An Exploratory Study Into the Feasibility of Magnetic Pulse Welding

An exploratory study into the feasibility of magnetic pulse welding

Kevin Loncke

Promotor: prof. dr. ir. Wim De Waele Begeleiders: Koen Faes (BIL)

Masterproef ingediend tot het behalen van de academische graad van Master in de ingenieurswetenschappen: bouwkunde

Vakgroep Mechanische constructie en productie Voorzitter: prof. dr. ir. Patrick De Baets Faculteit Ingenieurswetenschappen Academiejaar 2008-2009 De auteur en promotor geven de toelating deze scriptie voor consultatie beschikbaar te stellen en delen ervan te kopi¨eren voor persoonlijk gebruik. Elk ander gebruik valt onder de beperkingen van het auteursrecht, in het bijzonder met betrekking tot de verplichting uitdrukkelijk de bron te vermelden bij het aanhalen van resultaten uit deze scriptie.

The author and promoter give the permission to use this thesis for consultation and to copy parts of it for personal use. Every other use is subject to the copyright laws, more speciﬁcally the source must be extensively speciﬁed when using from this thesis.

Gent, Juni 2009

De promotor De begeleider De auteur

Prof. dr. ir. W. De Waele ir. K. Faes Kevin Loncke Acknowledgments

With the finishing of this master thesis, my time at Ghent University has almost come to an end. I am truly grateful to a large number of people who have helped me throughout my study and the fulfilling of this master thesis. First and foremost, I would like to express my appreciation to my promotor Prof. Dr. Ir. Wim De Waele, and my mentor Ir. Koen Faes. They were always prepared to give advice and feedback during the experimental research. I very much appreciate the time they took to read and re-read the numerous drafts of this thesis. Most importantly, I would like to thank them for the understanding of my sometimes over-booked schedule, and the freedom they gave me to plan the experiments as it suited my agenda best. I would also like to thank the people of the Belgian Welding Institute that helped me find my way around in the laboratory: Michel De Waele for explaining me the tools I needed to use; Anja Buyse and Gert Oost for taking their time to etch the samples; the people of the technical staff to make the workpieces; and all the others that were prepared to answer my countless questions throughout the semester. During the hundreds of hours that I spent behind this desk while writing this work and pondering about all possible aspects of magnetic pulse welding, there was one special person always ready for me. I would like to thank my girlfriend and future wife Jessica for encouraging me during the times when I felt the deadline was coming too close, for keeping me company on skype during my breaks, and for showing her sincere interest in the work I have done. I would also like to thank my fellow students of the VTK and IAESTE for sharing all those fun moments the past academic year, and more importantly making me realize that I was far from the only student combining a thesis with working at the student union. Last but not least, I would like to thank my parents for offering me the opportunity to complete these studies, for encouraging me during the hard first years of my studies, for appreciating my efforts done throughout this time, for understanding when I was cranky due to too big work - time ratios, and for buying me a coffee machine for my student home.

Kevin Loncke Ghent, 1 June 2009

iii iv An exploratory study into the feasibility of magnetic pulse welding

Kevin Loncke

Master thesis presented in fulﬁllment of the requirements for the degree of Master of Civil Engineering

Academic year 2008–2009

Promotor: Prof. Dr. Ir. Wim De Waele Mentor: Ir. Koen Faes (BIL) Faculty of engineering Ghent University

Department of Mechanical construction and production Chairman: Prof. Dr. Ir. Patrick De Baets

Summary

In this work a study was done of the magnetic pulse welding process. The influence of the stand-off distance, thickness of the flyer tube, position of the field shaper, the shape of the inner workpiece and the material were discussed, based on the results of a series of experiments. A series of formulas was set up to help understand the influence of the parameters. A literature study on the explosive welding process was used to get a better understanding of the bonding and the deformation behaviour.

Keywords magnetic pulse welding, principle, process parameters, weld characteristics An exploratory study into the feasibility of magnetic pulse welding Kevin Loncke Supervisor(s): Wim De Waele, Koen Faes

Abstract— This article displays the principle of the MPW process. The operations required, and the high quality of the weld. Limita- influence of the process parameters and the deformation behaviour is dis- tions of the process lay in the geometries that are possible to cussed. These are based on a series of experiments, a set of formulas and a comparison with explosive welding. combine, while the sensitivity to process parameters and the in- Keywords—magnetic pulse welding, principle, process parameters, weld accessibility of the welding zone are disadvantages. The process characteristics is very similar to the explosive welding process, apart from the energy source that is used to move the flyer material. I.INTRODUCTION III.EXPLOSIVEWELDING OINING of dissimilar materials offers great opportunities Jfor e.g. the automotive industry and heat-exchanging sys- A. Welding windows tems. Solid state welding processes like magnetic pulse welding Except for the source of energy, explosive welding (EXW) (MPW) offer this possibility. But because research in MPW is is very similar to MPW. The EXW theory learns that the weld fairly new, the knowledge about and experience with the process holds the best strength properties when a wavy interface is cre- is concise. In order to use the process in industrial applications ated without an intermediate layer [2]. The quality of the weld on a wide scale, a good understanding of the bonding and the depends on the impact velocity v and impact angle α. Graphs influence of the process parameters is required. c show welding windows with the required range of these two pa- II.PRINCIPLE rameters. The angle α should be within a certain range, and vc should be in the subsonic range in order to allow jet formation. The magnetic pulse welding process starts by charging a bank The impact velocity vc should also be high enough to create a of capacitors upto a desired energy level. This energy level is wavy interface. The welding windows can serve as a tool to characterized by the voltage level in the capacitors. When the compare the required parameters of different material combina- capacitors are charged, the current is instantaneously released tions. through a coil (figure 1[1]), this way inducing a magnetic field. A field shaper is placed inside the coil to concentrate the mag- B. Wavy interface netic field in the desired area. The field will be blocked by the flyer tube, depending on the skin depth of the material. Through The waves at the interface show an increase in the wavelength simple calculations, it was shown that for the experiments in this and amplitude for an increasing vc. For an increasing α the work, more than 80% of the field is blocked. The difference in amplitude will first increase and immediately decrease, while magnetic field causes a radial pressure on the flyer tube, due to the wavelength will only show an increase. These relations are which it will impact to the inner workpiece. used to guess the change in vc and α in the MPW experiments.

IV. SETOFFORMULAS The voltage level V could be calculated starting from the required impact velocity vc with a set of equations. The chosen value of the impact velocity gives the total pressure that is required to deform and accelerate the flyer tube. This pressure leads to a value of the magnetic field, taken into account the diffusion of the magnetic field through the tube. From the magnetic field density, the required current peak in the coil is calculated. This leads to the required voltage level, by using the equipment’s Fig. 1. Magnetic Pulse system layout. characteristics. The formulas were used to get an idea of the impact veloci- So far the research has been limited to tubular workpieces and ties that were achieved in the experiments. Due to the simplifi- few experiments with sheets. A great variety of material combi- cations, the pressure required for the deformation of the work- nations has already been successfully welded which illustrates pieces is underestimated. The formulas are therefore not accu- the main advantage of the process: the possibility to combine rate for workpieces with higher thicknesses t and higher stand- dissimilar materials. Other advantages are a reduction of the off distances s, because these require more deformation energy. manufacturing costs because there are no pre or post welding Moreover, the effect of the field shaper overlap is not taken into account. The formulas are therefore not accurate, but give a rea- residual stresses in the flyer tube. These porous welded areas sonable estimation of the impact velocity and show the influence should be avoided, as they clearly degrade the quality of the of the parameters for experiments with lower deformation. weld.

V. EXPERIMENTAL PROCEDURE D. Influence of the parameters The experiments showed that a A. Workpiece layout • Shape of the field shaper: straight surface of the inner workpiece, as shown in figure 2, The experiments were done using the general layout shown in should be preferred over a slant surface. figure 2. The varying parameters were the materials, stand-off The collar on the inner workpiece only influences the deforma- distance s [mm], thickness of the flyer tube t [mm], shape of the tion behaviour of the flyer tube if l exceeds a certain value. inner workpiece, overlap of the field shaper l [mm] and volt- F.S. F.S. Stand-off distance s: It was seen that there is an optimal age level V [kV]. The workpieces were cross-sectioned after • value for s. Based on the longer wavelengths, the aluminum welding and judged by microscopic examination. The different tubes with t = 1, 5 mm had an optimal s = 3, 0 mm. At higher welds were compared by their welding length. The experiments stand-off the stresses due to the high deformation will decrease were done with a Pulsar model 50 25. | the impact velocity. t A change in thickness of the flyer • Thickness flyer tube : tube of 1, 5 mm to 2, 0 mm showed similar weld lengths. The lower wavelengths show that the thicker workpieces impact at a lower vc. The welding windows in EXW confirm that these workpieces can impact at a lower vc, due to the higher weight. l l was found to be an impor- • Position field shaper F.S.: F.S. tant parameter, regardless the value of the other parameters, as it influences both the impact velocity and impact angle. It also influences the location of the impact point, which has a direct influence on the length of the weld. For different workpieces, a different range of possible positions of the field shaper exists. The position is best chosen in the middle of the possible range, which is for aluminum tubes with t = 1, 5 mm an overlap of Fig. 2. The positioning of the workpieces inside the field shaper. 11 mm. An increase of the voltage level increases v . • Voltage level: c For the welding of Al - Al, a decrease of the voltage level to B. Deformation behaviour values lower than 15 kV led to a much smaller range of possible The experimental results and the comparison with the wave parameters. formation in EXW led to an assumption on the impact behaviour Apart from the Al - Al experiments that led to the • Materials: of the flyer workpiece. The flyer tube will start moving as soon conclusions about the other parameters, also Cu - Al and Al - as the yield stress is reached at some point along the tube. A steel were succesfully welded. The formulas show that Cu - Al hinge will be formed and the flyer tube will move until it impacts was welded with a lower vc than Al - Al at the same voltage at its end. The impact point will progress along the surface, level, because of the higher weight of copper. The Al - steel while the hinge at the other end of the working zone moves down combination required an increase of the voltage level, and hence towards the inner workpiece. This causes an increasing α along was welded at higher vc than Al - Al. This corresponds with the the weld. As the surfaces were not cleaned prior to welding, the welding windows in EXW. The Cu - Al samples didn’t show the jet formation is sufficient to remove the surface contaminations. expected results, it is therefore possible that the residual stresses The radial compression results in large longitudinal and cir- in the copper tube resulted in failure of the weld after cross- cumferential stresses in the tube, which cause an increase in sectioning. thickness and buckling along the circumference. Consequently to this buckling effect, the weld won’t be equal all around the VI.CONCLUSION circumference, as the buckled parts lag behind during the move- The results from the experiments combined with the know- ment. ledge of the EXW interface characteristics showed how the flyer tube is deformed. This led to a better understanding of the in- C. Weld characteristics fluence of the process parameters. A series of formulas was Most of the welds had a wavy interface with wavelengths in presented, that can be used to relate voltage level V and impact the range 100 µm - 500 µm, and patterns similar as in EXW. velocity vc when the deformation of the flyer tube is limited. In some experiments, dark porous areas were seen at the interface. This indicate melting and subsequent solidification of a REFERENCES thin layer of the material. This melting was reported before, as [1] V. Shribman. Magnetic pulse technology for improved tube joining and forming. Tube & Pipe Technology, pages 91–95, November/December it forms brittle intermetallic compounds that are susceptible to 2006. cracking. In the experiments, the porosity of this layer was in [2] P.V. Vaidyanathan and AR. Ramanathan. Design for quality explosive weld- some cases so high, that cracking occurred, possibly due to the ing. Journal of Materials Processing Technology, 32:439–448, 1992. An exploratory study into the feasibility of magnetic pulse welding - Nederlandse samenvatting

Inleiding

De las tussen twee werkstukken is dikwijls de meest kwetsbare plaats van een onderdeel van een machine of installatie, en des te meer wanneer het over een combinatie van verschillende materialen gaat. De zoektocht naar betrouwbare manieren om verschillende materialen te combineren is daarom eindeloos.

Wanneer warmte gebruikt wordt om verschillende materialen te lassen, zorgt het verschil in thermische eigenschappen voor grote spanningsgradi¨enten. De brosse tussenlaag die dan gevormd wordt in de las tussen deze materialen of hun legeringen zal dan makkelijk scheuren.

De beste manier om deze brosse laag en de spanningsgradi¨enten te vermijden is door de materialen te lassen in vaste toestand. De materialen worden verbonden bij temperaturen beneden hun smeltpunt, door de warmte die normaal gebruikt wordt deels te vervangen door een drukkracht. Magnetisch puls lassen (MPL), dat zeer gelijkaardig is aan het explosief lassen (EXL), is hiervan een veelbelovend voorbeeld. De materialen worden hierbij verbonden door een grote impact.

De magnetische puls techniek wordt al sinds jaren gebruikt voor het vormen van materialen, maar recentelijk is er steeds meer interesse naar het gebruiken van de techniek om materialen te verbinden. Het verschil met het EXL is de manier waarop de druk wordt verkregen die de stukken impacteert. Bij EXL wordt deze geleverd door een explosie, bij MPL door een magnetische druk die wordt gegenereerd door een magnetisch veld in een spoel.

De techniek is uitermate geschikt voor het verbinden van buizen. Door de vele mogelijkheden van materiaalcombinaties die kunnen verbonden worden opent die mogelijkheden voor vele verschillende toepassingen: brandstoﬂeidingen, air conditioning systemen, koelkasten,

vii viii condensatoren,... Ook vanuit de auto-industrie is er een grote intersse om meer aluminium te gebruiken om zo het gewicht van de voertuigen te reduceren. Een eenvoudige verbinding tussen roestvast staal en aluminum zou dan ook de mogelijkheden enorm verruimen.

De kwaliteit van de las zal, net zoals in explosief lassen, afhangen van de impact hoek en impact snelheid waarmee de werkstukken impacteren. De vele procesparameters zullen deze twee parameters be¨ınvloeden. Deze parameters omvatten materiaalkarakteristieken, eigenschappen van de pulslasmachine, en de geometrie van de werkstukken. Het probleem met MPL is dat tot op heden de invloed van deze parameters niet voldoende gekend is, en er dan ook geen praktische aanbevelingen kunnen gegeven worden voor het verbinden van twee werkstukken.

Deze thesis kadert in het Magpuls project, dat een samenwerking is tussen de Belgisch Insti- tuut voor Lastechniek (BIL). Het doel is om een brede basis van kennis te voorzien inzake het MPL om het verder onderzoek te vereenvoudigen. Een goed begrip van de precieze werking van het proces en de invloed van de procesparameters staan dan ook voorop. Dit zal trach- ten bereikt te worden met een literatuurstudie over het MPL, een literaturstudie over het bindingsproces in EXL, en een reeks experimenten.

Literatuurstudie

Principe

In MPL zorgt een elekromagnetische kracht voor het impacteren van de twee werkstukken. Een magnetisch puls machine (figuur 1) bestaat typisch uit een reeks capaciteiten, een energievoorziening, een spoel, en een ontladingscircuit. De hoog voltage energievoorziening wordt opgeladen via het stroomnetwerk en bevoorraadt de capaciteiten. Wanneer deze het gewenste energieniveau bereiken wordt de stroom ontladen door de spoel, deze genereert vervolgens een magnetisch veld. Aangezien de experimenten in deze thesis met buizen worden uitgevoerd, wordt een soleno¨ıdemet meerdere windingen gebruikt. Een field shaper wordt binnenin de spoel geplaatst om het magnetisch veld te concentreren in de gewenste zone (figuur 2). Het magnetisch veld genereert wervelstromen in het uitwendige werkstuk, dat binnenin de spoel wordt geplaatst. Deze blokkeert het magnetisch veld grotendeels door reflectie, absorptie en secundaire reflectie, waardoor een verschil in magnetisch veld ontstaat tussen de binnen- en buitenzijde van de buitenste buis (figuur 3). Een korte berekening leert dat het veld hoofdzakelijk bepaald wordt door absorptie. Het afschermend effect kan vergroot worden door de frequentie van de stroom te verkleinen, of door dikkere werkstukken of materialen met een hogere conductiviteit te gebruiken. Het verschil in magnetisch veld zorgt voor een radiale magnetisch druk die het buitenste buisje vervormt en impacteert tegen het binnenste werkstuk. ix

Figure 1: Layout van een magnetisch puls machine [5].

Figure 2: De stroom die in de ﬁeld shaper ge¨ınduceerdwordt door de spoel [13].

De impact zorgt voor een verbinding tussen de werkstukken. Smelten, adhesie, en het vormen van een golfvormig oppervlak zijn de drie redenen die vermoedelijk het meest bijdragen in het binden van de materialen. Een transitie-zone wordt gevormd tussen de materialen door de grote plastische vervorming en/of smelten. Deze vertoont een golfvormig oppervlak (enkele typische voorbeelden zijn gegeven in figuur 4), dat zal verschillen afhankelijk van de materiaalcombinatie die gebruikt wordt. Dit smelten veroorzaakt intermetallische componen- ten in de bijgevolg brosse tussenlaag, waardoor de hardheid sterk gaat toenemen. Ook de hardheid van de werkstukken zal toenemen door de grote plastische vervorming. Het smelten wordt veroorzaakt door het Joule effect, de impact en het vormen van een jet langsheen de oppervlakken. Deze jet is een laagje materiaal dat vooruitgestuwd wordt en de oppervlakken voorbereid op de binding door het verwijderen van vuil en oxides, en het effenen van onregelmatigheden. x

Figure 3: Het magnetisch veld binnen en buiten het werkstuk.

Magnetisch puls las machine

De machine die in de experimenten van deze thesis werd gebruikt is een Pulsar model 50 25. | Deze kan 50 kJ aan energie opslaan, wat overeenkomt met een maximum voltage van 25 kV. Stroompieken van 500 kA kunnen bereikt worden met een frequentie van 14 kHz.

Eisen voor de werkstukken

Het materiaal van het binnenste werkstuk kan eender wat zijn, zolang het niet te veel gaat vervormen bij impact. Dit kan altijd vermeden worden door een cilinder te plaatsen binnenin het buisje. De buitenste buis wordt best vervaardigd uit een zo goed mogelijk geleidend materiaal. Voorbeelden werden teruggevonden van aluminium en wolfraam, aluminium en inconel, aluminium en koper, aluminium en titanium, aluminium en ijzer, aluminium en magnesium, aluminium en staal, aluminium en roestvast staal, aluminium en brons, en aluminium en molybdenum. Buisvormige werkstukken worden verkozen voor een optimale sterkte van de verbinding, maar ook plaatmeriaal kan succesvol gelast worden met de MPL techniek. De diameter van de mogelijke werkstukken is vooralsnog beperkt, omdat geen machines bestaan die een dergelijk hoog energieniveau kunnen halen. Een overzicht van enkele uitgevoerde experimenten met de gekozen procesparameters is terug xi

(a) Aluminium - Aluminium (b) Aluminium - Koper

Figure 4: Golfvormige las in experimenten bij MPL [2] te vinden op pagina 18.

Voordelen

De voordelen van het proces zijn in hoofdzaak de mogelijkheid om nieuwe materialen te verbinden, het reduceren van de productiekosten aangezien geen andere bewerkingen nodig zijn voor en na het lassen, de snelheid van het proces, en de hoge kwaliteit van de lassen door de snelle vervorming. Bovendien is het een ecologisch gunstig proces aangezien geen gas, straling of rook geproduceerd wordt.

Nadelen en beperkingen

De nadelen liggen vooral in de geometrie van de werkstukken. Enkel overlappende verbindin- gen kunnen gerealiseerd worden, en voorlopig is dit enkel mogelijk voor buizen en platen. Het materiaal van het bewegende werkstuk moet bovendien voldoende geleidend zijn. Het proces is enkel geschikt voor fabriekstoepassingen en de kwaliteit van de las is gevoelig aan de invloed van de procesparameters.

Vergelijking met explosief lassen

EXL is zeer gelijkaardig aan MPL, behalve dan de energie die de druk genereert. De impact van de werkstukken verloopt gelijkaardig, waardoor het belang van de jet duidelijk wordt. De impactsnelheid en impacthoek blijken de twee cruciale parameters te zijn voor het slagen van de las. De gelijkenissen kunnen gebruikt worden om de invloed van de parameters bij MPL te linken aan een impacthoek en impactsnelheid.

Explosief lassen

EXL wordt reeds sinds enkele decennia op grote schaal toegepast. Dit heeft geleid tot een reeks lasvensters van verschillende parameters zoals impactsnelheid vs. impacthoek en impactdruk xii vs. impacthoek. Het bindingsmechanisme wordt best geassocieerd met een hydrodynamisch stromingsmodel.

Het bindingsoppervlak komt in verschillende vormen voor: vlakke oppervlakken, kleine golven, grote golven, golven met wervels, golven met gestolde concentraties smelt, en continue gesmolten oppervlakken. Een golvend oppervlak zonder smeltzones toont de beste sterktekarakteristieken. Het vermijden van deze gesmolten tussenlaag is daarbij belangrijker dan het vormen van de golven.

De hoge drukken aan de zone van impact planten zich voort in het materiaal aan de snelheid van het geluid. Omdat het materiaal zelf trager voortbeweegt, worden er zeer grote drukken gecre¨eerdaan de oppervlakken juist voor de impactzone, die voldoende zijn om een laagje van het materiaal af te brokkelen en weg te stuwen. Deze jet zorgt ervoor dat vuil van het materiaal verwijderd wordt.

Een theoretisch voorbeeld van een lasvenster is gegeven in figuur 6. Deze is gegeven met de impactsnelheid vc op de abscis en de impacthoek α op de ordinaat (te vinden op figuur 5). De grenzen 3 en 4 voor de impacthoek verwijzen naar de hoek die vereist is voor het vormen van een jet. Grenzen 1 en 6 zijn gebaseerd op het vormen van een golvend oppervlak uitgaande van respectievelijk elastisch-plastisch gedrag en gedrag bij vloeien. De maximale snelheid van grens 2 is de snelheid van het geluid in het materiaal, aangezien de impactsnelheid hieronder moet blijven om een jet te creëren. Grens 7 werd experimenteel vastgelegd. Voor grens 5 werd er in de literatuur geen verklaring gevonden.

Figure 5: Het vormen van een jet tijdens de impact, met 1) het bewegende werkstuk, 2) het vaste werkstuk en 3) het jet materiaal. xiii

Figure 6: Een theoretisch lasvenster in EXL [26].

Een hogere plastische rek en schuifspanningen veroorzaken de vorming van een golfoppervlak. Dit oppervlak zal er anders uitzien met wijzigende impactsnelheid en impacthoek. Voor een stijgende vc zal ook de amplitude en de golﬂengte van de golven mee stijgen tot op het punt waar de supersonische snelheden bereikt worden. De impacthoek α varieert tijdens het proces en zal stijgen naarmate de las vordert. Bij stijgende α zal ook de golﬂengte van de golven stijgen, de amplitude zal aanvankelijk stijgen tot een piekwaarde om daarna weer af te nemen.

De golven zijn ook groter bij lagere sterktekarakteristieken van het materiaal, bij grotere stand-oﬀ afstand tussen de werkstukken, en bij een grotere dikte van het te impacteren werkstuk.

De conclusies over het uitzicht van de las kunnen verder aangewend worden om waarden van vc en α te vergelijken in verschillende MPL experimenten.

Een diepere blik op het MPL proces

In dit deel werd een serie formulas opgesteld die de vereiste impactsnelheid linken aan het voltage niveau dat kan ingesteld worden in de magnetisch puls machine.

1. De gekozen waarde van de impactsnelheid levert de druk die vereist is om het materiaal xiv

te vervormen (P1), en te versnellen richting het binnenste werkstuk (P2). Er werd hierbij een factor 2 toegevoegd aan de druk voor de versnelling omdat deze druk slechts een piekwaarde is en de versnelling constant aangenomen wordt. · · 2 2 σy t vc P = P1 + P2 = α · + 2 · t · ρ · (1) R 2 · s

T q 2 · s 2. Er moet voldaan worden aan de voorwarde tacc = 2 · tdeform tflyer = , 2 − ≥ a 2 2 · π · · waarbij tdeform kan gevonden worden uit P1 = sin ( T tdeform) Ptot.

3. De vereiste druk levert de nodige sterkte van het magnetisch veld: 2 · t B2 − P = ( 0 )(1 e δ ) (2) 2µ0 −

4. De waarde van het magnetisch veld leidt vervolgens tot de nodige waarde van de stroompiek in de spoel:

I = iwz + 2 · iend (3a) φ · l s iwz (1 ) + 2 · ∆i (3b) ≈ 2 · π · µ0 · R · s − 2 · R φ ∆i = 2 (1 ln(2)) (3c) π · µ0 · R − φ π · R iend 2 ln( ) C0 + Ω1 (3d) ≈ π · µ0 · R 2 · s −

5. Uit de waarde van de stroompiek kan het vereiste voltageniveau bepaald worden op basis van de eigenschappen van de lasmachine: I Uset = (4) 20

Deze formules werden verder gebruikt om een idee te krijgen over de impactsnelheden die werden bereikt in de experimenten. Omgekeerd kunnen ze ook aangewend worden om via waarden van de impactsnelheid die voorgesteld worden in het explosieﬂassen een waarde te bekomen van het vereiste energieniveau.

Procesparameters

De verschillende procesparameters zullen allen een invloed hebben op de impactsnelheid en de impacthoek. De impactsnelheid bepaalt de energie die vrijkomt bij impact, terwijl de impacthoek cruciaal is voor het vormen van de jet. xv

Variabele parameters

Materiaaleigenschappen Verschillende materialen zullen gebruikt worden in de experimenten, en daarom is het eﬀect van de materiaaleigenschappen belangrijk voor het vormen van de las. De magnetische permeabiliteit µ en de elektrische conductiviteit σ hebben beide een invloed op de skin depth δ. Voor stijgende waarden van µ en σ zal δ toenemen. δ drukt de diepte uit waarop de wervelstromen terugvallen naar 1/e van hun waarde aan het oppervlak. Hoe kleiner deze waarde relatief t.o.v. de dikte van de buitenste buis, hoe groter het deel van het magnetisch veld dat geblokkeerd wordt en hoe groter de magnetische druk. Aangezien enkel met goed geleidende materialen als aluminium en koper wordt gewerkt, en de gebruikte stukken dikker zijn dan 1, 5 mm blijkt dit van weinig invloed te zijn.

De sterkteparameters bepalen het gemak waarmee de buis vervormd wordt. De vloeigrens

σy is hierbij de belangrijkste parameter. De elasticiteitsmodulus E kan een rol spelen bij de residuele spanningen ten gevolge van het verkleinen van de buisdiameter. Deze zullen groter zijn bij koper dan bij aluminium, aangezien koper een grotere elasticiteitsmodulus heeft.

De dichtheid ρ van het materiaal is van belang aangezien zwaardere materialen een grotere druk vereisen om te versnellen richting het binnenste werkstuk. Maar, ze zullen ook een grotere impact energie hebben door deze grotere dichtheid, wat het voorgaande nadeel opheft.

De thermische conductiviteit zal niet direct een invloed hebben op het proces, maar eerder op de las zelf. Hoe groter de conductiviteit, hoe groter de warmte die gecre¨eerdwordt door de jet en door de impact kan weggeleid worden. Voor lage waarden, zal de temperatuur aan het oppervlak meer toenemen en bestaat er een grotere kans op smelten.

Geometrische eigenschappen De geometrische parameters hangen af van de uitvoering van de experimenten. Een typische layout is getoond op ﬁguur 7 met aanduiding van de verschillende parameters.

De stand-oﬀ s is de ruimte tussen de twee werkstukken. Deze moet groot genoeg zijn opdat het buitenste werkstuk voldoende kan versnellen, maar mag niet te groot zijn opdat het niet opnieuw zou gaan vertragen door de grote vervormingen.

De dikte van het buitenste werkstuk t maakt het vervormen van het materiaal moeilijker, maar vergroot wel de impact energie.

Er werd gekozen om de ﬁeld shaper over de rand van het buitenste buis te plaatsen. De positionering wordt dan gekarakteriseerd door zijn overlap met deze buis.

Drie mogelijke vormen voor het binnenste werkstuk werden geprobeerd, zoals te zien is op ﬁguur 8. De kraag werd hierbij voorzien om de werkstukken concentrisch te houden, zodat xvi

Figure 7: Positionering van de werkstukken onder de ﬁeld shaper.

de impact rondom dezelfde is. Er werd voor volle stukken gekozen om vervormingen van het binnenstuk en de bijhorende be¨ınvloeding van het proces te beperken.

Figure 8: De verschillende vormen voor het binnenste werkstuk die gebruikt zijn in deze thesis: a)schuine configuratie, b)rechte configuratie, c)configuratie zonder kraag field shaper.

Voltage Deze drukt het energieniveau uit dat ingesteld kan worden. Hoe groter het energieniveau, hoe groter de magnetische druk en uiteindelijk de impactsnelheid. xvii

Constante parameters

Frequentie van de stroom Deze heeft net zoals de magnetische permeabiliteit en de elektrische conductiviteit een invloed op de skin depth en zal dus van minder belang zijn. De frequentie van het toestel dat hier gebruikt wordt is 14.000 Hz.

Lengte van de field shaper Hoe korter de field shaper, hoe meer het magnetisch veld zal geconcentreerd worden. Kortere field shapers kunnen dus grotere drukken leveren voor eenzelfde energieniveau. De lengte van de field shaper in deze thesis is 15 mm.

Diameter van de buitenste buis Aangezien de buitenste buis best zo dicht mogelijk geplaatst wordt tegen de ﬁeld shaper, hangt zijn diameter af van het ontwerp van deze ﬁeld shaper. In deze thesis ligt de diameter op 25 mm.

Experimentele resultaten

In het onderzoek bij deze thesis werden 76 experimenten uitgevoerd, waarvan er 31 bleken gelast te zijn. Na het lassen werden de stukken doorgezaagd, en indien een las was gevormd ingebed en gepolijst, om vervolgens aan een microscopisch onderzoek te onderwerpen. Enkele stukken werden ook ge¨etst,en hardheidstesten werden uitgevoerd op 3 stukken. Ook enkele SEM-testen werden gedaan om het materiaal van de jet te achterhalen.

De meeste experimenten werden uitgevoerd met aluminium EN AW-6060 als materiaal voor het binnen- en buitenstuk, maar er werden ook succesvolle lassen gemaakt met koper R220 als buitenstuk en aluminum als binnenstuk, en aluminum als buitenstuk en staal C45 als binnenstuk. De lassen werden beoordeeld op basis van hun lengte, aangezien enkel een microscopisch onderzoek werd uitgevoerd.

Een uitgebreide bespreking van de resultaten is uitvoerig uitgeschreven in de hoofdtekst van deze thesis. Hier zal zich beperkt worden tot een overzicht van de resultaten uit zowel de experimentele waarnemingen, als de resultaten van een vergelijking van de experimentele resultaten met de literatuurstudie, het EXL en de formules.

