A profitable implementation of 
collaborative robots 

A safe and affordable way of increasing production rate 
and quality 

Bachelor’s thesis in mechanical engineering 
 

John Bredvad-Jensen 

Jakob Johansson 
 

Department of industrial and material science 

CHALMERS UNIVERSITY OF TECHNOLOGY  

Göteborg, Sverige 2023 

www.chalmers.se 



 

 
 

 

 
 

Abstract 

At Volvo Powertrain in Skövde combustion engine cylinder heads are casted at foundry 2. Most of the 
manufacturing of their products is fully automated, but the assembly of the sand cores prior to casting 
for the D6 and D4 cylinder heads is manual, since they are produced in smaller quantities. At the 
current workstation the manual assembly creates a bottleneck, due to problems with the production 
and error rates. Volvo Powertrain therefore wants a study performed about the possibility of using 
collaborative robots for the gluing of the sand cores.  

To find a suitable concept for a new workstation a requirement specification was created. With idea 
generation methods, concepts were created and later evaluated using decisions matrices until a final 
concept remained. In unison with the decision matrices. A simplified risk assessment was performed 
and data from recordings of the current production was acquired. The data acquired was then 
statistically evaluated with the three-point method. This was then used in the visualisation. After the 
visualisation a simplified ergonomic and economic evaluation was performed.  

The final concept consisted of two collaborative robots of the model FANUC CRX-25iA, one with a 
moving pedestal and one with a permanently placed pedestal. The movable pedestal opens for the use 
of the FANUC robot at other parts of the factory. The end effector of the FANUC robot has a glue gun 
attached to it that will disperse glue. A camera mounted on a beam in the workstation is used to scan 
the position of where the glue should be applied. The current workstation uses a telfer with a manual 
glue gun attached, by keeping this the flexibility is kept. In addition, preventative measures used in the 
risk assessment is added to the workstation.  

The authors recommend the implementation of collaborative robots for the manufacturing of the D6 
and D4 cylinder heads. The findings in this thesis indicates a potential increase of production rate with 
44%, with less errors and high degree of safety. Also, the ergonomic evaluation points towards that the 
operators will experiencing less strain compared to the current workstation. The risk assessment 
reveals that this solution can be implemented with acceptable levels of risks to the operators and the 
property of Volvo Powertrain. The implementation of collaborative robots is assessed to be 
economically justifiable due to the payback time being 0,6 years. The authors assess that the positive 
effects that apply to the D6 cylinder head will also apply to the D4 cylinder head. An implementation 
of the solution that the authors present would increase the knowledge at Volvo Powertrain and enable 
new exciting projects with collaborative robots in the future. 

 

 

 

 

 

 

 

 

 



 

 
 

 

 
 

Acknowledgements 

This page is dedicated to all those that have helped us and made this thesis possible.  

 

First, we would like to thank Volvo Powertrain and our supervisors Annika Theander, Marcus 
Gustafsson and Nikos Valsamidis for allowing us to do this thesis and giving us guidance throughout 
the thesis. We would also like to thank Marcus Gottschlich for providing valuable insight about 
collaborative robots and risk assessment. We also thank the operators working on the D6 station, for 
allowing us to record them and giving us valuable input of the problems with the current station from 
the operator’s perspective. 

 

We would also like to thank Peter Hammersberg our examiner at Chalmers for giving us guidance 
with this report and how to approach the thesis and our supervisor Endre Erös for providing us with 
guidance throughout this thesis. 

 

Finally, we want to thank Visual Components and Christer Ericsson for providing us with a licence for 
Visual Components. 

 

 

 

 

 

 

 

 

 

 

 

 

 



 

 
 

 

 
 

Table of contents 

Executive summary ............................................................................................................................... 1 

1 Introduction and background ........................................................................................................... 3 

1.1 Purpose ......................................................................................................................................... 3 

1.2 Problem description .................................................................................................................... 3 

1.3 The edge of knowledge ................................................................................................................ 3 

1.4 Expanding the edge of knowledge .............................................................................................. 3 

1.5 Deliverables .................................................................................................................................. 4 

1.6 Limitations ................................................................................................................................... 4 

1.7 Company introduction ................................................................................................................ 4 

2 Pre-studies and theoretical foundation ............................................................................................. 5 

2.1 Data from Volvo .......................................................................................................................... 5 

2.1.1 Error rate .............................................................................................................................. 5 

2.1.2 Current ergonomic evaluation of the workstation ............................................................ 6 

2.1.3 Process flow ........................................................................................................................... 8 

2.1.4 Glue gun data ...................................................................................................................... 11 

2.2 Interviews with the workers ..................................................................................................... 11 

2.3 Collaborative robots .................................................................................................................. 11 

2.3.1 Differences between industrial robots and collaborative robots .................................... 11 

2.3.2 Collaborative operation ..................................................................................................... 11 

2.4 Rules and regulations ................................................................................................................ 12 

2.4.1 SIS ISO 15066: 2016 ........................................................................................................... 12 

2.4.2 SS-EN ISO 10218-1: 2011 .................................................................................................. 12 

2.4.3 SS-EN ISO 10218-2: 2011 .................................................................................................. 12 

2.4.4 SS-EN ISO 13849-1: 2016 .................................................................................................. 12 

2.4.5 SS-EN ISO 13855: 2010...................................................................................................... 12 

2.5 Collaborative robot operation .................................................................................................. 12 

2.5.1 Safety-rated monitored stop .............................................................................................. 12 

2.5.2 Hand-guiding ...................................................................................................................... 13 

2.5.3 Speed and separation monitoring ..................................................................................... 13 

2.5.4 Power and force limiting by design or control ................................................................. 15 

2.6 Programming of robots ............................................................................................................. 18 

3. Method .............................................................................................................................................. 19 



 

 
 

 

 
 

3.1 Choosing a robot solution ......................................................................................................... 19 

3.2 Assembly times .......................................................................................................................... 19 

3.3 Visualisation ............................................................................................................................... 20 

3.3.1 Programming of robot ....................................................................................................... 20 

3.4 Error rate indicators ................................................................................................................. 20 

3.5 Risk assessment.......................................................................................................................... 21 

3.6 Determining robot speed and force .......................................................................................... 21 

3.7 Economic analysis ...................................................................................................................... 22 

3.8 Determining advantages and disadvantages with collaborative robots................................ 22 

3.9 Determining production rate .................................................................................................... 22 

3.10 Verification of production rate ............................................................................................... 22 

3.11 Ergonomic analysis .................................................................................................................. 22 

4. Result part 1- Concept choice ......................................................................................................... 23 

4.1 Requirement specification ........................................................................................................ 23 

4.2 Concept Generation .................................................................................................................. 25 

4.2.1 Functional tree-diagram .................................................................................................... 25 

4.2.2 Morphological table............................................................................................................ 25 

4.3 Concept evaluation .................................................................................................................... 26 

4.3.1 Elimination matrix ............................................................................................................. 26 

4.3.2 Pugh-matrix ........................................................................................................................ 27 

4.4 Final evaluation of the concepts ............................................................................................... 28 

4.4.1 Pairwise comparison .......................................................................................................... 28 

4.4.2 Further ranking of the criteria .......................................................................................... 29 

4.4.3 Kesselring matrix................................................................................................................ 30 

4.4.4 Description of the final concept ......................................................................................... 31 

4.4.5 Further development .......................................................................................................... 32 

5. Result part 2 – Performance of final concept................................................................................ 33 

5.1 Visualisation of new workstation ............................................................................................. 33 

5.2 Production rate .......................................................................................................................... 34 

5.3 Verification of production rate ................................................................................................. 35 

5.4 Waiting times ............................................................................................................................. 35 

5.5 Gluing times ............................................................................................................................... 35 

5.6 Error rate result ........................................................................................................................ 36 



 

 
 

 

 
 

5.7 Risk assessment result ............................................................................................................... 36 

5.8 Ergonomic evaluation of new workstation .............................................................................. 36 

5.9 Economic evaluation of new workstation ................................................................................ 37 

6.Discussion .......................................................................................................................................... 38 

6.1 Visualisation of new workstation ............................................................................................. 38 

6.2 Assembly times .......................................................................................................................... 38 

6.3 Production rate .......................................................................................................................... 39 

6.4 Verification of production rate ................................................................................................. 39 

6.5 Waiting times ............................................................................................................................. 40 

6.6 Gluing times ............................................................................................................................... 40 

6.7 Error rate ................................................................................................................................... 41 

6.8 Risk assessment.......................................................................................................................... 41 

6.9 Ergonomic evaluation of new workstation .............................................................................. 41 

6.10 Economic evaluation of new workstation .............................................................................. 42 

6.11 Advantages and disadvantages with the application of collaborative robots for the 
workstation ...................................................................................................................................... 42 

6.12 Method ...................................................................................................................................... 43 

7.1 Analysis of requirements and desirables ................................................................................. 44 

7.2 Evaluation of the deliverables .................................................................................................. 45 

7.3 Recommendation ....................................................................................................................... 45 

7.4 The next step .............................................................................................................................. 46 

8. References ........................................................................................................................................ 47 

9. Attachments ..................................................................................................................................... 48 

Attachment 1, Derivation of payback time ................................................................................... 48 

Attachment 2, Force and speed calculations ................................................................................. 49 

Attachment 3, Robot speed ............................................................................................................. 49 

Attachment 4, Robot force .............................................................................................................. 50 

Attachment 5, Solutions .................................................................................................................. 50 

Attachment 6, Risk assessment prerequisites ............................................................................... 51 

Attachment 7, Risk analysis preventative measures .................................................................... 52 

Attachment 8, Risk assessment part 1 ........................................................................................... 53 

Attachments 9, Risk assessment part 2 .......................................................................................... 62 

 



 

 
 

1 
 

Executive summary 
 

At Volvo Powertrain in Skövde the combustion engine cylinder heads are cast in foundry 2. The 
production of cylinder heads is fully automated except for the Volvo Penta D6 and D4 cylinder head, 
in which prior to casting, the sand cores are assembled manually. Since the sand cores are difficult for 
a robot to assemble and because they are produced in a lower quantity compared to other cylinder 
heads. At the current workstation the manual assembly creates a bottleneck in the production of the D6 
and D4. This is due to a lower production rate and a high error rate when compared to the rest of the 
cylinder heads produced at foundry 2. Most of the production errors are gluing errors. This is a 
problem because it lowers the production rate at foundry 2 and because the high error rates result in a 
high quantity of discarded moulds.  
Another problem is the ergonomics of the workstation, which can lead to fatigue and physical damage 
to the operators. The goal is to decrease the strain on the operators. 
 
With the problems and their consequences as stated above, Volvo Powertrain wants to investigate if 
there is a possibility of implementing collaborative robots for the gluing of the sand cores in the 
assembly of the D6 cylinder head. 

