• No results found

MIG/MAG brazing with Cold Metal Transfer

N/A
N/A
Protected

Academic year: 2021

Share "MIG/MAG brazing with Cold Metal Transfer"

Copied!
107
0
0

Loading.... (view fulltext now)

Full text

(1)

MIG/MAG brazing with

Cold Metal Transfer

MARIE ALLVAR

Master of Science Thesis Stockholm, Sweden 2012

(2)
(3)

iii

MIG/MAG brazing with Cold Metal Transfer

Marie Allvar

Master of Science Thesis MMK 2012:14 MKN 070 KTH Industrial Engineering and Management

Machine Design SE-100 44 STOCKHOLM

(4)
(5)

v

Master of Science Thesis MMK 2012:14 MKN 070 MIG/MAG brazing with Cold Metal Transfer

Marie Allvar Approved 2012-06-04 Examiner Ulf Sellgren Supervisor Ulf Sellgren Commissioner Swerea KIMAB AB Contact persons

Joakim Hedegård, Kjell-Arne Persson

Abstract

In the automotive industry a commonly used material is thin steel sheets coated with a thin layer of zinc for corrosion resistance purposes. Welding of this material, with the high temperatures involved, causes problems with zinc burn-off leading to reduced corrosion resistance. The zinc evaporation also causes arc disturbances leading to spatter formation, pores and difficulties achieving good visual weld quality. The joints are in many cases visible or semi-visible (e.g. visible after opening a door) and “A-grade” quality is demanded, meaning no visible discontinuities are allowed. It also implies a smooth and generally appealing appearance of the joint.

An alternative to welding is brazing, and laser brazing meets the requirements but the process is associated with high costs. In the national project “LEX-B”, funded by Vinnova, the possibilities of using arc brazing, in particular one MIG/MAG brazing and two TIG brazing processes, for some automotive applications are investigated. This master thesis is connected to the first part of the project and aims at compiling data for making a selection of the most promising process to investigate and optimise further. LEX-B is conducted in cooperation between Swerea KIMAB, Volvo Trucks, Scania CV and University West. The joint of interest is a lap joint between bottom and upper sheets of 1.2 mm and 0.8 mm respectively that represents a joint on the side of a truck cabin. The requirements are visual A-grade quality and tensile shear strength of 300 MPa.

In the project the MIG/MAG process Cold Metal Transfer (CMT) was compared to the TIG processes forceTIG and Plasmatron. CMT was experimentally investigated while the results for the other two processes were obtained partly from a previous study and partly from Volvo Trucks where tests were performed simultaneously. A system for measuring data for the process was developed, test specimens were brazed and examined visually and mechanical destructive testing was performed to ensure the tensile shear strength. Parameter studies were done for further process optimisation.

Keywords: CMT, Cold Metal Transfer, forceTIG, Plasmatron, brazing, arc brazing, laser

(6)
(7)

vii

Examensarbete MMK 2012:14 MKN 070 MIG/MAG-lödning med Cold Metal Transfer

Marie Allvar Godkänt 2012-06-04 Examinator Ulf Sellgren Handledare Ulf Sellgren Uppdragsgivare Swerea KIMAB AB Kontaktpersoner

Joakim Hedegård, Kjell-Arne Persson

Sammanfattning

Inom fordonsindustrin är ett ofta använt material stål i tunna ark belagda med zink p.g.a. zinkets korrossionsskyddande egenskaper. Vid svetsning av detta material uppstår problem med zink som förångas vilket leder till minskad korrosionsbeständighet. Förångningen av zink orsakar också störningar av ljusbågen vilket leder till sprut, porer och svårighet att uppnå god visuell svetskvalitet. Fogarna är i många fall synliga eller delvis synliga (t.ex. synlig efter att ha öppnat en dörr) och "A-kvalitet" efterfrågas, vilket innebär att inga synliga diskontinuiteter tillåts. Fogen ska upplevas jämn och tilltalande. Lödning är ett alternativ till svetsning, och laserlödning är en process som uppfyller kraven men är förknippad med höga kostnader. I det nationella projektet "LEX-B", som finansieras av Vinnova, undersöks möjligheterna att använda båglödning, specifikt en MIG/MAG-process och två TIG-processer, för dessa applikationer. Detta examensarbete är anslutet till den första delen av projektet och syftar till ta fram underlag för att välja den mest lovande av dessa processer för vidare utredning och optimering. LEX-B sker i samarbete mellan Swerea KIMAB, Volvo Lastvagnar, Scania CV och Högskolan Väst. Den undersökta fogen är en överlappsfog mellan en undre plåt med 1,2 mm tjocklek och en övre plåt med 0,8 mm tjocklek. Detta representerar en fog på sidan av en lastbilshytt. Kraven är visuell A-kvalitet och en drag-skjuvhållfasthet av 300 MPa.

I projektet jämfördes MIG/MAG-processen Cold Metal Transfer (CMT) med TIG-processerna forceTIG och Plasmatron. CMT testades experimentellt medan resultaten för de övriga två processerna erhölls dels från tidigare tester och dels från Volvo Lastvagnar där tester utfördes parallellt. Ett mätsystem for att dokumentera processdata utvecklades och de framställda lödfogarna undersöktes visuellt och med mekanisk förstörande provning för att säkerställa drag-skjuvhållfastheten. Parameterstudier gjordes för vidare optimering av processen.

Nyckelord: CMT, Cold Metal Transfer, forceTIG, Plasmatron, lödning, båglödning,

(8)
(9)

ix

NOMENCLATURE

In this chapter abbreviations used in the master thesis are listed and explained, in alphabetic order.

Abbreviations

CMT Cold Metal Transfer DAQ Data Acquisition DOE Design Of Experiments

DP Dual Phase

HAZ Heat Affected Zone MAG Metal Active Gas

MAX Measurement and Automation Explorer MIG Metal Inert Gas

Nd:YAG Neodymium-doped Yttrium Aluminium Garnet PTFE Polytetrafluoroethylene (trade name: Teflon) TIG Tungsten Inert Gas

(10)
(11)

xi

TABLE OF CONTENTS

Abstract ... v Sammanfattning ... vii NOMENCLATURE ... ix 1 INTRODUCTION ... 1

1.1 Background, problem description ... 1

1.2 Organisation and time plan ... 1

1.3 Objective ... 1

1.4 Delimitations ... 1

1.5 Methodology ... 2

1.6 Specification of requirements ... 2

1.7 Application ... 2

2 STATE OF THE ART ... 3

2.1 Zinc coated steel ... 3

2.2 Brazing ... 3

2.3 Conventional brazing techniques ... 4

2.4 Special processes applied on arc brazing ... 9

3 TESTING OF COLD METAL TRANSFER ... 13

3.1 Experimental set-up ... 13

3.2 Measuring system ... 17

4 EVALUATION AND RESULTS ... 29

4.1 Specification of visual A-grade quality ... 29

4.2 Evaluation and comparison ... 29

4.3 Optimisation analysis CMT... 37

5 DISCUSSION AND CONCLUSIONS ... 41

5.1 Discussion ... 41

5.2 Conclusions ... 44

6 RECOMMENDATIONS AND FUTURE WORK ... 45

ACKNOWLEDGEMENTS ... 47

REFERENCES ... 49

Appendix A. Brazing protocol ... 51

Appendix B. Program code, measuring program ... 55

Appendix C. Extended optimisation analysis ... 59

Appendix D. Visual results CMT ... 63

Appendix E. Cross section views CMT ... 85

Appendix F. Manual Measure & log ... 89

(12)
(13)

1

1 INTRODUCTION

In this chapter the outlines of the project are presented; why and how the work was executed.