Vervormingsgedrag

De buitenste buis zal enorm gaan vervormen door de inkrimping van zijn diameter. Deze vervorming resulteert in longitudinale spanningen die er voor zorgen dat het materiaal gaat opstuiken, en spanningen in de omtreksrichting die er voor zorgen dat de buis langsheen zijn omtrek gaat uitknikken. Ten gevolge van dit knikken, zal de las niet geheel dezelfde zijn xviii omheen de omtrek van het stuk. De residuele spanningen die achterblijven in de buitenste buis bleken voldoende te zijn om de las te doen bezwijken na het doorzagen. Tijdens de vervorming oefent de buis een kracht uit op de kraag van het binnenstuk, waardoor het binnenstuk naar binnen zal schuiven voor en na de impact, tot op het moment dat er een verbinding ontstaat.

De vervorming van de buitenste buis zal starten wanneer de scharnier gevormd wordt ergens langsheen de buis. Bij werkstukken met een kraag, zal op die plaats de scharnier vormen ten gevolge van de lokale reactiekracht op de buis. Bij de werkstukken zonder kraag vormt de scharnier op het punt waar de vloeigrens eerst bereikt wordt. De buis impacteert eerst aan zijn uiteinde, waarna het punt van impact opschuift terwijl de las gevormd wordt. De scharnier beweegt op hetzelfde moment in de richting van het binnenste werkstuk. Deze beweging begint voor de impact plaatsvindt.

Jet

De jet bleek voldoende te zijn om het materiaal vrij te maken van onzuiverheden. Metingen met de SEM wezen uit dat de jet bestaat uit zowel materiaal van het binnen- als buitenstuk.

Golfoppervlakken

De lassen waren ofwel vlak, of toonden een golvend oppervlak. Twee verschillende patronen van de golfvorm werden waargenomen. Eénwaarbij de amplitude en golflengte nagenoeg constant bleken langsheen de las, en éénwaarbij de golflengte toenam en de amplitude eerst toenam om daarna onmiddellijk terug af te nemen. Dit komt overeen met de patronen uit het EXL, en toont aan dat de impacthoek stijgt naarmate de las vordert.

Hardheid

De hardheidstesten toonden aan dat de hardheid van de tussenlaag veel groter was dan van het basismateriaal. Ook de hardheid van de twee werkstukken was toegenomen ten gevolge van de grote plastische vervormingen.

Smelten

In de golfoppervlakken van sommige experimenten werden zones waargenomen die gesmolten en opnieuw gestold waren. Zowel lokale smeltzones als een continue gesmolten laag werden waargenomen. Deze zones bleken zeer poreus te zijn, en vertoonden scheuren ten gevolge van deze porositeit en de residuele spanningen in de omtrek van de buis. Deze zones bleken dus zeer gevaarlijk voor de kwaliteit van de las. xix

Invloed van de parameters

Materiaal Voor de las tussen koper en aluminium bleek de golﬂengte veel kleiner te zijn dan voor experimenten met aluminium - aluminium. Dit duidt op een lagere impactsnelheid, wat inderdaad bevestigd wordt door de formules. Deze is te wijten aan de grotere massadichtheid van koper. Een las werd toch gevormd omdat door het zwaardere materiaal de impactenergie toch voldoende was. De experimenten met koper waren minder succesvol dan verwacht, zelfs een toename van het energieniveau bleek geen oplossing te zijn. De mogelijkheid bestaat dat de lassen bezweken zijn na het doorzagen, omdat de residuele spanningen in koper groter zijn dan in aluminium door de grotere elasticiteitsmodulus. De combinatie aluminium - staal was enkel mogelijk op hogere energieniveau’s. Dit valt ook af te lezen in lasvensters voor het EXL.

Vorm van de werkstukken De kraag op de binnenste werkstukken bleek geen invloed te hebben op de kwaliteit van de las, behalve bij grotere overlap van de ﬁeld shaper, omdat op die manier de impact hoek werd gewijzigd. Het gebruik van de binnenstukken met schuine conﬁguratie bleek geen voordelen te bieden op de gewone rechte stukken.

Positie field shaper De positie van de field shaper bleek een belangrijke parameter te zijn in het proces. Een verschillend venster van mogelijke posities bleek te bestaan voor verschillende werkstukken. De optimale overlap voor aluminium stukken met een dikte van 1, 5 mm bleek 11 mm te zijn. De positie be¨ınvloedt zowel de impactsnelheid als de impacthoek. Er werd gezien dat het golfpatroon opschoof bij een verplaatsing van de field shaper. Dit toont aan dat de impacthoeken verkleinen bij een grotere overlap van de field shaper.

Stand-off De golflengtes namen toe met een toenemende stand-off, om vanaf een bepaalde waarde weer af te nemen. Dit toont aan dat er een optimale waarde bestaat waarbij de grootste impactsnelheid bereikt wordt. Voor aluminium buizen met een dikte van 1, 5 mm was deze optimale waarde 3, 0 mm.

Dikte van de buis De golﬂengte nam af met een toename van de dikte van de buis van 1, 5 mm tot 2, 0 mm, wat duidt op een afnemende impactsnelheid. Door de grotere dikte zal het materiaal moeilijker vervormen, wat deze lagere snelheden verklaart. De lassen waren echter van dezelfde kwaliteit.

Voltage Een toename van het ingestelde voltage resulteerde in een verhoging van de impact snelheid, wat nodig bleek om aluminium met staal te verbinden. xx

Toepasbaarheid formules

De formules bleken nuttig om het effect van sommige parameters op het proces aan te tonen. Er wordt evenwel geen rekening gehouden met de positie van de field shaper. Ook met de grote spanningen in de omtreksrichting en de longitudinale spanningen wordt geen rekening gehouden. Dit zorgt voor een overschatting van de impact snelheden bij stukken die meer moeten vervormen, zoals de stukken met een grotere dikte en grotere stand-off. Voor kleinere vervormingen geven de formules wel realistische resultaten. De formules zijn m.a.w. niet bruikbaar om exacte waarden van de impactparameters te bekomen.

Conclusie

Deze thesis biedt meer inzicht in het vervormen van de buitenste buis bij het MPL, en de invloed van de verschillende parameters op het proces. Een vergelijking met het EXL blijkt nuttig te zijn om parameters in te schatten, of lassen in verschillende experimenten te vergelijken. De formules blijken, hoewel niet accuraat, nuttig te zijn voor het inschatten van het eﬀect van parameters. Bij toekomstige experimenten, en in een verdere stap praktische toepassingen, zal er op moeten toegezien worden dat smeltzones vermeden worden, gezien de grote porositeit die hier kan in ontstaan. Contents

Overview iv

1 Introduction 1 1.1 General ...... 1 1.2 Problem statement ...... 3 1.3 Goal ...... 3 1.4 Summary ...... 4

2 Literature study 5 2.1 Principle ...... 5 2.1.1 From start button to magnetic pulse ...... 6 2.1.2 From magnetic pulse to impact ...... 10 2.1.3 From impact to weld ...... 12 2.2 Equipment ...... 15 2.3 Workpiece requirements ...... 15 2.3.1 Materials ...... 15 2.3.2 Dimensions ...... 16 2.3.3 Practical examples ...... 17 2.4 Advantages ...... 17 2.4.1 New designs are made possible ...... 17 2.4.2 Reduction of manufacturing costs ...... 17 2.4.3 Higher quality of the weld ...... 19 2.4.4 Other advantages ...... 19 2.5 Disadvantages and limitations ...... 19 2.6 Comparison with explosive welding ...... 21 2.7 Summary ...... 22

3 Explosive welding 24 3.1 The bonding process ...... 24 3.2 The bonding interface ...... 25

xxi Contents xxii

3.3 Jet formation ...... 27 3.4 Bonding conditions ...... 28 3.5 Wave formation ...... 31 3.6 Variation of wavy interface with impact velocity ...... 34 3.7 Variation of wavy interface with impact angle ...... 35 3.8 Other parameters that aﬀect the wavy interface ...... 37 3.8.1 Strength characteristics ...... 37 3.8.2 Stand-oﬀ distance ...... 37 3.8.3 Flyer plate thickness ...... 38 3.9 Conclusion ...... 38

4 Profound view on the MPW process 40 4.1 From start button to pulse ...... 41 4.2 From pulse to pressure ...... 44 4.3 From pressure to impact ...... 46 4.3.1 Calculation of required pressure ...... 46 4.3.2 Check whether maximum velocity is reached ...... 48 4.4 From impact to weld ...... 51 4.5 Use of the formulas ...... 52

5 Process parameters 55 5.1 The impact welding parameters ...... 55 5.1.1 Impact velocity ...... 55 5.1.2 Impact angle ...... 56 5.2 Variable parameters ...... 57 5.2.1 Material properties ...... 57 5.2.2 Geometrical properties ...... 61 5.2.3 Voltage level ...... 67 5.3 Invariable parameters ...... 68 5.3.1 Frequency ...... 68 5.3.2 Length ﬁeld shaper ...... 69 5.3.3 Diameter of the ﬂyer tube ...... 69 5.4 Summary ...... 70

6 Experimental results 71 6.1 Experimental outline ...... 71 6.2 Material characteristics ...... 75 6.2.1 Aluminum EN AW-6060 ...... 75 6.2.2 Copper R220 ...... 75 6.2.3 Steel C45 ...... 76 Contents xxiii

6.2.4 Steel E235 ...... 76 6.3 Experimental results ...... 77 6.3.1 Deformation behaviour ...... 77 6.3.2 Jet formation ...... 84 6.3.3 Welded zone characteristics ...... 86 6.3.4 Interface morphology ...... 88 6.3.5 Inﬂuence of the parameters ...... 99 6.4 Summary ...... 108

7 Discussion 110 7.1 Impact behaviour ...... 110 7.2 Jet formation ...... 113 7.3 Melting ...... 113 7.4 Hardness increase ...... 114 7.5 Comparison with welding windows ...... 115 7.6 Influence of the parameters and comparison with the formulas ...... 117 7.6.1 Calculated impact velocities ...... 117 7.6.2 Influence of the thickness of the outer tube ...... 118 7.6.3 Influence of the stand-off distance ...... 118 7.6.4 Materials ...... 118 7.6.5 Position field shaper ...... 119 7.6.6 Applicability of the formulas ...... 119 7.6.7 Ratio of the pressure components ...... 120 7.7 Wave formation ...... 120 7.8 Summary ...... 121

8 Conclusion 123

9 Advice for future experiments 125 9.1 More material combinations ...... 125 9.2 Use of smaller ﬁeld shaper ...... 125 9.3 Adjusting workpieces ...... 126 9.4 Testing methods ...... 126 9.5 Numerical simulations ...... 127

A Experimental results 128 A.1 Series A: Aluminum to steel ...... 128 A.2 Series B: Aluminum to aluminum ...... 129 A.3 Series C: Aluminum to aluminum ...... 130 A.4 Series D: Aluminum to aluminum ...... 132 Contents xxiv

A.5 Series E: Aluminum to aluminum ...... 133 A.6 Series F: Aluminum to steel ...... 134 A.7 Series G: Aluminum to aluminum ...... 135 A.8 Series H: Steel to steel ...... 136 A.9 Series I: Aluminum to steel ...... 137 A.10 Series J: Copper to aluminum ...... 138 A.11 Series K: Aluminum to aluminum ...... 139 A.12 Series L: Copper to aluminum ...... 140

B Microhardness tests 141 B.1 Microhardness results C1 ...... 141 B.2 Microhardness results D1 ...... 143 B.3 Microhardness results G5 ...... 145

Bibliography 147 Chapter 1

Introduction

1.1 General

Joints between two workpieces are often the most vulnerable parts of a machine component or installation, especially when the joint is applied between dissimilar materials. Therefore, the search for more eﬃcient and more reliable ways to join similar and dissimilar material is continuous.

In many branches of the industry, it is necessary to use a combination of different materials in order to optimise the design of the final products. In the automotive industry for example, manufacturers desire to use more aluminum in combination with steel. This would reduce the vehicle’s weight, which increases the fuel efficiency and decreases the exhaust emission. There are opportunities for weight reduction in many automotive parts: drive shafts, suspension links, space frames, fuel and air supply, pressure and vacuum vessels, fuel cells,... [1]

When welding dissimilar materials, it is in most cases impossible to use fusion welding techniques, unless the two materials exhibit comparable characteristics. Because of the difference in melting temperature, thermal conductivity, volumetric specific heat, thermal expansion coefficient etc. (see table 1.1), large stress gradients will occur after the fusion welding. The brittle interlayer which is inevitably formed in the weld between two dissimilar metals or alloys will then show cracks [2].

There are several possibilities to avoid these brittle intermetallic compounds between two materials. This could be by inserting a metallic interlayer, using a ﬁller metal when fusion welding, or braze the material with the lowest melting temperature onto the other [2]. The solution that oﬀers most advantages however, is by using a solid-state bonding process. In that case, coalescence of two materials is produced at a temperature below the melting temperature of either of the workpieces. In solid-state welding, the heat conventionally used to join the materials is partly replaced by applying pressure on the materials. Both stress gradients,

1 Chapter 1. Introduction 2

Table 1.1: Thermal properties of materials

Material Melting Temp. Thermal Conduct. Expansion Coeﬀ. Speciﬁc Heat (°C) (W/m.K) (10-6/K) (kJ/kg.K) Aluminium 660 237 23.0 0.91 Copper 1084 353 - 386 17.0 0.39 Titanium 1670 16 - 22 11.0 0.54 Carbon steel 1425 - 1540 36 - 54 10.8 0.49 Stainless steel 1510 16.3 17.3 0.50

caused by differences in thermal properties, and the brittle interlayer are avoided this way. The different ways to apply this pressure led to a variety of solid-state welding techniques: cold welding, diffusion welding, explosion welding, forge welding, friction welding, hot pressure welding, roll welding, ultrasonic welding and magnetic pulse welding [3]. Of these welding techniques, Magnetic Pulse Welding (MPW) is often referred to as one of the most promising.

Magnetic pulse technology has been known for decades. At first it was mainly used for forming processes, but recently there has been a growing interest for the welding process. Magnetic pulse welding is similar to the explosive welding process, as in both techniques the necessary pressure to join the workpieces is created through impact. The difference is that in MPW the applied force results from a magnetic field instead of an explosion. Magnetic pressure is generated by discharging a tremendous amount of energy through a coil during a short lapse of time. The process typically takes less than 100 µs, while the recharging time takes about 6 s, depending on the required energy level. This means that with automatic magnetic pulse welding, it is possible to reduce the cycle time to less than 10 s [4]. Because no heat or additives are used it results in a clean weld. This can be seen in figure 1.1, where the difference between a MP-weld and a MIG1 weld in an automotive air conditioning accumulator is shown.

The nature of the MPW process makes it perfectly suitable for the welding of tubes. Seen as the great potential of this process for dissimilar materials, it opens up new and immediate opportunities for a variety of applications: fuel lines in automotive and aerospace applications, air conditioning systems, refrigerators, condensers,... Many domestic devices or heat-exchange systems require joining of aluminum with titanium or copper, while applications in power transmission often require the joining of aluminum to steel or stainless steel. A fast and cost-eﬀective method for joining dissimilar metals and their alloys might in these cases be perfect for mass production.

1Metal inert gas welding. A semi-automatic or automatic arc welding process in which a continuous and consumable wire electrode and a shielding gas are fed through a welding gun. Chapter 1. Introduction 3

Figure 1.1: Comparison of magnetic pulse welding (left) with MIG welding (right) [5]

1.2 Problem statement

As research in magnetic pulse welding is fairly new, most literature covers the technique itself, rather than the practical use of it. Few papers mention case studies of diﬀerent material combinations and/or the physical eﬀect of a varying parameter. No applicable range of parameters has been examined, even though the process parameters are of crucial importance to ensure a good joint quality. In order to use MPW on a large scale for various material combinations and workpiece dimensions, a ready-to-use window of parameters that need to be applied is required.

The two crucial parameters to ensure a good bonding quality are, similarly to explosive welding, the impact velocity and the impact angle. The impact velocity is the velocity at which the flyer workpiece collides with the fixed workpiece. The impact angle is the angle under which the workpieces collide. These two parameters are influenced by several parameters that are inherent to the process: parameters regarding the electrical circuit (energy level, capacity, frequency,...), parameters regarding the material characteristics (electrical conductivity, density, yield strength,...) and parameters regarding the geometry of the workpieces (diameter, thickness, stand-off,...). This wide variety of parameters makes it complicated to find the conditions for an optimal weld of 2 given workpieces, and surely to find a practical weldability window.

1.3 Goal

This thesis has to be situated in the Magpuls project, which is a cooperation between the research department of the Belgian Welding Institute, which is the coordinator of the project, and Belgian companies with an interest in the magnetic pulse technique. The main goal of the project is to acquire technological knowledge for useful industrial applications. Chapter 1. Introduction 4

In order to acquire fundamental knowledge, two things are required: a good understanding of the process, and enough experience with experiments. As we are at the start of this project, in this thesis a broad basis is provided for the further research on the magnetic pulse welding process. The goal of this work is to provide a better understanding of the process and more in particular of the influence of its parameters. Knowing the influence of the parameters on the resulting weld and/or its influence on the impact velocity and impact angle is essential for the creation and understanding of welding windows. Three different aspects are used to reach this objective: a literature study on the theory of the process, a literature study on the bonding in explosion welding, and a series of experiments.

1.4 Summary

This thesis is done at the Belgian Welding Institute and the Laboratory Soete of Ghent University. The practical use of the MPW technique is at this point too difficult due to both a lack of knowledge about and experience with the process. This is mainly due to the numerous process characteristics that influence the two crucial parameters to form a good weld. Experimental research in combination with a theoretical study of the MPW process and a profound study on the bonding in explosion welding should make it possible to achieve the goal of this thesis: understanding the influence of some of the process parameters. Chapter 2

Literature study

2.1 Principle

In magnetic pulse welding, an electromagnetic force is used to impact two materials at high speed against each other. A charging system is used to charge a bank of capacitors. When the required amount of energy is stored in the capacitors, it is instantaneously released into a coil which is placed at a small distance to one of the workpieces. This current induces a strong transient magnetic field which in turn induces eddy currents in the outer workpiece. These eddy currents prevent the magnetic field to diffuse through the material, and cause a difference in magnitude of the magnetic field on both sides of the piece. This difference is expressed in a magnetic pressure, which causes the outer workpiece to impact with the second workpiece. The following sections describe the process more in detail.

Figure 2.1: Magnetic Pulse system layout [5]

5 Chapter 2. Literature study 6

2.1.1 From start button to magnetic pulse

The electromagnetic welding setup consists of an energy-storage capacitor bank, a high- voltage charging power supply, a work coil and a discharge circuit (figure 2.1). The high- voltage charging power supply receives its power from the power grid and supplies it to the energy storage capacitor bank. The capacitor bank stores the energy until it reaches the predefined target level. The amount of stored energy is measured through the voltage level, which is the parameter that needs to be set before starting the process. When the required energy level is reached, the capacitor bank discharges a current pulse through a secondary circuit containing the coil. The capacitor bank needs to possess a sufficiently high capacitance in order to store enough energy, while the inductance of the discharge circuit should be low enough, in order to ensure a fast energy release and thus a higher current pulse when discharging the current through the coil [2]. The damped sinusoidal current set up in the work coil then produces a transient magnetic field [6]. A field shaper is used to concentrate the magnetic flux in the region required, which is more economical and quicker than using a special coil. Moreover, the same coil can be used in combination with several field shapers for different geometries. The whole setup must be electrically isolated, otherwise the fields generated by the conductors and coil can interact with adjacent tooling. Also, the coil and field shaper need to be of sufficient strength, as they will undergo a very high pressure opposite to the magnetic pressure on the tube.

Most experiments are done with tubular geometries as they are the easiest concerning the energy consumption and coil geometry [7]. The coil is in this case a multi-turn solenoid. The primary current running through the coil induces a strong magnetic field (see figure 2.2), with field lines parallel to the tube in the area within the solenoid. When inserting a conductive material into the coil, these field lines will be blocked by the workpiece. When the electromagnetic field penetrates the conductive workpiece, mobile charges will start to oscillate at the same frequency of the field causing an alternating current, the so-called eddy current. The magnitude of the eddy currents will be maximal the surface, and decline exponentially further inwards the material. These eddy currents hinder the further penetration of the magnetic field through the workpiece [8; 9], as can be explained physically.

There are different ways to interpret the effect of a conductive material blocking a magnetic field. It is the theory of magnetic shielding that was used before to describe this effect [10]. The material can be seen as a conductor for the magnetic field, so that the magnetic field completes its path through the material instead of through the inside of the workpiece. An- other way to look at the magnetic shielding effect is by seeing the workpiece as a compensator of the external field from the coil by means of fictitious magnetic charges. This interpretation is mostly used in process descriptions of the magnetic pulse welding and forming. A third interpretation is by seeing the inner workpiece as a solenoid that compensates the external Chapter 2. Literature study 7

Figure 2.2: Magnetic ﬁeld caused by an electrical current in a solenoid coil

ﬁeld by means of molecular currents in the opposite direction. However, these three diﬀerent physical mechanisms taken separately can’t account for all the aspects of the magnetic shielding. A more complex approach is required for an adequate description, but the mechanism described here can give a better insight on the importance of the process parameters.

The effectiveness of the shielding equation(2.4) can be quantified by splitting it into a component for the reflection of the magnetic field on the outer surface equation(2.1), a component for the absorption equation(2.2), and a component for the secondary reflections inside the material equation(2.3), which can have a positive or negative effect on the effectiveness. An overview of these shielding effects can be seen in figure 2.3 [11]. These effects are quantified by the following formulas [12]:

s 11, 735 r µ G · f R = 20 · log10( · + 0, 00535 · d · + 0, 354) (2.1) d G · f µ

p A = 0, 131 · t · µ · f · G (2.2)

2 (K 1) − A −j,227A B = 20 · log10( 1 − · (10 10 ) · (e ) ) (2.3) | − (K + 1)2 |

SE = R + A + B (2.4)

With: SE = 20 · log Bin = shielding effectiveness [dB] 10 Bout R = the reflection losses [dB] A = the absorption losses [dB] B = the secondary reflection losses [dB] Chapter 2. Literature study 8

d = the distance between the source of the magnetic ﬁeld and the material [mm] G = the electrical conductivity of the workpiece material relative to the one of copper [-]

µr = the magnetic permeability of the workpiece material relative to the one of copper [-] f = the frequency of the current [Hz] t = the thickness of the material [mm] Z µ 1 S · 2 K = = 33 ( · · 2 ) | ZH | f G d ZS = the impedance of the workpiece material [Ω]

ZH = the impedance of the magnetic ﬁeld [Ω]

Figure 2.3: Loss in magnetic ﬁeld trough shielding eﬀects

The conclusion of this, is that most of the magnetic ﬁeld is blocked by a conductive material, in this case the outer workpiece, and only a small part diﬀuses through it. Therefore, the

magnetic ﬁeld density inside the workpiece Bdiff will be much lower than the density between

the workpiece and the coil Bgap, as is illustrated in ﬁgure 2.4.

Successful magnetic pulse welding experiments have also been performed for on metal sheets [14]. These require a different shape of the coil, as the magnetic field required to generate the accelerating force needs to be shaped different from the one used with tubes. The coils used so far have either an H-shape or an E-shape (figure 2.5). The experiments showed that the strength of the weld declines near the edges of the sheet, which could be the result of a change in direction of the eddy currents [14].

The use of tubular workpieces surrounded by a tubular coil is at this moment the most eﬃcient way to experiment, due to the ease of controlling the magnetic ﬁeld inside the tubes. Chapter 2. Literature study 9

Figure 2.4: Magnetic ﬁeld in and around the workpiece

A multi-turn coil is often inductively coupled to a field shaper. If properly designed the entire current flux that is created in the primary current can be transferred to the bore of the shaper. Field shaper based coils are used to develop high electromagnetic pressure while being able to increase the inductance of the coil. The principle is shown in figure 2.6. This is done at the cost of being less efficient than a well-designed single-stage coil would be, as more energy is dissipated in the field shaper [15]. But nevertheless it is more efficient, as the field shaper homogenises the magnetic pressure acting on the tube, increases the pressure value and decreases the distributional gradient of pressure at the end of the tube [13].

Figure 2.5: Examples of a double layer H-shaped ﬂat coil (a) and one layer E-shaped ﬂat coil (b). [14] Chapter 2. Literature study 10

Figure 2.6: The current in the ﬁeld shaper induced by the primary coil [13]

2.1.2 From magnetic pulse to impact

The difference in magnetic field between the inner and outer side of the workpiece can be found through the shielding effectiveness in the above expressions (2.1), (2.2), (2.3) and (2.4). These formulas estimate this gradient in magnetic field, when changing several parameters. A comparison of the percentage of the magnetic field that is shielded, is given in table 2.1 for aluminum and copper at different frequencies and workpiece thicknesses.

Table 2.1: Percentage of the magnetic ﬁeld that is shielded

Aluminum Copper frequency [Hz] t = 1 mm t = 1, 5 mm t = 2 mm t = 1 mm t = 1, 5 mm t = 2 mm 10,000 66 81 90 78 90 95 28,000 87 95 98 94 98 99 50,000 95 98 99 98 100 100

The range of parameters that is used, is the same as what will be used in the experiments1. When applying the formulas on these examples it became clear that shielding happens mostly by absorption in the workpiece and that the component of the secondary reflection is negligible. For example at a frequency 28.000 Hz and with an aluminum piece of 1, 5 mm thick, the absorption component is 96% of the shielding effectiveness. A higher effectiveness, and hence a higher difference in magnetic field is reached with increasing frequency, thickness and electrical conductivity. Applying the formulas on a ferromagnetic material with high magnetic permeability such as steel show an increase as well.

1In the experiments the current will always ﬂow be at 14.000 Hz. The examples here show the possible inﬂuence of changing the frequency. Chapter 2. Literature study 11

A magnetic pressure is associated with a magnetic field. A gradient in the magnetic field strength causes a force due to the gradient in magnetic pressure. Because of the difference in magnetic field inside and outside of the workpiece, such a magnetic pressure will compress the flyer tube radially. The eddy currents in the workpiece, caused by the magnetic field, need to be large enough and be concentrated close to the edge of the workpiece to block the magnetic field sufficiently. If not, the impact force resulting from the difference in magnetic field inside and outside the tube won’t be strong enough to create bonding.

Figure 2.7: Forming of wavy interface [5]

(a) Aluminum - Aluminum (b) Aluminum - Copper

Figure 2.8: Wavy interfaces in experimental magnetic pulse welds [2] Chapter 2. Literature study 12

2.1.3 From impact to weld

The collision causes bonding between the two elements through several binding mechanisms. Good bonding between the materials is ensured when the distance between their atoms becomes smaller than the range for their mutual attractive forces. In that case, electrons are shared between the two materials and an intermetallic phase with high hardness is formed. Previous experiments show that in general a wavy morphology bond interface is formed with occasional laminar interface morphology. This transition zone might be caused by mechanical mixing, intensive plastic deformation and/or melting. The temperature increase to cause this melting occurs due to joule effects and the collision itself. Because the process takes place in a short lapse of time, the heating is not enough to generate a temperature rise in a wide area, so there is no significant heat affected zone, as can be concluded from previous experiments. [16, 5]

Figure 2.9: Microstructure of the welded interface for diﬀerent material combinations [17]. (a)aluminum tube to aluminum core (b)aluminum tube to mild steel core (c)aluminum tube to copper core

The exact microstructure of the welded zone depends on the process parameters and the materials used. The formation of the wavy interface is similar as in the explosion welding process, and can be seen in figure 2.7. Typical examples of the wavy interface are shown in figures 2.8 and 2.9. Figure 2.9 shows the microstructure of the welded interface for experiments executed using the same parameters but with different materials. This illustrates the material Chapter 2. Literature study 13

Figure 2.10: Hardness distribution across the interface of a welded joint of aluminum [17]

Figure 2.11: Hardness distribution across the interface of welded joints [17]. (a)aluminum tube to mild steel core (b)aluminum tube to copper core (c)aluminum tube to titanium core

dependence of the interface structure. The aluminum - aluminum connection in figure 2.9(a) has a simple wavy pattern. The aluminum - mild steel interface in figure 2.9(b) shows a wavy pattern on the boundary of aluminum and the intermediate layer, and a flat boundary between the intermediate layer and the mild steel. In figure 2.9(c) the wave inclines and swallows a part of the interlayer.

The interface layer shows an increase in hardness relative to the base material. This can be attributed to severe plastic deformation or to a new fine-grained microstructure produced by melting and rapid solidification of the weld interface. This interface layer may support two possible mechanisms of bond formation: 1. bonding as a result of solid-state processes based on accelerated mass transfer due to intensive plastic deformation at very high rates, and 2. bonding as a result of solid-liquid interaction and based on the formation of a thin layer of molten metal between the components. [16] Examples of this hardness increase that may evidence these two mechanisms are shown in figures 2.10 and 2.11. Mark that the hardness distributions shown in figures 2.10, 2.11(a) and Chapter 2. Literature study 14

Figure 2.12: Distribution of element concentrations across the interfaces of welded joints [17]. (a)aluminum tube to iron core (b)aluminum tube to mild steel core (c)aluminum tube to copper core

2.11(b) correspond with the samples of the microstructures shown in figures 2.9(a), 2.9(b) and 2.9(c) respectively. The hardness increase in the aluminum - aluminum weld (figure 2.10) is clearly the consequence of the strain hardening of the material. The maximum hardness occurs at the interlayer, but is not that much higher than that of the tube. The tube hardness is higher than the core hardness, as larger deformation occurs in the tube. Much larger hardness increases occur in the welded joints of dissimilar materials, as it has been found for aluminum to mild steel (figure 2.11(a)), aluminum to copper (figure 2.11(b)) and aluminum to titanium (figure 2.11(a)). This increase in hardness indicates the origin of an intermetallic compound layer. This is proven by X-ray microanalysis of these interfaces, which shows a zone in which the concentration of both elements are at certain constant levels. The thickness of this compound layer varies between 2 µm and 20 µm and is smaller than those produced by other solid state welding processes. [17]

Microscopic roughness of the materials and surface contaminants could prevent the materials from being brought close enough together. Cleaning of the surfaces before the start of the Chapter 2. Literature study 15 process would therefore be required. However, the workpieces are impacted with a high velocity under a certain angle, which generates a jet force along the materials surface before they contact. This jet force is able to remove surface contaminants such as oxide ﬁlms, which reduces the need for pre-process cleaning. Also, due to the intense plastic deformation, mostly in the more ductile material, microscopic roughness isn’t necessarily an obstacle when bringing the workpieces together. In general it is assumed that no pre-weld cleaning is required at all [4].

The bonding mechanism in a solid-state welding process is thought to be the result of several processes. Melting, diffusion, recrystallisation and interfacial reaction are heat-assisted processes that can cause bonding [18, 2]. The two most important processes however are adhesion and the forming of the wavy structure between the two materials [18]. Although there is no unified theory which explains the adhesion phenomena, Nenakhov gives a good summary of the many definitions describing adhesion:“a heterogeneous body consisting of two dissimilar condensed phases brought into contact; moreover, these bodies are bound with one another by intermolecular forces through the interface” [19]. The wavy structure increases the possibility to transfer forces through the interface, and thus has the same positive effect as mechanical interlocking, which is realized on a smaller scale. Although often referred to as one, mechanical interlocking is not an adhesion process [19]. As it is not known what exactly the importance of each of these processes is, it is hard to determine the physical effects of the process parameters in a theoretical way. Experimental results are therefore of great importance.