 
The use of both industrial and collaborative robots is well known in the world today. At Volvo 
Powertrain in Skövde the manufacturing in dominated by industrial robots, industrial robots are used 
in both foundry 1 and 2. Whereas the company's knowledge of industrial robots is substantial, the 
company's knowledge of collaborative robots is limited. This includes the different standards that are 
used, the risk assessment and general information for collaborative robots. 
 
Several methods were used to broaden the knowledge of the uses of collaborative robots. A 
requirement specification was created where desirables and requirements were established. With the 
use of idea generation methods as well as decision matrices a suitable concept was established. A 
simplified risk assessment was performed along with data acquisition of the time it takes for the 
operators to perform their tasks at the current workstation. The data was then statistically evaluated 
with the three-point method. With this information the visualisation of the new workstation could be 
performed. Lastly an ergonomic and economic analysis of the new workstation was conducted. 
 
In the requirement specification 2 main measurement values where defined. The production rate and 
the error rate. They are especially important because they tell if the implementation of the 
collaborative robots at foundry 2 is possible and if it is economically justifiable. The workflow of the 
thesis is closely related to these 2 measurement values. 
 
In the thesis relevant information that is needed to implement collaborative robots is presented. 
Summarised information of different standards, rules and regulations that can be applied to 
collaborative robots can be found. A python script for calculating the collaborative robots speed 
according to ISO 15066´s force and power limiting can be found in attachment 2. The risk assessment 
presented in the thesis is simplified but can be used as a background or a starting point for a full risk 
assessment of the new workstation. 
 
 
 



 

 
 

2 
 

The value provided by this thesis is that it shows that collaborative robots are suitable for application 
at the D6 working station. This thesis also provides value in that it expands the edge of knowledge 
meaning that Volvo can use this knowledge in their production at other places than the D6 working 
station. The implementation of collaborative robots at the D6 working station will have a payback time 
of 0,6 years and will increase production rate with 44%, as well as lowering the gluing related error 
rate. This thesis will be of interest for engineers and supervisors seeking to implement collaborative 
robots in a production flow. 
 

This thesis has shown what is possible to achieve with collaborative robots within the scope of the D6 
working station. To implement this a full risk assessment must be made, a new economic analysis 
must be made, the equipment must be purchased, and a detailed construction of the station must be 
made. After the working station has been implemented the production rate and error rate calculated in 
this thesis can be validated by measuring the error and production rate after the collaborative robots 
has been implemented. The error rate will be lowered, and the production rate will be increased as 
shown in this thesis.  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



 

 
 

3 
 

1 Introduction and background 
Here the purpose, problem description, deliverables, limitations and company introduction is 
presented. 

 

1.1 Purpose 
The purpose of this thesis is to conduct a pilot study of the implementation of collaborative robots at 
the D6 working station at Volvo Powertrain in Skövde. The cause of why the pilot study was 
performed is stated in segment 1.2. Where, except from the deliverables, the aim is to increase 
productivity, reduce error rates and ergonomic strain while still maintaining a high safety between 
operators and collaborative robots. 

 

1.2 Problem description 
The cylinder heads for the Volvo Penta D6 & D4 are casted in Skövde at foundry 2. Today this 
process is mostly automated except for the assembly of the sand cores. The cylinder heads 11, 13 and 
16 are also casted at foundry 2 and are fully automated. The D6 & D4 cylinder heads are produced in 
smaller quantities than the other cylinder heads. Today the D6 cylinder head production is a bottleneck 
in the production due to problems with high error rates and a lower production rate compared the rest 
of the production line. The foundry could produce more D6 cylinder heads if it had capacity but due to 
the problems stated above this is not possible in the current workstation. 

Another problem is the ergonomics of the workstation, where the aim is to reduce the strain on the 
operators. Volvo Powertrain wants to investigate the possibility of using collaborative robots in the 
assembly of the D6 cylinder head moulds, specifically for the gluing of the sand cores. 

 

1.3 The edge of knowledge 
The knowledge of industrial robots at Volvo Powertrain is substantial, as industrial robots is used in 
both foundry 1 and 2. However, the company's knowledge about collaborative robots is limited in 
Skövde. This includes the risk assessment, standards, and general information for collaborative robots. 
Therefore, Volvo Powertrain wants a study performed as stated in segment 1.2 to increase the edge of 
knowledge. 

 

1.4 Expanding the edge of knowledge 
With the knowledge edge defined in segment 1.3, the methods of further broadening the knowledge 
edge for the company can be established. The methods used are described in segment 3. The two main 
measurement values where defined, the production rate and the error rate. The workflow of the thesis 
is closely related to these values as they are the main indicators for if it is possible and economically 
justifiable to implement collaborative robots to produce D6 cylinder heads.  
In the thesis relevant information needed to implement collaborative robots is also presented. In the 
thesis summarised information regarding rules, regulations and standards is presented, that are 
applicable for the implementation of collaborative robots. As well as a python script to calculate the 
speed of the collaborative robot according to power and force limiting described in ISO 15066. The 
risk assessment presented is simplified but can be used as a starting point and background for when a 
full risk assessment of the workstation is conducted. 
 



 

 
 

4 
 

The value of presenting this is that the edge of knowledge regarding collaborative robots and the use 
of them in production broadens. The value of this thesis is that Volvo can use the information provided 
in the thesis in projects at foundry 2 and in other parts of Volvo Powertrain in Skövde to implement 
collaborative robots. 
 

1.5 Deliverables 
In the thesis the following shall be delivered. 

 The thesis should lead to a solution which is implemented virtually. 
 An assessment of the risk between collaborative robot, human, and equipment. 
 Advantages and disadvantages with the application of collaborative robots within the scope of 

the D6 cylinder head sand core mounting. 
 

1.6 Limitations 
Before beginning the thesis, limitations were set due to time restrictions and to confine the scope of 
the thesis. The following bullet points are the limitations set for the thesis. 

 The thesis will only consider solutions with collaborative robots. 
 The thesis shall only consider the Volvo Penta D6 cylinder head. 
 The thesis limits itself to only visualise the solutions in Visual Components. 
 The solution will not be used directly in production. 
 No physical tests will be performed to verify requirements or goals. 
 Verifications of goals and requirements will if possible be performed by reason, visualisation, 

or calculations. 
 The thesis will only discuss advantages and disadvantages with the solution. 
 In the thesis a fundamental evaluation of the risks of the presented solution is performed with 

respect to humans and other production equipment. 
 The thesis limits itself to make an economical evaluation of the solution using a fictional price 

for the cylinder head. 
 The thesis will not collect any own data regarding fault rates. 
 The thesis will use data provided by Volvo regarding fault rates. 
 The position and measurements of the floor and conveyor belts are not to be changed. 
 The white cores and black cores geometry are not to be changed. 
 The lower and upper part of the mould are not to be changed. 

 

1.7 Company introduction 
Volvo Powertrain AB is a Swedish subsidiary to AB-Volvo group that was founded in 1897. Volvo 
Powertrain AB develops and manufactures drivelines for all companies within AB Volvo e.g., Volvo 
Trucks (Wikipedia, 2023). A supervisor at Volvo Skövde stated that currently about 9500 persons are 
employed by Volvo Powertrain AB. Whereas 3800 are employed at Skövde. The development work is 
performed in Gothenburg and Lyon while the manufacturing is in Skövde and Köping. The 
manufacturing plant in Skövde produces the diesel engines for Volvo Trucks and Volvo Penta. These 
are casted in various sizes (Wikipedia, 2023). 

 

 



 

 
 

5 
 

2 Pre-studies and theoretical foundation 
In the pre-study the data given by Volvo, standards that apply to collaborative robots and 
programming methods are presented. 

 

2.1 Data from Volvo  
Here the data that was given by Volvo is presented. 

 

2.1.1 Error rate 
The error data (Volvo Powertrain, 2022-b) showed that the glue errors was the most prominent driver 
of the error rates. Where 30 % of the errors are gluing errors. The error data from the D6 cylinder head 
is presented in figure 1. The error data (Volvo Powertrain, 2022-a) from cylinder head 11 is presented 
in figure 2. 

There are 4 causes of the gluing errors:  

 Too little glue is applied. This results in the mould falling apart when the protective coating is 
applied as it is turned upside down. 

 The glue dries before the sand cores are applied, this results in the mould falling apart.  
 Too much glue is applied. This results in the glue pouring out into the mould which results in 

a distorted geometry of the cylinder head.   
 The glue is applied to the wrong location. This can result in both the mould falling apart and a 

distorted geometry of the casted part. 

These four causes in turn are caused by operator error, glue gun coking and glue gun losing its 
calibration or an error with the glue itself.  

When comparing the data from the D6 cylinder head and cylinder head 11 in figure 1 and 2, it shows 
that the gluing errors for cylinder head 11 is 0,57% compared to 30% for the D6 cylinder head. The 
mould for cylinder head 11 is glued and assembled by robots and the D6 cylinder head is glued and 
assembled by operators. 

 

Figure 1: Error rates for the Volvo Penta D6 cylinder head (Volvo Powertrain, 2022-b). 



 

 
 

6 
 

 

Figure 2: Error rates for cylinder head 11 (Volvo Powertrain, 2022-a). 

 

2.1.2 Current ergonomic evaluation of the workstation 
Data about the ergonomics of the workstation was collected from an interview with 2 operators and an 
evaluation of the ergonomics of the workstation (Volvo Powertrain, 2021). 

The interview can be read under segment 2.2. The ergonomic evaluation can be summarised by the 
following bullet points: 

 Half of the lifting are performed in the yellow zone and the other half is performed in the 
green zone. 

 The weight of each component is estimated to be under 2kg. 
 The gluing occurs in the yellow zone in 3 sets and are performed in static.  
 The largest contributor to ergonomic strain according to the ergonomic evaluation is the 

gluing of the sand cores. 
 If the operators rotate tasks within the station the gluing of the mould becomes green. 
 Operators that are short are more exposed to ergonomic strain in the yellow zone. 
 It is highly recommended that the operators rotate working tasks with each other. 
 If the operators rotate within the station, the workstation becomes green according to Volvos 

ergonomic guidelines. 

Following the bullet points an explanation of what the zone is affected by and what they mean: 

 Green zone: Normal zone 
 Yellow zone: Intervention zone  
 Red zone: Danger zone 

The different zones are governed by certain parameters. In the ergonomic evaluation these parameters 
are the frequency of the tasks per hour and for the entire day. The parameters also consist of the 
weight of the component the operators are working on/lifting, the working position, and the angles of 
the operators’ body. These values were then calculated, and several tables were used to find what type 
of zone a working task is in, this can be seen in figure 3 and 4. 

 



 

 
 

7 
 

 

Figure 3: The table of how different working angles and frequency per hour and day affect the ergonomic evaluation. (Volvo 
Powertrain ,2021) Printed with permission. 