1.1 Background, problem description

In the automotive industry a commonly used material is thin steel sheets coated with a thin layer of zinc for corrosion resistance purposes. Welding of this material, with the high temperatures involved, causes problems with zinc burn-off leading to reduced corrosion resistance. The zinc evaporation also causes arc disturbances leading to spatter formation, pores and difficulties achieving good visual weld quality. The joining method brazing is interesting to investigate because of its significantly lower temperature and heat input, leading to lower distortions, reduced impact on the zinc layer and less disturbances.

Laser brazing presents very good results when joining zinc coated steel sheets; however it is associated with high costs. The investment cost is considerably higher than for other brazing processes, why it is of interest to find arc brazing processes with the same good visual and mechanical results. New processes for MIG/MAG and TIG brazing have shown potential of meeting the requirements but the results obtained in previous projects have not been verified sufficiently.

Previous projects at Swerea KIMAB have shown how to reduce instabilities and cracking in arc brazing. Remaining tasks are to manage both visual A-grade quality joints and sufficient joint strength.

1.2 Organisation and time plan

The project LEX-B (Lean EXterior brazing, part B), to which this degree project (also referred to as master thesis) is connected, is a project performed in cooperation between Swerea KIMAB, Volvo Trucks, Scania CV and University West. LEX-B follows the pre-study KEEX, where interesting results were obtained but not fully investigated due to the limited time frame and budget.

Supervisors for the degree project are at Swerea KIMAB Joakim Hedegård and Kjell-Arne Persson, and at KTH Ulf Sellgren. The time reserved for the project is 20 weeks of full-time work, and the starting date is January 23, 2012.

1.3 Objective

The main objective of LEX-B was to evaluate some arc brazing processes as possible cost efficient alternatives to laser brazing. Both technical and economical evaluations of the MIG/MAG process Cold Metal Transfer (CMT) and the TIG processes forceTIG and Plasmatron will be made. The objective of this master thesis was to investigate CMT and meet the requirements in visual quality, cost efficiency and mechanical strength.

1.4 Delimitations

The literature study included the three arc brazing processes CMT, forceTIG and Plasmatron while the experimental part was focused on CMT, prioritised due to its

(14)

2

availability at Swerea KIMAB. Other delimitations concern the amount of different materials, brazing filler wires and processes being tested, as well as settings of variable parameters of the brazing equipment and joint configurations. This was limited due to limited time for conducting the tests. The processes and materials investigated were selected based on previous test results. In the LEX-B project, when one process is chosen, further optimisation is still to be done. This is not part of the master thesis, due to the limited time assigned for master thesis work.

1.5 Methodology

A literature study was performed in the initial phase of the project, which resulted in the starting point for the later experiments. A measuring system was developed and lap joints of zinc coated steel sheets were brazed with the CMT process. The brazed joints were analysed and compared to the demands and at least partly compared to the results obtained for the other two processes. Following optimisation trials were performed but only to a limited extent.

1.6 Specification of requirements

The main requirements set for the specific joint are nominal tensile shear strength σ≥300 MPa, see Figure 1, and visually high-quality brazed seams with so-called A-grade quality (specified in chapter 4.1 Specification of visual A-grade quality). This also includes a requirement for a dense surface layer, demanded for making painting of the surface with good results possible. The area used for calculating the nominal strength is the cross section area of the thinner sheet.

Figure 1. Tensile shear testing.

1.7 Application

The application for the studied arc brazing processes is a lap joint on the side of a truck. However, if good results can be produced and verified using at least one of the three processes, other possibilities are created since many applications use the same type of material and experience the same type of problems.

(15)

3

2 STATE OF THE ART

In this chapter existing knowledge within the area of the master thesis is presented. This chapter is the result of a literature study performed in the beginning of the project.

2.1 Zinc coated steel

A zinc layer is applied to the steel sheet in order to prevent corrosion. The zinc can protect the steel in two ways; by a protective physical surface layer and by cathodic protection. When the new zinc layer is exposed to the atmosphere the zinc reacts with oxygen and forms zinc oxide, which in turn reacts with water molecules to form zinc hydroxide. The zinc hydroxide reacts with carbon dioxide and forms a thin layer of zinc carbonate which protects the surface of the steel. [1] If the physical layer of zinc is damaged, the steel will still be protected cathodically, meaning that the zinc layer acts as a sacrificial anode. However, it is important to keep the layer as intact as possible since the cathodic protection cannot span wide gaps in the layer.

The zinc coating on the steel sheets melts at 419°C and evaporates at 907°C. The melting temperatures of copper based fillers used when brazing are typically between 800°C and 1100°C, meaning that zinc melting and vaporisation occurs even when brazing, however not at the level that is present when welding, where the wires used have melting temperatures of more than 1500°C. The arc temperature is significantly higher still, between 6000°C and 20000°C in both cases of brazing and welding, but with brazing it is possible to melt the filler wire with lower energy input.

2.2 Brazing

Brazing is a generic name for a variety of processes for joining metal. Similar methods are welding and soldering, with the difference being the temperature at which the process is performed. When welding the base material is molten with or without a filler metal being added. In common for all soldering and brazing processes is that the melting temperature of the filler is lower than the melting temperature of the base metal. Consequently, only the filler metal is molten and wets the base metal. For soldering the melting temperature of the filler is below 450°C and for brazing the temperature is above 450°C. The processes and equipment used for brazing are essentially the same as for welding, meaning that there is a wide variety of processes. Some of the most common processes are presented here, see Figure 2.

(16)

4

2.3 Conventional brazing techniques

As shown in Figure 2 the common brazing techniques presented here are divided into two categories, arc brazing and laser brazing. Arc brazing includes MIG/MAG, TIG and plasma brazing and have in common that an electric arc provides the heat. Another classification can be made from the melting of filler wire, where MIG/MAG differs from laser, TIG and plasma. In MIG/MAG the filler wire is used as an electrode conducting the electricity, whereas in the other processes the wire is fed independently of the heat source.

2.3.1 MIG/MAG

MIG/MAG brazing is an arc brazing process where the continuously fed filler wire is also one of the electrodes. The difference between MIG and MAG is in the shielding gas used; MIG uses an inert gas while MAG uses an active gas to stabilise the arc. The active gas component can also react with the molten metal. The softer wires used in brazing increases the demands on hose package and wire feed system. A plastic or PTFE liner is required for the hose and an additional wire feed motor on the torch (push-pull system) might be required if the hose is long (>1.5 m for automatic brazing and >3 m for semiautomatic). Other than this the equipment is the same as for MIG/MAG welding, see Figure 3. In MIG/MAG brazing short arc or pulsed arc is mainly used in order to keep the heat input low. [3]

Figure 3. MIG/MAG concept. [2]

Since the filler wire is an electrode, the melting rate and consequently the deposition rate will depend on supplied arc energy. It is therefore more difficult to alter the

(17)

5 geometry of the bead than with a process where the wire feed is independent of the heat source. The bead is often rather large when MIG/MAG brazing since the melting efficiency of filler wire is very high. The process has very good gap bridging capabilities, see Figure 4.

