observed very carefully and, again, with a lot of common sense. News outlets are filled with scientifically

9

illiterate reporters who praise futuristic vision without prove of implementation.

Thus they were mostly avoided.

2.3.3 Interviews

According to Saunders et al., interviewing is a good method to obtained so called primary data [5]. This data is directly observed from a given source rather can from a report or book, in which case it would be a secondary source. Of course even primary sources can report findings from secondary sources, in which case it can be difficult to determine the degree of source for a given piece of informa- tion. Qualitative interviews are especially suitable when covering complex topics according to Denscombe [4]. The design of an interview, further influences the type of data that can be extracted and thereby influence the quality of an inter- view. The choice is very dependent on the situation and the relationship with the person to be interviewed. These formats can be structured, semi-structured or unstructured. For quantitative research where large data sets are required, structured interviews are a good choice. Many answers can be compared to each other and conclusions can be based on who said what and when. For more qualitative studies, these highly structured interviews can be a roadblock. They do not allow for more spontaneous directions when it comes to topic choice as semi-structured interviews. These semi-structured interviews still rely on well defined topics and a general direction. However, they can wander off to different directions than originally planned. This can of course be for better or for worse.

Unstructured interviews are based on very broad ideas and let the interviewer decide where the interview is going. This can be very useful when wanting to understand the subtopics which the interviewer wants to discover and thereby eliminate the risk of talking about something which maybe the interviewer finds irrelevant [4].

The type of interview chosen for this thesis is semi-structured. This is a very common type of interview for these kinds of topics and enable a good amount of freedom when it comes to direction of the interview without making it too unstructured so that the valuable time of the interviewee is wasted.

Interviews can further be divided in internal and external interviews. These differ in that internal interviews are conducted with people working for the industrial tool provider which has initiated this thesis. They are more easily obtainable interviewees compared to external interviewees. For external people who work at companies that conduct business with the tooling provider, inter- views are conducted even more on the structured than semi-structured side to avoid loosing focus and leaving a seemingly unprofessional impression.

The internal interviews were conducted with people working for the industrial

tooling provider. They were chosen in collaboration with the supervisor at the

company in order to ensure a good expertise hit rate. The professional roles of

the persons interviewed at the companyare are listed below in Table 2.

Table 2: Interviewees from internal collegues

Name Role Format

Interviewee 1 Proposal Engineer Video conference Interviewee 2 Business Developer Face to face Interviewee 3 Procuct Specialist Face to face Interviewee 4 Business Mangager Face to face Interviewee 5 Powertrain Expert Face to face Interviewee 6 Product Manager Face to face

The external interviews were conducted with industry professionals from var- ious car manufactures and collaborative robot manufacturers. The car manu- factures include Volvo, Ford, General Motors and Renault. Collaborative robot manufacturers include Univeral Robots, Techman, Fanuc and ABB. In order to saviour the integrity of each of these interviewees, it will not be revealed which interviewee belongs to which company. This is furthermore not relevant to the research question of this thesis. Further external interviewees include a Tier 1 supplier and cobot integrator. Details about the external interviewees are listed in Table 3 below.

Table 3: Interviewees from external industry professionals

Name Role Format

Interviewee 7 Plant Manager - OEM Face to face Interviewee 8 Innovation Manager - OEM Video conference Interviewee 9 Line responsible - OEM Face to face Interviewee 10 Line Purchase - OEM Video conference Interviewee 11 COO - Cobot manufacturer Face to face Interviewee 12 Product Director - Cobot manufacturer Face to face Interviewee 13 Integration - Cobot manufacturer Face to face

Interviewee 14 Cobot integrator Face to face

Interviewee 15 Tier 1 - Plant Manger Face to face

2.4 Data analysis

Denscombe suggests that qualitative data cannot be presented without structur- ing it [4]. The processing and analysis of the data must be conducted with care.

Without this, the data itself is useless and prone to lead to false conclusions.

There are a few steps that need to be included when analyzing data, according to Denscombe. These steps are preparation of data, getting familiar with the data (familiarizing), coding and interpreting the data, verification of data and finally presentation of data.

2.5 Quality of the research

In order for research to be of high quality, it is important that the result can be trusted. This is only possible if the result can be reproduced. Denscombe [4] means that there are three criteria in order to verify the quality of research.

These three are validity, generalizability and reliability and are presented below.

11

2.5.1 Validity of research

Validity means that theory and/or methods are suitable for any given research.