 

 

Figure 4: The table of how different working angles and frequency per hour and day affect the ergonomic evaluation. (Volvo 
Powertrain 2021) Printed with permission. 

 

 

 

 

 

 



 

 
 

8 
 

2.1.3 Process flow 
To get a better understanding of the workstation several videos were recorded of the workstation 
during production and an interview with 2 workers were conducted. With this information and the data 
received (Volvo Powertrain, n.d.), a process analysis was made including the process flow, the current 
layout for the workstation and for the whole production line of the Volvo Penta D6 cylinder head.  

 

Figure 5: The process flow for the production of the cylinder heads (Volvo Powertrain, n.d.) printed with permission. 

In figure 5 the process flow for the manufacturing of the cylinder heads is presented. The process 
begins with the preparation of new and old sand, that will be used in the manufacturing of the sand 
cores and the lower and upper parts of the mould. After the cores and mould halves has been made in a 
core shooter and have been detoxified, they are assembled by manual labour.  This results in a 
complete mould that is ready to be used for casting. Melted iron is poured into the mould. The mould 
is then cooled until the metal has solidified. Thereafter the cylinder head gets separated from the 
mould. The sand can then be reused in a new mould. The cylinder head then goes to after-treatment 
and cleaning where it gets separated from the internal cores. This sand is later reused as well. The last 
step is additional processing. 

 

 

 



 

 
 

9 
 

 

Figure 6: The current layout of the workstation. 

In figure 6 the current layout of the workstation can be seen. The current layout consists of two 
conveyor belts on each side with 3 pallets. In the direction of the process, the first pallet contains the 
white sand cores, the second one contains the lower mould and the last contains the black cores. To 
ensure that the workers don’t fall into the conveyor belt when it's operating, a light grid is placed on 
both conveyor belts. On the outer side of the conveyor belt there are ventilation pipes that sucks out 
the sulphur dioxide to reduce the level of that chemical. Above the ventilation pipe there is a mirror 
and lights so that the operator can clearly see the other side of the lower mould. This is used for 
quality control for the fully automated cylinder heads in startups or after longer breaks in production. 
Diagonally behind the mirror is a telfer that holds up the gluing tool and hoses. The telfer can be 
turned so that the operator can have full control over the gluing tool. This is the same on both sides. 
The floor between the conveyor belts can be adjusted in height. In the middle of the workstation, there 
is a special assembly tool for the black cores. After the assembly of the black cores there is a special 
lifting tool that aids the operator when the assembly of the black cores are to be mounted in the lower 
mould. This lifting tool is allowed to be moved anywhere in the workstation with the help of a traverse 
above the workstation. 

 

 



 

 
 

10 
 

 

Figure 7: The process flow for the assembly of the white and black cores. In the top right corner, an explanation of the 
acronyms used in the process flow is present. 

The current process begins with the pallets rolling into the station. The operators do a quality check for 
defects in the cores and mould. If a defect core is found the core is thrown into a cassation bin. If they 
have saved cores from earlier cassation where not every core was damaged, they will replace the 
damaged core with one of these. 

Workflow operator 1 

If the cores and mould has passed the quality control operator 1 starts by retrieving vk1-4 (white core) 
and puts them onto the lower mould in a temporary placement. Operator 1 then retrieves vk5 and 
temporary places it on the lower mould. After that operator 1 glues and assembles the white cores, 
starting with vk1, then vk2-4 and lastly vk5. Operator 1 then starts gluing for the mounting of the 
black core assembly done by operator 2. After the assembly of black cores has been mounted operator 
1 retrieves and temporarily places vk6 and vk7 one at the time onto the lower mould. The operator 
then glues and mounts vk6 followed by vk7. Afterwards operator 1 retrieves vk8-9, temporarily places 
them, glues and mounts them separately. The assembly of the lower mould is then completed and is 
ready to be sent away to the next station. This is done by operator 1 by pressing a few buttons. The 
process then repeats on the other conveyor belt. 

Workflow operator 2 

After the cores have passed the quality control, operator 2 starts by retrieving sk1 (black core) and 
mounting it on the special assembly tool. Thereafter the same is done for sk2. Operator 2 then 
retrieves sk3-4 and mounts them in the assembly tool. After mounting sk3-4 the operator presses a 
button on the floor with their feet to hold sk3-4 in place. The same is repeated for sk5-6. The assembly 
of the black cores are then completed, and a lifting tool is brought down to pick up the assembly. The 
lifting tool with the assembly fixed to it is now moved from the assembly tool to the lower mould, 
where the black core assembly is mounted in the mould. The process then repeats on the other 
conveyor belt. 

The assembly of the black cores is done by operator 2 and is done in parallel while operator 1 is 
working on assembling the white cores. 



 

 
 

11 
 

2.1.4 Glue gun data 
The glue gun and glue hose are estimated to have a combined maximum weight of 5kg, but this 
information has a low degree of accuracy. Therefore, a safety factor of 3 was used resulting in a 
combined weight of 15kg for the glue gun and glue hose for the purpose of calculating the robot speed 
and force. The opening time for the glue gun is 0,1s per dose and 0,05 s were added for the glue gun to 
open and close resulting in 0,15s for a single dose and 0,25s for a double dose. The glue gun referred 
to is the handheld glue gun which is currently used in the manual gluing. 

 

2.2 Interviews with the workers 
According to 2 operators at foundry 2 in Skövde, the overall impression with the layout of the 
workstation is good. When gluing or placing the sand cores that are placed the furthest from the 
worker, the worker is forced to work far away from their body which results in a greater strain on the 
lower back and hips. Sometimes there can also be problems if the two workers have different height 
since the floor height is adjustable and cannot be adjusted properly for both. The operators also wanted 
brighter lights and they consider the light grids problematic since they slow down the work quite a lot. 
If the light grids are broken, the operators must go up to the control room to once again, start the 
conveyor belts. If the station is operating at full production rate the operators don’t have time to have 
conversations with each other. They sometimes feel stressed by the short drying time of the glue, 
especially when mounting the large black core assembly. The common errors according to the 
operators are gluing errors, miscast cores and that the cores are easily broken in the assembly tool. The 
sand cores cannot be pushed too hard into the lower part of the mould since then it will break. 
Sometimes there can also be problems with the glue guns as they may coke if not used for some time 
or they may lose their calibration. Therefore, the operators must continually check the amount of glue 
coming out of the glue gun. 

The operators want a bigger working area and they want to remove the light grids. Since the work also 
includes quality control, it is important to be able to work calmly since otherwise defects will appear.  

 

2.3 Collaborative robots 
Here information about the collaborative robots and operations is presented. 

 

2.3.1 Differences between industrial robots and collaborative robots 
A collaborative robot differs from an industrial robot in that they can perform collaborative operations. 
These robots must comply with ISO 10218-1 Svenska institutet för standarder, (2011-b). Collaborative 
robots are generally smaller and have rounded forms to reduce injury upon collision, they are also 
designed to reduce the risk of entrapment. They generally have shorter range, lower speed and can 
carry less load than an industrial robot. 

 

2.3.2 Collaborative operation 

Collaborative operation is described in ISO 10218-2 as an operation between robot and a person that 
share the same workspace. For a collaborative operation to be allowed it must be used for a 
predetermined task, all protective measures must be active and it must use robots with features 
specifically designed for collaborative operation, meaning that the robot must comply with ISO 
10218-1 (Svenska institutet för standarder, 2011-b). 



 

 
 

12 
 

2.4 Rules and regulations  
Here the standards that were used are introduced. 

 

2.4.1 SIS ISO 15066: 2016  
ISO 15066 describes robot operations where the robot and people share the same workspace. It shows 
how to implement collaborative robots and collaborative modes of operations, that are used so that the 
people working around the robot are safe. These are described in segment 2.5. To use ISO 15066 a 
thorough risk assessment must be made. The robot integration must meet ISO 10218-2 and the robot 
must comply with ISO 10218-1 (Svenska institutet för standarder, 2016-b). 

 

2.4.2 SS-EN ISO 10218-1: 2011 
 
ISO 10218-1 describes how a robot should be constructed to assure a safe design since this influences 
the safety of the collaborative robot implementations (Svenska institutet för standarder, 2011-a). 
 

2.4.3 SS-EN ISO 10218-2: 2011 
ISO 10218-2 describes the robot system and the robot cell. This part of ISO 10218 describes how to 
implement and make sure that the robot system and robot cell is safe (Svenska institutet för standarder, 
2011-b). 

 

2.4.4 SS-EN ISO 13849-1: 2016 
ISO 13849-1 describes safety requirements for control systems and gives guidance in how to design 
them (Svenska institutet för standarder, 2016-a). 

 

2.4.5 SS-EN ISO 13855: 2010 
Describes where safeguards should be placed considering the approach speed of the human body parts 
and how to calculate safe separation distance for machines (Svenska institutet för standarder, 2010). 

 

2.5 Collaborative robot operation  
Different kinds of collaborative modes that can be used when implementing a collaborative robot are 
introduced. At least one of the collaborative modes in segment 2.5.1-2.5.4 must be used when 
designing a collaborative operation (Svenska institutet för standarder, 2011-b). 

 

2.5.1 Safety-rated monitored stop 
A safety-rated monitored stop is described in ISO 10218-1 as follows, if no person is inside the 
collaborative workspace, then the robot operates autonomously. When a person enters the 
collaborative workspace, the robot will stop moving. The robot can resume automatic operation when 
the person leaves the collaborative workspace (Svenska institutet för standarder, 2011-a). 
 



 

 
 

13 
 

2.5.2 Hand-guiding 
The operator transfers motions to the robot via a hand-operated device, these motions are then 
converted to commands which the robot will perform.  
The robot is guided by hand and operates with the safety rated monitored speed, which is determined 
by the risk assessment (Svenska institutet för standarder, 2016-b). 
 

2.5.3 Speed and separation monitoring 
The robot will maintain a safe separation distance between the operator and itself. When the safe 
separation distance is broken the robot will stop. The safe separation distance is a function of the robot 
speed so when the robots speed decreases the safe separations distance also decreases. When the 
operator moves away from the safe separation distance, the robot will resume motion at such speeds 
that the safe separation distance to the operator is maintained. 

 

The safe separation distance is calculated using the formulas below. 

  𝑆௉(𝑡଴) = 𝑆௛ + 𝑆௥ + 𝑆௦ + 𝐶 + 𝑍ௗ + 𝑍௥  (1) 

 

𝑆௉(𝑡଴) = 𝑇ℎ𝑒 𝑝𝑟𝑜𝑡𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑒𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑎𝑡 𝑡𝑖𝑚𝑒 𝑡଴. 