Figure 4. MIG/MAG brazed joint with 1mm gap, above view (left) and cross section (right). [2]

2.3.2 TIG

TIG (Tungsten Inert Gas) brazing is an arc brazing process which utilises the same equipment as is used for TIG welding, see Figure 5. The non-consumable tungsten electrode is used for creating an arc between the electrode tip and the work piece and the inert shielding gas used is often pure argon. The filler wire is fed into the arc independently of the heat input making it relatively easy to alter the geometry of the seam.

Figure 5. TIG concept. [2]

The TIG is sensitive to changes in distance between electrode and work piece, due to the bell-shaped geometry of the arc. TIG brazing is a relatively slow process which causes

(18)

6

low working productivity as well as long times of heat impacting on the base material which leads to distortions.

2.3.3 Plasma

Plasma brazing is an arc brazing process with similarities to the TIG process. It uses a non-consumable tungsten electrode but as opposed to TIG it uses two gas flows; one plasma gas and one shielding gas. The inner nozzle of the plasma torch surrounds the electrode so that it does not stick out. The filler wire is fed into the focused plasma arc independently of the power source, allowing large variation of the geometry of the seam. Water cooling is often used for plasma torches, to carry away the heat created in the process and to lengthen the life of the torch. [4] The plasma brazing concept is shown in Figure 6.

Figure 6. Plasma brazing concept. [2]

The plasma gas is heated by the arc and is ionised, consequently conducting electricity. The ionised gas is what is known as plasma. This is however not specific for the plasma process; ionised gas is present with all arc brazing and -welding processes. The outer nozzle provides the shielding gas, which is often the same gas as the plasma gas (commonly argon) but only used for shielding. Compared to TIG, the plasma process is much more energy dense and less sensitive to changes in distance between electrode and work piece. This is explained by the shape of the arc. Figure 7 shows the constricted plasma arc in comparison with the TIG arc.

(19)

7

Figure 7. Comparison plasma arc (left) and TIG arc (right). [5]

Examples of plasma brazed joints are given in Figure 8.

Figure 8. Plasma brazed joints. [2]

2.3.4 Laser

Laser brazing uses a laser to provide the heat, either Nd:YAG, or diode lasers are used. The concept of laser brazing is shown in Figure 9.

Figure 9. Laser brazing concept. [2]

(20)

8

A laser beam has only one wavelength and its divergence is minimal. This makes it possible for the beam to travel relatively long distances still providing a narrow focal spot. Due to the high power density of the beam it does not lose much of its power over long distances. By defocusing the beam the laser spot size can be altered.

The quality of the joints is very high and the heated area is narrow resulting in little heat effect on the adjacent base metal, a small heat affected zone (HAZ) and low distortions. High brazing speed, of more than 3 m/min, is documented and the need for post-treatment can be avoided making it possible to paint directly on the brazed joint. In the Audi TT Coupé there is a fully visible laser brazed joint, see Figure 10.

Figure 10. Visible laser brazed joint in Audi TT Coupé. [2]

The main disadvantages of laser brazing are the high investment cost and the high demands on precision in fixturing. More examples of laser brazed joints are presented in Figure 11.

(21)

9

2.4 Special processes applied on arc brazing

2.4.1 CMT, Fronius.

Cold Metal Transfer is a MIG/MAG process developed by Fronius [7] for welding and brazing. The idea is to minimise both spatter formation and the heat input, thus causing less damage to the base metals being joined. This is achieved by mechanically retracting the wire in the moment short circuiting occurs. Due to the retraction, the current rush that occurs with conventional MIG/MAG equipment is avoided or decreased. One drop of metal is transferred, by mechanical retraction and not by electromagnetic pinch-effect. The process is claimed to be, due to the controlled droplet transfer, spatter-free. The main feeder feeds wire at a constant speed, and the retraction is done by the second wire drive placed on the torch, up to 90 times per second. The excess, retracted wire is stored in a wire buffer on the hose, see Figure 12.

Figure 12. Wire buffer.

One advantage stated by the manufacturer is the ability to join dissimilar metals such as steel and aluminium, where the steel is brazed and the aluminium welded. Another possibility the manufacturer accentuates is of great interest to this master thesis, the “almost spatter-free brazing of hot galvanised […] sheets using a filler wire made from an alloy of copper and silicon.” [8] The retraction sequence is shown schematically in Figure 13.

Figure 13. CMT retraction sequence. [9]

2.4.2 forceTIG, ewm.

ForceTIG®, a registered trademark and patented by ewm [10], is a TIG process where the arc is constricted by extreme water cooling of the electrode, thus increasing the energy density of the arc. The area for electron emission is narrowed by the cooling of the electrode and the arc is focused. This enables high brazing speed, up to 5 m/min,

(22)

10

with good arc stability. [11] Ewm accentuates advantages such as low investment cost (compared to laser), high speed, high efficiency and robust design. [12] Figure 14 shows the torch and wire feed device as well as a comparison between forceTIG and conventional TIG. The energy consumption of the forceTIG is stated to be 90 % less than for a corresponding laser weld joint. [11] Distortions are claimed to be low, due to the focused arc and reduced heat transfer.

Figure 14. forceTIG, comparison with TIG (left) and torch with wire feed system from Scansonic (right). [11]

2.4.3 Plasmatron, Inocon.

The Plasmatron®, a registered trademark by Inocon Technologie [13], is a variant of a TIG process with some features of a plasma process. It can be referred to as a “constricted TIG”, meaning that instead of two gas flows it uses only one, like the TIG process, and “constricted” refers to the narrower orifice of the Plasmatron nozzle, compared to a conventional TIG nozzle. The tip of the non-consumable electrode is situated outside of the nozzle, as opposed to plasma, and the distance between electrode tip and work piece is short. This can cause contamination of the electrode from air, dust and evaporated zinc. [14] Another difference is that the plasma is produced outside of the Plasmatron torch and not inside, causing less thermal stress on the equipment. The constriction of the plasma beam comes from an angled nozzle creating a focus point far away from the nozzle.

On Inocon’s web page [13] advantages like low insertion of heat and low distortion, high joint stability, low operation costs, and an “optical joint” - meaning a clean and smooth surface - are listed. The total efficiency of the process is claimed to be up to 80 % as compared to 30 % of the conventional plasma process. Plasmatron allows high brazing speeds, up to 3 m/min, which is advantageous when brazing zinc coated steel since less zinc burn-off will take place, due to less heat input. Figure 15 shows the concept and the Plasmatron torch.

(23)

11

(24)
(25)

13

3 TESTING OF COLD METAL TRANSFER

In this chapter testing of the CMT is presented.