2.2 Equipment

The equipment that is used for the experiments described in this thesis is a Pulsar model 50 25, which is shown in figure 2.13. As mentioned before, this consists of an energy-storage | capacitor bank (in the middle on the figure), a high-voltage charging power supply (left on the figure), a work coil and a discharge circuit (right on the figure). The maximum energy storage of the capacitor is 50 kJ which corresponds with a maximum charged voltage of 25 kV. The value of the current can be up to 500 kA and the frequency of the current is 14 kHz. [20]

2.3 Workpiece requirements

2.3.1 Materials

Theoretically, the inner tube can be any material, as long as it is prevented from deforming. Of more importance are the properties of the outer tube material. These need to have a high electrical conductivity, in order to create larger eddy currents and thus induce a larger accelerating force. Materials with a lower electrical conductivity would hold smaller eddy Chapter 2. Literature study 16

Figure 2.13: Magnetic Pulse system machine [20]

currents and therefore have a smaller impact velocity. This could be solved by either increasing the energy or by applying a conductive driver around the outer tube, which can be removed afterwards, or be kept as an integral part of the workpiece. Materials that have an electrical resistivity of 15 µΩ/cm or less lend themselves to direct welding. That includes aluminum, copper, low carbon steels and most precious metals [21].

Earlier research has shown that it is possible to join several combinations of dissimilar materials, as long as the process parameters are chosen correctly. Successful case studies of MPW between aluminum and tungsten [18], titanium and inconel [18], aluminum and aluminum [16, 22, 17], aluminum and copper [2, 22, 17], aluminum and titanium [2, 23, 17], aluminum and iron [23, 17], aluminum and magnesium [23] aluminum and steel [14, 22, 17], aluminum and stainless steel [2, 8, 17], aluminum and bronze [17], aluminum and molybdenum [17] have been realised before. Other sources also mention the possibility of joining copper to copper, copper to steel, copper to brass, nickel to titanium, nickel to nickel, steel to steel [7].

2.3.2 Dimensions

As far as the geometrical shape of the workpieces is concerned, axi-symmetrical parts are preferred. For an optimal joint strength, a tubular section is required [2]. As mentioned before, successful experiments have also been done with sheets [14]. The diameter of the tubes is limited by the coil, which gets more expensive for larger coils. As MPW lends itself perfectly for mass production, it is most interesting to apply on smaller diameters. The diameter has to be big enough to prevent the magnetic ﬁeld from interacting with itself. Chapter 2. Literature study 17

The range of thicknesses of the outer tubes is limited by the maximum weight of the tube, which increases the necessary energy to initiate the acceleration of the tube, and the reduced formability of the tube. The thickness also has a minimum value in order to allow eddy currents to run in the material and prevent the magnetic ﬁeld from diﬀusing through it.

2.3.3 Practical examples

In table 2.2, an overview can be found of examples of successful magnetic pulse welds that were found in literature. This table gives an idea of the dimensions and materials that are possible to join with the magnetic pulse welding technique.

The MPW technique is already used on a larger scale in a few industrial applications. For example at TI Automotive in Milan MPW is used for the serial production (200.000 pieces/year) of receiver driers for air conditioning systems [24]. At Hirotec America Inc., MPW is used to join aluminum and steel in hybrid crash beams for automobile doors [1].

2.4 Advantages

2.4.1 New designs are made possible

Possibility to join dissimilar materials. [5]

No additional materials are used. [4]

The materials can be pre-ﬁnished (paint, alodine2, anodize3) before the assembly is realized.

2.4.2 Reduction of manufacturing costs

No tools need to be used for the cleaning or ﬁnishing of the workpiece. [5]

The process is instantaneous and highly repeatable, which makes the production rate higher than for any other conventional joining process. [4]

Because the large amount of energy is released so quickly, the actual energy expenditure is up to 10 times less then conventional welding processes. [20]

No additional materials are needed. [5]

No need for pre or post weld cleaning or deburring, post weld thermal treatment or slag removal. [20]

2A brand of Chromate conversion coating products used on aluminum alloy and to a lesser extent magnesium alloy metals, primarily in the aerospace industry. 3An electrolytic passivation process used to increase the thickness of the natural oxide layer on the surface of metal parts.

Chapter 2. Literature study 18

various Al99,5 c - 10 10 15 [17] 1,5 1,0 bar 10,0 11,2

Al99,5 Al99,5 - 10 10 15 [17] 1,0 bar 11,2 1,5 - 2,2 8,6 - 10,0

40 brass, commercial pure

Al99,5 Al99,5 - 10 10 15 [17] 0,6 bar var, 11,2

0 - 1,5

l6061 Al l6061 Al -

15 25 96

[22] 2,3 1,7 bar 88,0 17,7

Al99,5 Cu99,5 15 25 25 [22] 2,4 2,5 1,5 21,8 17,7 tube , 0,45% carbon steel (JIS S45C), austenitic stainless

e te S355J2H Steel AlMgSi0,5 mm

15 25 25 / [22] 4,5 1,7

17,7 tube 25,0

te C45 Steel AlMgSi0,5 - = 780 N 22 25 [22] 2,4 2,1 bar 16,0 19,8 15,7

AlMgSi0,5 AlMgSi0,5 5 mm and higher. -

, 22 25

[22] 2,4 2,1 bar 16,0 19,8 15,7

Ti-6,1Al-3,9V Al99,2 - [2] 20 12 20 2,0 bar var, 12,25 0,5 - 5

d Al99,2 Cu99,9 - [2] 20 12 20 2,0

bar var, Overview of magnetic pulse welding examples

12,25 0,5 - 5

tils steel Stainless Al99,2 2 [2] 20 12 20 2,0 2,0 12,0 tube 12,25

Table 2.2:

noe 625 Inconel Ti-3Al-2,5V

b a

22 [18] 2,2 0,9

16,0 1,25 40,5 22,5 tube

l6061 Al use K1700 Tunsten - 16 [18] 1,8 0,9 bar 11,0 40,5 22,5 [kV] h Gap [mm] Material Material Reference Thickness [mm] Thickness [mm] Geometry Geometry Energy level [kJ] Copper driver sheet isInconel used, is an austeniticSuccessful superalloy experiments composed were of done mainly with nickel, chromium, mild iron, steel molybdenum (JIS and SS41), niobium, high strength steel ( The inner bars wereThis conical, is hence the the voltage gap level width that changes would along have the to overlap, be Optimal set result to was reach found the for same this energy gap level width, as in these tests, Variable - inner bars wereThe conical, outer hence tube the wasVarious gap deformed gap width to widths changes allow have along overlap, been the tried. overlap. Optimal results were found for a gap of 1 Peak Voltage Overlap length [mm] f Internal tube Outer diameter [mm] Outer diameter [mm] b c g e a h d External tube steel (JIS SUS 304), ductile cast iron, malleable cast iron, Al-Mg alloy (JIS 5052), commercial pure copper, phosphor bronze, 60 molybdenum, commercial pure titanium and commercial pure magnesium. Chapter 2. Literature study 19

Because of the negligible process failure, almost no inspection or rework is required. [4]

2.4.3 Higher quality of the weld

There is no signiﬁcant heat aﬀected zone (HAZ) which eliminates the risk for local annealing and degradation of the material. [17]

Because of the high speed, the formability of the materials is increased.

Inertia forces during the process reduce wrinkling.

The weld is stronger than the weakest material. [5]

The deformation is plastic and not elastic so the springback is reduced.

The joints are pressure tight. [5]

No damage of the workpiece’s surfaces as there is no contact with the welding tool.

No corrosion development in the welding area. [5]

2.4.4 Other advantages

There is less heat and no radiation, gas or smoke produced [5].

No fumes are emitted, so the process does not form a health hazard for the welder

The work equipment doesn’t contain any moving parts, which increases its safety level.

2.5 Disadvantages and limitations

The magnetic pulse welding process imposes some limitations considering the workpieces be welded. The geometry is set by the shape of the coil, while the size is set by the field shaper. Some shapes, such as a rectangle, are always difficult to weld as the sides are much easier to move than the corners. Also the size of the welded parts is limited. Magneform mentions possible diameters ranging from 0, 64 cm to 25, 4 cm [21]. The largest tube diameter that is welded until now was 12, 1 cm, larger sizes have not been tried due to a lack of demand [7]. With diameters chosen too small, the magnetic field may interact with itself. The maximum size is limited by the cost of the machine, which rises significantly for bigger sizes. As discussed in the material section on page 15 the outer tube needs to be a good electrical conductor, otherwise a conductive driver could be used to increase the impact velocity. When the thickness of the inner tube is insufficient to withstand the impact of the outer tube, a mandrel is required to prevent deformation. Chapter 2. Literature study 20

Lap joints are required since the outer member must impact the inner to create the weld. The actual welding zone is therefore not accessible, which makes it unlikely to test the quality of the weld with non-destructive tests. The process can not be used in-field due to the size of the welding machine. Seen as the potential applications for the process, which are mostly factory- made parts, this is not necessarily a disadvantage. Including the discontinuous magnetic pulse welding process in a continuous production line is however difficult, and will be a challenge for the future. Also, one should be precautious because of the influence of the magnetic field on internal components and other plant equipment.

During the process, very high electrical currents are handled. This method is theoretically capable of reaching velocities of the order of 15.000 m/s, but in practice is limited to lower velocities of less than 1.000 m/s due to problems associated with power management, for example coils that tend to vaporise [25]. However, this does not cause limitations for the small geometries that are used. Because the process is so fast, it does not lend itself to deep drawing of the outer tube since the material does not have the time to stretch [21].

The parameters of the process are very part specific as can be seen from previous experiments. The quality of the weld is sensitive to fluctuations in the collision conditions, such as coaxial positioning of the parts or the angle of impact. The sensitivity to these fluctuations is smaller when the interface is more wavy [18].

Summary of the limitations:

Only lap joints can be formed.

Only tubes and sheets have been possible to weld until now.

The size of the tubes is limited.

Deep drawing of the material is not possible.

The process only works with a high conductive ﬂyer material.

The impact velocity is limited.

Summary of the disadvantages:

The process is not suitable for in-ﬁeld applications.

The welding zone is not accessible.

The process is very sensitive to changes in process parameters. Chapter 2. Literature study 21

2.6 Comparison with explosive welding

The principles of magnetic pulse welding and explosive welding are the same: coalescence of the workpieces through high velocity impact. A typical layout of an explosive welding experiment is shown in figure 2.14. The parameters that can influence the resulting weld and that are inherent to the merging process of the materials are therefore the same. Collision angle and impact velocity play a crucial role for the weld quality, and the resulting welds have the same characteristics. As a jet force is created along the surface, bonding is not disturbed by oxides or surface influences in either of the techniques. In explosive welding, the weld can be formed for a certain welding window containing a required range of impact velocity and impact angle. This range of parameters might be used here as well, as both the impact velocity and angle are also characteristic to the magnetic pulse welding. An example of such a weldability window is shown in figure 2.15. The physical meaning of the borders will be discussed in the following chapter. [18]

The generation of the impact force is different in both techniques. While MPW uses a magnetic force to deliver a sufficient amount of energy for impact, explosive welding uses an explosive. Obviously, parameters regarding the generation of the magnetic field cannot be related to the explosive welding process.

Figure 2.14: Explosive welding process

The main advantages of explosive welding are similar to those of magnetic pulse welding and are inherent to a solid state bonding process: dissimilar metals can be joined, there is no HAZ that can cause degradation of the materials’ properties and the strength of the weld is stronger than the weakest material. The most important limitations again are the same: the workpieces need to have a simple geometry and a high enough impact resistance. Explosive welding can be successfully used for simple geometries like flat, cylindrical and conical workpieces, while in magnetic pulse welding even the flat workpieces require a more Chapter 2. Literature study 22 complex shape of the coil and field shaper. The explosive welding process is determined by three parameters: the explosive velocity and load, and the stand-off distance between the plates. These have an influence on the penetration depth, plate velocity and impact energy. The angle of impact is determined by the stand-off distance and the flyer plate properties, and influences the jetting force. With good knowledge of the characteristics of the explosive used, it is rather easy to obtain the right parameters. This in contrast with MPW, where the choice of the coil geometry, field shaper, energy level and frequency of the current, and conductivity of the workpiece material influence the strength of the magnetic field and therefore the pulse force.

Figure 2.15: An example of a weldability window [26]

The similarities between explosive welding and magnetic pulse welding might be useful to help understanding the actual bonding process and more important to obtain the required impact velocity and stand-oﬀ distance between two materials with known thicknesses. The parameters for the MPW setup could then be estimated starting from these required parameters. A more profound view on the explosive welding process will be given further in this work.

2.7 Summary

In this chapter an overview was given of a literature study on the magnetic pulse welding process. It was explained how a current pulse leads to a magnetic pulse, which is blocked by the outer of 2 workpieces. This leads to a magnetic pressure on the flyer workpiece, pushing it towards the inner workpiece at high velocity. The result is a solid state bond between the two pieces, often with a wavy interface. MPW offers the possibility to produce new material combinations, increase the welding quality and reduce the manufacturing costs. The disadvantages so far lay in the geometry and the limitations for the flyer material. The research Chapter 2. Literature study 23 done so far is mostly on tubular workpieces, but with a variety of material combinations. In this thesis, experiments will be done with a magnetic pulse system model 50 25 delivered by | Pulsar. Chapter 3

Explosive welding

As the explosive welding process is known and used in large-scale applications in the manufacturing industry already for several decades, considerable progress has been made in the search for the optimum operational and physical parameters required to produce an accept- able bond. This led to a series of welding windows (of various parameters such as ﬂyer plate velocity vs. impact angle and impact pressure vs. impact angle etc.) proposed by diﬀerent authors. In this chapter an overview is given of a literature study on the bonding process in explosive welding, together with some welding windows that can be used as a start for magnetic pulse welding experiments. The conclusions from articles [27, 28, 29, 30, 31, 32] are experimental, the conclusions from [33] are numerical, and the conclusions from [34, 35] are both numerical and experimental. The conclusions for explosive welding usually refer to the impact of plates.

3.1 The bonding process

Up to now, there is still no quantitative theory capable of accurately describing and predicting the parameters and the characteristics of the explosive welding process. The process has mostly been considered to be a solid-state welding process, but according to some authors it is also a fusion welding process. [35]

The fusion welding theory relies on the dissipation of the kinetic energy of the workpiece at the interface. This serves as a source of heat that causes bilateral melting across the interface and diffusion within the molten layers. This is followed by cooling at a rate of 105 K/s creating an ultrafine grain size. However, fluid diffusion leads to a gradual transition from one material to another, while in explosive welding the transition is very sudden. Computer simulations ruled out the possibility of fusion welding, as it showed that the temperature increase at the surface is not sufficient to melt the material. But, the temperature increase is high enough for phase changes to occur, i.e. similar to what happens in tempering in steels. [35]

24 Chapter 3. Explosive welding 25

The pressure welding theory is based on the assumption that large plastic deformations occur at the interface which allow clean surfaces to be formed and a solid diffusion process to take place. Therefore, the high pressure should be maintained for a sufficiently long time. However, in explosion welding, the peak pressures are maintained only for a few microseconds and the coefficient of diffusion is small. These two reasons reduce the chance that sufficient solid diffusion occurs. Moreover, the phase changes that are shown in explosively welded specimens are not predicted by the solid state theory. These phase changes occur due to high temperatures (but less than the melting temperature) at the collision point. As there is no time for heat transfer in metals, the result is an ideal metal-metal bond without melting or diffusion. [35]

In addition, the interfacial waves, vortices and/or melt pockets often observed cannot be accounted for by either pressure welding or pure melting mechanisms. The interfacial grain deformation and wave formation suggest that the mechanism of welding must be associated with a ﬂow process. [35] In previous calculations, the metal behaviour in the vicinity of a contact point at high-velocity oblique 1 impact was satisfactorily described from the hydrodynamic point of view. In fact, thin pre-boundary metal layers with a thickness of some microns transform into a liquid phase. [31] Hence, a hydrodynamic ﬂow process is the best way to describe the bonding.

3.2 The bonding interface

Depending on the impact velocity and angle, and thus the initial process parameters, the interface of explosive welded metals can vary widely. Straight smooth waves, small waves, large waves, waves with vortices (figure 3.1), waves with solidified melt pockets, and continuous fused molten layers are the most important possible types of interfaces. The wavy bond without an apparent intermediate layer is preferred, as this type generally yields the most desirable properties. The advantages of these interfaces are a solid phase bond, absence of metallurgical defects and its associated problems. The advantages of the wave formation are an increase of the contact surface between the materials, higher mobility of atoms and dislocations, and effective removal of contaminations. [32]

The continuous fused layer is generally limited to the center of the interface. In the combination of titanium and steel, a layer was found with a thickness around 0, 05 mm to 0, 2 mm and includes very ﬁne grains of about 0, 01 mm. [35] These intermetallic compounds degrade the quality of the bond, cause although the hardness of this interlayer is usually higher than the base material, this material is brittle and tends to crack easily. [27]

1Normal shock wave loading only accounts for normal stresses, while oblique shock wave loading also accounts for shear stresses. Chapter 3. Explosive welding 26

Figure 3.1: (a) Wave formation for impact of titanium ﬂyer plate (3 mm) on titanium base plate (30 mm) with an impact velocity of 486 m/s (b) Formation of vortices for impact of titanium ﬂyer plate (3 mm) on titanium base plate (30 mm) with an impact velocity of 606 m/s [35]

In previous experiments it was found that tensile failure for specimens with waveless interfaces that didn’t show intermetallic formation occurred in the aluminum, away from the weld interfaces. In the same experiments, for samples with wavy interfaces containing Fe-Al in- termetallics, the tensile failure occurred at the weld interfaces. This shows that avoiding the formation of an interlayer can be of greater importance than generating a wavy surface. And even if the intermetallic compounds form, they do not necessarily play a signiﬁcant role in determining the bond strength if the layer thickness is within several micrometers. [18]

Figure 3.2: Theoretical boundaries of wave formation for collisions (a) of ﬂat streams of Newtonian liquids, and (b) of ﬂat plates of elastic-plastic solids. (c) Typical observed boundary of wavy bond zone in metal cladding. [18]

Cowan proposed a mechanism that stated that the generation of waves in explosive welding Chapter 3. Explosive welding 27 depends on the collision point velocity and the collision angle. The conditions for the transition from smooth to wavy interfaces are shown in ﬁgure 3.2. The actual boundary (the dashed line in ﬁgure 3.2) is a combination of the behavior of the metals as solids (at low velocities) and as liquids (at high velocities). At low velocity boundary, the transition velocity is largely independent of the impact angle α and the limit is mainly determined by the yield stress of the material. At high velocity boundary, as shown by the dashed line, the behavior becomes more complicated because of possible melting and extremely high pressure at the collision point. In practice, the constant transition velocity vt for elastic-plastic solids is mainly used as the boundary to set up welding windows. [18]

3.3 Jet formation

The high-localised pressures created at the collision point spread away at the sound velocity. Because the collision is moving forward at a subsonic rate, pressures are created at the approaching adjacent surfaces which are sufficient to crumble a thin layer of metal from each surface and eject it away in the form of a jet (figure 3.3). A minimum angle is required to ensure a pressure of sufficient magnitude so that it exceeds the dynamic elastic limit of the material and deformation of the metal surfaces into a jet can occur. It has often been noted that most of the jetting material, which is sometimes referred to as backflow material, comes from the flyer part. [35, 28]

Figure 3.3: Jet formation during plate impact, with 1) the ﬂyer plate, 2) the ﬁxed plate and 3) the jet material. [30]

The thickness of the removed layer is generally less than 0.05 mm. The thickness of the Chapter 3. Explosive welding 28 jet is greater than the height of microroughness on the base material and is comparable with the amplitude of the interface wave proﬁle. This way the jet can severely deform the microroughness and damage the inert surface layer. [30, 28, 35]

Despite the small thickness, up to 50% of the kinetic energy of the flyer plate can be concentrated in this backflow of material due to its high velocity, which reaches values 30-50% larger than the impact velocity. Apart from preparing the surface of the materials, the jet also increases the temperature of the materials and this way lowers its strength. One can say that due to this effect the effective time of interaction of the welded surfaces increases, because the jet already starts heating the surface prior to impact. [30]

3.4 Bonding conditions

The impact pressure that is associated with the dissipation of kinetic energy must be higher than a threshold value dependent on shear wave velocity, and be maintained for a suﬃcient length of time to achieve the stable inter-atomic bonds. The impact velocity (vc on ﬁgure

3.3) then determines the pressure, while the velocity at the collision point (vcp on ﬁgure 3.3) determines the time available for bonding. However, experimental evidence shows that successful welding also depends on the impact angle. [35]

Therefore, the conditions for a good quality explosive weld are generally given in terms of impact or collision point velocity and impact angle. In diagrams containing the impact angle

α in the ordinate and the collision point velocity vcp in the abscissa, a certain window of these parameters that can result in a good weld is then shown. An example of such a welding window is given in ﬁgure 3.4. The physical meaning of the boundaries 1, 2, 3, 4, 6 and 7 which form the weldability window is explained below. The physical meaning of boundary 5 was not found in the literature study and is therefore believed to be of lesser importance than the other boundaries.

It should be noted that the formulas used here are in general not used to obtain welding windows, as most welding windows are obtained through experimental research. The boundaries 6 and 2 however are sometimes used as controlling conditions for the impact velocity that is needed to be used in the process. As the impact angle is a variable during the process, its boundaries are usually not controlled. In other words, the formulas 3.1 and 3.5 can lead to a simple estimation of the impact velocity. This velocity should then be chosen as close as possible to the lower boundary 1.

Boundaries 1 and 6 The lower boundary of the impact velocity is the minimal velocity required to establish a wavy interface. These correspond with the boundaries presented by Cowan and illustrated in ﬁgure 3.2. In other words, this is the transition velocity from a Chapter 3. Explosive welding 29

Figure 3.4: The weldability window in explosive welding [26]

smooth to a wavy interface. As it was mentioned in chapter 3.2, a wavy interface yields the best weld properties.

Boundary 6 was based on the mechanism for wave forming proposed by Cowan, when the material behaves as a liquid. The following condition is used for the minimal value of the impact velocity. This relationship was based on a hydrodynamic model of the weld, where the smooth - wavy boundary corresponds to the laminar-turbulent transition. The transition is characterized by a Reynolds number as given in the following relation: [18, 34, 32]

(ρa + ρb) · 2 RT = vcp (3.1) 2 · (Ha + Hb) With:

RT = Reynolds number [-] ρ = the density of the materials [kg/m3] H = the Vickers hardness of the material[MPa] vcp = the collision point velocity [m/s]

In [34] it was tried to ﬁnd a relationship between impact velocity and impact angle to fulﬁll the condition of a wavy interface. The Reynolds number in this work becomes dependent of the impact angle and can be given through the following relationship based on a model for Chapter 3. Explosive welding 30

Fe-Fe, Cu-Cu and Al-Al couples: [34]

2 3 RT = 93, 02( 9, 62) 13, 45( 2, 06)α + 0, 71( 0, 14)α 0, 012( 0, 03)α (3.2) ± − ± ± − ± With:

RT = Reynolds number [-] α = the collision angle [◦]

The impact angle that corresponds with a Reynold number of 10,6 is 12◦.

The equation presented in [34, 30] is similar to the condition 3.1:

s · H α = k · 2 (3.3) ρ vc With: k = a constant [-] ρ = the density of the less strong material [kg/m3] H = the Vickers hardness of the less strong material[MPa] vc = the impact velocity of the collision [m/s]

In [30] the factor k is deﬁned as 1.14. In [34] the factor k is deﬁned as a value between 0.6 for high quality pre-cleaning of the surfaces and 1.2 for imperfectly cleaned surfaces. In that case, the cleaner the surfaces are previous to the welding, the lower the boundary will lay and the bigger the weldability window will be.

Usually, the practically used condition is one value for the minimum impact velocity. The Reynold number for this transition value is mostly set to 10.6. This corresponds with an elastic-plastic behaviour of the materials and is represented by boundary 1.

Boundary 2 The upper boundary of the collision point velocity is determined by the transition from the subsonic to the supersonic regime. This transition velocity equals the speed of

sound c0 inside the base material. Above this value no jet will be formed during the impact and consequently no bonding will take place. This leads to the following condition: [34, 32]

s E vc < c0 = (3.4) 3 · ρ · (1 2ν) − With: c0 = the speed of sound inside the base material [m/s] k = a constant [-]

It should be noted that also the flow velocity of the flyer plate should be kept lower than the local speed of sound to allow a jet to be formed. This condition can only be taken into Chapter 3. Explosive welding 31 account when the welding window is put in function of the impact velocity. The collision point should then be seen as a fixed point, and the workpieces will have a velocity as is shown in figure 3.3. In that case the condition for the formation of the jet will be as following: [18]

−1 vc α = tan ( ) if u = c1, vcp < c2 (3.5a) c1 −1 vc α = sin ( ) if u < c1, vcp = c2 (3.5b) c2 With: c1 = the speed of sound inside the flyer material [m/s] c2 = the speed of sound inside the base material [m/s] u = the flyer plate flow velocity [m/s]

Boundaries 3 and 4 These boundaries relate to the value of the impact angle that is required to induce jet formation along the surface of the materials.

Boundaries 5 The physical meaning of this boundary was not found during the literature study.

Boundary 7 The upper boundary for the formation of a weld is an experimentally found relationship between the impact velocity and impact angle and is given by: [34]

α k3 sin( ) = 0.25 · 1.25 (3.6) 2 t vc With: ρ = the density of the weakest material [kg/m3] H = the Vickers hardness of the weakest material [MPa] q E k3 = 12 · ρ · (1−2ν) [-] E = Young’s modulus [MPa] ν = Poisson’s ratio [-]

3.5 Wave formation

The formation of waves is a different physical process than the formation of a weld, although their parameters partly overlap. For the combination of similar metals, a bigger difference between the ultimate strength σt and the yield strength σy leads to an increase of the interval for waveless welding. For dissimilar welds, the interval of vc and α for waveless welding increases with increasing difference between the materials’ strength properties, and more in particular their hardness. When trying to obtain a wavy interface in material combinations with a far different strength (for example aluminum and high strength steel), the welding Chapter 3. Explosive welding 32 window would become small and the range of parameters to use would become narrow. Ad- justing the process parameters to meet the conditions of this small welding window is rather difficult, as the impact velocity and angle cannot be predicted precisely. This could result in a velocity and an angle outside of the welding window, but not in the zone which results in a waveless weld. In other words, for material combinations that are do not tend to form a wavy interface easily, narrowing the range of parameters down to the wavy welding zone could result in weaker welds. [30]

At the collision point where the plates collide, the high impact velocity causes severe plastic deformation and high shear stresses. Numerical models show that the highest values of the velocity occur at the collision point (figure 3.5(b)). This high-velocity oblique impact causes intensive shearing deformation (due to the relative sliding of the plates) (figure 3.5(e)), cumulative jets to be formed and the preboundary layers of the materials to be strongly heated (figure 3.5(d)). The absolute value of shear stress in the flyer plate is higher than in the base plate and this is the reason for the concave wave produced in this flyer plate (figure 3.5(h)). The metal behavior in the area around the contact point can be described from the hydrodynamic point of view. In addition, the compressibility of the metal is very small under the high pressures that occur at the collision point. Due to the periodic disturbance of the velocity profile and material deformation, a wavy interface is created. These periodic disturbances are induced by the oblique collision. [35, 31, 32]

Interfacial waves are the most discussed aspect of explosive welding. Numerical models show that wave formation is the result of variations in the velocity distribution at the collision point and periodic disturbances of the materials. While the former causes the waves to be formed behind the collision zone, the latter causes wave formation ahead of the collision zone. The weld interfaces often contain vortices following the wave peaks. [35]

The evidence that the variations in the velocity distribution result in a wavy interface, are the simulations that show the creation of surface waves ahead of the collision point. This supports the stress wave mechanism of wave formation that was proposed by El-Sobky and Blazynski(1975), who investigated the problem of surface disturbances caused by successive interference from rarefaction 2 waves in both plates. In this mechanism, the waves were generated in front of the collision point, while the vortices are created subsequently. [35]

In the same simulations, velocity discontinuities were noticed across the surfaces of the plates behind the collision zone. This supports the Robinson theory of wave formation, which states that the waves are an example of Kelvin - Helmholtz instability. This occurs when there is a diﬀerence of velocity between adjoining streams, similar to e.g. wind blowing along a water surface and creating waves. Hereby it is assumed that for suﬃciently high values of the shear

2Decompression. A decrease in density and pressure in a medium caused by the passage of a sound wave. Chapter 3. Explosive welding 33 strain rate, the metals act as inviscid ﬂuids. Below this critical strain rate, the metals act as viscid ﬂuids, and this viscosity removes the instability. Consequently, no wave formation will occur. According this theory, the waves are created behind the collision zone. [35, 33]

(a) Material distribution (b) Velocity distribution

(e) Plastic strain distribution (f) Plastic strain rate distribution

(g) Normal stress distribution (h) Shear stress distribution

Figure 3.5: Results from a simulation of two 6 mm thick stainless steel plates impacting at a velocity

of 650 m/s and an inclination of 15◦ [35]

The transition from a straight to a wavy interface appears to be related to an increase in the plastic strain and shear stress. Higher plastic strain and shear stress correspond with a more wavy interface. As was seen in the welding windows in chapter 3.4, this can be reached at a Chapter 3. Explosive welding 34 high enough impact velocity. The wavelength along the interface is in general not uniform. More shallow waves appear at the beginning of the bond, while deeper waves occur at the end. The results suggest that the change of the wavelength was due to changes in the impact angle along the interface. The wave shape has also been found to be perfectly symmetric across the interface for metals with similar density, and to become more asymmetric as the density diﬀerence increases. [35]

3.6 Variation of wavy interface with impact velocity

An increase of the velocity vc in the subsonic regime leads to an increase of loading pressure and elastic shearing deformation in the contact zone. The mass of metals entrapped into a jet flux will be greater and the disturbance amplitude (i.e. the amplitude of the wave) will increase. This can be seen in figure 3.6, where the relation between the amplitude of the wave as a function of the relative velocity vc , with c the speed of sound in the material, is shown. c0 0 This relation is calculated using a dynamic yield strength, which characterises the ability of a material to resist shear deformations. At the critical velocity vmax for individual metal pairs, characterising a transition from subsonic to supersonic flux, the wave amplitude reaches its maximum value. A further velocity vc increase is accompanied by a decrease of the wave amplitude, as can be seen in the decreasing branch on figure 3.6. In this oblique impact regime the waves are formed under Kelvin - Helmholtz instability development, as it was explained in the previous section. This results in wave formation on a separation boundary between the two metals. With this increase of the velocity, the velocity of the relative metal sliding increases as well, which reduces the contact time. The waves fail to form completely and their amplitude decreases. This increase can also cause the rarefaction wave separating the plates to tear the welded joint. A certain part of the flyer metal (up to 0.1 mm in thickness) remains attached to the fixed plate surface in the form of a porous layer. [31]

In the subsonic regime, the ascending branch in figure 3.6, the wave amplitude varies as fol- 3 lows. At contact point velocities , vcp < 1500m/s, on a separation boundary of two arbitrary metals the waves are not formed. In this case the loading pressure is not high enough. At vcp > 1500m/s, on a contact boundary, symmetric waves occur. A subsequent velocity growth results in the increase of loading pressure and elastic shearing deformation intensity in the contact zone. The mass of metals entrapped into a jet flux is greater and the wave amplitude increases. With a further increase, the waves lose symmetry, create vortices and the wave amplitude increases. With a further velocity increase, transformation of the contact boundary geometry is seen. It becomes a layer of turbulently mixed materials, instead of being wavy (see also figure 3.1). Further increase leads to velocity values in the supersonic regime. [31]

3The contact point velocity is the speed at which the contact point between the materials moves. This is the horizontal component of the impact velocity, and has a higher value than the vertical impact velocity. Chapter 3. Explosive welding 35

Figure 3.6: Generalised relation between the relative impact velocity and the amplitude of the wave disturbance for 3 metals (aluminum, copper and steel) at a constant impact angle. [36]

Similarly to the increase of wave amplitude, also the wavelength and hardness of the material will increase. The hardness increase is due to the higher shock induced and the deformation caused by impact. [28, 29, 27]

3.7 Variation of wavy interface with impact angle

Figure 3.7: Increasing wave amplitude along the weld for stainless steel plate (3 mm) to steel plate at impact velocity 584 m/s. [32]

The collision angle is not a constant, but increases with weld propagation. As the collision point moves along the plate the impact angle increases due to initial plastic hinging. In other words, the impact angle is dynamic. This changing impact angle determines the conditions for deformation near the point of impact as the smaller collision angles result in a higher impact pressure and higher shear strain rate, which is required for a suﬃcient deformation to induce bonding. Due to this change in pressure, the change in collision angle is directly responsible for the change in interface morphology from wavy to smooth at the welding front. Moreover, the Chapter 3. Explosive welding 36

Figure 3.8: Eﬀect of the collision angle α on the amplitude a and the wavelength λ of the wavy weld interface. [32]

wavelength and amplitude of the waves will change gradually along the direction of welding. An example of such an increasing wave is shown in ﬁgure 3.7 for the combination of a stainless steel plate to steel plate. [33, 35, 34]

The variation of the wave amplitude is generally as shown in ﬁgure 3.8. There is a critical

minimal collision angle αmin for the forming of waves, and there is a maximum value αmax after which no jetting will take place and no bond will occur. [32] states that within this range of angles, the amplitude of the waves will become larger with increasing angles up to a

certain value, and thereafter start decreasing until a certain value αw < αmax. For collision

angles between αw and αmax the interface will be smooth. Hence, the amplitude of the waves is strongly dependent on the collision angle. [32]

The workpieces are often welded under a certain initial angle. A possible configuration to do so can be seen in figure 3.9. Because the impact angle in function of time varies for different values of the initial angle, there are clear differences between the wavelengths of the welds for the different values of the initial angle. The smaller the intitial angle, the more rapid will be the increase of the impact angle (figure 3.10) and the impact velocity. [34]

That the value of the impact angle is of importance for the quality of the weld, can be seen in its inﬂuence on the jet formation. In the numerical simulations for aluminum-aluminum in [34] it was noted that jets only form for initial angles between 4° and 15°. However, the jets are not created at the same time during experiments and they probably do not carry the Chapter 3. Explosive welding 37

Figure 3.9: Example of workpiece conﬁguration under initial angle. [34]

same quantity of matter. [34]

3.8 Other parameters that aﬀect the wavy interface

The conclusions in this section are valid for plates. Due to the more difficult deformation of tubes, the conclusions regarding these parameters might change for MPW as these parameters (strength characteristics, stand-off and thickness) also affect the deformation behaviour of the tubes. The conclusions regarding the impact itself that were described before however stay valid.