𝑡଴ = 𝑇ℎ𝑒 𝑝𝑟𝑒𝑠𝑒𝑛𝑡 𝑜𝑟 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑡𝑖𝑚𝑒. 

𝑆௛ = 𝑇ℎ𝑒 𝑐𝑜𝑛𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 𝑡𝑜 𝑡ℎ𝑒 𝑝𝑟𝑜𝑡𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑎𝑡𝑡𝑟𝑖𝑏𝑢𝑡𝑎𝑏𝑙𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟’𝑠 𝑐ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛. 

𝑆௥ = 𝐼𝑠 𝑡ℎ𝑒 𝑐𝑜𝑛𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 𝑡𝑜 𝑡ℎ𝑒 𝑝𝑟𝑜𝑡𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑎𝑡𝑡𝑟𝑖𝑏𝑢𝑡𝑎𝑏𝑙𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑠𝑦𝑠𝑡𝑒𝑚’𝑠 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒. 

𝑆ௌ = 𝐼𝑠 𝑡ℎ𝑒 𝑐𝑜𝑛𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 𝑡𝑜 𝑡ℎ𝑒 𝑝𝑟𝑜𝑡𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑑𝑢𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑠𝑦𝑠𝑡𝑒𝑚’𝑠 𝑠𝑡𝑜𝑝𝑝𝑖𝑛𝑔 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒.  

𝐶 = 𝐼𝑠 𝑡ℎ𝑒 𝑖𝑛𝑡𝑟𝑢𝑠𝑖𝑜𝑛 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒, 𝑎𝑠 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑖𝑛 𝐼𝑆𝑂 13855;  𝑡ℎ𝑖𝑠 𝑖𝑠 𝑡ℎ𝑒 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡ℎ𝑎𝑡 𝑎 𝑝𝑎𝑟𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑏𝑜𝑑𝑦 𝑐𝑎𝑛 𝑖𝑛𝑡𝑟𝑢𝑑𝑒  

𝑖𝑛𝑡𝑜 𝑡ℎ𝑒 𝑠𝑒𝑛𝑠𝑖𝑛𝑔 𝑓𝑖𝑒𝑙𝑑 𝑏𝑒𝑓𝑜𝑟𝑒 𝑖𝑡 𝑖𝑠 𝑑𝑒𝑡𝑒𝑐𝑡𝑒𝑑. 

𝑍ௗ = 𝐼𝑠 𝑡ℎ𝑒 𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛 𝑢𝑛𝑐𝑒𝑟𝑡𝑎𝑖𝑛𝑡𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑜𝑙𝑙𝑎𝑏𝑜𝑟𝑎𝑡𝑖𝑣𝑒 𝑤𝑜𝑟𝑘𝑠𝑝𝑎𝑐𝑒, 𝑎𝑠 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑏𝑦 𝑡ℎ𝑒 𝑝𝑟𝑒𝑠𝑒𝑛𝑐𝑒  

𝑠𝑒𝑛𝑠𝑖𝑛𝑔 𝑑𝑒𝑣𝑖𝑐𝑒 𝑟𝑒𝑠𝑢𝑙𝑡𝑖𝑛𝑔 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑠𝑒𝑛𝑠𝑖𝑛𝑔 𝑠𝑦𝑠𝑡𝑒𝑚 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡 𝑡𝑜𝑙𝑒𝑟𝑎𝑛𝑐𝑒.  

𝑍௥ = 𝐼𝑠 𝑡ℎ𝑒 𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛 𝑢𝑛𝑐𝑒𝑟𝑡𝑎𝑖𝑛𝑡𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑠𝑦𝑠𝑡𝑒𝑚, 𝑟𝑒𝑠𝑢𝑙𝑡𝑖𝑛𝑔 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑎𝑐𝑐𝑢𝑟𝑎𝑐𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡  

𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡 𝑠𝑦𝑠𝑡𝑒𝑚. 

 

 𝑆௛ = න 𝑉௛(𝑡)𝑑𝑡 
௧బା ೝ்ା ೞ்

௧బ

(2) 

 

𝑇௥ = 𝐼𝑠 𝑡ℎ𝑒 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑠𝑦𝑠𝑡𝑒𝑚, 𝑖𝑛𝑐𝑙𝑢𝑑𝑖𝑛𝑔 𝑡𝑖𝑚𝑒𝑠 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑 𝑓𝑜𝑟 𝑑𝑒𝑡𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟 𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛, 

𝑝𝑟𝑜𝑐𝑒𝑠𝑠𝑖𝑛𝑔 𝑜𝑓 𝑡ℎ𝑖𝑠 𝑠𝑖𝑔𝑛𝑎𝑙, 𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑎 𝑟𝑜𝑏𝑜𝑡 𝑠𝑡𝑜𝑝, 𝑏𝑢𝑡 𝑒𝑥𝑐𝑙𝑢𝑑𝑖𝑛𝑔 𝑡ℎ𝑒 𝑡𝑖𝑚𝑒 𝑖𝑡 𝑡𝑎𝑘𝑒𝑠 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑡𝑜 𝑐𝑜𝑚𝑒 𝑡𝑜 𝑎 𝑠𝑡𝑜𝑝. 

𝑇ௌ = 𝑇ℎ𝑒 𝑠𝑡𝑜𝑝𝑝𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡, 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑡𝑜𝑝 𝑐𝑜𝑚𝑚𝑎𝑛𝑑 𝑢𝑛𝑡𝑖𝑙 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 ℎ𝑎𝑠 ℎ𝑎𝑙𝑡𝑒𝑑. 

 𝑇ௌ 𝑖𝑠 𝑛𝑜𝑡 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, 𝑏𝑢𝑡 𝑟𝑎𝑡ℎ𝑒𝑟 𝑎 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑟𝑜𝑏𝑜𝑡 𝑐𝑜𝑛𝑓𝑖𝑔𝑢𝑟𝑎𝑡𝑖𝑜𝑛, 𝑝𝑙𝑎𝑛𝑛𝑒𝑑 𝑚𝑜𝑡𝑖𝑜𝑛, 𝑠𝑝𝑒𝑒𝑑, 𝑒𝑛𝑑 𝑒𝑓𝑓𝑒𝑐𝑡𝑜𝑟 𝑎𝑛𝑑 𝑙𝑜𝑎𝑑. 

𝑉௛ = 𝑇ℎ𝑒 𝑑𝑖𝑟𝑒𝑐𝑡𝑒𝑑 𝑠𝑝𝑒𝑒𝑑 𝑜𝑓 𝑎𝑛 𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑜𝑙𝑙𝑎𝑏𝑜𝑟𝑎𝑡𝑖𝑣𝑒 𝑤𝑜𝑟𝑘𝑠𝑝𝑎𝑐𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑚𝑜𝑣𝑖𝑛𝑔 

𝑝𝑎𝑟𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑎𝑛𝑑 𝑐𝑎𝑛 𝑏𝑒 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑜𝑟 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑑𝑒𝑝𝑒𝑛𝑑𝑖𝑛𝑔 𝑜𝑛 𝑤ℎ𝑒𝑡ℎ𝑒𝑟 𝑡ℎ𝑒 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑖𝑠 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔  

𝑜𝑟 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔. 



 

 
 

14 
 

𝑡 = 𝐼𝑠 𝑡ℎ𝑒 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 𝑖𝑛 𝐹𝑜𝑟𝑚𝑢𝑙𝑎𝑒 (2), (4) 𝑎𝑛𝑑 (6). 

𝑆௛ = 1,6(𝑇௥ + 𝑇௦) (3) 

 

𝑆௥ = න 𝑉௥

௧బା ೝ்

௧బ

(𝑡)𝑑𝑡 (4) 

 

 𝑆௦ = න 𝑉௦(𝑡)
௧బା ೝ்ା ೞ்

௧బା ೝ்

𝑑𝑡 (5) 

 

(Svenska institutet för standarder, 2016-b). 

 

To judge if speed and separation monitoring is a suitable collaborative robot operation for the working 
station, a rough and ideal calculation of the minimum safety distance was calculated. The main 
formula is taken from ISO 15066 and is presented above in equation 1. Equation 1 is the formula that 
is supposed to be used when calculating the protective separation distance. Due to assuming an ideal 
scenario the formula can be reduced to the formula in ISO 13855, which is presented below in 
equation 6. This is possible when neglecting the penalty factors 𝑍ௗ, 𝑍௥ and 𝐶. It is assumed that there 
are no uncertainties regarding the position of the operator (𝑍ௗ) and the robot (𝑍௥). It is also assumed 
that the intrusion distance (𝐶) is 0 m. This can be achieved by placing the scanner in such a way that 
the height of the upper edge of the detection zone is 2,6 m. This can be seen in table 1, page 18 in ISO 
13855. In essences the scanner will have a detection zone so that an intruding body part won't be 
possible thus 𝐶 is 0. 

𝑆 = 𝐾(𝑇ெ + 𝑇ௌ) (6) 

  

Where: 

𝑆 = 𝑀𝑖𝑛𝑢𝑚𝑢𝑚 𝑠𝑎𝑓𝑒𝑡𝑦 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒   

𝐾 = 𝐴𝑝𝑝𝑟𝑜𝑎𝑐ℎ 𝑠𝑝𝑒𝑒𝑑  

𝑇ெ = 𝑆𝑡𝑜𝑝𝑝𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑚𝑎𝑐ℎ𝑖𝑛𝑒 𝑜𝑟 𝑠𝑦𝑠𝑡𝑒𝑚  

𝑇ௌ = 𝑅𝑒𝑠𝑝𝑜𝑛𝑠𝑒 𝑡𝑖𝑚𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑎𝑓𝑒𝑡𝑦 𝑙𝑎𝑠𝑒𝑟 𝑠𝑐𝑎𝑛𝑛𝑒𝑟  

(Svenska institutet för standarder, 2010). 

When calculating 𝑇ௌ a SICK s3000 scanner was used. It has as best a response time of 60ms when 
neglecting penalty factors (SICK, 2022). The approach speed 𝐾 of the operator is 1,6 m/s Svenska 
institutet för standarder, (2016-b). Assuming that the robot is running at the TCP speed of 250 mm/s 
then the stopping time 𝑇ௌ is 840 ms for joint 2 (FANUC CORPORATION, 2020).  

Inserting the values gives us: 

𝑆 =
1,6(60 + 840)

1000
= 1,44 𝑚 



 

 
 

15 
 

This means that the robot will stop performing its operation if an operator enters closer than 1,44m of 
the robots end effector. Due to not knowing where position of the robot is, this means that the 
operators cannot be in the marked zones when the robot is operating as seen in figure 8.   

 

Figure 8: Workstation with speed and separation monitoring, the grey circles are where the operator are not allowed to 
stand during robot operations. 

 

2.5.4 Power and force limiting by design or control 
Power and force limiting is a mode of operation where the robots speed and force is limited so that the 
operator will not be hurt if collision would occur between operator and robot. In this mode of 
operation, the operator and the robot can work in collaboration. To be allowed to operate in this mode 
the robot must comply with ISO 10218-1. 

