Download Experimental research on magnetic pulse welding of dissimilar

Experimental research on magnetic pulse welding of
dissimilar metals
Jan Broeckhove, Len Willemsens
Promotor: prof. dr. ir. Wim De Waele
Begeleiders: Koen Faes (BIL), Thomas Baaten (BIL)
Masterproef ingediend tot het behalen van de academische graad van
Master in de ingenieurswetenschappen: werktuigkunde-elektrotechniek
Vakgroep Mechanische constructie en productie
Voorzitter: prof. dr. ir. Patrick De Baets
Faculteit Ingenieurswetenschappen
Academiejaar 2009-2010
“De auteurs en de promotor geven de toelating deze masterproef voor consultatie beschikbaar te
stellen en delen van de masterproef te kopiëren voor persoonlijk gebruik. Elk ander gebruik valt
onder de beperkingen van het auteursrecht, in het bijzonder met betrekking tot de verplichting de
bron uitdrukkelijk te vermelden bij het aanhalen van resultaten uit deze masterproef.”
“The authors and the promoter give permission to make this master dissertation available for
consultation and to copy parts of this master dissertation for personal use. In the case of any other
use, the limitations of the copyright have to be respected, in particular with regard to the obligation
to state expressly the source when quoting results from this master dissertation.”
Gent, mei 2010
De promoter
De begeleider
De auteurs
Prof. dr. ir. W. De Waele
dr. ir. K. Faes
Jan Broeckhove
Len Willemsens
ii
Acknowledgments
With these words, our master thesis is almost finished. We would like to acknowledge the
help of several people during the course of this year.
First of all, we would like to express our sincere gratitude to our promotor Prof. dr. ir. Wim
De Waele, and our mentor dr. ir. Koen Faes. Despite their busy schedules, they were always
prepared to help us and push us in the right direction. We very much appreciate all the
effort they put into revising several drafts of this work.
We would like to thank the people of the Belgian Welding Institute, which helped us a lot
during this year: Michel De Waele for showing us how to embed and etch our specimens,
and the technical staff for making the workpieces and for helping us with many practical
problems.
Also, we much appreciate the help of Prof. dr. ir. Luc Dupré and dr. ir. Lode Vandenbossche,
during the development of the magnetic field measurement probe.
Finally, we would like to show our deepest gratitude to our parents, for offering us the
opportunity to complete these studies and for their support throughout the years.
Jan Broeckhove
Len Willemsens
Ghent, 27 May 2010
iii
Experimental research
on magnetic pulse welding
of dissimilar metals
by
Jan Broeckhove and Len Willemsens
Master thesis presented in fulfillment of the requirements for the degree of
Master of Science in Engineering
Academic year 2009-2010
Promotor: Prof. dr. ir. Wim De Waele
Mentor: dr. ir. Koen Faes (BIL)
Faculty of Engineering
Ghent University
Department of Mechanical Construction and Production
Chairman: Prof. dr. ir. Patrick De Baets
Summary
This thesis describes the MPW process, the main parameters and some of the occurring phenomena.
Furthermore and of more importance, a number of experiments have been conducted in order to
obtain practical knowledge concerning MPW. The major goal of this work is to develop weldability
windows which can be used as a tool when welding with magnetic pulses. For parts which are
constructed by magnetic pulse welding, no testing methods are readily available and thus some
methods have been developed and others were examined. This provided the ability to evaluate the
weldability and can also be useful for the development of testing methods for the industry.
Keywords:
Magnetic Pulse Welding, analytical model, experiments, evaluation methods
iv
Experimental research on magnetic pulse welding
of dissimilar metals
Jan Broeckhove, Len Willemsens
Supervisor(s): Wim De Waele, Koen Faes
Abstract – This paper describes magnetic pulse welding (MPW)
and the influence of different parameters on the process. A
proposed analytical model is investigated and some limitations
are described. Further, some experiments are conducted to
develop weldability windows for copper-brass and copperaluminium connections. To evaluate the quality of these welds,
different testing methods were used (both NDT and DT) and a
leak test was developed.
Keywords – Magnetic Pulse Welding, analytical model,
experiments, evaluation methods
shaper is placed inside the coil. The changing magnetic field
will induce eddy currents in the outer work piece, also named
the flyer tube. Further due to the shielding effect of an
electrical conductor the flyer tube will prevent the magnetic
field of passing through. So considering the Lorentz force, the
magnetic field outside the flyer tube will exert a force on the
flyer tube due to the eddy currents, thrusting the tube inward in
radial direction. The high velocity of the inward motion and
thus the high-energy impact between outer and inner work
piece will result in a cold weld.
I. INTRODUCTION
The contemporary construction industry is evolving with a
rapid pace and is pushing technological boundaries. Joints of
dissimilar metals are becoming more important as they offer
numerous advantages: light weight constructions, corrosion
resistant parts, etc.. Magnetic pulse welding (MPW) is a cold
welding process which is able to create bonds between
dissimilar metals. Yet only a small number of experimental
data is available nowadays. To confirm whether the process is
suited for industrial applications, it is important that a large
number of experiments is performed.
II. PRINCIPLE
III. ANALYTICAL MODELLING
Magnetic pulse welding is a cold welding process which
uses the energy of a high velocity impact to join two parts. The
process can be compared to explosion welding, but using
magnetic force to accelerate the object instead of explosives.
Unlike conventional welding processes no melting is involved
and thus no major changes in material properties take place.
The working principle is based on the theory of the Lorentz
force, dictating that an electrically charged particle, moving in
a magnetic field, undergoes a force normal to the direction of
the magnetic field and to the direction of movement:
)
= ( Figure 1: System schematic [1]
(1)
In equation (1) F is the force (in Newton), q is the electric
charge (in Coulombs), B is the magnetic field (in Tesla) and v
the speed of the particle (in m/s). The force exerted by an
electric field has been neglected since no significant electric
field will be present in this application.
The main components of the welding machine can be
schematically depicted as shown in Figure 1 .
First a bank of capacitors is charged to an energy level chosen
by the operator. Once the bank is fully charged, the high
current switch can be closed, sending a current through the
coil. This current will induce a magnetic field in the coil. To
concentrate the magnetic field in the desired region, a field
An analytical model which was proposed by the
manufacturer of the welding machine, is discussed in this
work. It appears that too many simplifications were used
throughout this model. The MPW process is a high velocity,
elastic-plastic process. The proposed model neglects the time
dependence of different parameters and it uses formulas which
describe linear elastic processes.
The development of an analytical model should start with a
proper description of the electrical circuit. A RLC-circuit is
proposed which enables the calculation of the current in the
circuit. The knowledge of the value of this current leads to the
calculation of the magnetic field. To understand the influence
of the field shaper on the magnetic field, it is proposed to
conduct finite element simulations which can lead to a
correction factor. This factor can then be implemented in the
analysis of the process.
Once the magnetic field is known, the magnetic pressure on
the flyer tube can be calculated. The magnetic pressure will
accelerate the flyer tube to the desired velocity. If it were not
for the complicated deformation behaviour of the flyer tube,
the time-functions of acceleration and velocity could be
estimated by integration of the time-dependant magnetic
pressure. Again, FE simulations can be applied to determine
the deformation behaviour of the tube. For example, the
Johnson-Cook model can be used to model the strain
v
hardening and the strain rate hardening on the flow stress
during the process. If these simulations could generate a set of
analytical equations relating the pressure to the impact
velocity and angle, the analytical model would be complete. It
should be noted that in that case, the model would not be valid
in general, but only for the specific coil/field shaper and
workpiece geometry applied in the experiments.
IV. EXPERIMENTS
In the MPW process, a large number of parameters are
important. Given the MPW machine (capacitance, coil
inductance and field shaper geometry) and the choice of tube
and rod materials (electrical conductivity, magnetic
permeability, mechanical properties), geometrical parameters
of the workpiece can be varied. The most important
geometrical parameters are the overlap length and the standoff distance. Experiments were performed on material
combinations copper-aluminium and copper-brass.
A. Weld evaluation methods
Both destructive and non-destructive testing methods were
applied to evaluate the quality of the welds.
Non-destructive tests
Because it is an important quality criterion for the welds to
be leak free, a simple leak test setup was established using
pressurised air. The leak test is quick and easy, and gives a
straight-forward measure of weld quality. Both computerized
tomography and ultrasonic inspection were performed
externally. Although the tested workpieces were not welded,
no flaws could be detected using these two inspection
techniques.
Destructive tests
Microscopic examination is used to evaluate the weld length
and wave formation. Both compressive and torsion testing
were applied to determine the shear strength of the weld zone.
The torsion test was abandoned because also flawed
magnetic pulse welds passed the test.
From the compressive test it was observed that the shear
strength increases with the voltage level. For high-quality
welds, the shear strength exceeds the buckling resistance of
the flyer tube. It was concluded that the compressive test is
more suitable to evaluate weld strength.
It is important to emphasize the need for multiple testing
methods when evaluating weld quality. For example, several
welds did not show leakage, but the tube separated from the
rod
after
cross-sectioning.
Other
welds
leaked,
notwithstanding a weld with wave pattern was observed
during microscopic examination.
B. Weldability windows
1) Copper-Aluminium
None of the copper-aluminium experiments resulted in
high-quality welds, only some were partially welded. It was
observed that stand-off distance values of 2,5 and 3 mm were
too large. The inner rods were severely deformed, which
indicates that the impact velocity was very high. Also, the
impact angle was too large. The experiments conducted with a
smaller stand-off distance (2 and 1,5 mm) showed smaller
leakage. At higher voltage levels (19 kV), several
workpieces were partially welded. that further experiments
should be performed at low stand-off distances and high
voltage levels.
A fourth series of experiments was planned but could not
be performed due to the field shaper damage. It is
recommended that these experiments are conducted in future
research.
2) Copper-Brass
The experiments in this thesis were performed with an
overlap length of 10 mm, because previous experiments
showed that this is the optimal value for the copper-brass
combination.
It was observed that a stand-off distance of 1,5 mm
produced higher quality welds than stand-off distance 2 mm.
At high voltage levels (18 kV up to 20 kV), high-quality
copper-brass welds were produced, without leakage and with
sufficient shear strength.
By breaking the welds, it was observed that the weld
length varied around the circumference. So, the value
measured by microscopic examination is not necessarily
representative for the entire weld zone.
It should be noted that many workpieces are labeled
“partially welded”. These welds showed irregularities in the
weld zone, caused by the large cracks in the field shaper. The
damaged field shaper probably influenced the reproducibility
experiments, which showed significant variation with
regards to welds performed at the same parameters.
Future experiments are necessary to confirm the
reproducibility of the MPW process with an undamaged field
shaper. Also, tests should be performed using a smaller
stand-off distance to further develop the weldability window.
V. CONCLUSION
A proposal was made for the development of a more
realistic analytical model. This method includes the use of
finite element simulations and experiments to gain a better
understanding in the process.
Experiments on copper-aluminium and copper-brass
material combinations showed that a stand-off distance of
2 mm can be considered to be too large and that the required
voltages are rather high. Welds started to form at a voltage
around 18 kV.
Inspection of the weld quality showed that no evaluation
method was able to confirm the quality with absolute
certainty. In the process of developing weldability windows,
it is recommended to combine different evaluation methods.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the support of the
technical staff of the Laboratory Soete and the Belgian
Welding Institute.
REFERENCES
1. Shribman, V. Magnetic Pulse Welding. s.l. : Pulsar ltd.
Magnetic Pulse Solutions, 2007.
vi
Experimental research on magnetic pulse
welding of dissimilar metals
- Nederlandse samenvatting
Inleiding
Dit werk kadert in het onderzoeksproject “SOUDIMMA” dat wordt uitgevoerd voor de Waalse
industrie door het Belgisch Instituut voor Lastechniek en CEWAC (Centre d’Etude Wallon de
l’Assemblage et du Contrôle des Matériaux). Het project onderzoekt voor verschillende materiaal
combinaties de lasbaarheid met behulp van het magnetisch puls lasproces.
Magnetisch puls lassen is een “koud” lasproces dat twee metalen stukken met elkaar verbindt door
middel van de energie die vrijkomt bij impact aan hoge snelheid, zonder de materialen echt te doen
smelten. Het proces kan dus vergeleken worden met explosielassen maar er wordt een magnetische
kracht gebruikt om het werkstuk te versnellen. Vermits er geen smelt optreedt, zal er ook geen
verandering van de materiaaleigenschappen optreden. Het werkingsprincipe is gebaseerd op de
Lorentz kracht die zegt dat op een geladen deeltje dat beweegt in een magnetisch veld, een kracht
wordt opgewekt die loodrecht staat op zowel de richting van het magnetisch veld als op de
bewegingsrichting.
Het proces is in staat om verschillende metalen met elkaar te verbinden. In de auto-industrie en de
koeltechniek is er steeds meer vraag naar zulke verbindingen. Een voorbeeld hiervan is een cardanas
bestaande uit een aluminium buis die aan een stalen uiteinden wordt gelast met behulp van
magnetisch pulslassen. Door de vaak grote verschillen in de smelttemperatuur van deze materiaal
combinaties is het niet mogelijk om zulke verbindingen te maken met traditionele lastechnieken.
Het principe van magnetisch puls lassen is reeds decennia gekend maar wordt toch nog niet op grote
schaal toegepast. Dit is zeker deels te wijten aan het feit dat het onderzoek naar dit proces tot op het
heden vaak van theoretische aard was en dat de praktische bruikbaarheid van het proces nog niet
uitgebreid beschreven is. Gestandaardiseerde methoden om de kwaliteit van de las te onderzoeken
zijn niet voorhanden en vaak worden geïmproviseerde testen gebruikt.
Deze thesis zal eerst een uitgebreide beschrijving van het proces geven alsook van de invloed van
verschillende parameters. Verder worden ook de resultaten besproken van proeven uitgevoerd op
zowel koper-aluminium als op koper-messing verbindingen. De doelstelling was om de industrieel
relevante lasvensters te bepalen voor deze materiaalcombinaties. Om de kwaliteit van deze proeven
te controleren werden enkele evaluatiemethoden ontwikkeld. Ook werd de toepasbaarheid van
enkele algemene beproevingsmethoden besproken.
vii
Het magnetisch puls lasproces
Het magnetisch puls lasproces gebruikt een magnetische kracht om een werkstuk te versnellen en te
doen impacteren op een ander werkstuk. De machine die hiervoor gebruikt wordt kan opgesplitst
worden in drie grote delen: de energievoorziening, een condensatorbank en een ontlaadcircuit met
spoel. Een schematische voorstelling van dergelijke machine wordt gegeven in Figuur 1. Het
magnetisch puls las proces kan opgedeeld worden in een elektrisch/magnetisch en in een
mechanisch gedeelte.
Figuur 1: Een schematische voorstelling van de belangrijkste componenten van een magnetisch puls lasmachine. [1]
Zoals reeds aangehaald, wordt de verbinding verwezenlijkt onder invloed van een impact aan hoge
snelheid. De kracht die het werkstuk versnelt, is magnetisch van aard. Via de energievoorziening van
de machine wordt een condensatorbank opgeladen. Zodra het gewenste energieniveau bereikt is,
wordt een schakelaar gesloten zodat de bank in verbinding wordt gebracht met het ontladingscircuit.
Er ontstaat aldus een wisselstroom doorheen dit circuit alsook door de spoel die zich erin bevindt. De
stroom die door de spoel vloeit zal in die spoel een wisselend magnetisch veld opwekken. Dit
magnetisch veld kan indien gewenst nog geconcentreerd worden door een zogenaamde field shaper.
Dergelijke field shaper bestaat uit een geleidend materiaal dat een specifieke vorm bezit om zo het
magnetisch veld te concentreren naar de zone van interesse.
In de spoel worden twee concentrische cilindrische werkstukken aangebracht. Het wisselend
magnetisch veld zal tijdens het proces wervelstromen opwekken in het buitenste werkstuk (dat dus
uit een geleidend materiaal moet bestaan). Door het zogenaamde skin effect dat optreedt bij een
geleider, zal dit buitenste werkstuk ook het magnetisch veld blokkeren. Zo ontstaat er een verschil
tussen het magnetisch veld binnen en buiten de cilinder. Door het verschil in magnetisch veld zal
volgens het principe van de Lorentz kracht, een kracht uitgeoefend worden op de geïnduceerde
wervelstromen, die naar binnen toe gericht is. Er wordt dus een kracht gegenereerd op het buitenste
werkstuk die dit stuk een inwaarts gerichte versnelling zal geven tot het stuk botst tegen het
binnenste werkstuk. Indien de juiste parameters worden gebruikt, zal door de hoge energie die
vrijkomt tijdens de botsing, een verbinding worden gecreëerd tussen de twee werkstukken. De
afstand tussen de atomen van de verschillende materialen wordt dan zo klein dat het delen van
elektronen mogelijk wordt.
viii
In vorige paragraaf werd aangehaald dat het buitenste werkstuk uit een geleidend materiaal moet
bestaan. Zo zijn er nog wel andere beperkingen die eigen zijn aan het magnetisch puls lassen. Deze
beperkingen zijn vooral op te leggen aan de te lassen onderdelen. Naast de nood aan voldoende
elektrische geleidbaarheid, is de geometrie van de werkstukken beperkt tot cilinders en vlakke platen
met beperkte grootte. De grootste diameter waarvoor resultaten van lasexperimenten
gerapporteerd zijn in de literatuur, bedraagt 121mm. Ook zijn sommige gedeeltes van de
werkstukken (zoals hoeken en randen) moeilijker om te lassen en moet er steeds een overlap zijn
tussen de twee stukken.
Het proces bezit natuurlijk ook enkele voordelen ten opzichte van conventionele lasmethoden. Eerst
en vooral is de warmte-inbreng tijdens het proces minimaal. Dit zorgt ervoor dat de door warmte
beïnvloede zone minimaal is en dat alle materiaaleigenschappen behouden worden. Verder moeten
de te lassen werkstukken niet voorbereid worden en is ook een nabewerking meestal niet nodig.
Samen met het feit dat het eigenlijke lasproces slechts een fractie van een seconde duurt, kan er dus
een zeer hoge productiviteit bekomen worden. Magnetisch puls lassen is dus een veelbelovende en
kostenefficiënte manier om verschillende niet lasbare metalen met elkaar te verbinden.
Golfpatroon
Bij een impactlas kan vaak een golfpatroon worden waargenomen in het lasoppervlak. Dit principe
kan vergeleken worden met het effect van wind over een wateroppervlak. Zodra er twee fluïda over
elkaar bewegen met een relatief snelheidsverschil, worden er in de buurt van een Kelvin-Helmholtz
instabiliteit golven gecreëerd, met massaoverdracht van het zwaarder naar het lichter materiaal.
Door de hoge snelheid die het werkstuk krijgt gedurende het proces, kunnen de materialen tijden de
impact als vloeistoffen beschouwd worden. De eigenlijke instabiliteiten ontstaan door de
interferentie van drukgolven doorheen het materiaal.
Aan het voortschrijdend punt van impact ontstaan drukgolven die zich zowel in het binnenste als het
buitenste stuk voortplanten. Deze golven worden gereflecteerd door het uitwendige oppervlak van
het stuk, terug richting het lasoppervlak. Daar waar er interferentie is tussen de gereflecteerde
golven en het punt van impact (dat nieuwe drukgolven voortbrengt) ontstaan Kelvin-Helmholtz
instabiliteiten en dus ook golven. Dit principe is in zeven stappen weergegeven in Figuur 2.
Analytisch model
Door de fabrikant van de puls las machine (Pulsar) werd een analytisch model voor het MPW proces
opgesteld. Dit model beschrijft wel het multidisciplinair karakter van het proces maar het is op al te
veel vereenvoudigingen gebaseerd. Het MPW proces is een hoogdynamisch, plastisch vormgevend
proces met een grote tijdsafhankelijkheid. Het vooropgestelde model houdt hier geen rekening mee:
de tijdsafhankelijkheid van verschillende parameters wordt verwaarloosd en het mechanisch
materiaalgedrag wordt lineair elastisch beschouwd, wat zeker niet het geval is.
In dit werk worden enkele bouwstenen gegeven voor een meer realistisch analytisch model.
Het elektrisch circuit wordt beschouwd als een RLC-keten waarvan de componenten kunnen worden
berekend aan de hand van experimentele metingen. Eens de waarde van de componenten gekend is,
kan de ontladingsstroom bepaald worden voor verschillende instellingen van de machine.
ix
Figuur 2: Kelvin-Helmholtz instabiliteiten liggen aan de basis van de golfvorm die het lasoppervlak verkrijgt. [20]
De kennis van de ontladingsstroom leidt tot een berekening voor het magnetisch veld in de spoel. De
field shaper zal het magnetisch veld echter nog verder concentreren tot in een kleine zone. De
invloed van deze concentratie op de grootte van het magnetisch veld is moeilijk analytisch te
bepalen. In dit werk wordt voorgesteld om met behulp van eindige elementen simulaties de invloed
van de field shaper te bepalen en hiermee een vormfactor op te stellen. Deze vormfactor kan dan in
een analytisch model het verband tussen het magnetisch veld in de field shaper en dat in de spoel
uitdrukken. Een voorbeeld van een simulatie van de invloed van de field shaper op het verloop van
het magnetisch veld wordt gegeven in Figuur 3. Merk op dat deze invloed afhankelijk is van de vorm
van de field shaper: voor elke field shaper dient een eigen vormfactor opgesteld te worden.
Figuur 3: De invloed van de field shaper op het magnetisch veld [27].
x
Eens de grootte van het magnetisch veld gekend is, kan de magnetische druk op het buitenste
werkstuk berekend worden. De vervorming van dit buitenste werkstuk gebeurt aan een zeer grote
snelheid en is volkomen plastisch. Hierdoor kunnen geen lineair elastische formules gebruikt voor de
vervorming van het werkstuk. Er wordt voorgesteld om dit vervormingsproces te beschrijven aan de
hand van het Johnson-Cook model.
De magnetische druk staat in relatie met de versnelling van het werkstuk. Hier moet echter
opgemerkt worden dat het werkstuk weerstand biedt tegen het vervormingsproces en dus ook tegen
de versnelling. Indien geen rekening wordt gehouden met de weerstand tegen vervorming, is het
snelheidsprofiel van het buitenste werkstuk zoals aangeduid op Figuur 4.
Figuur 4: Het snelheidsprofiel van het buitenste werkstuk indien geen rekening wordt gehouden met de weerstand die
het uitoefent tegen de vervorming.
Indien deze vervormingsweerstand wel in rekening wordt gebracht, zal de snelheid van het buitenste
werkstuk terug beginnen afnemen als de magnetische kracht kleiner wordt dan de kracht die nodig is
voor de vervorming van het werkstuk. Het snelheidsprofiel ziet er dan uit als getoond in Figuur 5.
Merk op dat het dus de bedoeling is dat de impact plaatsvindt rond de eerste piek van de
werkstuksnelheid.
Figuur 5: Experimenteel opgemeten snelheidsprofiel van het buitenste werkstuk waarin de weerstand tegen vervorming
inherent is vervat [23].
xi
Er wordt aangeraden om het analytisch model op te stellen in nauwe samenwerking met simulaties
en experimenten. Met de hulp van simulaties kan het tijdsafhankelijke vervormingsgedrag
bestudeerd worden. Zo kunnen correcties aangebracht worden op de theoretische tijdsfuncties die
geen rekening houden met de weerstand tegen vervorming. De resultaten van experimenten kunnen
de geldigheid van deze correcties bevestigen of kunnen gebruikt worden om modelparameters aan
te passen.
Metingen van het proces
Om een beter begrip van het systeem te verkrijgen werden enkele methoden bedacht om metingen
uit te voeren tijdens het proces. Er zijn twee moeilijkheden die zich opdringen bij eventuele online
procesmetingen. Het proces duurt slechts een honderdtal microseconden en de plaats waar het
magnetisch veld wordt opgewekt is zeer moeilijk bereikbaar.
In de literatuur werden de volgende meetmethoden gevonden:
•
•
•
•
hoge snelheidscamera
meting van de procestijd met behulp van een elektrisch circuit
meting van de vervorming van het werkstuk met behulp van lasers
meten van het magnetisch veld met behulp van optische vezels
Metingen die daadwerkelijk werden uitgevoerd gedurende dit werk zijn:
•
•
Metingen van de ontlaadstroom met behulp van een Rogowski spoel.
Metingen van het magnetisch veld met behulp van een probe die hiervoor werd
ontwikkeld.
De probe die gebruikt werd om het magnetisch veld te meten, bestaat uit een buis uit kunststof
waarop een koperen wikkeling werd gelegd. Een schematische voorstelling van deze probe in het
systeem wordt gegeven in Figuur 6.
Figuur 6: Een probe werd ontwikkeld voor het meten van het magnetisch veld in de spoel of field shaper. Deze probe
wordt geplaatst tussen de spoel en het buitenste werkstuk.
xii
Het meetprincipe van deze probe is gebaseerd op de wet van Lenz. Een wisselend magnetisch veld
dat door een geleidende wikkeling stroomt, zal in deze wikkeling een stroom opwekken die de
verandering van het magnetisch veld tegenwerkt.
Het wisselend magnetisch veld loopt axiaal door de spleet tussen de field shaper en het buitenste
werkstuk. Indien de probe in deze spleet wordt aangebracht, zal het veld door de koperen wikkeling
stromen. In een elektrische wikkeling waardoor een wisselend magnetisch veld loopt, zal een stroom
opgewekt worden. De grootte van deze stroom staat in relatie tot de grootte van het magnetisch
veld. Door de opgewekte stroom te meten, kan dus de grootte van het magnetisch veld bepaald
worden. Om de opgewekte spanning te beperken, werd de wikkeling slechts rond een kwart van de
omtrek van het buisje gelegd.
De juiste afmetingen van de meetspoel moesten bepaald worden door ijking met behulp van een
Helmholtz spoel. Door de kleine afmetingen van de wikkeling bracht dit enorme moeilijkheden met
zich mee. De veldsterkte die de Helmholtz spoel genereert, is afhankelijk van de voedingsstroom.
Echter zal de veldsterkte nooit de waarden benaderen die gehaald worden tijdens het puls lassen. De
spanning die werd opgewekt tijdens de ijking was erg beperkt en moeilijk te meten met de
voorhanden spanningsmeters. Het was na de ijking onmiddellijk duidelijk dat de verkregen
oppervlakte (36,95mm²) veel te klein was ten opzichte van de geometrisch berekende oppervlakte
(73 mm²). Voor de metingen werd uiteindelijk geopteerd om de oppervlakte te gebruiken die
geometrisch werd berekend.
Na enkele experimenten was duidelijk dat de meetspoel naar behoren functioneert. Er werden
simultaan metingen gedaan van zowel de ontlaadstroom als van de door het magnetisch veld
geïnduceerde stroom in de meetspoel. Figuur 7 geeft een voorbeeld van dergelijke metingen. De
vorm van de stroom die werd opgewekt in de meetspoel was net zoals de ontlaadstroom een
gedempte sinus met een faseverschuiving ten opzichte van die laatste.
Figuur 7: Opgemeten tijdsverloop van de ontlaadstroom (blauw) en de opgewekte stroom in de meetspoel (paars). Het is
duidelijk dat beide stromen een gedempte sinusoïdale vorm hebben en in fase verschoven zijn ten opzichte van elkaar.
xiii
De veldmetingen werden uitgevoerd bij verschillende laadspanningen van de magnetisch puls
lasmachine. Het werd snel duidelijk dat de veldsterktes die werden opgemeten niet correct waren.
Bij sommige voltages was het verschil tussen de gemeten en gesimuleerde waarden enorm. Deze
verschillen kunnen niet te wijten zijn aan een slecht gekozen meetoppervlakte want er is geen vaste
factor voor de fout. De ene keer is het verschil positief, de andere keer negatief.
Deze eigenaardigheid deed vermoeden dat er een fout was opgetreden in de lasmachine zelf. Na
inspectie bleek dat de field shaper enorm beschadigd was. Dit fenomeen zal uiteraard aanleiding
geven tot foutieve metingen, zelfs met correct geijkt materiaal.
Methoden om de laskwaliteit te evalueren
Er zijn in de literatuur geen gestandaardiseerde evaluatiemethoden terug te vinden om de kwaliteit
van magnetisch puls gelaste proefstukken te beoordelen. In dit werk werden enkele algemene
testmethoden en hun toepasbaarheid op het magpuls proces bestudeerd. Deze methoden omvatten
zowel destructieve als niet-destructieve testen. Ook werd een eenvoudige opstelling voor lektesten
ontwikkeld.
De drie niet-destructieve methoden die werden bestudeerd, zijn de lektest, ultrasoon onderzoek en
computer tomografie (de zogenaamde CT scan).
Tijdens een lektest wordt lucht met een druk van 4 bar in het werkstuk (ondergedompeld in water)
gebracht en wordt er gekeken naar eventuele lekken. Deze methode geeft een zeer goede indicatie
over de kwaliteit van de las maar blijkt ook niet feilloos te zijn. Eerst en vooral zijn de moleculen in de
lucht eerder groot en worden kleine lekken dus moeilijk zichtbaar. Een stuk dat goed scoorde op de
lektest bleek toch aanzienlijk te lekken wanneer helium werd gebruikt in plaats van lucht. Een ander
stuk was ook lekdicht, maar viel toch uit elkaar nadat het axiaal werd doorgeslepen voor
microscopisch onderzoek.
Ultrasoon onderzoek werd toegepast op enkele stukken door de firma “Brutsaert”. Ondanks het feit
dat de ultrasone methode werd uitgevoerd op stukken die behoorlijke lekken vertoonden, werden er
geen fouten gevonden. Het blijkt dat de gebruikte tasters en de manuele hantering ervan niet
geschikt zijn voor deze lassen. Er wordt voorgesteld om in een laboratoriumopstelling een
aangepaste taster te gebruiken die gefocust kan worden.
Werkstukken met behoorlijk uitgestrekte fouten werden gebruikt om computer tomografie te
evalueren. Deze testen werden uitgevoerd door CEWAC en de resultaten waren negatief. Ondanks
de aanwezigheid van lasfouten, werden er door de CT scan geen fouten ontdekt. De gebruikte
resolutie van het systeem was dus te klein. Ook de medische afdeling van de Gentse universiteit kan
CT-scans nemen van voorwerpen maar enkel indien de wanddikte van de werkstukken niet te groot
is. De werkstukken die in dit werk gebruikt werden, voldoen niet aan deze voorwaarde en konden
niet onderzocht worden.
Er werden ook enkele destructieve methoden gebruikt om de laskwaliteit te evalueren:
microscopisch onderzoek, een torsietest en een doordrukproef.
Alvorens een werkstuk microscopisch kan onderzocht worden, dient het axiaal doorgeslepen te
worden. Vaak vallen de werkstukken spontaan uit elkaar tijdens dit proces wat wijst op een slechte
xiv
las. Als de stukken toch aan elkaar blijven hangen, kan het werkstuk in kunststof worden ingebed. Na
zorgvuldig polijsten, kan het stuk dan worden bekeken onder een microscoop. Het wordt dan snel
duidelijk of de twee materialen al dan niet aan elkaar hangen. Ook deze methode bleek niet feilloos
te zijn. Een stuk dat lekken vertoonde, bleek onder de microscoop toch aan elkaar gelast te zijn. Deze
beperking is te wijten aan het feit dat je het lasoppervlak slechts zeer lokaal kan bekijken op twee
posities van de volledige omtrek. Afhankelijk van de plaats waar de axiale snede wordt gemaakt,
wordt een ander beeld verkregen. Het is dus mogelijk dat toevallig een goed deelt van een slechts
gedeeltelijke las wordt onderzocht. Het microscopisch onderzoek volstaat dus niet om uitsluitsel te
brengen omtrent de kwaliteit van de las.
De torsietest en de doordrukproef werden met succes uitgevoerd op verscheidene stukken. Deze
methoden bepalen de kracht die nodig is om een afschuiving te verkrijgen in het lasoppervlak. Indien
een goede las werd gemaakt zal de laszone sterker zijn dan het zwakste basismateriaal en zal het
werkstuk tijdens deze testen falen op een andere plaats dan het lasoppervlak. Deze
evaluatiemethoden zijn uiteraard ook niet feilloos. Ze zijn een goede indicatie voor de mechanische
sterkte van de las maar vertellen niets over eventuele lekken.
De conclusie van deze bevindingen is dat geen van de genoemde evaluatiemethoden volstaat om
éénduidig de kwaliteit van de las te beoordelen. Bij twijfel omtrent de kwaliteit van een las, zullen
steeds meerdere soorten testen moeten uitgevoerd worden.
xv
Experimenten
In dit werk werden experimenten uitgevoerd op twee materiaalcombinaties: koper-aluminium en
koper-messing.
Het doel van deze proeven was om de invloed van verschillende parameters te onderzoeken. Er zijn
veel verschillende procesparameters die kunnen gevarieerd worden tijdens het magnetisch puls
lasproces. De geometrische parameters worden weergegeven in Figuur 8. Verder zijn er ook nog de
materiaal-afhankelijke parameters en het spanningsniveau waarmee de machine wordt opgeladen.
Figuur 8: Het magnetisch puls lasproces bevat veel parameters die het proces kunnen beïnvloeden. Op deze figuur
worden volgende geometrische parameters weergegeven: de dikte van het buitenste werkstuk (t), de grootte van de
luchtspleet (s), de overlaplengte tussen field shaper en buitenste werkstuk en de overlaplengte tussen de twee
werkstukken [17].
Al deze parameters beïnvloeden de waarde van twee zeer belangrijke grootheden: de
impactsnelheid en de impacthoek. Enkel indien deze grootheden binnen een nauw procesvenster
gelegen zijn, zal een lasverbinding verwezenlijkt worden. Het belang van deze parameters werd
reeds duidelijk bij onderzoek omtrent explosielassen. Zoals reeds gezegd, zijn er grote
overeenkomsten tussen beide processen en kan het belang van de impacthoek en –snelheid dus
veralgemeend worden naar het magnetisch puls lasproces. Vooropgesteld wordt dat de waarde van
de impacthoek zich tussen 6 en 14° moet bevinden. De waarde van de impactsnelheid is
vermoedelijk materiaal gebonden.
Gedurende de koper-aluminium experimenten werden twee verschillende groottes van luchtspleet
gebruikt, 2,5 en 3 mm. De variabele parameters tijdens deze experimenten waren de lengte van het
buitenste werkstuk (en dus de overlaplengte met de field shaper) en het spanningsniveau.
De resultaten van deze testreeksen waren negatief. Er werden geen lassen met hoge kwaliteit
geproduceerd. Hieruit kan geconcludeerd worden dat de gebruikte luchtspleet te groot was. Deze
conclusie werd gecontroleerd door schatting te maken van de impacthoek bij eerste impact. Het
bleek dat de impacthoek inderdaad te groot was om een goede las te verkrijgen. Ook wordt er
xvi
verondersteld dat de grote luchtspleet zorgt voor meer tijd waarin het werkstuk kan versneld
worden en zal de impactsnelheid dus te hoog liggen. Bij deze hoge snelheden wordt een deel van de
impactenergie verbruikt door vervorming van het binnenste werkstuk en wordt slechts een klein deel
van de vrijkomende energie gebruikt om een verbinding te realiseren.
Tijdens de koper-messing experimenten werd de lengte van het buitenste werkstuk constant
gehouden (48 mm). Uit voorafgaand onderzoek bleek dit een optimale waarde te zijn. De parameters
die tijdens de testreeksen werden aangepast, zijn het spanningsniveau en de grootte van de
luchtspleet. Het spanningsniveau werd gevarieerd van 14 tot 20 kV, in stappen van 0,5 kV en dit voor
luchtspleten van 1,5 en 2 mm. Ook hier bleek een grote luchtspleet nadelig: de experimenten met
een luchtspleet van 2 mm produceerden geen goede lassen.
De experimenten met een luchtspleet van 1,5 mm leverden wel enkele goede resultaten. Alle
proeflassen die met een spanning hoger of gelijk aan 18 kV werden gemaakt, bleken na
microscopisch onderzoek van hoge kwaliteit te zijn. De las die werd gemaakt met een laadspanning
gelijk aan 18 kV was echter niet volledig lekvrij. Nader onderzoek van dit werkstuk toonde dat er een
zone aanwezig was waar de las van mindere kwaliteit was. Deze zone was op de omtrek
gepositioneerd op precies 180° van de gleuf in de field shaper. Deze slecht gelaste zone was ook
zichtbaar bij de lassen met een lagere kwaliteit. Figuur 9 toont deze zone: een deel van de omtrek
bevat geen golfpatroon wat wijst op het feit dat op die positie geen las gevormd werd.
Dit verschijnsel was enkel te verklaren door een onregelmatigheid tijdens de ontwikkeling van het
magnetisch veld. Samen met de onregelmatige veldmetingen leidde de vaststelling van dit
terugkerend fenomeen tot de beslissing om de magnetisch puls lasmachine te inspecteren. Na deze
inspectie bleek inderdaad dat het koper van de field shaper enorm beschadigd was (Figuur 10).