3.8.1 Strength characteristics

It has been found that the amplitude of the waves increases with decreasing strength characteristics (at equal impact velocity and a constant impact angle) and thus higher formability. [31]

3.8.2 Stand-oﬀ distance

Many references [28, 29, 27, 32] mention an increase of the wave amplitude with increasing stand-off distance between the workpieces relative to the flyer plate thickness and increasing explosive energy relative to the flyer plate weight. The experiments in these works were done with copper - stainless steel and aluminum - copper. This is similar to the general effect of an increasing wave amplitude with increasing impact speed, as both the increase in energy and the stand-off distance result in a higher impact speed. Stand-off distances are varied between 0.5 and 3 times the thickness of the flyer plate, and can all result in good welds. The hardness Chapter 3. Explosive welding 38

Figure 3.10: Variation of the impact angle in time for diﬀerent values of the initial angle. [34]

increase of the materials is said to reach a maximum at a stand-oﬀ equal to 2 times the ﬂyer plate thickness. [27]

3.8.3 Flyer plate thickness

An increase of the amplitude and wavelength of the interface profile with increasing flyer workpiece thickness is noted. [35] The higher weight of the flyer piece results in a higher impact energy, which has the same effect of an increasing impact velocity. From the welding windows suggested in [32] it can also be seen that the required impact velocity for thicker flyer pieces is lower, which agrees with a higher impact energy due to the increasing thickness. Apart from that, the thickness of the flyer plate also influences the dynamic impact angle. [34]

3.9 Conclusion

The eﬀects on the explosive welded bond can be related to the magnetic pulse welding process. The interface types that occur are the same for both processes. Similarly to explosive welding, this means that the parameters of the magnetic pulse welding process need to be optimized to result in wavy or smooth interfaces without intermetallic compounds.

The effect of the impact velocity and impact angle on the interface of the weld is of high importance. As there is no standard testing method yet to judge on the quality of the magnetic pulse weld, the welds will be examined through microscopic investigation and hardness tests. Knowledge of the effect of the impact velocity and impact angle allows to estimate the values of these parameters in different experiments relative to each other. This can give an indication of Chapter 3. Explosive welding 39 the effect of the initial parameters in the magnetic pulse welding (energy level and geometry), and more important this can help to optimize these initial parameters.

By comparing the values of impact velocity and impact angle from the welding windows for different material combinations in explosive welding, it should be possible to find an estimation for the initial parameters in the magnetic pulse welding experiments when trying different material combinations. Chapter 4

A more profound view on the magnetic pulse welding process

The magnetic pulse welding process is the result of a complex combination of phenomena in electromagnetism, mechanics and heat transfer (figure 4.1). The current computer technology makes it possible to conduct a numerical analysis of the process, but up to now it is not possible to calculate any required process parameters manually. The effect of the different parameters is difficult to interpret, as they often influence the process through several physical phenomena.

Figure 4.1: Interaction between physical phenomena in EM welding [36]

As was discussed in chapter 3, the impact velocity and impact angle are the two important parameters in the forming of the weld. The initial process parameters will inﬂuence these 2 parameters and thus the formation of the weld. In the magnetic pulse welding machine, the only parameter that can be set is the charging voltage level of the capacitors. Based on the basic calculations from the manufacturer of the MPW system in [26], it is assumed that this voltage level is the only variable parameter that inﬂuences the impact velocity when the geometry is chosen. In this chapter a series of formulas is given, which can lead to a roughly estimated value of the required voltage level starting from the desired impact velocity. This will further be compared with the experimental results.

40 Chapter 4. Profound view on the MPW process 41

4.1 From start button to pulse

After aligning the workpieces in the coil, the only parameter which needs to be set before starting the process, is the required voltage in the capacitor bank. This voltage determines the energy that is stored in the capacitors through the following expression:

C · U 2 E = set (4.1) 2 With: E = stored energy [J] C = bank capacitance [µF]

Uset = the set voltage level [kV]

When the process is started, the capacitor bank is charged upto the required energy level. When this level is reached, a switch is triggered to close the discharging circuit and release the stored energy instantaneously through the coil. This current takes the form of a damped sine wave, which is typical for a LRC resonance circuit. It is only the ﬁrst peak of this current that is of importance in the deformation of the outer workpiece [37] and therefore only this value is considered in further calculations. The frequency of this current depends on the number of capacitors used [38]. The amplitude of this current depends on the chosen energy level. In the magnetic pulse machine used in this work, the capacitance equals 160 µF and the frequency of the current is 14.000 Hz. The peak current generated by a capacitor bank discharge, Imax, can be estimated from standard LRC equations for the primary circuit [26]: I Uset = q (4.2) CΣ · δ LΣ

With: I = the current peak in the coil [A]

CΣ = the total capacitance of the system [F]

LΣ = the total inductance of the system [H] δ = system attenuation coeﬃcient [-]

For the magnetic pulse welding machine used here, it is known that the maximum voltage is 25 kV and the corresponding maximum current is 500.000 A. Hence, the correlation between the peak current and voltage level is known, and this can be used further when calculating the required voltage level: I Uset = (4.3) 20

The coil induces a transient magnetic field around the workpiece. The magnetic field and the magnetic pressure caused by it are concentrated by a field shaper in the desired working Chapter 4. Profound view on the MPW process 42 zone. Several approaches have been proposed to calculate the magnetic field when knowing the peak current and the characteristics of the magnetic system. It is assumed that the value of the magnetic field is uniform everywhere between the coil and the flyer workpiece. The simplest relation between the current in a solenoid coil and the magnetic field it induces, is the one found through Ampère’slaw:

I · N · µ0 B = (4.4) l With: B = the magnetic ﬁeld inside the coil [T] N = number of windings [-] l = length of the coil [m] −7 µ0 = 4 · π · 10 H/m= permeability of free space

Figure 4.2: A configuration with field shaper: 1) the coil, 2) the field shaper and 3) the workpiece. The zone where the magnetic flux density is concentrated is marked in red. [13]

However, a magnetic field shaper is used to concentrate the magnetic flux in the desired welding zone (figure 4.2). As part of the energy is dissipated inside the field shaper, the above formula can’t be applied here. Although there is a certain energy loss, the field shaper still offers a significant advantage due to the concentration of the magnetic field. The field shaper used during the experiments in this thesis has a working zone of 15 mm and an inner diameter of 27 mm as can be seen in figure 4.6. In [39] a simple expression is suggested to express the correlation between the current and the magnetic field it induces:

B = K · I · f(Aeff ) (4.5) Chapter 4. Profound view on the MPW process 43

With: B = the magnetic ﬁeld inside the coil [T] K = constant depending on the geometrical, material and electrical characteristics [-] 2 f(Aeff ) = a function of the ﬁeld shaper length [H/m ]

The value of f(Aeff ) is presented as a function of the ratio of the length of the field shaper nodule (which is the length of the working zone) to the whole length of the coil (figure 4.3) as presented in [39]. The factor K can then be derived from the experimental results. However, different experiments will be done with a variety of materials and geometries. A new value for K would then be required for each of these cases. It would therefore be more convenient to use formulas that are a direct function of the geometrical, electrical and material characteristics. These formulas were provided by the manufacturer of the magnetic pulse welding system. [26]

Figure 4.3: The factor f(Aeff ) as a function of the length of the ﬁeld shaper. [39]

In these formulas the current in the coil is split in 2 components, being the current in the working zone iwz and the current at the ends of the coil iend. The current in the working zone accounts for the current that is concentrated in the ﬁeld shaper, while the current at the ends of the coil accounts for the energy loss in the edges of the ﬁeld shaper. The values of these currents are given through the following set of equations: Chapter 4. Profound view on the MPW process 44

I = iwz + 2 · iend (4.6a) φ · l s iwz (1 ) + 2 · ∆i (4.6b) ≈ 2 · π · µ0 · R · s − 2 · R φ ∆i = 2 (1 ln(2)) (4.6c) π · µ0 · R − φ π · R iend 2 ln( ) C0 + Ω1 (4.6d) ≈ π · µ0 · R 2 · s − With:

iwz = the current in the working zone of the ﬁeld shaper [A]

iend = the current in the end zone of the field shaper [A] ∆i = a correction factor for the current [A] φ = B · A = the magnetic flux [Wb] A = the surface between the coil or field shaper and the workpiece [m2] l = the length of the working zone [m] R = the outer radius of the flyer workpiece [m] s = the initial gap, or stand-off distance, between the workpieces [m]

C0 = 0.577 = Euler coeﬃcient

Ω1 = 1 = a constant

The length of the working zone is the length of the field shaper along which the current is concentrated. Hence it equals l = 15 mm. The internal diameter of the field shaper is 27 mm, but the insulation layer has a thickness of 1 mm. This leaves a diameter of 25 mm for the workpiece. The flyer workpieces in the experiments will always be chosen to match the outer diameter of 25 mm, so that the axial alignment of the pieces is assured. The surface between the field shaper and the workpiece will then equal A = π(R2 R2) = π(0, 0272 0, 0252) = coil − − 8, 17 · 10−5 m2.

With these equations it is possible to calculate the required current peak, when the required magnetic ﬁeld is known.

4.2 From pulse to pressure

A magnetic pressure is associated with a magnetic field. A gradient in the magnetic field strength (as is shown in figure 2.4) over the thickness of the workpiece then causes a radial force due to the gradient in magnetic pressure. Because of the difference of the magnetic field inside and outside of the workpiece, such a magnetic pressure will compress the flyer tube radially. This pressure will similarly to the magnetic field vary in time, and can be found Chapter 4. Profound view on the MPW process 45 with the following expression [40]:

1 2 2 P (t) = (Bgap(t) Bdiff (t) ) (4.7) 2 · µ0 − With: P = magnetic pressure [Pa]

Bgap = B = magnetic ﬁeld outside the workpiece [T]

Bdiff = magnetic ﬁeld diﬀused through the workpiece [T]

The value of the magnetic pressure can be expressed in terms of material properties instead

of the unknown magnetic ﬁeld inside the tube Bdiff [9]. First, the radial magnetic pressure per unit volume is calculated as the Lorentz force which interacts on a current carrying body that is exposed to a magnetic ﬁeld:

P = iΘ · Bgap (4.8)

With: P = magnetic pressure[N/m2] 2 iΘ = the current density in the workpiece [A/m ]

The value of iΘ and Bgap, which are both caused by the current in the solenoid, are calculated from Faraday’s law of induction, one of Maxwell’s equations, which describes how a changing magnetic field can create an electric field: E = ∂B . In a body with a certain electrical ∇ × − ∂t conductivity, the electric field intensity E leads to a current flow i = σ · E, with σ the electrical conductivity of the material. This leads to a new expression for the magnetic pressure on the workpiece, as a function of the material properties [9]: 2 · t B2 − P = ( 0 )(1 e δ ) (4.9) 2µ0 − With:

B0 = Bgap = the magnetic ﬁeld at the outer surface of the tube [T] t = the plate thickness [mm] δ = the skin depth [mm]

δ is calculated with: 1 δ = (4.10) √π · σ · µ · f With: σ = the electrical conductivity [m/Ω] µ = magnetic permeability of the workpiece [H/m] f = the frequency of the current[Hz] Chapter 4. Profound view on the MPW process 46

In formula 4.9, which is used in many works [8], it can be seen that the magnitude of the magnetic pressure is determined by the skin depth and the magnetic field outside of the tube. According to these formulas, the magnetic field will decline exponentially in the workpiece, as can be seen in figure 2.4. The magnetic pressure is then determined by the absolute value of the magnetic field outside the workpiece, and by the part of the field that is diffused through the workpiece, which is characterized by the skin depth. Hence, the same parameters which are of importance in magnetic shielding determine the value of the skin depth.

The formulas 4.9 and 4.10 can be used further to estimate the influence of other parameters. The physical meaning of these parameters is discussed in the brief description of the magnetic shielding theory. It is interesting to note that the material parameters also have an influence on the primary current in the coil and thus the absolute value of the magnetic field outside the workpiece. In [40] it is estimated that an increase of conductivity from 10, 0 MS/m to 17, 8 MS/m leads to an increase in magnetic pressure of 12%, only due to an increase of the magnetic field outside the workpiece.

4.3 From pressure to impact

The magnetic pressure has to be large enough to deform the part of the flyer workpiece in the working zone, and provide it with enough kinetic energy so that the impact with the inner workpiece causes a good bond. This leads to two components of the required pressure. In order to work with practically applicable formulas, it is assumed that the magnetic field is constant along the length of the welding zone, which is a reasonable assumption if the workpiece is placed close to the center of the coil or the field shaper when it is used.

4.3.1 Calculation of required pressure

The first component is the pressure that is necessary for the circumferential deformation of the tubular workpiece. This is calculated as a combination of the hoop and the longitudinal stresses. The pressure which is used in the theory of thin-walled cylinders is adjusted with a factor α [-] that ranges from 1 to 10 to take into account the inertia effect, the workpiece length, the workpiece anisotropy, strain and strain-rate hardening, and the through thickness t stresses when the workpiece is not thin-walled (i.e. R > 0, 1) [37]. This parameter α will be kept equal to 1 unless mentioned differently. This leads to the following expression:

2 · σy · t P1 = α · (4.11) R With: 2 P1 = pressure required for deformation [N/m ] α = factor for eﬀects that limit deformation [-] Chapter 4. Profound view on the MPW process 47

2 σy = the yield strength of the material [N/m ] t = thickness of the workpiece [m] R = radius of the workpiece [m]

The second part of the pressure is used to accelerate the working zone of the outer tube suﬃciently to cause bonding at impact with the inner part. Following the guidelines for the relation between impact velocity and the pressure proposed in [26], it is assumed that the velocity of the ﬂyer tube will increase linearly from 0 to the speed vc reached at the moment of impact. This means that the acceleration of the workpiece is assumed to be constant. The correctness of this assumption and an adjustment of the formulas presented here is given in section 4.3.2.

The speed at impact can be related to the gap between coil and workpiece, when using the acceleration for a linearly accelerated object: v2 02 s = c − (4.12) 2a With: s = the gap, or stand-oﬀ distance, between the two workpieces prior to welding [m] vc = the component of the impact speed perpendicular to the surface [m/s] a = the acceleration of the ﬂyer tube [m/s2]

The pressure can then be directly related to the velocity: · 2 m a vc P2 = = t · ρ · (4.13) A 2 · s With: 2 P2 = pressure required for acceleration [N/m ] m = mass per unit of length of the outer tube [kg/m] A = surface of the intersection of the ﬂyer tube per unit of length [m2/m] ρ = density [kg/m3] s = the gap, or stand-oﬀ distance, between the two workpieces prior to welding [m]

It is hereby assumed that the speed of the outer tube equals the speed perpendicular to the inner tube. In other words, the horizontal speed component along the inner tube is ignored. This assumption is correct if the impact angle doesn’t reach values over 10◦. The assumption that the acceleration is constant doesn’t correspond fully with reality, as the velocity will decrease after reaching a peak value. In case the tubes would impact after reaching the peak velocity, this would require a more complex relation between acceleration and velocity by taking into account the resistance to the movement. However, this does not aﬀect the conclusion that the pressure and impact velocity are in linear relation with each other. Chapter 4. Profound view on the MPW process 48

The total pressure required to impact the outer tube against the inner tube at a required velocity vc can then be estimated by the following formula: · · 2 2 σy t vc P = P1 + P2 = α · + t · ρ · (4.14) R 2 · h It has to be noted that the workpiece deformation increases the air gap between the field shaper and the workpiece, and therefore can affect the value of the magnetic flux density [39]. This is not taken into account in the formulas.

4.3.2 Check whether maximum velocity is reached

As soon as the yield stress is reached in the welding zone of the outer workpiece, it will start its movement towards the inner workpiece. Previously, it was assumed that the velocity increases upto the required velocity at impact, due to a constant acceleration during the whole movement until impact. The acceleration is in reality not constant, because the pressure applied on the tube is a pulse and not a constant load (figure 4.4), as it is directly related to the value of the magnetic pulse. The acceleration similarly will occur as a pulse acceleration (figure 4.5). The kinematic behavior of the outer workpiece will be far different from the behaviour with the constant acceleration that was proposed. This simplification could be disadvantageous if the actual impact speed that is reached doesn’t meet the impact speed that was proposed. Apart from that, the flyer workpiece will deform more while being compressed, inertia forces have not been taken into account and the part of the workpiece that doesn’t deform might offer resistance to the movement. Although the importance of these factors might be negligible due to the high speed nature of the process, they could also lead to a decrease of the energy and velocity of the workpiece. In order to get an idea whether the assumption of a constant acceleration might lead to an overestimation of the impact speed, a check will be made based upon the time that the flyer workpiece deforms.

The required acceleration is associated with the maximum value of the pressure P1 + P2. T 1 With a current frequency of 14.000 Hz it is reached at time t = 4 = f · 4 = 17, 8 µs when the maximum pulse pressure is reached. The workpiece starts accelerating at the time tdeform when the pressure reaches a value equal to the pressure required for deformation P1 (figure 4.4). This time is derived by assuming a quadratic sinusoidal pressure pulse, as it is related to the square of the magnetic field: · 2 2 π P1 = sin ( · t ) · Ptot (4.15) T deform T The acceleration will then increase until its maximum value at t = 4 and decrease back until T it reaches 0 at time t = tdeform. The total time that the workpiece is accelerated then 2 − equals: (figure 4.4) T tacc = 2 · tdeform (4.16) 2 − Chapter 4. Profound view on the MPW process 49

Figure 4.4: The pressure pulse on a tubular workpiece in a coil. The workpiece will accelerate due to the arched part of the pressure pulse.

Assuming that the acceleration is constant, the ﬂyer workpiece will reach the inner workpiece

at time tflyer, given by: (ﬁgure 4.5) r 2 · s t = (4.17) flyer a A comparison between the required time to move under a constant acceleration, and the time actually being accelerated is made in table 4.1 for steel, high strength steel and aluminum at diﬀerent gap widths s and impact velocities vc. When looking at this table, it can be concluded that when the pressure is calculated based

on a smaller required velocity, the time that the workpiece is accelerated acc will be smaller than the time the workpiece should undergo a constant acceleration tflyer (marked in red in the table). An example of this situation is shown in figure 4.5. In these cases the flyer workpiece will never reach the desired impact velocity with the pressure that is calculated in the formulas given previously. This is first because the average acceleration will be lower than the one assumed, and second because the acceleration time is shorter than the time required. However, this risk only exists when trying to reach lower impact velocities. In the cases at higher required impact velocities, the acceleration time is long enough, but as the average acceleration is lower than the one required it is still likely to occur that the desired impact velocity won’t be reached.

Assuming that the relation between the applied pressure and the set voltage level is perfectly known, it is important to obtain a good estimate of the relation between impact velocity and Chapter 4. Profound view on the MPW process 50

Table 4.1: Comparison of acceleration time and ﬂyer time

Material σy ρ s vc tacc tfly N/mm2 kg/m3 mm m/s µs µs High strength steel 690 7850 2 200 19,8 20,0 400 26,7 10,0 600 29,5 6,7 3 200 17,5 30,0 400 24,9 15,0 600 28,2 10,0 Steel 255 7850 2 200 24,9 20,0 400 30,0 10,0 600 31,9 6,7 3 200 23,0 30,0 400 28,8 15,0 600 31,0 10,0 Aluminum 120 2700 2 200 23,5 20,0 400 29,1 10,0 600 39,3 6,7 3 200 21,4 30,0 400 27,7 15,0 600 30,3 10,0 Chapter 4. Profound view on the MPW process 51

Figure 4.5: A comparison of the constant acceleration that is assumed in the original formulas, and the actual acceleration that will occur due to the magnetic pressure. It is obvious that the assumed acceleration is an overestimation of the actual acceleration. Therefore the magnetic pressure should be higher to accelerate the workpiece suﬃciently.

pressure in order to use values of the impact velocity from explosive welding experiments. Working with the actual varying value of the acceleration would complicate these rather easy to use formulas too much. Therefore, the following two adjustments are made to the above formulas:

1. Applying a factor 2 to the pressure required for the acceleration that is given in formula 4.13. This in order to take in account that the pressure is calculated based on the maximum acceleration, while the average acceleration is half of this maximum value (based on a quadratic sinusoidal function).

2. The condition tacc tflyer, with tacc and tflyer as given in equations 4.16 and 4.17 ≥ respectively, should be true.

4.4 From impact to weld

When the workpieces impact at a velocity vc, a pressure will originate at their interface. The impact pressure for two semi-inﬁnite elastic bodies equals [15]:

ρ1 · ρ2 · c1 · c2 Pinterface = · vc (4.18) ρ1 · c1 + ρ2 · c2 With:

Pinterface = the pressure at the interface of the workpieces during impact [MPa] 3 ρ1 = density of the ﬂyer workpiece [kg/m ] 3 ρ2 = density of the stationary workpiece [kg/m ]

c1 = longitudinal wave speed of the ﬂyer workpiece [m/s] Chapter 4. Profound view on the MPW process 52

c2 = longitudinal wave speed of the stationary workpiece [m/s] vc = impact velocity [m/s] And the longitudinal wave speed is a material characteristic given by: s 3 · K · (1 ν) c = − (4.19) ρ · (1 + ν)

With: c = longitudinal wave speed [m/s] ρ = density [kg/m3] K = bulk modulus, which measures the substance’s resistance to uniform compression [Pa] ν = Poisson’s ratio [-]

The high values of pressure that can be reached through impact are the reason why the workpieces can bind at lower temperatures. For example, aluminum - aluminum impacting at 50 m/s generates an impact pressure of 680 MPa. However, this impact pressure won’t be discussed further, as the required impact velocity can be found directly using results from explosive welding experiments.

4.5 Use of the formulas

The above formulas can now be used to calculate the required voltage level, starting from an impact velocity that is used in explosive welding experiments. Due to the simplifications and assumptions made when composing this set of formulas, it is not known whether they are fully accurate. The voltage level found through these formulas will be compared with the minimum voltage level found in the experiments for several geometries and material combinations, in order to test their accuracy. It has to be noted that the geometry and the placement of the workpieces inside the field shaper are also of crucial importance, as they will influence the impact angle. This angle is of major importance in the forming of a weld, as was discussed in the previous chapter.

Apart from that, applying these formulas can give a good insight of the inﬂuence on the required voltage level when changing the material or the geometrical properties. As a closure of this chapter, a summary is given of the formulas that are required to use.

1. The chosen value of the impact velocity gives the total pressure required. Note that a factor 2 is used in the second term, according to the discussion in section 4.3.2. · · 2 2 σy t vc P = P1 + P2 = α · + 2 · t · ρ · (4.20) R 2 · s

2. The condition tacc tflyer needs to be fulﬁlled. ≥ Chapter 4. Profound view on the MPW process 53

3. The required pressure leads to a value of the magnetic ﬁeld: 2 · t B2 − P = ( 0 )(1 e δ ) (4.21) 2µ0 −

4. The magnetic ﬁeld density leads to a required current peak in the coil:

I = iwz + 2 · iend (4.22a) φ · l s iwz (1 ) + 2 · ∆i (4.22b) ≈ 2 · π · µ0 · R · s − 2 · R φ ∆i = 2 (1 ln(2)) (4.22c) π · µ0 · R − φ π · R iend 2 ln( ) C0 + Ω1 (4.22d) ≈ π · µ0 · R 2 · s −

5. And ﬁnally the required voltage level is easily determined through this current peak: I Uset = (4.23) 20 Chapter 4. Profound view on the MPW process 54

Figure 4.6: The ﬁeld shaper that is used during the experiments Chapter 5

Process parameters

Just like in explosive welding, the quality of the MP weld is dependent on the impact angle and the impact velocity during coalescence. Numerous process parameters influence these 2 parameters in some way. These include properties of the materials, the electrical properties of the magnetic pulse welding machine, and the geometry of workpieces and field shaper. The majority of these parameters will be invariable throughout the experiments, either because they are inherent to the magnetic pulse welding equipment, or because they are chosen to be kept constant. In this chapter first the impact velocity and angle are discussed, followed by an overview of parameters that will be varied during the experiments, and the way they relate to the impact velocity and angle. This chapter can be seen as an introduction to the experiments that are discussed in the following chapter.

5.1 The impact welding parameters

5.1.1 Impact velocity

The impact velocity is of crucial importance in the coalescence of the 2 workpieces, as it is this kinetic energy that is transformed to the energy used for bonding of the two materials. The impact velocity is directly related to the pressure that is required to move the flyer workpiece, and hence to the energy level used when conducting the weld. The geometry of the workpiece will also have its influence on the impact velocity, just like the material characteristics. The formulas given in chapter 4 cover the influence of these parameters, and so they will be used to link the impact velocity with the energy level.

The impact velocity is important in that way that it needs to ensure a high enough pressure between the two materials, in order to allow joining. The conclusions regarding the bonding process in explosive welding in chapter 3 learn that if the velocity is too low, the impact energy is insuﬃcient to initiate bonding. On the other hand, when the velocity is too high and reaches supersonic values, no jet force will occur and the materials won’t bind. Because

55 Chapter 5. Process parameters 56 the welding zone is not accessible and the occurring velocities are very high, it is in this thesis impossible to measure the velocities during the experiments. No exact values of the impact velocity are known, but this parameter will be used to help understand the inﬂuence of other parameters.

Figure 5.1: The backﬂow of material during impact welding when viewing the proces with the impact point as a ﬁxed point. [31]

5.1.2 Impact angle

The angle between the two impacting surfaces is the second crucial parameter. This parameter will depend on the geometry of the workpieces. As it was explained in chapter 3, the angle needs to be in a certain range in order to induce the jet along the surfaces. A backflow of material will then move along the surface ahead of the collision point (figure 5.1), deform the microroughness and damage the inert surface layer so that the surface is prepared for interaction. If the workpieces would impact with an angle out of this range, the surface wouldn’t be prepared by the backflow and consequently there would be no bonding.

Judging the influence of this parameter quantitatively is almost impossible. At first, the angle of incidence varies during the welding process [34]. In explosive welding, these variations are within a small range, and the angle is therefore in calculations usually assumed constant. It is not known so far how the angle varies in MPW. Secondly, the angle is impossible to measure as this would require images of the two surfaces during the welding process. Due to the speed and inaccessibility of the process these are simply impossible to capture. An indirect way is to calculate the angle through the horizontal and vertical component of the velocity. But as these are also impossible to measure, the exact value of the angle during the process is an unknown. However, this parameter will be useful in judging the influence of other parameters. Chapter 5. Process parameters 57

5.2 Variable parameters

5.2.1 Material properties

Magnetic permeability

The permeability of a material defines the degree of magnetisation. For all diamagnetic or · −7 H paramagnetic metals the value equals the permeability of air µ0 = 4π 10 m . For ferromagnetic metals, the value will be a factor 10 to 10.000 higher, depending on the magnetic state. Most materials that are commonly used (e.g. aluminum, copper, magnesium, titanium) are dia- or paramagnetic, the only ferromagnetic material that is of any importance is iron. Therefore steel and stainless steel have a much larger permeability and thus smaller skin depth. In this case, the material will conduct the magnetic field better, which prevents it from diffusing through it.

To understand the inﬂuence of the magnetic permeability of the material, it is necessary to take a closer look at the skin depth. As described in section 4.2, the skin depth is given by:

1 δ = (5.1) √πσµf The magnitude of the alternating eddy currents in the workpiece will be highest at the surface, and will decline exponentially further inwards the material. The skin depth is the depth at which the amplitude of the eddy current falls to 1/e of its original value. In order to maximize the influence of the eddy currents, the skin depth should be small enough so that the current is concentrated at the interface and the magnetic field doesn’t diffuse through the workpiece. In order to maximize the magnetic pressure the skin depth should be small relative to the workpiece thickness, as could be seen in formula 4.9. When the thickness of the tube is the same as the skin depth, the pressure equals 86% of its maximum value. When the thickness of the tube is twice the skin depth, the pressure reaches 98% of its maximum value.