3.1 Experimental set-up

Testing of CMT was performed at Swerea KIMAB in Stockholm. The approach was to not follow a pre-defined test plan according to a “design of experiments” (DOE) but rather to use a qualitative approach where each test is evaluated, based on the behaviour of the process and appearance of the joint, during and immediately after brazing. Parameters are then altered accordingly. This is due to the extensive amount of alterations possible. Altering all of the parameters in a consistent manner would result in several hundreds of tests for each wire which was not achievable within the limited time frame, and many of the tests produced in that way would be unusable.

3.1.1 Brazing equipment and robot

The equipment used was Fronius CMT Advanced with power source TPS 4000 with maximum welding current of 400 A. The CMT equipment is shown in Figure 16.

Figure 16. CMT equipment.

Beyond short arc and spray arc, the CMT Advanced system offers possibilities to weld and braze with four other techniques; CMT, CMT + pulse, CMT Advanced (reversed polarity) and CMT Advanced + pulse. In this project only the CMT technique was used. The robot, Motoman HP20, and positioner used are shown in Figure 17.

(26)

14

Figure 17. Robot (with fixture).

3.1.2 Material

The base material was the same for all tests, sheets of mild steel, hot dip galvanized with 7 μm zinc coating. Ultimate strength of the steel is around 320 MPa. A 1.2 mm thick bottom sheet was joined with a 0.8 mm thick upper sheet. Both sheets are 1200 mm in length and 100 mm in width, with a lap of approximately 50 mm. Filler wire materials are listed in Table 1, and all of the wires were 1 mm in diameter. The IDs for the wires given in the table are used in the thesis for identifying the different tests.

Table 1. Filler wire material.

Producer Name ID Main composition Ultimate strength

[MPa] Elongation [%] Melt temp. [°C]

ESAB 19.30 A CuSi3Mn 350 40 965 – 1035

ESAB 19.40 B CuAl7.8Mn0.5Si0.1 420 40 1030 – 1040

UTP A 3421 C CuAl5Ni2Mn0.2 >350 (has reached 600) 45 1060 – 1085

UTP A 320 D CuSn12P0.3 300 25 825 – 990

Bedra Bercoweld B60 E CuSn6SiMn 359 44 900 – 1040

Castolin Castomag 45703 F CuSn8 260 – 295 20 – 25 875 – 1025

The filler materials chosen have differences in composition, melting temperature and their ability to wet the base metal, which affects the strength of the joint. The viscosity of the molten filler metal will affect the appearance of the joint, as well as the strength. All filler metals are copper based, but with rather big differences in alloying elements. The shielding gases used were MISON® 2 and MISON® Ar. MISON® is a registered trademark by AGA. [15]

(27)

15 MISON Ar consists of Ar + 0.03 % NO and is often used for TIG welding. It is also used for MIG welding of aluminium and stainless steel, as well as MIG brazing of galvanized steel. It is said to provide a stable arc. [16]

MISON 2 consists of Ar + 2 % CO2 + 0.03 % NO and is used for MAG welding of stainless

steel and especially for “common” austenitic, ferritic and duplex steel. It is suitable for short-, spray-, and pulsed arc welding. The active component help stabilise the arc. [16]

3.1.3 Fixture

A fixture for holding the work pieces in place when brazing was manufactured. In order to obtain optimal and verifiable results it is important that the two sheets are precisely placed relative each other and that the joint is precisely placed relative the torch on every trial. The fixture is presented in Figure 18.

Figure 18. Fixture CMT, in “correct” position (left) and upright position (right).

The sharp edge, from cutting the steel sheet, of the upper plate is consistently placed facing the bottom plate, to obtain similar condition for each trial and to create a small gap between the sheets where evaporated zinc can flow out. The sheets are brazed as delivered, without cleaning.

(28)

16

3.1.4 Parameters and synergic brazing

A number of basic trials were made in order to get a rough span in which the parameters can be altered. The parameters were altered in intervals according to Table 2.

Table 2. Parameter intervals.

Parameter Min value Max value

Gas flow [l/min] 15 18

Travel speed [cm/min] 120 150

Wire feed speed [m/min] 5 8.5

Stick-out [mm] 14 18

Offset (x-position) [mm] -2 0

Torch angle α [degrees] 20 35

Torch angle β [degrees] 0 0

Arc length correction [%] -30 25

Dynamic [%] -5 3

Tilt angle γ [degrees] 40 40

Wire diameter [mm] 1 1

Parameters that are not self-explanatory are illustrated in Figure 19.

Figure 19. Illustration of parameters.

All of the tests with the CMT machine were performed in “synergy” mode. Different wire materials and diameters in combination with different shielding gases have different pre-set parameter settings, so called synergy settings or synergy lines. Synergy lines are developed by the power source manufacturer. They are made up by optimised parameter settings for a number of wire feed speeds. The user only has to set a wire feed speed and the power source automatically sets all other parameters. However, since the application may differ from the application used by the manufacturer when creating the synergy lines, there are some adjustment settings available for the user.

Different synergy lines have different possible adjustment settings. Using a certain synergy setting the “boost” parameter can be altered between -5 and 5 from minimum to maximum value. This implies that the “boost current” is altered which gives control of the heat transferred to the base metal. Using another synergy setting the “hot start” time can be altered between -5 and 5 (0 and 200 ms, respectively). The different parts of the current-, voltage- and wire feed speed curves are presented in chapter 4.3.2

Characteristic parameters. All of the synergy settings have a parameter called arc length

correction, spanning from -30 % to 30 %, which affects the time during which the wire is retracted. In the (simplified) test protocol appended in Appendix A. Brazing protocol, the parameter with different names for different settings is called “dynamic” for all wires and is given in the scale from -5 to 5, in accordance with the settings options. On each set

(29)

17 of steel sheets four tests with different settings were made. However, for some of the wires there was no synergy line developed, why the available synergy lines for other wires were tested in order to find out which line works best.

3.2 Measuring system

In order to fully understand the process and the possible formation of discontinuities it is important to interpret the output from the power source. This is also important for optimising the process parameters.

3.2.1 Voltage and current

To measure the current and voltage instantaneously was a first step since they are the two most significant parameters and are used to calculate the heat input. This was done with hardware from National Instruments [17], a data acquisition (DAQ) unit consisting of a chassis with one module for measuring. An application was programmed in LabVIEW. LabVIEW is a graphical programming environment for developing systems for testing, measuring and controlling. [18] The full structure of the programmed application is appended in Appendix B. Program code, measuring program, and also explained in chapter 3.2.5 Programmed application, code. The original system developed for measuring and logging of current and voltage consisted of an Ethernet connected chassis (NI-cDAQ9188) with one module (NI-9229) that offers a maximum sampling rate of 50 kS/s and can handle voltage of maximum 60 V. The graphical interface is presented in Figure 20. Specifications of the hardware are presented in Table 3.

Figure 20. Graphical interface.

In the program, a number of settings are required. These are chosen in the box in the upper left corner and are shown in Figure 21 - Figure 25.

(30)

18

Figure 21. Graphical interface, logging settings

Figure 22. Graphical interface, saving settings

(31)

19

Figure 24. Graphical interface, channel 2 settings

Figure 25. Graphical interface, counter settings

This program is executed during brazing, stores all data points and presents a graph of current and voltage when the data acquisition is finished. In the graph it is possible to choose an interval within which to calculate mean values of voltage, current and heat input, and to zoom and scroll through the acquired data points.