To determine maximum speed and force of the robot a risk assessment is made to determine which 
body parts that the robot may collide with. Calculations are then done to determine maximum speed 
and force that the robot may use. These calculations do not need to consider the body parts where the 
risk is assessed as acceptable in the risk assessment.  

Table 1 below shows the maximum pressure and forces that may be applied to different body parts, for 
both quasi static contact and transient contact. 

 



 

 
 

16 
 

Table 1: Table A2 from ISO 15066 (Svenska institutet för standarder, 2016-b). 

 

Table 2: Table A3 from ISO 15066 (Svenska institutet för standarder, 2016-b). 

 

 



 

 
 

17 
 

Using values from table 1 and 2. The maximum amount of energy that can be transferred to each body 
part can be calculated. 

𝐸 = 𝑇𝑟𝑎𝑛𝑓𝑒𝑟 𝑒𝑛𝑒𝑟𝑔𝑦. 

𝐹௠௔௫ = 𝑀𝑎𝑥𝑖𝑚𝑢𝑚 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑓𝑜𝑟𝑐𝑒 𝑓𝑜𝑟 𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝑏𝑜𝑑𝑦 𝑟𝑒𝑔𝑖𝑜𝑛. 

𝑃௠௔௫ = 𝑀𝑎𝑥𝑖𝑚𝑢𝑚 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑓𝑜𝑟 𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝑏𝑜𝑑𝑦 𝑎𝑟𝑒𝑎. 

𝐾 = 𝐸𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑝𝑟𝑖𝑛𝑔 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑓𝑜𝑟 𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝑏𝑜𝑑𝑦 𝑟𝑒𝑔𝑖𝑜𝑛. 

𝐴 = 𝐴𝑟𝑒𝑎 𝑜𝑓 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑟𝑜𝑏𝑜𝑡 𝑎𝑛𝑑 𝑏𝑜𝑑𝑦 𝑟𝑒𝑔𝑖𝑜𝑛.  

𝐸 =
𝐹௠௔௫

ଶ

2𝐾
=

𝐴ଶ𝑃௠௔௫
ଶ

2𝐾
 (7) 

   

 

Then the maximum relative speed and force can be calculated using the formulas below.  

𝑚ு = 𝑇ℎ𝑒 𝑒𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑡ℎ𝑒 ℎ𝑢𝑚𝑎𝑛 𝑏𝑜𝑑𝑦 𝑟𝑒𝑔𝑖𝑜𝑛. 

𝑀 = 𝑇ℎ𝑒 𝑡𝑜𝑡𝑎𝑙 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑚𝑜𝑣𝑖𝑛𝑔 𝑝𝑎𝑟𝑡𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡. 

𝑚௅ =  𝐸𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑝𝑎𝑦𝑙𝑜𝑎𝑑 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑠𝑦𝑠𝑡𝑒𝑚. 

𝜇 = the reduced mass of the two − body system, which is expressed by Formula 9 

𝑚ோ =  
𝑀

2
+ 𝑚௅  (8) 

 

𝜇 = ൬
1

𝑚௛
+

1

𝑚ோ
൰

ିଵ

 (9) 

 

𝐸 =
𝐹ଶ

2𝐾
=

1

2
𝜇𝜈௥௘௟

ଶ  (10) 

 

𝐹 =  𝑇ℎ𝑒 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑓𝑜𝑟𝑐𝑒 𝑡ℎ𝑎𝑡 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑚𝑎𝑦 𝑢𝑠𝑒. 

𝜈௥௘௟ = 𝐼𝑠 𝑡ℎ𝑒 𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑠𝑝𝑒𝑒𝑑 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡ℎ𝑒 𝑟𝑜𝑏𝑜𝑡 𝑎𝑛𝑑 𝑡ℎ𝑒 ℎ𝑢𝑚𝑎𝑛 𝑏𝑜𝑑𝑦 𝑟𝑒𝑔𝑖𝑜𝑛. 

(Svenska institutet för standarder, 2016-b). 

 

 

 

 

 

 

 

 



 

 
 

18 
 

2.6 Programming of robots 
Robot programming can be done in two main ways, offline programming and online programming.  

Offline programming is when a robot program is created without using the robot, using computer 
software to create programs. This is often done using virtual simulation software.  

Online programming is when the robot is used to create the robot program. Online programming can 
be done using both lead through programming and teach pendants.  

Lead through programming is used to manually guide the robot arm by hand and then using these 
motions to create the robot program. 

Teach pendants are commonly used to program robots. They are handheld devices which are included 
with the robots control systems. They usually have a keypad or a touch screen which is used to enter 
instructions to the robot (Robots Done Right, 2023). 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



 

 
 

19 
 

3. Method 
Here the methods used in the project is presented, varying from establishing the final concept to 
determining the production rate. 

 

3.1 Choosing a robot solution 
During the thesis several different methods has been used to choose a suitable robot for the 
application. First a pre-study was conducted to broaden and increase the understanding of the problem 
and desirables with the workstation. This can be read in segment 1 and 2. From this information the 
workstations functions, requirements and desirables were established. Part solutions for each part 
function was established and summarised in a functional diagram. Requirements and desirables were 
summarised in a requirement table. To generate concepts a program called Morpheus was used, which 
systematically combines the part solutions from the functional diagram with each other to generate 
concepts.  

When the concepts had been created, they were ranked by 3 different matrices. This was done to 
eliminate concepts until a final concept was acquired. First an elimination matrix was used to 
eliminate all solutions that doesn’t meet the requirements from the requirement table. Afterwards a 
Pugh-matrix was used to evaluate the concepts with respect to the desirables and their ranking from 
the requirement table. Thereafter a Kesselring matrix was used where certain criteria was given further 
evaluation of their own ranking as well as being compared to each other. From these 3 matrices a final 
concept was acquired. The matrices are described in segment 4. 

 

3.2 Assembly times 
To get a better understanding for the process flow of the workstation, each operation that the workers 
did was broken down into groups. These groups were then broken down further to find each single 
task of the workstation. These tasks were then timed. The timing of each task was done by recording 
13 videos of the workstation when it was in use. Then looking at the recordings and writing down the 
time for each task in Excel. This was done for all 13 videos and the results were compiled and the 
average time was taken. Then the three-point estimation method was used to find the expected time for 
each task. The three-point estimation method has the following formula: 

𝑒 =
𝑎 + 4𝑚 + 𝑏

6
 (11) 

 

Where e is the expected value of each task, a is the optimistic value that has an occurrence of  
ଵ

ଵ଴଴଴
,     

b is the pessimistic value that has an occurrence of 
ଵ

ଵ଴଴଴
 and m which is the most likely value which in 

this case is the average time per task. The three-point estimation method uses a beta distribution. The 
advantage of using this is that it accounts for the variation and not only the most likely value. Usually, 
the most likely case is closer to the optimistic case than the pessimistic. The three-point estimation 
method accounts for this and gives a more realistic value which is the expected value. To note here is 
that the values used for a and b were the fastest and slowest times recorded. The time for each task was 
measured in seconds.  

From the three-point estimation method the standard deviation s, is calculated by the formula:  

𝑠 =
𝑏 − 𝑎

6
 (12) 



 

 
 

20 
 

Then the variance is calculated by the formula: 

𝑣 = 𝑠ଶ (13) 

   

The variance is then summed up from all tasks 𝑣௦௨௠. To find the total standard deviation 𝑠௦௨௠ it is 
calculated by taking the square root of the total variance. To then find the high range and low range of 
the expected time the following formula is used.  

𝐻𝑖𝑔ℎ 𝑟𝑎𝑛𝑔𝑒 = 𝑒 + 3 ∗ 𝑠 (14) 

 

𝐿𝑜𝑤 𝑟𝑎𝑛𝑔𝑒 = 𝑒 − 3 ∗ 𝑠 (15) 

(Hammersberg, n.d). 

3.3 Visualisation 
Visual Components was used to visualise the assembly of the moulds. This program was used since 
there was a lot of available material making it easy to learn. The visualisation uses 12 different nodes. 
There are 2 feeder nodes which create parts, 7 work nodes where the workers pick up or assembles 
parts, 2 sink nodes are used to remove finished moulds from the simulation and one node is used to 
decide which side the operators should work on. The simulations also use 11 lamps which toggle 
between true and false, these are used as global variables for the simulation. The nodes use a set of 
predefined statements from Visual Components to create and delete parts. These 12 nodes also control 
when the workers work on different tasks and when the robots will start each gluing sequence. The 
gluing sequence have been programmed using jogging in Visual Components. The visualisation uses 6 
conveyor belts, 3 on each side to transport the parts between the different nodes.  

The tasks that the operators perform in the visualisation comes from the three-point estimate. The 
operators use a walking and turn speed that is based on how fast the operators turn and walk in reality. 
The velocity of the operators was calculated with the measured distance and the data from the three-
point estimate. 

To optimize the visualisation an effort was made to minimize the waiting times by redistributing some 
tasks so that they could be performed during the previous waiting time. 

 

3.3.1 Programming of robot 
The robot was programmed using jogging in Visual Components. By moving the robot in the 
visualisation and then saving its position this generates a script which the robot will run from.  

 

3.4 Error rate indicators 
It will not be possible for this thesis to show an actual reduction of error rates since that would require 
implementing the solution. It will however be possible to use indicators to show a likely quality 
outcome. To do this the D6 cylinder head error rates will be compared to the error rates of cylinder 
head 11 whose production is fully automated. 

 



 

 
 

21 
 

3.5 Risk assessment 
A full risk assessment will not be done since that would be out of scope for this thesis. Important to 
note is that the risk assessment in this thesis is simplified and cannot be used in an implementation of 
the solution proposed in this thesis. The results achieved may differ greatly from a full risk assessment 
and as such can only be used as an indication of the results a full risk assessment may yield. 

The first step in doing the risk assessment was to layout the prerequisites for the risk assessment, 
meaning laying out the layout, components, and work steps. Then brainstorming was used to 
determine the risks. The risk where the robot may hit or entrap the operators were divided into 
different body parts according to ISO 15066. The risks were then evaluated and received a rating 
between 1 and 5 for probability and consequence, were 1 is a low probability/low consequence and 5 
is a high probability/ high consequence. These were then multiplied to receive a risk value. Preventive 
measures were then identified, which was done before the second risk evaluation to reduce the amount 
of safety measures. In the second risk evaluation the goal was to rate all the risks as green. To get a 
green rating the risk value must be 3 or less or have risk value of 4 where both the probability and 
consequence rating is a 2. If the risk value was 4-9 then the risk was marked yellow unless both 
probability and consequence were marked as 2. A red rating was received if the risk value exceeded 9. 
A green rating means that the risk is assessed as acceptable and shall be reduced if opportunity 
presents itself. A yellow rating means the risk requires action and a red rating means that the risk 
requires direct action. The preventive measures were applied to all that that had a risk value of more 
than the goal value. New values for probability and consequence were then devised. When 
determining probability and consequence the reasoning for the different values were written down so 
that one could go back and see how a value was chosen. When doing the risk assessment, reduced 
speed and force is to be avoided as a safety measure since the robot will be limited in speed and force 
by using this preventative measure. After the risk assessment is done the robots speed and force will 
be calculated. Using power and force limiting; all the risks where body impact will occur that are not 
limiting with respect to the robot speed and force will receive reduced speed and force 
countermeasures, further lowering the risk value since the robot will be moving at a lower speed and 
force. 

