Optimization of the polishing procedure by using a robot assisted polishing equipment

(1)

OPTIMIZATION OF THE POLISHING PROCEDURE BY USING A ROBOT ASSISTED POLISHING EQUIPMENT

av

Anaïs Faure-Vidal •••• 2008 06 26

Handledare: Sabina Rebeggiani Examinator: Bengt-Göran Rosén

Ett examensarbete utfört enligt kraven för Högskolan i Halmstad för en Magisterexamen i Teknisk Produkt- och Produktionsförbättring

(2)

Acknowledgment

I would like to thank all the people I met here in Sweden; who helped me to adapt myself in this country and who transformed these five months in an amazing experience; especially:

• Mr. Bengt-Göran Rosén, professor and director of the laboratory, for welcoming me in his department and helping me in my project.

• Ms. Sabina Rebeggiani, PhD student, for always being patient and joyful in the work and for sharing with me her knowledge about polishing.

• Mr. Zahouani, professor at the ENISE, for giving me the opportunity to work in Halmstad University during this second internship.

• Ms. Levy, English teacher at the ENISE, for being always available in case of problems.

• Mr. Frédéric Cabanettes, PhD student, for helping me before and after my arrival in Sweden, as well on a personal as on a professional level.

• Mr. Stefan Rosèn and Ms. Karin Westerberg, from Toponova AB, for their patience and always giving me advices about the use of the interferometer.

• Mr. Jens Grønbæk and Mr. Lars Sørensen, for trusting me in this project and for their help and their welcome during my work at STRECON A/S.

• All the people from the university and Kåren, for their help during these five months.

• All the students I met and who transformed this internship in an unforgettable human experience.

(3)

Abstract

Nowadays, the polishing process is one of the most important steps of the manufacturing of moulds and dies. Occupying up to 40% of the total production time and cost, it is decisive for the final appearance and quality of a surface. Because of its complexity, the polishing is mainly carried out manually; and the final quality depends of the expertise of the operator. That is why for a mass production, automating this process is necessary.

The purpose of this project was to find out the optimized sequence of polishing for a Mirrax ESR steel (Uddeholm Tooling AB) using the Strecon RAP-200 (robot assisted polishing equipment). Using a Design of Experiment, the machine parameters were tested in order to better understand their influences and interactions. The report starts with a description of general polishing knowledge and ends up with the results from a Design of Experiment.

The information from this test are a first step in the evaluation of the Strecon RAP-200. Even if many results have been found out, only four parameters have been tested, and to be able to optimize the polishing sequence, further studies need to be carried out.

(4)

Introduction

I did my second industrial placement in the frame of my 4^th year at the ENISE. After a first experience in Eurocopter (Marignane, France), European company of helicopter production and design, I wanted to have a different experience; working abroad in a research laboratory.

The first challenge was to work abroad, using another language and living in a different country with other traditions and culture. I had to adapt myself in an international world too, by living and working with people from many countries.

The second challenge was to understand the work in a research laboratory. I have always been attracted by research but never known exactly how the work could be the work in a laboratory; this industrial placement was a perfect opportunity for me to find out if the research could correspond to my expectations.

Attracted by Scandinavian countries and helped by the partnership between the ENISE and the Halmstad University in Sweden, I choose to make my second industrial placement in this mechanical laboratory managed by professor Bengt-Göran Rosén.

(6)

I. Presentation of companies

1.1. HHaallmmssttaadd

The city of Halmstad, located on the west coast of Sweden between Göteborg and Malmö, is well known for sport and tourism. After 300 years being Danish, Halmstad finally became Swedish. Now capital of the Halland county, it is one of the most touristic area in Sweden. Many of the 90 000 inhabitants are students and members of the armed force.

2

2.. HHaallmmssttaadd UUnniivveerrssiittyy

Around 7000 students are following humanistic, social sciences, economics, behavioral and health sciences, technical and scientific subjects at Halmstad University.

Really close to the industry, the university has also a well- developed research environment.

The university celebrates its 25^th anniversary this summer.

Figure 1: Trade Center, Halmstad University Halmstad

(7)

3 3..

U

Uddddeehhoollmm TToooolliinngg AABB ¹¹

The firm of Uddeholm Tooling AB, created in 1668, is nowadays on of the world’s leading supplier of tool steel and related services.

The company produces different kind of steels and tool steels all over the world, and is present in more than 100 countries. Uddeholm employs 4000 persons.

4.4. SSttrreeccoonn AA//SS ²²

In the end of the 70´s, the technical university of Copenhagen started a project about safe tools.

This project reavealed the strength of the strip-wind container technology and the first prototype of the STRECON container was created.

From the beginning of the 80’s, this technology was developed and applied, for example, to cold forging operations.

Nowadays, located in Sønderborg (Denmark), the company has a strong competence engineering and designing pre-stressing systems and other tooling parts.

1www.uddeholm.com

2 www.strecon.com

(8)

II. Literature study

I started my project by reading some articles about polishing. This subject was unknown for me and I needed to learn more about it. I focused on two parts; polishing methods and surface characterization.