Figuur 9: Op 180° van de gleuf in de field shaper bevindt zich geen golfpatroon in het oppervlak van één van de
werkstukken. Dit wijst erop dat deze zone niet naar behoren werd gelast.
xvii
Figuur 10: Een afbeelding van de schade aan de field shaper die werd ontdekt tijdens een inspectie van de machine.
Installatie van een nieuwe spoel
Vermits de field shaper ernstig beschadigd was, moest een nieuwe field shaper of een andere spoel
geïnstalleerd worden. Er werd geopteerd voor een nieuwe spoel met 1 winding zonder field shaper.
Zoals zal blijken uit het volgende was de installatie van deze spoel niet zonder problemen.
De installatie van een nieuwe spoel op de transformatorbank moet zeer nauwkeurig uitgevoerd
worden. De geleidende oppervlakken van de transformatorbank en van de voet van de spoel moeten
spelingsvrij met elkaar in contact gebracht worden om doorslag en vonken te vermijden. Vonken die
tijdens het proces optreden, kunnen de geleidende oppervlakken ernstig beschadigen. Door de grote
stroomintensiteit die optreedt gedurende het proces, zullen kleine beschadigingen in het materiaal
snel groter worden. In de installatiehandleiding van de spoel wordt beschreven dat tijdens de
inloopfase wel enige vonken kunnen optreden. Door de bevestigingsbouten verder aan te spannen
zouden deze vonken echter moeten verdwijnen.
Tijdens de installatie van deze spoel bleven er echter vonken optreden. Het was niet mogelijk om de
bouten nog verder aan te spannen want deze werden al tot aan grens van hun sterkte aangedraaid.
Er moest dus een andere oplossing gevonden worden om de vonken te doen verdwijnen.
Er werd geopteerd om extra isolatie aan te brengen tussen de delen die niet moeten instaan voor het
doorgeven van de stroom. De stroomoverdracht gebeurt door koperen staven die ingelegd zijn in de
voet van de stalen spoel (Figuur 11). Figuur 11 toont ook dat er grote holtes aanwezig zijn tussen de
stalen voet en het koperen oppervlak van de transformator. Het is net in deze holtes dat vonken
werden vastgesteld.
Om de ontwikkeling van de vonken tegen te gaan werden de volgende stappen ondernomen:
•
•
de koperen staven werden verder uit elkaar geplaatst
de holtes werden geïsoleerd met behulp van kunststof en silicone (Figuur 12).
xviii
Figuur 11: De stroom vloeit van de transformatorbank naar de spoel via koperen staven die zijn ingelegd in het staal.
Figuur 12: Extra isolatie materiaal werd aangebracht om de ontwikkeling van vonken te voorkomen.
Zelfs na deze ingrepen werden er nog vonken ontwikkeld aan de voet van de spoel. Inspectie van de
materialen toont aan dat er wel degelijk beschadigingen zijn opgetreden. Het is dus zeer belangrijk
om verdere stappen te ondernemen om vonken te voorkomen.
De volgende werkwijze wordt voorgesteld:
•
•
•
de oppervlakken die met elkaar in contact staan, moeten bijgeschaafd worden om zoveel
mogelijk oneffenheden en fouten te verwijderen. Dit zorgt ervoor dat het contact tussen de
materialen verbeterd wordt.
er dient materiaal verwijderd te worden van de onderkant van de spoel. Door een
hoeveelheid staal te verwijderen zullen de koperen staven verder uit de voet uitsteken en
wordt het dus moeilijker om een boog te trekken tussen het staal van de spoel en het koper
van de transformator.
tenslotte moet er voldoende isolatie worden aangebracht tussen het staal van de spoel en
het koper van de transformator. Als isolatiemateriaal wordt een flexibele kunststof plaat
voorgesteld die in de opening geperst kan worden. Door de flexibiliteit van het materiaal
zullen spleten opgevuld worden.
xix
Aanbevelingen voor verder onderzoek
Hieronder volgen enkele aanbevelingen voor onderzoek in de toekomst:
•
De magnetische veldmetingen zouden moeten herhaald worden met een intacte field
shaper. Deze metingen kunnen de correcte werking van de probe bevestigen en ze kunnen
de B-I relatie bepalen. Deze relatie is belangrijk voor de verdere ontwikkeling van een
analytisch model.
•
Verdere koper-aluminium lasexperimenten zijn nodig voor de ontwikkeling van de
lasvensters. De vierde reeks koper-aluminium experimenten (die werd voorbereid tijdens dit
werk) dient ook uitgevoerd te worden.
•
Om de kwaliteit van de partieel gelaste koper-messing verbindingen te bevestigen, zijn
verdere experimenten noodzakelijk. Ook zijn extra experimenten nodig om het lasvenster bij
een buislengte van 48 mm af te werken.
•
Het bleek dat de gebruikte evaluatiemethoden niet in staat waren om de laskwaliteit van een
proefstuk met zekerheid te bevestigen. Het is dus aangewezen om verder onderzoek te
verrichten naar gepaste evaluatiemethoden voor de laskwaliteit. Dit kan leiden tot nieuwe
methoden of tot aanpassingen van de methoden die werden gebruikt in dit werk.
•
Het binnenoppervlak van de field shaper of van de spoel (met één winding) moet regelmatig
geïnspecteerd worden om de initiatie van scheuren op te sporen.
xx
Contents
Chapter 1
Introduction............................................................................... 1
1.1
Magnetic Pulse Welding technology .......................................................................................1
1.2
Applications and their requirements .......................................................................................1
1.3
Problem statement ..................................................................................................................3
1.4
Overview ..................................................................................................................................3
1.4.1
Chapter 2: Magnetic Pulse Welding .................................................................................3
1.4.2
Chapter 3: Analytical Model .............................................................................................3
1.4.3
Chapter 4: Process Measurements ..................................................................................3
1.4.4
Chapter 5: Weld Quality Evaluation .................................................................................3
1.4.5
Chapter 6: Experiments....................................................................................................4
1.4.6
Chapter 7: Single turn coil installation .............................................................................4
1.4.7
Chapter 8: Conclusions and recommendations for further research ..............................4
Chapter 2
Magnetic Pulse Welding ............................................................ 5
2.1
The MPW machine ...................................................................................................................5
2.2
The electrical process ...............................................................................................................5
2.3
Shielding ...................................................................................................................................7
2.4
The impact..............................................................................................................................10
2.5
Comparison with explosion welding ......................................................................................12
2.6
Material properties ................................................................................................................13
2.6.1
Electrical properties of the material ..............................................................................13
2.6.2
Mechanical properties of the material ..........................................................................14
2.7
Main geometrical parameters ...............................................................................................15
2.7.1
Stand-off distance, s .......................................................................................................15
2.7.2
Thickness of the flyer tube, t..........................................................................................15
2.7.3
Overlap length of the field shaper with the workpiece, LF.S. ..........................................16
2.7.4
Overlap length of the workpieces, Lwp ...........................................................................16
2.7.5
The relative position of the field shaper ........................................................................16
2.8
Weld interface........................................................................................................................17
xxi
2.8.1
The mechanism of wave creation ..................................................................................17
2.8.2
Influence of parameters on the wave pattern ...............................................................19
2.9
Advantages of the process .....................................................................................................22
2.10
Limitations of the process ......................................................................................................23
Chapter 3
Analytical model ...................................................................... 24
3.1
Introduction ...........................................................................................................................24
3.2
Model by Pulsar .....................................................................................................................25
3.2.1
Introduction ...................................................................................................................25
3.2.2
Structure of the model ...................................................................................................25
3.2.3
Analysis .........................................................................................................................26
3.2.4
Discussion .......................................................................................................................33
3.3
Electrical circuit ......................................................................................................................34
3.3.1
Introduction ...................................................................................................................34
3.3.2
RLC circuit .......................................................................................................................34
3.3.3
Other electrical circuit models .......................................................................................40
3.4
Electromagnetic energy transfer ...........................................................................................41
3.4.1
Coil..................................................................................................................................41
3.4.2
Magnetic pressure .........................................................................................................43
3.4.3
Field shaper ....................................................................................................................46
3.4.4
Magnetic field measurement .........................................................................................51
3.5
Deformation, acceleration and impact velocity.....................................................................52
3.5.1
Deformation pressure ....................................................................................................52
3.5.2
Time-dependant acceleration and impact velocity .......................................................54
3.5.3
Conclusion ......................................................................................................................59
3.6
Conclusion ..............................................................................................................................60
Chapter 4
Process Measurements ............................................................ 61
4.1
Introduction ...........................................................................................................................61
4.2
Literature Survey ....................................................................................................................62
4.2.1
High speed camera .........................................................................................................62
4.2.2
Process duration measurement .....................................................................................62
4.2.3
Tube deformation measurement ...................................................................................63
4.2.4
Photon doppler velocimetry ..........................................................................................63
xxii
4.2.5
4.3
Magnetic field measurement .........................................................................................66
On-line Measurements ..........................................................................................................69
4.3.1
Discharge current measurement ...................................................................................69
4.3.2
Magnetic Field Measurement ........................................................................................71
Chapter 5
Weld Quality Evaluation .......................................................... 80
5.1
Introduction ...........................................................................................................................80
5.2
Non-Destructive Testing Methods .........................................................................................81
5.2.1
Leak test .........................................................................................................................81
5.2.2
Ultrasonic Testing...........................................................................................................82
5.2.3
Computerised Tomography ...........................................................................................86
5.3
Destructive Testing Methods .................................................................................................88
5.3.1
Microscopic investigation ..............................................................................................88
5.3.2
Torsion test ....................................................................................................................89
5.3.3
Peel test..........................................................................................................................95
5.3.4
Compression test ...........................................................................................................97
Chapter 6
Experiments........................................................................... 101
6.1
Introduction .........................................................................................................................101
6.2
Overview ..............................................................................................................................102
6.2.1
Test configuration ........................................................................................................102
6.2.2
Weld quality evaluation ...............................................................................................102
6.3
Material Characteristics .......................................................................................................104
6.4
Welding parameters.............................................................................................................106
6.4.1
Stand-off distance ........................................................................................................107
6.4.2
Charging voltage...........................................................................................................107
6.4.3
Overlap length ..............................................................................................................108
6.5
Field shaper damage ............................................................................................................109
6.5.1
Occurrence of consistent but unexpected weld defects .............................................109
6.5.2
Nature and cause of field shaper damage ...................................................................110
6.6
Copper-Aluminium experiments ..........................................................................................113
6.6.1
Introductory comments related to field shaper damage.............................................113
6.6.2
Series 1 (SD-CA-1) ........................................................................................................113
6.6.3
Series 2 (SD-CA-2).........................................................................................................116
xxiii
6.6.4
Series 3 (SD-CA-3).........................................................................................................118
6.6.5
Series 4 .........................................................................................................................119
6.7
Discussion on the Copper-Aluminium experiments ............................................................120
6.7.1
Non Destructive evaluation .........................................................................................120
6.7.2
Destructive evaluation .................................................................................................123
6.7.3
Welding Parameters.....................................................................................................132
6.8
Copper-Brass experiments: Test series 1 .............................................................................137
6.8.1
Test series 1.1: Stand-off distance = 0,75 mm .............................................................137
6.8.2
Test series 1.2: Stand-off distance = 1,0 mm ...............................................................137
6.8.3
Test series 1.3: Stand-off distance = 1,5 mm ...............................................................138
6.8.4
Test series 1.4: Stand-off distance = 2,0 mm ...............................................................139
6.8.5
Test series 1.5: Stand-off distance = 2,5 mm ...............................................................139
6.9
Copper-Brass experiments: Test series 2 .............................................................................140
6.9.1
Test series 2.1...............................................................................................................140
6.9.2
Series 2.2: Subset of series 2.1 .....................................................................................143
6.9.3
Series 2.3: Grooved collar ............................................................................................150
6.9.4
Series 2.4 ......................................................................................................................152
6.10
Discussion on the copper-brass experiments ......................................................................156
6.10.1
Destructive evaluation .................................................................................................156
6.10.2
Investigation of the wave interface .............................................................................169
6.10.3
Weldability window .....................................................................................................175
6.11
Conclusions of the experimental research...........................................................................176
6.11.1
Copper-Aluminium .......................................................................................................176
6.11.2
Copper-Brass ................................................................................................................176
6.11.3
Effect of the field shaper slit ........................................................................................177
6.11.4
Wave formation ...........................................................................................................177
6.11.5
Deformation of the inner workpiece ...........................................................................177
6.11.6
Weld strength...............................................................................................................177
Chapter 7
Single turn coil installation..................................................... 178
7.1
Introduction .........................................................................................................................178
7.2
Installation of the coil ..........................................................................................................179
7.2.1
Extra insulation - step 1................................................................................................179
7.2.2
Extra insulation- step 2 ................................................................................................181
xxiv
7.2.3
Extra insulation - step 3................................................................................................182
7.2.4
Conclusion ....................................................................................................................183
Chapter 8
8.1
Conclusions and recommendations for further research ........ 184
Summary and conclusions of this work ...............................................................................184
Process analysis ............................................................................................................................184
On-line process measurements ...................................................................................................185
Experimental research .................................................................................................................185
Weld evaluation methods ............................................................................................................186
Weldability windows ....................................................................................................................187
Single turn coil..............................................................................................................................188
8.2
Future research ....................................................................................................................189
xxv
Abbreviations
BWI
Belgian Welding Institute
CEWAC
Centre d’Etude wallon de l’Assemblage et du Contrôle des Matériaux
CT
Computerised Tomography
DT
Destructive Testing
EXW
Explosion Welding
HVAC
Heating Ventilation Air Conditioning
MAG
Metal Active Gas (welding)
MIG
Metal Inert Gas (welding)
MPW
Magnetic Pulse Welding
NDT
Non-Destructive Testing
PSD
Position Sensitive Controller
PVD
Photon Doppler Velocimetry
SEM
Scanning Electron Microscope
UGCT
Ghent University Centre for X-ray Tomography
UT
Ultrasonic Testing
xxvi
Chapter 1
Introduction
1.1 Magnetic Pulse Welding technology
Magnetic Pulse Welding (MPW) is a “cold” welding process which uses the energy of a high velocity
impact to join two parts. The process can be compared to explosion welding, but using magnetic
force instead of explosives to accelerate the object. Unlike conventional welding processes no
melting is involved and thus no major changes in material properties take place. The working
principle is based on the theory of the Lorentz force, dictating that an electrically charged particle,
moving in a magnetic field, undergoes a force normal to the direction of the magnetic field and to
the direction of movement:
)
= ( (1.1)
F is the force (in Newton), q is the electric charge (in Coulombs), B is the magnetic field (in Tesla) and
v the speed of the particle (in m/s). The force exerted by an electric field has been neglected since no
significant electric field is present in this application.
The main components of the welding machine can be schematically depicted as shown in Figure 1.1.
A bank of capacitors is charged to an energy level chosen by the operator. Once the bank is fully
charged, the high current switch is closed, sending a current through the coil. This current will induce
a magnetic field in the coil. If it is necessary to concentrate the magnetic field in the desired region, a
field shaper is placed inside the coil (not shown on the figure). The changing magnetic field will
induce eddy currents in the conductive outer workpiece, also named the flyer tube. Further, due to
the shielding effect of an electrical conductor the flyer tube will prevent the magnetic field of passing
through, creating a difference in magnetic field between the inside and the outside of the flyer tube.
So considering the Lorentz force, the magnetic field outside the flyer tube will exert a force on the
flyer tube due to the eddy currents, thrusting the tube inward in radial direction. If correctly
executed the high velocity of the inward motion and thus the high-energy impact between outer and
inner workpiece will result in bonding. During the collision, the atoms of the adjacent surfaces are
brought together, overcoming the repulsion force which drives them apart. The distance between
the atoms is now small enough to enable sharing of electrons and the creation of an intermetallic
phase, creating a bond.
1.2 Applications and their requirements
Magnetic pulse welding is a cold welding technique which is able to join dissimilar materials.
Welding of dissimilar materials is becoming more desired in the automotive industry and for HVAC
installations. In the automotive industry for example, aluminium-steel drive shafts can be welded by
magnetic pulse welding (Figure 1.2). These welded connections are not always easily to obtain by
1
traditional welding techniques due to the sometimes large difference in the material properties of
the materials to be joined.
Figure 1.1: System schematic [1]
The main demands with respect to these connections are that they are structurally strong and leakfree. To ensure a strong and safe construction, necessary in the automotive industry, the weld should
be sufficiently strong. Regarding the substances which are used in the HVAC industry (sometimes
containing CFC’s) the joints obviously need to be leak-free. MPW could provide an alternative for
conventional welding techniques, such as MIG/MAG-welding, for the production of these welded
joints.
Although the process has been known for several decades, it is not yet commonly used in industry
and certainly not when compared to conventional welding techniques. The largest part of the
research that has been conducted regarding this technique is of a theoretical nature. An
experimental approach could be interesting to provide more clarity concerning the practical use of
the process.
Figure 1.2: A magnetic pulse welded drive shaft. (aluminium-steel connection) [1]
2
1.3 Problem statement
This work is executed in the framework of the research project “SOUDIMMA”, which is performed for
the Walloon industry by the Belgian Welding Institute (BWI) and CEWAC (Centre d’Etude wallon de
l’Assemblage et du Contrôle des Matériaux). The project which is funded by the Walloon
government, the weldability of several material combinations is investigated. The results of this study
will be used for welding prototypes for the industry.
This thesis describes the MPW process, the main parameters and some of the occurring phenomena.
Furthermore and of more importance, a number of experiments have been conducted in order to
obtain practical knowledge concerning MPW. The major goal of this work is to develop weldability
windows which can be used as a tool when welding with magnetic pulses. For parts which are
constructed by magnetic pulse welding, no testing methods are readily available and thus some
methods have been developed and others were examined. This provided the ability to evaluate the
weldability and can also be useful for the development of testing methods for the industry.
1.4 Overview
1.4.1 Chapter 2: Magnetic Pulse Welding
This chapter provides a more detailed description of the MPW process and the parameters which
play an important role. A phenomenon which often occurs when welding pieces with the MPW
process is a wavy weld interface. The mechanism of this interface formation will also be explained in
this part. Furthermore, an analogy with another impact welding process, explosion welding, will be
discussed.
1.4.2 Chapter 3: Analytical Model
This chapter is devoted to the analytical description of the process. Since the process is multidisciplinary, this model exists out of both a mechanical part and an electromagnetic part. A model
which was found in literature is discussed and some recommendations are given to increase its
accuracy.
1.4.3 Chapter 4: Process Measurements
In order to obtain a better understanding of the process, some measurements can be conducted
during the experiments. Measurements of the magnetic field and the current were conducted during
the experiments. Also some ideas are presented for possible future measurements, more specific the
flyer tube velocity and the process time.
1.4.4 Chapter 5: Weld Quality Evaluation
Chapter 5 describes the methods which were used to test the welded workpieces. Distinction is
made between Destructive Testing (DT) and Non-Destructive Testing (NDT). The methods that were
used are:
•
•
DT: microscopic investigation, torsion test, compressive (“push through”) test
NDT: leak test, ultrasonic investigation, computerized tomography
Also a peel test is described. However, this peel test was not performed in this work.
3
1.4.5 Chapter 6: Experiments
The actual experiments which were conducted in the framework of this thesis are described in
chapter 6. The experiments can be divided in two major groups: copper-aluminium and copper-brass.
An overview is given of the different parameters which were used as well as the results of the
experiments. These results are derived by using the testing methods which were described in
chapter 5.
Three series of copper-aluminium welding experiments were performed using different parameters
and geometries. As an example, the third series of copper-aluminium experiments were conducted
with custom collars on the internal workpieces. Note that no qualitative welds were constructed
during all these experiments. Although a fourth series was planned, further experiments were not
possible due failure of the field shaper.
The copper-brass experiments were more extensive than the copper-aluminium experiments. These
were a continuation of tests which were conducted by dr. ir. Koen Faes of the Belgian Welding
Institute. This work investigates these welds and additional welds were performed to define the
“window” of suitable welding parameters in order to obtain a high-quality weld. Also a
reproducibility test was performed in this series. Ten magnetic pulse welds were performed with the
same parameters to gain some insight in the repeatability of the process.
1.4.6 Chapter 7: Single turn coil installation
Since the field shaper of the multi-turn coil was damaged during the experiments, the coil needed
replacement. It was decided to install a single-turn coil for welding parts with an outer diameter of
60 mm. During the installation some problems occurred. Sparks appeared between the transformer
and the coil, damaging the parts. Due to the high current nature of the process, it is important to
prevent damage to the parts as they will then deteriorate quickly by spark erosion.
This chapter describes the problems which occurred and formulates some guidelines to follow during
the installation of a new coil.
1.4.7 Chapter 8: Conclusions and recommendations for further research
This chapter gives an overview of the most important results of this work. Also some
recommendations are made on which experiments should be performed in the future. These can
include series of experiments which were already planned but could not have been conducted due to
the field shaper damage.
4
Chapter 2
Magnetic Pulse Welding
2.1 The MPW machine
The machine used during the experiments was constructed by Pulsar (Figure 2.1). This machine can
be divided into three major parts. The first part of the machine is a high-voltage power supply which
is used to charge the capacitor bank. This capacitor bank is the second major part of the machine and
it stores the energy which will be used to accelerate the flyer tube. The last part of the machine
consists of the discharge circuit. This circuit mainly consists out of a coil which generates the
magnetic field necessary for accelerating the flyer tube and producing the weld. The maximum
voltage to which the machine can be charged is 25 kV (equals an energy of 50 kJ) and the current can
be up to 500 kA. The discharge will occur with a frequency 14 kHz. This frequency is specific to the
machine and cannot be altered. The only possibility to change the frequency is by disconnecting
some capacitors. This however will also decrease the energy level at which the machine can be
charged.
Figure 2.1: Photograph of the magnetic pulse machine, type MPW 50/25,
with power supply (1), capacitor bank (2), coil (3)
2.2 The electrical process
As a start, a required energy level has to be selected. This level can be chosen by charging the
capacitor bank to a certain voltage which can be selected on the control panel of the welding
machine. An AC current is rectified and charges the capacitors to the selected voltage. Once the bank
has been fully charged, a switch is closed. Often one or more vacuum switches are used to
simultaneously release the stored energy towards the coil [2].
5
The current which then flows through the coil, will produce a magnetic field as shown in Figure 2.2.
The discharge of a capacitor over a solenoid will create a damped alternating current through the
latter one. The dampening effect is due to the internal resistance of the system (Figure 2.3). Because
of this effect, also the magnetic field will be damped and alternating.
Figure 2.2: Magnetic field induced by a current through a coil
Figure 2.3: Current through the coil during the MPW process[3]
To concentrate the magnetic field close to the flyer tube, the geometry of the coil needs to be
chosen carefully or optionally a field shaper has to be used. A field shaper is a core made out of a
strong material with preferably an excellent electrical conductivity, which concentrates the magnetic
field of the coil to the region of importance. The material of the field shaper which was used during
the experiments is copper beryllium (CuBe2). The emphasis was not only placed on the electrical
conductivity of the material (for a good efficiency) but also on its strength. This strength is necessary
to obtain an acceptable lifespan of the field shaper which undergoes heavy loads during the process.
Indeed, the forces which are generated on the workpiece will also be exerted on the field shaper.
Furthermore, the currents that flow through the field shaper are several hundreds of kilo ampères.
At these high currents, any existing flaw will quickly deteriorate through spark erosion. The same
effects take place on the coil and thus the same material requirements are placed upon the coil if no
6
field shaper is used. The main advantage of using a field shaper is an economical one: the same coil
can be used for different geometries by simply placing a proper field shaper into the coil [4].
The workpieces, which are made of a conductive material, are positioned into the field shaper or coil.
When a magnetic field is created, it will induce eddy currents in the flyer tube material. Figure 2.4
shows the flow of the currents through the system where i1 is the current in the coil, i2 in the field
shaper and i3 in the flyer tube. The eddy currents will prevent the magnetic field lines to go through
the material of the flyer tube. This phenomenon is called “shielding”. A difference in magnetic field is
now created between the space inside and outside the flyer tube. The magnetic field outside the
field shaper will be larger and thus the resulting Lorenz force will is oriented inwards.
Figure 2.4: The flow of currents through the system[4]
2.3 Shielding
To enable shielding, a certain thickness of the material is required because the efficiency of the
shielding phenomenon is linked to the skin depth of the material. The skin depth of a material is
defined as the depth below the surface of the material at which the current density decays to 1/e of
the current density at the surface. Figure 2.5 shows the distribution of the current throughout a
section of the flyer material. Jo and Jt respectively stand for the current density at the outer and inner
surface. Ei and Et stand for the electric field which is present on both sides of the material. When the
thickness of the material equals the skin depth, 86% of the magnetic field is shielded. If the thickness
of the material equals two times the skin depth, already 98% of the magnetic field is shielded
[5][6][7].
The skin depth δ of a material is given by:
=
With :
1
. . . (2.1)
σ = the electrical conductivity of the material [m/Ω]
µ = magnetic permeability of the workpiece [H/m]
f = the frequency of the current [Hz]
7
Using formula (2.1), the skin depth at 14 kHz was calculated for several relevant materials. The
results can be seen in Table 2.A. Due to the high magnetic permeability of steel, the skin depth of
steel is much smaller than the skin depth of the other materials. A steel flyer tube can thus be
thinner than a flyer tube of for example brass.
Copper
Aluminium
Brass
Steel
electrical
conductivity [m/Ω]
magnetic
permeability [H/m]
skin depth [μm]
59,6 . 106
1,26 . 10-6
550
37,8 . 10
6
1,26 . 10
-6
691
15,6 . 10
6
1,26 . 10
-6
1076
5,56 . 10
6
8,75 . 10
-4
68
Table 2.A: The skin depth for several relevant materials is given in this table. The frequency which was used to calculate
these skin depths is 14 kHz.
Figure 2.5: Distribution of the current throughout the material [6]
Note that the shielding percentages above are a rule of thumb which provides a quick estimation of
the value. If however the shielding effectiveness has to be calculated more precisely the following
method can be used.
The phenomenon of shielding consists out of three mechanisms, illustrated on Figure 2.6:
1. Incident electromagnetic waves are reflected at the surface.
2. Electromagnetic waves are absorbed in the material.
3. The waves reflect on the back surface of the material.
Also waves are transmitted and thus travel through the entire workpiece. Since this phenomenon
does not account for an amount of shielding, it will not be used in the further calculation of the
shielding effectiveness.
The total effectiveness of the shielding operation can be calculated by the sum of the shielding by all
three mechanisms. When the shielding factors of mechanism 1,2 and 3 are expressed in dB
respectively as R, A and B, the shielding effectiveness, S.E., can be calculated using the following
formulas [6]:
8
7. 8. = + 9 + (2.2)
= + % + )
9 = 3,338 . 10
= 20 |1 − >
:"
(2.3)
. ; . !
A
(, − 1)$
? . @10: B . (C :D.$$E.A )|
$
(, + 1)
(2.4)
(2.5)
With RE, RH, and RP are the reflection factors for the electric, magnetic, and plane wave fields
expressed in dB:
= 353,6 + 10 (
% = 20 (
!
"
# $
)
0,462
!
+ 0,136 # (
+ 0,354 )
(
#
!
) = 108,2 + (
! . 10+
)
(2.6)
(2.7)
(2.8)
In these equations the following symbols were used:
•
•
•
•
•
•
•
•
G = the relative conductivity of the material referred to copper [-]
f = the frequency [Hz]
μ = is the relative permeability referred to free space [-]
r1 = is the distance between the source and the shield [mm]
t = the thickness of the shielding material [mm]
Zs = the impedance of the workpiece material [Ω]
ZH = the impedance of the magnetic field [Ω]
.
, = -. / - = 33. ( 2
0
5
1
6
)
3 45 6
Figure 2.6: Three shielding mechanisms[8]
9
2.4 The impact
As stated above, the difference in the magnetic field around the flyer tube will result in a force with
as inward direction. This force will accelerate the flyer tube towards the inner workpiece. During that
phase, the process is thus comparable to the explosion welding process (EXW). In both processes a
bond is created by high velocity impact of two surfaces. This high energy impact is able to overcome
the repulsive forces between the atoms of both materials and to decrease the distance between the
surfaces to a value small enough so that electrons can be shared between the two materials. This
process creates the bond.
Greases, oxide films and other surface contaminants could provide a protective film on the surfaces
which counters the attempts to bring the surfaces close together. It is assumed that no cleaning of
the workpieces is needed prior to welding. This can be contributed to the fact that a jet is formed
during the impact of the materials. Although the process appears to be instantaneous, it is in fact a
very fast progressive action. The surfaces are collapsed against each other with a high relative
velocity. This will participate in surface jetting if the collision angle and the collision velocity are in
the range of the weldability window. Often it is prescribed that the collision angle should be between
6-14°. The collision velocity depends more on the materials which are to be welded. This jet will now
remove the contaminants of the surface. Figure 2.7 shows the jet, the impact angle (α) and the
collision velocity (vc) [2][9].
Figure 2.7: Jet formation during high speed impact [10]
In the interface between two bonded materials some phenomena can be observed. Both an
intermetallic phase and a wavy pattern can be created during the bonding process. It is important to
notice that both phenomena are not a necessary condition for a high-quality weld. In Figure 2.8, a
typical wavy interface is shown. This image is taken during microscopic investigation of a copperbrass weld. A more detailed description of the wavy interface will be given in § 2.8 [11].
As mentioned above, also an intermetallic phase can be formed in the weld interface (see Figure 2.9).
Note that the weld in this figure (aluminium-copper) does not possess the typical waves pattern. This
intermetallic zone is created by the high temperature which occurs during the highly energetic
10
collision. However, the thickness of this layer is only 2 to 20 μm and is thus smaller than the interface
zone which is created by solid state welding processes.
Figure 2.8: A typical wavy interface which can occur in welds performed by the MPW process
Figure 2.9: Copper-aluminium weld. The darker grey zone is an intermetallic phase
Published investigations have also led to the knowledge that the intermetallic layer in magnetic pulse
welds has a higher hardness than both of the parent materials, though the difference can be
minimal. The following figures show the hardness distribution throughout the interface of welded
parts (Figure 2.10) and the distribution of the chemical composition across the interfaces (Figure
2.11). Note that the Vickers hardness of the intermetallic layer between the Cu-Al weld and the Ti-Al
weld appears to be around 400 HV in both cases. Although the hardness of the titanium is twice the
hardness of the used copper, the interface hardness is even slightly higher in the Cu-Al weld [12].
Figure 2.10: Hardness distribution across the interface of welded joints
(a) mild steel core, (b) copper core, (c) titanium core [12]
11
Figure 2.11: Distribution of the chemical composition across the weld interface; around 12 μm away from
the interface the amount of the other material (non base material) becomes zero [12]
2.5 Comparison with explosion welding
Explosion welding (EXW) is a “cold” welding technique that relies on a high energetic impact to
obtain a bond. It is often used for the cladding of materials with another material. The main principle
of MPW and EXW is thus the same. The difference however lies in the way the flyer material is
accelerated towards the other workpiece. In the MPW process the force is of a magnetic nature and
in the EXW process it is obtained by the detonation of explosives. Figure 2.12 gives an impression of
the explosion welding process. The main parameters of this process are the impact velocity and the
angle of impact. These parameters should be chosen so that a jetting action occurs. Phenomena that
occur during the MPW process, like wave formation in the interface, also occur in the explosion
welding process. This once again shows the analogy between the two processes [9][13][14].
Figure 2.12: Principle of explosion welding: flyer material is accelerated by detonation of an explosive[15]
For the EXW process more experimental research has been reported and a number of weldability
windows are available. An example of a weldability window is shown in Figure 2.13. This figure shows
both the lower and the upper welding limits, and a transition zone between smooth and wavy
interface [13].
During the research on the MPW process, these experimental data of the EXW process could provide
some guidance. It appears that in both impact welding techniques the same parameters are
influential in whether or not a good weld is obtained. The weldability windows of the EXW process
could thus be used as a starting point to obtain specific weldability windows for magnetic pulse
12
welding. It should be noted however that the two processes are not entirely the same. It is presumed
that the angle of impact is a constant during the explosion welding process. In magnetic pulse
welding, this angle is more difficult to control and changes during the welding process. Hence, if an
EXW weldability window prescribes certain values for the angle of impact, the weld produced by
magnetic pulse welding will only occur in that part of the workpiece where these conditions are met
[11].
Practically this means that there are three zones in the MPW weld :
1. Run-in zone: angle of impact is too small at the start of the weld. => no weld
2. Weld zone: angle of impact is in the correct range.
3. Run-out zone: angle of impact is too large at the end of the weld. => no weld
This is in direct contradiction with the EXW process where the chosen angle of impact is constant and
thus a weld is created throughout the entire work piece.
Figure 2.13: Weldability window for 6061 T0 aluminium alloy which also shows a transition zone from smooth to wavy
interface [13]
2.6 Material properties
The properties of the materials to be welded have an influence on the whole process. The properties
can be divided into two groups : electrical and mechanical properties.
2.6.1 Electrical properties of the material
Following electrical properties of the material are important for the MPW process:
•
•
electrical conductivity
magnetic permeability
13
As explained above, the shielding efficiency of a workpiece is dependent on the skin depth of the
given material. This skin depth should be small enough to enable a good shielding operation of the
thin-walled flyer tube.
If the product of the electrical conductivity and the magnetic permeability of a given material is too
small, the skin depth of that material will be too large and the flyer tube will not sufficiently shield
the magnetic field. Insufficient shielding will lead to a smaller radial force on the workpiece and thus
the process will not be efficient. When a material with a low electrical conductivity is to be welded, a
thin but highly conductive so-called driver material can be wrapped around the workpiece to enable
a smaller skin depth.
2.6.2 Mechanical properties of the material
The main mechanical properties which are of importance for the welding process are:
•
•
•
yield strength
strain hardening
strain rate hardening
As the outer workpiece undergoes severe permanent deformation during the welding process, the
mechanical properties of that workpiece are very important . The yield stress of the material (σy) and
the strain hardening can be related to the pressure which is required to deform the outer workpiece.
Due to the fact that severe deformations take place at a very high velocity, the strain rate
dependence of the elastic-plastic properties of the material will be of importance. A model that is
often used in the description of such processes, is the Johnson-Cook model. The influence of strain
hardening and strain rate hardening on the flow stress is obvious in the following equation:
J
FG = H9 + . IFG
K. H1 + L. MNIFO PK. Q1 −
With :
σpl
IFO
εpl
A, B, n
C
m
θ
θtrans
θmelt
R − RS4TJU
X
RVWGS − RS4TJU
(2.9)
: von Mises equivalent flow stress [MPa]
: the relative equivalent plastic strain rate [-]
: the equivalent plastic strain [-]
: yield and strain hardening constants [-]
: strain rate constant [-]
: thermal softening constant [-]
: absolute temperature when the stress is applied [K]
: transition temperature defined as the one at or below which there
is no temperature dependence on the expression of the yield stress [K]
: melting temperature [K]
This model describes the influence of strain hardening and strain rate hardening on the flow stress
[16].
14
2.7 Main geometrical parameters
As mentioned before, both the impact velocity and the angle of impact are of importance in creating
a high quality weld. To obtain prescribed values, the geometrical parameters of the process should
be chosen carefully. The main geometrical parameters are shown in the following Figure 2.14 [7].
Figure 2.14: The main geometrical parameters: stand-off distance (s) , flyer tube thickness (t), overlap length between
the field shaper and the workpiece (LF.S.) and the overlap length of the workpieces (Lwp)[17]
2.7.1 Stand-off distance, s
The stand-off distance is the space between the flyer tube and the inner workpiece. To obtain a
proper impact velocity a certain distance is necessary to provide enough space for the flyer tube to
accelerate. If s is too small, the impact velocity will be insufficient to obtain a weld.
However, a stand-off distance which is too large is not desired either. A large stand-off distance
allows the part to accelerate to a velocity which can exceed the optimal impact velocity. At this high
velocity, the kinetic energy of the flyer tube will now deform the inner workpiece rather than
creating a bond. This phenomenon was also observed during the aluminium-copper experiments
which were conducted for this work (§6.6).
2.7.2 Thickness of the flyer tube, t
When the thickness of the flyer tube increases, its resistance against deformation increases. The part
will also be heavier when the wall thickness increases and so a larger mass is to be accelerated. The
pressure (p) required for the deformation of a thin-walled ring increases with increasing thickness:
Y =
Z#
;
(2.10)
Note that this formula is only correct for deformations within the elastic region of the material, but it
gives an indication of the influence of the wall thickness.
15
Also the pressure necessary for the acceleration of the thin-walled workpiece will become larger. To
obtain a prescribed impact velocity the energy level should thus be increased for a larger wall
thickness [7][18].
2.7.3 Overlap length of the field shaper with the workpiece, LF.S.
This length describes how far the outer workpiece is placed in the field shaper. It is of importance
that this region is large enough to grant the flyer tube with enough energy for deformation. The
overlap length is in relation with the axial length of the field shaper.
This axial length equals the length of the field shaper over which the magnetic field is concentrated.
Hence it is a measure for the intensity of the magnetic field in the working region. It is important that
the axial length of the field shaper exceeds the overlap length, as the magnetic field might show
irregularities at the end of the field shaper.