A theory in itself can be valid but when applied to the wrong phenomenon make a study not valid. It is important to contextualize research and understand purpose and validity of a theory before applying it to a study [3]. Empirical data also needs to be verified in order to carry validity. Denscombe means that triangulation is a good way of ensuring validity of empirical data. This means that one empirical data source can point in a direction. Still, it can be difficult to draw precise conclusions from that. If another empirical source points in another direction, their results have the possibility to align. More and more empirical data points can further triangulate the data in confirming or disproving prior input. In this study, multiple OEMs and cobot manufacturers were interviewed to decrease the possibility of results that are not shared within the industry. A downside of this method is that results can contradict each other, meaning that that results from interview sources contradict the data from another source.

This can make it difficult to know what data to trust [4]. In this study, this was sometimes a problem and it is discussed later in the report.

2.5.2 Generalizability of research

Qualitative studies are built on longer interviews and more open discussions with people who are judged very knowledgeable in the field. This means that sample size is often way smaller compared to quantitative studies. The reason of course being time limitations for this kinds of thesis work. In order to gener- alize correctly, big data sets are preferred [3],[4]. This downside can somewhat be mitigated be conducting multiple case studies. In order for something to be generalizable, it must be possible to transfer results and conclusions from one case to another. However is is never as general as a quantitative study with a large sample size which is well representable of the industry. This is another downside of qualitative case studies. Their representability is not certain. In this case, representability was approximated by recommendations from people with knowledge about the industry. Denscombe points out that even though a study can take measures to increase generalizability, it should be clearly under- stood that it will not be as generalizable compared to research methods with larger sample size and less qualitative nature [4].

2.5.3 Reliability of research

Blomkvist is pointing out that reliability is the main determining factor if a

given piece of research can be trusted [3]. All sources have to be critically

reviewed in order to reduce the risk of over valuing a statement which can lead

to a false conclusion. The best method to do this is transparency. This means

that the reader can individually decide if a piece of information is relevant and

reliable for a give conclusion.

3 Literature review

The literature review provides an overview of research that has been conducted which is relevant to the purpose of this thesis.

3.1 Asimov’s laws

The American author Isaac Asimov is believed to be the first creator of ethical rules for robots. They are written down in this story ”Runaround” in 1942 and were later included in a collection of stories, ”I, Robot” [6].

The three laws he wrote are:

• Law one: A robot may not injure a human being or, through inaction, allow a human being to come to harm.

• Law two: A robot must obey orders given to it by human beings, except when such orders conflict with law one.

• Law three: A robot must protect its own existence as long as such protec- tion does not conflict with law one or law two.

Isaac Asimov made an addition to the three laws by adding another law zero.

• Law zero: A robot may not injure humanity or, though inaction, allow humanity to come to harm.

It’s important to remember that these laws have their origin in fiction and are not in fact actual laws. However, their significance is thought to be quite large and their implementation is a quite tricky subject [6]. One difficult example of the implementations of these laws are military robots. They must contradict some laws to not hurt humans or humanity if they are used in wars. These ethical rules were never intended for cobots. With the evolving AI, cobots that device themselves what to do are not too far off. One case study in a later section of the thesis includes such AI that can device what work to do with a cobot.

3.2 Cobots

Collaborative robots, or cobots have existed for around 20 years now. Their form and function differ a bit depending on who is manufacturing them. The same can be said for their intended use. Cobots have the ability to preform work together, or in collaboration, with humans [7]. The name implies that collaboration is always a part of a cobot when implemented. However, the collaborative part is a mere possibility. When broken down, a collaborative robot is a robot arm that in one way or another has the possibility to detect interference. The ISO definition according to ISO 8373 of a robot is a automatically controlled, programmable multipurpose manipulator, programmable in three or more axis [8]. Furthermore it should be either mobile or fixed in place for use in industrial applications.

When looking at factories in the motor vehicle industries, robots are plentifully

and the ISO description fits them pretty well. Though they are almost always

fixed in place and fenced in due to safety concerns. The cobot on the other

hand is trying to break this fence - or barrier - between humans and robots that

today is a literal, physical fence [9].

3.2.1 Safety

In order to not need this safety barrier which more conventional industrial robots are subjected to, cobots need to ensure that they will in fact not damage other factory equipment or operators working in the plant. Danish Universal Robots was one of the first to commercially launch a cobot in 2008 with their UR5 model (see Figure 5 below).

Figure 5: UR5 demo station with tightening tool, built by Gabriel Tobler Also in 2008 another company called Rethink Robotics opened for business.