1.1. GGeenneerraall ppoolliisshhiinngg iinnffoorrmmaattiioonn

Polishing is a major step of the manufacturing process. Nowadays, polishing is mainly done manually, because of the complexity of the process. A human being can easily feel and see the surface and so adapt his/her movements to obtain the best surface. Therefore the process is quite long and costly, and so more and more firms and laboratories try to create a machine able to polish as well as (or better than) a human, but quicker.

1. Polishing methods.

To obtain a smooth or shiny surface, many different methods can be used:

- Laser polishing [1]

- Chemical mechanical polishing [2], [3]

- Magnetic polishing [4]

- Electro polishing [5], [6]

- Mechanical polishing

These methods use different techniques: chemistry, electricity, magnetism ... I will focus on mechanical polishing, which is the one I used, in the next chapter.

2. Mechanical polishing,

Mechanical polishing is the removal of material to produce a scratch-free, specular surface using fine abrasive particles. [7]

This process is performed in two steps:

- Grinding - Polishing

The grinding, a rapid removal of material to remove large irregularities from the surface.

The polishing, to produce a scratch-free surface.

For each of these steps, different equipments need to be used: clothes, stones …

(9)

Basically, as we want smoother surfaces, we use finer grain sizes. A grinding step is mainly done with grinding stones, directly in contact with the surface; the polishing step is done with fine abrasive particles (less than 20 µm) used with polishing clothes or other carriers such as copper or wood.

The main problem by automating polishing is to know when to change the polishing equipment (stone or diamond paste, etc). Indeed, one of the major problems about polishing is the “over polishing”.

In the beginning of each step, the surface roughness decreases, but after a certain time with the same abrasive, it increases due to an excessive polishing.

A experienced polisher can see when to stop and change the equipment, and that is on reason to why they are more efficient. [8]

Figure 2: Over polishing phenomenon

2.2. SSuurrffaaccee aannaallyyssiiss

Even if the visual aspect of a surface is an important proof of its quality, we need numerical parameters to quantify it and so, instruments able to measure them.

1. Roughness parameters [9]

Roughness is not an easy characteristic to describe. A lot of parameters are used, but none of them can entirely describe a surface.

OVERPOLISH

The Birmingham 14 are the main parameters used to describe surface textures.

These parameters are based on 3D measurements and are named S.. The same exist for profiles but are named R..

Figure 3: The Birmingham 14

(10)

They describe the height deviation of a surface and are divided in two categories; the average of ordinates and the extreme peak and valley parameters.

Sz, Sp and Sv are extreme peak and valley parameters.

Sz is the difference between the highest peak and the deepest valley, which gives an idea of the maximum roughness. Sp is the difference between the mean line and the highest peak; Sv is the difference between the mean line and the deepest valley.

But these are not really representative of the surface; we can improve it by taking the five highest peaks and the five deepest valleys.

Figure 4: Amplitude parameters

Sq and Sa are average of ordinates parameters.

Sa is the arithmetic mean of the magnitude of the derivation of the profile from the mean line. We can see on the figure how we obtain it. (1) The medium line of the profile is calculated. (2)The arithmetic (positive) value of the profile is taken. (3) The medium line of this new profile is calculated; the Ra value is the difference between these two medium lines.

Figure 5: Average parameter: Ra

But even if Ra is the most commonly used parameter, we can see on figure 5 that it can not entirely describe a surface. This figure shows three surfaces with the same Ra, but which look really different.

1

2

3

(11)

Figure 6: V-parameter set

- Linear area material ratio curve parameters These parameters are often used in the automotive industry.

Figure 7: Linear area material ratio curve parameters

It indicates the part of the surface which is valleys, or peaks. This curve allows for example, if we need surfaces with as less peaks as possible, to find these surfaces, which have small Rpk values.

2. Measurement instruments [9]

To measure the parameters, we dispose of many kinds of instruments, using many kinds of techniques:

- contacting techniques - optical techniques

- scanning probe microscope.

-

The V-Parameter Set is designed to analyse the material volume and void volume. The surface is split into three zones; the peak zone, the core zone and the valley zone, and then, calculation are made about the volume distribution.

(12)

Halmstad University uses the equipment of a company located near the university, Toponova AB.

Among other things, they have a mechanical stylus and an interferometer.

- The mechanical stylus - a contacting method

A diamond cone or a steel ball for example, runs over the surface, and the vertical movement is recorded and transformed into electrical signals.

The result is a surface profile and the possibility to calculate roughness parameters.

The precision of the stylus depends of its dimensions (dimensions of the diamond cone), and if the surface is too smooth, the performances can decrease.

Moreover, the stylus is a contacting technique which can damage the surface.

Figure 8: Mechanical stylus

- The interferometer - an optical technique

A beam is divided in two beams by a mirror. One is reflected on the surface to measure, the other on a reference surface. The two beams are combined into optical fringes called interferogram.

A software (for example MountainsMap [9]) can show the surface in 3 Dimensions and calculate roughness parameters.

Figure 9: Left: Interferometer principle; Right: Interferometer fringes

The resolution is about the wavelength of the beam used.