2.7.4 Overlap length of the workpieces, Lwp
A sufficient overlap length is necessary to obtain a sufficient weld length. Obviously the weld length
can never exceed this overlap length. Also, if the overlap length increases, the magnetic field is
exerted over a larger part of the material and thus the formability at a certain energy level increases.
Measurements of the magnetic field are described in §4.3.2.
2.7.5 The relative position of the field shaper
The position of the workpiece in the field shaper plays an important role in the impact behaviour of
the workpieces. Two possible ways of field shaper placement are shown in Figure 2.15. In
configuration (a), the central part of the flyer tube will first impact on the inner workpiece. Thus a jet
is created in two directions and the weld will propagate in both directions. In the “end joint”
configuration (b), the end of the tube will impact first and the weld will propagate in only one
direction and less material needs to be deformed. This last configuration will thus require less energy
[19].
Figure 2.15: different ways of positioning the field shaper [19]
(a) middle joint
(b) end joint
16
2.8 Weld interface
2.8.1 The mechanism of wave creation
During microscopic investigation often a wavy pattern can be observed in the welded area. Figure
2.16 shows a typical example of such an interface of a copper to brass weld. Note the variation in
amplitude and period of the wavy pattern along the weld zone. This again demonstrates the analogy
with explosion welding. It is important to note that an impact weld does not necessarily require a
wavy pattern to obtain a good weld quality. The parameter window (c.f. collision angle versus
collision velocity) in which waves are created and the welding window in which a good quality is
obtained partially overlap but are not equal. Trying to obtain a wavy interface can in some cases
even result in a weld of lower quality [11].
Figure 2.16: Typical wavy pattern at the weld interface between copper and brass.
In literature concerning explosion welding and in general impact welding , several mechanisms are
described which can be the cause of the formation of this pattern [20].
1. Penetration of the jetting material
This theory dictates that the jet penetrates into the surface of the internal part. Due to this
penetration, movement of material is created and humps are randomly formed. This model is
considered to be inaccurate for the formation of a continuous wave. If such a continuous wave, with
no large jumps in wavelength, would be created by random hump formation, coincidence would be a
too important factor.
2. Kelvin-Helmholtz instability
This mechanism describes the interaction of pressure waves and the collision point velocity Vc (the
axial speed of the contact point). As the weld starts to form, an impact takes place which induces
compressive pressure waves through the internal workpiece. At the same time the contact point
moves in the axial direction with velocity Vc. Since the internal workpiece is not moving and the outer
is, a velocity difference is created between the two materials which can be considered as fluids at
these high velocities. When two fluids move over each other with a speed difference, waves are
created in the region of a Kelvin-Helmholtz instability, with mass transfer from the heavier to the
lighter material. One can see the analogy with wind over water. The instability necessary for the
formation of the waves is created by the interaction of the compressive waves with the collision
17
point. Since the speed of the waves is always at its maximum at the collision point, the pressure peak
will be situated at that point [21].
Figure 2.17 describes the mechanism in seven steps:
Figure 2.17: Kelvin-Helmholtz instability versus shockwaves [20]
a) First, compressive shockwaves are generated at the impact point, which move in a radial
direction with an angle proportional to α (= the impact angle).
b) The compressive shockwaves reflect on the back surface of the flyer material and thus
refraction waves are formed, which are depicted as the blue arrows.
The compressive waves in the internal solid workpiece collide and are reflected as
compression waves towards the interface of the materials. (black arrows)
c) The first wave can only be formed at the location where an interaction of the refraction and
compression waves takes place, which happens when their periods match. Furthermore, this
interaction should take place in the vicinity of the collision point. That point is under extreme
pressure and heat is generated, so the interaction of the shockwaves, in combination with
the mutual movement of the metals, creates the source of the interface waves.
d) From the moment the first wave is formed, a Kelvin-Helmholtz instability takes place and
waves are created periodically. The instability will generate waves as long as nothing decays
it.
e) New collision points are created and thus new waves are generated continuously. Due to the
massive metal movements across the material interface a new interference cannot be
created, so the waves form one continuous wave pattern rather than separate waves. The
pattern will thus be created by the interference continuity.
18
As the weld progresses, the propagation velocity Vc will decrease severely and the
interference point of the shockwaves now will meet the collision point further along the
interface. This will make the wavelength increase.
g) At last the propagation speed will drop to a amount so small that the interferences will now
take place ahead of the collision point and new waves can no longer be generated.
f)
This mechanism is considered to be accurate and has been experimentally verified on different
geometries. The shape of the inner part was varied between a solid rod, a tube and a tube in which
an eccentric internal hole was placed. The experiments proved that the wavelength of the interface
waves are proportional to the free path of the shockwave propagation in the inner part of the weld
[20].
2.8.2 Influence of parameters on the wave pattern
In the MPW process a lot of different process and geometrical parameters should be considered.
These parameters will affect the weld and thus will also have an influence on the wave pattern
formation. Next, the role of different parameters will be described. The energy level will not be
considered separately because its influence will appear in the other parameters (impact velocity for
example).
2.8.2.1 Thickness of the workpieces
The wavelength will increase with increasing thickness of the internal piece. This is due to the fact
that the compressive waves will have to cross a larger distance through the material before their
reflections reach the surface[22].
Table B gives an example of the same phenomenon, describing the changes in wavelength when the
thickness of the flyer plate/tube increases. These values were derived from finite-element
simulations of the impact of a copper flyer plate at a collision velocity of 650 m/s with an angle of
impact of 15° onto a 15 mm thick copper base plate[22][23].
Table B: Flyer plate thickness influences the wavelength[22]
2.8.2.2 Impact velocity
Generally, it is presumed that the wavelength correlates with the impact velocity. To obtain the
wave formation, a critical velocity has to be reached. Below this velocity no interference of
shockwaves will take place and thus no waves will be formed. Investigation has shown that the best
weld quality is obtained when the impact velocity is around this critical velocity [9].
As stated above, the flyer plate/tube thickness (t) will also influence the wave shape, the wavelength
(λ) and the amplitude (A). For this reason these parameters are often normalised as λ/t and A/t.
Figure 2.18 shows a chart of the normalised amplitude as function of normalised wavelength for
different impact welding methods. It indicates that amplitude and wavelength tend to scale with
19
each other. Large energy levels will not only result in larger wavelengths but also in larger amplitudes
[23].
This correlation can be explained by the fact that both the loading pressure and the elastic shearing
deformation (the deformation of the surface material) increases with increasing velocity. The jet will
then contain a larger mass of material and the amplitude of the disturbance will increase as well as
the amplitude. This theory is valid when the impact velocity is in the subsonic region. When the
impact velocity exceeds a certain value, the amplitude will no longer increase. This critical value of
the velocity seems to be in relation with the mach number of the impact velocity. This mach number
equals the impact velocity divided by the speed of sound in air. Figure 2.19 shows the relation
between the wave amplitude and the impact velocity for different materials (copper, steel and
aluminium) at a constant angle of impact, α=12°. The velocity of impact has been normalised with c0,
the speed of sound in that material, so that the relation can be compared for the different materials.
It shows clearly that the critical value takes place in the region around M=1 to 1,5 [24].
Figure 2.18: normalised amplitude vs. normalised wavelength for different impact welding processes (EXW= explosion
welding, LIW= Laser Impact Welding, MPW= Magnetic Pulse Welding)[23]
Figure 2.19: Relation between wave amplitude and normalised impact velocity[24]
This phenomenon can be explained as follows: at the critical value, the jet will shift from subsonic to
supersonic regime, resulting in a maximum wave amplitude. A further increase of impact velocity will
20
also further increase the sliding velocity between the two metal surfaces. The contact time between
the surfaces becomes too small and the waves can no longer be formed completely.
2.8.2.3 Angle of impact
There is a lower and an upper limit on the value of the collision angle for which jetting between two
materials occurs. Thus the collision angle is very important for the weld quality. The angle of impact
should generally be between 6 and 14°. It is important to notice that the impact of the outer tube
with the internal workpiece does not occur with a constant angle of impact. During the process, the
angle will increase while the weld propagates (Figure 2.20). Figure 2.21 shows the effect of the
collision angle on the amplitude and wavelength (a and λ respectively). First, both the amplitude and
the wavelength increase with an increasing angle of impact. After the collision reaches a certain
value, only the wavelength will increase further and the wave amplitude will decrease. Finally the
wave amplitude will reach zero and the weld interface in that zone will not have a wavy pattern[9].
Figure 2.20: The collision angle increases during the propagation of the weld. With α0 the initial impact angle and αe the
angle of impact further in the weld
Figure 2.21: Effects of collision angle α on the parameters a and λ[9]
2.8.2.4 Stand-off distance
Experiments have been conducted using several stand-off distances ranging from 0,5 to 3 t, t being
the flyer plate thickness [25]. The results of this investigation showed that the waviness of the
interface increases when the stand-off distance increases. Both the wavelength and the amplitude of
the waves increase. Figure 2.22 shows the interfaces of a copper-stainless steel weld which were
welded with a stand-off distance of 2 and 3 t. In these experiments, explosion welding was used. The
weld with stand-off distance s = 2 t resulted in a wavelength of 750 μm and in an amplitude of
130 μm. In the weld with s = 3 t, a wavelength of 950 μm and an amplitude of 150 μm was measured.
21
Together with the amplitude, also the penetration depth of the metals increases resulting in a
stronger mechanical lock between the materials. When the stand-off distance is further increased,
the wavelength will also further increase and the weld will seem to be free of waves.
Figure 2.22: Stand-off distance on the waviness; s= stand-off distance, t= thickness of the flyer material[25]
2.9 Advantages of the process
The main advantages of MPW can be summarized as follows [1] [7] [19]:
•
Since the MPW process is a “cold” welding process, the heat affected zone (HAZ) is only
generated very locally over a thickness of about 20-50 µm. All material properties will be
maintained. Any heat treatments that were performed prior to the welding will not be
influenced. Also the parts can be finished prior to welding.
•
Due to the low temperature, the part can be unloaded immediately after welding and no cooling
down phase has to take place prior to further processing.
•
No preparation of the parts is required. The parts do not have to be cleaned prior to welding.
•
Joining of non-weldable or dissimilar materials in a quick and cost-effective manner.
•
No post weld finishing or cleaning has to be carried out. Figure 2.23 compares a MPW weld to a
MIG/MAG-weld. The left part (MPW) is has a very clean appearance and does not need finishing
for both structural or aesthetical reasons.
•
It is possible to improve the work conditions of the welder or operator, since the technology is
environmentally clean (no heat, radiation, fumes, shielding gases) and the hardest labour is
performed by machines. However, since the process uses a strong magnetic field precautions
should be taken. The distance between the operator and the coil should be sufficient and people
who use a pacemaker should not be allowed anywhere near the MPW machine.
22
•
The process welds the parts together in less than 100 μs. With a good choice of weld parameters
and a good clamping device, the process should be able to obtain a large reproducibility. Due
these factors the process allows a great degree of automation.
Figure 2.23: Automotive AC accumulator. Left = MPW, Right=MIG weld
2.10 Limitations of the process
Magnetic pulse welding imposes some limitations regarding the workpieces to be welded [7] [19]:
•
Since the process relies on the formation of eddy currents in the flyer material, only materials
with a high conductivity can be used for the outer piece.
•
The geometry of the parts to be welded is limited to tubes and sheets. No other shapes have yet
been welded. Also some parts of the workpiece are more difficult to weld than others (corners
for example).
•
The size of the parts is limited from 5 to 121 mm. In the open literature, it is reported that the
largest diameter of tubes that has been welded until now is 121 mm. The maximum size is also
limited by the cost of the machine, which increases significantly with increasing size of the parts.
•
The parts must overlap to generate the joint and thus no flat surface can be produced.
•
Due to the size of the welding machine, the welds can only be carried out in a workshop. Hence
the process is not suited for in-field applications. However since one of the main advantages of
the process is the high degree of automation, the main application lies at large quantities of
factory-made parts. So this remark should not necessarily be regarded as a vast disadvantage.
23
Chapter 3
Analytical model
3.1 Introduction
An analytical model is essential to gain insight in the parameters governing the MPW process, and to
make a quick estimation of the parameter values required to obtain a successful weld. However, it is
not straightforward to develop a set of equations that is able to accurately model the MPW process.
The discharge current is a damped sinusoidal wave, which results in a time-dependent magnetic
pressure. Often a field shaper is used to increase the amplitude of the magnetic field. The pressure
exerted on the tube will theoretically be a function of axial and circumferential position, as well as
time.
Furthermore, the calculation of the plastic deformation and acceleration of a cylindrical tube under a
variable radial pressure (acting only on a part of the tube) is very difficult. The complex deformation
behaviour of the tubular workpieces and high speed deformation both add to the problem of finding
equations that have reasonable accuracy, as well as sufficient simplicity.
This chapter intends to give an overview of available analytic equations, with a discussion on their
applicability. A relatively simple model by Pulsar is discussed, and the inaccuracies of this model are
exposed. It is not possible to develop a complete an improved model within the limits of this
dissertation. But, the key analytical concepts necessary for an accurate model are discussed.
24
3.2 Model by Pulsar
3.2.1 Introduction
The analytical model discussed in this section was developed by the manufacturer of the welding
machine (Pulsar). This model should allow users to choose the parameters of the welding process
needed to obtain a successful weld. Although the structure of this model is essentially correct, a
multitude of simplifications result in decreased accuracy of the model. The simplifications are
described, and afterwards alternatives are proposed. The Pulsar model is quite simple because of the
applied simplifications. This is an advantage, because it is intended to be used as a tool for quickly
determining optimal parameters. However, corrections are necessary to improve the accuracy of the
model.
3.2.2 Structure of the model
The structure of the simplified analytical model proposed by Pulsar is schematically shown in Figure
3.1 [26].
Figure 3.1: Schematic overview of charging voltage calculation
i.
The collision velocity, vc, is first chosen depending on the materials to be welded. Considering
the analogy between MPW and explosion welding, this data can be derived from experiments
carried out with EW [13][9]. As discussed in Chapter 2, values for the impact velocity and
collision angle should be chosen from weldability windows to obtain a successful weld. The
required acceleration can then be calculated, under the assumption that the velocity increases
linearly from zero to vc when travelling a distance equal to the stand-off distance, the distance
between the flyer tube and the inner work piece.
ii.
This acceleration is generated by an electromagnetic pressure, exerted on the flyer tube. The
required pressure can be found as the sum of two components: the pressure necessary to
accelerate- and the pressure needed to deform the flyer tube.
25
iii.
The exerted electromagnetic pressure is caused by the magnetic field. The magnitude of the
magnetic field is calculated from the required pressure.
iv.
The magnetic field originates from the discharge current, which flows through the coil
windings. The current, and subsequently the charging voltage over the capacitor bank are
calculated. The voltage is the only machine parameter that can be set.
The calculations associated with this structure are discussed in detail in the next section ‘Analysis’.
3.2.3 Analysis
3.2.3.1 Collision velocity (i)
This velocity is chosen out of a range of velocities which should lead to a good weld. These values are
the same as for the explosion welding process and can be found for different material combinations
like steel-steel, copper-steel, aluminum-steel, … [26]
It is assumed that the flyer tube is accelerated radially inwards until it impacts the inner workpiece.
The acceleration needed to reach the chosen velocity after travelling a distance equal to the standoff distance, is calculated as:
[=
\ $
2. ]
(3.1)
With: a = acceleration [m/s2]
vc = collision or impact velocity [m/s]
s = stand-off distance = length of air gap between the inner rod and the flyer tube [m]
This formula is based on the theory of linear motion with constant acceleration, as a part of the
formula:
2. ^]
\
=
=[=
$
(^;)
^;
_
(3.2)
Based on these equations, the process time is determined as:
2 Δs
t= (
a
(3.3)
This theory is based upon the assumption that the acceleration during the process is constant, which
in the case of the MPW process is not valid.
3.2.3.2 Required pressure (ii)
The transient discharge current flows through the coil windings, thus generating a magnetic field
inside the coil. This magnetic field is concentrated to a narrow axial zone, where the workpiece is
located. The transient magnetic field induces currents in the flyer tube. Consequently, the magnetic
field exerts a Lorentz force on the current conducting tube. This Lorentz force acts on the part of the
tube, where currents are induced. The axial length over which the force acts, is equal to the length
that the tube overlaps with the field shaper, hence the name ‘overlap length’. As the Lorentz forces
26
are a direct consequence of the magnetic field, they are associated with a magnetic pressure. This
magnetic pressure, exerted on the flyer tube both deforms and accelerates the tube.
So, the total required pressure (P) is calculated as the sum of these two components:
d = deW2 + dT\\
(3.3)
With: Pdef = Pressure required for the deformation of the flyer tube
Pacc =Pressure required for the acceleration of the flyer tube
Deformation Pressure
The calculation of the pressure required to deform the flyer tube is based on the theory of thin
walled pressure vessels (equation 3.4). A radial pressure acts on the tube, which is regarded as a
thin-walled vessel with radius R and thickness t.
deW2 =
2 . ; . fgh
(3.4)
With: R = average flyer tube radius = (Rout+Rin)/2 [mm]
t = flyer tube thickness[mm]
σUTS = ultimate tensile strength of the material [N/mm2]
Several remarks can be made on this formula proposed by Pulsar. First of all, the formula for thinwalled vessels is valid only for elastic deformations. The tubes deform plastically during
electromagnetic compression. It is also assumed that during compression, the circumferential
stresses in the flyer tube reach the ultimate tensile strength of the tube material (at which fracture
should occur). The tube will deform plastically when the compressive circumferential stresses reach
the material’s yield strength.
Finally, the formula assumes that the pressure is exerted over the entire axial length of the tube or
vessel. In the MPW process, the radial pressure acts only on the overlap length (Figure 3.2).
The diameter of the tube is thus not uniformly reduced. In reality, the tubes deformation behaviour
is very complicated.
Figure 3.2: Geometry of the flyer tube and inner rod. Magnetic pressure acts only on the part of the tube that overlaps
with the field shaper.
27
Acceleration Pressure
As mentioned before, the flyer tube velocity at impact is chosen. Below a certain limit, no jet will be
created and the process will not result in a weld [9] [1].
The tube must be accelerated from standstill to the chosen impact velocity (vc) over the stand-off
distance. The pressure required to accelerate the tube, is calculated based on the assumptions that
the movement of the tube is linear and that no other forces act on the tube.
Equation (3.5) states that when a total force F acts on an object of mass m, this object will be
accelerated with acceleration a (in the direction of the total force). The mass of the flyer tube that is
accelerated is calculated as the product of the tube volume and the density of the tube material. It is
important to note that the Lorentz forces only act on the overlap length, so it is assumed that only
this part of the tube is accelerated.
= _ . [ = (2 ; i). [
(3.5)
With: F = force [N]
R = average flyer tube radius[m]
l = overlap length [m]
t = tube thickness [m]
ρ=density of tube material [kg/m3]
a = acceleration [m/s2]
The force F, which causes the acceleration, originates from magnetic pressure acting on the tube’s
outer surface A.
dT\\ =
=
9 2 (3.6)
; . i . j\ $
2 .]
(3.7)
So, the pressure (Pacc) required for the acceleration can be calculated by substituting equation (3.1)
and (3.5) into equation (3.6):
dT\\ =
In the MPW process, the mass of the flyer tube is not accelerated linearly. First of all because the
magnetic pressure acting on it is strongly time-dependent (as discussed further in this chapter). In
addition, the tube resists against deformation by deforming elastically and (mostly) plastically.
Furthermore, the influence of the static part of the flyer tube is not taken into account. The radial
pressure acts on the overlap length, and the tube material located to the left of the inner rod collar
on Figure 3.2 cannot move. The static part will exert forces on the accelerated part of the flyer tube,
causing resistance against deformation. The open end of the tube (right side of the tube on Figure
3.2) will accelerate the fastest and impact first, because the resistance against deformation is the
smallest at this location. Because the deformation process is continuous, the entire tube does not
impact the rod at the same moment in time. An FE simulation of the deformation profile of the flyer
tube during electromagnetic compression is shown in Figure 3.3 [19].
28
Figure 3.3: Deformation profile of the flyer tube during electromagnetic compression
Magnetic pressure
The magnetic pressure accelerates and deforms the tube. By substituting equations (3.4) and (3.7) in
equation (3.3), the magnetic pressure equals the sum of both components:
d = dVTkJWSl\ = ; . m
i . j\ $ 2 . fgh
+
n
2 .]
(3.8)
3.2.3.3 Magnetic Field (iii)
As previously explained, the transient magnetic field is the cause of the induced currents in the tube,
and of the Lorentz force acting on the conducting tube (which can be regarded as a current loop).
The Lorentz forces are associated with a magnetic pressure. So, the pressure exerted on the tube is a
direct consequence of the magnetic field.
Magnetic pressure is in fact an energy density associated with a magnetic field. The magnetic field
strength between the tube and the rod is much smaller than the field strength between the tube and
field shaper. The gradient in field strength gives rise to a magnetic pressure that compresses the tube
radially inwards.
The magnetic pressure is calculated by Pulsar with the assumption that the field strength between
the flyer tube and inner rod is negligible:
d=
$
2. (3.9)
29
With: Pmagn = magnetic pressure [Pa]
B = magnetic field [T]
μ0 = magnetic permeability of free space = 4π×10−7 H/m
Solving equation (3.9) for B yields:
= 2. . . d
(3.10)
Substituting the required magnetic pressure from equation (3.8) into equation (3.10) yields:
= (2. . . ; . m
i . j\ $ 2 . fgh
+
n
2 .]
(3.11)
3.2.3.4 Current and charging voltage (iv)
The magnetic field originates from the discharge current that flows through the coil and field shaper.
The MPW machine allows setting only one parameter: the charging voltage over the capacitor bank.
The machine, field shaper and workpiece form one electrical discharge circuit. The charging voltage is
thus directly related to the discharge current that flows through the coil. The coil current induces a
current in the field shaper, which is located inside the coil. The discharge current flowing through the
coil windings generates a magnetic field. The field shaper concentrates the field to a narrow axial
zone where the flyer tube is located, thus increasing the field magnitude at the welding zone.
The relation between the discharge current and the magnetic field strength inside the field shaper is
very difficult to calculate analytically. The Pulsar calculations regarding the currents are not fully
understood. They involve the calculation of the current density on the inside field shaper surface.
The total current flowing on the inside surface of the field shaper Itotal [A], is calculated as the sum of
separate components:
oSxSTG = oy.z. + 2. oWJe = o + 2. o△ + 2. oWJe
(3.12)
The following equations allow for the calculation of the total field shaper current.
p. ]
q1 − r
2 2
p
t1 − ln (2)w
o∆ = $
p
oWJe = $
Qln @ B + 0,423X
2]
o=
With the magnetic flux:
$
p = . 9 = . . (\xlG
− $ )
(3.13)
(3.14)
30
With: B = magnetic field between field shaper and flyer tube [T]
Φ = magnetic flux [Wb]
Rcoil = radius of the coil [m]
Combining equations (3.12) up to (3.14), the following relation is obtained:
oSxSTG =
$
2. . (\xlG
− $ ) ]
1
. Q }1 − ~ +
(1 − M2 + ln @ B + 0,423)X
4]
2
2]
(3.15)
The equations in (3.13) describe the current distribution on the inner surface of the field shaper. The
shape of the current distribution can be motivated by multi-physics FE simulations. The exact
analytical calculations (3.13) are however not fully understood.
Alternating electric current has the tendency to distribute itself within a conductor so that the
current density near the surface of the conductor is greater than at its core. This is called the skin
effect [16]. In several FE simulations, the current density on the field shaper inner surface was
calculated. It was observed that the current density is much higher towards the edges [16] [27]. An
example is shown further in this chapter, in Figure 3.23.
The only figure given by Pulsar to clarify their proposed calculations is shown in Figure 3.4 (left) [26].
Because of the simulation results indicating higher current densities at the edges and the analogy
with Figure 3.23, it was concluded that the Figure 3.4 represents a simplification of the current
density of the inner field shaper surface. The right figure of Figure 3.4 clarifies this. The partition of
the current density into Iwz, Iend, IΔ, and I, is probably used to simplify calculations. There was no
discussion on the analytical calculations in the documents provided by Pulsar.
Figure 3.4: Simplified current distribution in field shaper [26].
31
It can be seen that in equation (3.13) and (3.15) the dimensions are not correct. Pulsar claims that
the dimension of all the currents in (3.13) is Ampère [A]. In equation (3.16), the SI unit of the ratio
ϕ/μ0 is calculated in brackets:
„
q r
€ t‚w
9
=
=
= t9. _w
ƒ
…
q r
q $r
_
9
(3.16)
It is clear that of the three currents in equation (3.13), only IΔ and Iend are expressed in [A]. Current I is
expressed in [A.m]. Therefore, equation (3.15) is not valid, as it is a sum of terms with different units.
In addition, it would only make sense to calculate the current at the inner field shaper surface, with
the assumption that the magnetic field between the tube and the inner field shaper surface is
generated by this current. Indeed, in the calculation of the magnetic pressure, the field strength
inside the flyer tube was assumed negligibly small. Therefore the parameter Rcoil should be chosen as
the inner radius of the field shaper, not as the coil radius.
Several attempts were undertaken to receive clarification by Pulsar, unfortunately without satisfying
response.
The frequency of the discharge current can be calculated as:
=
1
2. . √L. ‡
(3.17)
With: f = frequency of the discharge current [Hz]
C = total capacitance of capacitor bank [C]
L = total inductance of the system [H]
The final step is to calculate the charging voltage from the total discharge current.
j=
With:
=
oSxSTG
. ˆ
L
‡
(3.18)
oSxSTG,,VWTU‰4We oSxSTG,,VWTU‰4We
=
oSxSTG,,SŠWx4WSl\
2. . . L. j
(3.19)
I1 = first current peak [A]
δ = system attenuation coefficient [-]
V= charging voltage [V]
Another definition for the attenuation coefficient was given:
With: I1 = first current peak [A]
I3= third current peak [A]
= >
oSxSTG,
?
oSxSTG,"
.$‹
(3.20)
32
The charging energy is the energy (E) stored in the capacitor bank after the charging process and can
be calculated as:
8=
L j$
2
(3.21)
3.2.4 Discussion
The model proposed by Pulsar is (at this moment) the only analytical model available to describe the
entire MPW process. However, after critical evaluation it is obvious that several simplifications and
assumptions made in this model, limit the accuracy of its predictions. The most important
simplification, on which the entire Pulsar model is built, is that the time-dependency of the MPW
process is completely neglected.
First of all, the acceleration is assumed to be constant. This would require the magnetic pressure
exerted on the flyer tube to be constant. However, this magnetic pressure originates from the
damped sinusoidal current through the coil. Because the magnetic field is a sinusoidal damped wave,
the magnetic pressure exerted on the flyer tube is time-dependent (pressure pulses, discussed
further in this chapter). As a consequence, the acceleration is also time-dependent. In addition, the
magnetic pressure wave will decrease in amplitude because the air gap increases during the
compression of the tube. For accurate determination of the impact velocity, velocity as a function of
time should be calculated. This is only possible if the deformation behaviour of the tube is fully
understood and can be modeled analytically.
In the Pulsar model, the pressure required for deformation is determined in a simplified way. It is
calculated as the pressure for which the ultimate tensile strength is reached in a thin walled
cylindrical tube subjected to radial compression. This formula can solely be used in case of linear
elastic deformations and the simplification would suggest that the pressure compresses the entire
tube with a radial displacement equal to the stand-off distance. In reality only one end of the tube is
plastically compressed, and this at an extremely large deformation speed.
As stated before, several calculations in this simplified model were not fully understood. Because
insufficient support was given by Pulsar, uncertainty still exists regarding the correctness of these
particular calculations.
33
3.3 Electrical circuit
3.3.1 Introduction
The electrical discharge circuit consists of the capacitor bank, wires, the coil and the workpieces. The
circuit has been modeled with the goal of calculating the discharge current.
Different types of circuit models can be found in literature. There is no consensus as to which model
would be optimal. It is a trade-off between a large number of parameters, leading to a precise model
but rather difficult to apply and with less insight in which effects actually occur, and on the other
hand a rather elementary model with few parameters. The RLC circuit is the most cited model in
literature.
3.3.2 RLC circuit
3.3.2.1 Circuit description
The electrical discharge circuit can be modeled as an RLC circuit. In this electrical circuit, C represents
the total capacitance of the capacitor bank, R the equivalent resistance of the electrical circuit (circuit
wires, coil and workpiece) and L the equivalent inductance (circuit, coil, field shaper and workpiece).
The advantage of this model is that it is easy to work with, as there are a very limited number of
parameters. The relationships between variables are obvious, which leads to an improved insight of
the key process variables.
The RLC circuit, shown schematically in Figure 3.5, is e.g. used in [28], [29] and [30].
Figure 3.5: RLC circuit
An essentially similar representation of the electrical circuit is shown in Figure 3.6.
R0 and L0 are the natural resistance, respectively the natural inductance of the discharge circuit. LE is
the equivalent inductance and RA the active resistance of the work piece-inductor system. The term
workpiece-inductor system is used to describe the system combining the workpiece, coil and field
shaper (if present). In the figure of the circuit, an additional resistance RF is depicted. This is a very
34
small (and thus negligible) resistance, used for the current measurement. RF will not be taken into
account, as the current measurement in this thesis uses a Rogowski coil [31] [32].
Figure 3.6: RLC circuit, similar to Figure 3.5
The analogy with the RLC model of Figure 3.5 can be explained because these resistances and
inductances should be modeled in series. Indeed, when the workpiece is put in the discharge circuit,
the current in the workpiece is induced by the coil current. According to the law of electromagnetic
induction, the current in the workpiece is reverse to the current in the coil. So the work piece should
be regarded as an inductor installed in series with the coil in the discharge circuit.
Therefore the total inductance and total resistance are given by:
‡ = ‡ + ‡
= + A
(3.22)
This reduces this model to the original RLC circuit [31].
3.3.2.2 Circuit Analysis
Kirchoff’s second law states that the directed sum of the electrical potential differences around any
closed circuit must be zero. The electrical circuit does not contain an external voltage source. The
capacitor bank is charged to a certain initial voltage, as required for a successful welding process.
Applying this to the RLC circuit yields:
Œ (;) +  (;) + Ž (;) = 0
∀;
(3.23)
Deriving this equation to time leads to the differential equation of an RLC circuit without an external
voltage source:
$‘
‘
1
(;) + . (;) +
‘(;) = 0
$
;
‡ ;
‡L
∀;
(3.24)
$‘
‘
(;) + 2’. (;) + “$ ‘(;) = 0
; $
;
∀;
(3.25)
This differential equation can be simplified by introducing two parameters: β (damping factor) and ωc
(current angular frequency). This substitution yields the following equation governing the RLC
circuit:
35
with:
’=
“ =
2. ‡
1
(3.26)
√‡. L
As stated previously, there is no external voltage source in the circuit. The operator of the MPW
machine manually sets the initial voltage across the capacitor bank. At the start of the process, there
will be no current flowing through the circuit. Equation (3.25) can be solved for the current
waveform, using the two initial conditions: at the start of discharge no current flows through the
circuit and the capacitor bank is charged to voltage V0.
‘ (; = 0) = 0
\ (; = 0) = ‡.
‘
(; = 0) = j
;
→
‘
j
(; = 0) =
;
‡
(3.27)
The current can be expressed as a time function:
‘(;) =
j
. C :•S . sin (“Œ ;)
“Œ ‡
“Œ = ˆ“$ − ’ $
(3.28)
The discharge current is a damped sinusoidal wave as shown in Figure 3.7.
Three variables determine the circuit (R, L and C), however only the capacitance of the capacitor
bank is known in advance. The producer of the MPW machine can provide its value. The capacitance
is in most cases not fixed, as the capacitor bank exists of multiple capacitors connected in parallel.
Therefore separate capacitors can be disabled and it is possible to obtain a different capacitance. In
practice this total capacitance will be a multitude of the capacitance of a single capacitor.
Values of resistance and inductance are in most cases not available beforehand. Analytical equations
to estimate the inductance of multi-turn coils and of field shapers were found in[33]. The two
parameters (R and L) can also be estimated by a current measurement. The measurement of the
discharge current waveform is discussed in Chapter 4: a Rogowski coil is placed around the wires
connecting the capacitor bank and the coil. The current passing through the Rogowski coil will be
measured and visualised on the computer by a digital oscilloscope.
Measuring the current is essential in the RLC circuit model used in this thesis, because it is possible to
extract the resistance and the inductance from the current waveform. Given the measured current
waveform of Figure 3.7, the two parameters β (damping factor) and ωc (angular frequency of the
discharge current) can be extracted.
The current is a damped sine wave, and the relation between the frequency (fC), the angular
frequency (ωC) and the period (TC) is:
“Œ =
2
= 2Œ
—Œ
(3.29)
The parameter ωc can be found by extracting Tc from the current measurement. The local peak
values (positive and negative) are reached at fixed time intervals: Tc/4, 3Tc/4, 5Tc/4, 7Tc/4, etc. and
36
the current is momentarily zero at multiples of Tc/2. So, Tc/2 is simply the time interval between two
peak values of the current, or between two points of zero-current. As the amplitude of the wave
decreases exponentially, only the first couple of peak values are accurately measurable.
Figure 3.7: Measured current waveform[34].
The value of β can be determined by taking the ratio of the first two peak values of the current. The
first peak is positive and occurs at tpeak,1=Tc/4; the second peak is negative and occurs at tpeak,2=3 Tc/4.
According to the calculated time function i(t), this ratio can be expressed as:
•g™
‘FWT˜ ‘ (;FWT˜ )
=
=C $
‘FWT˜$ ‘ (;FWT˜$ )
(3.30)
Using equation (3.30), it is possible to make a good estimate of the damping factor β:
‘FWT˜
B
‘FWT˜$
—Œ
(3.31)
1
+ ’$)
= 2. ‡. ’
(3.32)
’=
2. ln @
The experimentally determined values for β and ωc can be used to determine values for the
inductance L and resistance R, using equations (3.26) and (3.28)
‡=
L. (“Œ$
In conclusion: the value of C is given by the capacitance of the MPW machine, and L and R can be
obtained by a current measurement (using a Rogowski coil). These three values R, L and C determine
the complete electrical circuit, representing the machine and the workpiece. By choosing the initial
voltage, the current waveform can be obtained as shown above.
37
3.3.2.3 Time-dependency
The equivalent inductance and resistance of the workpiece-inductor system are time-variant during
the electromagnetic forming process. Therefore the electrical circuit can strictly not be regarded as a
RLC circuit.
However, studies show that most of the energy is transferred to the workpiece during the first period
of the electromagnetic pressure wave [31]. As the pressure frequency is twice the current frequency,
this suggests that most energy is transferred in the first half period of the current. In addition, the
stand-off distance is small so the deformation of the workpiece will be relatively limited. Taking these
two arguments into account, it can be assumed that inductance and resistance will not show
extreme variations during the process, so an RLC circuit gives a qualitative model to calculate the
discharge current [31].
The effect of flyer tube deformation on the current waveform is of importance to realise which
effects are neglected when using the RLC model. Due to Lenz’s Law, the current in the workpiece is
reverse to the current in the coil. The workpiece can be regarded as an inductor installed in against
the coil in the discharge circuit. The presence of a workpiece essentially reduces the equivalent
inductance [31]. During the process, the flyer tube is compressed, which results in an increase of the
radial gap between the workpiece (flyer tube) and the coil (or field shaper). This motion leads to an
increase of the equivalent inductance. Both current amplitude and frequency are affected by this, as
illustrated in Figure 3.8 [29].
The influence of the distance (denoted as h) between the workpiece (flat sheet) and the inductor
(spiral coil) extensively investigated in [31] by means of experiments. The resistance and inductance
were calculated as described in § 3.3.2.2, for different values of h (which was kept constant during
each experiment). The experimental data was fitted to an exponential function, as shown in Figure
3.9.
Figure 3.8: Influence of flyer tube deformation on coil current waveform [29]
38
Figure 3.9: Variation of L and R with h [31]
The results clearly show that the inductance increases and the resistance decreases when the
distance between workpiece and inductor increases. Both curves converge to a constant value,
representative for the situation without a work piece. Note that the relative increase of L with h is
significantly larger than the relative decrease of R with h. For example, comparing the measurements
at h=0 mm to those at h=150 mm: L increases from 6 μH to 23 μH (283% increase), while R decreases
from 104 mΩ to 72 mΩ (only 31% decrease). So, the resistance will not vary as much as the
inductance when the workpiece is deformed.
As a result the peak current will decrease slightly, as illustrated in Figure 3.10 (left). The damping
factor of the current decreases very sharply, which is also illustrated in Figure 3.10 (right). By the
same comparison (h=0 mm to h=150 mm), the peak current decreases with 22,5% and the damping
factor decreases with 76%. It should be noted that the calculated variations render qualitative
insight, but that the exact values cannot be transferred to another MPW machine (as they are
dependent on the geometry of the setup).