For further analysis an additional program was developed. In this program the data points saved from the data acquisition is read and displayed in graphs showing different relations as shown in the following list.

 Voltage as a function of current  Voltage as a function of time  Current as a function of time

 Voltage/current as a function of time  Arc power as a function of time

 Time derivative of voltage as a function of time  Time derivative of current as a function of time

Pictures of graphs (with white background for printing reasons) can be saved, as well as mean and RMS values for example. Data for chosen intervals can also be exported to other file formats for further analysis in other programs such as Matlab.

(32)

20

3.2.2 Wire feed speed

The measuring program is prepared for measuring wire feed speed using a pulse encoder on the feed unit on the torch. The pulse encoder, see Figure 26, is to be placed on the driving wheel and gives 4096 pulses/revolution.

Figure 26. Pulse encoder.

It is of great interest to see how the retraction of the wire affects the seam and how different settings affect the retraction. Using a digital input module (NI-9421) three signals are read into LabVIEW. Two of the signals (A and B in Figure 27) are slightly shifted and therefore enables the inbuilt counter of the chassis to count up if the wire feed speed is positive, and count down if the speed is negative (wire is retracted). The third signal sends a pulse for each full revolution.

Figure 27. Encoder pulse trains.

The signal from the counter can subsequently be visualised in a graph. Due to lack of time the wire feed speed measuring unit was not fully developed but is left to further work. When the wire feed speed measurement part is completed the analysis program can be expanded with graphs according to the following list.

 Wire feed speed as a function of time  Voltage as a function of wire feed speed  Current as a function of wire feed speed

(33)

21

3.2.3 Upgraded voltage measurement

The measuring system was also expanded with a module (NI-9221) for measuring voltage with a sample rate of maximum 800 kS/s. However, this module was not used for this thesis since the module was obtained after most of the trials were finished. Calculating instantaneous heat input is more complex when measuring voltage and current with different sample rates, since the vectors of data will be of different lengths. A possible solution is to copy each current value (with lower sample rate) and insert again to make the vectors equally long, but that would only work if one sample rate is a multiple of the other. However, it is possible to calculate heat input from mean and/or RMS values of current and voltage. Another option is to measure voltage with both sample rates, one used for calculating heat input (lower rate) and one used for analysing the curve (higher rate). The chassis with the three modules is shown in Figure 28.

Figure 28. Chassis with three modules. Table 3. Specification hardware. [17]

Module NI-cDAQ-9188 NI-9229 NI-9221 NI-9421

Type AI/AO, DI/DO, counter AI, voltage AI, voltage DI

Sample rate - 50kS/s 800kS/s Max clock rate 10kHz

Channels Max 256 (8 slots) 4 8 8

Resolution 32-bit (counters) 24bit 12bit -

Range - ±60V ±60V -

Simultaneous input - Yes No -

Meas. type - Differential Single ended -

3.2.4 Camera

Some trials were made with camera surveillance; the camera being mounted and set-up by University West. Figure 29 shows the torch with camera mounted. A few trials with a high speed camera were also performed in order to see the movement of the wire and analyse the behavior of the mechanical wire retraction.

(34)

22

Figure 29. Torch with camera mounted.

3.2.5 Programmed application, code

The entire program is depicted in Figure 30 and explained in detail in the following sections. Note that the visual structure is altered in the following “zoomed” pictures, but the connections are the same. Note: the program code presented here is not the latest version; however the principle is the same. Manuals for the latest versions of the measuring and analysing programs are appended in Appendix F. Manual Measure & log and Appendix G. Manual Read & analyse.

Figure 30. Overview of measuring program

Set-up module 1

The ”first” module, NI-9229, is the module that times the acquisitions for all three modules since its “start task” virtual instrument (VI) (1), see Figure 31, is to the far left in the “flat sequence structure”. In a flat sequence structure events occur from left to right. Sample mode, sample rate, samples per channel (which in the case of continuous sampling implies buffer size) and task for which to acquire data are chosen in the graphical interface of the program. The tasks are defined in a program called “Measurement and Automation Explorer” (MAX) where the specific channels of the modules are chosen. In MAX a custom scale can be defined for scaling the acquired data. When using more than one channel on the same module, it is recommended to use tasks

(35)

23 instead of choosing the physical channels directly in the programmed application. The property node “Xscale.Multiplier” (2) divides the X-scale of the plotted graph with the sample rate resulting in an X-scale showing the time in seconds.

Figure 31. Set-up module 1

Set-up module 2

The start task VI (3), see Figure 32, for the second module (NI-9221) for faster voltage measurements is placed to the right in the flat sequence structure and is also in a case structure meaning that if the boolean control is set to “false” no data acquisition for the module will take place. This control is a setting in the graphical interface. For this module the physical channel is chosen and a custom scale can be applied to it. The “create task” VI (4) then creates a task from the chosen settings. The “unit” control has to be set to “from custom scale” if a custom scale is used. The controls for samples per channel and sample mode are connected to both this module and module 3 (for wire feed speed). This is in order to time the acquisitions accurately. The property node (5) assigns the chosen name to the plot and the scale label of the graph.

(36)

24

Set-up module 3

The third module is for counting pulses from the pulse encoder and this is as mentioned not fully developed and tested. The start task is in similarity with module 2 placed in a case structure and to the right in the flat sequence structure. An angular encoder (6), see Figure 33, is introduced into which settings for “counter”, meaning channel, initial angle, decoding type, pulses per revolution, units and custom scale name are wired. These settings are made in the graphical interface. The sample clock (7) is connected to the sample clock for module 2 for timing reasons, but the sample rate and source is set manually. The source should be chosen as “/cDAQ1/ai/SampleClock”, to access the built-in counter of the chassis.

Figure 33. Set-up module 3

Reading and writing data

Directly ”after”, to the right of, the flat sequence structure containing the start task VIs the “read” VIs (8), see Figure 34, are placed in a while loop that executes its content until the boolean control ”stop” is pressed in the interface. The arrays are converted to “dynamic data” and stored by the “write file” VIs. Filenames are assigned for storing the data in binary files (.tdms). The stop button should be pressed when the data is acquired, making the “read file” VIs read all of the data points stored in the binary files and plot them in waveform graphs, one graph for each file. The case structures for module 2 and 3 are connected to the boolean controls in the set-up deciding if the modules should be used. Without the case structures the “read file” VIs return error messages due to non-existent files to read.

(37)

25

Figure 34. Reading and writing data

Zoom and calculate

The next part of the program is for analysing the data. A while loop is placed around the rest of the program; when this while loop is stopped the execution of the program is terminated and no further handling of the data can be done unless the file is read in the additional analysis program. The case structure within the while loop contains calculations of mean, RMS, minimum and maximum values. The five property nodes to the far left in the case structure are used to zoom in on particular intervals in the graphs. The upper graph in the interface decides the interval for the lower graphs, why the property node for that graph (9), see Figure 35, reads the values of the X-scale and writes it to the property nodes for the two lower graphs. The interval is chosen in the graph with the mouse pointer in the interface. If the box “Update results continuously” is ticked, the result indicators will change immediately as the interval is changed. Subarrays (10) with the chosen data are created from which all of the subsequent calculations are made. For the calculations it is possible to choose whether the mean or the RMS values should be used, which efficiency factor should be used for heat input calculations and what travel speed was used.