(13)

III. Background

The collaboration between Halmstad University, Strecon and Uddeholm started during the second semester of 2007. The beginning of this work was to test the prototype of the Strecon RAP-200, an automated polishing system developed by Strecon.

The Strecon RAP-200 - a robot assisted polishing equipment

The sample is in rotation through the medium of a bit. The polishing equipment (stone or diamond paste) is translated on the sample through a flexible bar controlled by the robot.

The robot also gives a horizontal vibration movement to the bar and so the polishing equipment, which seems to be the key of the machine.

During the process, a lubricant is thrown on the work piece.

In the beginning of the tests, the decision was taken to use a Design of Experiment in order to test the machine. This kind of experiments reduces the test and evaluation time and gives information about the influence of some parameters and their interactions.

A similar Design of Experiments (DoE 1) was performed in 2007 but could not be completely finished because of a machine breakdown.

Bit

Lubricant Polishing equipment

Bar

Control computer

Robot

Work piece

(14)

Results found after the DoE 1: [10]

Although the prototype has a breakdown, some samples were made and analysed.

The followings remarks were made:

- The bit induces a rotation movement to the sample and so the polishing equipment works on the surface according to a rotation speed.

- On the samples made during the DoE 1, it appeared that after the polishing step, turning marks were visible. Now, it is important to get rid of these marks in order to have a high surface quality.

Figure 10: DoE 1 – Turning marks after polishing

In the beginning of the first semester of 2008, Strecon received the new machine which needed to be tested.

Linear speed Rotation

speed The machine can control the

rotation speed but not the linear speed.

So depending on the position on the surface, some points will be polished with a high linear speed and other with a lower one.

It is important to know if this difference really influences the final surface and to quantify it.

µ m

-2 .5 -2 -1 .5 -1 -0 .5 0 0 .5 1

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 6 5 0 7 0 0 µ m

L e n g th = 7 0 6 µ m P t = 1 .3 µ m S ca l e = 3 .5 µ m

100 µm

(15)

IV. Prestudy and DoE 2

1

1.. IInnfflluueennccee ooff tthhee lliinneeaarr ssppeeeedd

1. Presentation

It was decided, as a first step, to test the influence of the linear speed. Three disks were included.

For each disk, all parameters (force, vibration, diamond size ...) were fixed with normal levels according to Strecon. Three rotation speeds were tested, one for each disk.

This configuration permitted to test the influence of the linear speed. Indeed, if the rotation speed is fixed, the linear speed will change depending on which distance we are from the centre.

Strecon prepared three disks with three different rotation speeds:

- 200 rpm (A1) - 500 rpm (C2) - 1000 rpm (B2)

and the measurements were done on three radius for each disk so we could study nine linear speeds.

It is important to notice that for these samples; just two repetitions (chapter IV. 2 Results pretest) have been performed (instead of 20 at least for a normal polishing step). Indeed, the purpose here was not to have the best surface but just to compare different linear speed.

Fixed rotation speed

Linear speed depending on the distance to the centre

(16)

2. Results pretest The diagram below presents the results of this test.

Sa values

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8

0 1000 2000 3000 4000 5000

Linear speed (mm/s)

Sa (µm) C2

B2 A1

First observation: the highest rotation speed (B2) produces the best surface. This observation is explained by the fact that for the three samples, two repetitions have been performed which means that the polishing equipment covered two times the distance from the outer of the disk until the centre and from the centre until the outer part. If the rotation speed is higher, each point of the surface will be polished more times than for a lower rotation speed.

Second observation: For each disk, the best surface was performed with the lowest linear speed. It can be explained by the fact that the points in the center are polished during a longer time (due to lower distance needed to be polished in case of lower radius).

The usual values for the rotation speeds are about 200 until 500 rpm. For these values, the difference of Sa is about 20%.

9-10 mm 27-28 mm 40-41 mm

Distance to the centre

A1: rotation speed 200 rpm B2: rotation speed 1000 rpm C2: rotation speed 500 rpm

(17)

It will be interesting to know if the result is the same for smoother surfaces (after an entire polishing step) and if the difference can be a problem. Strecon needs to polish entire disks with, to a final roughness of 10 nm (Sa), and if these 10 nm are just performed in the centre of the piece, the aim is not reached on the whole surface. It is important to quantify the difference.

2.2. PPrreeppaarraattiioonn ooff tthhee DDooEE 2 2

The test of the Strecon RAP-200 will proceed in different steps. For the first one, the decision was made to perform around 20 trials and to use a Design of Experiment.

To test all the parameters and their interactions, we should need to perform hundreds of tests. To reduce this number and gain in time and money, we chose a DoE, which goal is to find a lot of information about parameters with as few tests as possible.

To choose a Design of Experiment, many factors have to be considered: [12]

- the resolution of the DoE - the number of trials available

- the number of level for the parameters - the number of parameters we want to test

Resolution Number of trials Number of factors Number of levels

4 32 1-5 2

4 27 1-3 3

4 16 1-4 2

4 9 2 3

4 8 1-3 2

3 32 6 2

3 16 5 2

For the first test (of the new machine), we wanted to have a good resolution about the main parameters (resolution of 4 - information about the influence of all the parameters and their interactions) but the number of levels was not important. Because of the possibility of non-linear results, a 3 levels design would have given better results but this non-linearity can be test in a further step. We could handle with 2 levels. Moreover, with a 2 levels test, it is possible to have an extra sample with medium values which can give clues about this non-linearity.