Generally, the thickness of the workpieces used in this thesis is sufficient to prevent the magnetic field from diffusing through of the workpiece. For example for aluminum (σ = 37, 7 · 106/mΩ and µ = 1, 257 · 10−6H/m) and a frequency of 14.000 Hz, the skin depth is 0.69 mm. For steel with a low magnetic permeability (σ = 6, 0 · 106/mΩ and µ = 12, 57 · 10−6H/m), the skin depth is 0, 57 mm. As we will be dealing with thicknesses of 1, 5 mm and 2, 0 mm, the magnetic pressure will always be close to its maximum value. When using very thin sheets (< 1, 0 mm) it is likely that a great share of the magnetic field is diffused, the only way to reach a skin depth smaller than the thickness of the material in that case is by adjusting the frequency of the current [8]. This might be necessary when a thin conductive driver is wrapped around a less conductive workpiece. Hence this parameter is, together with the skin depth, of far less importance than the other parameters in the experiments in this thesis. Chapter 5. Process parameters 58

Electrical conductivity

The electrical conductivity of the flyer workpiece is expressed in m−1Ω−1. Its only effect on the process that is known and considered here is its influence on the magnitude of the eddy currents in the workpiece. This influence is expressed in the skin depth. A higher conductivity means that the induced electrical currents will be greater in value, and thus cause a greater opposing magnetic field. Consequently the magnetic pressure will be larger, which is important to cause bonding.

In table 5.1, some values of the electrical conductivity are given for several metals and metal alloys. The influence of the conductivity will now be discussed considering the value of the skin depth. These are calculated in the assumption that the current’s frequency equals 14.000 Hz and that the materials are non-magnetic. For aluminum and especially copper, no problems are to be expected due to the conductivity. As mentioned in the previous paragraph, for aluminum the skin depth is only 0, 69 mm and the magnetic pressure will still reach over 94% of its maximum level for thicknesses of the flyer tube 1.0 mm. For copper the skin depth ≥ will even be less critical: 0, 55 mm. However, when using alloys like Inconel (non-magnetic) or stainless steel 1 as the flyer material, the skin depth can be up to a factor 7 larger than the one for copper. The magnetic pressure is then reduced to 40% of its maximum value for a thickness of 1,.0 mm.

Table 5.1: Electrical conductivity of materials

Material Electrical conductivity. · 106/mΩ Inconel 1,0 - 3,0 Stainless steel 2,0 - 3,0 Carbon and low-alloy steel 8,0 - 12,0 Iron 9,9 Nickel 14,3 Tungsten 18,9 Aluminum 37,7 Copper 59,4

From these basic calculations it can be seen that using low conductive alloys as flyer material can significantly reduce the magnetic pressure, and thus the impact velocity. In this thesis, only copper and aluminum will be used as flyer material to make sure welding is not prevented due to low conductivities. Moreover, these 2 materials are the most important in practical

1non-magnetic in case nickel is added to stabilize the austenite structure of iron Chapter 5. Process parameters 59 applications of the magnetic pulse welding process. Hence, the inﬂuence of the electrical conductivity will be rather low for the experiments done in this thesis.

Mechanical properties

The mechanical properties of the workpiece are of importance mainly in the deformation of the material. The yield strength σy [MPa] is directly related to the pressure required on the workpiece, as can be seen in formula 4.14.

Strain hardening and strain rate hardening are two other important material characteristics that inﬂuence the formability of the workpiece. It is well known that the fundamental constitutive behavior (stress, strain, strain-rate relations) for most metals change qualitatively at strain rates above about 104 s−1. Above these strain rates, the apparent strain rate sensitivity of the material increases markedly. [15] The strain and strain rate hardening can be included into the material constitutive law. The Johnson-Cook model for yield stress expresses these inﬂuences:

· n · · · θ θtrans m σpl = [A + B pl] [1 + C ln˙p] [1 − ] (5.2) − θmelt θtrans − With:

σpl = von Mises yield stress [MPa] −1 ˙p = the plastic strain rate for ˙0 = 1.0 /s [s ]

pl = the equivalent plastic strain [-] A, B, n = yield and strain hardening constants [-] C = strain rate constant [-] m = thermal softening constant [-] θ = absolute temperature when the stress is applied [K]

θtrans = transition temperature deﬁned as the one at or below which there is no temperature dependence on the expression of the yield [K]

θmelt = temperature when the stress is applied [K]

The first 2 factors comprised in this equation express the influence of strain and strain rate hardening on the yield stress, while the third factor expresses the effect of the temperature. The material dependent parameters A, B, C, n, m are determined from straining tests. At the melting temperature, the stress approaches zero for all strains and strain rates. [35]

Also the elastic modulus will be of importance in the deformation of the outer tube. The tube will be compressed to a smaller diameter, which includes a big deformation along its circumference. For example for a stand-oﬀ distance of 2, 5 mm the diameter of the outer tube changes from 25 mm to 20 mm. The circumference decreases in that case from 79 mm to 63 mm, which might cause large circumferential stresses even though the material can Chapter 5. Process parameters 60 expand in the radial and longitudinal direction. These stresses might partly remain in the material as residual stresses after the weld is achieved. The strains due to this circumferential compression will cause higher residual stresses for materials with a higher elastic modulus, as can be seen from Hooke’s law stress-strain relationship σ = E · that is valid in the elastic region. The residual stresses in a copper tube (E = 120 GPa) will be higher than the residual stresses in an aluminum tube (E = 69 GPa).

Another important mechanical property is the elongation at break factor. This parameter deﬁnes the ductility of the material. Brittle materials or materials with low ductility can be cracked or fractured during the process.

The only strength parameter that is explicitly taken into account in this thesis is the yield stress that is used to calculate the required magnetic pressure to deform the flyer workpiece. However, this part of the pressure is much smaller than the part required to accelerate the workpiece and hence the influence of the yield strength seems to be much smaller than for example the density. Other properties can be taken into account to explain different experimental results for different materials.

Density

The density ρ [kg/m3] of the ﬂyer workpiece is of importance as heavier materials require a higher pressure to accelerate them to reach the required impact velocity. As can be seen in formula 4.13, the required pressure is in direct relation to the density.

However, the eﬀect of the density is less signiﬁcant than it seems to be expected from this formula. Due to the higher mass of thicker workpieces, their kinetic energy at impact will also be higher. It is the kinetic energy which is responsible for the severe deformation of the workpieces and the forming of the jet. As this kinetic energy is also dependent on the impact velocity of the workpiece, the velocity can be smaller and still result in a good bond. Hence, heavier workpieces require a lower impact velocity, as their kinetic energy is higher due to this higher mass.

These both effects (higher required pressure and lower required impact velocity) oppose each other, the influence of the density on the process is expected to be insignificant.

Thermal conductivity

The thermal conductivity of the materials has no influence on the welding process itself, but it can influence the interlayer between the two materials. During the process, the eddy i2 currents will generate a Joule heat proportional to σ in the flyer workpiece, with σ being the thermal conductivity. More importantly the severe plastic deformation due to the impact and interface wave creation and especially severe deformation of the metal surfaces during the jet Chapter 5. Process parameters 61 action will cause heat. Also, the exothermal creation reaction of the intermetallic layer is a source of heat. This last effect however is additional after the material has already molten. [41]

This heat needs to dissipate through the material [23]. When the thermal conductivity of the materials is not suﬃcient, this heat will cause a temperature rise at the interface, as the density of the currents is biggest here, which will result into local melting of the least conductive material. Melting has already been found as one of the bonding mechanisms [5], as the heat resulting from the plastic deformation of the area adjacent to the interface and the deformation of the jet increase the interface temperature much above the melting temperature2. [41] The lower the thermal conductivity, the more narrow the molten interface. [29] Melting and subsequent solidiﬁcation results in intermetallic compounds that have a higher hardness, but increase the cracking susceptibility. In general, they decrease the quality of the weld. [27]

Because the process takes place in about 100 µs, the heating takes place only for this short time. Therefore, the melting is limited to the boundary layer and was found mainly in the middle part of the welding zone [17]. Previous MPW experiments were successfully conducted with stainless steel [2] and titanium [18], materials with thermal conductivities as low as 12 W/mK. Melting is most likely a negative eﬀect for the forming of the bond, but can be avoided when the energy of the system is not chosen too high. [41]

With very thin materials, which might be the case when using a conductive driver sheet, the heat can’t spread over a big enough volume and consequently the temperature may increase too much and the driver could vaporize. This eﬀect is experienced during experiments with tube forming [42].

5.2.2 Geometrical properties

Stand-oﬀ distance

The standoﬀ distance is the distance between the workpieces that are to be welded prior to discharge. The standoﬀ distance, or gap, should be high enough so that the outer workpiece will have the time to accelerate upto the desired velocity.

If the stand-off distance becomes too high, the velocity might start decreasing again. This effect was noticed in [8], where the standoff distance has one optimum value, which gives the maximum tensile shear strength of the weld. This maximum tensile shear strength corresponds with the maximum kinetic energy, or maximum velocity that the workpiece reaches

2This eﬀect was described for aluminum - magnesium experiments, which are materials with a rather high thermal conductivity: aluminum = 220 W/mK, magnesium = 160 W/mK Chapter 5. Process parameters 62

at impact. This effect corresponds with the condition tacc tflyer that was proposed in the ≥ formulas in chapter 4. This condition expresses that the acceleration should last as long as the workpieces haven’t impacted yet, otherwise the impact velocity could decrease. At higher stand-off distances the minimum impact velocity to meet this condition will be higher. For aluminum workpieces of 1, 5 mm thick, this minimum impact velocity increases from 370 m/s to 500 m/s for an increase of the stand-off from 2, 5 mm to 3, 5 mm. This means that increasing the stand-off doesn’t automatically correspond with a lower energy level to meet the required impact velocity. Because at higher stand-off distances the outer tube will spend a longer time moving towards the inner tube, it is possible that the pressure peak passes before the outer workpiece reaches the inner, and its velocity starts decreasing again. It is therefore not sure that the energy level can be chosen lower.

It is clear that at a given energy level the standoff distance is a crucial parameter for the value of the impact velocity and thus the quality of the weld. A higher standoff gives the workpiece the chance to accelerate during a longer time before impact, and thus reach higher impact velocities at the same energy level. This effect is shown in figure 5.8. However, because it will take longer until impact, the pressure peak might have passed before the impact and the velocity might start decreasing again. As it is very difficult to quantify these effects, the total effect of changing the stand-off distance will be researched in the experiments.

Apart from its influence on the velocity, it will also require more energy to deform the workpiece. The material needs to be compressed from an inner diameter di to an inner diameter di 2 · s, with s the stand-off distance. With greater stand-off, the material will need to − be compressed more and the energy required to do this will increase. Clearly, this effect is a disadvantage due to the tubular shape of the workpieces. This effect is not taken into account in the formulas of chapter 4.

Shape of workpieces

The shape of the workpieces in this thesis is limited to tubular forms, as this is the easiest geometry to join. In practical applications, it would be most convenient to connect 2 simple tubes in order to prevent any special manufacturing operations on the workpieces prior to the welding. However, the eventual configuration of the workpieces in the experiments will differ from the configuration with 2 tubes (figure 5.2).

Due to the high force of the impact, there is a risk that the inner workpiece would deform. Therefore, a mandrel can be used inside the inner tube to prevent this deformation. In experiments it is easier to work with full cylindrical parts as inner workpieces.

Apart from that, also the geometry of the magnetic pulse welding machine should be taken into account. The inner workpiece will be placed in a holder that serves to push the workpiece Chapter 5. Process parameters 63

Figure 5.2: The difference between an easy practical configuration of the workpieces, and the experimental configuration.

Figure 5.3: The different geometries of inner workpieces that are used in the experiments of this thesis: a)slant configuration b)straight configuration c)configuration without collar

inside the ﬁeld shaper. This holder has a diameter of 16 mm and a length of 31 mm and hence the end part of the inner workpiece should be made to ﬁt this shape.

If the workpieces wouldn’t be concentric, part of the circumference of the outer tube would have to cover a bigger distance towards the inner workpiece. Consequently, this part would also have to deform more. The impact velocity would then differ over the circumference of the workpieces, and the quality of the weld would not be the same everywhere. Clearly this has to be avoided. In the experiments in this thesis, the outer workpieces are made to fit the insulation inside the field shaper and thus the concentricity of outer workpiece and field shaper is guaranteed. To ensure the concentricity of the inner workpiece with the outer, the inner workpiece is designed with a collar, as can be seen in figures 5.3 a) and b). If there is no easier way to ensure the geometrical positioning of the tubes, it is possible to use a positioning and insulating plug. This plug maintains the tubes’ mutual concentricity, the standoff distance and isolates the tubes from the coil. High density and high molecular weight polymers are best used as material for these plugs. [43]

The zone between the collar and the holder is the zone where the workpieces will impact, as can be seen in the experimental layout on ﬁgure 5.4. This zone can be designed straight, or under a certain angle. With a straight interface, the impact velocity is expected to be practically constant along the weld. The impact angle will increase towards the end of the weld as was Chapter 5. Process parameters 64

Figure 5.4: The positioning of the workpieces inside the ﬁeld shaper.

shown in chapter 3.7. Putting the workpiece under a certain initial angle both influences the impact angle as the impact velocity. The configuration as shown in figure 5.3 a) decreases the impact angle in comparison with the straight surface. The opposite configuration is also possible in order to increase the process’ impact angle. The impact velocity will now vary along the length of the surface, as the standoff distance is now a variable as well. Both these configurations, with straight and slant surface will be used in the experiments.

As it is easiest for the design to keep the shape of the workpieces as simple as possible, conﬁ- gurations without the collar, as shown in ﬁgure 5.3 c) will be tested as well. The concentricity might in this case be ensured by the clamping in the holder.

Thickness outer workpiece

An increase in the thickness of the outer tube means that the formability of the piece decreases, and it will have a bigger mass that needs to be accelerated towards the inner workpiece. Both the pressure required for the deformation of the outer tube and the pressure required for the acceleration of the deformed part will increase as both these pressures are directly related to the thickness. This can be seen from the formulas 4.11 and 4.13. When the energy level and Chapter 5. Process parameters 65 hence the magnetic pressure is chosen constant, the impact velocity will decrease for higher thicknesses of the outer tube. This effect is graphically shown in figure 5.5 for experiments with aluminum flyer tubes with a stand-off distance 2, 5 mm and a voltage level 15 kV, and with a stand-off distance 3, 5 mm and a voltage level 12 kV. These values were calculated by using the formulas from chapter 4.

Figure 5.5: The inﬂuence of the thickness on the impact velocity.

From this graph it is seen that an increase of the thickness from 1, 0 mm to 2, 5 mm, decreases the impact velocity with about 30%. It has to be noted that this is just an indication of the importance of the thickness. It will also require more energy to compress the tube as the higher thickness causes higher circumferential stresses, but this eﬀect is not taken into account here. On the other hand, due to the higher mass, the same eﬀect as described for a higher density (see page 60) is valid: heavier workpieces require a lower impact velocity, as their kinetic energy is higher due to their higher mass.

As there are two eﬀects opposing each other, the total eﬀect of changing the thickness of the workpiece on the quality of the weld will be investigated in the experiments in this thesis.

Position ﬁeld shaper

The position of the field shaper above the workpiece plays an important role in the deformation of the outer tube and thus in the impact behavior. The two possible ways to place the field shaper are shown in figure 5.6. In the configuration with the field shaper entirely over the workpiece (figure 5.6 a), a central part of the tube will impact to the inner workpiece. From Chapter 5. Process parameters 66 here the weld will propagate to both sides. Also the jet will occur on both sides of the initial impact. In the configuration with the field shaper over the end of the outer tube (figure 5.6 b), the end of the tube will impact first, and the weld will propagate in one direction.

Figure 5.6: Diﬀerent possible positions of the ﬁeld shaper: [18] a) middle joint b) end joint

The deformation work will be lower when the field shaper is positioned over the end of the workpiece, as there is less material that has to be deformed. Hence, the end joint requires less energy than the middle joint. For this reason, this configuration will be used during the experiments in thesis, as could be seen before in the experimental outlay on figure 5.4.

The field shaper will overlap with the outer workpiece over a certain working zone lF.S.. The further they overlap, the longer the magnetic pressure zone on the outer workpiece, and the more material that will be deformed (see figure 5.7). Also, a longer working zone allows longer welds to be formed. This parameter will have an important effect on the impact angle and possibly as well on the impact velocity during the process, as the deformed zone will be bigger and the deformation will happen faster with a bigger overlap. The effect of the position of the field shaper will be examined in the experiments during this thesis.

The overlap of the field shaper can be adjusted by changing the length of the outer tube lo.t.. This is shown in figure 5.4. On its left side, the outer workpiece rests against a fixed element of the system. The distance between this left wall and the field shaper is 38 mm, as could be seen in the drawing of the field shaper in figure 4.6. Hence, the overlap of the field shaper depends on the length of the outer tube:

lo.t. = 38 + lF.S. (5.3) With: lo.t. = the length of the outer tube [mm] lF.S. = the length overlap of the ﬁeld shaper [mm] Chapter 5. Process parameters 67

Figure 5.7: Magnetic pressure on the outer tube caused by overlap of ﬁeld shaper.

5.2.3 Voltage level

The only electrical parameter that can be set during the experiments is the voltage level of the capacitor, which is directly related to the energy level in the system, as can be seen in formula 4.1. The voltage level is a crucial parameter in the forming of the weld, as it is this parameter that determines the impact velocity for a given geometry and given materials of the workpieces. If a good weld is not obtained in a chosen experimental setup and the cause is believed to be an insuﬃcient impact velocity, then bonding can be achieved by simply increasing the chosen voltage level. On the other hand, choosing a voltage level and hence an impact velocity that is too high could lead to failure of the weld as well, as was explained in the welding windows in chapter 3.4.

The formulas in chapter 4 allow to calculate the relationship between the impact velocity vc

and the set voltage level Uset for different materials and geometries. In figure 5.8 the relation is shown for a flyer tube thickness of 1, 5 mm and a stand-off distance between 2, 5 mm and 3, 5 mm, and for a flyer tube thickness of 2, 0 mm and a stand-off distance of 2, 5 mm. These values were calculated with the formulas from chapter 4. It can be seen from this graph that the bigger the stand-off distance and the thinner the workpiece (i.e. lighter), the more steep the increase in impact velocity with increasing voltage level. According to the formulas, when a bigger standoff is chosen the required voltage level can be chosen lower.

As an increase in the voltage level results in an increase of the magnetic pressure on the outer workpiece, also the deformation behaviour might diﬀer so it is not unlikely that the impact Chapter 5. Process parameters 68

Figure 5.8: The influence of the voltage level on the impact velocity according to the formulas in chapter 4 for an aluminum flyer workpiece with different thicknesses and different standoff distances.

angle would change at higher energy levels. The experimental results will have to conﬁrm this.

The voltage level is the parameter that is easiest to change in the experiments, and will come up as one of the most important parameters for a successful design.

5.3 Invariable parameters

5.3.1 Frequency

As mentioned before, the frequency plays a role in the skin depth of the material. Increasing the frequency reduces the induced current layer thickness. This can be a reason to adjust the frequency of the current when using thin sheets as ﬂyer material. For example when dealing with conductive drivers, the skin depth might become too large and therefore it might be necessary to increase the frequency. However, in this thesis the thicknesses of the tubes are suﬃciently high and the skin depth will always have values that are much smaller.

Apart from its effect on the skin depth the frequency also influences the space distribution of the magnetic field, the peak value of the magnetic pressure and the pulse width of the magnetic pressure. The pulse width of the magnetic pressure and hence the pulse width of Chapter 5. Process parameters 69 the acceleration is of importance in the acceleration time of the flyer workpiece. The longer it accelerates, the more the flyer tube will increase its velocity. Lower frequencies lead to larger pulse widths. But there are more effects of the frequency that are of importance, but difficult to quantify. A good illustration of this is that from numerical results in magnetic pulse forming it was seen that the maximum radial displacement, the maximum radial velocity and the maximum axial magnetic flux density were reached at different frequencies. [38]

Adjusting the frequency is only possible by making adjustments to the capacitor bank and consequently can’t be set in the magnetic pulse welding equipment that is used in this thesis. The frequency will be constant at 14.000 Hz throughout the experiments in this thesis. The maximum frequency that can be reached with this capacitor bank is 28.000 Hz.

5.3.2 Length ﬁeld shaper

The length of the field shaper is the length over which the magnetic field will be concentrated. When keeping the energy level constant, the magnetic field and hence the magnetic pressure will be higher when the length of the field shaper is smaller. The reason for this is that the energy will be concentrated in a smaller zone. In order to keep the same magnetic pressure with a longer field shaper, the energy level would have to increase.

The required length of the field shaper is dependent on the overlap of the field shaper with the outer workpiece lF.S.. lF.S. is defined as the distance between the side of the flyer tube that is about to be welded, and the edge of the field shaper that is placed furthest over the flyer tube. As the magnetic field might show discontinuities towards the edge of the field shaper, the field shaper should be longer than the overlap.

All the experiments are executed using the same field shaper with a length of 15 mm. If this would seem too big in comparison with the required overlap, energy might be saved by using a smaller field shaper. Working with a smaller field shaper can also be of use to generate a higher local magnetic pressure with the same energy level. In this case the working zone and hence the potential welding zone will be smaller. Opposite to this, if the magnetic pressure is high enough, it might be possible to work with the same energy level and a longer welding zone in order to produce longer welds. It should be noted that the length of the working zone also influences the impact angle. Welding with a shorter working zone in order to increase the magnetic pressure is therefore not a guarantee for success.

5.3.3 Diameter of the ﬂyer tube

The diameter of the ﬂyer tube will have an inﬂuence mainly on the formability of the weld. The use of bigger tubes means that there is more material that needs to be deformed and accelerated. So for bigger diameters, a higher required energy level is expected. As mentioned Chapter 5. Process parameters 70

before, the material needs to be compressed from an inner diameter di to an inner diameter di 2 · s, with s the stand-oﬀ distance. The decrease of the diameter relative to the original − · diameter di−2 s will be smaller for larger diameters. This makes them more deformable, as di the material needs to compress less.

The diameter is limited by the size of the field shaper. To obtain an equal magnetic pressure all along the circumference of the flyer tube, the diameter should be placed as close as possible to the field shaper. To ensure a good radial alignment, the outer diameter is chosen to fit the insulation surrounding the field shaper. In the experiments of this thesis, the outer diameter of the flyer tube will therefore always be set at 25 mm.

5.4 Summary

Numerous parameters will inﬂuence the impact velocity and impact angle during the process. These parameters can be inherent to the materials that are used, to the geometry of the workpieces, or to the electrical system. In the experiments that will be done in the following of this thesis, several of these parameters will be varied in order to learn what their inﬂuence on the process is, but also to understand the way of impacting during the process. An overview of these parameters is given here:

Material of the ﬂyer tube

Material of the inner tube

Energy level

Stand-oﬀ distance

Position ﬁeld shaper

Shape of the inner tube

Thickness of the outer tube Chapter 6

Experimental results

In this chapter an overview is given of the observations during the experiments. Assumptions and conclusions about these observations will be discussed. In the next chapter, the results of the experiments are compared with the theory of explosive welding and the formulas which were presented before.

6.1 Experimental outline

In this thesis, a total of 76 experiments performed at various settings of the process parameters. The experimental setup was always as shown in figure 5.4. Apart from the charging voltage level and the material, also the geometrical parameters were varied. After joining the two workpieces (figure 6.1(a)), the weld was examined visually. The welded zone was isolated and cross-sectioned (figure 6.1(b)). After embedding in epoxy the workpieces were prepared by standard metallographic procedures, such as mechanical polishing down to 3 µm and also some of the parts were chemically etched. The samples were examined by optical microscopy (figures 6.1(c) and 6.1(d)). Some samples (J4 and C21) were also examined with scanning electron microscopy (SEM) and hardness tests (HV0,2) were conducted on were welds (experiments C1, D1 and G5).

Of the 76 experiments that were done, 31 showed a successful weld after cross-sectioning of the workpieces. Most of the experiments were done for welding aluminum to aluminum, but also copper to aluminum and aluminum to steel were welded successfully. The experiments were split into series in which 1 or 2 parameters were varied. The results won’t be discussed per series, but will be used to illustrate the diﬀerent conclusions. The pictures which are shown in this thesis are the ones that are most representative for the conclusions, but they are only a small part of the total amount pictures that were taken.

The most important series is number C, as it is the most complete series of experiments. It was a continuation of tests which were performed previously. In this series, aluminum tubes

71 Chapter 6. Experimental results 72

(a) Two workpieces that are joined (b) The workpieces after cross- sectioning

Figure 6.1: The outline of the experiments, starting from the 2 joined workpieces until the microscopic view of the weld.

with thickness 1, 5 mm were joined with aluminum straight inner workpieces (figure 6.2 b) at a voltage level of 15 kV. The position of the field shaper was varied at several stand-off distances, and a range of possible positions of the field shaper was found. The other series can all be compared with the series C as they only differ in 1 or 2 parameters, usually the shape of the inner workpiece and/or the material. In the series A and B, experiments were done with a slant configuration of the inner workpiece for both the combinations aluminum-steel and aluminum-aluminum. In the series D it was checked whether the chosen voltage level in series C could be lowered. The series E was done to verify if the overlap length of the workpieces are of importance in the forming of the weld. Series F and H are tests where the field shaper was placed entirely over the workpieces. In series G it is checked what the influence of the outer tube’s thickness might be on the formation of the weld. In series I the combination aluminum-steel is tried for the same geometries that appeared to be successful in series C. In series J the same is done for the combination copper-aluminum. These series I and J were completed close to the finishing of this thesis, and therefore no further research has been done Chapter 6. Experimental results 73

Figure 6.2: The different geometries of inner workpieces which are used in the experiments: a)slant configuration b)straight configuration c)configuration without collar

on the welded samples apart from a visual examination whether the pieces were welded. In series K and L inner workpieces without a collar on the internal workpiece were used to study in what way this collar influences the deformation and impact behaviour. A short overview of the series is given in table 6.1. The variable parameters can be seen on figure 6.3. They are marked with a if they vary within the series. In the column referring to the geometry, ∼ the notations a, b and c correspond with the slant, straight and straight without collar inner workpieces of figure 6.2 respectively. Note that the overlap of the field shaper is sometimes bigger than the actual length of the field shaper, which equals 15 mm. This means the field shaper doesn’t reach over the edge of the outer tube.

Table 6.1: An overview of the experimental series.

Name Exp. Outer tube Inner tube Geom. V s lwp t lF.S. [kV] [mm] [mm] [mm] [mm] A 9 aluminum steel a 15 11 1,5 ∼ ∼ B 6 aluminum aluminum a 15 11 1,5 ∼ ∼ C 15 aluminum aluminum b 15 12 1,5 ∼ ∼ D 7 aluminum aluminum b 12 1,5 ∼ ∼ ∼ E 1 aluminum aluminum b 15 2,5 9 1,5 9 F 1 aluminum steel b 15 2,5 12 2,0 19 G 7 aluminum aluminum b 2,5 12 2,0 ∼ ∼ H 1 steel steel b 15 2,5 12 1,0 17 I 8 aluminum steel b 12 1,5 ∼ ∼ ∼ J 12 copper aluminum b 12 1,5 ∼ ∼ ∼ K 6 aluminum aluminum c 15 2,5 1,5 ∼ ∼ L 3 copper aluminum c 15 20 1,5 ∼ ∼

A full overview of the experiments can be found in appendix A. The numbering was kept Chapter 6. Experimental results 74

Figure 6.3: The variable parameters in the series of experiments.

the same as the original numbers of the experiments. The seemingly missing numbers are experiments that were cancelled because they wouldn’t oﬀer much extra information after the examination of other samples, or because they are expected to fail based on the results of other experiments. For every series, a drawing with the experimental layout is given in the attachment. The parameters that are varied are marked with their symbol on this drawing, parameters that are kept constant are marked with their value. The drawing in each series is accompanied with a table in which an overview is given of the experiments. The values of the varying parameters are given in this table, and it is noted whether the workpieces were successfully joined or not. In case a weld was achieved, the lengths of the run-in zone li, the weld itself lw and the run-out zone lo were given in this table, followed by the total length lt of these 3 zones.

The variable parameters that will be used in this and the following chapter are as follows:

lF.S. = overlap length of the ﬁeld shaper [mm]

s = stand-oﬀ distance [mm] Chapter 6. Experimental results 75

V = the voltage level [kV]

t = thickness of the outer tube [mm]

lwp = overlap of the workpieces [mm]

β = initial angle of inner workpiece surface [◦]

In the following, reference is often made to the experiment numbers. The parameters of the mentioned experiments will only be explicitly given if they are essential for the content. All other details of the experiments can be found in appendix A.

6.2 Material characteristics

The experiments are mainly done with both the flyer tubes and the inner workpieces made out of aluminum EN AW-6060. In the series J and b, copper R220 is used as flyer material. In the series A, F and I, steel C45 is used for the inner workpieces. In the experiment of series H, steel E235 is used for the flyer tube and steel C45 is used for the inner workpiece. The characteristics of the materials used are described below.

6.2.1 Aluminum EN AW-6060

Aluminum alloy 6060 is a general commercial alloy, and is the most commonly used. Its main alloying elements are Si and Mg (> 30%). The material characteristics are shown in the table below.

Yield strength 110 MPa Tensile strength 150 MPa Elastic modulus 69 GPa Elongation at break 8 % Density 2700 kg/m3 Thermal expansion coeﬃcient 23,4 · 10−6 K−1 Melting temperature 888-923 K Thermal conductivity 209 W/mK Electrical conductivity 32 · 106 m−1Ω−1

6.2.2 Copper R220

Soft copper R220 is a copper alloys that is composed for more than 99,9% out of Cu. Its characteristics are given below. Chapter 6. Experimental results 76

Yield strength 70 MPa Tensile strength 220 MPa Elastic modulus 120 GPa Elongation at break 40 % Density 8960 kg/m3 Thermal expansion coeﬃcient 17 · 10−6 K−1 Melting temperature 1356 K Thermal conductivity 364 W/mK Electrical conductivity 59,7 · 106 m−1Ω−1

6.2.3 Steel C45

The high grade steel alloy C45 mainly contains C ( 0,44%), Si ( 0,28%) and Mn( 0,60%) ∼ ∼ ∼ as alloying elements. Its characteristics are given below.

Yield strength 340 MPa Tensile strength 600 MPa Elastic modulus 210 GPa Elongation at break 16 % Density 7850 kg/m3 Thermal expansion coeﬃcient 11,7 · 10−6 K−1 Melting temperature 1540 K Thermal conductivity 46 W/mK Electrical conductivity 5,0 · 106 m−1Ω−1

6.2.4 Steel E235

The steel alloy E235+C mainly contains C ( 0,17 %), Si ( 0,35%) and Mn( 1,20%) as ∼ ∼ ∼ alloying elements. No heat treatment was done after the ﬁnal cold drawing process. The characteristics are given below.