(38)

26

Figure 35. Zoom and calculate

Another option is if module 1 or module 2 should be used for calculations. This is an option since the instantaneous heat input cannot be calculated for arrays of different lengths and the module with higher sample rate is going to be used for voltage measurements. The array of data acquired with higher sample rate will contain a larger amount of data than one with lower sample rate for the same time interval. Therefore, it is possible to use mean or RMS values acquired with module 2 instead of instantaneously calculating the heat input with module 1. The calculations will not be explained in detail since they are logical and quite repetitive. As can be seen in the code no calculations are made for the wire feed speed, module 3, since it is not finalised and the encoder has not been available for testing.

(39)

27

Saving results

The results of the calculations are presented in indicators on the graphical interface, exemplified in (11), see Figure 36, and an array is built with all of the results (12). For every result there is an option to save them to a separate file (or append to an existing file) by ticking a box, a boolean control. The choices are made in (13) (showing only a few of the 15 results to choose from) creating yet another array of results that is stored in a file specified in the case structure (14) when clicking the boolean control “save”. The file created does not have a specified file extension but can be chosen when creating the file.

Figure 36. Saving results

If further analysis with other programs is required, it is possible to save larger intervals of data in a file format specified in the graphical interface. The arrays saved are the same arrays as chosen in the graph with the mouse pointer, but the save buttons (current, voltage module 1 and voltage module 2) are separate from the save button for the results, making it possible to save results for some intervals and complete sets of data for other intervals. Figure 37 shows the part of the program for saving the chosen intervals.

(40)

28

(41)

29

4 EVALUATION AND RESULTS

In this chapter the results are evaluated and compared to the requirements.

4.1 Specification of visual A-grade quality

When evaluating the visual quality the requirements are to a high extent subjective, the customer or user decides what is aesthetically good quality and it can be difficult to quantify these properties. However, some basic requirements were set and these make up the foundation for the comparison between the different parameter settings and between the different processes. These requirements are presented in Table 4.

Table 4. Visual requirements.

Property Requirement

Spatter Not accepted

Bead shape Flat, even

Bead cross section area Small

Zinc burn off Not accepted

Zinc melting Accepted

Visual pores Not accepted

Soot Accepted, unwanted

Visual cracks Not accepted

Uneven fusion line Not accepted

Surface layer Dense (for painting)

All pictures of CMT brazed joints in this thesis were taken immediately after brazing and no cleaning was done. The soot and oxides on the sides of the joints is easy to remove but the amount of soot and oxides might be interesting for optimising and analysing purposes why it was not removed. It is a result of the thermal process and is affected by for example shielding gas, surface layer and possible dirt and oil on the surface.

The zinc burn-off requirement is not a purely aesthetical requirement but is important for the corrosion resistance of the joint. Looking at the rear of the joined sheets some partial melting of zinc is visible as a shiny stripe along the joint. This is acceptable and hard to avoid. Zinc “burn-off” however, is visible as white powder, zinc oxide, which falls off easily. This is not acceptable.

4.2 Evaluation and comparison

A visual comparison of the best joints was done, as well as mechanical destructive testing. Some samples were cut and molded into plastic cylinders and examined using a microscope.

4.2.1 Visual appearance CMT

Some of the visually best results obtained with CMT are shown in Figure 38. The samples in the figure were sent to Scania for tensile testing. More results are presented in Appendix D. Visual results CMT.

(42)

30

Figure 38. Examples of joint appearance. From the top down: A92, A93, B62, C7, D1, D3, E10, F1, F3.

Samples from selected specimens with relatively good visual appearance were cut, molded into transparent plastic cylinders, polished and etched, see Figure 39. This was not done for all of the specimens since no specific demands were set for the geometric features. The ones that were cut, polished and etched are just to be viewed as examples of how cross sections of CMT brazed seams look. More pictures of cross sections are presented in Appendix E. Cross section views CMT.

(43)

31

Figure 39. Polished and etched samples.

Using a microscope, results such as effective throat thickness, leg length and other geometric features can be measured, see Figure 40. The effect of the heat on the base material can also be seen in this figure (crystalline structure).

Figure 40. Cross section of brazed seam (CMT, wire ESAB 19.40 left, ESAB 19.30 right).

4.2.2 Visual appearance Plasmatron

The results presented for the Plasmatron equipment were received partly from Volvo Trucks in Umeå where the equipment was tested in project LEX-B, and partly from the pre-study KEEX. A typical result obtained with Plasmatron is shown in Figure 41.

(44)

32

The cross sections of two Plasmatron brazed seams are shown in Figure 42.

Figure 42. Cross section view Plasmatron. [19]

Figure 43 shows results obtained in the pre-study KEEX.

Figure 43. Plasmatron brazed joint (different set-up), cross section (left), above view (right) [6]

4.2.3 Visual appearance forceTIG

Due to circumstances beyond the control of the participating companies, the forceTIG equipment could not be tested according to time schedule and the results presented in this report are therefore from the pre-study KEEX. These trials were performed at ewm in Mündersbach, Germany, by employees from ewm, Swerea KIMAB and Scansonic. The set-up used in the pre-study was not the same as the current set-up why the results cannot be used for a direct comparison. Some properties such as visual appearance can nevertheless be judged. Some of the visually best results obtained with forceTIG are shown in Figure 44.

(45)

33

Figure 44. Joint appearance forceTIG.

4.2.4 Comments on visual appearance

As seen in the figures the cross section area of the bead is significantly smaller for forceTIG and Plasmatron than for CMT. This is a logical consequence of the much higher wire feed speed of CMT, around 5-6 times higher. The two TIG variants also provide a more flat and smooth surface than CMT. All of the processes have potential of producing no or very little spatter which was one of the demands for an A-grade joint. Visual pores, cracks and zinc evaporation are also preventable. CMT gives the largest amount of soot apparently, but this is believed to be easily removed in the process of cleaning that takes place before painting. The CMT process sometimes shows a tendency of deviating from its linear path, which might be a consequence of the bent wire from the bobbin being fed through a hose package and twisting in different directions depending on the position of the robot.

4.2.5 Mechanical properties CMT

In order to get results on mechanical strength for the CMT brazed specimens, preliminary tests were performed by cutting the brazed sheets to proper size and tensile-shear testing them at Swerea-KIMAB. The reason for the preliminary testing was that a batch of tests was sent to Scania for testing but the results were expected after the publication of this report. However, the results obtained give a sufficient clue on the

(46)

34

actual performance of the CMT process and the different wires. One test was chosen for each wire material, and three samples from each test were cut and mechanically tested, except for wire F2 where only two samples were tested. Tensile shear tests were performed on test specimens with 48 mm width. With the thickness of the thinner sheet of 0.8 mm the nominal area was . The tests were performed with a speed of 10 mm/min. The results in ultimate strength are presented in Table 5.