So the best choice of DoE was a 16 trials, 4 parameters and 2 levels design.

The process parameters we could test are the following: force, vibration pulse time and vibration pulse length, robot speed, repetitions, rotation speed, lubrication, carrier and diamond paste, etc.

(18)

Because of the prestudy and the knowledge of the influence of the linear speed, it was decided to keep the rotation speed constant and to take the linear speed as a factor.

The key of the Strecon RAP-200 seems to be the vibration. To know what its real effect is, it was important to test it; the vibration pulse time was the second factor.

Finally the repetitions and the force are major factors which influence the polishing.

These were the parameters tested in this DoE 2.

In the future, we could consider to test:

- the rotation speed: to confirm the prestudy where we found out that the highest rotation speed performed the best surface

- the robot feed speed - the pulse length

- the carrier and diamond paste to know how smooth surfaces we can get

So for the DoE 2, 16 trials were planned.

But these trials just concerned the first polishing step. It was needed to find a good way to grind the surfaces to be sure that all turning marks were removed. In general it is also of interest to find the cheapest, the more efficient and the quickest way of grinding. In the beginning of the work at Strecon, different grinding processes were tested.

(19)

V. Work at Strecon

We were at Strecon from the fifth until the seventh of May, in order to test the Strecon RAP-200. The purpose was to test, in a first part the grinding step, and in a second part the polishing process.

1.1. GGrriinnddiinngg sstteepp

We first wanted to test the grinding to reduce the time of this step and to be sure that no turning marks were left. Indeed, we wanted to know if the process could be improved.

We tested two stones: MF 600 and DF 900 and some numbers of repetitions, in order to find the best combination. For this part we didn’t perform a DoE, but just tried a few samples and we finally agreed to the following grinding steps.

Parameter Unit Step 1 Step 2

Pulse pulse/min 2000 2000

Pulse length mm 1,5 1,5

Force g 1000 1000

Repetitions 30 20

Bar No. 4 4

Carrier, JOKE MF600 DF900

Area of the carrier mm2 18 18

Robot speed mm/s 2 2

Rotation speed Rpm 500 500

Lubrication* times/min 2 2

We could notice a problem; the samples included didn’t have the same initial roughness after a turning step; some were around 0,5 µm and some around 1,2 µm (Ra values). After the grinding process, this difference was smaller. The influence of this initial roughness will be studied in detail in the analysis part. One solution could be to put the initial roughness as a fourth parameter if the influence of the linear speed is not too important. The initial roughness should then be simplified to include two levels, 0,5 µm and 1µm.

2.2. PPoolliisshhiinngg sstteepp 11

After finding a good way to prepare the surfaces, we could begin the DoE 2. (cf. Appendix 1 ) The parameters we choose to test were:

1st level 2nd level

Force (g) 500 1100

Repetitions 10 40

Pulse (pulse/min) 1500 2500

Linear speed (mm/s) 470 1100

(20)

To test the linear speed we kept the rotation speed constant and measured at two different radiuses (15 and 35 mm).

We also prepared another surface with medium values; the so called centre point, which is very helpful to analyse in this kind of Design of Experiment.

- Force: 800g

- Pulse: 2000 pulse/min - Repetitions: 20 repetitions - Linear speed: 630 mm/s

We chose these values because they seemed to be normal levels, and we choose the 2 levels of the DoE 2 to have higher and lower values than these “normal levels”.

3.3. PPoolliisshhiinngg sstteepp 22

Finally, we had time for some extra samples, so we chose to test two other parameters. For each one (and for the reference) was performed on the outer part of the disks an additional step of polishing with a 3 µm diamond paste.

1. On the G2 sample, we tried a higher rotation speed. The analyse of the previous samples showed that the highest rotation speeds gave the best surfaces. Now, at Strecon, they have noticed the opposite, so we needed to have a sample to compare with the 300 rpm used in the DoE. We tested 700 rpm.

2. On the D2 sample, we tried to polish without vibration. We tested one disk without any pulsation. Before analysing the results, we can see that the surface appearance is different with and without vibration. According to the stylus measurements, the surface seems better (low Ra), but it seems that the vibration is a main factor in the grinding step more than in the polishing step (stones or diamond paste).

Ref.2 F2 Ref.2 F2 outer G2 G2 outer D2 D2 outer Parameter Unit Polish 1 Polish 2 Polish 1 Polish 2 Polish 1 Polish 2

Pulse pulse/min 2000 2000 2000 2000 0 0

Pulse length mm 1,5 1,5 1,5 1,5 – –

Force g 800 800 800 800 800 800

Repetitions 20 20 20 20 20 20

Carrier, JOKE soft wood soft wood soft wood soft wood soft wood soft wood

Area of the carrier mm2 28,3 28,3 28,3 28,3 28,3 28,3

DP** um 9 3 9 3 9 3

Robot speed mm/s 2 2 2 2 2 2

Rotation speed Rpm 300 500 700 500 300 500

Lubrication* times/min 2 2 2 2 2 2

The values of the fixed parameters in the DoE are the same as for the reference sample.