Yield strength 480 MPa Tensile strength 640 MPa Elastic modulus 210 GPa Elongation at break 6 % Density 7850 kg/m3 Thermal expansion coeﬃcient 11,7 · 10−6 K−1 Melting temperature 1540 K Thermal conductivity 46 W/mK Electrical conductivity 5,0 · 106 m−1Ω−1 Chapter 6. Experimental results 77

6.3 Experimental results

6.3.1 Deformation behaviour

The magnetic pressure pulse causes the flyer tube to be compressed and impact the inner workpiece. The flyer tube will undergo large deformation due to the radial compression and will deform more for a larger stand-off distance. The impact causes severe deformation of the T inner workpiece as well. The duration of the pressure pulse theoretically equals 2 . With the 1 system’s frequency f = T = 14.000 Hz, the pressure pulse will take 35, 7 µs. Measurement of the current pulse during the experiments confirmed this. For example the measurement in figure 6.4 shows a duration of 37, 7 µs. The peak pressure that corresponds with the current peak will in this case be reached at 17, 5 µs. The deformation that was observed due to the pressure pulse is described in this section.

Figure 6.4: Measurement of the current through the coil in experiment J4.

Symmetry

After cross-sectioning it was noticed that the welding lenghts on both sides of a sample were not always equal. It even occurred that only one side of the specimen seemed to be welded, or that one of the two halfs was welded while the other half was not. This means that even though the magnetic pressure is equally distributed over the circumference of the tube, the weld will not be equal over the whole surface. It is assumed that the weld is not symmetric due to the deformation of the workpiece, and more in particular the suspected buckling, during compression. This phenomenon is described in what follows.

After cross-sectioning of experiment G7, at ﬁrst both the halfs of the workpieces showed welds on both sides. Several seconds after the piece was cross-sectioned, a click was heard and 1 of the halfs of a sample appeared to be cracked. This cracking was probably the consequence of the residual stress in the ﬂyer tube, due to its high deformation. This indicates that failure of the weld can occur because of the residual stresses that are released after cross-sectioning. When all the sides that are visible after cross-sectioning crack open, these experiments will be categorised as unsuccessful. Even though there has been a weld, it must have been of poor quality because it cracked open. Chapter 6. Experimental results 78

Flyer tube

The flyer tube deforms due to the radial compression caused by the magnetic pressure. The diameter of the tube will decrease, and the radial stresses cause according to Poisson’s law large circumferential and longitudinal stresses in the flyer tube. When the stand-off distance and thus the decrease in diameter is larger, these stresses will be higher. From the experiments, it was seen that these stresses cause an increase of the thickness of the flyer tube on the one hand, and buckling (possibly due to the circumferential stresses) along the circumference of the flyer tube on the other hand.

Figure 6.5: An exaggerated view of the buckling in the ﬂyer tube due to the high circumferential pressure during compression. The 4 symmetric buckling zones on the drawing are just a way to illustrate the eﬀect, and not representative for the actual buckling zones.

Buckling along the circumference means that the material tends to move outwards at one or several locations, as is shown in figure 6.5. The direction of movement at these locations is opposite to the magnetic pressure. Due to the magnetic pressure and their acceleration, these parts are also impacted onto the inner workpiece. However, due to this buckling effect, these areas lag behind on the rest of the tube, and consequently the jet concentrates in these areas. Evidence of this can be seen in figure 6.6, where the jet remainder on the inner and the outer workpiece is shown. The jet remainder form a certain pattern, which indicates that also the buckling occurred. Chapter 6. Experimental results 79

(a) Jet concentrations on the inner work- (b) Jet concentrations on the ﬂyer workpiece piece

Figure 6.6: Concentration of the jet at regular distributed areas in aluminum-steel experiments (series A), creating a wavy form of jet material on both the inner and ﬂyer workpiece.

Figure 6.7: The arrows mark the location where the gap in the field shaper was located. At this place, the flyer tube is compressed less firmly against the inner workpiece due to the lower magnetic pressure.

The flyer tube doesn’t buckle randomly along its circumference. As can be seen in figure 2.6, there is a split in the field shaper. Consequently, at the place of the split there is no magnetic field and no magnetic pressure. Therefore, the tube will tend to buckle at this location. In all experiments, a location was observed where the materials were pressed less firmly together (figure 6.7). Although the location of the gap in the field shaper wasn’t marked on the workpieces prior to welding, it is expected that the gap was located in this zone. Starting from this place, a buckling pattern develops along the circumference of the weld (figure 6.5). Chapter 6. Experimental results 80

(a) Jet concentrations after a copper- (b) Jet concentration on the alu- aluminum experiment minum inner workpiece

Figure 6.8: Concentration of the jet material in copper-aluminum experiment J4. The jet is more equally spread, with high concentration in one particular zone. SEM investigation shows that the jet material consists of both Cu from the ﬂyer tube and Al from the inner workpiece.

In the experiments where copper was used as the flyer tube material, the jet was more equally spread around the surface, but with a high concentration at the location of the gap in the field shaper (figure 6.8). This could possibly be due to the higher elastic modulus of copper, that causes the circumference only to buckle at one position.

Due to the decreasing diameter also longitudinal stresses will exist in the flyer tube, apart from the radial and circumferential stresses. These stresses cause the flyer tube to become longer and thicker (figure 6.11). Towards the end of the flyer tube, where the workpieces impact, the tube thickness shows a small decrease over a narrow zone. In the middle of the working zone the flyer tube increases in thickness (figure 6.9(b)). If only the thickness would increase, and no stretching of the flyer tube would occur, the final thickness would reach values shown in table 6.2.

These effects are illustrated in experiment G4 (figure 6.9(b)) where the thickness in the end zone decreases from 2, 00 mm to 1, 88 mm, and reaches a maximal value of 2, 48 mm in the middle. Figure 6.9(a) shows the thickness increase in experiment C18, which has an original flyer tube thickness of 1, 50 mm. The thickness was different on both sides of the workpiece, and also the angle between the flyer tube and the internal workpiece was different. On the side where the angle is 44◦, the thickness increase reaches a maximum of 1, 75 mm, on the side where the angle is 30◦ the thickness increase reaches a maximum of 1, 91 mm. The weld was longer on the side with the highest angle and the smallest thickness increase. Chapter 6. Experimental results 81

(a) A different thickness increase on both sides of the flyer workpiece in the sample of experiment C18, with an original thickness of 1, 5 mm and a stand-off distance of 3, 5 mm.

(b) Thickness increase in experiment G4, with an original thickness of 2.0 mm and a stand-oﬀ distance of 2, 5 mm.

Figure 6.9: Examples of the thickness increase in the ﬂyer tube.

Table 6.2: The theoretical thickness increase of the ﬂyer tube, calculated according to the conserva- tion of volume.

Thickness before welding Stand-oﬀ Theoretical thickness after welding [mm] [mm] [mm] 1,5 2,5 2,02 1,5 2,5 2,13 1,5 3,0 2,25 2,0 3,5 2,71 Chapter 6. Experimental results 82

The thickness increase is less then the values in table 6.2. This shows that the flyer tube also stretches to a longer length. Stretching happens at the end of the flyer workpiece, where the workpieces impact, and also at the beginning, where the flyer tube is moved under a certain angle (figure 6.9(b)).

Inner workpiece

In general, the inner workpieces were deformed severely during the impact. The highest deformation occurs at the outermost point of flyer tube, where the inner workpiece is compressed in a narrow zone (figure 6.10). From there on, the deformation of the inner workpiece declines gradually towards the end of the weld (from right to left in figure 6.10). The high deformation at the right end of the flyer tube indicates that the impact of the workpieces occurs here first. The inclination in this zone was found to be between 0, 5 mm and 1 mm. At lower energy levels and stand-off distances, the deformation was smaller. The width of this severely deformed zone becomes larger for a longer overlap of the field shaper, which is shown in figure 6.28 for the copper-aluminum experiments.

(a) Welded experiment K3

(b) Zooming in on the severe defor- (c) A similar severe deformation in mation at the end of the ﬂyer tube experiment C3. in experiment K3.

Figure 6.10: The deformation of the inner workpiece in some of the experiments.

These high deformations show that in case a tube would be used for the inner workpiece, a mandrel should be placed to prevent deformation. Chapter 6. Experimental results 83

Displacement of the workpieces

When looking at the welding lengths in the overview of the experiments (appendix A), it can be seen that the total length of the overlapping zone lt is always longer than the initial overlap of the workpieces. This means that the flyer tube and the inner workpiece are displaced relative to each other during the process. This displacement was confirmed in several experiments: the initial position of the inner workpiece was marked on the flyer tube before welding, and after the welding it was seen that the inner workpiece was displaced relative to this mark.

Figure 6.11: The deformation behaviour of the outer tube. Due to the radial compression, a force is generated on the collar of the inner workpiece (1.) which pushes it inwards the flyer tube. Consequently, the flyer tube will reach further than the point where it initially impacted (4.). The circumferential and longitudinal stresses (3.) cause the thickness of the flyer tube to increase (2.).

The flyer tube suffers from large radial forces due to the compression. Due to the deformation, this radial compression causes a force on the collar of the inner workpiece (figure 6.11). This force pushes the inner workpiece to the left, inwards the flyer tube. The movement starts before the impact, because the impact point was found to be at a further distance from the collar than the initial overlap length of 12 mm. The flyer tube reaches further than the point where it impacted, which means that the workpieces are also moved after the impact. This is seen in figure 6.9, where the flyer tube reaches further than the point with the highest deformation of the inner workpiece. This movement occurs until the workpieces are joined at one point, so they are unable to move relative to each other.

The displacement varies between 0, 0 mm and 4, 0 mm, causing overlap lengths lt of 12 mm to 16 mm. The overlap lengths in series C show that the length increases with increasing stand-oﬀ distance s. Comparison with series G shows that an increasing ﬂyer tube thickness

t also causes an increase of the overlap length lt. Series C and G show the same correlation Chapter 6. Experimental results 84

between the field shaper overlap lF.S. en the eventual overlap of the workpieces lt. This leads to the conclusion that with increasing flyer tube thickness t or field shaper overlap lF.S., the radial force of the flyer tube on the collar of the inner workpiece will be bigger. The longer

overlap lt for an increasing stand-oﬀ distance s can be explained by the fact that the force on the collar is applied longer since it will take longer for the ﬂyer tube to reach the inner workpiece.

A rough estimation of the displacement speed can be given by assuming that the displacement T happens during the entire pressure peak, which lasts 2 = 35, 7 µs. A displacement of 2 mm 2 mm during that time takes place at a velocity of 35,7 µs = 56 m/s = 202 km/h.

6.3.2 Jet formation

During the welding, a jet is created along the surface of the weld. Remainders of this jet material are found in all the experiments. A substantial amount of material is moved by the jet force. This is illustrated in figure 6.12, where a discolouration of the workpieces without collar is seen due to the jet along its surface. In the 5 workpieces on the bottom row, the jet was blocked by the collar and remains trapped between the workpieces. This can be clearly seen on figure 6.8(b). As mentioned before, the jet material will concentrate at the places where the flyer tube buckles.

Figure 6.12: Diﬀerence between welded pieces without a collar from series a (above 3), and pieces with a collar from series C and G (down 5). The change of colour in the experiments of series a indicate a large amount of jet material that passed along its surface.

Because the jet material is blocked at the collar, the question arises if this can influence the process and hence the weld quality in a negative way. The experiments in series K and series C (for a stand-off distance s = 2, 5 mm) show no increase of the welding length due to the Chapter 6. Experimental results 85 removal of the collar. This indicates that the blocking of the jet had no influence on the welding process in these experiments.

In order to determine the composition of the jet, a SEM examination was performed on the jet remainders on the copper tube (figure 6.8(a)) and the aluminum workpiece (figure 6.8(b)) in experiment J4. The results for the jet remainder on the inside of the copper tube (figure 6.13) showed that part of the jet material consists of the aluminum of the inner workpiece. The results for the jet remainder on the aluminum inner workpiece (figure 6.14) showed that also copper is an important element in the jet material.

(a) Result of the SEM test. (b) Another measuring point on the inside of the Cu tube.

(c) Surface of the remainder inside the copper (d) Surface of the remainder inside the copper tube. (magn. 400x) tube. (magn. 1000x)

Figure 6.13: Results of the SEM investigation of the jet remainders on the copper ﬂyer tube of experiment J4, showing that the aluminum of the inner workpiece is an important component of the jet material.

Most of the tests were done without pre-cleaning of the workpieces. Only when the surface was covered with oil which remained from the manufacturing of the workpieces, it was cleaned Chapter 6. Experimental results 86 roughly with a paper tissue. This shows that indeed the magnetic pulse welding can be done without pre-cleaning of the surfaces, and hence the oxide layer and surface contaminants are removed by the jet. Hence, the jet consists out of material from both the inner workpiece and the ﬂyer tube, and is capable of removing the surface contaminants suﬃciently to ensure good joining of the workpieces.

(a) Surface of the remainder on the aluminum in- (b) Surface of the remainder on the aluminum inner workpiece. (magn. 50x) ner workpiece. (magn. 400x).

Figure 6.14: Results of the SEM investigation of the jet remainders on the aluminum inner workpiece of experiment J4, showing that the copper of the ﬂyer tube also is an important component of the jet material.

6.3.3 Welded zone characteristics

Location of the welded zone

The working zone after welding can be divided up in 3 zones, which are shown in ﬁgure 6.15. Joining of the materials will only occur in the middle part of the working zone. As the Chapter 6. Experimental results 87 workpieces will impact from right to left in the ﬁgure, the zone on the right with no weld will be called run-in zone li , and the left zone will be called the run-out zone lo.

No clear correlation was shown between the lengths of the run-in zone and run-out zone and the initial process parameters.

At the end of the run-out zone, the ﬂyer tube is positioned under a certain angle γ (ﬁgure 6.15). This angle was measured for several specimens. It was noticed that the angle becomes higher for experiments with higher energy levels, and for experiments with longer overlap of

the ﬁeld shaper lF.S..

(a) The welding experiment G4 for inner workpieces with a collar.

(b) The welding experiment K2 for inner workpieces without a collar.

Figure 6.15: Illustration of the run-in zone li, the weld, and the run-out zone lo for inner workpieces with and without collar.

◦ ◦ ◦ For example in the series G, the angle equals 25 , 31 and 37 for lF.S. of 8 mm, 10 mm and 12 mm respectively. The angle for the series K and L, where the inner workpieces didn’t have a collar, was similar to the angles in the series where the inner workpiece did have a collar. These angles all had values of 30◦( 2). However, the effect of an increasing angle for a longer ± field shaper overlap was not noticed in the experiments without the collar. This indicates that the collar on the inner workpiece has a certain influence on the deformation behaviour only when the field shaper overlap exceeds a certain value. Chapter 6. Experimental results 88

Length of the welded zone

Microscopical investigation is the only way for determining the weld quality in this thesis, and the length of the welded zone is the only parameter to quantify the welding quality. To characterise the length of the weld, the average value of the welding length on the two sides of the specimen is used. As described before, the welded length is not necessarily the same everywhere around the circumference. The diﬀerences in welding length that were seen between two sides were however small (in the order of 2, 0 mm), so the average value of the two sides is a good representative value.

The inﬂuence of the parameters will be discussed based on the length of the welded zone, as will be seen in ﬁgures 6.29, 6.30, 6.33 and 6.32. The welding lengths that were encountered varied between 3, 0 mm and 9, 5 mm.

6.3.4 Interface morphology

Wavy interface

In most of the samples, prior to etching, the interface was indistinguishable of the base materials. A few examples however showed a clear wavy welding interface. After etching, the wavy interface could also be noticed in other samples. The wavelength and the amplitude of the waves often appeared in a regular pattern. A few examples are given in the figures 6.16. The pattern will be described from right to left in the images, as this is also the direction of the weld formation. The most frequently found pattern of the waves, was with an increasing wavelength towards the end of the weld, and an amplitude that rises until a peak value before decreasing again. Examples of this pattern can be seen in figure 6.16(b) and figure 6.16(d). The latter shows that this pattern evolves to a smooth interface. Welds also occurred with a more or less constant or slightly varying wavelength and amplitude. Figures 6.16(a) and 6.16(c) are examples of this.

The wave pattern of the welded samples in the diﬀerent experiments will be compared according to their appearance, and according to the occurring wavelengths. Because it is not always clear how to interpret where a wave starts and ends, there is a fault margin of about 30 µm1. The size of the amplitudes are not compared as these values are much smaller than the wavelengths, and not much bigger than the fault margin. Often the run-in zone and sometimes the run-out zone also showed a wavy appearance, but this was not studied intensively.

In table 6.3, an overview is given of the patterns of wavy interfaces that occurred and the range of wavelengths that were found for the samples in which they were measured.

When taking a closer look at these results, several phenomena can be seen. Most of the

1This is the maximum diﬀerence that occurred when measuring the same wavepeak several times. Chapter 6. Experimental results 89 waves that occurred in the weld, hold the same pattern of wavelengths and amplitudes: a wavelength which increases towards the end, and an amplitude that ﬁrst increases and then decreases. The interfaces in the welds that didn’t have this pattern are all holding constant wavelengths and amplitudes. The waves that were measured in the run-in zone (K3, B5, G6) showed a similar pattern as well: an increasing wavelength with a decreasing amplitude.

In the experiments C5, C6 and C7 the same wave pattern occurs. However, the waves will occur more towards the end of the weld for experiment C6 than for C5, and even more for

C7. This shows a remarkable correlation with the ﬁeld shaper overlap lF.S., that increases from C5 to C7.

Table 6.3: Overview of the wavy interfaces observed in some of the experiments. The pattern of the wavelength and the amplitude is described from the beginning to the end of the weld: % indicates an increase of the parameter, indicates a decrease, indicates an increase & %& followed by a decrease, indicates that no clear tendency was observed. ∼ Number Location weld Wavelength Amplitude λ [µm] C25 weld 210-180 ∼ ∼ C30 weld 440-170 % %& C31 weld 320-230 ∼ ∼ C1 weld 450-170 % %& C5 beginning weld 300-200 % %& C6 weld 600-400 % %& C7 weld 600-270 % %& C20 weld 340-160 % ∼ E1 weld 570-430 % D1 weld 170-120 ∼ %& J2 weld 80 ∼ ∼ G4 run-in + weld 220-100 % ∼ G5 run-in + weld 180-130 ∼ ∼ G6 run-in 110-70 %& K2 weld 350-180 % %& K3 run-in 220-150 ∼ & B5 run-in 340-250 %& Chapter 6. Experimental results 90

(a) Experiment C25: both the wavelength and amplitude stay more or less constant along the weld.

(b) Experiment C30: the wavelength increases towards the end of the weld, the amplitude increases and suddenly decreases at the end.

(c) Experiment C20: the wavelength increases slightly towards the end of the weld, the amplitude stays more or less constant.

(d) Experiment C5: the wavelength increases and the amplitude increases and decreases again at the beginning of the weld, from there on the interface is smooth.

Figure 6.16: An overview of diﬀerent wavy interfaces. Chapter 6. Experimental results 91

When comparing the wavelengths in the interface of C25 (s = 2, 0 mm), C1 - C30 - C31 (s = 2, 5 mm) and C5 - C6 - C7 (s = 3, 0 mm), it is observed that the wavelength increases with increasing stand-oﬀ distance. The wavelength of experiment C20 however, where the stand-oﬀ was 3, 5 mm, is an exception, since the wavelength decreases again.

The experiments without a collar on the inner workpiece (K2 - K3) show a similar range of wavelengths as the equivalent experiments with inner workpieces with a collar (C1 - C30 - C31).

The experiments with a higher flyer tube thickness (G4 - G5 - G6) show smaller wavelengths for the same stand-off distance. The wavelength in these samples could only be measured in the run-in zone, or the zone where the weld apparently failed due to melting (G4 - see following section). Also the experiment at a lower energy level (D1) showed lower wavelengths than the experiments with the same stand-off distance (C1 - C30 - C31), but at higher energy level. The shorter workpiece overlap in experiment E1 led to a significant increase in the wavelength compared to experiment C31 with the same parameters, except for the workpiece overlap lwp.

The wavy interface in the experiment J2, where copper was welded to aluminum, was much less pronounced than the waves in the other experiments. This is shown in ﬁgures 6.17.

(a) The interface of the weld.

(b) A close-up of the waves before etching. (c) And a close-up after etching of the sample.

Figure 6.17: Results of the copper-aluminum experiment J2. Chapter 6. Experimental results 92

Melting

In figure 6.18 an example is shown of a wavy interface that has a more or less constant amplitude and wavelength in the first half of the weld, and that has the typical increase in wavelength and increase-decrease in amplitude in the second half of the weld. The first half of the weld is characterised by dark zones that follow the wavy pattern. These darker areas were often found in the wavy interface, but never on the smooth interfaces. Also the run-in zones where no weld was formed often contained these darker areas. Microscopical investigation of these areas showed that they were the result of melting and rapid solidification.

When zooming in on the weld of figure C7 (figure 6.19), it becomes clear that different molten zones occur along the interface, rather than a continuous molten zone. These melt zones can reach thicknesses of 44 µm (figure 6.19(b)). They are well separated from the zone that was not molten, as the change in colour is abrupt and not gradual. This indicates that the temperature rise, which must have been higher than the melting temperature for aluminum (888 K), was only noticeable in a narrow zone. This could have only happened if the heat increase only lasted for a very short period of time.

(a) An overview of the wavy interface.

(b) Zoom on the ﬁrst half of the wavy interface, where melting occurs.

Figure 6.18: Images of the wavy interface in the C7 experiment. In the ﬁrst half of the weld, both the amplitude and wavelength are more or less constant. In the second half, the wavelength increases while the amplitude rises to a peak and then decreases again.

Zones which were molten were found in several samples, but not all of them showed welding zones in the welded area itself. Often, the evidence of melting was found in the run-in or run-out zone of the weld. The samples with a clearly distinguishable molten zone along the weld interface were B2, C3, D1, G4, G7 and K3. The length of the welded area in B2 and G4 clearly decreased due to the melting and subsequent solidifaction. Chapter 6. Experimental results 93

(a) Melt pockets with a width of 27 µm. (b) Melt pockets with a width of 44 µm.

Figure 6.19: Zooming in on the interface of the C7 experiment shows that the interface consists of a series of melt pockets, rather than a continuous molten zone.

(a) The diﬀerence between the two sides of the B2 (b) A zoom on the molten zone, that doesn’t seem sample. Comparison of the both sides indicates to show any defects. that the molten zone on the one side corresponds with the weldless zone on the other side.

Figure 6.20: The welding interface in experiment B2. Chapter 6. Experimental results 94

(a) The diﬀerence between the interfaces on both sides of the sample shows that the welded area is much longer on one side.

(b) The zone that is not welded is clearly the consequence of melting and rapid solidiﬁcation.

Figure 6.21: The molten welding interface in experiment G4.

The experiments B2 and G4 show a remarkable difference between the two sides of the sample, as can be seen in figures 6.20 and 6.21. While the sample is welded over a certain length on one side, the other side shows no coalescence of the workpieces in that same area. Figure 6.20(a) shows that the length which is not welded on the one side corresponds with the molten zone on the other side (figure 6.20(b)). This is similar to experiment G4 (figure 6.21(a)), where a Chapter 6. Experimental results 95 slightly molten zone on one side corresponds with a severely molten zone that is teared open on the other side (figure 6.21(b)). This indicates that as soon as melting occurs, the weld might quality is low. This doesn’t happen necessarily around the whole surface, but this can undoubtedly affect the welding quality.

When looking closer to the molten zones of experiment G4 in figures 6.21(c) and 6.21(d), it can be seen that the zones contain porosities. This is a consequence of the rapid solidification in these areas. These porosities were seen even more clear in the weld of experiment C3 (figure 6.22(a)). Figures 6.22(b) and 6.22(c) show porosities in the molten zone and a small crack that is induced by the connection of the pores.

(a) A full view of the weld.

(b) A closer look on the molten zone clearly (c) Porosities in the molten zone. shows the crack along the interface.

Figure 6.22: The welding interface in experiment C3, with a small molten zone in the middle.

In experiment K5 a small zone is observed where melting occurred and a crack is induced (figure 6.23(a)). As the melting occurred very local, this might be the consequence of the severely deformed zone of the inner workpiece at the point where the outer tube impacted. Two assumptions are able to explain this: either the impact at this point was more severe than along the rest of the interface, or the jet that was initiated in the first part of the weld was trapped in this zone and heated it up (figure 6.24). As melting occurs in both the workpieces and not only in the inner workpiece, heating by the jet can’t be the reason, which means that the impact between the workpieces at this point must have been higher. This illustrates that Chapter 6. Experimental results 96

the horizontal contact point velocity vcp must be higher than the vertical impact velocity vc, as melting didn’t occur at any other point along the weld (ﬁgure 6.24).

(a) A local melted zone can be seen in the beginning of the weld.

(b) A similar local melt that was seen on (c) A close-up of the molten zone shows the the other side of the sample. porosity and the induced crack.

Figure 6.23: The interface of experiment K5, with a very local molten zone.

Figure 6.24: A schematic view of the two possible reasons for the melting in experiment K5. Either

the high contact point velocity vcp is the cause, or the jet material that bumps into the deformed zone of the inner workpiece.

Figure 6.23(c) shows the porosities in this zone, while ﬁgure 6.23(b) oﬀers a good view on the crack. The crack is perpendicular to the welding interface, which indicates that the stress Chapter 6. Experimental results 97 that caused it is parallel to the interface.

No proof of melting was found in the copper-aluminum experiment J2. In the run-in zone, it can be seen that a crack is present in the aluminum base material. Hence a weld has been formed, but the joining of the materials failed due to failure of the aluminum inner piece in the run-in zone. No reason was found for this failure, but it should be noted that the temperature increase could be the reason, even though no indications were found. The higher melting temperature of the copper would in that case explain why the failure occurred in the aluminum workpiece.

(a) A view on the start of the weld (left) (b) A zoom on the run-in one with alu- and the run-in zone (right). minum parts on the copper tube.

Figure 6.25: A view of the zone where the weld starts in experiment J2. It can be seen that the tear of the run-in zone is located in the base material.

Hardness increase

Hardness tests HV0,2 were done on the samples of the experiments C1, D1 and G5 in order to check if there was a hardness increase in the workpieces, and in order to compare the hardness of the weld with the base materials. The full results and location of the measuring points can be found in appendix B. The results are graphically shown also in ﬁgure 6.26. The measured hardnesses are shown in function of their location, which is characterised by its distance from the weld. The hardness of the reference points was measured well outside the welded and hence inﬂuenced area, and are also included in the graphs.

As the width of the welded interface is in the same order of 10 µm, measuring the hardness of this precise area with a HV0,2 test is more a matter of good luck than skills. In sample D1, a hardness increase was measured on the interlayer of almost 10 HV0, 2, which corresponds with about 100 MPa. In samples C1 and G5 this increase was not measured at the interface. The hardness in the vicinity of the interfaces was equal to the values in the zone next to these measuring points. Most likely, the hardness was not measured in the zone of the interface. Chapter 6. Experimental results 98

(a) Microhardness measurements for sample C1.

(b) Microhardness measurements for sample C1.

Figure 6.26: The microhardness (HV0,2) results for 3 experiments.

Both the ﬂyer tube and the inner workpiece show a similar hardness increase in all the samples. In experiments C1 and D1 there is an increase of about 3 HV0, 2 to 5 HV0, 2, which corresponds with about 30 MPa to 50 MPa in comparison with the reference points. Although these values are within the average spreading of results in hardness tests, a clear increase is Chapter 6. Experimental results 99 noticeable. In experiment G5, the increase is about 10 HV0, 2, which corresponds with about 100 MPa. The hardness increase in the inner workpieces is noticed in all the measured points up to a depth of 5 mm. In C1 and D1 the increase in the ﬁrst 2 mm is less than in the rest of the measured points.

This hardness increase is the consequence of the high deformation of the workpieces. This shows that the deformation in the G5 experiment was much larger than the deformation in C1 and D1. The deformed region of the inner workpiece is not limited to the immediate zone around the weld, but reaches deep in the inner workpiece and possibly even in the whole workpiece. The ﬁrst 2 mm with a smaller increase in C1 and D1 could not be explained, but might have something to do with the temperature increase due to the heat that is set free during the process.

6.3.5 Inﬂuence of the parameters

Position ﬁeld shaper

What immediately can be seen from the results of the experiments is the importance of the positioning of the field shaper, characterised in this thesis by the overlap of the field shaper with the workpiece lF.S.. This is most clearly seen in the experiments of series C. The positions that were tried at different stand-off distances are shown in figure 6.27. The green lines indicate that a weld was achieved with this position of the field shaper, the red lines indicate the unsuccessful experiments.

Figure 6.27: The different positions of the field shaper that were tried in series C. The green bars show the positions of the field shaper that led to a successful weld, the positions at the

red bars didn’t lead to a good weld. The overlap of the ﬁeld shaper lF.S. is indicated above the bars. Chapter 6. Experimental results 100

It can be seen that for different stand-off distances there is a different possible range of positions of the field shaper. This means the field shaper overlap must have an influence on the impact velocity and/or the impact angle. The zone with magnetic pressure increases with increasing field shaper overlap, as was seen in figure 5.7. This means that the total pressure will be higher, so it will have a certain influence on the impact velocity. But, there is also a maximum possible overlap of the field shaper. As the impact velocity is in this case definitely large enough, the impact angle must have reached values outside the possible range for welding. This indicates that the overlap also influences the impact angle.

It was also seen that the impact zone that was described in section 6.3.1 becomes wider with a longer field shaper overlap. This indicates that the deformation behaviour is different, and hence the impact angle as well. This can also be seen in the experiments L3, L2 and L1, which were not welded. The impact zone becomes wider with increasing field shaper overlap. More interesting about these experiments is the remainders of the jet material on the copper tube. It can be seen from figure 6.28 that for experiment L3, which had the largest field shaper overlap, the jet wasn’t formed at the start of the weld. This indicates that the impact angle is indeed influenced by the position of the field shaper.

Figure 6.28: Results of the experiments (from left to right) L3, L2 and L1. The red lines mark the ultimate position of the field shaper. Due to the different positions of the field shaper, the jet wasn’t generated along the whole surface in L3.

From the measurements of the welding lengths it could be concluded that the field shaper position always influences the process, regardless the other parameters. However, a clear tendency was not observed. Several graphs that were used in this chapter to compare welding lengths illustrate this. In figure 6.29 its influences can be seen in the experiments with slant inner workpieces of series B. In figure 6.30 the influence is shown for the inner workpieces without a collar in series K. In figure 6.32 the influence is shown on some of the experiments Chapter 6. Experimental results 101 at higher thickness in series G. In figure 6.33 its influence again is shown on the series C. From this graph it might be concluded that the welding length is maximal for the field shaper overlap that is closest to the maximum possible. However, it should be noted that for the experiments with the highest overlap, C3, C7 and C20, the weld all showed proof of melting. In general, it appears that the optimal field shaper overlap is dependent on the other parameters.

Shape inner workpiece

Figure 6.29: The inﬂuence of the initial angle on the inner workpiece in the aluminum - aluminum experiments by comparison of the welding length in series B and C.