Table 5. Results ultimate strength CMT

Test

no ID no Wire Nom. area Max force [kN] strength [MPa] Nom. ultimate Comment Fracture location

1 A89 19.30 38.4 12.57 327.3 Base 2 A89 19.30 38.4 12.53 326.3 Base 3 A89 19.30 38.4 12.48 325.0 Base 4 B57 19.40 38.4 12.47 324,7 Base 5 B57 19.40 38.4 11.59 301.8 FL 6 B57 19.40 38.4 11.73 305.5 FL 7 C13 A3421 38.4 12.39 322.7 Base 8 C13 A3421 38.4 12.50 325.5 Base 9 C13 A3421 38.4 12.49 325.3 Base 10 D2 A320 38.4 8.87 231.0 <300 FL/Braze 11 D2 A320 38.4 6.83 177.9 <300 FL/Braze 12 D2 A320 38.4 7.18 187.0 <300 FL/Braze 13 E8 B60 38.4 12.43 323.7 Braze 14 E8 B60 38.4 12.32 320.8 Braze 15 E8 B60 38.4 10.84 282.3 <300 Braze 16 F2 45703 38.4 12.55 326.8 Base 17 F2 45703 38.4 12.48 325.0 Base

Four of the tests did not meet the requirement of 300 MPa. The strength of test D2 was much lower than average, implying that the particular wire might not be suited for the purpose. The different fracture locations are presented in Figure 45 - Figure 47 with denotations corresponding to Table 5.

(47)

35

Figure 46. Fracture location “FL”, fusion line

Figure 47. Fracture location “Braze”, braze material

In some cases the fracture location is a combination of the categories fusion line “FL” and braze material “Braze”. In these cases there is a significant amount of braze material “spilling over” on the top sheet, helping to keep the sheets joined. The fracture goes partially through the braze material but is located in the fusion line, see Figure 48.

(48)

36

Figure 48. Combination of fracture locations “FL” and “Braze”

4.2.6 Mechanical properties Plasmatron

Tensile-shear tests were performed on Plasmatron brazed test specimens with the same dimensions as for CMT. Data was only obtained for one wire. Stress-strain curves for a test series of 10 specimens are shown in Figure 49.

Figure 49. Stress-strain curves for specimens brazed with wire Bercoweld B60. [19]

(49)

37

Table 6. Results ultimate strength Plasmatron

Test no Wire Nom. area Max force [kN] Nom. ultimate strength [MPa] 1 B60 38.4 13.10 341.1 2 B60 38.4 13.03 339.3 3 B60 38.4 12.76 332.3 4 B60 38.4 13.92 362.5 5 B60 38.4 13.05 339.8 6 B60 38.4 13.12 341.7 7 B60 38.4 13.97 363.8 8 B60 38.4 14.87 387.2 9 10 B60 B60 38.4 38.4 14.03 13.55 365.4 352.9

The minimum nominal strength of the series was = 332.3 MPa and the mean

nominal strength = 352.6 MPa. All of the tests met the requirement of 300 MPa.

4.2.7 Comments on mechanical properties

Since the forceTIG equipment was not available during the time of the master thesis work, there are only the results from the pre-study KEEX available. KEEX does not contain any results on mechanical strength why no comparison with forceTIG can be made at this point. The results from CMT and Plasmatron are not complete, but they prove that both processes have the potential of meeting the requirements and do not have to be excluded because of insufficient strength. For a discussion on strength calculations, see chapter 5.1.1 Mechanical strength.

4.2.8 Comments on economy and productivity

As stated prior to the project, the investment cost for arc brazing equipment is much lower than for laser brazing equipment, about one tenth (for forceTIG) according to ewm. The wire consumption of the TIG processes and laser process is approximately the same, as well as the gas consumption of the TIG processes. Laser brazing is often carried out without shielding gas [6]. Additional investments for the systems are robot equipment, seam tracking device and fixtures. All of the processes have potential of brazing with the same travel speed of at least 1.2 m/min. CMT is the most beneficial in investment cost and it requires the least accuracy. Nevertheless, the significantly higher wire consumption (5-6 times higher) cannot be omitted if larger series are to be produced. The productivity of the processes depends on how large series are being produced, and which wires can be used. The most expensive wire tested in this project is up to approximately 6-7 times more costly than the least expensive one, which implies that choices of process and filler material need to be made before any potential gain can be calculated accurately.

4.3 Optimisation analysis CMT

4.3.1 Empirical study

All of the tests were performed using the synergy mode, and analysing the graphs of voltage and current for different parameters shows that the most significant setting is the choice of synergy line. The curves presented in Figure 50 show the current during 200 ms of brazing with the same wire and same wire feed speed, but with different

(50)

38

synergy lines. In the upper graph, test F1, the current has a maximum value of around 230 A, compared to around 420 A in the lower graph, test F3.

Figure 50. Current for test F1 (upper) and F3 (lower). See Appendix A.

The voltage curves also present much higher maximum values for test F3 than test F1 (voltage and voltage/current curves are presented in Figure C 1 and Figure C 2 in

Appendix C. Extended optimisation analysis). The heat input

in kJ/mm, calculated with the mean values of voltage and current , and an efficiency factor of 0.8, is in this case similar for the two tests, 0.059 kJ/mm for F1 compared to 0.040 kJ/mm for F3, despite the higher maximum values. Travel speed is in this case 120 cm/min. The mean value of the heat input calculated instantaneously, however, shows a greater difference between the two tests, 0.095 kJ/mm for F1 and 0.050 kJ/mm for F3. This implies that current peaks and other features of the curves are of great importance when calculating arc energy and heat input. Another example of the importance of synergy settings is shown in Figure 51. The upper graph shows current curves over 100 ms for two tests, A92 and A93, plotted in the same graph and the lower graph shows voltage curves for the same tests.

(51)

39 The differences between A92 and A93 are both synergy settings and wire feed speed. The wire feed speed of A92 is 8 m/min compared to 5.5 m/min for A93. The heat input calculated with mean values are 0.061 kJ/mm for A92 and 0.038 kJ/mm for A93, while the mean values of the instantaneous heat input are 0.084 kJ/mm for A92 and 0.073 kJ/mm for A93. Looking at the brazed seams they show little visual difference, besides some spatter on A92, although the settings differ greatly. Figure 52 shows the seams of tests A92 and A93.

Figure 52. A92 (upper) and A93 (lower), front (left) and back (right)

These examples show some of the features to consider when optimising the process. Other analysis possibilities are frequency measurements; any deviations from the normal frequency or amplitude can be found, and if implemented with a measuring system logging the robot coordinates, defects and discontinuities can be found and the behaviour of the process at that exact time and position can be analysed. However, as shown in Figure C 3, Figure C 4, Figure C 5 and Figure C 6 in Appendix C. Extended

optimisation analysis, the synergy setting gives the curve its characteristics and other

parameters such as dynamic and arc length correction make rather little difference on the shape of the curve. In order to fully be able to optimise the process for any given wire and gas more tests are required so that a clear connection between the parameters and the results can be made, and a custom made synergy line can be created. More curves are shown in Appendix C. Extended optimisation analysis.

4.3.2 Characteristic parameters

Each synergy line has a characteristic that can be broken down into a number of parameters as shown in Figure 53. If not using the pre-set synergy settings, at least two but preferably between 6 and 10, and maximum 19 of these parameters have to be set manually. Beyond the 9 parameters shown in the figure there are also parameters for pulse and reversed polarity. These are features available in the CMT Advanced but they were not used in this project.