Soon after other global robot manufacturers such as KUKA, ABB or Fanuc followed with their own models. Safety features are enabled by measuring the torque in every joint of the cobot. This way the controller can understand when something is out of order. For any give load, the drive which provides power to a joint of the cobot is sending a certain amount of current to the motor. This provides the joint with enough torque to perform a movement. If the cobot is colliding with something or someone, the amount of current will increase. If this is sensed, the cobot can stop. Models such as the Dual-arm Baxter cobot can even physically flex upon impact to further reduce potential damage to any ob- ject by absorbing energy [9], see Figure 6. System errors or control malfunctions can activate physical brakes and stop the cobot from moving further until safety is assured. Furthermore, a manual brake release needs to be in place, even if the system does not have power. Systems complying with these requirements are commonly referred to as ”power and force limited by inherent design” [10]. This is described in even more detail in the ISO 10218-1:2011 section 5.10.5. This

15

standard does not include the power and force limitations itself but refers to ISO 10218-2 which is called ”safety requirements for industrial robot - part 2 [11].

This includes risk assessment of the complete cobotic application. More de- tails are available in the technical specification ISO/TS 15066:2016 [12]. These standards have also been adopted internationally with ANSI RIA R15.06-2012 [13] which is the US adoption of ISO 10218-1 and 10218-2. In order to fulfill these requirements, a quite comprehensive evaluation is required. Not only of the cobot itself but also of the complete environment of the station. These standards specify the speed of the cobot depending on the separation distance between an operator and the cobot. If a task is considered potentially dangerous, these standards recommend ways to reduce risk [9]. For these safety machine standards ISO 12100, ISO 13850 and ISO 13855 can be referenced [14, 15, 16].

Figure 6: Dual-arm Baxter cobot, [17]

Another kind of safety function can be found in one of Audi’s factories, see Fig-

ure 7. Here coolant tanks for cars are handled. The robot in orange is an MR

5 SI and has a safety system developed by KUKA. The outer most layer of the

cobot - sort of its skin - is soft and fitted with dampening pads. It includes touch

sensitive capacitive sensors which detect collision with humans by a change in

the dielectric constant before an actual collision occurs. In case they would fail,

another safety function will be activated; tactile sensors which react on physical

touch [9].

Figure 7: Cobot station, [1]

3D cameras can also be used to constantly analyze the surroundings of a cobot and detect a potential collision. This is technically also possible with a more traditional robot arms and does not comply with higher safety standards at this point.

In Table 4 a list of ISO standards regarding collaborative robots and safety is presented.

Table 4: ISO standards regarding cobots and safety

Standard Name Reference

ISO 10218-1:2011

Robots and robotic devices

Safety requirements for industrial robots Part 1: Robots

[10]

ISO/TR 20218-1:2018

Robotics

Safety design for industrial robot systems Part 1: End-effectors

[18]

ISO 10218-2:2011

Robots and robotic devices

Safety requirements for industrial robots Part 2: Robot systems and integration

[11]

ISO 11593

Manipulating industrial robots

Automatic end effector exchange systems Vocabulary and presentation of characteristics

[19]

ISO 12100:2010

Safety of machinery

General principles for design Risk assessment and risk reduction

[14]

ISO 14539:2000

Manipulating industrial robots

Object handling with grasp-type grippers Vocabulary and presentation of characteristics

[20]

ISO/TS 15066:2016 Robots and robotic devices

Collaborative robots [12]

All together there are close to 30 active EU directives and 60 standards regarding safety for manufacturing robots. Below, a summary of the most critical safety

17

standards (mainly from ISO 10218-1, ISO 10218-2 and ISO/PDTS 15006) can be found [9, 18, 11, 12].

• Control system performance: Single faults in the safety related parts of the control system are not to compromise safety. This can for example be acceded with redundancy in control signals.

• Stop functions: Robots needs to have an emergency stop function. On top of this a protection stop function is required. Protection stop brakes the robot to prevent motor from turning without cutting power. This enables quick restart of operation.

– Emergency Stop safety levels are divided into different categories depending on the ISO standard used and the PFHd (probability of dangerous failure per hour.

• Robot TCP speed control: The speed of the very most front part of the robot needs to be controlled. For collaborative work space a speed of 250 mm/s of the tool center point (TPS) should not be exceeded.

• Collaborative operation indication: If the robot is designed to be used collaboratively, a visual indication of this should be available when in collaborative mode. An exam of such a visual indication could be a light visible from all angles.

• Safety monitored stop: The cobot needs to stop when in shared work space with human operators and may resume operation upon departure of human operator.

• Handheld operator equipment: Equipment made for human operation such as screen or panels need to be equipped with an emergency stop.

• Speed and position monitoring: The ability to keep a distance be- tween operator and cobot needs to be maintained during operation.

• Power and force limiting by inherent design: The ISO standards for maximum power and force are to be kept. If exceeded e-stop should be triggered. Inherent design means the safety aspect is an irremovable element of the design.

• Power and force limiting by control system: A control system need to be in place to keep the power and force limit in place.

• Motion limits: The area in which the cobot can operate need to be de- finable with software. This mitigates the need of physical fencing, limiting the motion path the robot can use during execution of any program.