Figure 3.10: Variation of peak current and damping factor with h [31]
3.3.2.4 Conclusion
The RLC model for the electrical discharge circuit can be used for the magnetic pulse compression of
tubular workpieces. It renders a quick estimation of the waveform of the discharge current. V0, the
39
initial voltage, is chosen for each experiment. The capacitance (C) is inherent to the MPW machine.
The resistance (R) and the inductance (L) need to be determined experimentally through a
measurement of the current waveform.
One must realize that the inductance and resistance depend on the geometry of the field shaper and
workpiece, and will even change during the process. Based on the discussion in the previous
paragraph, the variation of the resistance is expected to be of less importance than the variation of
the inductance.
These values, L in particular, will show variation if there is a difference in:
•
flyer tube material (shielding)
•
field shaper geometry or material
•
workpiece geometry (stand-off distance, overlap length)
The variations do not imply that the RLC circuit should be abandoned.
When using a single MPW machine, an effective method would be to perform current measurements
for a set of geometries and materials for each field shaper. Using curve fitting, a function could be
extracted to model the influence of the above parameters on the value of the equivalent inductance
L. Once this function has been determined, an accurate prediction of the current waveform can be
made for each set of future experiments.
3.3.3 Other electrical circuit models
More complex circuit models are found in literature. They are solely used for FE calculations, never
for analytical equations. An example of a detailed circuit model is shown in Figure 3.11. This electrical
circuit models contains thirteen parameters and is used in [33] and [16].
Figure 3.11: Circuit model used for FE simulations
This model approaches reality more accurately, but the large number of parameters makes it
unsuitable for analytical models. The determination of these parameters is a rather elaborate task, as
discussed in [16]. The frequency dependent parameters, are calculated both analytically and using
the multi-physics FE model (for comparison). The analytical formulae for the inductance of the coil
use the Nagaoka coefficient. [16] states that quite large differences are found between the FE results
and the Nagaoka formula.
40
3.4 Electromagnetic energy transfer
3.4.1 Coil
The discharge current, as calculated in the previous section, flows through the coil. Within the coil an
axial magnetic field ― proportional to the current ― will originate. Figure 3.12 shows the magnetic
field lines. Because the current is time-dependent, the magnetic field will also be a function of time.
The presence of a field shaper makes the analytical model more difficult to establish. The field shaper
is generally used to concentrate the magnetic field, by increasing the magnetic field amplitude in a
smaller region. The effect of the field shaper will be discussed in the next section. In this section we
will not take its effect on the magnetic field into account.
Figure 3.12: Magnetic field lines in a solenoid coil [35]
A solenoid coil with length L and N turns, conducting a current i(t) will induce a magnetic field, which
can be calculated according to Ampère’s Law:
(;) =
…o(;)
(3.33)
With: B= magnetic flux density [T]
I= current [A]
l=coil length (axial length) [m]
N=number of windings
μ=μ0 . μr = magnetic permeability [H/m]
μ0 =4π.10-7 H/m = magnetic permeability of vacuum [H/m]
μr = relative magnetic permeability
In an ideal solenoid, with large length relative to its diameter and no separation between windings,
the magnetic field can be assumed nearly uniform (constant in magnitude and along the axial
direction) inside the cross-section of the coil [36]. However, in the larger magnetic pulse welding
machines, the currents are extremely high (>100 kA). Therefore solenoid coils are no longer
applicable. For the MPW process single turn coils or multi-turn coils with a low number of windings
are mostly applied, often in combination with a field shaper (inserted in the coil). On the left of
Figure 3.13, the single-turn coil is shown, and on the right a multi-turn coil with 5 windings.
41
Figure 3.13: Single-turn coil (left) and multi-turn coil (right)
The problem with these larger coils is that they have no standard geometry, like solenoids. Currents
are very high, so the conducting coil metal is no longer in the form of wires (Figure 3.13), and the
spacing between windings is relatively large. Also, the coil length relative to its diameter is smaller,
which increases the stray fields. The magnetic field is strictly a function of axial and radial position.
This is probably too complicated to be taken into account in the analytical model. In conclusion, the
coils are often custom made and as a result no general formulae are available for the expression of
the magnetic field.
In theory the relation between magnetic field and current is determined only by the geometry of
both coil and workpiece (which determines air gaps).
(;) = (C_C;#š ›‘ & #žZ‘C›C). o(;)
(3.34)
The magnetic field in the gap between coil and workpiece is proportional to the discharge current,
and the conversion factor is a constant, based on geometrical parameters.
An analytical equation for the conversion factor was found in [37]. Figure 3.14 shows the equation
and geometry of the multi-turn coil. Note the many geometrical parameters that have been taken
into account.
The formula is presumably no longer valid if a field shaper is used. It is not known if the formula is
valid in the case of a single turn coil (only 1 broad coil winding and the workpiece length exceeds coil
length). If it is valid, it could be directly used in the analytical model for the single-turn coil where no
field shaper is present.
42
Figure 3.14: Analytical equation for the conversion factor between current and magnetic field in between coil and
workpiece [37].
3.4.2 Magnetic pressure
The damped oscillating current through the coil (i1 in Figure 3.15) generates an axial transient
magnetic field inside the coil. Often a field shaper is used to concentrate the magnetic field in the
welding zone. The field shaper is built with a radial slit. The induced current flows on the surface.
According to Lenz’s law, eddy currents will be induced in the field shaper (i2). This current flows on
the inner surface (near the workpiece) because of the radial slit in the field shaper, as shown in
Figure 3.15. A current is also induced in the tube (i3). The magnetic pressure will be discussed for the
configuration with only a forming coil, but no field shaper. The field shaper will be discussed in the
next section.
Figure 3.15: Coil(1), field shaper(2) and tubular workpiece(3) [4]
43
The induced currents in the tube flow in the circumferential direction and opposite to the coil current
(Lenz’s Law). The magnetic field between the coil and the tube is directed axially, and thus
perpendicular to the tube current. An electromagnetic Lorentz force acts on the flyer tube, which is
directed radially inwards (direction of the vectorial product of the tube current and the magnetic
field). As a consequence, the tube is accelerated away from the coil and collides with the inner tube.
The Lorentz force is useful for a single wire, but not for the analytical modelling of the process.
Therefore, the concept of magnetic pressure is applied, because the pressure acting on the tube is a
consequence of the transient magnetic field.
In the Pulsar model, the following equation is proposed for the magnetic pressure exerted on the
flyer tube:
dVTkJ =
With: Pmagn = magnetic pressure [Pa]
μ=magnetic permeability [H/m]
B= magnetic flux density [T]
1 $
2
(3.35)
This equation is derived from the energy density associated with the magnetic field, taking into
account that B=μ0H:
With: B = magnetic flux density [T]
H = magnetic field intensity [A/m]1
dVTkJ =
1
. ƒ
2
(3.36)
Equation (3.35) assumes that the magnetic field between the flyer tube and the inner workpiece is
negligibly small. In reality, this is not always accurate. Taking into account that the magnitude of the
magnetic field inside the flyer tube is of significant value, a more correct equation is proposed in [38].
It should be noted that in these equations the magnetic field is time-dependent, as it is caused by a
time-dependant current.
dVTkJ =
1
1
(‰$ − l$ ) = (ƒ‰$ − ƒl$ )
2
2
(3.37)
With: Pmagn = magnetic pressure [Pa]
μ =magnetic permeability [H/m]
Bu = magnetic flux density between flyer tube and coil (or field shaper, if present) [T]
Bi = magnetic flux density between inner tube and flyer tube [T]
Hu = magnetic field intensity between flyer tube and coil (or field shaper, if present) [A/m]
Hi = magnetic field intensity between inner tube and flyer tube [A/m]
In order for the previous equation to be useful, an analytical expression must be obtained for the
magnetic field between inner workpiece and flyer tube. The magnetic field between the coil and the
1
The term ‘magnetic field’ is used for two different vector fields, denoted B and H. There are many alternative
names for both. To avoid confusion in the formulas in this section, B will be denoted as magnetic flux density
and H as magnetic field intensity. The term magnetic field is used in the text for the B-field [79] [80] [81]
44
flyer tube differs from that between the flyer tube and the inner workpiece, due to the shielding
effect [34].
:S
With: t = thickness of flyer tube [m]
δ = skin depth [m]
l = ‰ . C Ÿ
(3.38)
The skin depth of a material is defined as the depth below the surface of the material at which the
current density decays to 1/e of the current density at the surface, as discussed in Chapter 2. It can
be calculated as:
=
1
(3.39)
With: δ = skin depth [m]
f = current frequency [Hz]
μ = magnetic permeability [H/m]
κ = electrical conductivity of flyer tube material [S]
Quantification of the skin depth δ simplifies the expression for the magnetic pressure [34]:
dVTkJ =
:$S
1 $
‰ (1 − C Ÿ )
2
(3.40)
With: Pmagn = magnetic pressure [Pa]
Bu = magnetic flux density between flyer tube and coil (or field shaper, if present) [T]
t = thickness of flyer tube [m]
δ = skin depth [m]
It is clear that the magnetic pressure exerted on the tube is proportional to the square of the
magnetic field in the air gap (between tube and coil/field shaper). Because the discharge current is a
sinusoidal damped wave, so is the magnetic field. The pressure wave is shown in Figure 3.16,
together with the current waveform.
Figure 3.16: Discharge current and magnetic pressure [9].
45
The pressure wave has a frequency which is double of the current frequency, and is positive at all
times. This can be explained by the fact that when the current changes direction, both the magnetic
field and the induced current in the tube change direction. The Lorentz force consequently remains
directed inwards.
It should be noted that there is a small decrease in magnitude and frequency of the magnetic field
due to the deformation of the workpiece (§ 3.3.2.3).
3.4.3 Field shaper
The purpose of a field shaper is to concentrate the magnetic field in the welding zone. It essentially
increases the amplitude of the magnetic field, in a smaller region (axially). In most cases it will be
used with multi-turn coils, for focusing the widely-spread current from many windings onto a small
work-zone. In single-turn coils, there is usually no need for a field shaper, as the current flows
directly to the work zone of the coil. The coil shape can be directly modified in this case. The field
shaper material must have a combination of sufficient electrical conductivity and mechanical
endurance against pulse loads and thermal shock. Tantalum and Copper-Beryllium are widely used.
Its lower cost (despite the slightly lower service life) makes Cu-Be the most economical choice[39].
The principle of the field shaper was briefly discussed in the previous section (Figure 3.15). The field
shaper is located inside the multi-turn coil, as shown in Figure 3.17.
Figure 3.17: The field shaper is located inside the multi-turn coil. The magnetic flux is concentrated, resulting in a larger
B-field in the inside of the field shaper (at the welding zone) [16].
The discharge current flowing by the coil windings induces a current in the field shaper, which is
located inside the coil. The conducting coil generates a magnetic field. The axial length of the field
shaper on the inside is much smaller that at the outside. As a result, the magnetic flux, generated by
the coil, flows through a smaller cross-section. The field shaper thus concentrates the field to a
narrow axial zone, thus increasing the field magnitude at the welding zone.
Figure 3.18 shows a comparison of the magnetic field with and without field shaper. The curve with
dots indicates the field inside a coil with length 40 mm and the curve with squares indicates the field
inside a field shaper with length 15mm, which is placed in the center of the coil.
46
Figure 3.18: The magnetic field at the welding zone is increased in magnitude by the field shaper[27]
3.4.3.1 Analytical equations
The main question for the analytical model is by how much does the magnetic field increase because
of the field shaper?
In [27] an expression is determined by FE analysis to express the correlation between the current and
the magnetic field it induces:
= ,o (9W22 )
(3.41)
With: I = current amplitude [A]
B = magnetic flux density inside the coil [T]
K = constant depending on the geometrical, material and electrical characteristics
f (Aef f ) =enhancement factor [H/m2]
The factor K depends on geometrical, material and circuit parameters. The enhancement factor is
shown in Figure 3.19 as a function of the ratio of the length of the field shaper nodule (= length of
the working zone) to the length of the forming coil. The enhancement factor is almost independent
of the inner radius of the field shaper. Because no further information is given for the calculation of
K, and it is not clear from the article which current is denoted as I, this formula is not applicable in
practice [27].
Pulsar suggests the following formula to calculate the field shaper efficiency (I), based on its
geometry (Figure 3.20) [39]:
I=
With: B=axial length of field shaper[m]
a= work-zone radius[mm]
b=work-zone width[mm]
α=angle of focusing[˚]
[. sin (¡)
‚+
1 − cos (¡)
(3.42)
47
Figure 3.19: Enhancement factor of the magnetic field [27]
Figure 3.20: Geometrical parameters for the calculation of the field shaper efficiency [39].
The article states that greater efficiency results in a stronger magnetic field in the work-zone, for any
given energy. However it is not clear which efficiency is defined by this equation. The maximum
efficiency is reached for α= 45˚. For α<45˚there will be more current losses on the conical sides, and
for α>45˚, a higher impedance reduces the current. [39] But the article states that for best magnetic
focusing, α=90˚ is preferred, as shown on the left of Figure 3.21. The only problem with this kind of
shape is its mechanical strength. Due to shock stresses, the field shaper will not last for many cycles.
Cracks will form in the sharp corners and the work-zone edges will flatten. As a trade-off, the angle is
chosen between 45˚ and 90˚, shown on the right of Figure 3.21.
However, for the two types of field shaper examples given in this document, the magnetic field level
only changes by about 4% [39]. Again, despite several attempts, no clarification was received by
Pulsar.
48
°
°
Figure 3.21: Angle of focusing; left α =90 - right 45º<α<90 [39]
3.4.3.2 Radial slit
Field shapers are constructed with a radial slit for functionality. The slit allows the current to flow as
shown in Figure 3.15. The complication is that the magnetic field will be significantly reduced at the
slit region.
The effect of the slit was reported in several articles based on results of FE calculations. The magnetic
field strength around the periphery of the gap between the field shaper and the workpiece is not
uniform, as can be seen in Figure 3.22. The field strength is lower around the slit region. On the right
of Figure 3.22, the Lorentz force distribution over the workpiece circumference is shown. The Lorentz
force and consequently the magnetic pressure exerted on the tube are reduced at the slit region
[40].
At OCAS, an FE model is implemented for the equipment at BWI. Figure 3.23 shows the current
distribution as a function of the axial position along the inner surface of the field shaper: the higher
curve far away from the radial slit, the lower curve at the radial slit region. The axial length of the
inside surface of the field shaper is 15mm. The current density, and consequently the magnetic field,
is lower at the slit region. It is also observed that the higher current density is higher at the sides of
the surface [16].
This detrimental effect of the radial slit might result in a deviation of the deformation of the flyer
tube. In the experiments performed in this thesis, possible phenomena associated with the decrease
in magnetic pressure at the slit region were investigated. The observations are discussed in detail in
Chapter 6. The weld defects were consistently found at the crack positions of the damaged field
shaper, but very few were observed at the slit region. If weld interruptions occurred, they were very
small (< 2 mm). In addition, roundness measurements did not confirm the buckling effect, suggested
in [17].
So, it is probably not necessary to invest much more research on the phenomena associated with the
slit. In addition, it is too difficult to incorporate the effects of the slit region into the analytical model.
49
Figure 3.22: Distribution of magnetic field strength along the periphery in the gap between the field shaper and the
workpiece (left), and vectorial depiction of the Lorentz force distribution over the workpiece circumference (right) [40].
Figure 3.23: Current distribution as a function of the axial position along the inner surface of the field shaper. The higher
curve is calculated far away from the radial slit, the lower curve at the radial slit region. The current density, and
consequently the magnetic field, is significantly lower at the slit region [16].
50
3.4.3.3 Conclusion
Similar to the custom coils, no general analytical equations were found for field shapers (which can
be applied to any geometry). However, finite element simulations offer an alternative.
The linear relation between the B-field and the current exists in the case of coil and workpiece
(equation 3.34). The linearity also exists in the case of coil, field shaper and workpiece. Because only
the amplitude is affected by the field shaper presence, only the conversion factor will be different:
(;) = (C_C;#š ›‘, ‘C ]ℎ[ZC# & #žZ‘C›C). o(;)
(3.43)
The magnetic field in the gap between coil and workpiece is still proportional to the magnetic field,
and the conversion factor is a constant, based on geometrical parameters.
The model introduced into any FE simulation is of course based on the specific geometry of the coil,
field shaper and workpiece. So, FE simulations are specific for the equipment used at each research
facility. At OCAS, a model is developed for the MPW equipment at the Belgian Welding Institute. The
model includes the exact coil, field shaper and workpiece geometry. In their report the effectiveness
of the field shaper is determined, expressed as the ratio of the current in the coil over the current on
the inner surface of the field shaper. For the Pulsar coil and field shaper geometry used at BWI, this
ratio was determined to be 1,67. This is much less than the ideal factor 5 one would expect, based on
the ratio of 5 windings of the coil over 1 “winding” of the field shaper [16]. Besides the ratio of the
field shaper current to coil current, the calculation of the ratio of the magnetic field (between
workpiece and field shaper) to the primary coil current would be useful to continue the analytical
model. It has been requested to perform computations to obtain a value for the conversion factor,
which characterizes the relation between magnetic B-field and current.
3.4.4 Magnetic field measurement
In the course of this thesis a probe was developed to measure the magnetic field waveform between
the workpiece and the field shaper. Combining the magnetic field measurement with the discharge
current measurement, it was attempted to determine the relation between B-field and current
experimentally.
Some problems were encountered during the measurements. First, the calibration of the probe area
was difficult. In addition, at high voltages, consecutive measurements at the same voltage level
showed quite different signals. At low voltage levels, the measurement of the magnetic field as a
function of time resulted in sinusoidal damped waveforms with the same frequency as the current.
Also, the linearity between the current and the magnetic field was only found consistently at voltage
levels lower than 12kV. The magnitude of the magnetic field differed slightly from that calculated
with finite element analysis. The measurements are discussed in detail in Chapter 4.
The difficult calibration of the probe area is related to the high field strength in the MPW process, as
the same problems were reported in literature. The irregularities found at higher voltage levels are
possibly caused by the field shaper damage. Therefore, it is recommended to perform the
measurements again with an undamaged field shaper.
With the measurements, the development of the analytical model could be continued. In addition,
they could be used as a tool to improve the accuracy of finite element simulation.
51
3.5 Deformation, acceleration and impact velocity
3.5.1 Deformation pressure
In the model proposed by Pulsar, the acceleration is assumed to be constant. This assumption is not
realistic, but has the important advantage that calculations can be simplified significantly. The
magnetic pressure exerted on the tube both deforms and accelerates the tube radially inwards. With
a constant acceleration, the pressure required for acceleration can also be assumed constant in time.
The magnetic pressure can be calculated as a sum of two terms: acceleration pressure and
deformation pressure. So, if an expression is found for the pressure required to plastically deform
the tube and the magnetic pressure is calculated from the predicted discharge current, the
acceleration pressure can be calculated.
The equation suggested by Pulsar to calculate the magnitude of the required radial pressure to
deform the workpiece (equation 3.4) is based on the linear-elastic theory of thin-walled vessels.
Another equation was found in [40]:
deW2x4V = [.
With:
ž = 0,5. exp (
[ ∈ w1,10t
−0,5. y
)
¥xy
2
.
2−ž
¥
$
Q xy − 1X 3. (1 − ž + ž )
ℎy
. Y
(3.44)
(3.45)
σy = yield stress of tube material [N/mm2]
Dow= tube outer radius[mm]
hw= tube thickness[mm]
lw = tube length[mm]
Equation 3.44 is only valid for materials that do not possess strain-rate hardening. It is said that the
factor a takes into account inertia effects, strain-hardening, through thickness stresses, anisotropy,
etc. However, no analytical equation is given for a, so the formula cannot simply be applied.
In reality, the tube deformation is mostly plastic and it occurs at high strain-rates. The pressure
required to deform is difficult to calculate analytically, because simple linear-elastic models are not
valid. In addition, only a part of the tube length (the overlap length) is subjected to the radial
pressure, as illustrated in Figure 3.24. As a consequence, the tube will not deform uniformly over its
length, as shown in Figure 3.25. The tube impacts the inner workpiece at a certain angle.
52
Figure 3.24: In magnetic pulse welding experiments, the magnetic acts only on the part of the tube that overlaps with the
field shaper [33].
Figure 3.25: Cross-section of copper-aluminium tubular weld
Because of the complicated nature of the deformation process, finite element analysis is often used
to predict the tubes deformation behaviour, as shown in Figure 3.26 [41]. The FE techniques take
high strain rates into account by applying the Johnson-Cook model (Chapter 2). [42]
Figure 3.26: FE analysis is used as a tool to model the deformation behaviour of the tube [41]
53
3.5.2 Time-dependant acceleration and impact velocity
The complex deformation behaviour adds difficulty in establishing an analytical expression for the
impact velocity, because deformation and acceleration occur simultaneously.
If the deformation pressure is neglected, equations can be formulated that describe the motion of
the flyer tube, when subjected to a radial magnetic pressure. The calculations that are presented
were developed in this thesis. Similar calculations were found in [28], for MPW welding of flat sheets.
The discharge current is a sinusoidally damped wave, as shown in Figure 3.27:
o(;)~ exp(−’;) . sin (“Œ ;)
(3.46)
(;)~ exp(−’;) . sin (“Œ ;)
(3.47)
Because the magnetic field in the gap between the tube and the field shaper is proportional to the
discharge current:
The conversion factor between the magnitude of B and I is still to be determined, as discussed in the
previous paragraphs.
Figure 3.27: Current waveform [28]
The magnetic pressure is proportional to the square of the magnetic field, so:
Z(;)~ exp(−2’;) . sin$ (“Œ ;)
(3.48)
The magnetic pressure is a wave with twice the frequency of the current, as shown in Figure 3.28.
54
Figure 3.28: Pressure waveform [28]
If it is assumed that the pressure results only in acceleration (deformation is neglected), then
according to Newton’s second law, the acceleration is:
And by considering equation (3.48):
[(;) =
(;) 9
= . Z(;)
_
_
[(;)~ exp(−2’;) . sin$ (“Œ ;)
(3.49)
(3.50)
The time dependency of the acceleration is too important to neglect. Certainly if we consider that it
is not known in advance after which time interval the flyer tube impacts the inner tube. Of course the
value of the stand-off distance is known, as it is a parameter that can be chosen for each experiment.
Integrating the acceleration yields an equation for the velocity as a function of time:
S
(;) = « [(;) ;
(3.51)
The velocity as a function of time is shown in Figure 3.29: the velocity profile is theoretical because
deformation is not taken into account.
Subsequent integration of the velocity as a function of time leads to an equation for the radial
distance travelled by the flyer tube as a function of time.
S
S
](;) = ¬ [(;) ; = « (;) ;
(3.52)
55
Figure 3.29: Velocity as a function of time [28]
An example of the calculated displacement as a function of time is shown in Figure 3.30. This graph
was calculated in Maple, using the equations in this paragraph.
Figure 3.30: Displacement as a function of time
Impact with the inner tube occurs after the flyer tube has travelled over a distance equal to the
stand-off distance. By expressing that the distance at impact equals the stand-off distance, a value
for the time interval can be obtained. This value is then substituted in the velocity function to obtain
a value for the impact velocity.
It is assumed that the sheet impacts within the first half period of the current wave, or equivalently
within the first pressure pulse [28] [43]. Although the magnetic pressure acting on the tube is
positive at all times, it decreases sharply at the end of the first pulse. When the magnetic pressure is
small (but positive), the tubes resistance against deformation causes the tube to decelerate [40]. This
is not taken into account in the calculations, nor in Figure 3.29, where in reality the velocity would
start decreasing after approximately 70 μs or 80 μs.
56
Figure 3.31 shows a measurement of both primary discharge current and flyer plate velocity profile
in a sheet welding experiment[23]. The flyer plate impacts at about 15 μs, which is slightly after the
first current peak.
Figure 3.31: Measured primary discharge current and flyer plate velocity in sheet welding (Cu-Al joint)[23].
By comparing the measured data from Figure 3.31 to Figure 3.29, it would suggest that impact would
occur at approximately 80 μs in Figure 3.29. The calculated waveform of Figure 3.29 does not take
deformation pressure into account, which would cause the velocity to start decreasing after 70 μs, as
previously explained. Up to that moment in time, the shape of the measured velocity profile is in
agreement with the calculated profile.
The following graphs are simulation results of magnetic pulse tube compression by OCAS [16]. Figure
3.32 shows the radial velocity of the tube (left) and the radial displacement of the tube (right), during
‘free’ compression. Figure 3.33 shows the radial velocity of the tube (left) and the radial
displacement of the tube (right), during compression against an inner workpiece. Both simulations
were calculated in the axial center of the pressurised zone.
During free compression the velocity decreases after reaching its maximum value (Figure 3.32). In the
compression against an inner workpiece, impact occurs at 22 μs. The velocity has not yet started
decreasing when the tube impacts the inner workpiece.
The shape of the calculated velocity (Figure 3.29) and displacement (Figure 3.30) curves is similar to
the simulated curves, but only during acceleration. The calculations do not take the tubes resistance
against deformation into account. After the first pressure pulse, when the pressure becomes very
small, the calculated velocity increases only slightly in the calculations. But in the simulations, it
starts decreasing
It should be noted that during the period where the velocity increases, the shape of both velocity and
displacement curves are in agreement, but that they are both overestimated in magnitude by the
calculation. The magnetic pressure is large enough not to cause deceleration, but the pressure to
57
deform the tube is still required, and not taken into account in the calculation. So, the acceleration is
overestimated, and consequently also velocity and displacement.
Figure 3.32: Radial velocity of the tube (left) and the radial displacement of the tube (right) during ‘free’ compression
Figure 3.33: Radial velocity of the tube (left) and the radial displacement of the tube (right) during compression against
inner workpiece
If the stand-off distance is too small, the workpiece might not yet have reached sufficient velocity at
impact. On the other hand, if the stand-off distance is too large, the workpiece might have already
started decelerating when the impact occurs. It is important to realise that the occurrence of the
deceleration for a given stand-off distance is dependent on the frequency of the current, and of its
amplitude (which affects the pressure and acceleration). The frequency is primarily determined by
the capacitance of the capacitor bank and by the transformer (if used). The amplitude is primarily
determined by the voltage level.
In [43], FE modelling is applied to investigate the frequency as a ‘tool’ to optimize the
electromagnetic forming process. The frequency can be changed easily by changing the capacitance
of the capacitor bank. Figure 3.34 shows the effect of frequency on the radial displacement profile in
tubular crimping.
Again, it can be seen in all curves that the velocity starts decreasing near the end of the displacement
(v=dsr/dt). Similar to the previous comparison, it can be seen that the calculated deformation profile
58
(Figure 3.30) has the same shape as the simulated profile (Figure 3.34) during a certain period, after
which the velocity starts decreasing.
Figure 3.34: Effect of frequency on radial displacement radial profile [43]
3.5.3 Conclusion
The equations shown in this paragraph incorporate the time-dependency of the acceleration. A
shortcoming of the proposed equations is that deformation is not taken into account. The tube is
also assumed to accelerate uniformly over its length.
If the complex deformation behaviour is not taken into account, an analytical expression for the
acceleration as a function of time could be established. Integration then leads to a time function of
radial velocity and displacement. As the stand-off distance is set in advance, a value for the time
interval can be obtained and used to estimate the impact velocity.
If the stand-off distance is too small, the workpiece might not yet have reached sufficient velocity at
impact. On the other hand, if the stand-off distance is too large, the workpiece might have already
started decelerating when the impact occurs.
Further research is needed to integrate the effect of deformation in the analytical model.
59
3.6 Conclusion
Further research is needed to continue the development an analytical model, which takes the timedependency of the process into account. The RLC-circuit analysis can be applied to generate
estimates for the discharge current. The variation of the parameters L and R can be investigated by
analysing the current measurements.
On-line measurements such as time-measurement and deformation measurement (Chapter 4) could
provide valuable information. Finite element simulations, customised to the geometry of the
experiments, can also be a valuable source of information. Simulation results found in literature are
often not valid in general, and cannot be applied to develop an analytical model. In the currently
applied workpiece geometry, FE simulations can be used to determine the relation between the
magnetic field inside the air gap (between field shaper and flyer tube) and the discharge current.
Also, estimates of the angle at which the tube impacts the inner rod can be generated. The
calculations are currently being performed at OCAS.
It is suggested to first perform simulations for simplified workpiece geometries, such as a tube, which
is accelerated uniformly over its length (pressure acts on entire length). The tube does not even
necessarily have to impact an inner rod. With this geometry, the analytical calculations are more
accurate (no impact angle). If the simulation can provide functions of acceleration and/or velocity
and/or displacement versus time, these results can be compared to the calculations using the
analytical model as proposed in this section. Perhaps a correction factor could be found to correct
the acceleration (and consequently velocity and distance functions) for the deformation pressure. So,
first the analytical model should be adjusted based on the suggested simple workpiece geometries.
Afterwards, it can be further developed to incorporate the deformation of a tube, pressurised only
on the overlap length.
60
Chapter 4
Process Measurements
4.1 Introduction
Several options are possible to gain new information about the MPW process. The first method is
studying results from other research found in literature. Another approach is to perform experiments
in the attempt to draw conclusions based on the results. The workpieces can be analysed after the
MPW welding process is completed and can comprise a lot of information of what occurred during
the process. This information is limited however. Many aspects of the MPW process – the magnitude
of the magnetic field, the deformation behaviour of the tube and its velocity profile – cannot be
determined from the analysis of the workpieces. The only way to gather more information on these
parameters (which have an important influence on the weld) is to perform process measurements.
Measurements during the course of the process are difficult for two reasons. The first major
limitation is the duration of the welding process. The entire process, from the start of the current
discharge to the final impact of the flyer tube on the internal workpiece, takes a few ten
microseconds. No exact value of the total duration is known, but the frequency of the discharge
current is 14 kHz (period=71,4 μs). The impact is assumed to occur within the first half period of the
current [44]. The second problem with on-line measurements is the fact that the access to the inside
of the coil, where the welding process occurs, is extremely difficult. The isolation and clamping
devices of the tube and rod, hinder the use of measuring devices. In addition, there is almost no
empty space inside the coil.
Very few on-line measurements for the MPW process were found in literature. The measurement
methods discussed in literature include:
•
•
•
•
•
High speed camera.
Process duration measurement, using an electrical circuit.
Tube deformation measurement, using laser beams.
Photon doppler velocimetry, using laser beams.
Magnetic field measurement, using fibre optic sensors.
The second part of the chapter describes the two on-line measurements used in this thesis:
•
•
Discharge current measurement, using a Rogowski coil and integrator.
Magnetic field measurement, using a custom developed probe.
61
4.2 Literature Survey
4.2.1 High speed camera
The development of ultra-high speed cameras opens new possibilities. Cameras were found with a
speed up to 1 million frames per second. This means that an image can be taken every microsecond.
Even if the total process duration is only 40 μs, this means that 40 photographs can be taken during
the process. This would allow investigating the deformation behaviour of the flyer tube, as well as
determining the exact duration of acceleration (from standstill to impact).
The only problem with the use of a camera is that the inside of the coil is not within the line of sight.
In applications of the MPW process for sheet joining, a flat spiral coil is located at only one side of the
sheets, and photographs can be taken. In tube welding, the inside of the coil is not accessible for
photography.
4.2.2 Process duration measurement
A method to measure the collision speed in sheet welding experiments is described in [45]. In reality,
the setup measures the duration of the process, rather than the collision speed. Using the measured
time interval and assuming a uniform acceleration, the collision speed is calculated.
The measurement circuit is shown in Figure 4.1. The voltage difference over the two sheets is
measured using a digital oscilloscope. The cables connecting the sheets and the oscilloscope are
coaxial cables, to prevent the magnetic field from disturbing the measurement.
Figure 4.1: Circuit used to measure the time that the flyer plate accelerates during the MPW process for flat sheets. [45]
When the impulse discharge current passes through the coil, a voltage is induced on the two work
pieces by magnetic coupling between the coil and these work pieces. The magnitude of the voltage
difference between the sheets is of no importance. When the two sheets are in contact (after
impact) the circuit is shorted, and there will be no voltage difference. The oscilloscope will generate a
graph of voltage versus time, and the duration that a voltage was measured is equal to the process
(and acceleration) duration. Assuming that the sheet movement is a uniform acceleration motion,
the collision speed just before impacting can be estimated by using the time travelling and stand-off
distance.
The time that the flyer tube travels before impacting the inner rod, can give valuable information.
Especially in combination with the current measurement, the point in time (during the discharge)
when impact occurs could be determined. This information could then be used to estimate the
impact velocity more accurately, based on theoretical calculations that take into account the timedependant pressure. Then, optimal values for the stand-off distances could be determined.
62
This measurement uses a very simple circuit and it is realistic to implement in the MPW machine for
tube welding.
4.2.3 Tube deformation measurement
The inner and outer contours of the flyer tube can be measured after the forming operation using a
microscope. The obtained geometry at the end of the deformation is not sufficient to allow accurate
assumptions about the behaviour during the forming process. For this purpose, a setup was designed
to measure the radial displacement of the flyer tube during compression in [44].
The on-line measurement system for tube compression is based on an optical principle. The setup is
shown in Figure 4.2.
A light source is located at one side of the workpiece in the axial direction. The amount of parallel
light shining through the sample is detected by a PSD (position sensitive detector) and depends on
the actual inner radius of the sample. This radius decreases during the deformation process. The
output voltage of the PSD is proportional to the displacement of point A, in which the maximum
deformation occurs. Because the forming process takes only a few ten microseconds, a very high
time resolution was realized using a laser diode with line generator. The measurement has an
accuracy of approximately 0,1 mm.
It is important to note that the tubes in the experiments in [44] are compressed, but not with the
intention of welding the tube to an inner workpiece. So, there is only one workpiece (the tube) and
free space on the inside to allow the light from the source to reach the detector. The pressure is
exerted in the middle of the tube (in the axial direction), causing it to deform symmetrically.
Therefore, the deformation of one axial cross section can be measured.
The same method was used in [38] to estimate the impact velocity.
The position of the laser is not discussed in the measurement setup. In the MPW machine used in
this thesis, clamping devices on both sides of the workpieces block the light from the source. In
addition, the collar on the inner workpiece prevents the light to reach the detector.
In conclusion, this method is not suitable to measure the tube deformation in asymmetrical welding
experiments. It can be used for crimping experiments, but the clamping devices should be
redesigned to for the laser light to pass.
4.2.4 Photon doppler velocimetry
Photon Doppler Velocimetry (PDV) was used as a method to measure the impact velocity, as
explained in [46] and [47].
The basic physics of the PDV is illustrated in Figure 4.3. A moving surface produces Doppler shifted
light, which is then recombined with the incident light signal to produce a ‘beat frequency’. This beat
frequency is proportional to the velocity of the moving surface, and can be analysed with digital
equipment to create velocity versus time profiles. Though the principles are simple, the actual
measurements are non-trivial and require very modern electronics.
63
The key components of the PDV system include:
•
•
•
•
•
•
Laser: High power fiber laser with a very narrow spectral line width.
Splitters: Divide laser output to several fiber optic ports for multi channel operations.
Circulators: Directional fiber optic device guides light from the laser out to the probe and
reflected light from the probe to the detector.
Detectors: Short rise time biased photodetector, with high bandwidth.
Probes: Collimating or focusing with built in reference partial reflection surface.
Oscilloscope: This allows data recording (1 GHz frequency) for periods up to 2 ms on multiple
channels.
Figure 4.2: Setup for the measurement of the tube deformation during the magnetic pulse crimping process. The amount
of parallel light shining through the sample is detected and depends on the actual inner radius of the sample. The output
voltage of the detector is proportional to the displacement of point A, in which the maximum deformation occurs. [44]
Figure 4.3: Schematic diagram of a Photon Doppler Velocimetry system. The arrows indicate the direction of the incident
and reflection beam. [46]
64
In regards to data analysis, a Fourier Transform can be performed to analyse the changes in beat
frequency in order to generate velocity versus time profiles.
The PDV technique was used to determine the flyer velocity profile in impact seam welding (welding
of sheets under a certain angle) and in electromagnetic ring expansion [46]
The electromagnetic ring expansion process is briefly discussed as an example. The measurement
setup is shown schematically in Figure 4.4. Electromagnetic ring expansion is very similar to the tube
compression process. The capacitor bank discharges and the current is conducted by a coil, which is
located inside the ring (instead of outside the tube in the compression process). The magnetic
pressure causes the ring to expand radially outwards to a larger diameter.
The measurement instrumentation includes the PDV system and two Rogowski coils. R1 measures the
primary discharge current. A second Rogowski coil (R2) with fine wire loops is placed directly around
the sample ring to measure the induced current. The PDV system measures the ring position with
temporal resolution on the order of nanoseconds and spatial resolution on the order of microns. This
data can be singly or doubly differentiated with time to generate the velocity – or acceleration profile
of the ring. The PDV method is easy to implement in the ring expansion process because only a thin
fiber optic line (with an inexpensive probe on the end) needs to run between the instrumentation
and the target.
The measurement results are shown in Figure 4.5. The first graph shows the raw signal from the light
detector. The primary current (measured by R1) and the induced current (measured by R2) as a
function of time are shown in the second and third graph.