In order to examine the influence of using a slant inner workpiece, the welding lengths of series B and C were compared for the combination of aluminum with aluminum. As the stand-off distance in series B varies along the length, the welding lengths are compared for the experiments in series C at various stand-off distances. These results show that manufacturing the slant inner workpiece doesn’t result in any improvement for the welding lengths. On the contrary, the welding lengths that were achieved in the experiments of series B were lower than with straight inner workpiece surfaces of series C. Additionally, two experiments failed at ◦ ◦ initial angles of 8 and 12 for a field shaper overlap lF.S. of 11 mm, while similar experiments with a straight interface and with an initial angle of 4◦ did were successful.

The experiments of series A between aluminum and steel all failed as well. This could be compared with the experiments of series I between aluminum and steel, but those were conducted with voltage levels 15 kV. It is therefore not known if the shape of the inner workpiece is ≥ the only reason that leads to the failure of the weld, or if an insuﬃcient energy level was also of importance.

From these experiments it was concluded that the initial angle of the inner workpiece doesn’t Chapter 6. Experimental results 102 improve the welding process. Above that, the more complex shape makes it more diﬃcult to manufacture. After these results, it was decided to continue the experiments with straight inner workpieces alone.

Collar

Figure 6.30: The inﬂuence of the collar on the inner workpieces by comparison of the welding lengths in aluminum-aluminum experiments at a stand-oﬀ distance of 2, 5 mm in series C and a.

In series K, some experiments were done completely similar to the ones in series C, but with inner workpieces without a collar (figure 6.30). The experiments K1, K2, K3 and K4 are meant to be compared with experiments C30, C1, C3 and C8 respectively. The welding lengths for a field shaper overlap of 8 mm (K1 - C30) and 10 mm (K2 - C1) seem to be comparable. For a field shaper overlap of 14 mm both the experiments K4 and C8 failed. The experiments K3 and C3 however, show a big difference in welding length: the welding length of experiment C3 (9, 4 mm) is significantly larger than the length in experiment K3 (2, 3 mm).

When looking more closely to the weld in experiment K3 (figure 6.31), it can be seen that the whole run-in zone exists out of a molten layer that is cracked. In other words, a bond between the 2 workpieces was formed, but it was too weak due to the melting and subsequent solidification. As it was seen from figure 6.22(a), also the weld in experiment C3 showed melted zones. As this melting didn’t occur in the experiments K1, K2, C30 or C1, this indicates that the 2 welds K3 and C3 are not that different from each other.

The conclusion is that the usage of a collar on the inner workpiece has no influence on the forming of the weld for a smaller overlap of the field shaper lF.S.. But, it can influence the process when the field shaper is placed further over the workpiece and hence closer to the collar. As successful welds were created with the workpieces without a collar, it can be concluded that a collar is not required for the radial alignment of the workpieces. In fact, Chapter 6. Experimental results 103 as they are much more easy to manufacture, it can be advised to continue experiments with regular bars without collar as inner workpieces.

(a) A total view of experiment K3.

(b) (c)

(d) (e)

Figure 6.31: Images of the run-in zone of experiment K3, which shows that there was a molten layer joining the pieces but that is cracked open.

Thickness outer tube

The influence of the thickness of the outer tube can be seen by comparing the series C and G for the stand-off distances of 2.5 mm. The welding lengths of these experiments are compared in figure 6.32. From this it becomes clear that the thickness does not influence the length of the weld. Even though the workpieces are less deformable, and they should require more acceleration to reach the inner workpiece, welds are established at the same energy level used Chapter 6. Experimental results 104 as in series C.

Figure 6.32: The influence of the flyer tube thickness on the length of the weld in aluminum- aluminum experiments with a stand-off distance of 2, 5 mm in series C and G.

Stand-oﬀ distance

The stand-off distance is one of the parameters that was varied the most in the experiments. Again, series C is the best reference to illustrate this influence. In this series, welds were successfully achieved for stand-off distances of 2, 0 mm, 2, 5 mm, 3, 0 mm and 3, 5 mm. This series was a continuation of experiments which were done previously. Experiments with a stand-off distance of 1, 5 mm and the same voltage level and positions of the field shaper were found unable to be welded. As can be seen on the graph with the welded lengths (figure 6.33), also for a stand-off distance of 2, 0 mm a weld was only achieved for 1 position of the field shaper. The experiments with a stand-off distance of 3, 5 mm were successful, but the welding lengths were 1 mm shorter and the range for possible positions of the field shaper ± was more narrow. The optimal stand-off distance for aluminum workpieces with a thickness of the outer tube of 1, 5 mm and a voltage level of 15 kV is then in the range 2, 5 mm-3, 0 mm.

Figure 6.33: The inﬂuence of the stand-oﬀ distance on the length of the weld in the aluminum- aluminum experiments in series C. Chapter 6. Experimental results 105

The results of series I (page 137), where aluminum was welded to steel, show that a minimum stand-oﬀ distance of 3, 0 mm is required for this material combination.

The massive deformation that is required to decrease the tube’s diameter and the high stresses which are induced by this deformation, are a possible reason why the welds are of lower quality at higher stand-oﬀ distances. This could also be the reason why the experiments in series B with higher initial angles of the inner workpieces failed. In these cases the stand-oﬀ distance could range up to 4, 3 mm for an initial angle of 12◦.

Energy level

The energy level of an experiment is set through the charging voltage level of the capacitor bank. As the eﬀect of this parameter on the welding process is simply increasing the magnetic pressure, it was chosen to keep it constant in most of the experiments and focus on the inﬂuence of other parameters. Because the experiments prior to this thesis only showed good welds at a voltage level of 15, 0 kV, this level was held throughout most of the experiments.

A voltage of 15, 0 kV is also the maximum voltage level that can be used without further control of the current flow. Because the current running through the coil is an alternating current, the current flow reverses direction after the first half period. That means that the second peak runs back towards the capacitor bank. Due to damping, this peak will be smaller than the first peak. When using voltage levels of more than 15, 0 kV, the ratio of these 2 peaks has to be measured. The maximum allowed voltage level can be calculated from formulas that were provided with the equipment.

The voltage level was increased over 15, 0 kV only in the series I and J. In series I it was chosen to work with higher voltage levels because the experiments between aluminum and steel from series A were not successful. The maximum allowed voltage level here appeared to be 18, 0 kV. In series J the combination of copper to aluminum was tried. The voltage level was increased because welding at a voltage of 15, 0 kV seemed to be more diﬃcult than welding of 2 aluminum pieces. The maximum allowed voltage level here was 18, 5 kV.

As mentioned before, the welded lengths of series I were not measured, but the welds that were successful (page 137) clearly show that the higher voltage levels were necessary to establish a weld.

In the series D and G, it was checked if the voltage level could be lower to establish a weld for aluminum tubes of thickness 1, 5 mm and 2, 0 mm. In both cases, only one experiment was successful at a voltage level of 12, 5 kV. Working with voltage levels lower than 15, 0 kV signiﬁcantly decreases the chance for a good weld for the combination of aluminum to aluminum. Chapter 6. Experimental results 106

In the copper-aluminum experiments of series J, welding was tried to be established first at lower voltage levels, as the literature study showed that copper is easier weldable due to its higher conductivity. At a voltage level of 8, 0 kV, the copper tube did not impact to the inner workpiece. Moreover, only a weld could be formed at voltage levels of 15, 0 kV. Oddly enough, a further increase of the voltage level wasn’t successful. This shows that increasing the voltage level and hence the magnetic pressure, doesn’t necessarily guarantee a good weld. This could be the consequence of a different deformation behaviour due to the different material characteristics of copper flyer tubes. A reason for the failure of these experiments was not found, but the experiments do indicate that the solution probably has to be found in the geometry of the workpieces or the field shaper overlap.

It is remarkable to notice that in series D only the weld at a stand-off distance of 2, 5 mm was achieved for a voltage of 12, 5 kV. The field shaper position in these experiments was chosen in the middle of the range of possible positions that were found in series C. This leads to the conclusion that increasing the voltage level leads to a broader range of stand-off distances.

Materials

3 diﬀerent combinations of materials were tried in the experiments. A broad range of aluminum-aluminum experiments was done, which were extensively discussed in the previous sections. The combinations copper-aluminum and aluminum-steel are compared with these experiments in this section. The steel-steel experiment of series H will not be discussed as impact was not reached for the chosen parameters.

Aluminum - steel

Two series of experiments were done with aluminum as ﬂyer material and steel as inner workpiece material. In series A, no successful welds were achieved with this combination. The previous conclusions show that this is possibly due to 2 reasons: the geometry of the inner workpiece that is more diﬃcult to weld than straight surfaces, and the voltage level that should be higher for aluminum-steel combinations.

The images of these experiments show that melting occurred in the aluminum, while no heat aﬀected zone can be distinguished in steel. Figure 6.34(a) shows a part of the interface in one of the experiments and so the porous molten zone in the aluminum can be seen clearly. When looking closer at the interface (ﬁgure 6.34(a)) it can be seen that a thin aluminum layer is welded to the steel surface. This indicates that it is possible to join these 2 materials, but the molten zone should be avoided as this is believed to be the main reason for the failure. The aluminum has been molten, leaving a broad gap between this molten zone and the steel workpiece, that still has a very thin layer of molten aluminum stuck to it. The aluminum must have impacted to the steel, causing a temperature rise that only melted the aluminum. Chapter 6. Experimental results 107

Due to this, no weld was formed.

As there was no gap left between the collar of the inner workpiece and the outer tube, the jet material couldn’t escape and got stuck behind the collar. Figure 6.34(c) shows that the jet consists mainly out of aluminum that was partly molten, as can be seen from its darker colour.

(a) Part of the interface in one of the expe- (b) riments.

Figure 6.34: Images of the failed experiments in series A. The darkest material in the images is the steel inner workpiece.

The series I with the combination of aluminum to steel was more successful. Several welds were achieved at higher voltage levels. Because a weld wasn’t achieved at a stand-oﬀ distance of 2, 5 mm, and the voltage level was higher than in the aluminum-aluminum experiments, it can be concluded that higher impact velocities are required to join aluminum to steel.

Copper - aluminum

Two series of experiments were done with copper as flyer material and aluminum as inner workpiece material. Experiments with copper were done in series J and in series L. Only 2 Chapter 6. Experimental results 108 welds were achieved for this combination. In both the series, the parameters were the same as experiments in series C and K which showed good welds for the combination of aluminum to aluminum. The successful experiment J2 has been discussed before, and images were already shown in figures 6.17 and 6.25. The reason for the failure of these experiments is not known but as an increase of the energy didn’t help, the solution should be searched for in the geometry of the workpiece and the position of the field shaper.

6.4 Summary

In this thesis, 76 MPW experiments were done, of which 31 resulted in a successful weld. The experiments were mostly aluminum - aluminum, with varying position of the field shaper and stand-off distance. Also the effect of a changing voltage level, thickness of the flyer tube and shape of the inner workpiece was checked. Apart from these, there were also some experiments done with the aluminum - steel and copper - aluminum combinations. The welds were judged on their length, that was retrieved from microscopical investigation.

It was seen that the flyer tube will deform a lot due to the radial compression. This deformation results in large longitudinal and circumferential stresses in the tube, which cause an increase in thickness and buckling along the circumference. Consequently the weld won’t be equal over the circumference, as the buckled parts lag behind during the movement. The residual stress in the flyer tube are high enough to cause the weld to fail after cross-sectioning in case the weld is of lesser quality. The impact will cause large deformation of the inner workpiece. The reaction force of the radial pressure on the collar of the inner workpiece causes it to move inwards the outer tube. This displacement was found to be higher for higher stand-off distances and bigger thicknesses.

During the impact, a lot of material is jetted along the surfaces of the workpieces. This jet consists of material of both the inner workpiece and the flyer tube, and cleans the surface from oxides and contaminants. Pre-cleaning of the workpieces was not required. The jet was blocked by the collar in most of the experiments, but this did not seem to have effect on the quality of the weld. The jet will concentrate in the areas where the flyer tube buckles.

The welds either had a smooth or a wavy interface. Two different patterns were found for the wavy interface: one where both amplitude and wavelength stay more constant along the weld, and one where the wavelength increases and the amplitude reaches a peak value and then decreases again. It was seen that the wave pattern moves with the position of the field shaper. The wavelength increased for increasing stand-off until a certain value of the stand-off. It decreased with increasing thickness of the flyer tube and with decreasing energy level. The wavelengths in copper - aluminum were also found to be much smaller.

In the wavy interfaces of some experiments, zones were found that were molten and solidiﬁed Chapter 6. Experimental results 109 during the process. This results in a very porous zone which easily cracks. These zones, that could be either melt pockets or a continuous molten layer, were also seen in areas were no weld occurred. Although this zone is very narrow and it doesn’t necessarily occur all around the surface, it is clearly a risk for the quality of the weld.

The zone at the interface was seen to have a higher hardness. Also the hardness of the whole ﬂyer tube and up to a depth of at least 5 mm into the inner workpiece increased due to the plastic deformation. The hardness increase was highest for the experiment with a thicker ﬂyer plate.

The collar on the inner workpiece only influences the deformation behaviour of the flyer tube if the field shaper overlap exceeds a certain value. That is, if the field shaper is placed close enough to collar. The use of inner workpieces with a slant surface didn’t offer an advantage over the straight surfaces.

The position of the field shaper was found to be an important parameter, regardless the value of the other parameters. For different experimental layouts, a different range of possible positions of the field shaper exists. The position of the field shaper influences the deformation behaviour and thus both the impact velocity and impact angle.

A change in thickness of the flyer tube of 1, 5 mm to 2, 0 mm didn’t seem to have an effect on the results. There is an optimal range for the stand-off, which equals 2, 5 mm - 3, 0 mm for aluminum tubes of 1, 5 mm thickness. The position of the field shaper is best chosen in the middle of the possible range of positions, which is an overlap of 11 mm.

Welding of aluminum to steel was only possible at voltage levels higher than 15 kV. For welding of aluminum to aluminum, a decrease of the voltage level to values lower than 15 kV led to a much smaller range of possible parameters. For copper however, increasing the voltage level didn’t result in longer welds. The solution to successfully weld copper to aluminum has to be found in the geometry of the pieces and the position of the ﬁeld shaper. Chapter 7

Discussion

In this chapter the results of the welding experiments will be compared with the study of the explosive welding process, the results of other magnetic pulse welding researches, and the series of formulas composed in chapter 4. First the general impact behaviour and melting is discussed, followed by a discussion on the applicability of welding windows from explosive welding. In the ﬁnal section, the applicability of the formulas is checked for diﬀerent parameters, based on the experiments that were done. The results of the literature study correspond perfectly with the observations from the experiments. These will be mentioned throughout this chapter.

7.1 Impact behaviour

From the experiments it could be seen how the outer tube and the inner workpiece are deformed. This helped the understanding of the eﬀect of several parameters. In order to get a better understanding of why the experiments succeed or fail, the impact behaviour and more in particular the impact velocity vc and impact angle α, need to be studied more closely.

The movement of the flyer tube will start when a hinge is formed at a certain point in the tube. When using inner workpieces with a collar, the hinge will form at the collar where it suffers a reaction force (figure 7.1(b)). In the experiments without a collar (figure 7.1(a)) the hinge will form at the location where the yield stress is reached first. For the experiments with aluminum tubes with a thickness 1, 5 mm as flyer workpieces, this point was at 15 mm to 20 mm from the edge of the tube, and will be located further away from the tube end with an increase of the field shaper overlap lF.S.. This explains why the experiments with and without collar only differ at higher field shaper overlap.

The flyer tube will impact first at its edge (point B in the figures 7.1(c) and 7.1(d)). This was concluded from the severe deformation of the inner workpiece in this area. These impact zones are also the evidence that the hinge is formed at point A, outside the welded area. If

110 Chapter 7. Discussion 111 the deformation would be as shown in ﬁgure 7.2, which is more similar to the behaviour in explosive welding, there would be no clear impact point at the inner workpiece.

(a) Forming of a hinge A at the point where the (b) Forming of a hinge A at the collar. yield stress is reached ﬁrst.

(e) Displacement of the hinge and the impact (f) Displacement of the hinge and the impact point. point.

(g) The completed welding process.) (h) The completed welding process.

Figure 7.1: The impact behaviour of the workpieces.

Figure 7.2: Possible way of deformation which is ruled out due to the impact zones on the inner workpieces.

After the first impact, the impact point B will move progressively towards the end of the weld. If the hinge A would stay at its place, the impact angle is very easy to calculate as s tan(α) = l (figure 7.1(d)). The angle will in that case increase along the weld. For several Chapter 7. Discussion 112 experiments that were done, this angle is calculated for the zone where the weld was formed. This led to the ranges of impact angles that would have occurred during the process (see table 7.1). These values are much bigger than the values that are proposed in the welding windows for explosive welding of aluminum (5◦ - 20◦) and the values that were found in literature of magnetic pulse welding (5◦ - 15◦) [5, 18]. Due to this it was concluded that the angles from table 7.1 are too high and hence the deformation behaviour with a fixed hinge A and a moving point B is not correct.

Table 7.1: The values of impact angles along the weld if the hinge A stays at the same location.

Experiment Initial angle [◦] End angle [◦] C25 11,4 33,4 C30 15,9 32,0 C31 13,9 33,6 E1 17,1 43,0 C1 12,6 34,0 C3 13,2 51,3 C6 14,0 34,3 C7 13,9 45,0 G4 22,6 35,5 G5 11,2 28,5

For this reason, it is assumed that the hinge A moves progressively towards the tube end and radial inwards until it reaches the inner workpiece (point A on the ﬁgures 7.1(e) and 7.1(f)). Due to this, the impact angle will be smaller and in the range of the necessary angles found in literature. As the impact point B is moving faster than the hinge A, the impact angle will still increase along the weld, but in a lower range than when the hinge A didn’t move.

The movement of the hinge also explains the influence of the positioning of the field shaper. When the field shaper overlaps more, the total force at the hinge will be bigger and it will move faster towards the inner workpiece. The impact angle will in this case be much smaller. This also explains the wider impact zone that was seen at the beginning of the weld. With a further increase of the field shaper overlap lF.S., the impact angle will become smaller. Also the angle at the end of the weld seems to be bigger for a longer overlap. As the angle is defined by the point where hinge A hits the inner workpiece, this seems also a logical observation.

A possible reason why a weld is not formed is that the impact angle becomes too low and consequently no jet is created. This explains the result of experiment L3, where the jet was only formed at the end of the weld, where the impact angle is bigger than in the beginning.

A second possibility why a weld is not formed at higher ﬁeld shaper overlap is the location of Chapter 7. Discussion 113

the impact point. When the field shaper overlap lF.S. is bigger, the hinge A can start moving before the impact of point B. In that case, the length l on figure 7.1(d) will be shorter. This can be seen on figure 7.3. Due to this shorter length, the flyer tube will need to have a bigger increase in thickness. As this required more energy, the impact energy will decrease and the weld will be smaller. This corresponds perfectly with the observations in experiment C18 (figure 6.9(a)). In this experiment the thickness increase on one side of the sample was bigger. The total overlap length lt on this side was smaller, the weld was shorter and the angle at the end of the weld was bigger. In this case the reason is not a different field shaper overlap, but the asymmetry of the weld. The important thing is that the link between shorter overlap, bigger thickness, higher end angle and shorter weld was seen and can be explained with the deformation behaviour that is explained here.

Figure 7.3: The difference impact points for short field shaper overlap (dashed line - point D) and longer field shaper overlap (full line - point C).

The increase of the wavelength combined with an increase and immediate decrease of the amplitude along the weld correspond, according to the explosive welding theory, with an increasing impact angle. In other words, the waves that were noticed in the experiment match the impact behaviour that is described in the above. More importantly, the similar interfaces in explosive welding show that these techniques are very similar.

7.2 Jet formation

In literature it is claimed that welding is possible without pre-cleaning of the weld, due to the jet of material that will clean the surfaces. This was conﬁrmed by the experiments, as none of the workpieces used were cleaned prior to the welding. It was also conﬁrmed that the jet exists mainly out of material from the workpieces.

7.3 Melting

It was seen in the experiments that there is a considerable risk for melting during the process. The sharp transition zone that was seen is typical for melting and rapid solidification, and is generally not observed in a diffusion bonding process [41]. The literature study of both Chapter 7. Discussion 114 magnetic pulse welding and explosive welding showed that continuous molten layers or melt pockets can occur. In more recent papers, the jet has been found to be the main source for the heating of the materials, apart from the heating due to the collision itself [41]. Also the presence of pores due to the turbulent jet and the rapid solidification have been mentioned before. However, melting has only been called a disadvantage due to the intermetallic compounds it causes to form. These compounds form a hard and brittle interlayer, that is generally susceptible to cracking. In this thesis the voids themselves were so numerous that they induced cracking. As the cracks run parallel with the surfaces, the stresses inducing them should have been perpendicular to them. This means that the shear force that is the cause of the jet formation can’t be the reason for cracking. Most probably, the porous interlayer is not able to withstand the residual stresses in the flyer tube, and therefore cracks. Apart from that, cracks might have induced due to shrinking of the material during solidification.

The temperature of the interface will increase due to the jet, which is dependent on the impact angle, and the impact energy, which is dependent on the impact velocity. That is why in [41] it is mentioned that melting can be avoided by either decreasing the energy level or by decreasing the impact angle. However, the experiments C3, C7 and C20 all showed evidence of melting. These were the experiments of series C with the highest field shaper overlap, so these experiments are believed to have the smallest impact angle. This means that decreasing the impact angle is not an option here to avoid melting. Hence, in order to avoid melting, the energy level should be decreased. Experiments at a lower energy level than the 15 kV used in these experiments, weren’t successful for the welds with a lower field shaper overlap. But, in these experiments with a larger overlap, the magnetic pressure will reach over a longer zone of the flyer tube and hence increase the impact velocity. It was already mentioned that the field shaper overlap influences the impact velocity because of this. Probably, these high velocities due to the long field shaper overlap are the reason for melting.

7.4 Hardness increase

The experiments showed a hardness increase of the interlayer and both the workpieces that was of the same order than the values for aluminum - aluminum experiments in [17]. The reason for the lower increase in the ﬁrst 2 mm of experiments C1 and D1 was found in [2]. The heating causes softening of the material of the area adjacent to the interface. The hardness decreases in this area due to dynamic recovery and recrystallisation, which are both thermally activated mechanisms. Chapter 7. Discussion 115

7.5 Comparison with welding windows

The welding windows of explosive welding can be used for a comparison of different material combinations relative to one another. The exact values given in the welding windows won’t be used because the values that were available were for plates with higher thicknesses. However, they can give an idea of the range of impact velocities vc and impact angles α that are required. the following general effects were observed in the welding windows. An example of the welding windows for aluminum - steel for different flyer material thicknesses is shown in figure 7.4.

The range of impact angles is much wider for copper than for aluminum.

Copper (range 250 m/s - 700 m/s) can be welded at lower impact velocities than aluminum (range 400 m/s - 1000 m/s).

Aluminum - steel is welded at higher impact velocities than aluminum - aluminum.

Copper - steel is welded at higher impact velocities than copper - copper.

For higher thicknesses of the ﬂyer material the impact velocity can be lower.

Figure 7.4: Welding windows for aluminum - steel for diﬀerent ﬂyer piece thicknesses [32]. Chapter 7. Discussion 116

In this thesis it was confirmed that copper - aluminum and aluminum - steel combinations can be welded with the MPW process. However, they seem to be more difficult to weld than aluminum - aluminum workpieces. In other works, these combinations both show intermetallic phases in the interlayer. This interlayer is hard and brittle and often shows cracks. Reports of aluminum - steel combinations mention these cracks as being a general effect for this combination [2, 22]. The interfaces of these weld combinations in this thesis were not examined. Welding windows for combinations with aluminum or copper as flyer material will be compared to the results of the experiments in what follows.

In the welding windows that are given in [32] it was seen that the range for the impact angle for copper flyer tubes is much wider than for aluminum flyer tubes. In the experiments however, a small change in the position of the field shaper could lead to failure of the weld in experiments with copper as the flyer tube. The experiment J2 showed a good weld, but the experiment L1 with exactly the same parameters and without collar on the inner workpiece didn’t show a good weld. Removing the collar showed only one possible effect in the experiments, and that was a change of the impact angle. This might indicate that the impact angle does have an important influence on the quality of copper - aluminum welds, which is contrary to the conclusion from the welding windows. More experiments for the copper - aluminum combination with varying parameters are required to make general conclusions.

Apart from that, the copper - aluminum also showed a remarkable behaviour because experiments J1 and J9 failed. These were performed with the same parameters as the successful experiment J2, apart from the voltage level that was decreased to 12, 5 kV in J1 and increased to 17 kV in J9.

Because the results of the copper - aluminum experiments are not as expected, the option that the welds failed after cross-sectioning of the workpieces should be considered. As copper has an elastic modulus that is twice as big as the one of aluminum, the residual stresses due to the deformation will be twice as high. This could have led to the fracture of the welds during the cross-sectioning.

The copper-aluminum experiment that was successful(J2) was conducted at the same energy level as the ones for aluminum. Due to the higher density of copper, the impact velocity must have been lower. This eﬀect can also clearly be seen in the formulas. This conﬁrms the conclusion from the welding window that copper - aluminum experiment require a lower impact velocity.

The welding windows in [32] also showed that the required impact velocity of aluminum - steel welds is higher than for aluminum - aluminum welds. This was conﬁrmed in the experiments, as the energy level had to be raised to successfully weld aluminum to steel. Chapter 7. Discussion 117

The welding windows also show the general effect that the required impact angle and impact velocity are lower for thicker workpieces. Not enough experiments are done to draw general conclusions regarding the impact angle, but the results for the impact velocity were confirmed in the experiments. Although their higher thickness, the experiments in series G had the same minimal energy level as the experiments in series C with the same stand-off distance. The impact velocity must have been lower due to the lower deformability and higher weight of the workpieces in series G, so this confirms the conclusion from the welding window.

7.6 Inﬂuence of the parameters and comparison with the formulas

In order to check their accuracy, in this section the formulas discussed in chapter 4 will be applied to the experiments that were conducted. The inﬂuence of the parameters is discussed, together with the applicability of the formulas.

7.6.1 Calculated impact velocities

The only parameters that can be set in the formulas are the material properties, the stand-off distance, the thickness of the flyer tube and the charging voltage level. This leads to 6 values of the impact velocity for the experiments that were done. The voltage level that was used to calculate these velocities is the lowest charging voltage level that succeeded to weld the workpieces. They are shown in table 7.2. The velocities at lower stand-off distances are all realistic values, when comparing with the welding windows of [32]. These gave for aluminum a range 400 m/s - 1000 m/s and for copper a range 250 m/s - 700 m/s. Also the ratio of the component of the pressure required for the deformation to the total required pressure is calculated with the formulas.

Table 7.2: The impact velocities in the experiments according to the calculations.

Outer Inner t s V v P1 c P1+2 [mm] [mm] [kV] [m/s] [% ] Al Al 2,0 2,5 12,5 470 6,9 Al Al 1,5 2,5 12,5 545 5,2 Al Al 1,5 3,0 15,0 850 2,6 Al Al 1,5 3,5 15,0 1050 2,0 Cu Al 1,5 2,5 15,0 365 2,3 Al steel 1,5 2,5 17,0 750 2,8 Chapter 7. Discussion 118

7.6.2 Inﬂuence of the thickness of the outer tube

As it was described in chapter 5, thicker workpieces have a higher weight, and therefore their impact energy will be higher. But on the other hand, there is more energy required to deform the workpieces, which causes a decrease of the impact velocity. Consequently, the impact velocity can be lower for these workpieces. This decrease of the impact velocity is also seen in the formulas. That these thicker workpieces form a weld could be seen from the experimental results, as both the thicknesses 1, 5 mm (series C and D) and 2, 0 mm (series G) hold the same minimal voltage level. This eﬀect was also seen in the welding windows of explosive welding.

7.6.3 Inﬂuence of the stand-oﬀ distance

It was experimentally observed that the stand-off distance has an optimal value for a certain combination of workpieces. This corresponds with the results found in the literature [17]. The reason for the optimal value is that the impact velocity will be lower for higher distances. This was not the result of the acceleration peak that is over before impact, but of the flyer tube’s deformation. The more the flyer tube has to be compressed, the more energy that will be lost in its deformation and consequently the impact velocity will decrease.

Because the values in table 7.2 are the minimal voltage levels that allowed welding, the impact velocities should equal the minimal velocity that is required. Because the same material combination is used in these experiments, these values are expected to be the same. Hence, the big differences at different stand-offs proof that these values are not accurate.

Based on the formulas in chapter 4 this optimal impact velocity value cannot be calculated. The deformations due to the radial compression are not taken into account, as the pressure required for deformation does not take the circumferential and longitudinal stresses into acount. This means that the higher the stand-oﬀ distance, the less accurate the formulas become. The impact velocity will be overestimated because the deformations in the ﬂyer tube are neglected.

7.6.4 Materials

The formulas show that the impact velocity for a copper tube will be much lower than for an aluminum tube with identical dimensions at the same voltage level. This means that the inﬂuence of the higher mass density of copper is more important than the inﬂuence of the lower yield stress. Not only because the density increase is bigger than the decrease of the yield stress, but also because the pressure component for the acceleration is much higher than the pressure required for the deformation. Chapter 7. Discussion 119

7.6.5 Position ﬁeld shaper

The position of the ﬁeld shaper was found to be a very important parameter in the experiments. It was seen that it aﬀects both the impact velocity and the impact angle. No reports on this parameter have been found in the literature study, and neither have they been included in the formulas.

The magnetic pressure is calculated in the formulas per unit of length. This pressure will occur along the length of the field shaper. Only the zone where the field shaper overlaps is compressed by the magnetic pressure, so theoretically only here the deformation and acceleration can take place. The zone where the field shaper doesn’t overlap has to be compressed by the pressure on the rest of the tube, which means that the impact velocity will be lower and the value that is expected is overestimated.

7.6.6 Applicability of the formulas

The stress component that is used in the formulas is the resulting stress from both the circumferential and the radial component. It is assumed in the formulas that the yield stress has to be reached everywhere in the tube, but that is not correct. The deformation behaviour shows that the flyer tube only yields at the point where the hinge moves and of course during impact. The large radial compression induces longitudinal and circumferential stresses in the flyer tube. These were seen to influence the deformation behavior by causing a thickness increase of the flyer tube and buckling effect along the circumference, and leave large residual stresses. This deformation requires a lot of energy and consequently reduces the velocity of the flyer tube, but was not explicitly taken into account in the formulas.

At lower deformations this buckling and thickness increase is limited, which explains the realistic values of the impact velocity that are found. However, for workpieces that require more deformation, for example with higher stand-oﬀ distances or higher thicknesses, the formulas will be less accurate. Hence, the formulas can serve as a tool to give indications on the process parameters at lower deformations, but a more realistic approach of the pressure component P1 is required before using the formulas to ﬁnd exact values of the impact speed.

Replacing the yield stress σy by the ultimate tensile stress σt might increase the accuracy.

Also the strain rate dependence of the stresses is not taken into account, which is a severe simpliﬁcation at these high deformations in such a short time.