• Collision detection: If the current positions of humans or robots get so small, the separation distance decreases, this need to be detected.

• Collision avoidance: Upon detection of a potential collision, a function is needed to avoid it. This can be by altering speed or motion trajectory.

• Ergonomics: No sharp or hard surfaces should come in contact with any

human in case of a collision. Work space requirements for cobots dictate

that there needs to be sufficient space for humans to maneuver around the

cobot.

3.3 Tightening station types

As discussed in the background section, there are different types of tightening stations each with their own needs when it comes to tooling. This section will provide an overview of the most common types of stations and what defines them.

3.3.1 Manual tightening stations

Manual tightening stations are station where an operator is to hold the tool, see Figure 1. These hand held tools are similar in design to what can be expected from commercial hand held tightening tools. Traceability, productivity and durability of the tooling equipment greatly differs from tools made for other types of stations. These stations use operators to pull a trigger that starts the tightening program and also position the tool in front of the joint. The operators are also responsible for the movement of the tool, although the tool itself can be carried by a balancer which takes some of the load away from the operator.

3.3.2 Semi-automatic stations

Semi-automatic stations are still dependent on an operator to start the process.

However the movement of the tooling equipment can be handled by actuators or motors. The operator can start the process by pushing a button or dragging down a fixture with tools installed in it by pulling down handles. These stations are also referred to as pulldown stations, see Figure 8.

19

Figure 8: Semi automatic station [1]

3.3.3 Automatic stations

Automatic station completely remove the operator from the station. The trig- gering of the process is started by a signal from the line PLC. Movement of the tooling is handled by motors, actuators, robots and in some cases cobots.

These station types are typically seen in very high volume lines due to the their huge complexity and therefore cost. Rebalancing a line with many automatic stations is also more complex and costly compare to manual stations. High vol- ume lines, for example engine lines with half a million engines per year or more are a common occurrence of these stations. Furthermore, stations where multi- ple tools are used which eliminates the possibility of an operator due to weight restrictions are a common situation where automatic stations are applied.

3.3.4 Robot stations

Robot stations are a type of automatic stations. The use of heavy industrial

robots, from companies such as ABB, Fanuc, KUKA or Kawasaki, enables in-

creased flexibility due to access from multiple angles. They are always fenced

in to prevent any human from interfering with the process and thereby become

injured. An example of a robot cell can be seen in Figure 2.

3.3.5 Cobot stations

Cobot stations are getting more common in the automotive industry. They are very limited in weight capacity compare to larger industrial robots. Furthermore they are restricted in speed because they have to be able to stop in case someone get into their working environment. Then they have to be able to stop quickly and with the inertia that the movement of object bring with it, speed has to be limited. This station type is examined in more detail in Section 5.

3.3.6 XYZ-stations

XYZ-stations have three degrees of freedom. They use electric motors to move a tool. They are a slightly more cost effective solution compared to collaborative robots. This is largely due to spare part management. Since all three motors are of the same brand and model, a manufacturer only needs one motor in the spare parts inventory. Motors for this are for example offered by Yamaha and cost around $5000. A cobot costs five time that cost, or more.

3.4 Torque reaction force and cobots

The torque capacity of the tightening tooling equipment is limited when using cobots. This is due to the shut-off torque of the motors the cobots use and due to their safety functions. The joints of a cobot become smaller further out from the base of the cobot. This is because the first joint has to hold the complete cobot, while consecutive joints only have to hold the arms and motors further out from their location. These smaller motors are weaker by design. The reac- tion force caused by the tightening of a joint by a tool mounted on the cobot, will be applied on the outer most joint of the cobot. Since this is the weakest join on the cobot, it is dimensioning for the torque that can be applied on any given tightening. There are two reasons why this reaction force is bad for the cobot. Firstly, it can damage the components of the cobot because it is to weak to withstand the force. Secondly, it will cause the current monitoring function to think a collision has occurred and therefore stop and lock the cobot arm.

These are both unwanted things to happen in an production environment.

There are several ways, or methods, to reduce the strain on a cobot caused by reaction forces. They all aim to take load off the cobot. Here follow some examples.

A reaction bar is a physical bar that is designed to receive the reaction force.

The idea of a reaction bar is nothing new. It is commonly used for hand held tool equipment with torques that exceed what an operator can withstand. The bar is mounted on a fixed point in close proximity of the tool/cobot. The other end of the bar is attached to the very front of the cobot. Once the tightening is done and the toque has been applied, the reaction force will ”travel” or move.

From the bolt, through the socket and socket holder, to the mounting bracket of the tool and finally, instead of through the cobots outer most joint, it will take the path through the reaction bar. Thereby the cobot is not affected by the large force that just went through the tool.