The radial ring velocity, shown in the fourth graph, is measured based on the period of each optical
oscillation. This plot contains thousands of independent measurements of ring velocity, so
acceleration values could also be estimated accurately.
Figure 4.4: PDV can be used to measure the radial velocity of a ring in the magnetic pulse expansion process [46]
65
Figure 4.5: From top to bottom – raw PDV signal, discharge current, induced current, ring velocity profile [46]
The PDV measurement method is not suitable for tube compression, because the coil blocks the path
of the the laser beam. From all the measurements found in literature, it seems that the tube
compression is the most difficult variant of the magnetic pulse process to perform on-line
measurements for.
4.2.5 Magnetic field measurement
The magnetic field strength between a workpiece and the coil of an electromagnetic high speed
metal forming device is of special interest as it is the source for the resulting forces during the
forming process. The small gap and the high frequency and magnitude of the field make the
measurement of the field extremely complicated.
A high-tech method for measuring the field strength in the magnetic pulse sheet welding process was
found in [48]. The magnetic field between the coil windings and the accelerated workpiece was
measured, as shown in Figure 4.6. Small potential free optical sensors were used because of the
small gap and the high field strength.
Figure 4.6: The magnetic field between the coil windings and the accelerated workpiece was measured. Small potential
free optical sensors were used because of the small gap and the high field strength. [48]
66
Fibre optic current sensors use the magneto-optic Faraday effect. An applied longitudinal magnetic
field induces circular birefringence in the optical material. Therefore, a linear polarised light wave,
propagating through the material, will receive a change in its polarisation angle.
Besides the standard application of current measurements the Faraday effect can also be used for
direct measurements of magnetic fields. Regarding the electromagnetic and dimensional conditions
at a typical setup, only non-conductive miniature field probes are applicable. The problem with
conventional optical point sensors in the special case of field determination inside the small gap of an
electromagnetic high-speed forming device is their saturation field strength (about 20 Tesla have to
be measured) or their dimensions. Nevertheless, the functional principle stays the same, similar to
that of the current sensor. Linear polarised light is entering a sensitive fibre or crystal with a defined
angle and is subject to circular birefringence. After passing through the optic material, the rotation of
the polarisation angle is converted into a variation of the light's intensity.
One sensor consists of two crystal blocks with an edge length of 500 µm each. Both blocks have a
polarising surface glued to a multimode fibre. They are placed in the gap between the coil and a fixed
workpiece.
The recorded signals of the field measurement using optic fibres are shown in Figure 4.7, together
with the measured discharge current, conducted through the coil[48]. This transient current causes
the magnetic field in the gap. As discussed in the analytical model, the magnetic field is proportional
to the discharge current.
(;) ~ ‘(;)
(4.1)
This linearity is roughly recognised in the measurements.
Figure 4.7: Measurements of the magnetic field in the gap between a fixed workpiece and the coil of a magnetic pulse
sheet welding device, using miniature fibre-optic magnetic field sensors. [48]
67
Although, the principle functionality of the measurement is proven, some limitations are mentioned
in [48]:
•
The sensor could not be calibrated before, so only the measured voltages from the
evaluation unit were given in Figure 4.7. Accurate calibration of the sensor was not possible
due to the high field strength in the MPW process.
•
Consecutive measurements show different signals. The following reasons were mentioned:
The field configuration might be very sensitive to small changes or mechanical coupling may
disturb the sensor.
The use of these fibre-optic sensors is too complicated to be applied in the magnetic pulse tube
welding process. The valuable information that can be generated based on a magnetic field
measurement was an inspiration to develop a custom magnetic field measurement probe. The field
measurement performed in this thesis is discussed further in this chapter.
68
4.3 On-line Measurements
4.3.1 Discharge current measurement
In [49], a test setup was constructed to measure the discharge current waveform that flows through
the coil (in which the workpieces are located).
The measurement setup consists of a Rogowski coil, an integrator and a digital USB oscilloscope. The
helical Rogowski coil is placed around the cables that conduct the discharge current from the
capacitor bank to the forming coil, as illustrated in Figure 4.8. The magnetic field produced by the
current in the conducting cables induces a voltage in the Rogowski coil. The induced voltage (E) is
proportional to the rate of change of the conductor current. This voltage is then integrated by a
passive integrator, thus producing an output voltage that is proportional to the current. [50]
Figure 4.8: Rogowski coil and integrator, used for the measurement of the discharge current. [50]
The output voltage is then used as an input for the digital oscilloscope, which allows visualising the
measurements on a computer. Using the software of the oscilloscope, a digital low-pass filter is
applied to the measured signal to reduce the noise. The cut-off frequency was set at 100 kHz, which
is significantly higher than the current frequency of 14 kHz. Figure 4.9 shows a screenshot from the
oscilloscope, with the raw signal (above) and the filtered signal (below). The filtered signal clearly
shows the damped sinusoidal current waveform.
The discharge current was measured in each experiment. This is required because based on the
current measurement a maximum allowable voltage level should be calculated. This is discussed in
the chapter on the experiments (Chapter 6).
69
Figure 4.9: Screenshots taken from the current measurement - Raw signal (above) and filtered signal (below).
70
4.3.2 Magnetic Field Measurement
4.3.2.1 Principle
When the capacitor bank is discharged, a high frequency current passes through the coil, generating
a magnetic field with the same frequency. This magnetic field is concentrated onto the flyer tube
with the help of a field shaper. The magnetic field lines will flow in the axial direction through the gap
between the field shaper and the flyer tube (Figure 4.10).
Figure 4.10: Magnetic field Lines flow in the gap between the isolation and the flyer tube
As explained in the analytical model (Chapter 3), it would be interesting to be able to measure the
magnitude of the magnetic field. However, narrow tolerances and the difficult access to the work
zone of the machine, prohibits the use of the measuring probes which are commercially available. A
custom probe had to be developed.
The developed probe consists of a tube, made of non-conductive plastic, which is placed around the
flyer tube and fills the gap between the flyer tube and the field shaper. The probe tube supports a
single turn of copper wire through which the magnetic field lines will flow. The measurement
principle is based on Lenz’s law: due to the changing magnetic field that flows through the turn of
copper wire, an electric current is induced. By measuring this current with an oscilloscope, the
magnitude of the magnetic field can be derived.
Due to the expected high magnitude of the magnetic field (about 20 T), a large voltage will be
induced in the copper wire. An excessive voltage can lead to several problems: the insulation on the
copper wire is not sufficient; the value of the generated current is too high for the oscilloscope… To
limit the voltage, the measuring turn is not wound around the entire circumference of the probe
tube but merely around a quarter of its section. A cross-section of the measurement device and
location are shown in Figure 4.11.
Although only a quarter of the probe’s section was wound, calculations had to be performed to
exclude the possibility of failure of the copper wire:
1. When charging the capacitor bank to a voltage of 16 kV, finite-element simulations show that
the maximum value of the magnetic field is Bmax= 24 T [51]. Since the test can also be
performed with voltages up to 19 kV, Bmax = 30 T will be chosen to include extreme
circumstances.
71
2. To simplify the calculations, the entire section of the probe was considered to be wound.
Hence the real final voltage should be divided by four. The dimensions of the probe are ri=21
mm and r0=24,5 mm.
Figure 4.11: Measurement of the magnetic field
The total section through which the flux will flow is:
9 = .( #
$
− #l $ ) = 500,3 __² = 500,3 . 10:+ _²
(4.2)
For a magnetic field of 20 T, the total flux through the section of the probe becomes:
Φ = . 9 = 30— . 500,3 . 10:+ _² = 15,01 _‚
(4.3)
With a frequency of 14 kHz the voltage induced in the copper wire is:
V = 2 . Φ = 2 . 14000 ƒ¯ . 15,01 _‚ = 1320 j
(4.4)
Using the entire section of the probe, a voltage of 1320 V would be induced. This is too high for the
insulation on the copper wire. By using only a quarter of the total section, the induced voltage
becomes (1/4). 1320 V = 330V, which is an acceptable value due to the short time of occurrence. The
quarter winding was applied in practice, as shown in Figure 4.11.
The magnetic field will exert a force on the copper wire when it is conducting the induced current as
described by the Lorentz force law. Since the copper wire is connected to the tube by electrical
insulating tape, it is necessary to check whether the tape is sufficient to keep it in its place. The
internal resistance of the oscilloscope is 1 MΩ and thus the maximal current in the measuring circuit
can be calculated as follows:
o=
j
330j
=
= 0,33_9
10+ Ω
(4.5)
72
The total length of the copper winding is approximated by 2 times a quarter of the circumference of
the average diameter of the measuring probe:
=
2#T±W4TkW
. 2 = . #T±W4TkW = 71 __
4
(4.6)
The force on the wire becomes:
= . o . = 7,03 . 10:³ …
(4.7)
The force is small enough be carried by the insulating tape.
4.3.2.2 Calibration
Once constructed, the probe was calibrated using a Helmholtz coil, a coil which generates a uniform
magnetic field. The coil is connected to a source which can send various values of excitation current
through the coil. The probe is placed inside the Helmholtz coil and the voltage that is induced in the
probe is measured for different values of the excitation current. A relationship can be found between
the excitation current and the induced voltage which leads to the area of the copper winding.
The formula to calculate the area was supplied by the manufacturer of the Helmholtz coil:
9=
j
2 . 0,560 . 10:" . o
(4.8)
The calibration of the probe is performed with a frequency of 50 Hz. Due to the linearity of the
process, this calibration will also be correct at the frequency of 14 KHz during the actual experiments.
An image of the calibration test rig is shown in Figure 4.12.
During the calibration some severe problems occurred:
1. The cables connecting the digital multimeters picked up a lot of magnetic stray fields leading
to erroneous measurements. To solve this, a coax-cable was connected to the copper
winding.
2. Due to the small area of the copper loop, the measured voltages were extremely small and
hence also very inaccurate. The excitation current through the coil is limited (to 25 A) by
constructional reasons, so increasing the output voltage by further increasing this current
could not be performed. Increasing the frequency of the excitation current is another
possibility to increase the output voltage but this could not be conducted neither. The large
impedance of the Helmholtz coil necessitates a very powerful source at high frequency and
this was not available in short notice.
The only solution was to perform a large number of measurements with the most accurate
voltmeter that was available, which was an analogue one.
73
Figure 4.12: The calibration test rig
Table 4.A shows the measurements of the induced voltage in function of the excitation current.
Figure 4.13 plots these values and adds the trend line that describes the relationship between the
two magnitudes.
The relation between the excitation current and the Induced voltage appears to be:
∆j = 6,5 . 10:+
j
. ∆o
9
(4.9)
Substituting this in the equation (4.8) that was delivered by the supplier, the area of the copper
winding is calculated:
9=
6,5 . 10:+
= 36,95 __²
2. 50 . 0,560 . 10:"
(4.10)
With the knowledge of this area the calibration is finished. The probe can now be used in the
machine to measure the magnetic field in the gap between the field shaper and the flyer tube.
0,18
0,16
0,14
0,12
y = 0,0065x
Voltage (mV)
0,10
0,08
0,06
0,04
0,02
0,00
0,0
5,0
10,0
15,0
20,0
Current ( A)
25,0
30,0
Figure 4.13: Excitation current versus induced voltage
74
Excitation
current (A)
Induced
voltage(mV)
9,0
0,04
9,6
0,05
10,0
0,05
11,0
0,05
11,8
0,06
12,0
0,06
12,4
0,07
13,0
0,07
13,8
0,07
14,2
0,08
15,0
0,08
15,4
0,09
16,2
0,09
16,6
0,10
17,4
0,10
18,0
0,11
18,6
0,12
19,0
0,13
19,6
0,14
20,0
0,15
20,4
0,15
21,0
0,15
21,4
0,15
21,8
0,15
22,4
0,16
23,0
0,16
23,6
0,17
24,0
0,17
24,4
0,17
Table 4.A:: Excitation current versus Induced Voltage.
4.3.2.3 Calculated Conversion Factor
To perform the measurements, the probe is placed over a steel bar. This bar replaces the flyer tube
and shields the magnetic field in such a manner that the entire field will flow through the probe. By
clamping the bar in the field shaper, the probe is kept in place.
The probe wires are connected to the same USB oscilloscope that is used for the current
measurements (Tiepie HS3). To decrease the voltage which is placed on the oscilloscope, a voltage
divider with a resistance of 10 MΩ is used. The internal resistance of the oscilloscope is 1 MΩ. The
oscilloscope also has an internal capacity of 20 pF in parallel with the internal resistance (Figure
4.14).
75
Figure 4.14: Vindusced vs Vmeasured
Since the capacitor is in parallel with the 1 MΩ resistance, the total impedance of the oscilloscope
drops significantly at a high frequency. The total impedance of the oscilloscope at 14 kHz becomes:
µxU\ =
With:
µŽ . µŒ
= 0,3624 ¶·
µŽ + µŒ
(4.11)
ZR =total internal impedance of the oscilloscope
ZR =internal resistance of the oscilloscope
ZC = internal impedance of the capacitor in the oscilloscope = (2π.14kHz.20pF)-1
The correlation factor between the measured voltage and the induced voltage can then be computed
from the voltage divider:
jxU\lGGxU\xFW
0,3624 ¶·
=
= 0,03497
jlJe‰\We
(0,3624 ¶· + 10 ¶·)
(4.12)
Combining equation (4.12) and the following formula for the magnetic field, the conversion factor for
the magnetic field can be calculated:
=
j
2 . 9
(4.13)
The conversion between the measured voltage, the induced voltage and the magnitude of the
magnetic field is shown in Table 4.B.
Vosc [V]
Vinduced [V]
B [T]
1
28,5929
4,4502
Table 4.B: Correlation factor between measured and induced voltage
76
4.3.2.4 Measurements
Using the oscilloscope software, a digital low-pass filter was applied to the signal to reduce noise.
The cut-off frequency was chosen at 100 kHz, much higher than the signal frequency of 14 kHz. The
discharge current and the magnetic field were simultaneously measured on different channels.
The tests were performed for a voltage of 1 kV to 20 kV in steps of 1 kV. The measurement of the
magnetic field as a function of time resulted in sinusoidal damped waveforms with the same
frequency as the current, but only at low voltage levels (Figure 4.15). Also, the linearity between the
current and the magnetic field was consistently found at voltage levels lower than 12kV.
A small phase shift was found between the current and the induced voltage.
Some complications were encountered with the conversion from the measured voltage to the
magnetic field magnitude. When the machine was set to a voltage level of 19 kV, a peak voltage of
3,36 V was measured. Using the conversion factor from the calibration, the magnetic field would only
be 3 T. Simulations suggest that at this voltage level, the magnetic field should be around 20-25 T
[51]. However, with the conversion factor using the theoretically calculated winding area, the
measured 3,36 V results in a magnetic field strength of 15 T.
The measurements indicated that something was inaccurate. These phenomena show that the
measuring coil works properly but that the calibration was not accurate enough. After consulting
EELAB [52], the department of Ghent University which is specialized in electromagnetic energy
conversion, the calibration was put aside and the area of the coil was determined geometrically to be
73,04 mm². This value appears to be more realistic than the value that was computed through
calibration.
Figure 4.15: Current (blue) and magnetic field (purple)
The results of the experiments are shown in Figure 4.16 and Table 4.C. The maximal measured
magnetic field strength increases linearly with an energy level up to a level of 12 kV. When the
energy level is increased further, some random phenomena seem to occur. It can be seen that the
time at which a maximum is reached decreases drastically when the energy level of 13 kV is reached.
77
It should be noted that at voltage levels higher than 12 kV, the measured waveform was different for
consecutive measurements at the same voltage level.
The measured magnetic field strength at 19 kV is 15 T. At voltage levels slightly higher or lower than
19 kV, the measured magnetic field strength is higher than at 19 kV. This could indicate an
irregularity of the 19 kV measurement. If the trend of the surrounding measurements is continued,
the field strength at 19 kV would be around 23 T. This result is in accordance with the simulation
result of 20 T to 25 T at this voltage level. The simulation results in [51] have not yet been verified by
experimental data, so their accuracy is uncertain. The measurement is definitely promising.
The irregular measurements at voltages higher than 12 kV suggested that some irregularities
occurred inside the machine or field shaper. After investigation, it became clear that in fact the field
shaper was severely damaged (Chapter 6). This could be an explanation for the fact that the
measured magnetic field was not always equal for different measurements at the same voltage level.
The cracks in the field shaper will grow during every pulse and so a certain energy level will produce a
different magnetic field every pulse. Additionally the current will not follow the same path during
every pulse and this will also influence the magnetic field measurements. Experiments with an
identical energy level should be carried out again when a new field shaper is available.
Voltage level [kV]
Vosc[V]
Vinduced[V]
B [T]
time of maximum [μs]
1
0,158
4,516
0,703
0
2
0,109
3,115
0,485
51
3
0,152
4,351
0,677
0
4
0,247
7,074
1,101
56
5
0,245
7,011
1,091
54
6
0,297
8,505
1,324
55
7
0,427
12,215
1,901
54
8
0,433
12,393
1,929
56
9
0,546
15,598
2,428
60
10
0,599
17,130
2,666
60
11
0,561
16,046
2,497
59
12
0,661
18,905
2,942
63
13
1,988
56,829
8,845
8
14
0,867
24,790
3,858
8
15
5,045
144,238
22,449
8
16
3,152
90,121
14,026
6
17
3,800
108,656
16,911
7
18
4,317
123,427
19,210
8
19
3,359
96,034
14,947
7
19,5
5,742
164,169
25,551
8
20
8,720
249,342
38,807
8
Table 4.C: Results of the B-measurements
78
B-measurement
45
40
35
B [T]
30
25
20
15
10
5
0
0
2
4
6
8
10
12
14
16
18
20
Voltage level [kV]
Figure 4.16: Maximum magnetic field (measured) versus voltage level
4.3.2.5 Conclusion
The measurement of the magnetic field as a function of time resulted in sinusoidal damped
waveforms with the same frequency as the current, but only at low voltage levels. Also, the linearity
between the current and the magnetic field was consistently found at voltage levels lower than 12
kV.
It should be noted that the problems that were encountered during the measurements, more
specific the difficult calibration and the fact that consecutive measurements at the same voltage
level showed different signals, were also reported in [48] during the magnetic field measurement
with optical fibres. Because the method of field measurement in [48] is completely different from
that used in this thesis, the difficulties of the field measurement are probably inherently related to
the extreme high magnetic field strength in combination with the high frequency in the MPW
process.
79
Chapter 5
Weld Quality Evaluation
5.1 Introduction
The tubular welds produced by the magnetic pulse welding process must be objectively
characterised. The applied methods for evaluating the welded workpieces are discussed in this
chapter. In this thesis both non-destructive and destructive were used.
The first section of this chapter describes non-destructive testing methods (NDT) which can be used
to examine magnetic pulse welded pieces. These tests are used to detect bonding flaws throughout
the weld and can thus provide an indication of the weld quality. The NDT tests have the main
advantage that they do not destroy the specimen. Even in a production environment, parts which are
meant to be sold afterwards can be examined by these testing methods.
The following NDT methods are described: leak testing, ultrasonic testing (UT) and computerised
tomography (CT). The leak test, which renders a quick and easy measure of the weld quality, was
performed on all workpieces. UT and CT were only performed on one or two workpieces.
In the second section, destructive tests are discussed. These tests render a measure of the weld
quality, but the workpiece is destroyed during the evaluation. The destructive tests described are
microscopic examination, torsion, peel and compressive testing.
Microscopic examination requires a longitudinal cross-sectioning of the workpiece for the weld zone.
This method was used to investigate the wave pattern of the weld interface.
The other three destructive tests (torsion, peel and compressive testing) were used to determine the
strength of the weld. An important criterion in the selection of the appropriate tests is the possibility
(and ease) of clamping the workpieces. The torsion test and compressive test were applied on
several workpieces. The peel test was investigated, but not executed.
80
5.2 Non-Destructive Testing Methods
5.2.1 Leak test
The first NDT method that was developed in the framework of this dissertation is a leak test. In the
test setup, the parts are connected to a source of pressurised air and are checked for leaks. A
schematic of the test rig is shown in Figure 5.1.
Figure 5.1: Leak test system – 1) pressurised air source,
2) valve, 3) pressure gauge and 4) coupling piece
For this test, a coupling piece had to be developed in order to connect the tubes to the pressurised
air source (Figure 5.2). The drawings of this part can be found in Appendix A. All drawings and
3D-models were made in “SolidWorks 2009 SP3”. A section view of the coupling and installed welded
specimen is given in Figure 5.3.
The vertical pipe was brazed upon the coupling and serves for the connection to the pressurised air
circuit. Due to the fact that the air will exert a force on the workpiece in its axial direction, also a
clamping device had to be developed. This exists merely out of a steel profile with a bolt to adapt to
the length of the tubes (Figure 5.4).
Figure 5.2: Coupling piece for pressurised air
Using this test set-up, both a qualitative and a quantitative evaluation of weld quality can be
performed. The qualitative inspection consists of nothing more than submerging the pressurised
workpiece in a water basin. If any leak is present, bubbles will form, revealing the leak. This method
can be carried out very quickly and gives a good first impression concerning the presence of any weld
defects. However, the severity of the defects cannot be quantified very accurately by a counting the
amount of bubbles.
81
Figure 5.3: section view of the coupling piece + workpiece
The second method can be used to quantify the magnitude of the leak. This method is illustrated in
Figure 5.1. The welded tubes (4) are connected to a pressurised air circuit which is depicted as a
pump (1). When the entire assembly has been brought to a prescribed pressure value, the valve (2) is
closed. The pressure gauge (3) will measure the pressure in the circuit. If it indicates that there is a
pressure drop, the welded connection is not leak free. The time necessary to obtain a certain
pressure decay can be measured and hence the severity of the leak can be described.
Figure 5.4: clamping device for the leak test
5.2.2 Ultrasonic Testing
Another NDT method is Ultrasonic Testing (UT). UT uses high frequency sound energy to detect flaws
in the material or connection. Ultrasonic investigation is the most widely used NDT method for welds
made by explosion welding [53]. A general description of the ultrasonic testing principle will be given
and afterwards some problems will be described which occur for magnetic pulse welded workpieces.
The system consists of a transducer, pulser/receiver and display devices. The pulser will order the
transducer to generate high frequency ultrasonic energy. In the form of sound waves, this energy is
transmitted into the material. When a flaw or a discontinuity is in the wave path, some of the wave
energy will be reflected to the surface. This reflection can be detected by the transducer, which will
transform it into an electrical current. The current will be gathered by the receiver and is displayed
on a screen. Of course, not only flaws will generate an echo but also the back surface of the material
will create an echo as well. The process and the different echoes are illustrated in Figure 5.5 [54].
Since there are multiple sources for echoes, it is important that the tests are conducted by a trained
person who can distinguish the signals and detect flaws. An investigation has been conducted were
technicians were ordered to investigate welds (complete joint penetration groove welds) that had
82
embedded flaws of known size, location and orientation, using the UT method. The outcome showed
that on average approximately 25% of the known discontinuities were missed. This stresses once
more the need of thorough testing by a skilled technician [55].
Figure 5.5: Ultrasonic testing principle [54]
The UT method is best used for searching flaws or cracks which are positioned normal to the
propagation direction of the waves. The perpendicular orientation will lead to the maximal possible
surface on which the waves will reflect, causing an echo. When positioned parallel to the wave
direction, the flaws are very difficult to discover (Figure 5.6). For testing of the MP welded tubes, this
does not create a problem because the main goal is to find connections with insufficient bonding.
These bonding flaws will be parallel to the surface of the workpiece, hence normal to the wave path.
However a bonding flaw does not necessarily mean a volumetric flaw. Where bonding is not
attained, the materials can still be in close contact with each other, making the flaw more difficult to
detect.
Figure 5.6: Normally positioned flaws can be found with more ease than parallel positioned flaws.
During the magnetic pulse welding process often two different materials are connected to each
other. The interface between those two materials will also deliver a reflection even when no flaws
are present. Indeed, the different materials will possess different properties (acoustic impedance,
density) and thus the waves will reflect on the interface. This will make the detection of flaws a lot
more difficult. As shown in Figure 5.7, the reflection of the waves will occur almost simultaneously on
83
the interface and the flaw. Values of acoustic impedance for several metals are given in Table 5.1.
The unit of acoustic impedance is the Rayleigh [Rayl]: 1Rayl = 1 kg/ms² or 1 MRayl = 106 kg/ms². [56]
Figure 5.7: Flaw in an interface between two different materials
The difference in acoustic impedance of two materials at an interface is referred to as the impedance
mismatch. The greater this impedance mismatch, the greater the percentage of acoustic energy that
will be reflected at the interface. As can be seen in Table 5.1, the values of acoustic impedance can
greatly differ. An aluminum-copper connection will for example reflect sound waves a lot more than
a copper-steel interface. Nevertheless all interfaces between different materials will generate a
reflection of the sound waves and will thus induce a difficulty in finding flaws[54].
Another difficulty of investigating MP welded specimens is created by their geometry. The circular
workpiece has no flat surface on which the transducer van be placed. To investigate the viability of
using handheld transducers, workpieces were sent to “Brutsaert Ingenieurs”. This company is
specialised in non-destructive and material testing. It became clear that these handheld transducers
are not suited to investigate these small diameter circular workpieces. Positioning of the transducer
cannot be performed precisely and the contact fluid is not evenly distributed.
To sum up the main problems of investigating a MP weld with ultrasonic investigation:
1. The interface will also reflect the sound waves
2. Positioning of a transducer onto the circular workpiece
A possible solution for these problems is the use of a focusing transducer and a basin of water, which
will serve as contact fluid. The focusing transducer will improve the signal-to-noise ratio, creating a
signal which is more detailed. Thus the difference between reflections by the interface and
reflections by flaws can be determined more easily. Furthermore, by submerging in water both the
workpiece as the transducer an even distribution of the contact fluid can be assured. This method
should be investigated experimentally in the future. Figure 5.8 shows the difference between a
focusing transducer and a transducer without the ability to focus. The transducer works with a
frequency of 1 to 10 mHz, depending on the precise model.
84
Material
Acoustic impedance
[Mrayl]
Aluminum
17,0
Brass
36,7
Copper
41,6
Steel, mild
46,0
Steel, stainless
45,4
Table 5.1: the acoustic impedances of various materials[57][58]
Figure 5.8: A normal transducer (left) vs. a transducer with the ability to focus the beam (right): the transducer with focus
generates a signal with significantly less noise and thus a much more detailed measurement
85
5.2.3 Computerised Tomography
Computerised Tomography, or CT-scan, is a technique which uses a large quantity of flat X-ray
images of an object, taken around a single axis, to produce 3-D cross-sectional images of an object.
This imaging technique is based on the difference of absorption of the X-rays in different materials.
Due to this difference in absorption, shadowgraphs can be made which then ultimately can be put
together to generate the 3D-model. Internal defects and other internal structures of the object can
be found on these images [59].
The method has already successfully been used in the evaluation of bonds which were made by
cladding through explosion welding. It was observed that the CT scan is capable of revealing the wavy
pattern which is often created during an impact weld. [53] These facts lead to the assumption that
computerised tomography can be a very good method for investigating the quality of a MP weld.
Figure 5.9 describes the components which are used during a CT-scan. The X-ray tube emits the X-ray
beam which then travels through the workpiece to the image intensifier. The computer captures the
signal that is generated in the intensifier and also controls the turntable on which the workpiece is
positioned [59].
Figure 5.9: A schematic view of a CT scan test rig with X-ray tube, turntable, image intensifier and PC with capturing
hardware[59]
Figure 5.10 shows a system which is commercially available for CT in industry. It allows workpieces
with maximum dimensions of 0,4 x 0,3 m and up to 10 kg. The maximum voltage which is placed
upon the X-ray tube is 240 kV. The resolution of this system is dependant on the size of the
workpiece to be investigated. However, the manufacturer assures a resolution lower than 1 μm,
which is surely enough to detect any significant flaws in the MP weld. Since our workpieces are well
within the range of dimension and weight, this machine could serve well for evaluating of the MPW
process [60].
In this project, two workpieces were sent to CEWAC 2 for examination using computerised
tomography. Those parts had first been examined with a leak test. One of the pieces was almost free
of leaks, the other leaked significantly. This would allow us to compare the tomography images of a
2
Centre d'études wallon de l'assemblage et du contrôle des matériaux [www.cewac.be]
86
relatively good weld and a low quality weld. The results of the tomography however were the same
for both workpieces; no flaws were found.
Figure 5.10: A 240kV CT system. It allows workpieces up to 10 kg and guarantees a resolution of 1 μs [60]
Figure 5.11 gives an example of a workpiece that underwent a CT-scan. No flaws could be identified.
The only coloured part is the cavity in between the flyer tube and the shoulder of the inner work
piece. This cavity is coloured green and can be easily observed on the image. One must keep in mind
that this is a cavity that can also be seen with the naked eye once the part has been cut through, so
no prove of the ability of finding small flaws is provided.
Figure 5.11: Example of a CT image made of a MP welded specimen. It shows now flaws and thus is not able to prove the
reliability of this method.
Presuming that computerised tomography should be able to reveal the flaws, this leads to the
conclusion that the resolution of the system which is used at CEWAC is far from sufficient to get a
proper view of any flaws which could occur.
87
5.3 Destructive Testing Methods
5.3.1 Microscopic investigation
In order to investigate the weld interface a destructive testing method is used: microscopic
investigation. During microscopic examination the tubes are cut through in the axial direction. After
cleaning the parts, they are embedded in epoxy resin, as shown in Figure 5.12. This embedding is
done manually at room temperature. When the resin is fully hardened, the sample is polished in
several steps to attain a surface in which the presence of flaws is minimized. On copper-aluminum
welds, the polishing has to be carried out very carefully. Since both metals are relatively soft,
scratches are made very easily and which will affect the image.
Figure 5.12: Copper-aluminium workpiece embedded in epoxy, after longitudinal cross-sectioning.
The polished surface is then placed under a microscope which can be used as a normal microscope
but it can also be used together with a computer. The data then will be sent to the computer and
software allows taking pictures. This method allows visual inspection of the weld in levels up to
micrometers. During the investigation it becomes clear whether the two materials are connected
and/or whether metallurgical/physical changes have taken place. These phenomena include the
formation of an intermetallic layer and a wave pattern of the weld interface.
Additionally the workpieces can be inspected with SEM (scanning electron microscopy) to investigate
the composition of certain layers and changes of the base material.
Some examples of microscopically inspected welds are given in Figure 5.13. The left image of the
figure shows a connection that was properly welded although no wave pattern was found. The two
materials are joined together and an intermetallic layer was formed. The right image shows a part
that was not bonded. Although some aluminium has been deposited on the copper surface, it is clear
to see that the connection did not suffice. It should be noted that with visual inspection with the eye
only this connection appears welded.
Figure 5.14 shows an image of a copper-brass weld. This connection shows a wavy interface and no
intermetallic layers. The mechanism of this wavy pattern is discussed in Chapter 2.
88
Figure 5.13: Copper-aluminum welded joint (left) and copper-aluminum not welded (right)
Figure 5.14: Copper-brass weld with wavy interface
5.3.2 Torsion test
5.3.2.1 Principle
Torsion testing is widely used to determine the shear strength of workpieces with a circular cross
section. In a conventional torsion test, a cylindrical specimen is twisted by a torque acting around its
axis, as shown in Figure 5.15. The shear stresses can be calculated from the measured torque M, and
the strain from the twisting angle Ψ [61].
Figure 5.15: Conventional torsion test [61].
In contrast to uniaxial tension tests, the stresses are not distributed uniformly over the cross section.
For a circular cross-section, in the absence of other loads, a pure shear stress state exists in each
89
point. Torsional elastic shear stresses vary linearly from zero at the axis of twist to a maximum at the
outer surface. Thus, in a solid circular bar, yielding will start at the surface of the bar.
The welded workpieces consist of a bar, joined to a hollow thin-walled tube. In these hollow thinwalled tubes, the entire cross-section of the tube is approximately at the same stress. If not
supported by a cylindrical bar inside the hollow tube, buckling failure could occur in the thin-walled
tube, due to the large diameter to thickness ratio. In addition, clamping both sides of the workpiece
is much easier with this support, as shown in Figure 5.18.
The objective of the torsion test is to determine the shear stress at which failure of the weld occurs
and the outer tube separates from the inner rod.
This shear stress at failure will be referred to as the ultimate shear strength of the weld. It should be
noted that our point of interest here is the shear strength at the interface of the two parts, and not
necessarily the distribution of stresses in the inner bar and the tube.
During the torsion test, the applied torque is continuously measured, and each test is conducted until
failure. It will end in fracture of the inner rod or of the hollow tube, or in shearing of the welded
zone. If failure occurs first in the base material of the inner rod and the weld stays intact, it can be
concluded that the ultimate shear strength of the weld exceeds the ultimate shear strength of the
rod base material. In this situation the weld strength cannot be determined accurately, but the shear
stress at failure is a lower bound for the ultimate shear strength of the weld. If the base material fails
before the weld, it can be concluded that the weld is sufficiently strong.
On the other hand, if the weld fails first, the ultimate shear strength of the weld can be determined
through an estimation of the weld length and the measured torque at fracture (as discussed in
§5.3.2.2).
Torsion testing equipment for tubular workpieces found in literature is shown in Figure 5.16. Both
sides of the welded workpiece are clamped in the chucks (the hollow tube with a cylindrical insert).
The torque and angle of rotation are measured continuously, while the applied torque is increased
incrementally [62] [63].
Figure 5.16: Tinius Olsen torsion testing equipment [64].
If the torsion test renders interesting information about the weld, a similar test setup could be
developed at Laboratory Soete.
90
5.3.2.2 Determination of the ultimate shear strength
Subjecting the welded workpiece to torsion stress is used as an objective test to determine the weld
quality. The applied torque at fracture yields a direct measure for the shear strength of the weld.
When one end of the welded work piece is clamped (no movement) and a torque is applied at the
other end, shear stresses will develop in the welded zone. The welded zone can be approximated by
a cylindrical surface with radius r (the inner rod outer radius) and length a (axial weld length), as
shown in Figure 5.17. The applied torque is measured continuously. When the shear stress reaches
the ultimate shear strength of the weld, the two surfaces will separate. Using formula (5.1), the
measured torque at fracture yields the ultimate shear strength of the weld.
The axial weld length must be determined using non-destructive evaluation methods, such as
ultrasonic inspection (UT) or computed tomography (CT). An alternative is to estimate the weld
length by conducting experiments with similar parameters and performing a microscopic inspection
or by post-mortem measurements.
¸=
With:
—
2# $ [
(5.1)
r = outer radius inner rod [mm]
a = axial length of welded zone [mm]
τ = ultimate shear strength of weld [N/m2]
T = torque required to separate the surfaces (in torsion) [Nm]
Figure 5.17: Schematic representation of torsion test
The inner rod radius (r) is measured before welding using a caliper. The high velocity impact of the
flyer tube causes significant deformation of the inner workpiece. The radius at the weld zone is
consequently slightly smaller than the original rod radius. For a more accurate calculation of the
shear strength, an average weld radius can be used. This average value should also be determined by
non-destructive testing, or by an estimate from similar experiments or by post-mortem
measurements.
Some concerns should be mentioned regarding the accuracy of the strength determination. First of
all, performing a non-destructive evaluation using UT or CT techniques to determine the weld length
(and indentation) is both time-consuming and expensive. Microscopic investigation on the other
hand renders two values for the weld length. These values only represent the weld length at that
particular location of the workpiece. As discussed in the section on the experiments (Chapter 6), it
91
was observed in several workpieces that the weld length can vary significantly over the
circumference. Therefore, the question rises how accurate these weld length measurements are,
being determined by microscopic investigation at a single location of the circumference.
Nevertheless, an estimate must be made regarding the weld length.
5.3.2.3 Preliminary torsion tests
In order to develop a torsion test setup, an estimate of the shear strength of the welded tubes is
required. The value of the ultimate shear strength relates directly to the necessary torque that has to
be exerted by the machine. In addition it determines the required clamping force on both ends.
In a first attempt to determine the shear strength of the weld, a M8 bolt was inserted in a tapped
hole in the inner rod of workpiece SD-CA-2.2 ( see § 6.6.3). A cylindrical steel rod was inserted in the
copper tube, as shown in Figure 5.18.
The copper tube (with cylindrical insert) was firmly clamped, preventing slip. Subsequently a torque
was applied on the bolt using a torque wrench. However, the bolt failed at a torque of approximately
55 Nm, preventing the realisation of sufficient torque to shear the welded zone.
As the shear strength of the bolted connection did not suffice, two flat zones were machined in the
inner rod, as shown in Figure 5.19. The clamping system and torque measurement using a torque
wrench are shown in Figure 5.20.
Figure 5.18: Preliminary torsion testing using a bolt in the inner rod. The tube is supported by a cylindrical bar.
Figure 5.19: Flattened zones to apply torque.
92
Figure 5.20: Clamping and torque measurement.
The workpiece was clamped and subjected to a gradually increasing torque. The aluminum inner rod
failed at a torque of about 140 Nm, as shown in Figure 5.21. This indicates that the weld shear
strength exceeds the shear strength of the aluminum rod.