(52)

40

Figure 53. Characteristic parameters for CMT welding/brazing. [20]

The parameters are explained in the following list. (1) Set-up value for current before short circuiting

(2) Set-up value for wire feed speed before short circuiting (3) Set-up value for current at short circuiting

(4) Slope up for current at arc ignition (5) Slope up change before “Boost” current (6) Set-up value for current at “Boost” (7) Time in “Boost” phase

(8) Slope down from “Boost” (9) Slope down change

With this information in addition to the curves obtained with the measuring system, it is fully possible, however a complex task, to create synergy lines and optimise the process.

(53)

41

5 DISCUSSION AND CONCLUSIONS

In this chapter the frame of reference, procedure and results are discussed and conclusions are listed.

5.1 Discussion

5.1.1 Mechanical strength

Determining the ultimate strength of the joint is a question of definition. The main problem is to know what area to use for the calculations. The position of the fracture makes a big difference in what the actual strength is and what the fracture mechanism is. Using the nominal cross section area of the braze is an option, for example where the bead has its maximum height. To accurately measure the true strength each specimen has to be thoroughly examined after the mechanical testing. In effect, the most important feature to verify is that the joint can take a predetermined load without fracturing, why it was decided that the thinner sheet thickness was to be used for calculating the strength, as a comparable value.

All mechanically tested CMT brazed joints except four met the requirement of 300 MPa and in most cases the fracture location was the base material. Wire D, UTP A320, did not meet the requirement in any of the tests. Whether this is due to the filler material properties or the geometry of the joint is not easy to say with full certainty, but it implies that the wire is too weak for the purpose.

5.1.2 Calibration

The arc characteristics depend on arc current and –voltage. The power source, however, uses the output values and compensates for losses in cables etc. by a calibration procedure. The electrical resistance depends on cable lengths but also on torch position which varies rather much due to the relatively long sheets to be brazed. The procedure for resistance and inductance compensation was done in the middle of the sheets, lengthwise, since the resistance and inductance differ from one end to the other. The difference was quite small and the middle value was regarded as the most correct.

5.1.3 Wrong synergy settings

Since there were no synergy settings available for three of the wires and the settings for another one did not suit the application it was difficult achieving satisfying results. Even if a setting for another wire would give satisfying results it is hardly scientifically correct to suggest a procedure that works based on luck and trial and error. The right software seems to be required in order to achieve good results since the technology is “hidden”, meaning that without synergy lines it can get very complicated to set the parameters. It is difficult to say how synergy lines, optimised for the wires that did not have any, would affect the results. It would probably be possible to get better results, since good results are the base on which the optimised settings are created. It greatly depends on the wire and for which application the synergy line is created.

The synergy settings used are all made for brazing with pure argon as shielding gas. In the trials the shielding gases were MISON 2 and MISON Ar, both containing mostly argon

(54)

42

but with small additives. Pure argon was not tested - if this makes a significant difference is unknown.

5.1.4 Implementation of new technique

The implementation of a new technique can cause extra costs for the user – in the automotive industry it is probably easier to implement a MIG/MAG variant than a TIG because of the already widely spread use of MIG/MAG in the industry. This has to be taken into consideration when finally evaluating the processes – e.g. if it brings extra costs for education and other investments.

5.1.5 Distortion and restraints

The distortion that can be measured when brazing the full length of ~1200 mm can be used in comparing between different settings, but it is largely dependent on the fixture. In this case the sheets are clamped on one side while the other one is free. The size of the sheets is also a factor, the sheets for testing are significantly smaller than the real part would be, and the larger size of the sheets would help counteract distortions. However, with more clamping more residual stress would remain in the joint. This has not been regarded in the thesis work since there are no requirements stated. Comparing the levels of distortion it seems the two TIG processes produce less distorted sheets, due to smaller heat affected areas and less filler material that shrinks when solidifying.

5.1.6 Instability sources

Although CMT has good bridging ability and therefore requires relatively low precision in joint set-up compared to laser brazing there are still demands on positioning and repeatability. Some problems arose in the beginning due to the fixture. The fixture is relatively long, fastened in the short end, and made mainly of steel making it heavy and therefore inclined to deflect. This deflection causes the joint to become poorly positioned and the braze bead to end up outside the wanted joint. One solution would have been to build a more centered fixture, “balanced” on the positioner instead of fastened at the short end. The problem was corrected in this case by welding on extra stabilising parts in the attachment, but there were still some minor positioning problems occasionally.

Besides the fixture there are other noise sources possibly generating discontinuities or defects. A new ventilation system caused large amounts of soot and zinc burn off on the base metal, since it was more powerful than the old ventilation system and thus caused the shielding gas flow to be disrupted. The problem, when realised, was easily fixed by increasing the distance between ventilation and torch.

The camera when mounted added an extra 0.5 kg, approximately, some distance from the torch, generating an oscillating motion of the torch making the bead uneven. This occurred only for some positions of the camera.

Since the wire is on a bobbin it comes out of the torch slightly bent which affects the positioning of the seam. The wire is aimed at the correct position, but since the wire hose package moves with the robot arm, there is a risk of the wire coming out bent in another direction at certain positions, which adds to the problems with positioning. A

(55)

43 solution is to keep the hose as still as possible, possibly by centering the fixture so that it is closer to the robot base.

Another aspect is the positioning accuracy of the robot. All of these small variations in positioning might add up to a large positioning error and need to be taken into consideration.

5.1.7 Quality of the CMT brazed joints

The wires containing Sn (wires D, E and F) all resulted in a lower, wider and smoother bead. The wire containing Si (wire A) got a high bead in all cases and the two Al-containing wires (wires B and C) showed results between the two extremes. The amount of spatter depends to a large extent on the synergy line used.

The quality of the joints brazed with CMT is not yet as visually high-quality as laser brazed joints. The work has shown, however, that there is room for improvements and with the measuring system available it is most likely possible to obtain even better results.

References

Related documents

46 Konkreta exempel skulle kunna vara främjandeinsatser för affärsänglar/affärsängelnätverk, skapa arenor där aktörer från utbuds- och efterfrågesidan kan mötas eller

Uppgifter för detta centrum bör vara att (i) sprida kunskap om hur utvinning av metaller och mineral påverkar hållbarhetsmål, (ii) att engagera sig i internationella initiativ som

In the latter case, these are firms that exhibit relatively low productivity before the acquisition, but where restructuring and organizational changes are assumed to lead

The increasing availability of data and attention to services has increased the understanding of the contribution of services to innovation and productivity in

Närmare 90 procent av de statliga medlen (intäkter och utgifter) för näringslivets klimatomställning går till generella styrmedel, det vill säga styrmedel som påverkar

• Utbildningsnivåerna i Sveriges FA-regioner varierar kraftigt. I Stockholm har 46 procent av de sysselsatta eftergymnasial utbildning, medan samma andel i Dorotea endast

Den förbättrade tillgängligheten berör framför allt boende i områden med en mycket hög eller hög tillgänglighet till tätorter, men även antalet personer med längre än

På många små orter i gles- och landsbygder, där varken några nya apotek eller försälj- ningsställen för receptfria läkemedel har tillkommit, är nätet av