21

The major downside of this solution is accessibility since the reaction bar is a limiting factor for the movement of the cobot. It cannot move to a point where the reaction bar cannot reach. Neither can it move in a way that will cause a collision with the cobot and the bar. See Figure 9 for how a reaction bar looks on a cobot.

Figure 9: Reaction bar mounted on cobot

A Reaction lip is a special type of front part that can be mounted on a tool.

It is adjustable and can use a second socket to engage an adjacent bolt that was tightened previously at the same time as it its engaging the bolt that is is tightening. This way the reaction force will not even travel through the tool mounting but instead go directly into the adjacent bolt. Reaction lips do not have to be equipped with a socket that engages a adjacent bolt. Sometimes its simply a short angled bar that touches an adjacent bolt or other fixed part. See Figure 10 for an tool equipped with a reaction lip socket. The downside here is less flexibility due to the complexity involved in the setup of this device. As well as the limitation in the movement of the tool due the larger physical footprint of the front part.

Figure 10: Hand held tool with reaction lip [1]

A reaction plate is another solution. Here, a fixed plate is mounted between the operator/cobot and the part where a bolt is to be tightened. The plate has a hole with splines in it exactly where the operator/cobot will move in the tool to perform a tightening. The front part of the tool has the same splines as the plate and when the tightening is performed, the reaction force travels through the plate and finally the mounting points of the plate. This solution is quite complex and requires a lot of setup to get up and running. It also hinders access to the part because it creates a physical barrier, especially if many bolts are to be tightened. In that case the plate becomes rather large and blocks visibility and accessibility to the part.

Hold and drive (HAD) is another way of solving this. In order for this to work, a special kind of bolt and nut is needed, see Figure 11. The head of the bolt is usually the part of the bolt that is turned in order to tighten it. In this case the bolt has a special indent on the other side, allowing a HAD front part to grip it and prevent rotation of the bolt.

Figure 11: Hold and drive side view [1]

A hold and drive front part has two independent axes which are inline with each other (coaxial), see Figure 12. The inner axis (yellow) is stationary and holds the bolt it order not to rotate it. At the same time and on the same axis, the green part rotates to tighten the bolt. Instead of the usual way, where the head of the bolt is turned, now the nut is turned around the bolt.

23

Figure 12: Hold and drive side view [1]

This HAD bolt together with the HAD front part make it possible to hold and tighten (drive) the joint from one side. This is useful when accessibility is limited on one side of a joint. One primary example where this is used are airplane wings, where a very tight construction prevents access from the underside of the wing. In the case of reducing reaction forces this method is also quite useful. When the nut is being turned, the joint starts to be tightened.

The the end of the tightening, the nut is clamping down the two parts which

are being tightened from one side. At the same time, the head of the bolt is

applying a clamping force from the order side. This is exactly the same as in

any other joint where a bolt and nut is used. The difference is here, the tool

that is performing the tightening and the bolt are fixed in position in relation

with each other. This is because the indent on the front of the bolt (Figure 11)

and the screw holder of the tool (yellow part - Figure 12) are engaged. Once

the tightening is near completion (when the reaction force occurs) the clamping

force is large enough to handle a large part of the reaction force which then

continues trough the two parts that are being joined. This method of reaction

force limitation is able to reduce the reaction force by around 70 %.

4 Empirical data

The following empirical data is created by observing two engine assembly line of a car manufacturer. The empirical data used in this thesis is composed of line layouts in order to have an accurate model to use later. A few case studies are also presented in the later part of this section.

4.1 Motor vehicle powertrain assembly

In order to asses the main target of this thesis, it is important to define the production environment of the motor vehicle industry. This section will define the stations used in powertrain manufacturing lines. A powertrain is considered to be the components in a vehicle that provide power to the wheels. This includes engine, transmission, differential and axles. The line layout that will be focused on here is the engine line. It is by far the most complex line type.

It is divided into the short block line and the long block line. Furthermore the volume of these lines can greatly influences the takt time (the average time between the start of production of one unit and the start of production of the next unit) and therefore the validity of the calculations later performed. The engine line considered is of very high volume, i.e 500’000+/year.

4.1.1 Engine assembly line

This section shows the layout of a typical engine assembly plant. The layout is determined by visiting two engine plants, one in the United States of America and one in Europe. Some features of these engine plants are different, due to the make and configuration of the engine, i.e. an inline four engine is not produced in the exact same order and with the same steps as an V8 engine. Still the basic steps included and operations used to perform these steps are the same. The following description of an engine line is aimed to be representative of engine lines with similar volumes. All engine manufacturer specific steps have been removed to provide a general applicable overview. Some basic understanding of the inner workings of a vehicle and internal combustion engine is required for optimal understanding. Most factories have split the engine production line into two parts. The first is the short block line which included the most basic components such as block and crank-piston assembly. The second part is the long block line which includes everything else needed to run the engine.