 

3.6 Determining robot speed and force 
To make it possible for the robot to use collaborative operation while gluing the sand cores, the power 
and force limiting mode was chosen. It was chosen because speed and separation monitoring would 
require a separation distance which is too large for the station as can be seen in figure 8. Hand guiding 
could not be used since it cannot work with a human in the collaborative workspace. This mode is 
only for teaching the robot movements and cannot be used in production. Safety rated monitored stop 
could not be used because the separation distance would be too large for the station. 

To determine the robot speed and force a python script was written to calculate the speed and force 
using equations and table values from segment 2.5.4. The python script can be seen in attachment 2. 

In the formulas used to calculate the maximum speed that the robot is allowed to run, it is not actually 
the robot speed that is calculated. Instead, it is the relative speed of the robot that is calculated. An 
assumption was made that the operators speed was set to 0 and thus the relative speed became the 
robot speed. 

 



 

 
 

22 
 

3.7 Economic analysis 
A simplified economic analysis will be performed using percentages since production rate cannot be 
published in this thesis and the production costs are unknown. As such a formula will be delivered 
which is used to calculate the payback as accurately as possible with the known data and assumptions. 

 

3.8 Determining advantages and disadvantages with collaborative robots 
The advantages and disadvantages with collaborative robots are determined with the knowledge 
acquired in doing this thesis. 

 

3.9 Determining production rate 
The production rate was determined from the visualisation. This was done by measuring the time 
between sending away the first mould to sending away the second mould. This was done by adding the 
18 first completed moulds on the left and right side excluding the first mould. The first mould was 
discarded since it can be treated as an anomaly. After adding the time for the 18 first moulds, the time 
were divided by 18 to acquire the mean production rate. The reason that the first mould can be treated 
as an anomaly is that the workers start working on the same side at the same time, this makes it so that 
the mould takes longer to assemble. For the following moulds operator 2 will start working before 
operator 1 meaning that the assembly time will be faster. 

 

3.10 Verification of production rate 
To verify the production rate in the visualisation it was compared to a production rate calculated in 
Excel. This was done by adding the expected values from the three-point estimate and then adding the 
times from the new tasks. The times for the new tasks were taken from the visualisation. The 
visualised production rate and calculated production rate were then compared to each other. The 
reason why the verification was performed was to see if the production rates coincided with each 
other, to check that the visualisation was programmed correctly. 

 

3.11 Ergonomic analysis 
To evaluate the ergonomics at the new workstation, the ergonomic analysis was divided into four 
parts. The strain on each operator and the number of bends of operator 1s back on the left and right 
side of the workstation. The strain on each operator is defined as the number of tasks the operators 
perform. These criteria are then used in the decision matrices. 

 

 

 

 

 

 

 



 

 
 

23 
 

4. Result part 1- Concept choice 
From the information gathered in the pre-study, which can be in segment 2, the workstations 
functions, requirements and goals could be established. These were summarised in a functional tree-
diagram, morphological table, and a requirement table. With the morphological table, concepts could 
be generated. These were later evaluated to find the most suitable concept for the workstation. Several 
methods were utilized, some of them are the morphological-matrix, elimination-matrix, Pugh-matrices 
and a Kesselring-matrix. How the methods work together are discussed in segment 3.1. 

 

4.1 Requirement specification 
In the requirement table the requirements and desirables of the workstation are presented which can be 
seen in table 3. They are formulated in such a way that they are measurable which can be seen in the 
columns to the right. Some criteria are requirements and other are desirables, the reasoning behind this 
is that the requirements are criteria’s that are so important that they must be fulfilled. Desirables on the 
other hand must not be fulfilled. The desirables are graded on a scale from 1-4 with 1 being the least 
desirable and 4 being the most desirable. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



 

 
 

24 
 

 

 

 

 

 

Table 3: Requirement specification. 

 

 

 



 

 
 

25 
 

4.2 Concept Generation 
Here the concept generation is presented and how it was performed. 

 

4.2.1 Functional tree-diagram 
In the functional tree-diagram the workstation within the scope of the thesis was split into 
subfunctions and each subfunction is then broken down into sub-solutions. The main function within 
the scope of the thesis is gluing of sand cores, which splits into the subfunctions: ability to apply glue, 
to reach, to be placed, to move and position tool and lastly identifying the position of the lower mould. 
The functional tree-diagram can be seen in figure 9. 

 

Figure 9: Functional tree-diagram for the workstation. 

 

4.2.2 Morphological table 
The morphological table derives from the functional tree-diagram. The table consists of subfunctions 
and sub-solutions. One sub-solution from each subfunction was systematically combined to create 
concepts. This was done with a program called Morpheus. A total of 42 concepts was created which 
can be seen in attachment 5. The table can be seen in table 4. One of the 42 concepts are also presented 
below in table 5. More information regarding Morphological tables can be found in Johannesson et al. 
(2013). 

Table 4: Morphological table. Position 1,2 and 3 refers to the robot's position behind each table, position 1 is behind the 
white core table, position 2 is behind the mould table and position 3 is behind the black core table. 

 



 

 
 

26 
 

Table 5: Concept 34. 

 

 

4.3 Concept evaluation 
Here the concept evaluation process in presented. This includes elimination-, Pugh-matrices and also 
the reasoning behind the evaluation. 

 

4.3.1 Elimination matrix 
The elimination matrix evaluates the concepts with regards to the requirement specification. The 
concepts that did not fulfil the requirements were eliminated. The concepts that were eliminated, were 
eliminated because they had to low reachability. This resulted in 36 concepts being eliminated. The 
elimination matrix can be seen in table 6. More information regarding elimination matrixes can be 
found in Johannesson et al. (2013). 

Table 6: Elimination matrix for the concepts. 

 

 

 

 

 

 

 



 

 
 

27 
 

4.3.2 Pugh-matrix 
After the elimination matrix there were 6 concepts remaining. The remaining concepts were then 
evaluated using two Pugh-matrices. The Pugh-matrices evaluates the concepts comparing them with a 
reference concept and the desirables from the requirement table, where + means the concept is better 
than the reference, - if it's worse and 0 if they are equal. The Pugh-matrix 1 and 2 can be seen in tables 
7 respectively 8. More information regarding Pugh-matrixes can be found in Johannesson et al. 
(2013). 

 

Table 7: Pugh-matrix 1. 

 

Table 8: Pugh-matrix 2. 

 

 

 



 

 
 

28 
 

4.3.2.1 Reasoning in Pugh-matrix 
In the Pugh-matrices the criteria’s production rate and error rate were set as secondary criteria or SC 
due to them not being able to get evaluated until later in the Kesselring matrix, after the power and 
velocity calculations from the risk assessment was completed. Most of the criteria for each concept 
had the same score due to the concepts being equal. The criteria of being able to be mounted vertically 
was evaluated with data that was collected from the different manufacturers. Where the Yaskawa 
performed worse because it lost operating range for an axis if the robot was tilted more than 30 
degrees (Yaskawa, n.d). In the movable robot criteria, a concept performed better if it was possible to 
move the robot in comparison to only having a fixed base. To move the robot is beneficial because the 
foundry doesn’t manufacture the D6/D4 all the time, so when the D6/D4 isn't produced the robot can 
be used elsewhere. The cost of having a movable pedestal were estimated by the calculations made in 
table 9. The total cost was estimated at 20000 SEK (about 2000 €). This cost was then used in unison 
with the robot prices acquired by the manufacturers to evaluate the cost criteria. In the first Pugh-
matrix the concepts that performed the worst can be eliminated, but due to having so few concepts the 
choice of keeping them to the second Pugh-matrix was made. 

Table 9: Cost calculations for the moveable base. 

 

 

4.4 Final evaluation of the concepts 
In this chapter a pairwise comparison, further ranking of the criteria, the Kesselring matrix, and a 
description of the final concept will be presented.  

 

4.4.1 Pairwise comparison 
In the pairwise comparison the criteria are evaluated with respect to each other. The objective is to 
find which criteria that are more important than others. The ranking scale is 1 if the criteria is better, 
0,5 if it's equal and 0 if it's worse. Each criteria row was summed up creating a sum-column. That 
column was then summed up as well to a total sum-value (45). To find the relative sum (W) each 
criteria’s summed row was divided by the sum-value. The greater the value, the more important the 
criteria are. The table can be seen in table 10. 

 

 

 

 

 



 

 
 

29 
 

Table 10: The pairwise comparison for the criteria. 

 

 

4.4.2 Further ranking of the criteria 
In the further ranking of the criteria, each criterion is divided into values that are given a certain 
ranking value. The ranking value are from 1 to 6, where 1 is least desirable and 6 is most desirable. 
The value 1 is based on the current workstation’s performance. Some criteria were only divided into 1 
or 6 because either they can achieve the criterion or not. The objective of this is to further quantify 
what is more desirable within each criterion and in comparison to other criteria. The secondary criteria 
production rate and error rate could now be quantified due to the risk assessment and visualisation 
being completed. By completing the risk assessment, the maximum velocity and force that the 
collaborative robot could use was acquired. These parameters in unison with the visualisation of the 
new workstation yielded in that the secondary criteria could be quantified. Later in the Kesselring 
matrix this was used to find the final concept. The table of the further ranking of the criteria can be 
seen in table 11. 

 

 

 

 

 

 

 

 



 

 
 

30 
 

Table 11: Table of the further ranking of the criteria. 

 

 

4.4.3 Kesselring matrix 
With the use of the pairwise comparison and the further ranking of the criteria the Kesselring matrix 
can be established. The concepts were weighted with the help of table 10 and 11. An ideal concept was 
created so that the concepts could be compared to it. The Kesselring matrix can be seen in table 12. 
More information regarding Kesselring matrixes can be found in Johannesson et al. (2013). 

Table 12: Kesselring matrix. 