The torsion test was also performed on the copper-brass workpiece SD CuMs1.99 (see §6.9.2). The
workpiece was clamped in a bank screw using a cylindrical insert inside the copper tube, and a
torque wrench with a limit of 150 Nm was used. At this maximum torque, the workpiece did not
remain fixed in the bench screw. A lathe was used in an attempt to fix the tube more rigid, but also
here the clamping force on the copper tube was insufficient. Finally, the tube was successfully
clamped in a bench screw with curved clamping plates. A torque wrench with a higher maximum
torque was used. The brass rod failed at a torque of 280 Nm, as shown in Figure 5.22. The fact that
the rod failed before the weld zone indicates that the weld strength exceeds the strength of the
brass base material.
Figure 5.21: Aluminum inner rod failed before the weld in the torsion test (SD CA 2.2)
93
Figure 5.22: Brass inner rod failed before the weld in the torsion test (SD CuMs 1.99)
5.3.2.4 Conclusion
Both the copper-aluminium weld and copper-brass weld showed failure of the inner rod. From these
results, the torsion test seems to be able to determine if the shear strength of a weld is stronger than
the base material. Only if the weld strength is lower, a value for the ultimate shear strength of the
weld can be obtained.
In addition, some difficulties were experienced with clamping the flyer tube properly. Developing a
torsion test with a bush around the flyer tube (with cylindrical insert) would also result in insufficient
clamping strength. These considerations lead to the conclusion that the torsion test is not very
suitable to determine a value for the weld shear strength. The only application of the torsion test is
to verify if the weld is stronger than the rod base.
94
5.3.3 Peel test
Peel testing is used to determine the strength of adhesive joints. The adhesive strength of bonded
strips of metals or plastics is determined by peeling or pulling strips off and recording the required
force. It is important to note that peel tests are normally used to compare rather than to measure
properties. The peel test also introduces a different stress state in the weld zone, as the angle at
which the material is peeled generally equals or exceeds 90 degrees. Unlike the torsion test, the weld
is not subjected to pure shear stress [65].
Several DIN and ASTM standard peel testing procedures for the quality control of adhesive joints are
described. A schematic overview is given in Figure 5.23. The standards differ mostly in the angle at
which the layer is peeled off [66].
Figure 5.23: DIN and ASTM peel testing procedures [66].
In a similar peel test method to evaluate the weld strength of spot welded sheets is described in [67].
The idea of peeling off a layer of a bonded structure can be transferred to evaluate the quality of
tubular magnetic pulse welds. A method is illustrated in Figure 5.24 [2]. Axial grooves are machined
through the welded zone of the workpiece, thus creating “strips”, which can be peeled off
separately. The proposed method includes a clamping head mounted on the end of a shaft.
Subsequently a torque is applied (and measured) on this shaft, which rotates and peels off the strip
of the welded flyer tube material. The torque required to peel off the welded material over a certain
angle is a measure for the peel strength of the material. Again, the concept peel strength is not a
universally accepted material characteristic. The peel test is intended to evaluate and compare the
quality of the welded joints.
The result of a peel test on a tubular welded work piece is shown in Figure 5.25. Strips of flyer tube
welded to the inner rod have been peeled off [68].
95
Figure 5.24: Peel test method for tubular welded joints[2]
Figure 5.25: Result of a peel test on a tubular welded joint [68]
96
5.3.4 Compression test
5.3.4.1 Principle
The compression test is very similar to a conventional tensile test. In the compressive test, the force
is reversed and the inner rod is pushed into the flyer tube, opposite to being pulled out of the flyer
tube in a tensile test.
The idea of the compressive test was found in the literature [38]: the setup shown in Figure 5.26 was
used to investigate the strength of electromagnetically formed joints made of aluminum tubes under
cyclic loads.
Figure 5.26: Pull-out test found in literature [38]
The compressive test is preferred over the tensile test for two main reasons. The first involves
clamping, which is rather difficult for tensile testing of magnetic pulse welds (clamping devices must
be custom made due to the limited specimen length). In the compressive test, the only requirement
is that both ends of the workpiece must be perfectly flat and perpendicular to the specimen axis, to
ensure that the workpiece does not bend during compression.
Secondly, in all experiments, the inner rod is machined with a collar. In a tension test, this collar
would have to be pulled against the material of the flyer tube. The collar cannot be reached for
removal, as it is located well inside the flyer tube. An additional operation to remove the collar would
thus be difficult to realise.
Prior to the compressive test, only a small additional operation has to be performed. Due to the high
energy impact, the flyer tube will make an indentation in the rod. The part of the inner tube, which is
not impacted, forms a small edge that would hinder the flyer tube material as the inner rod is pushed
into the flyer tube. Therefore, this small edge has to be removed before the compressive test is
carried out (by turning). Figure 5.27 shows the edge on left, and the removal of the edge by turning
on the right.
97
Figure 5.27: The left figure shows the ‘edge’ that is formed by the indentation in the rod. This edge must be removed by a
turning operation before compressing testing is executed.
During compressive testing, the inner rod is pushed out of the flyer tube with a force (F),
schematically shown in. Shear stresses (τ) develop in the weld zone, which has a quasi-cylindrical
surface. The radius (r) and the weld length (a) must be estimated, as discussed in the section on the
torsion test.
Figure 5.28: Schematic representation of the compressive test
The compression force is gradually increased until the weld ultimately fails, when the shear stress
reaches the ultimate shear stress of the weld:
¸=
2#. [
(5.2)
With: r =outer radius inner rod [mm]
a = axial length of welded zone [mm]
τ = ultimate shear strength of weld [N/m2]
F =compression force [N]
Figure 5.29 shows a photograph of the compression test setup. The rod is pushed down and the weld
zone is sheared. To avoid instability during the compression, it is important that the tube length is
not too large.
During compression, the force and displacement are continuously monitored. A graph of the applied
force versus the displacement can be generated for each push test.
98
Figure 5.29: Compression test
5.3.4.2 Preliminary compression testing on copper-aluminium weld
Workpiece SD-CA-2.2(see §6.6.3) was first subjected to the torsion test. Because only the aluminium
failed, the same workpiece could also be subjected to the compression test after cutting off the
fractured part of the rod and flattening the surface. The weld sheared at a compressive force of
12 kN. The displacement at this maximum force was 0,46 mm.
Figure 5.30 shows the force-displacement graph measured during the push test on workpiece SD-CA2.2. The graph clearly shows a linear relationship between the force and displacement, indicating the
development of elastic shear stresses. At a maximum force of 12 kN, the weld sheared and the inner
rod is physically separated from the outer tube. After the weld has failed, a certain force is still
required to completely push the inner rod out of the flyer tube.
The maximum force is used to determine the ultimate shear strength of the weld. It is difficult to
determine a very accurate value for the shear strength, because the weld was not formed over the
entire circumference, as discussed in the experiments (Chapter 6). It is assumed that the welded
zone of workpiece SD-CA-2.2 extended over approximately half of the circumference. The weld
length in this welded zone varied between 2,5 and 4,5 mm (average=3,5 mm), and the average
diameter measured 17,5 mm. Using these values (with an estimated correction for the partial weld),
the shear strength of the weld zone is calculated as 124,7 MPa.
The ultimate shear strength for aluminium alloys is approximately 65% of the ultimate tensile
strength [69]. The minimal ultimate tensile strength of aluminium EN AW-6060 in the T6 condition is
170 MPa, so the minimum ultimate shear strength of the aluminium rod is approximately 110.5 MPa.
The strength of the connection thus exceeds the strength of aluminium, which was confirmed by the
fact that the aluminium rod failed first in the torsion test.
99
Force [kN]
SD CA 2.2
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
0
0,2
0,4
0,6
0,8
1
Displacement [mm]
Figure 5.30: Force-displacement curve recorded during a compression test on workpiece SD-CA-2.2
5.3.4.3 Compression testing of copper-brass welds
Compressive testing was performed on several copper-brass workpieces. The results and discussion
can be found in §6.10.1.2.
5.3.4.4 Conclusion
The advantage of the compressive test is that shear stresses are induced in the weld, while the base
material of the rod and tube is subjected to compressive stresses. As a result, the rod will not fail
during the compression test (as opposed to the torsion test). The force-displacement curve
measured during compression allows calculating the weld shear strength. In addition, it shows the
displacement before fracture.
In conclusion, the compression test is more suitable from a practical point of view to evaluate the
strength of the welded tubes than the torsion test. Therefore, several copper-brass workpieces were
subjected to the compressive test in
100
Chapter 6
Experiments
6.1 Introduction
The magnetic pulse welding process has been applied to tubular workpieces in several series of
experiments. As mentioned in the introduction, the MPW process is very suitable for joining metal
tubes for HVAC applications. The most obvious application is welding of two tubes. Traditional
welding techniques such as MIG and TIG allow joining two tubes with equal diameter. Using the
MPW technology, the outer tube diameter must be larger than the inner tube diameter. The
difference in diameter determines the stand-off distance, which is an important parameter in the
pulse welding process. Because a certain overlap length is necessary for the MPW process to be
successful, the length of both tubes must exceed the length that is required for the final application
of the joined tube. For example, two tubes with a length of 50 mm welded by the MPW process will
result in a joined tube with length 90 mm (not 100 mm). This is illustrated in Figure 6.1 (left). Due to
the high velocity impact of the flyer tube, the inner tube will also deform plastically. A mandrel
should be used inside the inner tube to prevent severe deformation.
Figure 6.1: Using the MPW process to weld two tubes, a certain overlap length and stand-off distances are required. If
the inner part is a tube, an internal mandrel should be used for support. In the experiments performed in this thesis,
solid rods are used as inner parts instead of tubes. The rod has a collar, which fits into the outer tube. The diameter of
the rod at the welding zone determines the stand-off distance.
In the experiments performed in this thesis, a solid internal workpiece was used, as shown on Figure
6.1 (right). This configuration is much easier to work with, as no mandrel is required to support the
rod. An application example of a tube welded to a solid rod could be an end cap for small copper
pipes.
101
6.2 Overview
6.2.1 Test configuration
The collar of the rod has the same diameter as the inner diameter of the flyer tube. At the welding
zone the inner rod is machined to a specific diameter. The flyer tube thickness together with the
welding zone diameter of the inner rod determine the stand-off distance, because the outside
diameter of the flyer tube is constant in all experiments. The outside flyer tube diameter is equal to
the inside diameter of the isolated field shaper, to ensure that the workpiece is accurately centered.
This configuration is depicted in Figure 6.2. The copper tube and inner rod are then inserted into the
center of the coil (with field shaper) and clamped, thus preventing movement of the workpiece
during the welding process. Before conducting any experiment, the outer tube and inner rod
dimensions are measured to ensure the correct geometrical parameters. Subsequently both parts
are cleaned thoroughly using acetone. This removes all contaminations from the surface, such as
dust particles and lubricating oil from the grinding process of the rod. These contaminations could
possibly interfere with the welding process, as discussed further in this chapter.
Figure 6.2: An illustration of the test configuration
The flyer tubes used in the experiments are extruded copper tubes (Cu-DHP R290). Several series of
experiments are performed using aluminium inner rods (EN AW-6060 T6). These experiments are
discussed in the Copper-Aluminium section. Experiments are also performed using copper tubes and
brass (CuZn39Pb3 alloy) inner rods. These copper-brass experiments are a continuation of the MPW
research of dr.ir. K. Faes (BWI).
The workpieces will be named after the project “SOUDIMMA”(SD) as follows: SD + “material
combination ”+“series”+“number”. For example, the fifth workpiece of series 2 of copper-aluminium
experiments will be named “SD-CA-2.5”.
6.2.2 Weld quality evaluation
After the welding process, all workpieces are subjected to the custom leak test. The welded
workpieces are internally pressurised with air at 4 bar and submerged in water. The amount of leaks
102
can be determined visually and the rate of bubbles per second can be estimated. Based on these
criteria, a classification of severity of the leaks is proposed, by means of letters A through E. Only if
no bubbles were detected, the workpiece passes the leak test (class A). Class B indicates less than 5
bubbles per second, class C indicates 5 to 10 bubbles per second and D indicates 10 to 20 bubbles
per second. Class E marks the workpieces that show an extreme amount of bubbles (Table 6.1).
Class indicating
severity of the leak
A
B
C
D
E
Number of bubbles
per second
No bubbles
<5
≥5 and <10
≥10 and <20
≥20 (very severe leak)
Table 6.1: Classes indicating the severity of the leak detected by the leak test. The welded workpieces are internally
pressurised with air at 4 bar and submerged in water. The severity is determined by estimating the number of bubbles
per second formed at the weld.
This leak test renders a quick measure for the quality of the weld. Being leak free is surely important
for welding components, for example in air-conditioning or cooling applications. Other leak tests, for
example the helium leak test, can determine small leak rates more accurately. However, this
equipment is much more expensive. Five copper-brass workpieces were subjected to a helium leak
test at CEWAC.
Other non-destructive evaluation techniques were applied to some of the welded workpieces. Two
copper-aluminium welds were inspected at CEWAC using computed tomography, a technique using
X-ray technology to obtain 3-D images of the workpiece. The ultrasonic wave reflection technique
was used by Brutsaert to evaluate the quality of one of the workpieces. Both technique are discussed
in the chapter Non-Destructive Testing.
Two destructive tests were applied to determine the weld strength. The strength of the weld is also
an important requirement for the connection. A weld is considered to be of high quality if it is leak
free and has sufficient strength. The destructive tests applied in this thesis are the torsion test and
the compressive (push through) test. Both tests introduce shear stress in the weld zone. The
workpiece is loaded until failure occurs. The torque in the torsion test (and the compressive force in
the push through test) at failure are a measure of the strength of the welded connection. These
destructive testing methods used to evaluate the welds are discussed in the chapter Destructive
Testing.
Another destructive test is microscopic examination of the welds. A welded workpiece is cut through
longitudinally and embedded in an epoxy resin. The welded zone is examined using a microscope to
determine the weld length, the occurrence of weld defects and to investigate the presence of a wave
pattern at the weld interface, which is typical for the magnetic pulse welding process.
103
6.3 Material Characteristics
Experiments were performed using copper tubes and brass and aluminium inner workpieces. The
material properties are described below.
Brass:
The CuZn39Pb3 alloy is used for the brass inner workpieces; its chemical composition is given in the
table below.
Cu
56,5 – 58,5 %
Zn
39 %
Pb
2,5 – 3,5%
The physical and mechanical properties of the brass alloy are given in the table below.
Brass: CuZn39Pb3
o
Density (at 20 C)
3
8,47
g/cm
15 (12)
MS/m
Young’s modulus (at 20 C)
97
GPa
Yield strength (Rp 0,2)
250
MPa
Ultimate tensile strength (Rm)
430 (min.)
MPa
Elongation
10
%
Hardness
120
HB
o
Specific electrical conductivity at 20 C
o
(200 C)
o
Copper:
The material used for the copper tubes are extruded Cu-DHP R290, consisting of:
Cu
> 99 %
P
< 0,04 %
The following table lists the physical and mechanical properties of copper R290, after cold
deformation (extruding process).
Copper: Cu-DHP R290
o
Density (at 20 C)
3
8,94
g/cm
43 (30)
MS/m
Young’s modulus (at 20 C)
132
GPa
Yield strength (Rp 0,2)
250
MPa
Ultimate tensile strength (Rm)
290 (min.)
MPa
Elongation
5
%
Hardness
90 - 115
HB
o
Specific electrical conductivity at 20 C
o
(200 C)
o
104
Aluminium:
The material used for the aluminium used for the inner rods in the Cu-Al experiments is EN AW-6060,
in the T6 condition.
Al
> 98,4 %
Mg
0,3 – 0,6 %
Si
0,3 – 0,6 %
The properties of this aluminium alloy are:
Aluminium: EN AW-6060 (T6)
o
Density (at 20 C)
3
2,7
g/cm
34 - 38
MS/m
Young’s modulus (at 20 C)
69,5
GPa
Yield strength (Rp 0,2)
140
MPa
Ultimate tensile strength (Rm)
170 (min.)
MPa
Elongation
8
%
Hardness
60
HB
o
Specific electrical conductivity at 20 C
o
105
6.4 Welding parameters
The magnetic pulse welding process requires that several parameters are set to optimal values in
order to obtain a successful weld (see Figure 6.3). By combining these appropriate parameter values,
weldability windows can be determined.
Figure 6.3: Several important parameters of the magnetic pulse process.
As the welding mechanism of magnetic pulse welding is similar to that of explosive welding, the
parameters that determine the formation of a magnetic pulse weld are also the same. The impact
velocity and the impact angle affect the weld formation the most, as discussed in §2.5. These two
parameters cannot be set directly or individually, but are determined by the geometrical parameters
(stand-off distance, overlap length, tube thickness), material characteristics (density, conductivity)
and machine settings (charging voltage).
The most important parameters are briefly discussed in this section. The charging voltage is the only
machine setting that can be varied. The other variable parameters are determined by the
geometrical configuration of the workpieces and coil or field shaper. The coil (or field shaper) inner
diameter determines the outer diameter of the flyer tube and its width will determine the magnetic
pressure. This outer diameter is chosen equal to the internal diameter of the insulation of the field
shaper, in order to center the workpiece inside the coil. The thickness of the tube determines the
inner diameter. The inner rod has a collar, which fits exactly inside the flyer tube for alignment
purposes. The diameter of the rod at the welding zone can be chosen, and determines the stand-off
distance, which is the radial air gap between the inner rod and the outer tube. Finally, the length of
the flyer tube determines the position of the tube end in the field shaper.
106
6.4.1 Stand-off distance
The stand-off distance is calculated as:
With:
7;[M − ‘];[M›C =
¥x,S‰¹W − 2; − ¥4xe
2
(6.1)
D0,tube = outer diameter of flyer tube [mm]
t = flyer tube thickness [mm]
Drod = inner rod diameter at the welding zone [mm]
The stand-off distance is the distance over which the flyer tube is accelerated by the magnetic
pressure. If the magnetic pressure is assumed to be constant in time, as was assumed in the
calculations by the manufacturer of the machine, the acceleration is constant as well. The impact
velocity can therefore be calculated as:
lVFT\S = √2. ]. [
With:
(6.2)
vimpact = impact velocity [m/s]
s = stand-off distance [m]
a = acceleration due to magnetic pressure [m/s²]
The assumption of a constant acceleration implies that the impact velocity is proportional to the root
of the stand-off distance (assuming that the charging voltage is constant). In reality, the magnetic
field caused by the damped sinusoidal current is strongly time dependant. Thus, the pressure exerted
on the tube and its acceleration (proportional to the square of the magnetic field) will also be time
dependant. The formula above is then no longer valid, and neither is the simple relation between the
stand-off distance and the impact velocity. The velocity of the tube will be a function of time, and the
velocity at impact cannot be calculated unless the time function of both the magnetic pressure and
the deformation pressure are known. Note that the magnetic pressure will change when the spacing
between the flyer tube and the field shaper changes. For a given charging voltage, each stand-off
distance will correspond with a different velocity at impact, but no easily interpretable relation
between the two can be established. If the time functions of the magnetic pressure (which also
depends on the tube material) and the deformation pressure are known, an optimal value for the
stand-off distance (which will result in the optimal impact velocity for welding) could be calculated.
This was not possible within the limits of this master thesis, so the objective is to determine the
optimal value (or interval) for the stand-off distance experimentally. It is important to note that the
stand-off distance not only affects the impact velocity, but also the impact angle.
6.4.2 Charging voltage
The only electrical parameter that can be set during the experiments, is the voltage level of the
capacitors, which is directly related to the energy level in the system:
8=L
j$
2
(6.3)
With: E = energy stored in the system [J]
V = charging voltage[V]
C = capacitance of the capacitor bank (160 µF)
107
The amplitude of the discharge current through the coil is proportional to the charging voltage. The
magnetic pressure, responsible for the both the acceleration and deformation of the flyer tube, is
proportional to the square of the current (and hence also of the voltage). The voltage level only
affects the magnitude of the current, pressure and acceleration - not their time functions. So, for a
given geometrical configuration and tube material, an increasing voltage level will result in a larger
acceleration and larger impact velocity.
The voltage is limited to a certain maximum allowed value. The discharge current is damped
sinusoidal, and hence reverses direction in time. The current amplitude in the reversed direction is
restricted by the MPW machine. A procedure has to be followed to determine the maximum allowed
voltage of the capacitors. An experiment is performed at a voltage level of 15 kV, and the current
waveform is measured. The first two peaks of the damped sinus are measured with the software.
Their ratio is multiplied by 15 kV to determine the maximum allowed voltage level.
jVTº,TGGxyWe =
oFWT˜$
. 15 žj
oFWT˜
(6.4)
With: Vmax,allowed = maximum allowed voltage level [kV]
Ipeak1 = amplitude of the first current peak [kA]
Ipeak2 = amplitude of the second current peak [kA]
6.4.3 Overlap length
The overlap length is the length that the flyer tube overlaps with the field shaper and determines the
tube end position in the field shaper. It is an important parameter because the field shaper
concentrates the magnetic field to a small region, and the magnetic pressure will be exerted only on
that part of the tube that overlaps with the field shaper. The overlap length is determined entirely by
the flyer tube length as the tube is pushed in the same position every experiment by the clamping
mechanism. It can be calculated as:
»C#[Z CM;ℎ = šC# ;¼‚C CM;ℎ − 38 __
(6.5)
The overlap length is not frequently discussed in literature, although it has an important influence on
the impact angle. The deformation of the tube is also affected by the overlap length and therefore
also the pressure required for acceleration. It is believed however that the charging voltage and the
stand-off distance have a larger influence on the velocity.
108
6.5 Field shaper damage
6.5.1 Occurrence of consistent but unexpected weld defects
The field shaper was examined for damage after an unexpected weld pattern was consistently found
in the copper-brass welds. Prior to microscopic evaluation, a longitudinal cross-section was made of
the welded workpieces. One half was cleaned, degreased and embedded to perform a microscopic
examination. For the other half of the weld, the inner rod was deliberately separated from the flyer
tube (the weld was broken).
The wave pattern could be seen visually on the outer surface of the inner workpiece. The axial length
of the wavy zone varied over the circumference. However, weld defects were noticed in all of the
copper-brass welds. The wavy weld zone was interrupted at the position 180° relative to the field
shaper slit. The width of the interrupted zone was always about 3 to 5 mm. Remarkably, at the field
shaper slit position, where the magnetic field (and thus pressure) is locally reduced, weld defects
were only noticed at rare occasions. If however interruptions at the slit position were present, they
were significantly smaller than those at the 180˚ position, as shown in Figure 6.4.
Figure 6.4: Weld defects were noticed in all of the copper-brass welds. The wavy weld zone was interrupted at the
position of 180° relative to the field shaper slit. Remarkably, at the FS slit position, where the magnetic pressure is locally
reduced, weld defects were only noticed at rare occasions. If however interruptions at the slit position were present
(left), they were significantly smaller than those at the 180˚ position (right).
The weld defect at the 180˚position relative to the field shaper slit is also shown in Figure 6.5. The
fact that there was no wave pattern at this location indicates that there was an irregularity during
the process. In several welds, interruptions of the wave pattern were also noticed at the positions
90°and 270° relative to the slit. These weld defects did not occur for every weld.
Due to this recurrent weld defect, the decision was made to disassemble and inspect the field shaper
in order to determine the defect cause.
After inspection (§ 6.5.2), it was clear that the width of the 3 cracks greatly exceeded the width of
the field shaper slit. Especially the crack at the 180° position is extremely wide. The fact that the weld
pattern in the copper-brass welds is interrupted at the 180° position is a direct consequence of the
109
large crack at this location. This crack causes a distortion in the current path, resulting in a magnetic
field pressure reduction. As the width of the crack is a multitude of the width of the field shaper slit,
the reduction of the magnetic pressure is significantly larger at the 180° position. In fact, as no
significant defects were seen at the slit position, the pressure reduction at the slit region does not
have a very large influence on the weld quality at this location.
Figure 6.5: The wavy weld zone was interrupted at the position 180° relative to the field shaper slit in all copper-brass
experiments. The width of the interruption was 3 to 5 mm.
6.5.2 Nature and cause of field shaper damage
During the experiments the field shaper has been damaged severely. The damage was discovered
during inspection, after some unexpected welding defects were observed. Severe cracks developed
at the inside surface of the field shaper at the positions 90°, 180° and 270° relative to the radial slit,
as shown in Figure 6.6. The radial slit is located on the right side of the photograph, with isolating
tape to separate both sides. The cracks all initiated at the inside surface of the field shaper and
propagated in the radial direction.
Figure 6.7 shows a detail of the most severe crack, which was located opposite to the field shaper
slit.
At the 90° position relative to the field shaper slit, the crack was not so wide but very long. This crack
also initiated on the inner field shaper surface and propagated in a radial direction. When reaching
110
the end of the field-concentrating zone (with a smaller axial thickness), the crack further propagated
in circumferential direction, as shown in Figure 6.8.
Figure 6.6: Field shaper damage.
Figure 6.7: The most severe crack was located opposite to the field shaper slit.
Figure 6.8: Crack at 90° relative to the field shaper slit.
111
At the time that the damage was observed, the magnetic pulse welding machine had discharged
about 1500 pulses over its lifetime. Before this event, the condition of the field shaper had never
been checked. Consequently, no conclusions can be drawn regarding the first occurrence or the
evolution of the cracks.
Damage presumably initializes at these locations due to the large tensile stresses at the inner surface
during the shock load. Despite the absence of physical contact with the components, the field shaper
is subjected to a reaction force; a radial outward force directed oppositely to the force acting on the
flyer tube. As the field shaper has a slit, this pressure causes the two ‘halves’ to be pushed apart,
initiating a crack opposite to the slit. Due to symmetry, cracks also initiate at 90° and 270° relative to
the slit. The crack formation due to the reaction forces is schematically shown in Figure 6.9 [70] .
In addition to the cracks at in the inside surface of the field shaper, the insulation surrounding the
field shaper was partially burnt. High discharge currents (>100 kA) flow primarily near the surface of
the field shaper. Due to the narrow crack formation, the current flow pattern is distorted. The
current will try to take the shortest path, rather than follow the damaged inside surface around the
cracks. This causes the currents to pass the narrow air gap of the cracks, causing sparks. These sparks
cause further erosion of the cracks and burning of the insulation.
Figure 6.9: Field shaper damage due to reaction forces [70].
112
6.6 Copper-Aluminium experiments
6.6.1 Introductory comments related to field shaper damage
Three series of welding experiments were completed using copper flyer tubes and solid aluminium
internal workpieces. A fourth series was scheduled to be performed. Due to the discovery of the
damaged field shaper, this series was postponed. The damage might also have had a negative
influence on the quality of the welds in the first three series. The regular discharge flow of the
current through the field shaper is disturbed by the severe cracks located at the inside surface of the
field shaper. If the current path is deviated, irregularities in the magnetic field formation are
unavoidable. If the magnetic field is distorted by means of local reduction of field strength at the
cracks, the pressure exerted on the flyer tube will also be reduced at these positions. The nonuniform pressure distribution will result in a non-uniform impact on the inner workpiece, and
consequently the weld formation might be disturbed. This effect was noticed during the copperbrass experiments. In these experiments the welds were formed successfully, so the effect of the
field shaper damage was more obvious than in the case of the copper-aluminium experiments,
where the welds were mostly unsuccessful. Therefore, all of the CA-experiments (copper-aluminium)
will be discussed jointly in the following sections.
6.6.2 Series 1 (SD-CA-1)
CA Series 1
Material
Flyer tube
Inner rod
Copper
Aluminium
Outer
Diameter
25 mm
16 mm/17 mm
Thickness
1,5 mm
/
The first series of the copper-aluminium weld trails counted twenty experiments. The copper flyer
tubes all have an outer diameter of 25 mm and a thickness of 1,5 mm. The length of the tubes was
varied between 46 mm and 50 mm resulting in different overlap lengths of the tubes with the field
shaper (see table below). As the position of the clamping mechanism is constant, an increase of the
tube length increases the overlap length, on which the magnetic pressure acts. The inner rod
diameter at the welding zone was chosen equal to 16 mm or 17 mm. The rod diameter determines
the stand-off distance, i.e. the radial air gap between the inner rod and the outer tube. These
geometrical parameters define different configurations: five different overlap lengths and for each
overlap length, two different stand-off distances. Finally, for each geometrical configuration, the
charging voltage was set at 15 kV and at the maximum allowed voltage level (about 19 kV).
An overview of the experiments of the first series is shown in Table 6.2. For each experiment, the
geometrical parameters (internal workpiece diameter, stand-off distance, tube length and overlap
length) and the charging voltage are listed. The third column from the right indicates whether a weld
was formed in the workpiece. The results of the leak test are listed in the right column.
None of the experiments of the first series resulted in a successful weld: all workpieces failed the leak
test. The majority of the workpieces even showed large leaks (classes C,D and E). The workpieces
were cut in half after welding to perform microscopic examination. The inner workpiece separated
from the outer tube after this operation for almost every specimen. One workpiece (SD-CA-1.12)
113
remained intact after the cutting operation. During microscopic investigation, it appeared that the
workpieces were only welded at one side. Although no qualitative weld was created during this
series of experiments, a large quantity of deposition of aluminium was found on the inside surface of
the copper flyer tubes (see Figure 6.10 for specimen SD-CA-1.17). The aluminium is found on the
inner tube surface at the edge of collar and in the middle of the weld zone. No deposition was seen
at the end of the flyer tube (first point of impact).
Test
Number
Voltage
(kV)
Diameter
(mm)
Stand-off
(mm)
Tube length
(mm)
Overlap
(mm)
Weld ?
Leak
free?
Leak
Class
SD-CA-1.1
15
16
3,0
46
8
No
No
B
SD-CA-1.2
19
16
3,0
46
8
No
No
D
SD-CA-1.3
15
17
2,5
46
8
No
No
B
SD-CA-1.4
18.5
17
2,5
46
8
No
No
C
SD-CA-1.5
15
16
3,0
47
9
No
No
E
SD-CA-1.6
19
16
3,0
47
9
No
No
C
SD-CA-1.7
15
17
2,5
47
9
No
No
C
SD-CA-1.8
18,5
17
2,5
47
9
No
No
B
SD-CA-1.9
15
16
3,0
48
10
No
No
E
SD-CA1-.10
19
16
3,0
48
10
No
No
B
SD-CA-1.11
15
17
2,5
48
10
No
No
B
SD-CA-1.12
19
17
2,5
48
10
Partially
No
C
SD-CA-1.13
15
16
3,0
49
11
No
No
D
SD-CA-1.14
19
16
3,0
49
11
No
No
D
SD-CA-1.15
15
17
2,5
49
11
No
No
B
SD-CA-1.16
19
17
2,5
49
11
No
No
B
SD-CA-1.17
15
16
3,0
50
12
No
No
C
SD-CA-1.18
19
16
3,0
50
12
No
No
C
SD-CA-1.19
15
17
2,5
50
12
No
No
C
SD-CA-1.20
19
17
2,5
50
12
No
No
C
Table 6.2: First series of copper-aluminium weld trials
The indentation of the inner workpieces, due to the high velocity impact of the flyer tube, was very
large in most of the experiments of the first series (Figure 6.10, Figure 6.11). Several reasons are
possible for the excessive plastic deformation of the inner rods. The impact energy of the flyer tube
was clearly too large. This can be related to the mass of the flyer tube, or to the velocity at impact
with the inner rod. The flyer tubes all had an outside diameter of 25 mm and a thickness of 1,5 mm.
This thickness was used in all copper-aluminium and copper-brass experiments. In other words, the
impacting mass was kept constant. The impact velocity depends on both the charging voltage and
the stand-off distance. The voltage over the capacitor bank is proportional to the discharge current.
The magnetic pressure that accelerates the flyer tube is proportional to the square of the discharge
current. An excessive voltage could cause an acceleration (and thus impact velocity) which is too
large. Similar voltage levels were applied in the copper-brass experiments, but the plastic
deformation of the brass internal workpieces was not as large. This is of course also related to the
114
fact that the yield strength of the brass alloy is 250 MPa, while that of aluminium is only 140 MPa.
The other parameter that has a large influence on the impact velocity is the stand-off distance. This is
the distance over which the flyer tube can accelerate. If the stand-off is increased, the impact
velocity will also increase (assuming constant acceleration). The stand-off distance in the
experiments of the first series was chosen equal to 2,5 or 3 mm. These values were most probably
too large. Therefore, a second series of copper-aluminium experiments was performed where the
stand-off distance was decreased.
Figure 6.10: No weld was formed during the MPW process of workpiece SD-CA-1.17. The inner rod separated
spontaneously from the outer tube during cutting through. Deposition of aluminium on the inside surface of the copper
tube was found in all of the Copper-Aluminium experiments.
115
6.6.3 Series 2 (SD-CA-2)
CA Series 2
Material
Flyer tube
Inner rod
Copper
Aluminium
Outer
diameter
25 mm
18 mm/19 mm
Thickness
1,5 mm
/
The objective of the second series was to verify the influence of the stand-off distance on weld
formation and plastic deformation of the internal workpiece. This series was performed using inner
rods with a diameter of 18 and 19 mm, thus reducing the stand-off distance to 2 and 1,5 mm
respectively. The tubes had a length of 47, 48 and 49 mm, which is similar to the experiments of the
first series. The voltage was again set at 15 kV and at the maximum allowable level. The experiments
of the second series are listed in Table 6.3.
Test
Number
Voltag
e (kV)
Diameter
(mm)
Stand-off
(mm)
Tube length
(mm)
Overlap
(mm)
Weld?
Leak
Free?
Leak
Class
SD-CA-2.1
15,0
18
2,0
47
9
No
No
D
SD-CA-2.2
18,0
18
2,0
47
9
Partially
No
C
SD-CA-2.3
15,0
18
2,0
48
10
No
No
E
SD-CA-2.4
18,5
18
2,0
48
10
Partially
No
C
SD-CA-2.5
15,0
19
1,5
48
10
No
No
D
SD-CA-2.6
18,5
19
1,5
48
10
No
No
C
SD-CA-2.7
15,0
18
2,0
49
11
No
No
E
SD-CA-2.8
18,5
18
2,0
49
11
Partially
No
B
SD-CA-2.9
15,0
19
1,5
49
11
No
No
E
SD-CA-2.10
18,5
19
1,5
49
11
Partially
No
C
Table 6.3: Second series of copper-aluminium weld trails
The indentation of the inner rod was clearly much smaller by reducing the stand-off distance. Figure
6.11 shows a comparison of the plastic deformation of the inner rod for two different stand-off
distances. The stand-off distance in the experiment of the left picture was 3 mm (workpiece SD-CA1.14) and that of the right one was 2 mm (workpiece SD-CA-2.8). In both experiments the tube
length was 49 mm and the voltage level was about the same: 19 kV for workpiece SD-CA-1.14 and
18,5 kV for SD-CA-2.8. A more thorough discussion on the influence of the voltage level and stand-off
distance on the indentation of the inner rod can be found further in this chapter.
None of the workpieces of the second series passed the leak test. Nevertheless, some interesting
conclusions can be drawn from the leak classes, listed in the right column of Table 6.3. For each
geometrical configuration, two experiments with different voltage levels were performed: one at
15 kV and one at 18,5 kV. The workpieces formed at 18.5 kV showed significantly less leakage than
those formed at 15 kV. For instance, the leaks found in workpiece SD-CA-2.7 were class E, while
workpiece SD-CA-2.8, with the same geometry but higher voltage level, only had small leaks (class B).
116
After the leak test, the workpieces were cut longitudinally for microscopic examination. In most
experiments, the flyer tube spontaneously separated from the inner tube at both sides after the
cutting operation. Only in experiments SD-CA-2.4, SD-CA-2.8 and SD-CA
CA-2.10, separation did not
occur. These are the experiments performed at the highest voltage level (18,5 kV), which also
showed the smallest leaks. However, the flyer tube was only bonded to the internal workpiece at one
side, as shown in Figure 6.12.
6 . A weld was created between the tube and the rod at the right hand
side of the photograph, but not at the left side. Workpiece SD-CA-2.2 was subjected to a torsion and
a compressive test to determine the weld strength. After examination of the surface of the inner rod,
it was observed that also this specimen probably was only partially welded.
Figure 6.11:: The plastic deformation of the inner rod is much more severe with an increase in stand-off
stand
distance. The
stand-off
off distance in the experiment of the left picture was 3 mm and that of the right one was 2 mm. In both
experiments the tube length was 49 mm and the voltage level was about the same: 19 kV for workpiece SD-CA-1.14 and
18,5 kV for SD-CA-2.8.
CA-2.4, SD-CA-2.8 and SD-CA-2.10 were the only copper-aluminium
aluminium experiments where the
Figure 6.12: Workpieces SD-CA
tube did not separate from the rod after cutting. None of these workpieces was fully welded, as the tube was connected
to the rod only at one side of each half of the workpiece.
There were no clear wave patterns on the surfaces of the inner rod or flyer tube. Nevertheless, a
successful bond was formed between the tube and the rod. This joint might have been broken at one
side by the cutting operation itself, or by the release of residual stresses after cutting
c
(which were
introduced by the diameter reduction of the tube). Although a partially-welded
partially
workpiece is
preferred over a non-welded,
welded, this is not sufficient to be a successful weld.