(21)

VI. Analysis of the results

1.1. MMeeaassuurreemmeenntt aanndd aannaallyyssiiss memetthhooddss

Back to Halmstad, the samples were measured and analysed.

For this study of a polishing machine, I needed to measure smooth surfaces. In view of the performances, I choose to use the interferometer. Moreover, this is a non contacting technique, which limits the risks of damaging the surface. Finally this device can perform 3D measurements and analysis.

On each surface, fifteen points were measured to have representative values at magnification 10x.

A software called MountainsMap was used to analyse the surfaces.

This software is, for example, able to calculate roughness parameters, extract a profile of the surface, and calculate the main directions of the surface.

For each surface, a document was made to calculate these information.

Figure 11: Extract of a MoutainsMap document

(22)

2.2. RReessuullttss aanndd ddiissccuussssiioonn DDooEE 22

1. Best surfaces

In this diagram, we can see that two disks have really good final surfaces: H1 and J1. (cf. Appendix 2)

0 0,02 0,04 0,06 0,08 0,1 0,12 0,14

F2 H1

inner H1 outer

H2 inner

H2 outer

I1 inner

I1 outer

I2 inner

I2 outer

J1 inner

J1 outer

J2 inner

J2 outer

K1 inner

K1 outer

K2 inner

K2 outer

Sq (µm)

Force (g) 800 1100 1100 1100 1100 1100 1100 1100 1100 500 500 500 500 500 500 500 500 Vibration (pulse/min)2000 2500 2500 2500 2500 1500 1500 1500 1500 2500 2500 2500 2500 1500 1500 1500 1500 Linear speed 700 500 1100 500 1100 500 1100 500 1100 500 1100 500 1100 500 1100 500 1100

Repetitions 20 40 40 10 10 40 40 10 10 40 40 10 10 40 40 10 10

For these two surfaces, the linear speed does not really influence the final roughness; there is not a big difference between the Sq values got with a high and a low linear speed.

The four surfaces have two parameters in common:

- the vibration pulse time: 2500 pulse/min - the repetition: 40.

The good point with these two configurations is that the surface is almost the same on the entire disk, i.e. the difference between the inner and the outer part is negligible.

The I1 sample has a low Sq too, especially the inner part, but higher than those of the H1 and J1 samples; and moreover, there is a difference between the inner and outer part.

So it can be concluded that the best surfaces are the one with 40 repetitions and 2500 pulse/min, whatever is the force or the linear speed.

In view of these results, it could be interesting to test other values for the parameter repetition. Indeed, because of the over-polishing phenomena, the best surface could be obtained with a medium number of repetitions: 20 or 30.

(23)

2. Analysis of the parameters repetition and linear speed/force

0 0,02 0,04 0,06 0,08 0,1 0,12 0,14

F2 H1

inner H1 outer

H2 inner

H2 outer

I1 inner

I1 outer

I2 inner

I2 outer

J1 inner

J1 outer

J2 inner

J2 outer

K1 inner

K1 outer

K2 inner

K2 outer

Sq (µm)

Force (g) 800 1100 1100 1100 1100 1100 1100 1100 1100 500 500 500 500 500 500 500 500 Vibration (pulse/min)2000 2500 2500 2500 2500 1500 1500 1500 1500 2500 2500 2500 2500 1500 1500 1500 1500 Linear speed 700 500 1100 500 1100 500 1100 500 1100 500 1100 500 1100 500 1100 500 1100

Repetitions 20 40 40 10 10 40 40 10 10 40 40 10 10 40 40 10 10

This diagram shows the same Sq values, but each colour corresponds to a combination of force/vibration. Each time, the two firsts sticks are with 40 repetitions and the seconds 10.

The diagram presents that the best surfaces are always made by 40 repetitions. 10 repetitions are not enough. Again, it could be interesting to test other values for this parameter.

0 0,02 0,04 0,06 0,08 0,1 0,12

0 5 10 15 20 25 30 35 40 45

Repetitions

Sq (µm)

Except for the repetitions, no result can be found after the study of the parameters one by one. In the diagram below the influence of the repetitions is clearly shown. For the three other parameters, no general rule can be found; just the interactions between some parameters can be studied.

High force Low force

(24)

An interesting observation can be made from the diagram about the interaction of two parameters: the force and the linear speed. Depending on the force, two schemes can be done.

- With a high force (1100g): to have the best surface after 10 repetitions, it is better to have a high linear speed; to have the best surface after 40 repetitions, a low linear speed is more efficient.

- With a low force (500g): to have the best surface after 10 repetitions, it is better to have a high linear speed too; but to have a good surface after 40 repetitions, a high linear speed is a better solution.

The difference is still more important with 1500 pulse/min.