 



 

 
 

31 
 

In table 12 all criteria except cost and movable robot have the same ranked value. This is because the 
concepts behaved similarly to each other, the concepts had similar prerequisites to reach the criteria. 
The sum of the values T for each concept are similar, this is due to all the other concepts that are 
worse have already been eliminated, resulting in the remaining concept solutions being similar to each 
other. The main difference is if the robot will be having a movable pedestal or not and if the robot type 
will be a FANUC CRX-25iA or an UR20. The final concept is concept 2 which is implemented in the 
visualisation of the working station. However, the movable pedestal will only be used on the left side 
of the working station where the robot can easily be moved. On the right side of the working station a 
forklift cannot move the collaborative robot because its boxed in between other working stations and 
thus the robot is not easily moved and will use a fixed pedestal. 

 

4.4.4 Description of the final concept 
The concept contains a FANUC CRX-25iA with a glue gun mounted as the end effector. The position 
of the FANUC CRX-25iA is on the outside and placed in the middle, closest to the lower mould pallet. 
This allows the current telfer with the manual glue gun attached to be used when running tests or if the 
collaborative robot fails. The concept also involves a camera that is mounted on a beam above that 
will scan the position of the lower mould as well as the collaborative robot. There are several reasons 
for this. One is that the camera will have a broad view so that the robot applies glue to the right 
positions. Another reason is that when the camera is not mounted on the robot the scan time reduces 
significantly due to the robot’s speed limitation. In figure 10 the fixed and movable pedestal can be 
seen, where the movable pedestal is represented by the orange box.   

 

Figure 10: A screenshot of the new layout of the working station. 

 

 

 



 

 
 

32 
 

4.4.5 Further development 
To further develop the final concept, preventative measures that was implemented in the risk 
assessment is added to the workstation. This is presented in the bullet points below. 

 Protective elliptical housing around the glue gun. 
 Protective fences around the collaborative robots. 
 Guide rails for the collaborative robot's movable pedestal.  
 Control panel for the activation for the gluing sequence. 
 Control panel for the collaborative robots. 
 Switch for the telfer with manual glue gun. 
 Emergency stop connected to a line running across the inside of the conveyor belt.  
 Mirrors that can slide away when the collaborative robot is in use. 
 Change in the geometry of the ventilation pipes. 
 New fixture for the lights. 

 

The risk assessment is presented in attachment 6-9.   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



 

 
 

33 
 

5. Result part 2 – Performance of final concept 
Here the result of the thesis is presented including visualisation, production rate, ergonomic 
evaluation, and economic evaluation.   

 

5.1 Visualisation of new workstation  

 

Figure 11: New workstation. 

The new workstation layout was visualised in Visual Components. The visualisation consists of three 
conveyor belts placed on each side for a total of 6 conveyor belts. The lifting tool is represented by the 
yellow box in the middle. The moveable pedestal is represented by the orange pedestal and 2 
collaborative robots were added. 2 humans were also added to represent the operators working in the 
station. The humans are set to a speed so that the walking times are the same as in reality. The 
simulation uses 12 nodes. 2 feed nodes, 6 work nodes, 2 sink nodes, a node for the lifting tool and one 
controlling when the workers should start working on the other side. The new process flow can be 
seen in figure 12. 



 

 
 

34 
 

 

Figure 12: New process flow. 

 

5.2 Production rate  
The visualised total production rate for the new workstation could be increased by 51,2% but due to 
the rest of the foundry having a production rate of 144% the production rate of the new working 
station is set to 144% instead of 151,2%.  

The right and left side of the workstation differ slightly in time to complete the production cycle. The 
production rate for the left side is 49,8% and the right side is 54,7% in the visualisation. This is due to 
the lower mould's orientation being the same. Resulting in the robot having further to travel on the left 
side of the workstation.  

The production rates are summarized table 13: 

Table 13: The increase in production rate for the new workstation with respect to the measured production rate of the 
current workstation. 

 

 

 

 

 



 

 
 

35 
 

5.3 Verification of production rate 
To verify the visualisation's production rate, the production rate for operator 1 and operator 2 was 
calculated. This can be seen in table 14. The verification was made on the left side, but the same 
reason could be used for the right side. In table 14 it can be seen that the simulated production rate on 
the left side is increased by 49,8% while the calculated is 51,7% and 50,3% for operator 1 respectively 
operator 2.  

Taking a closer look at table 14 the production rate range calculated from the three-point estimation 
method for operator 2 is at best 74,2% and at worst 32,5 % For operator 1 it is at best 81,5% and at 
worst 30,9%. To note here is that the visualised production rate is when operators 1 and 2 are working 
in unison, while the calculated production rate for operator 1 and operator 2 is individual. 

Table 14: The verification of the left´s side visualised production rate and the calculated production rate for operator 1 and 
operator 2. 

 

 

5.4 Waiting times 
In table 15 the percentage of waiting time in comparison to the total time for each operator to complete 
one production cycle is presented. The waiting time has increased 16 percentage points for operator 1 
and for operator 2 it has decreased 8,4 percentage points. 

Table 15: Percentage of waiting time for each individual operator in the current and new workstation. 

.  

 

5.5 Gluing times 
The gluing times for each sequence that the collaborative robot executes are presented in table 16. The 
difference in time for the right and left side is due to that the collaborative robot must move a further 
distance due to the lower mould´s placement.  

Table 16: Gluing times for each glue sequence. Where glue sequence 1 is for the white cores, 2 for the black core assembly 
and 3 for the remaining white cores. 

 

 



 

 
 

36 
 

5.6 Error rate result 
The relative error rate for the gluing for cylinder head 11 which is fully automated is 0,57% which can 
be compared to 30% for the D6 cylinder head where the glue is currently applied by hand. This is an 
indication that the error rate will be lowered if the solution presented in this thesis is implemented 
since the gluing process will be automated. The difference in the error rate can be seen in figure 1 and 
2 in segment 2. 

 

5.7 Risk assessment result 
The risk assessment resulted in 113 risks being evaluated. 20 preventive measures were used to reduce 
the probability and consequence of the risks. After the preventive measures was used, 7 risks received 
a yellow rating and the remaining 106 received a green rating. A green rating means that the risk is 
assessed as acceptable and shall be reduced if opportunity presents itself. A yellow rating means the 
risk requires action and a red rating means that the risk requires direct action. The risk assessment was 
used to determine the robot speed and force. The robot speed was calculated to 150mm/s and the robot 
force was calculated to 110N. The robot speed was limited by the sternum in entrapment and the force 
was limited by the abdominal muscle in entrapment. The risk assessment including reasoning and 
preventative measures can be found under attachment 6-9 and the maximum permissible speeds and 
forces can be seen under attachments 3 and 4. 

 

5.8 Ergonomic evaluation of new workstation  
The difference between the new and the current workstation can be seen in table 17. By implementing 
the collaborative robots, operator 1 no longer needs to glue and temporary place the sand cores. This 
resulted in the strain on operator 1 being reduced from 21 operations to 17 operations. It also reduced 
the number of bends operator 1 needs to perform, from 13 to 12 on the right side and from 23 to 22 on 
the left side. The strain on operator 2 remained the same for the new workstation compared to the 
current workstation. 

Table 17: Difference between the criteria for the new and current workstation. 

 

 

 

 

 

 

 

 



 

 
 

37 
 

5.9 Economic evaluation of new workstation 
The payback time in years can be calculated using the formula below. The derivation of the formula 
can be seen in attachment 1. 

𝑁𝑃𝑅 = 𝑁𝑒𝑤 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑅𝑎𝑡𝑒 

𝑃𝑅 = 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑅𝑎𝑡𝑒 

𝑃𝑎𝑟𝑡 𝐶𝑜𝑠𝑡 = 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝐶𝑜𝑠𝑡 𝑜𝑓 𝑚𝑜𝑢𝑙𝑑 𝑒𝑥𝑐𝑙𝑢𝑑𝑖𝑛𝑔 𝑠𝑎𝑙𝑎𝑟𝑦 𝑓𝑜𝑟 𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟 𝑎𝑡 𝑠𝑡𝑎𝑡𝑖𝑜𝑛.   

𝐼𝑚𝑝𝑙𝑒𝑚𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛 𝐶𝑜𝑠𝑡 = 𝐶𝑜𝑠𝑡 𝑡𝑜 𝑖𝑚𝑝𝑙𝑒𝑚𝑒𝑛𝑡 𝑝𝑟𝑜𝑝𝑜𝑠𝑒𝑑 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 

The following relation could be made: 

𝑌 =
ே௉ோ

௉ோ
   

Using the following equation: 

𝑇𝑖𝑚𝑒 =  
𝐼𝑚𝑝𝑙𝑒𝑚𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛𝑠 𝐶𝑜𝑠𝑡

൫𝑆𝑎𝑙𝑎𝑟𝑦 ∗ (𝑌 − 1)൯
 (16) 

                  

Inserting the values gives us: 

൫38325 + (38325 + 2000)൯

(2 ∗ 3 ∗ 50000(1,44 − 1))
= 𝑃𝑎𝑦𝑏𝑎𝑐𝑘 𝑇𝑖𝑚𝑒 = 0,5958 = 0,6 𝑦𝑒𝑎𝑟𝑠 

 

For this calculation the following assumptions were made: 

 The new production rate is 144% due to the rest of the foundry has this as its maximum 
production rate. 

 The current production rate that was measured was set as 100%.  
 There are 2 operators at the workstation. 
 The salary of the operators is the same and is set to 50000 € a year. 
 The production of the Volvo Penta D6 cylinder head at the foundry is in 3-shift working all 

year around and all hours of the day. 
 Only the cost of the collaborative robots and a movable pedestal is included in the 

implementation cost. 

 

 

 

 

 

 

 

 



 

 
 

38 
 

6.Discussion 
In segment 6 the results from the thesis are discussed. 

 

6.1 Visualisation of new workstation  
The visualisation was performed in Visual Components. This made it easy to visualise and identify 
when the operators were waiting, allowing for easy optimisation. Alternatively, the workstation could 
be visualised on paper, this would be hard to oversee and would take a lot of time. It would also be 
very hard to identify errors if the visualisation was done on paper. Visual Components make it easy to 
find and identify these errors since everything is shown in a graphical interface. A problem that was 
encountered when creating the visualisation was that the workers could only use a fixed walking and 
turn speed. This was solved by adjusting turning and walking speed so that it matched most of the 
walking times. Where the walking time does not represent reality, a penalty factor was added to the 
following work task for the operator so that the total time would be correct. As described in segment 3 
effort was made to minimize the waiting times.   

 

6.2 Assembly times 
When collecting and analysing data from the 13 video recordings taken of the operators when 
working, the maximum accuracy of each task was measured in seconds. This could have an influence 
on the result.  

In most of the videos recorded there was not a continuous flow in production. With flow it is meant 
that the operators work on one side and directly go to the other side and start working immediately. In 
most of the videos there were complications with either small or larger stops or the cores where 
damaged which resulted in cassation of the cores. There were also instances with complications with 
the glue guns which halted the production. As a result of this the majority of the data collected were 
not collected when there was a continuous flow in production. This has most likely impacted the times 
for each task, most likely resulting in each task taking longer time.   