Other accessories not necessary for firing of the engine are installed in the final

assembly plant. Torque levels are an average of different manufacturers that

were visited for this study, they differ between different engines and furthermore,

torque and angle tightening strategies, where the final torque varies quite a

lot. This is because the tightening stops at a specific angle after the snug

point, which is when the head of the bolt is making contact with the part it

is tightening rather than only a specific torque. Below follows a visual over

of the lines followed by a step by step explanation. The lines have also been

created in Matlab in order to verify if a collaborative robot, in conjunction with

a tightening tool would be able to complete all tightenings in a station within

the time frame dictated by the takt time.

Figure 13: Short block line layout

The blocks in the schematic of the short block engine line layout, presented in Figure 13, is explained in greater detail below.

• Block load: The engine block is loaded onto the conveyor system. It has been machined in another part of the factory and has been transported to the start of the small block line. 3-4 bolts, 20-30 Nm.

• MBC B/O: The main bearing caps are already installed on the block.

They are backed out in order to make room for the crankshaft. The reason that they are already installed is because they are an exact fit with the block. Changing one MBC with another would not fit due to the lack of tolerances. 8 bolts, ~ 90 Nm.

• HDR Z/T: The harmonic dampener reluctor it attached and pre-tightened to the crankshaft. This round disc-shaped part dampens the rotational movement of the crankshaft. 3-4 bolts, ~ 0 Nm.

• HDR R/D: The dampener pulley is torqued down. 3-4 bolts, 10-20 Nm.

• MBC Z/T: The main bearings and crankshaft are inserted into the engine block. Now all main bearing cap bolts are run down to zero torque. The torque is not truly zero but rather represent the friction within the threads.

Compared to the final torque of the joint, this torque is almost zero. The number of these varies, but for modern engines it can in general be said to be (2 x the number of crank pinks) + 2. So for a inline four cylinder engine, it would be 10 bolts. For a V10 it could be 12, since two cylinders share one crank pin. Heavy engines sometimes use twice that, engines like this are used in heavy trucks or off road vehicles like from the likes of Caterpillar or Scania. (2 x number of crank pins) + 2 bolts, ~ 0 Nm.

27

• MBC R/D: Now that the main bearing caps are snug and the mating surfaces are parallel, the bolts can be run down to their final torque. (2 x number of crank pins) + 2 bolts, ~ 0 Nm.

• CPS install: The crank position sensor is secured to the engine block. 1

”bolts”, 10 Nm.

• CR Z/T: The connecting rod caps are secured to the connecting rod - piston assembly. 2 bolts per cylinder, ~ 0 Nm.

• CR R/D: Now the connecting rod caps are tightened to their final torque.

2 bolts per cylinder, 20 Nm.

• TTT: The crank-piston assembly is now installed in the block. Now the torque to turn test can be performed. Here the crankshaft is rotated and the torque required for this is measured. The torque specified for this varies but is often in the single digit Nm range.

• Ladder frame Z/T: The ladder frame is rundown. 8 bolts ~ 0 Nm.

• Ladder frame Z/T: The ladder frame is tightened to its final torque. 8 bolts ~ 60 Nm.

• Balance shaft cap B/O: For engines with balance shafts, they are now ready to be rundown and secured to the engine block. ~ 0 Nm.

• Balance shaft cap R/D: After rundown to zero torque, the balance shaft cap can be tightened to their final toque. 4 bolts, 25 Nm.

• Oil pump R/D: The oil pump is attached to the engine block. 1 bolt, 20 Nm.

• ROSR Z/T: The rear oil seal retainer can now be rundown and secured to the engine block. 6 bolts, ~ 0 Nm.

• ROSR R/D: Finally, the rear oil seal retainer is tightened to its final torque. 6 bolts, 10 Nm.

Now, the short block is completely assembled and is next step is the long block

line.

Figure 14: Long block line layout

Figure 14 shows a schematic of the long block engine line layout. It is explained in greater detail below.

• R/D Balance shaft cover: The short block is loaded onto the long block line. Here the balance shaft cover assembly is bolted onto the engine block to prevent leakage. 4 bolts, ~ 10 Nm.

• R/D Oil tube: The constant drain tube and oil separation assembly are installed and secured onto the block. 1 bolt, 10 Nm.

• R/D Oil pressure sensor: The oil pressure sensor is installed and hand started, then secured. 1 bolt, 20 Nm.

• R/D Oil pan: The oil pan is secured to the bottom of the engine block.