3. Analysis of the parameter force

0 0,02 0,04 0,06 0,08 0,1 0,12 0,14

I2 inner K2 inner

I1 inner K1 inner

I2 outer K2 outer

I1 outer K1 outer

F2 H2

inner J2 inner

H1 inner

J1 inner

H2 outer

J2 outer

H1 outer

J1 outer

Sq (µm)

Force 1100 500 1100 500 1100 500 1100 500 800 1100 500 1100 500 1100 500 1100 500

Pulse 1500 1500 1500 1500 1500 1500 1500 1500 2000 2500 2500 2500 2500 2500 2500 2500 2500

Repetitions 10 10 40 40 10 10 40 40 20 10 10 40 40 10 10 40 40

Lin. Speed 500 500 500 500 1100 1100 1100 1100 700 500 500 500 500 1100 1100 1100 1100

It can be observed through this diagram that a high force is always better or equal, except for the combination:

- 1500 pulse/min, 1100 mm/s.

To know the optimal value for this parameter (force), it could be interesting to test other values, keeping the combination of other parameters we had for the samples H1 and J1:

- 2500 pulse/min, 40 repetitions

1500 p/min – 500 mm/s

Better with a higher force (1100g)

1500 p/min – 1100 mm/s

Better with a lower force (500g)

2500 p/min – 1100 mm/s

Better with a higher force (1100g) 2500 p/min – 500 mm/s

Better with a higher force (1100g)

(25)

A medium value such as 800g could be tested as well as higher values, for example 1300g.

4. Analysis of the parameter vibration

0 0,02 0,04 0,06 0,08 0,1 0,12 0,14

K2 inner

J2 inner

K1 inner

J1 inner

K2 outer

J2 outer

K1 outer

J1 outer

F2 I2 inner H2 inner

I1 inner H1 inner

I2 outer

H2 outer

I1 outer

H1 outer

Sq (µm)

Force 500 500 500 500 500 500 500 500 800 1100 1100 1100 1100 1100 1100 1100 1100

Pulse 1500 2500 1500 2500 1500 2500 1500 2500 2000 1500 2500 1500 2500 1500 2500 1500 2500

Repetitions 10 10 40 40 10 10 40 40 20 10 10 40 40 10 10 40 40

Lin. Speed 500 500 500 500 1100 1100 1100 1100 700 500 500 500 500 1100 1100 1100 1100

A high vibration pulse time (2500 pulse/min) always performs the best surface, except for 10 repetitions with 1100 mm/s and 500g.

According to the previous results, this configuration is not an interesting one. It was demonstrated that a high force and more than 10 repetitions were essential.

As Strecon explained, the vibration is the key of this machine and performs best results. It could be interesting to test higher values for this parameter. The machine is able to perform 3000 pulse/min.

5. Analysis of the visual aspect

Until here, the analysis of the results is just based on Sq values. Indeed, the polishing standards are mostly expressed through average parameters such as Sq (most commonly Ra). However, another aspect of the polishing standards is the visual appearance. To polish surfaces is not just a way to get lower roughness values, but to have for example, mirror-like surfaces.

500g – 1100 mm/s

Better with high vibration pulse time (2500 pulse/min) for 40 repetitions.

500g – 500 mm/s

Better with high vibration pulse time (2500 pulse/min)

1100g – 1100 mm/s

Better with high vibration pulse time (2500 pulse/min) 1100g – 500 mm/s

Better with high vibration pulse time (2500 pulse/min)

(26)

That is why the analysis of the visual aspect is as much important as the roughness study. (cf. Appendix 3)

Figure 12: Comparison between two surfaces: left: H1L1 – Lower Sq value --- right: K2L1 - Higher Sq value

The figure above shows two surfaces with different Sq values (0,04 µm and 0,12 µm).

The two surfaces have scratches on one main direction. These scratches are not due to the turning process, indeed their frequency is not about 100 µm, which is the turning marks frequency (see chapter III, Results – DoE 1).

The observation of all surfaces (MountainsMap images) gives almost the same impression: one main direction of scratches. They all look the same at this scale.

The observation of the real surfaces gives other results. Some differences can be seen. The one with a high vibration pulse time seems less glossy because of the intersection of polishing marks (even if their Sq values are smaller).

Maybe for the finer polishing step (next step with for example a 3µm diamond paste) it could be interesting to use less (or maybe no) vibration to have glossy surfaces (closer to a mirror-like surface) and to keep the vibrations for the grinding and the first polishing step, to quickly decrease the roughness values.

Figure 13: Comparison between two surfaces: left: K2 – 2500 pulse/min --- right: J2 – 1500 pulse/min

(27)

6. Influence of the initial roughness

One problem we had during the work at Strecon, was the initial surfaces of the samples. After the turning process some surfaces were two times rougher than the others.

Sample Initial Ra [um]

H1 outer 0,561 H1 inner 0,561 H2 outer 1,18 H2 inner 1,18

I1 outer 1,2

I1 inner 1,2

I2 outer 0,56 I2 inner 0,56 J1 outer 0,52 J1 inner 0,52 J2 outer 1,14 J2 inner 1,14 K1 outer 0,38 K1 inner 0,38 K2 outer 0,71 K2 inner 0,71

During the experiments at Strecon, we could not use the interferometer (which is in Halmstad). To check the surfaces after the grinding step, we used the same method we used to know the initial roughness: a portable stylus which belonged to Strecon.