Another source of error is that in 10 of the 13 videos recorded a new operator was trained. Resulting in 
the data collected being influenced by this. The times for each task in these 10 videos were longer 
compared to when 2 experienced operators performed the same task in the 3 remaining videos. This is 
however not a problem since it only leads to longer times making the times more conservative, which 
makes the production rate less sensitive to complication that might occur since the operators have 
more time for each task. 

As stated in segment 3 the three-point estimation method was used to evaluate the assembly times. 
However, the values used for a and b (optimistic and pessimistic) are the fastest and slowest values 
collected from the video recordings. The probability of the time values occurring is most likely not 

ଵ

ଵ଴଴଴
. Looking at the formula 11 and 12 for the expected value respectively for the standard deviation it 

can be seen that the a and b value has a greater influence regarding the standard deviation and later the 
high and low range than the expected value. The result of the three-point estimate should be regarded 
with this in mind. Especially the standard deviation and later the high and low range.  

Even though there are some uncertainties in the values from the three-point estimate they are still used. 
This is due to the values for the expected time being conservative. When calculating the production 
rate in Excel and in the visualisation the expected value is used. When looking at the mean value that 
was collected by 13 videos it is lower than the three-point estimate´s expected value which means that 
the expected value is more conservative than the mean value.  



 

 
 

39 
 

6.3 Production rate 
As stated in the result the new stations visualised total production rate has increased by 52,1% but due 
to the rest of the foundry being limited to a production rate of 144% the new workstations production 
rate is set at 144%. This increase in production rate will enable the foundry to free up capacity to 
produce more products. The increased production also means that the workstation has a safety factor 

of  
ଵହଶ,ଵ

ଵସସ
= 1,056 for its production rate. This is important due to being a buffer to error sources when 

gathering data and calculating assembly times.   

To note is that the production rate established from this thesis will merely give an indication of what 
production rate that can be achieved. To verify this production rate, the solution must be implemented. 

 

6.4 Verification of production rate 
After the production rate from the visualisation was acquired, it had to be verified by the calculated 
production rate. The tasks that are not performed in reality, their time values are taken from the 
visualisation to create the calculated production rate. The tasks taken from the visualisation are 
presented in the bullet points below. 

 Operator 1 moving from the control panel when sending away the completed pallets to the 
white core pallet on the other side. 

 Pressing the button to start the gluing for the first, second and third sequence. This was 
estimated to take 1 second each. 

 Operator 1 moving from the white core pallet to the control panel when sending away the 
completed pallets on the other side. 

 Operator 2 waiting for the robot to finish its gluing sequence (2) before mounting the complete 
black core assembly. 

 Operator 2 waiting for the robot to finish gluing sequence 3. 

Since these values are taken from the visualisation, they have no standard deviation. This is due to the 
simulation always producing the exact same time values. 

From the three-point estimation method the production rate for each operator was calculated which 
can be seen in table 14. This was done to verify that the production rate in the simulation is correct. By 
simulating the production rate on the left side and comparing it to the operator’s individual production 
rates on the left side which were calculated in Excel. The simulated left side has an increase of 49,8% 
while the operators 1 and 2 has 51,7% respectively 50,3%. The production rates differ slightly but the 
difference is negligible. The visualised production rate is lower than the calculated production rate, 
which suggest that the visualised production rate should be higher, meaning that the visualised 
production rate is conservative. 

This indicates that the visualisation on the left side is correct, which indicates that the entire 
visualisation is correct. 

To note here is that the visualised production rate is higher than the visualised production rate on the 
left side. This because the robot on the right side moves a shorter distance during the gluing sequence. 
This can be seen in table 13. 

In table 14 it can be seen that operator 1 and operator 2 have a low range and a high range of their 
respectively production rates but a low range and high range for the visualised production rate is 
missing.  



 

 
 

40 
 

The reason is that there are uncertainties how different delays for operator 1 and/or operator 2 will 
affect each other and lastly the production rate. When the operators are working in unison, they   
depend on each other which makes it very difficult to calculate high and low range since relevant data 
is missing. It should be entirely possible to investigate this further but due to time constraints this is 
not regarded in the thesis.  

The high and low range for operator 1 and operator 2 is however possible to acquire, by adding the 
fastest respectively the slowest work times the low range respectively the high range can be acquired. 
When looking at one operator the production rate is not dependent on the other operator´s production 
rate in the same way as for the total production rate. They are still dependent on each other due to the 
waiting time. But the waiting time is taken from the simulation meaning it has the same m, a and b 
value, which means it doesn’t contribute to the low and high range.  

 

6.5 Waiting times 
Looking at table 15 the waiting times for the current workstation is 14,4% for operator 1 and 15,6% 
for operator 2. For the new workstation the waiting times are 30,4% for operator 1 and 7,2% for 
operator 2. The values are presented in table 15. 

The waiting time for operator 1 increased by 16 percent points while the waiting time for operators 2 
decreased by 8,4 percentage points. The reason why is because operator 1 is now waiting longer for 
operator 2 to finish mounting the black core assembly. This is mostly the case in the current 
workstation that was observed in the recorded videos. The waiting time has increased due to the robot 
speeding up the process for operator 1 due to operator 1 losing all the gluing and the temporary 
placement tasks of the sand cores. For operator 2 some of the waiting time was eliminated and some 
was redistributed. In the current workstation operator 2 often had to wait for operator 1 to finish the 
gluing for the black core assembly. This is to make sure that the glue doesn’t dry before the black core 
assembly has been mounted onto the lower mould. In the new workstation the time to wait for gluing 
sequence 2 is decreased. A small portion of the time is also distributed towards waiting to begin with 
the black core assembly, on the other side until operator 2 is finished with the current side. The latter 
waiting time however should not come up in reality but due to how the visualisation is programmed 
this occurs. This is regarded as positive due to the production time being lower resulting in a more 
conservative production rate.  

 

6.6 Gluing times 
In table 16 one can see the gluing times for each gluing sequence that the collaborative robot performs. 
In the current workstation operator 1 must glue as a separate task. However, in the new workstation 
the operator can work alongside the robot when it performs it´s gluing sequence. Meaning that 
operator 1 does not have to wait during gluing sequence 1 and 3 and can instead work. During gluing 
sequence 2, operator 1 must wait to start the gluing sequence so that when the gluing sequence is 
completed, operator 2 is ready to mount the black core assembly in the lower mould. For sequence 2 
this is the same scenario for the new workstation as the current one.  

The importance of the value for sequence 2 in table 16 is that it is below 30 seconds. Between 30 and 
40 seconds is the time that it takes for the glue at the current workstation to have an increased risk of 
drying. During gluing sequence 2 there is shortage of time to mount the black cores on the lower 
mould. Therefore, it is recommended that a glue with a slightly longer drying time is used. 

 



 

 
 

41 
 

6.7 Error rate 
Since the solution has not been implemented, the error rate cannot be judged properly since that would 
require the implementation of the new workstation. Instead, indications from other products have been 
used to determine the error rate for the gluing. The evaluation of the error rate does also not consider 
any new errors that may occur due to the suggested changes in the workstation since this would 
require implementation of the new workstation. The data that has been received only shows the 
number of errors and what kind of error that has occurred. This means that it is unknown what the total 
error rate is. However, it is known that cylinder head 11 is produced in larger quantities then the D6 
cylinder head and that both are produced in significant enough quantities that the different shares of 
errors should still be correct.  

 

6.8 Risk assessment  
In segment 3 an assumption of the relative speed was assumed. The velocity of the operator was set to 
0 so that the formula calculating the relative speed becomes the robot´s speed. The reasoning behind 
this assumption was that when the mean operator speed of 1,6 m/s from Svenska institutet för 
standarder (2016-b) is used, the calculations yielded that in the case of a contact between operator and 
robot, the robot had to be moving in the same direction as the operator. This means the entire 
workstation would be assessed as unsafe according to Svenska institutet för standarder (2016-b) if just 
the robot was completely still. With this reasoning you could say that any beam or fence that is in the 
current workstation that an operator could walk into would make the current workstation unsafe which 
isn’t the case. Therefore, the assumption was made to set the operators speed to 0.  

In ISO 15066 its clearly stated that the speed or force values that each body part is allowed to be 
subjected to, is below the minor injury threshold. It means that the operator should not feel pain. This 
is the reason a consequence becomes a 1, when preventative measure 1 and 2 is used. (Svenska 
institutet för standarder, 2016-b). 

According to ISO 15066 the risk that the operator is hit in the head by the collaborative robot must be 
negligible (Svenska institutet för standarder, 2016-b). If the risk is negligible then the risk is assessed 
safe. By using the preventative measures seen in the risk assessment in attachment 6-9, the probability 
of the head being hit is so small it can be considered negligible.  

 

6.9 Ergonomic evaluation of new workstation 
By implementing collaborative robots into the workstation operator 1 no longer needs to glue the sand 
cores as stated in segment 5.9. The number of bends of the back is only reduced by 1 bend. This is 
because the operator instead needs to press a button to start the gluing sequence of the robot. The 
control panel for this button is situated under the conveyor belt just as the control panel for sending 
away the pallets. This means that operator 1 must bend to press the button, if the control panel instead 
would be situated in a way that the operator does not need to bend the back then the number would be 
reduced to 9 on the right side and 19 on the left side. The strain however is not reduced by doing this 
and remains the same.  

According to the current ergonomic evaluation of the workstation the largest contributor to the 
ergonomic strain is the gluing of sand cores, due to the gluing being performed in static in 3 sets in the 
yellow zone. In the new workstation this is eliminated which should reduce the total ergonomic strain 
on the operator. It is however still recommended that the operators switch between working on the 
white cores and black cores several times during the day. 



 

 
 

42 
 

In the new workstation operator 1 has fewer tasks and the production rate has increased in comparison 
to the current workstation. This means that a new ergonomic evaluation must be made due to the 
evaluation accounting the number of lifts performed in one batch multiplied by the production rate. 

  

6.10 Economic evaluation of new workstation 
The economic evaluation shows how long it will take before the robots' production cost per product is 
equal to the old cost per product. The real payback time will however be shorter since this does not 
consider that the robots will also be used to glue the moulds for the D4 cylinder head which will 
shorten the payback time since the D4 cylinder heads most likely will be produced faster lowering the 
salary cost per product. Note that the evaluation does not consider the installation costs for the robot 
and procurement of glue gun, glue hose, and vision system.  

The payback time is calculated when the foundry is running at the new 100 % production rate 
constantly which most likely will not be the case at all times. This will result in the payback time 
being longer. Since the D6 is not produced at all times this will be an even longer time. 

The increased production rate will free up capacity to produce more cylinder heads. This will most 
likely be the