It covers the oil pickup, tubes and pump. The number of bolts varies a lot depending on the size of the engine. 15-25 bolts, 10 Nm.

• R/D Oil filter: The oil filter is installed. 1 bolts, 20 Nm.

• Z/T Cylinder head: The cylinder head is secured to the engine block.

The number of bolts, depends on the number of cylinders of the engine.

These bolts are often tightened to their yield point, to ensure maximum clamping force. The separation of these two very intricate parts can result in fatal engine faliure due to the oil lubrication system and water cooling

29

system interchanging fluids. Furthermore, water can get into the cylinders themselves. ~ 8 bolts, ~ 0 Nm.

• B/O Cam caps: The camshaft caps are backed out in order to make space for the camshaft to be installed in a later step. 12 bolts, ~ 20 Nm.

• R/D Cylinder head: The cylinder head is secured to its final torque. 8 bolts, ~ 100 Nm.

• R/D Cam caps: The camshaft caps are secured to their final torque.

Each camshaft cap usually has two bolts. 12 bolts, ~ 20 Nm.

• R/D VCT Phasers + bridges: The variable camshaft timing phasers and brides are installed with one and two bolts respectively. 1-2 bolts,

~ 30-50 Nm.

• Z/T Tensioner: The timing belt tenisioner assembly is secured to the engine block. It assures that the play of the timing belt (or chain) is within its specified range, in order not to slip or fall off. 1 bolt, ~ 0 Nm.

• R/D Tensioner + Z/T Guide pins: The timing belt tensioner is secured to its final torque. In the same station, the guide pins for the front cover are also installed. 1 bolt, ~ 40 Nm.

• R/D Spark plugs: The spark plugs are secured to their specified torque, one for each cylinder. 4 spark plugs, 20 Nm.

• Z/T Front cover: The front cover is placed over the guide pins and the bolts to hold it in place are rundown to zero torque. 10-16 bolts, ~ 10 Nm.

• R/D Damper pulley: The damper pulley is secured to the front of the crankshaft. It will facilitate and drive the motion of the timing belt. 1 bolts, 350-550 Nm.

• Z/T Cam cover: The camshaft cover is secured onto the cylinder head.

It prevents dirt and other unwanted substances from entering the engine and oil splashes from exiting it. The number of bolts greatly varies de- pending on the number of cylinders and especially the number of cylinder banks, which can double it. 10-16 bolts, ~ 0 Nm.

• R/D Cam cover: The camshaft cover is secured to its final torque. 10-16 bolts, 10 Nm.

• R/D Fuel rail: The fuel rail is installed onto the engine and secured. 2 bolts, 10 Nm.

• R/D Oil solenoid: The variable vale timing oil solenoid is secured. The solenoid can alter the amount of oil flow that get into the lifters. Thereby altering the lift to correspond with the respective valve timing. Mainly for optimal fuel consumption. 1 bolt, 10 Nm.

• R/D Cam position sensor: The camshaft position sensor is lubricated,

installed and secured. 1 bolt, 10 Nm.

• R/D Knock sensor and thermostat: The knock sensor is installed and secured. It is used to monitor the engine and allow the ECU to detect engine knocking (premature detonation). In the same station, the thermostat is also secured. 1 bolt, ~ 20 Nm.

• Z/T Intake manifold: The intake manifold is secured and installed to zero torque. 6 bolts, ~ 0 Nm.

• R/D front handle: The front engine handle is installed onto the engine block. It is used to lift and transport the engine later. 1 bolts, 60 Nm.

• R/D Intake manifold: The intake manifold is secured to its final torque.

6 bolts, 10 Nm.

• R/D Overflow tube: The overflow container of the cooling system has a tube connected to the engine itself. It is installed here. 1 bolt, 10 Nm.

• R/D ETB: The electronic throttle body is secured. It is connected only by electrical wires to the gas pedal of the driver. 4 bolts, 10 Nm.

• R/D Water outlet: The water outlet is secured onto the engine block.

2 bolts, 20 Nm.

• R/D CH t + R handle: The cylinder head temperature sensor is installed and torqued. In the same station the rear engine handle is also secured. It is used to lift and transport the engine. 2+2 bolts, ~ 15 Nm.

• R/D Coil packs: The coil packs are installed over the sparkplugs and secured with two bolts each. 8 bolts, 15 Nm.

• Z/T Flywheel: The flywheel is secured to zero toque onto the crankshaft.

8 bolts, ~ 0 Nm.

• Z/T Water pump: The water pump is secured onto the engine block with zero torque. It is driven by the timing belt or chain and seals the side of the engine block. 6 bolts, ~ 0 Nm.

• R/D Water pump: The water pump is tightened to its final toque. 6 bolts, 15 Nm.