Figure 14: Portable stylus used in Strecon

Many problems appeared with this measurement device:

- the surfaces we needed to measure were quite smooth and some measurements damaged the surfaces (hole created by the impact of the stylus, i.e. sample H2). So we could not take many measurements, we just took two for each surface, which is not enough to have representative results.

- The measurements were done on lines (approximately 5 cm) and were not area measurements.

- Finally this device does not have a good precision to measure such smooth surfaces.

After the turning process and before the DoE 2, two grinding steps were performed. The purpose of it was to have a finer surface before the polishing, where we used a 9 µm diamond paste (that we can not use directly on a surface with 0,5 or 1,2 µm as roughness).

In regard to the differences of initial roughness, we need to know if the grinding step could decrease this disparity.

(28)

So we can not use the values we got and compare them to interferometer values. But we can compare the surfaces between them.

After the two grinding steps, the surfaces had the following Ra values:

Sample

Ra after ginding steps

[um]

H1 outer 0,031 H1 inner 0,029 H2 outer 0,031 H2 inner 0,041 I1 outer 0,037 I1 inner 0,047 I2 outer 0,034 I2 inner 0,039 J1 outer 0,032 J1 inner 0,035 J2 outer 0,054 J2 inner 0,058 K1 outer 0,03 K1 inner 0,031 K2 outer 0,029 K2 inner 0,033

Initial roughness

(µm] 0,561 0,561 1,18 1,18 1,2 1,2 0,56 0,56 0,52 0,52 1,14 1,14 0,38 0,38 0,71 0,71 Final roughness

(µm) 0,014 0,012 0,02 0,025 0,021 0,02 0,025 0,03 0,017 0,021 0,033 0,035 0,017 0,019 0,022 0,024

% of reduction 98 98 98 98 98 98 96 95 97 96 97 97 96 95 97 97

0 0,01 0,02 0,03 0,04 0,05 0,06 0,07

Ra after grinding step (µm)

Ra after polishing step (µm)

Here the disparity is less important. Most of the surfaces have Ra values around 0,03µm; but some are around 0,05µm.

Even if the difference is just about 0,02µm, proportionally, it represents almost 50% of difference.

It is important to check if it influences the final results.

Calculating the percentage of decrease of the Ra values, another interesting result can be found out; it does not depend of the initial roughness. Some surfaces with a 0,5 µm initial roughness decrease about 98% their roughness when other just decrease about 95%.

This diagram showed clearly that a link exists between the initial and the final roughness.

For a better understanding of this phenomenon, it could be interesting to polish surfaces with different initial roughness with exactly the same polishing process, to study the exact influence of the initial surface.

Indeed, the surfaces are measured with a stylus (two measurements) and the process parameters included in the DoE 2 change, this result needs to be confirmed.

(29)

7. Analysis of the turning marks

After the first test in 2007, the observation has been made that some turning marks were visible on the samples after the polishing step. One purpose of this second DoE was to get rid of these marks.

During the experiments, especially the experiment of the grinding step, the measurements made with the stylus were used to get profiles.

Figure 15: Extract of a profile got by the stylus measurement

Figure 15 shows the profile from a surface before polishing.

It’s hard to distinguish if these marks are turning marks or not but it does not seem that the marks are regular, which means they do not result from the turning process.

It seems that the process chosen for the grinding step removed the turning marks.

(30)

On the first test of the machine (second semester 2007), the turning marks did not appear after the grinding step but appeared after the polishing process.

It was important to check the samples we got after the polishing step.

The study of the profiles showed some marks which did not seem to be turning marks because of their frequencies. They seemed to be polishing marks.

The software MountainsMap can for example give such a profile from an area measurement.

nm

-225 -200 -175 -150 -125 -100 -75 -50 -25 0 25 50 75 100 125 150

0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540 560 µm

Length = 562 µm Pt = 246 nm Scale = 400 nm

Figure 16: profile of one of the surfaces

On this profile (Figure 16) some deep valleys which can be interpreted as turning marks can be found.

But these marks are not regular, a frequency can not be found. Between two valleys are sometimes 70 µm sometimes 110 µ m.

It seems that the process chosen removed the turning marks.

(31)

8. Analysis of the extra samples (Result polishing 2)

Sq values

0 5 10 15 20 25

G2 outer G2 inner F2 outer F2 inner D2 outer D2 inner

Sq (µm)

G2 outer G2 inner F2 outer F2 inner D2 outer D2 inner

Vibration (pulse/min) 2000 200 2000 2000 0 0

Force (g) 800 800 800 800 800 800

Repetitions 20 20 20 20 20 20

Diamond paste (µm) 3 9 3 9 3 9

Rotation speed (rpm) 500 700 500 300 500 300

The analysis of the three extra samples should inform us about other parameters values:

- a high rotation speed of 700 rpm - the absence of vibration

- the surface quality after 3 µm diamond paste

The surface F2 is the reference used during the DoE 2. Nevertheless the outer part is polished with a 3µm diamond paste.