Another problem for applying the formulas is the lack of the effect of the impact angle. It was seen that the influence of the impact angle is very important in magnetic pulse welding, but this is not taken into account in the formulas. The influence of for example slant inner workpieces or the position of the field shaper can’t therefore be taken into account. Chapter 7. Discussion 120

When doing the same calculations without the factor 2 that was added in chapter 4, the impact velocities would be much higher than the ones calculated here. For example the aluminum tube of 1, 5 mm at a voltage of 12, 5 kV would impact at 770 m/s in stead of 545 m/s, and at a voltage of 17 kV it would impact at 1055 m/s in stead of 750 m/s. These values however are rather high in comparison with the minimal values from the welding windows. The range found there for aluminum was 400 m/s - 1000 m/s. This indicates that the values that were presented in table 7.2 were more accurate.

7.6.7 Ratio of the pressure components

When looking at the ratio of the pressure required for the deformation to the total pressure P deform , it can be seen that this component is always lower than 10%. Hence, most of the Ptotal energy is used to accelerate the workpieces according to these formulas. However, for workpieces with higher deformation (lower stand-oﬀ distance or lower thickness) the deformation component of the pressure is most likely underestimated. This leads to the conclusion that the pressure for the acceleration is higher than for the deformation in, with a decreasing diﬀerence for higher deformations.

7.7 Wave formation

The explosive welding theory showed that the wavelength of the interface waves will increase for an increasing impact velocity (as long as it’s lower than the sound velocity in the material). For an increasing impact angle the amplitude of the waves will ﬁrst increase and immediately decrease, while the wavelength will only show an increase. These eﬀects will now be compared with the waves that were seen in the experiments.

The wavelength showed an increase for an increase of the stand-off distance upto a value of 3, 0 mm, and decreased again for a stand-off of 3, 5 mm. This indicates that the impact velocity that was reached at a stand-off of 3, 0 mm was the highest possible. Therefore, a stand-off of 3, 0 mm is the optimal value for aluminum tubes of thickness 1, 5 mm.

The thicker tubes from series G showed a smaller wavelength in comparison with series C, which indicates a lower impact energy. The ratio of the masses of the tubes in series G and C 4 470 is 3 and the ratio of the impact velocities according to the formulas is 545 . The ratio of their · 2 EG kinetic energy E = m vc then is 1. The diﬀerence in wavelength can be explained by EC ≈ the higher deformation that is required for thicker pieces but not taken into account in the formulas. This shows that neglecting these deformations leads to higher overestimation of the impact velocity for thicker workpieces.

The lower wavelength for experiment D1 in comparison with C1, C30 and C31 conﬁrm that the impact velocity was lower due to the lower voltage level. The low wavelength in the Chapter 7. Discussion 121 copper - aluminum weld of experiment J2 shows that the velocity of the copper tube must have been much lower, which is conﬁrmed by the formulas.

The changing of the wave pattern with different positions of the field shaper in the experiments C5, C6 and C7 indicate that the range of impact angles must have changed. This confirms that the position of the field shaper reduces the impact angle, so that the values required to form the waves move towards the end of the weld.

7.8 Summary

Based on the required impact angles from the literature study and the welding windows in explosive welding, a detailed deformation behaviour of the flyer tube is presented. The deformation will start when a hinge is formed in the flyer tube. This explained why the experiments with and without collar are different at a higher field shaper overlap. The end of the flyer tube will impact the inner workpiece first, and then both the hinge at the end of the weld and the impact point will move until the weld is formed. The field shaper overlap influences the location of the impact point, which has a direct influence on the length of the weld.

The wave forms that were found in the experiments correspond with the waveforms that are typically seen in explosive welding. This indicates that the 2 techniques are very similar.

Both the cleaning of the surface due to jet formation, and the hardness increase of the workpieces were conﬁrmed by the experiments. Melting has been found in the literature to form a risk due to brittle intermetallic compounds it causes to form. In this thesis, the pores that resulted from the solidiﬁcation of the melted zones were found to contribute to cracking of the weld.

Welding windows from explosive welding can be used to compare the parameters for different material combinations, relative to each other. It was confirmed that the aluminum - steel combination requires higher energies than the aluminum - aluminum combination, and that thicker flyer tubes and flyer tubes made out of copper require lower impact velocities.

The results of the copper - aluminum experiments were not as expected. It is therefore possible that the residual stresses in the copper tube resulted in failure of the weld after cross-sectioning.

The formulas appeared to be useful to give an idea of the effect of the process parameters, or to roughly estimate the impact velocities for chosen energy levels. However, they can’t be used to calculate exact values. The effect of the field shaper for example cannot be taken into account, and the deformation is underestimated because circumferential and longitudinal Chapter 7. Discussion 122 stresses are ignored. Due to this, the stand-off distance doesn’t show an optimal value like it was seen in the experiments. The pressure required for the acceleration appeared to be more important than the pressure for the deformation.

The wavelengths that were measured confirm that lower impact velocities occur for copper tubes and for experiments at a lower voltage level. They also confirm the reducing of the impact angle when increasing the field shaper overlap. They also showed that for thicker workpieces, the underestimation of the deformations is bigger. The stand-off distance reaches an optimal value of 3, 0 mm for aluminum tubes of thickness 1, 5 mm. Chapter 8

Conclusion

The results of this thesis give more insight in the deformation behaviour and the influence of the different parameters. It is promising that these conclusions correspond with the results from explosive welding and that the formulas are, although not accurate, useful for the estimation of the influence of a change in voltage level or the effect of other material and geometrical parameters. The MPW process seems promising, but joining two workpieces wasn’t found to be that easy as it is generally represented. Special care has to be taken for welds with molten and solidified zones, as these can be very porous and a danger for the quality of the weld.

The results of the experiments all corresponded with the eﬀects of magnetic pulse welding that were found in the literature study. They also corresponded with the characteristics of the interface that were seen in the literature study of explosive welding. The series of formulas that was proposed can give a good idea of the impact velocity and the eﬀect of some of the process parameters, but can’t be used to retrieve desired values of the voltage level as they are not accurate enough.

The experiments were mostly done with a combination of aluminum flyer tubes and aluminum inner workpieces. Apart from that, also copper - aluminum and aluminum - steel workpieces were successfully welded. The experiments with copper flyer tubes were more difficult to weld than expected, as only 1 of the experiments was successful. The effect of the most important process parameters, being the position of the field shaper, the material, the thickness of the flyer tube, the stand-off distance and the voltage level, was discussed. For aluminum tubes of thickness 1, 5 mm, the optimal value of the stand-off was 3, 0 mm and the optimal field shaper overlap was 11 mm. The conclusions regarding the thickness of the flyer tube, the material and the voltage level corresponded with both the results from the literature study of explosive welding and the series of formulas. The position of the field shaper was a parameter that hadn’t been encountered during the literature study, and its effect is also not taken into account in the formulas. However, this parameter seemed to be of great importance in

123 Chapter 8. Conclusion 124 the process. The eﬀect of the stand-oﬀ distance corresponded with the conclusions from the literature study, but this wasn’t seen in the formulas.

The workpieces showed an increase in hardness due to the deformation, as was to be expected from the literature study. The jet formation indeed seemed to be sufficient to clean the surfaces of the workpieces and eliminate the need for pre-cleaning. In some of the experiments zones were observed that had been molten and solidified. Due to their high porosity, these are very susceptible to cracking. This porous molten interlayer, combined with its brittleness and the high residual stresses in the flyer tube form a risk for the quality of the weld and should be avoided.

A more extended overview of the results can be found in the summaries of the experimental results, section 6.4 and the discussion about the results, section 7.8. Chapter 9

Advice for future experiments

9.1 More material combinations

In this thesis experiments were done mostly with the combination of aluminum workpieces. As the eventual goal of the research is to find an applicable range of parameters for different material combinations with different dimensions, more experiments should be done with these other combinations. For now it seems for the best to keep using copper and aluminum as flyer tube, and use these in combination with inner workpieces of copper, aluminum, steel and stainless steel, as these are the materials that are most interesting for industrial applications. After mastering these combinations sufficiently, the combinations stainless steel - stainless steel and steel - steel can be studied. During these experiments, this thesis should be used as a tool in the understanding of the process and its various parameters. The new experiments will undoubtedly improve the insights on the impact behaviour and the effect of the parameters.

Microscopical examination of the workpieces are in these experiments deﬁnitely required. Special attention should be paid to melting zones and intermetallic phases, and whether these are or aren’t inﬂuenced by the process parameters.

9.2 Use of smaller ﬁeld shaper

From the experiments that were done in this thesis, it appeared that not the whole length of the field shaper is required for successful welding. It might therefore be interesting to work with a smaller field shaper in order to concentrate the magnetic field more in the desired zone. This would allow to reach higher pressures for lower energy levels. This might even appear to be necessary, taking in account that the experiments with the aluminum - steel combination were already close to the maximal possible energy level.

125 Chapter 9. Advice for future experiments 126

9.3 Adjusting workpieces

In order to get a better understanding of the process parameters and the deformation behaviour, it might be interesting to do experiments with slightly adjusted workpieces. It could be interesting to do experiments with 2 cuts in the flyer tube along the length of the working zone (figure 9.1), so that the tube loses it’s cylindrical behaviour in the deformed area. This can give an idea of the importance of the circumferential stresses in the workpiece. If this works for example for the copper - aluminum experiments J9 and J10 that failed, it would verify that these welds failed due to the residual stresses that were set free after cross-sectioning. Or, if this decreases the minimal voltage level, this would show the negative influence of the circumferential stresses on the impact velocity.

Figure 9.1: The ﬂyer tube with 2 cuts along the length of the working zone, seen from the side and from the front.

9.4 Testing methods

In this thesis the welds were judged on their length. However, strength tests are required to judge the quality of these welds. Ideally these tests should result in the shear strength of the weld. Several tests are mentioned in the literature, but so far there are no international specifications defining the quality control of the MP-Weld process [4]. [5] mentions axial or torsional tensile shear test, shear strength test of a sample section, peel test, pressure or vacuum leak test, vibration test, thermal shock test and the pressure burst test. Leak testing is performed by connecting parts to a supply of pressurized gas (for example air, helium). Leaks in the welded area will indicate distribution of welded and non-welded areas. Peel testing is performed by mechanically rolling back a strip section of the tube using a plier tool. Usually welded tubes cannot be separated. The peel test, leak test and pressure burst test are often applied in to explosive welded workpieces [4]. The main issue when dealing with these workpieces is their small scale, which makes them difficult to use in the available testing equipment.

It should also be taken into account that the welded length is generally more than twice as Chapter 9. Advice for future experiments 127 long as the thickness of the flyer tube. When using for example a torsional shear test it is most likely that the flyer tube will fail first due to its lower thickness, even if the shear strength of the weld would be lower. This was noticed in [16].

9.5 Numerical simulations

The series of formulas that was presented wasn’t accurate enough to realistically link the charging voltage level to the impact velocity. The more complex deformation behaviour of the tube, including a strain and a stress strain dependence could not be simplified to the general yielding of the working zone. In order to find a realistic value of the impact velocity starting from the geometrical parameters, material characteristics and a voltage level, the deformation behaviour should be taken into account without simplifications. A manual calculation is therefore impossible, and numeric simulations will be required. These simplifications can then be verified by experimental results. Appendix A

Experimental results

A.1 Series A: Aluminum to steel

Figure A.1: Experimental layout of series A.

Number V s lF.S. β Weld? [kV] [mm] [mm] [◦] 1 15 2.0-3.0 11 4 NO 2 15 2.0-3.0 14 4 NO 3 15 2.0-3.0 17 4 NO 4 15 2.0-3.6 9 8 NO 5 15 2.0-3.6 11 8 NO 6 15 2.0-3.6 14 8 NO 7 15 2.0-4.3 9 12 NO 8 15 2.0-4.3 11 12 NO 9 15 2.0-4.3 14 12 NO

128 Appendix A. Experimental results 129

A.2 Series B: Aluminum to aluminum

Figure A.2: Experimental layout of series B.

Number V s lF.S. β Weld? li lw lo lt [kV] [mm] [mm] [◦] [mm] [mm] [mm] [mm] 1 15 2.0-3.0 9 4 YES 7.1 3.5 3.1 13.7 2 15 2.0-3.0 11 4 YES 1.5 8.9 2.1 12.5 3 15 2.0-3.6 9 8 YES 3.8 5.8 2.4 12.0 4 15 2.0-3.6 11 8 NO 0.0 0.0 0.0 0.0 5 15 2.0-4.3 9 12 YES 3.2 4.0 4.8 12.0 6 15 2.0-4.3 11 12 NO 0.0 0.0 0.0 0.0 Appendix A. Experimental results 130

A.3 Series C: Aluminum to aluminum

Figure A.3: Experimental layout of series C. Appendix A. Experimental results 131

Number V s lF.S. Weld? li lw lo lt [kV] [mm] [mm] [mm] [mm] [mm] [mm] 23 15 2.0 6 NO 0.0 0.0 0.0 0.0 24 15 2.0 8 NO 0.0 0.0 0.0 0.0 25 15 2.0 9 YES 3.3 7.4 2.5 13.2 26 15 2.0 10 NO 0.0 0.0 0.0 0.0 27 15 2.0 12 NO 0.0 0.0 0.0 0.0 28 15 2.0 14 NO 0.0 0.0 0.0 0.0 29 15 2.5 6 NO 0.0 0.0 0.0 0.0 30 15 2.5 8 YES 3.8 6.2 2.0 12.0 31 15 2.5 9 YES 4.2 5.4 3.0 12.6 1 15 2.5 10 YES 3.7 7.5 2.9 14.1 3 15 2.5 12 YES 1.3 9.4 2.9 13.6 8 15 2.5 14 NO 0.0 0.0 0.0 0.0 4 15 3.0 8 YES 7.6 3.5 2.4 13.5 5 15 3.0 9 YES 4.8 9.2 2.4 16.4 6 15 3.0 10 YES 4.4 7.6 2.2 14.2 7 15 3.0 12 YES 3.0 9.1 2.3 14.4 12 15 3.0 14 NO 0.0 0.0 0.0 0.0 16 15 3.5 8 NO 0.0 0.0 0.0 0.0 18 15 3.5 10 YES 5.5 6.5 2.0 14.0 20 15 3.5 12 YES 5.3 7.7 2.4 15.4 21 15 3.5 14 NO 0.0 0.0 0.0 0.0 Appendix A. Experimental results 132

A.4 Series D: Aluminum to aluminum

Figure A.4: Experimental layout of series D.

Number V s lF.S. Weld? li lw lo lt [kV] [mm] [mm] [mm] [mm] [mm] [mm] 1 12.5 2.5 11 YES 4.1 6.6 3.1 13.8 2 10.0 2.5 10 NO 0.0 0.0 0.0 0.0 3 11.5 2.5 10 NO 0.0 0.0 0.0 0.0 5 12.5 3.0 10 NO 0.0 0.0 0.0 0.0 6 10.0 3.0 10 NO 0.0 0.0 0.0 0.0 8 12.5 3.5 10 NO 0.0 0.0 0.0 0.0 9 10.0 3.5 10 NO 0.0 0.0 0.0 0.0 Appendix A. Experimental results 133

A.5 Series E: Aluminum to aluminum

Figure A.5: Experimental layout of series E.

Number V Weld? li lw lo lt [kV] [mm] [mm] [mm] [mm] 1 15 YES 3.1 4.4 4.0 11.5 Appendix A. Experimental results 134

A.6 Series F: Aluminum to steel

Figure A.6: Experimental layout of series F.

Number V Weld? [kV] 1 15 NO Appendix A. Experimental results 135

A.7 Series G: Aluminum to aluminum

Figure A.7: Experimental layout of series G.

Number V lF.S. Weld? li lw lo lt [kV] [mm] [mm] [mm] [mm] [mm] 2 15.0 17 NO 0.0 0.0 0.0 0.0 4 15.0 12 YES 3.6 2.6 9.2 15.4 5 15.0 10 YES 4.6 8.0 2.7 15.3 6 15.0 8 YES 7.3 4.8 2.5 14.6 7 12.5 12 YES 5.7 5.6 3.6 14.9 8 12.5 10 NO 0.0 0.0 0.0 0.0 9 12.5 8 NO 0.0 0.0 0.0 0.0 Appendix A. Experimental results 136

A.8 Series H: Steel to steel

Figure A.8: Experimental layout of series H.

Number V Weld? [kV] 2 15 NO Appendix A. Experimental results 137

A.9 Series I: Aluminum to steel

Figure A.9: Experimental layout of series I.

Number V s lF.S. Weld? [kV] [mm] [mm] 1 15 3.0 10 NO 2 17 3.0 10 YES 3 18 3.0 10 YES 4 15 3.0 12 NO 5 17 3.0 12 NO 6 18 3.0 12 YES 7 18 2.5 10 NO 8 18 2.5 12 NO Appendix A. Experimental results 138

A.10 Series J: Copper to aluminum

Figure A.10: Experimental layout of series J.

Number V s lF.S. Weld? li lw lo lt [kV] [mm] [mm] [mm] [mm] [mm] [mm] 1 12.5 2.5 10 NO 0.0 0.0 0.0 0.0 2 15.0 2.5 10 YES 6.3 3.4 4.2 13.9 3 8.0 2.5 10 NO 0.0 0.0 0.0 0.0 4 15.0 2.5 14 NO 0.0 0.0 0.0 0.0 5 15.0 2.5 12 NO 0.0 0.0 0.0 0.0 9 17.0 2.5 10 NO 0.0 0.0 0.0 0.0 10 18.0 2.5 10 NO 0.0 0.0 0.0 0.0 6 15.0 3.0 10 NO 0.0 0.0 0.0 0.0 7 15.0 3.0 12 YES ? ? ? ? 8 15.0 3.0 10 NO 0.0 0.0 0.0 0.0 11 17.0 3.0 10 NO 0.0 0.0 0.0 0.0 12 18.0 3.0 10 NO 0.0 0.0 0.0 0.0 Appendix A. Experimental results 139

A.11 Series K: Aluminum to aluminum

Figure A.11: Experimental layout of series K.

Number V lwp lF.S. Weld? li lw lo lt [kV] [mm] [mm] [mm] [mm] [mm] [mm] 1 15 18 8 YES 2.8 5.4 2.6 10.8 2 15 18 10 YES 2.7 8.9 2.9 14.5 3 15 18 12 YES 2.1 2.9 9.5 14.5 4 15 18 14 NO 0.0 0.0 0.0 0.0 5 15 16 10 YES 2.5 7.5 2.0 12.0 6 15 20 10 YES 2.6 6.6 2.9 12.1 Appendix A. Experimental results 140

A.12 Series L: Copper to aluminum

Figure A.12: Experimental layout of series L.

Number V lF.S. Weld? [kV] [mm] [mm] 1 15 10 NO 2 15 12 NO 3 15 14 NO Appendix B

Microhardness tests

B.1 Microhardness results C1

(a) The location of the measuring points in (b) The location of the points viewed from the sample. closer to the weld.

Figure B.1: The measuring points of the microhardness test on sample C1.

Table B.1: Reference measuring points in the base material.

Flyer tube Inner workpiece Number Microhardness Number Microhardness [HV0,2] [HV0,2] 1 88,0 1 90,4 2 91,7 2 90,4 3 88,9 3 95,0

141 Appendix B. Microhardness tests 142

Table B.2: The microhardness results for experiment C1. The points of measurement are localized according to their distance from the weld.

Left column Right column Distance from weld Microhardness Distance from weld Microhardness [mm] [HV0,2] [mm] [HV0,2] -1,65 86,9 -1,5 92,8 -1,35 93,8 -1,2 95,4 -1,05 91,1 -0,9 94,2 -0,75 94,4 -0,6 90,5 -0,45 94,4 -0,3 92,6 -0,15 93,8 0,0 95,8 0,15 91,5 0,3 92,8 0,45 91,9 0,6 98,6 0,75 94,2 0,9 92,6 1,05 94,0 1,2 93,2 1,35 96,9 1,5 93,8 1,65 94,8 1,8 95,6 1,95 97,1 2,1 97,1 2,25 92,6 2,4 94,4 2,55 98,8 2,7 96,7 2,85 97,3 3,0 96,9 3,15 96,9 3,3 94,6 3,45 93,3 3,6 95,8 3,75 96,9 3,9 96,7 4,05 95,6 4,2 95,4 4,35 94,6 4,5 93,0 4,65 98,8 4,8 86,8 4,95 94,8 5,1 94,5 Appendix B. Microhardness tests 143

B.2 Microhardness results D1

(a) The location of the measuring points in (b) The location of the points viewed from the sample. closer to the weld.

Figure B.2: The measuring points of the microhardness test on sample D1.

Table B.3: Reference measuring points in the base material.

Flyer tube Inner workpiece Number Microhardness Number Microhardness [HV0,2] [HV0,2] 1 89,1 1 89,1 2 92,5 2 88,5 3 87,4 3 94,6 Appendix B. Microhardness tests 144

Table B.4: The microhardness results for experiment D1. The points of measurement are localized according to their distance from the weld.

Left column Right column Distance from weld Microhardness Distance from weld Microhardness [mm] [HV0,2] [mm] [HV0,2] -1,65 94,0 -1,55 91,7 -1,35 92,5 -1,25 91,3 -1,05 93,8 -0,95 93,6 -0,75 94,8 -0,65 94,0 -0,45 90,7 -0,35 90,2 -0,15 93,8 -0,05 98,1 0,15 96,9 0,00 103,0 0,45 92,1 0,25 91,4 0,75 91,3 0,55 93,0 1,05 89,8 0,85 94,8 1,35 89,1 1,15 91,1 1,65 89,8 1,45 90,2 1,95 94,2 1,75 93,2 2,25 96,0 2,05 92,3 2,55 94,2 2,35 94,1 2,85 92,6 2,65 94,6 3,15 92,5 2,95 86,3 3,45 95,6 3,25 93,6 3,75 94,6 3,55 93,8 4,05 98,3 3,85 92,6 4,35 98,3 4,15 96,3 4,65 94,2 4,45 96,3 4,95 92,5 4,75 88,5 5,05 98,6 Appendix B. Microhardness tests 145

B.3 Microhardness results G5

(a) The location of the measuring points in (b) The location of the points viewed from the sample. closer to the weld.

Figure B.3: The measuring points of the microhardness test on sample G5.

Table B.5: Reference measuring points in the base material.

Flyer tube Inner workpiece Number Microhardness Number Microhardness [HV0,2] [HV0,2] 1 84,3 1 106,0 2 84,8 2 103,0 3 88,7 3 94,5 Appendix B. Microhardness tests 146

Table B.6: The microhardness results for experiment G5. The points of measurement are localized according to their distance from the weld.

Left column Right column Distance from weld Microhardness Distance from weld Microhardness [mm] [HV0,2] [mm] [HV0,2] -2,1 88,0 -1,9 90,2 -1,8 94,2 -1,6 94,4 -1,5 96,0 -1,3 92,8 -1,2 94,0 -1,0 94,2 -0,9 92,8 -0,7 97,5 -0,6 97,5 -0,4 98,2 -0,3 98,6 -0,1 95,4 0,0 98,6 0,2 122,0 0,3 122,0 0,5 121,0 0,6 119,0 0,8 117,0 0,9 114,0 1,1 119,0 1,2 112,0 1,4 113,0 1,5 117,0 1,7 119,0 1,8 116,0 2,0 114,0 2,1 120,0 2,3 114,0 2,4 119,0 2,6 119,0 2,7 114,0 2,9 115,0 3,0 116,0 3,2 114,0 3,3 116,0 3,5 113,0 3,6 114,0 3,8 113,0 3,9 114,0 4,1 113,0 4,2 118,0 4,4 111,0 4,5 112,0 4,7 113,0 4,8 117,0 5,0 111,0 5,1 114,0 5,3 113,0 5,4 118,0 5,6 104,0 Bibliography

[1] Stephan W. Kallee, Steve A. Westgate, and Mathew Amos. Magnetic pulse welding as an enabler of light-weighting in the automotive industry. 1st Technical Conference on Industrialized Magnetic Pulse Welding and Forming, Munich, 3 July 2008. 52 slides.

[2] M. Marya, S. Marya, and D. Priem. On the characteristics of electromagnetic welds between aluminium and other alloys. Welding in the World, 49(5/6):74–84, 2005.

[3] Solid State Welding. http://www.keytometals.com/Article51.htm. webpage visited on 23/12/2008.

[4] V. Shribman. Magnetic pulse welding of automotive hvac parts. PULSAR Ltd. Magnetic Pulse Solutions, July 2006.

[5] V. Shribman. Magnetic pulse technology for improved tube joining and forming. Tube & Pipe Technology, pages 91–95, November/December 2006.

[6] S.D. S.D. Kore, P.P. Date, and Kulkarni S.V. Eﬀect of process parameters on electromagnetic impact welding of aluminum sheets. International Journal of Impact Engineering, November 2006.

[7] V. Shribman and B. Spitz. Magnetic pulse welding for tubular applications: Discovering new technology for welding conductive materials. The Tube & Pipe Journal, July 2001.

[8] S.D. Kore, P.P. Date, and Kulkarni S.V. Electromagnetic impact welding of aluminum to stainless steel sheets. Journal of Materials Processing Technology, pages 486–493, November 2008.

[9] A. Zhang, M. Murata, and H. Suzuki. Eﬀects of various working conditions on tube bulging by electromagnetic forming. Journal of Materials Processing Technology, 48:113– 121, 1995.

[10] Yu. Ya. Reutov. Physical interpretation of magnetostatic shielding. Russian Journal of Nondestructive Testing, 36(2):117–126, February 2000.

147 Bibliography 148

[11] James Colotti. Emc design fundamentals. Telephonics - Command Systems Division presentation, 2005. 70 slides.

[12] Robert B. Cowdell. Nomograms simplify calculation of magnetic shielding eﬀectiveness. EDM Magazine, February 1972.

[13] Haiping Yu Yu, Chunfeng Li, Zhiheng Zhao, and Zhong Li. Eﬀect of ﬁeld shaper on magnetic pressure in electromagnetic forming. Journal of Materials Processing Technology, 168:245–249, 2005.

[14] M. Kashani, T. Aizawa, and K. Okagawa. Application of magnetic pulse welding for aluminium alloys and spcc steel sheet joints. Welding Journal, 86, May 2007.

[15] Glenn S. Daehn. Asm handbook, volume 14b, metalworking: Sheet forming - appendix d. ASM International, pages 405–418, 2006.

[16] V. Shribman, A. Stern, Y. Livshitz, and O. Gafri. Magnetic pulse welding produces high-strength aluminium welds. Welding Journal, April 2002.

[17] Isao Masmoto, Koreaki Tamaki, and Masatoshi Kojima. Electromagnetic welding of aluminum tube with aluminum-and dissimilar metal cores (studies on electromagentic welding, report 1). Transactions of the Japan Welding Society, 12(2):69, 1981.

[18] M.S. Peihui Zhang. Joining enabled by high velocity deformation. Phd thesis, The Ohio State University, 2003.

[19] S.A. Nenakhov. Basic terms and deﬁnitions in adhesion. Polymer Science, 1(1), December 2008.

[20] Pulsar Ltd. Mpw 50 25 magnetic pulse system - research edition. User guide, 2007. | [21] Austin Weber. The cold welding process is being used for more and more high-volume applications. Assembly Magazine, August 2002.

[22] Heidi Cramer. Metallograﬁc investigation of mpw interfaces. 1st Technical Conference on Industrialized Magnetic Pulse Welding and Forming, Munich, 3 July 2008. 42 slides.

[23] M. Kashani, T. Aizawa, and K. Okagawa. Magnetic pulse welding (mpw) method for dissimilar sheet metal joints. Tokyo Metropolitan college of Technology, Department of Electronic and Information Engineering, Tokyo, Japan, 2007.

[24] Paolo Mussi. Magnetic pulse welding on receiver drier for hvac. 1st Technical Conference on Industrialized Magnetic Pulse Welding and Forming, Munich, 3 July 2008. 20 slides.

[25] Magnetic pulse welding. http://www.tribtech.com/Magnetic%20pulse%20welding. htm. webpage visited on 8/1/2009. Bibliography 149

[26] Pulsar Ltd. New applications development. Presentation for internal use, 2008.

[27] Ahmet Durgutlu, Hasan Okuyucu, and Behcet Gülenc. Investigation of effect of the stand-off distance on interface characteristics of explosively welded copper and stainless steel. Materials and Design, 29:1480–1484, 2008.

[28] Behcet G¨ulenc. Investigation of interface properties and weldability of aluminum and copper plates by explosive welding method. Materials and Design, pages 275–278, 2008.

[29] Ahmet Durgutlu, Behcet G¨ulenc,and Fehim Findik. Examination of copper/stainless steel joints formed by explosive welding. Materials and Design, pages 497–507, 2005.

[30] B.S. Zlobin. Explosion welding of steel with aluminum. Combustion, Explosion, and Shock Waves, 38(3):374–377, 2002.

[31] Oleg B. Drennov. About the state of two-metal contact boundary at a high-velocity oblique impact. International Journal of Impact Engineering, pages 205–213, 1999.

[32] P.V. Vaidyanathan and AR. Ramanathan. Design for quality explosive welding. Journal of Materials Processing Technology, 32:439–448, 1992.

[33] H.H. Yan and X.J. Li. Strain rate distribution near welding interface for diﬀerent collision angles in explosive welding. International Journal of Impact Engineering, pages 3–9, 2008.

[34] F. Grignon, D. Benson, K.S. Vecchio, and M.A. Meyers. Explosive welding of aluminum to aluminum: analysis, computations and experiments. International Journal of Impact Engineering, page 13331351, 2004.

[35] A.A. Akbari Mousavi and S.T.S. Al-Hassani. Numerical and experimental studies of the mechanism of the wavy interface formations in explosive/impact welding. Journal of the Mechanics and Physics of Solids, 53:25012528, June 2005.

[36] V. Robin, E. Feulvarch, and J.M. Bergheau. Modelling of processes involving electromagnetic phenomena. International Journal of Material Forming, 1:1375–1378, January 2008.

[37] A.G. Mamalis, D.E. Manolakos, A.G. Kladas, and A.K. Koumoutsos. Physical principles of electromagnetic forming process: a constitutive ﬁnite element model. Journal of Material Processing Technology, 161(1-2):294–299, 2005.

[38] Yu Haiping and Li Chunfeng. Eﬀects of current frequency on electromagnetic tube compression. Journal of Materials Processing Technology, 2007. Bibliography 150

[39] M.A. Bahmani, K. Niayesh, and A. Karimi. 3d simulation of magnetic ﬁeld distribution in electromagnetic forming systems with ﬁeld-shaper. Journal of Material Processing Technology, 2005.

[40] V. Psyk, C. Beerwald, A. Klaus, and M. Kleiner. Characterisation of extruded magnesium proﬁles for electromagnetic joining. Journal of Materials Processing Technology, 177:266269, 2006.

[41] A. Ben-Artzy, A. Stern, N. Frage, and V. Shribman. Interface phenomena in aluminum- magnesium magnetic pulse welding. Science and technology of welding and joining, 13(4):402–408, 2008.

[42] Filip Broekaert and Michael Deketele. The Applicability of Magnetic Pulse Forming. Master thesis, Ghent University, June 2009.

[43] Pulsar. Experimental setup guidelines. Internal report, 2008.