• R/D WP pulley: The water pump pulley is secured to the water pump.

It is used to transfer linear motion from the timing belt to rotational motion that drives the water pump. 3 bolts, ~ 25 Nm.

• R/D Harness + g: The ground wire of the wiring harness is secured to the engine block and acts as a ground for the electrical system. 1 bolt, 15 Nm.

• R/D Flywheel: The flywheel is torqued to its final specification. 8 bolts,

~ 90 Nm.

• Z/T Clutch: The clutch is installed with zero torque. 6 bolts, ~ 0 Nm.

• R/D Clutch: The clutch is secured and torqued to its final specification.

6 bolts, 35 Nm.

31

• R/D Exhaust studs: The exhaust studs are installed into engine block.

They are double sided threaded studs where the exhaust manifold can later be attached. The engine lable is also attached here and scaned to verify the engine. 8 bolts, 12 Nm.

• Block offload: The block is loaded off the line. The turntable can now mount the engine to the transmission face plate and be mated with its transmission later. 3 bolts, 25 Nm.

Now, the long block is complete. Still, many important features are missing from this engine. They are mostly add on components from Tier 1 sub-suppliers.

They will be assembled in the final assembly plant where the engine is then put into its vehicle.

4.2 Case study: SYMBIO-TIC, ASSAR, Sweden

A case study at ASSAR in Sk¨ ovde was conducted. The goal of this case study was to better understand the work that has been done with flexible collabora- tive automation. The project is called SYMBIO-TIC and is fully sponsored by the European Union [21].

Figure 15: SYMBIO-TIC project station

A production line was built for the project. This line can carry parts that are

to be worked on. The parts were selected by a Swedish car manufacturer which

is a partner of the project. At this stage of the project the part which is being

worked on is a balance shaft assembly. Consisting of two shafts that have to be

placed in a housing and a total of eight bolts that are to be tightened. This task,

and other tasks related to the part are in a ”task-management-system”. This

system, can intelligently decide, who should perform any given task. There are

operators involved which as tracked by cameras. Depending on what they are

doing, the system can give a task to an operator or to the cobot. The robot can

then perform some tightenings if the operator is doing something else. In theory such a system could be used to have an assembly ”line” with no programmed tasks. A system would dynamically decide what task should be performed when and by who.

Furthermore there is a 3D tracking system. This system is creating a 3D repre- sentation using a couple of Microsoft Kinect 3D cameras mounted in the roof.

This system can track operators and always knows were the robot is. The robot is not a cobot, but rather a ”regular” industrial robot. This robot is gaining the functionality of a cobot with the use of the 3D camera system. Not only can the system track operators and calculate when the robot needs to stop, in case that someone is getting to close. It can also re-direct the robot with a new path, circumventing the operator.

Figure 16: SYMBIO-TIC station with ABB robot and working parts

4.3 Case study: Engine builder, USA

An American engine builder was visited in a very large factory in Michigan. This engine builder has very high volumes of around 500’000 engines a year. The short block and long block lines were observed to see if any implementation of cobots had been done. In total, only one cobot was present and it was for the purpose of parts handling rather than tightening. Upon discussing the case of cobots with the person responsible for these lines, it became clear that cobots have not been in a focus for them. Partially because there is a certain tradition in how these large scale engine lines are built. There is very little need for improvement when it comes to cost efficiency of the production equipment when the volumes are this high. At the same time, the current implementation of highly automated fixtures and large industry robots is a quite cost effective solution. Although the possibility of implementing cobots in the future is recognized, at this point

33

it is rather secondary. In the future more flexible lines with a larger number of product variant could call for the need of implementing cobots. However, this type of plant usually benefit from large volume with low product variety in order to be as efficient as possible.

4.4 Case study: Mirror supplier, USA

A Tier 1 supplier for the automotive industry was visited. Tier 1 suppliers are first tier sub suppliers, meaning they provide parts which are then assembled by the car manufacturers. This Tier 1 supplier was a mirror manufacture. Pro- viding a large variety of different side mirrors for all sorts of makes and models.

The plant floor was designed in a way that multiple smaller assembly lines were placed all over the building. These smaller assembly lines were heavily operator dependent and had around 10 stations each. After these 10 small stations, the mirror was complete. Here multiple collaborative cobots have been used.

Around two in each of the lines in the plant. One of the stations is depicted in

Figuer 17.

Figure 17: T1 Cobot tightening station

Figure 17 shows (1) a collaborative robot. This is the UR10 which is used throughout the plant. This cobot is used to move the tightening tool (3) into position. While in operation a 3D tracking system (2) is monitoring the sur- roundings of the station. Here it is checked at all times that no operator is in close proximity of the cobot or tool.

35