The surface G2 is polished with a higher rotation speed during the first step (9µm diamond paste) and the same values as the reference for the second step (3µm).

The surface D2 is performed without vibration.

The first observation is that the surfaces got without vibration are smoother than the reference. The analysis of the DoE 2 showed that a high vibration pulse time was more efficient. It should be interesting to test a few vibration pulse time with all other parameters fixed to understand the real influence of the vibrations. Nevertheless, it can be noticed that the F2 sample had an initial roughness two times higher than the D2 sample, these result needs to be confirmed

The second observation is that a higher rotation speed performs a smoother surface. This result confirms the prestudy. Higher rotation speeds could be tested to find an optimum.

These samples present the best surface got until now with a Sq of 7 nm (sample G2).

Other tests need to be performed in order to refine the results.

(32)

VII. Conclusion of the project

The Design of Experiment performed revealed much information about the Strecon RAP-200:

- 10 repetitions are not enough to get a good surface

- a high vibration pulse time performs a smoother surface but a bad mirror-like surface

- two different configurations were found which give a smooth surface where the linear speed does not influence the final results

- a high force seems to perform better surfaces - a high rotation speed performs better surfaces

Finally, the best configuration was the combination of a high vibration pulse time and 40 repetitions, but that 20 repetitions might be enough.

The study of the profiles shows that the grinding step tested removes the turning marks; and that the marks do not re-appear after the polishing step.

The importance of the initial roughness has been revealed with this DoE 2. Indeed, it seems that the surfaces with the lower initial roughness are the one with the lower final roughness after the polishing step.

VIII. Future

This Design of Experiment was the first step of the evaluation of the Strecon RAP-200.

Now, other tests have to be done in order to improve our knowledge of this machine:

- test of other number of repetitions - test of higher force and rotation speeds - test more values of vibration pulse times - test other grade steels

- test of initial roughness

Moreover some parameters have not been studied at all during these five months such as:

- the lubricant - the robot speed

- the carrier and diamond paste - the repeatability

(33)

IX. Personal conclusion

These five months in Halmstad have been very enriching for me on different aspects:

• I discovered the work in a research laboratory. I have always been attracted by research and this internship permitted me to see it. Moreover my project was half way between research and industry and I really liked it; doing research with a direct purpose: improve the

knowledge about automated polishing for a machine which will be commercialized.

• I improved my knowledge about roughness and polishing. These subjects were almost unknown for me and I became conscious of surfaces at a different scale which is very enriching for an engineer, whatever the field he works in.

• I had the chance to discover the work environment in another country and to see the differences with France.

• Being in another country permitted me to improve my English since I worked on a project in Sweden in collaboration with a Danish company. Moreover I lived with people from other countries. So I could improve as well my technical as my personal English.

• I lived here surrounded by people from all over the world. I became more open-minded, living with European students. I discovered that even in Europe, people and ways of life are really different.

• We traveled in Sweden, Denmark, Norway and Leetonia so we learned about Scandinavian culture which is very different from the French.

To conclude, I can say that this internship has been one of the most amazing periods in my life. In five months I became aware of the diversity in Europe and in the world.

It has been an unforgettable and unique experience.

(34)

X. References

[1] Laser polishing of parts built up by selective laser sintering (2007) A. Lamikiz, J.A. Sanchéz, L.N. López de Lacalle, J.L. Arana

[2] Mechanics, Mechanism, and modeling of the chemical mechanical process (2001) Jiun-Yu Lai (Massachussets institute of technology)

[3] Material removal mechanism in Lapping and Polishing (2003)

C.J Evans, E. Paul, D,Dornfekd, D.A. Lucca, G. Byrne, M.tricard, F.Klocke, O.Dambon, and B.A.

Mullany

[4] Magnetic polishing of three dimensional die and mold surfaces (2007) Jeong-du Kim & Ill-hwan Noh

[5] Basic studies for the electro polishing facility at Desy (2003)

N. Steinhau-Kühl; R. Bandelmann, K. Escherich, D. Keese, M. Leenen, L. Lilje, A. Matheisen, H. Morales, P. Schmüser, E. Schulz, M. Seidel, J. Tiessen

[6] VESTA Sterile Technology (2007)

Brochure GEA process equipment division, Tuchenhagen (page 26) [7] Lapping and Polishing Basics

Applications laboratory report 54, South Bay Technology [8] Tool steel polishing (2008)

http://www.lucidaturastampi.it/upload/accesso/documenti/LUCIDATURA%20ENG.pdf [9] Rough Surfaces (1982)

Tom R. Thomas ISBN 1-86094-100-1

[10] Optimization of the polishing procedure using a robot assisted polishing equipment (2008) M. Gagnolet

[11] Digital Surf – France (2008) www.digitalsurf.com

[12] Taguchi Techniques for quality engineering (1988) Ross P.J.

ISBN 0-07-053866-2

(35)

Appendix

Appendix 1: DoE 2 table

Appendix 2: All Sq values DoE 2

Appendix 3: Sq values and visual aspect of the DoE 2

Optimization of the polishing procedure by using a robot assisted polishing equipment