Workpiece SD-CA-2.2 was subjected to a torsion test. The aluminium workpiece
workpiece failed instead of the
weld, indicating that the weld of the tube and the rod had significant strength. The same workpiece
was then subjected to the compressive test, where the weld failed at 12 kN. These destructive tests
117
are discussed further in this chapter. From the surface examination after the compressive test, it
could be concluded that the workpiece was presumably welded only at one side. This might suggest
that the other partially-welded workpieces (SD-CA-2.4, SD-CA-2.8 and SD-CA-2.10) were also welded
at one side due to a certain asymmetry in the welding process, rather than that the weld was broken
at one side by cutting.
6.6.4 Series 3 (SD-CA-3)
CA Series 3
Material
Flyer tube
Inner rod
Copper
Aluminium
Outer
diameter
25 mm
17 mm
Thickness
1.5 mm
/
The third series of experiments was performed to investigate the possible influence of the jet
material on the weld quality. It was investigated if the jet can get trapped in the welding zone and at
the collar of the inner rod. Experiments were performed using inner rods with a similar geometry as
in the previous experiments, but with axial grooves machined in the collar. The grooves create a path
for the jet material to exit the welding zone. The grooved collar is shown in Figure 6.13.
Figure 6.13: The grooves in the collar of workpiece SD-CA-3.1 create an exit for the jet. The photograph shows that less
jet material remains behind the collar and more importantly in the weld zone.
Four experiments were performed with the same parameters: voltage level 15 kV, tube length 46
mm and rod diameter 17 mm. Grooves were machined in workpieces SD-CA-3.1, SD-CA-3.2 and
SD-CA-3.3 prior to welding. These three experiments are exactly the same, allowing a good
comparison with workpieces with a regular collar. No changes were made to the collar of workpiece
SD-CA-3.4. The experiments of the third series are listed in Table 6.4.
None of these workpieces was welded, and all of them failed the leak test. The workpieces with the
grooved collars showed significantly less leakage than the one with a regular collar. A discussion on
the jet material can be found further in this chapter.
118
Test
Number
Voltage
(kV)
Diameter
(mm)
Standoff
(mm)
Tube
Length
(mm)
Overlap
(mm)
Weld ?
Leak
Free?
Leak
Class
Remarks
SD-CA-3.1
15
17
2.5
46
8
No
No
B
Grooves
in collar
SD-CA-3.2
15
17
2.5
46
8
No
No
B
“
SD-CA-3.3
15
17
2.5
46
8
No
No
B
SD-CA-3.4
15
17
2.5
46
8
No
No
D
“
No
grooves
Table 6.4:Third series of copper-aluminium weld trails
6.6.5 Series 4
Based on the experience gathered in the previous three series of experiments, the goal of the fourth
series was to develop a weldability window for welding copper tubes to aluminium workpieces.
However at the time the workpieces were delivered, the damage of the field shaper was discovered
and thus this series of experiments has not yet been completed. For this series, 45 experiments were
planned combining three different stand-off distances, three energy levels and five different lengths
of the flyer tube. An overview is given in Table 6.5.
Stand-off (mm)
2
1,5
1
Voltage (kV)
15
not specified
maximum
15
not specified
maximum
15
not specified
maximum
Tube Length (mm)
46
46
46
46
46
46
46
46
46
47
47
47
47
47
47
47
47
47
48
48
48
48
48
48
48
48
48
49
49
49
49
49
49
49
49
49
50
50
50
50
50
50
50
50
50
Table 6.5: the fourth series of copper-aluminium experiments
Thus, for every combination of stand-off distance and flyer tube length, three experiments would be
conducted. The first experiment is performed at a voltage of 15 kV, the second one at the maximum
charging voltage of the pulse. This maximum value is not constant but depends on the workpieces. At
last, the third experiment is performed at a voltage which lies in between 15 kV and the maximum
voltage. This third voltage is not specified but should be chosen after investigation of the workpieces
which were welded with the above mentioned parameters.
Due to the great importance of these experiments, it is strongly recommended to conduct them in
the future.
119
6.7 Discussion on the Copper-Aluminium experiments
6.7.1 Non Destructive evaluation
6.7.1.1 Leak test
During the leak test, the workpieces are pressurised with air at 4 bar and submerged in water. The
classes indicating the severity of the leakage are listed in Table 6.1. The leak test setup is discussed in
the chapter on Non-Destructive Testing.
None of the workpieces of the first series of the copper-aluminium experiments were leak free. The
classes associated with the leakage for each workpiece can be found in Table 6.2.
The second series of Cu-Al experiments also did not produce welds of satisfying quality. Workpieces
SD-CA-2.4, SD-CA-2.8 and SD-CA-2.10 are the only workpieces that were welded partially: the tube
was welded to the rod at only one side. The leak test confirmed that the weld of these workpieces is
of higher quality than the other workpieces of the series. The leak severity of the three
aforementioned workpieces was determined to be respectively class C, B and C. The remaining
workpieces of the second series showed leakage of class C, D or even E. Although the leak rate of
class B is significantly better than class C or higher, this leak rate is for some applications still not
acceptable.
The experiments in the third series were executed to investigate the geometry of the collar and its
influence on the weld quality. Internal workpieces were used with grooves in the collar and with a
normal collar. The grooves create an exit for the jet, so the jet material does not get trapped in the
weld zone.
Grooves were machined in the collar of the inner rod in experiments SD-CA-3.1, SD-CA-3.2 and
SD-CA-3.3. These three workpieces showed only a few bubbles in the leak test (class B). Experiment
SD-CA-3.4 was performed with the exact same geometrical parameters and charging voltage, but
with a conventional collar (no grooves). This workpiece showed more than 10 bubbles per second
(class D). So, the experiments with the grooved collar showed significantly lower leakage than those
with a normal collar. The leak test performed on these workpieces is shown in Figure 6.14. None of
these workpieces were properly welded, as the internal workpieces spontaneously separated from
the outer tube on both sides after cutting the workpiece in half.
In all of the leak tests, the position of the leaks relative to the field shaper slit was examined. It was
suggested in [17] that a “buckling” effect occurred, due to the lower magnetic pressure in the field
shaper slit region. For symmetry reasons, the buckling would occur at positions FS slit + i 90˚ (i=0, 1,
2, 3). If the weld was locally interrupted at these locations, it would be likely that the leaks were
found at these positions. Sometimes leaks were indeed found at these, but not consistently.
The leaks generally showed great variety in their appearance. In some cases only one or two leaks
were observed, but the number of bubbles per second was very large. In other specimens, a large
number of small leaks (low leak rate per leak) were found. So, statistically there is no relation
between the position of the field shaper slit during welding and the location of the leaks [17].
120
Figure 6.14: Leak test performed on two workpieces from the third series of copper-aluminium experiments. On the left,
the workpiece with grooves in the collar shows limited leakage (class B). This is significantly less than observed for the
experiment with the exact same welding parameters but with a normal collar, which is shown on the right (class D).
6.7.1.2 Computed tomography
The welds of workpieces SD-CA-1.9 and SD-CA-1.15 were evaluated with the computed tomography
technique at CEWAC. The CT scan is a non-destructive testing method for producing 2-D and 3-D
cross-sectional images of an object from flat X-ray images. The internal structure of the workpiece
can be visualised, identifying internal defects. Weld flaws such as insufficient bonding (volumetrical
flaw) can be detected by the CT technique.
No significant weld defects were detected with the CT evaluation. Only the gap (filled with air)
located just behind the collar of the inner rod was detected, as shown in Figure 6.15 (left). This is of
course not a defect, but a consequence of the rod shape.
Workpiece SD-CA-1.15 showed a slight deformation at the welded zone, also shown in Figure 6.15
(right). The deformation results in a distortion of the roundness of the welded tube, but there was no
volumetrical flaw associated with this flaw. No defects were found in workpiece SD-CA-1.9.
Although no weld defects were found using the CT scan, neither of the workpieces were welded, as
could be concluded from microscopic inspection. As previously discussed, the combination of a high
charging voltage level and a large stand-off distance resulted in severe deformation of the aluminium
inner rod. The impact energy is assumed to be too high to result in a good weld.
The fact that the CT scan did not show any weld defects in workpieces SD-CA-1.9 and SD-CA-1.15
although there was no weld in either of the workpieces, could have several reasons.
A possible cause is that the CT scan did not have a sufficiently high resolution to detect the weld
flaws. This could be related to the scanning equipment or to the fact that the weld flaws are very
small. The flyer tube is pressed very firmly against the inner rod due to the high energy impact. Even
if no weld is formed, the surfaces will still be in contact with each other after the impact process. This
is in agreement with the deposition of aluminium on the inside surface of the copper tubes that was
121
found in most of the Cu-Al experiments, shown in Figure 6.10. The tube is very tightly connected to
the inner rod, but no weld is created. Thus, the flaws resulting from the MPW process have an
extremely small volume. The consequence is that they are very difficult to detect using the CT scan.
Figure 6.15: Computed Tomography is a non-destructive evaluation technique that can produce 2-D and 3-D cross
sectional images of the workpiece. The CT technique was applied to workpieces SD-CA-1.9 and SD-CA-1.15. No weld
defects were noticed. The left figure shows the air gap located behind the inner rod collar, which is not a weld defect. A
slight deformation (out of roundness) was noticed in the weld zone of workpiece SD-CA-1.15 (right picture). This
deformation did not result in a volumetrical flaw.
The UGCT (Ghent University Centre for X-ray Tomography) was contacted to acquire more
information about the CT evaluation technique. The equipment at UGCT cannot be used for the
MPW welded workpieces with diameter 25 mm. The necessary energy level of the CT equipment
increases as the thickness of the metal part (subjected to the evaluation) increases. The instruments
at UGCT are limited to metal workpieces with thickness lower than 10 mm.
122
6.7.2 Destructive evaluation
6.7.2.1 Weld strength
The torsion test was performed on workpiece SD-CA-2.2, as discussed in the section Destructive
Testing.
The grooves in the internal workpiece for the mounting of the torque wrench were machined prior to
welding. The aluminium inner rod failed before the weld did, at a torque of 140 Nm. Because only
the aluminium failed, the same workpiece could also be subjected to a compressive test after cutting
off the fractured part of the rod and flattening the surface. The weld sheared at a compressive force
of 12 kN. The displacement at this maximum force was 0,46 mm.
The flyer tube was most probably not fully welded to the inner rod, as shown in Figure 6.16. On one
side of the rod, a discolouration was found where the workpiece was presumably welded (left). The
other side of the rod had a rough surface but no indications were found that a weld had been formed
there. No wave pattern was observed on the surface. Although no perfect weld was formed, the
connection of the workpieces showed sufficient strength, as it could withstand axial forces up to
12 kN. During the impact, not only a weld is formed but also some mechanical interlock is created
which also contributes to the strength of the connection.
It is very difficult to determine an accurate value for the shear strength of the weld, because the
weld was not formed over the entire circumference. The welded zone (left in Figure 6.16) extended
over almost 180˚. The weld length in this welded zone varied between 2,5 and 4,5 mm (average=3,5
mm), and the diameter measured 17,5 mm. Using these values, a rough estimate of the shear
strength of the weld zone is 125 MPa.
Figure 6.16: Workpiece SD-CA-2.2 showed a discolouration over almost half the circumference (left), which was
presumably welded. The other half had a rough surface but no indications that a weld had been formed there. No wave
pattern was observed on the surface. The weld sheared in the compressive test at a force of 12kN.
According to [69] the ultimate shear strength for aluminium and copper alloys is approximately 65%
of the ultimate tensile strength. The minimum ultimate tensile stress of aluminium EN AW-6060 in
the T6 condition is 170 MPa and 290 Mpa for copper Cu-DHP R290. The ultimate shear strength of
these materials is then approximately 110,5 Mpa and 188,5 Mpa respectively. The strength of the
123
weld thus exceeds the strength of the aluminium base material, which was confirmed by the fact that
the aluminium rod failed first in the torsion test.
No additional destructive tests were performed on copper-aluminium workpieces to measure the
weld strength (or rather the strength of the connection). Other workpieces of the second series were
examined microscopically.
6.7.2.2 Microscopic examination of the weld interface
As stated above, in none of these copper-aluminium experiments, a fully developed weld was
created. However some workpieces remained attached after they had been cut through
longitudinally. This allowed microscopic examination of these workpieces. This investigation showed
that the workpieces were merely welded on one side of the part. Figure 6.17 shows a workpiece
(SD-CA-1.12) which has been embedded in epoxy resin and carefully polished afterwards. It is clear
that the lower interface on the image shows a gap between the workpieces, thus this side has not
been welded.
Figure 6.17: A workpiece embedded in epoxy resin, ready for microscopic investigation. It is clearly visible that the a slit
exists between the copper and aluminium at the lower part of the figure.
A microscopic investigation of workpiece SD-CA-1.12 was performed. Figure 6.18 shows a part of the
upper interface of Figure 6.17. It is clear that this interface has been welded and an intermetallic
layer was created. Figure 6.19 however shows the lower interface which is clearly not welded.
Although no weld was formed at this location, adhesion of aluminum to copper can be observed and
even some material mixing occurred. Some aluminium is still attached to the copper and even a small
amount managed to travel into the copper. This could mean either that the impact energy was not
sufficient to generate a good bond or that the impact energy was too large and some rebound of the
flyer tube occurred which broke the connection.
In both figures it can be seen that no wavy interface was formed. The phenomenon of wave
formation could not be found in any of the interfaces of these partially-welded parts. Occasionally,
on workpieces which had fallen apart after sectioning, few waves could be found on the
circumference of the aluminium inner part. However the number of these waves was very low and so
was their amplitude.
124
Figure 6.18: The welded side of part SD-CA-1.12. An intermetallic layer has been formed but no waves can be found in
this interface
Figure 6.19: Image of the non-welded side of workpiece SD-CA-1.12. Although no weld was formed, adhesion and mixing
occurred.
This leads to the conclusion that the parameters which were used in the three preliminary copperaluminium test series were outside the process window in which the formation of waves takes place.
Although they are not the same, the weldability window and the window for wave formation
partially overlap. Thus one might conclude that the values of the used parameters also did not fall
within the weldability window, thus explaining the low quality of the performed tests. It is important
to notice that these conclusions do not take in account the possible effect of de field shaper damage.
This once more emphasizes the need of further experiments on copper-aluminium connections
(including the fourth series of experiments) to obtain useful weldability windows.
6.7.2.3 Inner rod deformation
Significant differences were seen in the indentation of the inner rod. The plastic deformation of the
rod is caused by the impact of the flyer tube. A larger impact velocity will result in a larger
indentation. According to the simplified formulas by Pulsar, the acceleration is constant in time and
therefore, with s=stand-off distance:
lVFT\S ~ √]
(6.6)
125
If the deformation pressure is neglected, the full magnetic pressure causes the flyer tube to
accelerate. The acceleration is then proportional to the square of the magnetic field, which is
proportional to the discharge current. Because the discharge current is proportional to the charging
voltage:
lVFT\S ~√[ ~ j
(6.7)
The previous two relations show that the impact velocity is proportional to the root of the stand-off
distance and to the charging voltage.
The plastic deformation of the inner rod is related to the impact velocity. The impacting mass of the
flyer tube can be approximated as:
_ = i. (. x . ;. )
(6.8)
With: m = mass[kg]
R0 = flyer tube outer radius – flyer tube inner radius [mm]
t = flyer tube thickness[mm]
l = overlap length [mm]
ρ=density of the material [kg/mm³]
The part of the flyer tube that is accelerated has a length equal to the overlap length (on which the
pressure is exerted). The density of the flyer tube material, its outer radius and thickness are
constant in all experiments. Therefore, the mass is proportional to the overlap length.
The impact energy can be calculated as the kinetic energy of the flyer tube just before impact:
8lVFT\S =
$
_. lVFT\S
2
(6.9)
The impact energy is therefore proportional to the overlap length, to the stand-off distance and to
the square of the charging voltage.
8lVFT\S ~ . ]. j $
(6.10)
No detailed analysis of the deformation of the inner workpiece was found in literature. The energy
that is accumulated in the capacitor bank is partly dissipated in Joule-losses, partly in the tube
deformation and partly in kinetic energy of the tube. This kinetic energy is converted into
deformation of the rod at impact.
With the intention of verifying these formulas, the indentation of the inner rod was measured for ten
workpieces. The results are shown in Table 6.6. The relative diameter reduction is listed in the sixth
column and the absolute indentation in the seventh column. The first four experiments in the table
were performed at a higher voltage (18,5 kV and 19 kV) and an overlap length of 11 mm. As the
stand-off distance (s) decreases linearly, the indentation does not decrease in proportion. In fact, the
absolute indentation at s=3,0 mm is equal to the indentation at s=2,5 mm. At s=2,0 mm the
indentation is significantly lower, but at s=1,5mm the indentation increases again to the same value
as in the experiments at s=3,0 mm and s=2,5 mm.
126
For the next three experiments, the overlap length was also equal to 11 mm but the charging voltage
was 15 kV. The decrease in stand-off distance from s=3,0 mm to s=2,0 mm results in a proportional
decrease in indentation. But at s=1,5 mm, the indentation is equal to that of s=2,0 mm.
The final three experiments had an overlap length of 10 mm and a voltage of 15 kV. Again it was
observed that the indentation decreases with a decreases by reducing the stand-off distance from
s=2,5 mm to s=2,0 mm. But by further reducing the stand-off to s=1,5 mm, the indentation increased
again.
Workpiece
Voltage
(kV)
Standoff
(mm)
Overlap
(mm)
Diamete
r inner
rod
(mm)
%
Diameter
Reduction
Indentation
(mm)
Normalised
Impact
Velocity
Normalised
Impact
Energy
SD-CA-1.14
19,0
3,0
11
16
16
1,3
1,8
3,2
SD-CA-1.16
19,0
2,5
11
17
15
1,3
1,6
2,7
SD-CA-2.8
18,5
2,0
11
18
7
0,6
1,4
2,0
SD-CA-2.10
18,5
1,5
11
19
13
1,3
1,2
1,5
SD-CA-1.13
15,0
3,0
11
16
9
0,7
1,4
2,0
SD-CA-2.7
15,0
2,0
11
18
4
0,4
1,2
1,3
SD-CA-2.9
15,0
1,5
11
19
5
0,4
1,0
1,0
SD-CA-1.11
15,0
2,5
10
17
11
1,0
1,3
1,5
SD-CA-2.3
15,0
2,0
10
18
5
0,4
1,2
1,2
SD-CA-2.5
15,0
1,5
10
19
8
0,8
1,0
0,9
Table 6.6: The indentation of the inner rod was measured for ten workpieces. As the charging voltage increases, the
impact velocity and the indentation increase. According to the calculations by Pulsar, a larger stand-off distance should
also result in a larger indentation. Several irregularities were seen with respect to this relation.
It can be seen that for a similar overlap length and stand-off distance, the decrease in charging
voltage consistently results in a lower indentation. For example, a comparison between workpieces
SD-CA-1.14 and SD-CA-1.13 shows that decreasing the charging voltage from 19 kV to 15 kV, which
equals a reduction of 37% in impact energy according to equation 6.10, results in a decrease in
indentation of 46%. The same result can be seen in workpieces SD-CA-2.8 and SD-CA-2.7: 35%
decrease in impact energy, results in a 33% decrease in indentation. The indentation of workpiece
SD-CA-2.10 is very large, so the same comparison does not apply here.
To check the validity of the relations between these parameters, the impact velocity and the impact
energy were calculated for each experiment. Workpiece SD-CA-2.9 with parameters s=1,5 mm,
l=11 mm, and V=15 kV, was chosen as a reference. The impact velocity and energy for these
parameters were chosen as unity. The normalised impact velocity and impact energy, also shown in
Figure 6.20, were calculated according to equations 6.6, 6.7 and 6.10:
]
j
lVFT\S,Jx4V = ˆ
.
1.5__ 15žj
8lVFT\S,Jx4V =
. ]. j $
(11__)(1.5__)(15žj)$
(6.11)
(6.12)
127
The indentation of the inner workpiece is plotted against the normalised impact energy in Figure
6.20, along with a linear trend line. Although the results deviate from the trend line, a linear relation
can be recognized. As discussed before, increasing the charging voltage results in an increase in
indentation. The voltage only affects the magnitude of the acceleration. The effect of the stand-off
distance is not quite understood. Several irregularities were observed in the relation between the
stand-off and the indentation. This is most likely caused by the acceleration, which is not constant in
time. Therefore the assumed relation that the flyer tube accelerates linearly from standstill to the
impact velocity is not valid, and the impact velocity will not be proportional to the root of the standoff distance.
In reality, the acceleration depends largely on the deformation of the tube. The pressure exerted by
the magnetic field both deforms and accelerates the flyer tube. Neglecting the pressure required for
deformation probably has a significant influence on the accuracy of the suggested equations. The
deformation will absorb more energy when the stand-off distance increases (larger strains because
the final diameter is smaller). In addition, the Pulsar calculations for deformation simplify the flyer
tube geometry as shown on the left of Figure 6.21. The tube diameter reduces uniformly until it
impacts with the rod. The real deformation of the tube is shown on the right.
The pressure acts only on the right part of the tube. This side will accelerate, but the part of the tube
on the left side, where the collar is located, will not. This section of the tube will resist against the
deformation. The tube will impact first at its free end, and the collision will continue towards the left
side. As a result, the acceleration and impact velocity will be different at each point. The deformation
behaviour of the tube that occurs in reality is simply too complicated to be described with
elementary analytical equations.
A last remark concerning the indentation of the internal workpiece: In [23]it is suggested that the
yield strength of the internal workpiece material should exceed the strength of the flyer tube
material to prevent severe deformations. This can be confirmed by the observations in this work.
During this test series the yield strength of the cores (aluminium) was smaller than the strength of
the flyer tubes (copper). It is recommended to investigate the feasibility of copper-aluminium welds
with copper internal workpieces in the future.
128
Inner Rod Indentation [mm]
1,4
1,2
1,0
0,8
0,6
0,4
0,2
0,0
0,00
1,00
2,00
Normalised Impact Energy [-]
3,00
Figure 6.20: The indentation of the rod is plotted against normalised impact energy, along with a linear trend line.
Although the results deviate from the trend line, a linear relation can be recognized roughly. An increase in impact
energy results in a larger indentation.
Figure 6.21: The Pulsar calculations for deformation simplify the flyer tube geometry as shown on the left. The tube
diameter reduces uniformly until it impacts with the rod. The real deformation of the tube is shown on the right and is
much more complicated.
6.7.2.4 Jet
During the high velocity collision of the flyer tube onto the inner rod, high air pressures are created
at the adjacent surfaces. This pressure is sufficient to crumble a thin layer of metal (< 0,05 mm) from
the metal surfaces and eject it in the form of a jet. This jet consists of material of both the inner
workpiece and the flyer tube (oxide layer), and also removes contaminants. A minimum collision
angle is required to ensure a pressure of sufficient magnitude for the deformation of the metal
surfaces and the formation of the jet. Also a maximum value for the collision angle exists, after which
no jetting will take place and no bond will occur [17].
The jet mechanism is shown schematically in Figure 6.22. The magnetic pressure causes the impact
of the flyer tube, with the collision point moving towards the right hand side of the figure. The jet
moves away from the collision point, towards the area of lower pressure (located at the collar of the
129
inner rod). The collision angle is the angle between the inner rod surface and the flyer tube,
measured at the instantaneous collision point [19].
Figure 6.22: The magnetic pressure causes the acceleration of the flyer tube towards the inner rod and the weld is
formed during impact. The tube first impacts the rod on the left side. The collision point then moves towards the right
hand side of the figure. The high velocity impact creates an area of very high pressure, causing the formation of a jet.
This jet consists of material of both the inner workpiece and the flyer tube (thin surface layer), and also removes
contaminants.
In [17] it is stated that pre-cleaning of the workpieces is not required due to the cleaning effect of the
jet, which is capable of sufficiently removing the surface contaminants to ensure good joining of the
workpieces. In all of the experiments performed in this master thesis, the tubes and the rods were
cleaned thoroughly using acetone. It is a small effort to clean the parts before welding and it reduces
the amount of contaminating particles in the jet material. Cleaning can only result in similar or better
welds.
The movement of the jet is in the same direction as the contact point between the tube and the rod:
towards the collar. The collar blocks the jet, causing the jet material to accumulate at the bottom of
the collar. This effect was noticed in the both the copper-aluminium and copper-brass experiments.
In [17] experiments were performed with two different inner rod geometries, to investigate the
effect of the jet on the weld. The first geometry is similar to the one used in all of the experiments in
this thesis (with collar). In the second configuration the collar was removed, so the inner rod was
cylindrical. In the experiments with a collar on the inner rod, the jet was blocked by the collar.
Without the collar, the jet is no longer trapped between both workpieces. No difference in weld
characteristics was noticed between the two configurations. The fact that the welds were similar is
quite peculiar, as the collar influences not only the jet flow pattern, but also the collision angle during
welding.
To verify if the jet has a significant influence on the welding process, a different inner rod geometry
was applied in the CA 3-series experiments. The dimensions of the rod (diameter and collar) were
kept constant, but three axial grooves were machined in the collar prior to welding. The two
configurations result in a similar collision angle and impact velocity. The only difference is that in the
experiments using the grooved collar, the jet did not impact the collar. Due to the high pressure
difference, the jet material is pushed through the grooves. Figure 6.23 shows the jet accumulation at
the base of the collar of workpiece SD-CA-2.3 (right). Almost no jet material is found at the base of
the grooved collar of workpiece SD-CA-3.2, indicating that the jet can exit the collar region. In the
130
grooved collar experiments, the entire outer surface of the rod and inner surface of the flyer tube
were smoother and less contaminated after welding.
A portion of the jetting material, which impacts the collar at high velocity, is possibly reflected on the
collar. Another assumption is that not all jet material moves at the same velocity. The flow pattern of
the jet is not perfectly known. It is reported that most of the jetting material comes from the flyer
tube. This is caused by a higher jet velocity at the flyer tube surface than at the inner rod surface.
Some jet material, moving at a lower speed, might remain behind on the rod surface in the weld
zone. The thickness of the jet layer is larger than the height of the micro-roughness peaks of the base
material and is comparable with the amplitude of the interface wave profile. As a result, jet material
remaining behind in the weld zone could possibly interfere with the welding process [71][22].
The weld zone of workpiece SD-CA-2.3 is noticeably darker than that of workpiece SD-CA-3.2 (or any
other workpiece without grooves in the collar), as shown in Figure 6.24. This darker colour is jet
material, as the discolouration was not noticed in the grooved collar experiments. Neither of these
workpieces was welded together. The inner rod separated spontaneously from the outer tube after
cutting through the workpiece.
Figure 6.23: In the third series of copper-aluminium experiments grooves were machined in the collar of the inner rod.
These grooves create an exit path for the jet. With the standard rod geometry, the jet material accumulates at the base
of collar. In the grooved collar experiments, the entire outer surface of the rod and inner surface of the flyer tube were
smoother and less contaminated after welding.
Figure 6.24: The weld zone of workpiece SD-CA-2.3 is noticeably darker than that of workpiece SD-CA-3.2 (or any other
workpiece without grooves in the collar). This darker colour is jet material, as the discolouration was not noticed in the
grooved collar experiments.
131
6.7.3 Welding Parameters
6.7.3.1 Impact angle
In this section an approximation of the initial impact angle will be calculated for the three series of
experiments which were performed in this work. This angle will be in direct relation with the
stand-off distance. Assuming that the flyer tube will bend starting at the point where it enters the
field shaper, a right triangle will be created (Figure 6.25). Note that this theory is only an assumption.
The real movement of the flyer tube is of a complicated nature and can only be described accurately
by finite element simulations.
Figure 6.25: Assuming that the flyer tube bends over the collar, a right triangle is formed between the flyer plate and the
inner workpiece. This triangle can be used to calculate the angle of impact.
The stand-off distance is known for every experiment and the amount of overlap between the flyer
tube and the field shaper can be calculated from equation 6.5. The initial impact angle of the process
can thus be calculated using the tangent formula. These values are given in Table 6.7.
As explained in the chapter concerning the weld interface, the angle of impact should generally be
between 6° and 14°. It can be seen that most of the impact angles at the process start are situated
close to 14° or even exceed it. Since the angle of impact will grow during the process, also the parts
with an initial impact angle of 12° will soon have an impact angle which is situated outside the 6-14°
window.
In series 1 the initial angle only the last two experiments have an initial angle of impact which is
situated in the 6°-14° interval. The low quality of the welds can thus be explained by a excessive
impact angle.
Series 2 was performed with a smaller stand-off distance than series 1 and 3. Thus, the initial angle of
impact in these experiments was also smaller (Table 6.7). This can explain the fact that series 2
produced more partially-welded specimens than the other series.
It can be concluded that a stand-off distance which exceeding 2,5 mm is too large. Further
experiments to obtain useable weldability windows should thus be performed with a stand-off
132
distance of 2 mm or smaller. Note that one could use the calculation of the initial impact angle to
choose a combination of the stand-off distance and overlap length for future experiments.
Test
number
SD-CA-1.1
SD-CA-1.2
SD-CA-1.3
SD-CA-1.4
SD-CA-1.5
SD-CA-1.6
SD-CA-1.7
SD-CA-1.8
SD-CA-1.9
SD-CA-1.10
SD-CA-1.11
SD-CA-1.12
SD-CA-1.13
SD-CA-1.14
SD-CA-1.15
SD-CA-1.16
SD-CA-1.17
Stand-off
distance
[mm]
3,0
3,0
2,5
2,5
3,0
3,0
2,5
2,5
3,0
3,0
2,5
2,5
3,0
3,0
2,5
2,5
3,0
overlap Angle of
impact
[mm]
[°]
8
20,6
8
20,6
8
17,4
8
17,4
9
18,4
9
18,4
9
15,5
9
15,5
10
16,7
10
16,7
10
14,0
10
14,0
11
15,3
11
15,3
11
12,8
11
12,8
12
14,0
Test
number
SD-CA-1.18
SD-CA-1.19
SD-CA-1.20
SD-CA-2.1
SD-CA-2.2
SD-CA-2.3
SD-CA-2.4
SD-CA-2.5
SD-CA-2.6
SD-CA-2.7
SD-CA-2.8
SD-CA-2.9
SD-CA-2.10
SD-CA-3.1
SD-CA-3.2
SD-CA-3.3
SD-CA-3.4
Stand-off
distance
[mm]
3,0
2,5
2,5
2,0
2,0
2,0
2,0
1,5
1,5
2,0
2,0
1,5
1,5
2,5
2,5
2,5
2,5
overlap
[mm]
12
12
12
9
9
10
10
10
10
11
11
11
11
8
8
8
8
Angle of
impact
[°]
14,0
11,8
11,8
12,5
12,5
11,3
11,3
8,5
8,5
10,3
10,3
7,8
7,8
17,4
17,4
17,4
17,4
Table 6.7:The initial angle of impact of the process for every experiment in the copper-aluminium series.
6.7.3.2 Preliminary weldability windows
The objective of the experimental research was to obtain weldability windows for welding of the
desired material combinations. These are combinations of the ranges of parameters such as standoff distance, overlap length and charging voltage, which result in a successful weld. Because the
majority of the copper-aluminium welds were not successful, very few weldability windows could be
established.
Experiments with the same overlap length and charging voltage, but with a different stand-off
distance were compared. It should be noted that none of the workpieces were completely leak free.
But, there were some significant differences in the severity of the leakage between the workpieces.
Table 6.8 shows the leakage for experiments with a tube length of 48 mm, both for the voltage levels
of 15 kV and 18,5 kV. The leakage for experiments with a tube length of 49 mm are shown in Table
6.9, both for voltage levels of 15 kV and 18,5 kV. In reality, the maximum allowed voltage for the
second series was slightly lower than that of the first series of experiments. The experiments with a
stand-off of 3 and 2,5 mm were performed at 15 kV and 19 kV, and those with stand-off of 2 and 1,5
mm at voltages 15 kV and 18,5 kV. As can be seen from these tables, there was no consistent
relationship between the leakage of the connection and the stand-off distance. This means that a
larger stand-off distance neither improves nor worsens the leakage of the connection.
133
Workpiece
SD-CA-1.9
SD-CA-1.11
SD-CA-2.3
SD-CA-2.5
Workpiece
SD-CA-1.10
SD-CA-1.12
SD-CA-2.4
SD-CA-2.6
Overlap Length 10 mm
Voltage 15 kV
Stand-off
Leakage
distance[mm]
3,0
E
2,5
B
2,0
E
1,5
D
Voltage 18,5 kV
Stand-off
Leakage
distance[mm]
3,0
B
2,5
C
2,0
C
1,5
C
Bond
No
No
No
No
Bond
No
Partially
Partially
No
Table 6.8: The results of the leak test are shown for experiments with a tube length of 48 mm (overlap length of 10 mm),
both for voltage levels 15 kV and 18,5 kV, but with different stand-off distances. No consistent relationship between the
leakage of the connection and the stand-off distance was seen. The right column indicates whether there was a
connection between the tube and the rod: the workpiece is labeled “No” if the tube separated from the rod after cutting
it in half, and “Partially” if the tube was welded at one side of each half after the cutting operation.
Overlap Length 11 mm
Voltage 15 kV
Workpiece
SD-CA-1.13
SD-CA-1.15
SD-CA-2.7
SD-CA-2.9
Stand-off
distance[mm]
3,0
2,5
2,0
1.5
Voltage 18,5 kV
Leakage
Bond
D
B
E
D
No
No
No
No
Workpiece
Stand-off
distance[mm]
Leakage
Bond
SD-CA-1.14
3,0
D
No
SD-CA-1.16
2,5
B
No
SD-CA-2.8
SD-CA-2.10
2
1,5
B
C
Partially
Partially
Table 6.9: Experiments with similar parameters as in the previous table, only now with a tube length of 49 mm (overlap
length of 11 mm). The results of the leak test indicate that there is no consistent relation between the stand-off distance
and the leak rate of the bond.
The occurrence of a bond is also added to Table 6.8 and Table 6.9. None of the workpieces
performed at 15 kV were welded. None of the workpieces with a stand-off distance of 3 mm were
welded. Only workpiece SD-CA-1.12 (stand-off 2,5mm) was partially-welded. Other partially-welded
workpieces were SD-CA-2.4, SD-CA-2.8 and SD-CA-2.10. So, three out of four experiments with the
lower stand-off distance (2 or 1,5 mm) and the maximum allowed voltage (18,5 kV) were partiallyconnected.
These results clearly indicate that the combination of a small stand-off distance and a high charging
voltage result in a better connection. In addition, the leakage found in the experiments of the second
series at a voltage of 18,5 kV was consistently lower than at 15 kV.
134
Preliminary weldability windows were generated based on these results. The parameter
combinations (voltage, stand-off distance) are plotted for each overlap length. Because the only
partially-successful experiments were found for an overlap length of 10 mm and 11 mm, only these
graphs are shown. The weldability window for an overlap length of 10 mm is shown in Figure 6.26.
The triangles mark the combinations of parameters (voltage, stand-off distance) that resulted in an
unsuccessful weld. The square marks the experiment for this overlap length that was partiallywelded: SD-CA-2.4. Figure 6.27 shows the weldability window for an overlap length of 11 mm.
Experiments SD-CA-2.8 and SD-CA-2.10 were partially-welded, while all the other welds at this
overlap were not successful.
It should be noted that experiment SD-CA-2.2, which was performed with an overlap of 9 mm, a
stand-off distance of 2 mm and voltage 18 kV, was not cut longitudinally. It was examined
destructively with both the torsion test and the compressive test. The weld had a sufficient strength,
as discussed in the section Weld Strength.
Without proof, it is very well possible that all the partially-welded workpieces (that were only welded
at one side) would also have shown significant strength if they had been tested destructively. The
destructive tests to determine the weld strength induce shear stresses in the weld zone. Even if the
workpiece is not fully welded, the indentation of the inner tube causes the tube and the rod to be
connected tightly, which results in good shear strength properties. However, when cutting the
workpiece in half, the residual stresses in the flyer tube caused by compression to a much smaller
diameter are released. These residual circumferential stresses are high, as the diameter reduction is
significant: with a stand-off distance of 2 mm, the inner diameter of the flyer tube is reduced from 22
to 18 mm, or even less if the rod indentation is taken into account. This is a reduction of more than
18%. So, the cutting operation might cause the weld to separate more easily than the shear stresses
from the destructive weld strength tests. But the cutting must be performed in order to examine the
weld microscopically. Note that it is also it is possible that the workpieces are pressed against each
other without the formation of a weld. In this case friction between the workpieces is able to carry a
certain load and hence this low quality specimen could for example pass the torsion test.
It can be concluded from both these weldability windows that a charging voltage of 18.5kV and an
overlap of 1.5mm or 2mm show the best welds (although these were only partially-connected).
Obviously, more experiments should be performed, at similar or even lower stand-off distances. Also,
a series of experiments voltage levels in the range between 15kV and 19kV should be performed.
135
Figure 6.26: The weldability window shows the experiments performed with an overlap length of 10 mm. The triangles
mark the combinations of parameters (voltage, stand-off distance) that resulted in an unsuccessful weld. The square
marks the experiment that was partially-welded.
Figure 6.27: The weldability window for an overlap length of 11mm shows that experiments SD-CA-2.8 and SD-CA-2.10
were partially-connected, while all the other welds at this overlap were not successful.
136
6.8 Copper-Brass experiments: Test series 1
The experiments performed in this thesis on the material combination copper-brass extend the
research performed by dr.ir. K. Faes of the Belgian Welding Institute. A short review of his
experiments will be given in this section. The experiments conducted within the frame of this master
thesis are discussed in § 6.9.
The MPW experiments, with copper flyer tubes and brass internal workpieces were performed using
the same equipment as in this master thesis. The flyer tube had a wall thickness of 1,5 mm and an
outer diameter of 25 mm. By varying the tube length between 46 and 51 mm, the overlap length
varied accordingly from 8 to 13 mm. The internal workpiece diameter at the weld zone was varied
between 17 and 20,5 mm, which results in stand-off distances between 0,75 mm and 2,5 mm.
Different combinations of the stand-off distance and the overlap length were tested for several
voltage levels. In total , 126 experiments were performed (series 1 and 2), which can be found in
Appendix B.
These experiments in series 1 can be divided into five series, each performed with a different
stand-off distance. In all series the overlap length and the energy level was varied.
6.8.1 Test series 1.1: Stand-off distance = 0,75 mm
These experiments were performed with a stand-off distance of 0,75 mm. With a flyer tube length of
50,0 mm, 48,0 mm and 46,0 mm, the charging voltage was chosen equal to 12 kV, 15 kV and 18 kV.
This results in a total of 9 experiments. None of the workpieces which were produced during this
series resulted in a good weld. This leads to the conclusion that a stand-off distance of 0,75 mm is
too small to obtain a high weld quality (see Figure 6.28).
Copper - Brass : stand-off distance = 0,75 mm
No weld
Weld
Length Flyer Tube [mm]
52,0
51,0
50,0
49,0
48,0
47,0
46,0
45,0
9
10
11
12
13
14
15
16
17
18
19
20
21
Charging voltage (kV)
Figure 6.28: The weldability window of the first series of copper-brass. No good welds were created during this series.
6.8.2 Test series 1.2: Stand-off distance = 1,0 mm
This series of weld trails contains twenty experiments with a stand-off distance of 1 mm. Again both
the voltage and the length of the flyer tube were varied, respectively at 12 ,15 ,18 and 20 kV and 51,
50, 48 and 46 mm. After microscopic investigation, it appeared that only four of the twenty
experiments resulted in a good weld, resulting in the following weldability window (Figure 6.29).
Note that three out of four good welds were performed with a flyer tube length of 48 mm (and thus
137
an overlap length of 10 mm) . Another five workpieces were partially welded. Since these welds
cannot be qualified as qualitative welds, they are also described as “No Weld” in the weldability
window.
No weld
Weld
Copper - Brass: stand-off distance = 1,0 mm
Flyer Tube Length [mm]
52,0
51,0
50,0
49,0
48,0
47,0
46,0
45,0
9
10
11
12
13
14
15
16
17
18
19
20
21
Charging voltage (kV)
Figure 6.29:The weldability window of the second series of copper-brass welds. Note that most of the good welds were
performed with a flyer tube length of 48 mm (overlap length = 10 mm).
6.8.3 Test series 1.3: Stand-off distance = 1,5 mm
During this series, a stand-off distance of 1,5 mm was used with varying flyer tube length and
changing voltages. In this series seven out of twelve experiments resulted in good welds and two
workpieces were only partially welded. The weldability window of this series is shown in Figure 6.30.
Note that all experiments executed with a voltage higher than 18 kV resulted in a good weld. At
18 kV two experiments resulted in a weld and two did not. It can be concluded that with a stand-off
distance of 1,5 mm, the transition zone is positioned around 18 kV.
Copper - Brass: stand-off distance = 1,5 mm
No weld
Flyer Tube Length [mm]
52,0
51,0
50,0
49,0
48,0
47,0
46,0
45,0
9
10
11
12
13
14
15
16
17
18
19
20
21
Charging voltage (kV)
Figure 6.30: The weldability window of the third series of copper-brass welds. All the tests which were performed at a
charging voltage greater than 18 kV resulted in a good weld.
138
6.8.4 Test series 1.4: Stand-off distance = 2,0 mm
This series of experiments uses again a fixed stand-off distance (=2,0 mm). The voltage and the flyer
tube length were varied. Only two experiments resulted in a good weld. This shows that a stand-off
distance of 2,0 mm is probably too large. The weldability window is shown in Figure 6.31.
No weld
Weld
Copper - Brass: stand-off distance = 2,0 mm
Flyer Tube Length [mm]
52,0
51,0
50,0
49,0
48,0
47,0
46,0
45,0
9
10
11
12
13
14
15
16
17
18
19
20
21
Charging voltage (kV)
Figure 6.31: The weldability window of the fourth series of copper-brass welds. Note that only two experiments resulted
in a high-quality weld. This shows that a stand-off distance of 2,0 mm is probably too large.
6.8.5 Test series 1.5: Stand-off distance = 2,5 mm
This last series of copper-brass experiments which were performed by K. Faes uses a stand-off
distance of 2,5 mm. None of the eight experiments lead to a good weld. Four of the eight specimens
were welded only partially and the other four were not welded at all. It can thus be concluded that a
stand-off distance of 2,5 mm is certainly too large for magnetic pulse welding this material
combination (Figure 6.32).
No weld
Copper - Brass: stand-off distance = 2,5 mm
Weld
Flyer Tube Length [mm]
52,0
51,0
50,0
49,0
48,0
47,0
46,0
45,0
9
10
11
12
13
14
15
16
17
18
19
20
21
Charging voltage (kV)
Figure 6.32: The weldability window of the fifth series of copper-brass welds. None of the experiments resulted in a
weld, proving that a stand-off distance of 2,5 mm is too large.
139
6.9 Copper-Brass experiments: Test series 2
CuMs
Material
Flyer tube
Internal
Workpiece
Copper
Outer
Diameter
25 mm
Brass
18 mm/19 mm
Thickness
1,5 mm
/
Several intervals of parameter combinations (overlap length, stand-off distance, and charging
voltage) of the previous experiments resulted in successful welds. The objective of the experiments
of the next series was to refine the weldability windows for this material combination.
The experiments which were described in the previous section (§ 6.8) showed that most of the highquality welds were created when using a flyer tube length of 48 mm (overlap length = 10 mm).
During the following experiments the length of the flyer tube will be fixed to this value. The
parameter which will be changed during these test is the charging voltage. It will be changed in small
steps to determine the transition between a good and a low-quality weld.
6.9.1 Test series 2.1
6.9.1.1 Introduction
Table 6.10 shows a small series of experiments which were selected from § 6.8.3. During these
experiments only the voltage was varied from 15 to 20 kV. All the other parameters were held
constant. It can be seen that the first weld is created at a voltage of 18 kV. However, the first
experiment which was performed with 18 kV did not create a weld.
SD-CuMs-1.37
Voltage
(mm)
15,0
Diameter
(mm)
19,0
Tube Length
(mm)
48
Stand-off
(mm)
1,5
Overlap
(mm)
10,0
SD-CuMs-1.38
18,0
19,0
48
1,5
10,0
No
SD-CuMs-1.76
18,0
19,0
48
1,5
10,0
Yes
SD-CuMs-1.77
19,0
19,0
48
1,5
10,0
Yes
SD-CuMs-1.39
20,0
19,0
48
1,5
10,0
Yes
Workpiece
Weld?
No
Table 6.10: A selection of previous experiments of § 6.8.3
The experiments which are described in this section use the same parameters as before but will be
performed for more voltage levels. Also the interval between the voltages will be reduced to 0,5 kV.
The goal is to find the transition voltage at which a weld is created and thus to obtain a more
detailed weldability window. The experiments are listed in Table 6.11.
The workpieces were evaluated using two testing methods: leak testing (§5.2.1) and microscopic
investigation. The first weld which passes both tests was created at 18,5 kV. Note that workpieces
SD-CuMs-1.92 up to 1.101 were welded with the same voltage. These identical experiments will be
further described in the next section (§ 6.9.2). Weldability windows will be discussed in § 6.10.3.
140
Workpiece
Voltage
(kV)
Diameter
(mm)
Stand-off
(mm)
Tube
Length
(mm)
Leak
Free?
Leak
Class
Weld?
SD-CuMs-1.82
14,0
19,0
1,5
48,0
No
D
No
SD-CuMs-1.83
SD-CuMs-1.84
14,5
15,0
19,0
19,0
1,5
1,5
48,0
48,0
No
No
D
C
No
No
SD-CuMs-1.85
15,5
19,0
1,5
48,0
No
C
No
SD-CuMs-1.86
SD-CuMs-1.87
16,0
16,5
19,0
19,0
1,5
1,5
48,0
48,0
No
No
B
B
Yes
No
SD-CuMs-1.88
SD-CuMs-1.89
17,0
17,5
19,0
19,0
1,5
1,5
48,0
48,0
No
No
B
B
Partially
Partially
SD-CuMs-1.90
SD-CuMs-1.91
18,0
18,5
19,0
19,0
1,5
1,5
48,0
48,0
No
Yes
B
A
Yes
Yes
SD-CuMs-1.92
SD-CuMs-1.93
SD-CuMs-1.94
SD-CuMs-1.95
SD-CuMs-1.96
SD-CuMs-1.97
19,0
19,0
19,0
19,0
19,0
19,0
19,0
19,0
19,0
19,0
19,0
19,0
1,5
1,5
1,5
1,5
1,5
1,5
48,0
48,0
48,0
48,0
48,0
48,0
Yes
Yes
No
Yes
No
Yes
A
A
B
A
B
A
Yes
Yes
Yes
Yes
|
Yes
SD-CuMs-1.98
SD-CuMs-1.99
SD-CuMs-1.100
SD-CuMs-1.101
19,0
19,0
19,0
19,0
19,0
19,0
19,0
19,0
1,5
1,5
1,5
1,5
48,0
48,0
48,0
48,0
Yes
Yes
Yes
Yes
A
A
A
A
|
|
|
|
SD-CuMs-1.102
SD-CuMs-1.103
19,5
20,0
19,0
19,0
1,5
1,5
48,0
48,0
Yes
Yes
A
A
Yes
Yes
Table 6.11: Experiments of test series 2.1; the voltage is varied. The values of the other parameters were based on results
of previous experiments (§ 6.8). The goal is to find the transition zone of the voltage at which a weld is created.
6.9.1.2 Non-destructive testing
A leak test was performed on every workpiece of this series. As stated in Table 6.11 the minimum
charging voltage to obtain a leak-free weld was equal to 18,5 kV. Experiments which were performed
with a lower voltage always showed leaks in the weld zone.
In the experiments with a voltage higher than 18,5 kV, only two welds showed small leaks. Since
these two experiments are a part of the reproducibility experiments, they will be described in § 6.9.2.
6.9.1.3 Destructive testing
Most of the workpieces in this series were examined by microscopic investigation. The results of this
examination are given in Table 6.11 in the column “Weld?”. Some welded specimens from the
reproducibility series (SD-CuMs-1.92 up to 1.101) were evaluated with other methods (torsion test,
compressive test). These results will therefore be described in § 6.9.2.
The microscopic investigation of the workpieces was sometimes contradicting the results of the leak
test. Workpieces SD-CuMs-1.86 and 1.90 appeared to be welded when examined microscopically
(Figure 6.33), while the workpiece did not pass the leak test. Figure 6.34 shows an image of a leak141
free weld (workpiece SD-CuMs-1.91). No significant difference can be observed between the
interface of a leak-free weld and a weld with leaks. This leads to the conclusion that microscopic
investigation is not able to evaluate the weld quality with absolute certainty. Microscopic
investigation can be considered as a local weld quality evaluation, while the leak test is a global weld
quality evaluation.
Figure 6.33: An image of the weld in workpiece SD-CuMs-1.86. This interface with typical wave pattern appears to be
welded although the leak test showed a significant leak in this test specimen.
Figure 6.34: An image of a leak free weld (workpiece SD-CuMs-1.91): it shows no significant differences with the image of
a leaking weld (Figure 6.33)
During the microscopic investigation, also the weld lengths were measured for several workpieces.
These values can be found in Table 6.12. It can be noted that the weld length in a workpiece is not
the same on two positions of the circumference. An average of both weld lengths is calculated to
enable a quick comparison between different workpieces.
Workpiece
SD-CuMs-1.86
SD-CuMs-1.91
SD-CuMs-1.92
SD-CuMs-1.93
SD-CuMs-1.94
SD-CuMs-1.95
SD-CuMs-1.97
SD-CuMs-1.102
SD-CuMs-1.103
Voltage Weld length Weld length Average weld
[kV]
Side 1 [mm] Side 2 [mm] length [mm]
16,0
18,5
19,0
19,0
19,0
19,0
19,0
19,5
20,0
3,90
3,90
1,13
9,76
1,59
3,82
7,22
5,62
4,95
4,00
6,95
7,55
3,85
6,80
4,48
2,27
4,44
2,80
3,95
5,43
4,34
6,81
4,20
4,15
4,75
5,03
3,88
Table 6.12: The weld length for several workpieces. Since the weld length differs on both sides of the weld, the average
weld length is also calculated to enable a quick comparison between different workpieces.
142
Although workpieces SD-CuMs-1.92 up to 1.97 were welded with the same voltage, there are great
differences in their weld lengths (see Table 6.12). To investigate the influence of the charging voltage
on the weld length, the average weld length produced with 19,0 kV was calculated (4,85 mm). Figure
6.35 shows the average weld length as a function of the voltage. No direct relation between the
voltage and the weld length can be found. All values are scattered from approximately 4,0 to 5,4
mm.
6,00
Weld Length [mm]
5,00
4,00
3,00
2,00
1,00
0,00
15,0
16,0
17,0
18,0
19,0
20,0
Charging Voltage [kV]
Figure 6.35 : A plot of the weld length versus the voltage. Although one would expect an increasing weld length with an
increasing voltage, no direct relation can be seen in this plot.
6.9.2 Series 2.2: Subset of series 2.1
6.9.2.1 Introduction
In order to investigate the reproducibility of the MPW process, a series of ten repeat tests was
executed for the material combination copper-brass. As can be seen in Table 6.11, experiments
SD-CuMs-1.92 up to SD-CuMs-1.101 were performed with the same welding parameters. These
parameters were chosen based on previous experiments. It was observed that the weld
SD-CuMs-1.77 was of high quality. Using the same geometrical parameters but a lower charging
voltage did not render a successful weld. Increasing the charging voltage up to 20 kV resulted in local
melt pockets, which is detrimental for the weld quality.
The parameters of experiment SD-CuMs-1.77 are listed in Table 6.13. The same parameters were
used for the 10 experiments to verify the reproducibility.
Welding parameters for reproducibility experiments
Flyer tube thickness
1,5
mm
Inner rod diameter
19,0
mm
Stand-off distance
1,5
mm
Flyer tube length
48,0
mm
Overlap length
10,0
mm
Charging Voltage
19,0
kV
Table 6.13: Welding parameters used in the reproducibility experiments
143
All workpieces were first subjected to leak testing using pressurised air. Only eight 8 of 10 were
found to be leak free, as SD-CuMs-1.94 and SD-CuMs-1.96 failed the leak test. Both workpieces
showed two small leaks (Class B).
This result of 80% reproducibility is not satisfying. In order to be an economically feasible production
process, a reproducibility of at least 98% or more is required. However, after the experiments it was
observed that the field shaper was damaged, which can be the explanation for the low
reproducibility (see §6.5). In future research, a similar series of reproducibility tests should be
performed, using a field shaper in good condition.
The workpieces SD-CuMs-1.96 and SD-CuMs-1.98 up to 1.101 were first subjected to a helium leak
test at CEWAC, and afterwards evaluated destructively to determine the weld strength. Torsion
testing was performed on SD-CuMs-1.99 and compressive testing on the other workpieces. For the
welds SD-CuMs-1.92 up to 1.95 and SD-CuMs-1.97 a longitudinal cross section was made and
microscopic inspection was performed. Table 6.14 gives an overview of both the results of the air
leak test, and further evaluation methods applied to the workpieces.
Workpiece
Standard
leak test
SD-CuMs-1.92
A
SD-CuMs-1.93
A
SD-CuMs-1.94
B
SD-CuMs-1.95
A
SD-CuMs-1.96
B
SD-CuMs-1.97
A
SD-CuMs-1.98
A
SD-CuMs-1.99
A
SD-CuMs-1.100
A
SD-CuMs-1.101
A
Leak class
(pressurised air)
Helium leak test &
Compressive test
Microscopic
evaluation
Table 6.14: Leak test results and further evaluation methods applied to the reproducibility experiments.
6.9.2.2 Non-destructive testing
In the helium leak test, the welded zone is exposed to a pressure difference: pressurised helium
tracer gas will flow through a defective weld from the high-pressure side to the low-pressure side,
where a detector is mounted. The detector measures the concentration of helium, indicating the
severity of the leak.
The use of helium to examine the weld for leakage has the advantage over air that it is possible to
detect smaller leaks. The ability of helium testing to detect very low leak rates results from the
relatively small molecular structure of helium, which allows the gas to pass easily through pores that
would block larger molecules of most other air component gases, such as oxygen and nitrogen.
However, its higher cost usually restricts it to specialised applications where those advantages are
essential [72].
144
The test using pressurised air in water, also referred to as the “bubble test”, is not able to detect very
small leaks and/or leak rates. In addition, correlation to other leak test methods is very difficult, as
the mechanisms of air transport into water and bubble formation as well as other sources of air
bubbles (leaks from fixtures, bubbles from air trapped on surfaces, etc.) are biased and prevent
reliable leak detections. Nevertheless, the custom air leak test can render a qualitative measure for
the leak-tightness of the welded tubes [73].
Leak rates are defined as the product of the volume of the detected gas and its pressure, per unit of
time. Gases are compressible, so the same leak volume at a higher pressure implies a greater number
of leaked particles. For example, a leak into atmosphere of 1 mbar.l/s is equivalent to a volume leak
of 1 cm3 per second [74] .
The results of the helium leak tests are shown Table 6.15. The table also shows the leak class from
the air leak test for comparison.
Workpiece
Helium leakage [mbar.l/s]
SD-CuMs-1.96
7,5 . 10
SD-CuMs-1.98
-5
B
-10
A
3,3 . 10
-10
A
-8
A
-8
A
SD-CuMs-1.99
5 . 10
SD-CuMs-1.100
2 . 10
SD-CuMs-1.101
Leak class
(pressurised air)
2 . 10
Table 6.15: Results from helium and air leak test
In the air leak testing only SD-CuMs-1.96 showed leakage. The helium test on this workpiece resulted
in a leak of 7,5 . 10-5 mbar.l/s. The other 4 workpieces passed the air leak test, but showed a leak
between 2 . 10-8 and 3,3 . 10-10 mbar.l/s. Despite the fact that the same parameters were used in all
the welding experiments, a large difference in leakage flow is observed.
These results indicate that the air leak test is able to detect leak rates larger than an equivalent of 106
to 10-7 mbar.l/s in the helium leak test.
In [70], it is stated that MPW joints with a helium leak rate in the range 10-9 mbar.l/s have been
achieved. This small leak rate was accomplished by optimisation of welding parameters. Two
important improvements are responsible for the reduction in leak rate. First of all, the inner rod was
fabricated with a conical end, as shown in Figure 6.36. From several series of experiments it was seen
that an increase in cone angle resulted in smaller leak rates. The reason for this was the increase in
weld quality caused by the slightly smaller collision angle. The second improvement was to polish the
surface of the rod after machining to its final dimension. This measure was taken to limit the passage
of helium through the joint due to the serrated surface profile left by the machining process.
Although these measures might improve the leak-tightness of the welds, it is also reported in [70]
that further improvement is needed to establish an acceptable grade of repeatability.
Another method to reduce or eliminate leaks in the MPW-joints is suggested on the website of
Magneform [75]. They suggest to replace the magnetic pulse weld by a magnetic pulse crimp joint.
Also an ‘O’-ring is placed in a groove in the inner rod before welding, as shown in Figure 6.37.
145
Figure 6.36: Conical inner rod to decrease the angle of collision
Figure 6.37: ‘O’-ring placed in a groove in the inner rod to eliminate leakage (left), and knurl pattern on the inner rod
surface to improve the torsion strength (right)
The flyer tube presses the o-ring against the internal workpiece, creating a leak proof connection. In
contrast to the idea of reducing the surface roughness to obtain better leak proof properties in [70],
it is suggested in [75] that a rougher knurl pattern on the inner rod surface combined with the O-ring
is a better combination. The knurl pattern should result in a stronger connection in torsion, while the
O-ring guarantees leak-tightness.
A ‘textured’ inner rod surface (knurl or screw thread) was also used in [76], to improve the torsion
strength of electromagnetic crimped connections. This method is not so useful for the welding
experiments, as the torsion strength of the tested welds already exceeds the strength of the base
materials. However, it could be used in future research on magnetic pulse crimping (for material
combinations that are difficult to weld).
146
6.9.2.3 Destructive testing
The 5 workpieces that were subjected to the helium leak test were afterwards evaluated
destructively to determine the weld strength. Torsion and compressive tests were performed. The
other 5 workpieces of the series were inspected microscopically.
Torsion test
Torsion testing was performed on weld SD-CuMs-1.99. As discussed in the destructive weld
evaluation methods (Chapter 5), some difficulties were encountered regarding the clamping of the
workpiece (as a result of the high applied torque). The tube was ultimately clamped in the bench
screw with curved clamping plates. The brass rod failed at a torque of 280 Nm, indicating that the
shear strength of the weld exceeds the strength of the brass base material.
Compressive test
Compressive testing was performed on weld SD-CuMs-1.101. The workpiece failed at a compressive
force of 25,1 kN. At this force the copper tube buckled. The fact that the copper tube fails before the
weld indicates that the weld’s shear strength exceeds the buckling resistance of the copper tube.
The results of the compressive test are discussed in more detail further in this chapter (§6.10.1.2)
Microscopic examination
As stated above, the workpieces SD-CuMs-1.92 up to 1.95 and 1.97 were cross-sectioned, embedded
and subjected to microscopic investigation. Examination suggested that all parts were properly
welded. The following figures show the typical weld interface (Figure 6.38 to Figure 6.41); a wavy
pattern was observed at every weld interface.
Although weld SD-CuMs-1.94 did not pass the air leak test, no weld defects were found during
microscopic examination (Figure 6.39). This clearly indicates that welds must be evaluated by both
methods to obtain both a local and a global evaluation of the weld quality.
The images on Figure 6.40 and Figure 6.41 show both sides of a weld. It can be observed that the
weld interfaces are very different. The amplitudes and the wavelengths of the waves on side 1 are a
lot smaller than those on side 2.
Figure 6.38: Weld SD-CuMs-1.93; unetched condition
147
Figure 6.39:Weld SD-CuMs-1.94; unetched condition
Figure 6.40: Weld SD-CuMs-1.95, side 1; unetched condition
Figure 6.41: Weld SD-CuMs-1.95, side 2; unetched condition
148
By breaking the weld after cross-sectioning, it was observed that the weld length varies around the
circumference. Evidence of this is shown in Figure 6.42. The wave pattern, which can be seen without
magnification, does not extend over the same length in every point of the circumference. This
indicates that the MPW process did not create a weld which is uniform throughout the entire
circumference. When the pieces are cross-sectioned, just one location is investigated. It is possible
that in another location on the circumference a weld defect is present, which could be the cause of a
leak.
Figure 6.42: By breaking the weld after cross-sectioning, it was observed that the weld length varies around the
circumference. The wave pattern, which can be seen without magnification, does not extend over the same length at
every location.
Interruptions of the weld pattern were found at the position 180˚ relative to the field shaper slit,
caused by the damaged field shaper. The leak that was observed during the air leak test was however
not located at this position.
This can also explain the difference in measured weld lengths for workpieces which were welded
with identical parameters (§ 6.9.1.3). The measured weld length can differ significantly depending
on the position at which the workpiece was cut through and examined microscopically.
149
6.9.3 Series 2.3: Grooved collar
6.9.3.1 Introduction
This series of experiments will repeat the reproducibility experiments of test series 2.2 but now the
internal pieces will have a grooved collar. These experiments will investigate the influence of the
grooved collar on the reproducibility and whether the 80% yield of the former section can be
improved. The experiments and the results are listed in Table 6.16. Note that no leak-free welds
were formed.
Workpiece
Voltage
(kV)
Diameter
(mm)
SD-CuMs-1.117
SD-CuMs-1.118
SD-CuMs-1.119
SD-CuMs-1.120
SD-CuMs-1.121
SD-CuMs-1.122
SD-CuMs-1.123
SD-CuMs-1.124
SD-CuMs-1.125
SD-CuMs-1.126
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
Stand-off Tube length
(mm)
(mm)
1,5
1,5
1,5
1,5
1,5
1,5
1,5
1,5
1,5
1,5
48
48
48
48
48
48
48
48
48
48
Leak
free?
Leak
class
No
No
No
No
No
No
No
No
No
No
B
B
B
B
B
B
B
B
B
B
Weld? Remarks
Grooves
No
in collar
No
“
No
“
No
“
No
“
No
“
No
“
No
“
No
“
No
“
Table 6.16: A listing of the experiments which will be discussed in this section. It is a repetition of the experiments in test
series 2.2 but the internal pieces have a grooved collar. Hence, the influence of the grooved collar on the reproducibility
can be investigated.
The principle of § 6.6.4 was repeated: a grooved collar was chosen to create a path through which
the jet can leave the interface. Figure 6.43 shows the difference between a workpiece with and
without a grooved collar. It can be seen that significantly less jetting material is present on the
workpiece with the grooved collar.
Figure 6.43: An image of two brass internal pieces. The left workpiece has a collar with grooves and the right workpiece
has no grooves. The grooved workpiece shows significantly less jetting material at its collar. On the right workpiece a
large quantity of jetting material is visible ( the dark zone at the collar).
150
6.9.3.2 Non-destructive testing
All the workpieces of this series were evaluated by a leak test. None of the workpieces was
completely leak-free. Hence, the reproducibility did not improve at all. Note that 80% of the
specimens without collar grooves were leak-free. It appears that a grooved collar deteriorates the
leaking behaviour.
The results of this leak test are in contradiction to the results in § 6.6.4. Unlike the copper-brass
experiments, the copper-aluminium experiments with grooved collar did show an improvement. The
grooved copper-aluminium pieces were granted leak class B while the one without the grooves was
class D.
6.9.3.3 Destructive testing
Compressive test
Workpieces SD-CuMs-1.20, 1.123 and 1.124 were examined by compressive testing. Due to the fact
that those workpieces were welded with identical process parameters, they should perform similar
in this test.
Indeed, the results of this compressive test were similar for the three workpieces. The welds did not
fail during the test and the maximum axial force that was reached was about 25 kN for all three
workpieces. At 25 kN buckling of the copper flyer tube started and thus the weld was stronger than
the copper base material buckling strength. The welds have thus passed the test. The results of this
test can be found in § 6.10.1.2.
Microscopic examination
Most of the workpieces were cut longitudinally for the microscopic evaluation. However, none of the
workpieces remained attached after the cutting operation. This led to the conclusion that no highquality welds were created during these experiments. Again it appears that the grooved collar
deteriorates the weld quality.
Once again this shows that two different evaluation methods can lead to contradictory results. The
compressive test showed that the welds were of high quality while they fell apart after cutting them.
The need of using several evaluation methods is thus emphasized again.
151
6.9.4 Series 2.4
6.9.4.1 Introduction
The experiments in this series are a continuation of a small series of experiments which were
described in § 6.8.4 (Test series 1.4). Only two welds showed a good quality: one was created with a
voltage of 15 kV and one with 20 kV. Every voltage between those values did not result in a weld (see
Table 6.17).
Workpiece
SD-CuMs-1.5
SD-CuMs-1.6
SD-CuMs-1.7
SD-CuMs-1.43
SD-CuMs-1.44
SD-CuMs-1.78
SD-CuMs-1.79
SD-CuMs-1.45
Voltage
(mm)
10,0
12,0
15,0
15,0
18,0
18,0
19,0
20,0
Diameter
(mm)
18,0
18,0
18,0
18,0
18,0
18,0
18,0
18,0
Tube Length
(mm)
48
48
48
48
48
48
48
48
Stand-off
(mm)
2
2
2
2
2
2
2
2
Overlap
(mm)
10,0
10,0
10,0
10,0
10,0
10,0
10,0
10,0
Weld?
No
No
Partially
Yes
No
Partially
No
Yes
Table 6.17: A selection of experiments which were already described in § 6.8.4. These experiments will form the base for
new experiments which will be conducted in test series 2.4.
The experiments which will be conducted in test series 2.4 are listed in Table 6.18. The parameters of
these experiments are identical to those listed in Table 6.17 but the voltage was increased in smaller
steps (0,5 kV). The goal of these weld trails is to further investigate the influence of the charging
voltage on the weld quality. The transition zone between a weld of low and high quality should also
become more clearly visible.
Workpiece
SD-CuMs-1.104
SD-CuMs-1.105
SD-CuMs-1.106
SD-CuMs-1.107
SD-CuMs-1.108
SD-CuMs-1.109
SD-CuMs-1.110
SD-CuMs-1.111
SD-CuMs-1.112
SD-CuMs-1.113
SD-CuMs-1.114
SD-CuMs-1.115
SD-CuMs-1.116
Voltage
(kV)
14,0
14,5
15,0
15,5
16,0
16,5
17,0
17,5
18,0
18,5
19,0
19,5
20,0
Diameter
(mm)
18,0
18,0
18,0
18,0
18,0
18,0
18,0
18,0
18,0
18,0
18,0
18,0
18,0
Stand-off
(mm)
2
2
2
2
2
2
2
2
2
2
2
2
2
Tube Length
(mm)
48,0
48,0
48,0
48,0
48,0
48,0
48,0
48,0
48,0
48,0
48,0
48,0
48,0
Leak
Free?
No
No
No
No
No
No
No
No
No
No
Yes
Yes
No
Leak
Class
C
C
C
C
D
C
B
B
B
B
A
A
B
Weld?
No
No
|
No
|
No
|
No
|
No
Partially
Partially
Partially
Table 6.18: A list of the experiments which were conducted in this section. Identical geometrical parameters are used as
in § 6.8.4 but the voltage is increased in smaller steps to gain a better understanding of the influence of the charging
voltage on the MPW process. Note: symbol “|” means that the workpiece was not examined microscopically.
152
6.9.4.2 Non-destructive testing
The welds were leak tested. The results of this leak test are also listed in Table 6.18. It can be seen
that the leaks become smaller when the welding voltage increases, except for the experiment at the
highest voltage in which the leak becomes bigger again. Only two workpieces appeared leak-free :
SD-CuMs-1.114 and 1.115, which were welded with respectively 19,0 and 19,5 kV. These results
differ from the results of test series 1.4 as both the experiments at 15,0 and 20,0 kV are not welded
in this series. To find the precise transition zone, more experiments need to be conducted.
6.9.4.3 Destructive testing
Compressive test
Four workpieces were tested with the compressive test: SD-CuMs-1.106, 1.108, 1.110 and 1.112.
These workpieces were welded with different voltages which should be noticeable in the results of
the compressive test.
The relationship between the charging voltage and the maximum axial force reached at failure of the
weld is given in Figure 6.44. Note that the maximum axial force increases when the charging voltage
increases.
Maximum Axial Force[kN]
25
20
15
10
5
0
14
15
16
17
18
Charging Voltage [kV]
Figure 6.44: The maximum axial force which can be carried during the compressive test versus the used charging voltage.
It can be seen that this force increases when the voltage increases.
Microscopic examination
The workpieces which were not evaluated by the compressive test, were investigated by microscopy.
Although some workpieces were leak-free, none of the former appeared to be welded on both sides.
The best result in this series was a partially welded specimen. Thus, the leak test is not sufficient to
evaluate the weld quality with 100% certainty. This shows that the magnetic pulse welded pieces
need several evaluation methods during the search for the optimal process parameters.
During the microscopic investigation, a dark weld interface was visible (Figure 6.45). This is an
intermetallic layer and not a crack. It was thought that a high concentration of lead in this layer
created the dark color, so the chemical composition was investigated.
153
Figure 6.45: The workpieces in this series of experiments showed a dark weld interface. Although it might resemble a
crack, this was not the case.
To investigate its chemical composition, this layer was investigated with scanning electron
microscopy; Figure 6.46 and Figure 6.47 are examples of photographs. It can be observed that a bond
was. Figure 6.47 is a close-up of the dark interface zone which is indicated in Figure 6.46 .
Investigation of the chemical composition lead to the knowledge that this dark intermetallic layer
indeed had a different composition when compared to the base material. The results of several
chemical composition measurements showed that the following change occurred:
The intermetallic layer exists out of:
1. more copper
2. less zinc
3. less lead
than the base material.
Hence, the dark colour was not created by a high concentration of lead in the intermetallic zone. A
different explanation of the colour could be that this zone was attacked more by the etching agent.
Figure 6.46: A photograph of a copper-brass weld interface which was taken by a scanning electron microscope.
154
Figure 6.47: A close-up of the dark material which is positioned in the copper-brass interface (detail of Figure 6.46). This
photograph was taken by scanning electron microscopy.
155
6.10 Discussion on the copper-brass experiments
6.10.1
Destructive evaluation
6.10.1.1 Torsion test
The discussion on the torsion testing of workpiece SD-CuMs-1.99, one of the experiments on
repeatability, can be found in the destructive weld evaluation methods (Chapter 5). The brass rod
failed at a torque of 280 Nm, indicating that the shear strength of the weld exceeds the strength of
the brass base material.
The compressive test was found to be more suitable for the strength evaluation of the tubular welds.
6.10.1.2 Compressive test
The workpieces that were subjected to a compressive test are listed in Table 6.19. For each
workpiece, the force-displacement curve was recorded. These can be found in Appendix C.
All the workpieces were welded with an overlap length of 10 mm. Weld SD-CuMs-1.101 was one of
the reproducibility experiments (test series 2.2). Welds SD-CuMs-1.120, SD-CuMs-1.123 and
SD-CuMs-1.124 had grooves machined in the collar. The stand-off distance and charging voltage for
each workpiece are also listed in the table. The right column shows the maximum compressive force,
at which the workpieces failed.
Workpiece
Stand-off (mm)
Voltage (kV)
Remarks
Fmax (kN)
SD-CuMs-1.106
2
15
7,3
SD-CuMs-1.108
2
16
9,7
SD-CuMs-1.110
2
17
17,0
SD-CuMs-1.112
2
18
20,1
SD-CuMs-1.101
1,5
19
Reproducibility
25,1
SD-CuMs-1.120
1,5
19
Grooved collar
23,5
SD-CuMs-1.123
1,5
19
Grooved collar
24,4
SD-CuMs-1.124
1,5
19
Grooved collar
26,1
Table 6.19: Results of the compressive tests of test for several specimens from test series 2.2, 2.3 and 2.4.
For workpieces performed with a stand-off distance of 2 mm, it was observed that the force at which
the weld shears, increases for a larger charging voltage. This can be seen on the graph in Figure
6.48.These workpieces were probably welded only partially or even not welded. The workpieces
SD-CuMs-1.104 up to SD-CuMs-1.113 (test series 2.4) that were examined microscopically, showed
no bond. The tube spontaneously separated from the rod after cross-sectioning. It is reasonable to
assume that workpieces SD-CuMs-1.106, SD-CuMs-1.108, SD-CuMs-1.110 and SD-CuMs-1.112 would
also show no weld after cross-sectioning. Nevertheless, the force needed to shear the weld was high,
especially for SD-CuMs-1.110 and SD-CuMs-1.112.
The linear trend line added to the graph suggests that at a charging voltage of 19 kV, the maximum
force would also be around 25 kN (similar to the welds with stand-off distance 2 mm).
156
Figure 6.48: The compressive force, at which the weld shears, increases with increasing charging voltage.
Some significant differences were observed in the force-displacement curves. Figure 6.49 and Figure
6.50 show the curves for SD-CuMs-1.108 and SD-CuMs-1.112 respectively (test series 2.4).
Figure 6.49: Force-displacement curve recorded during the compression test on workpiece SD-CuMs-1.108.
(test series 2.4)
As the charging voltage increases, the maximum force also increases. For welds SD-CuMs-1.106 and
SD-CuMs-1.108 (Figure 6.49), the force does not decrease instantly after reaching its maximum value.
On the other hand, for welds SD-CuMs-1.110 and SD-CuMs-1.112 (Figure 6.50), this instant decrease
does occur. It is assumed that the sudden force decrease indicates a sudden fracture of the weld. It
can thus be assumed that workpieces SD-CuMs-1.110 and SD-CuMs-1.112 were perhaps partly
welded, while workpieces SD-CuMs-1.106 and SD-CuMs-1.108 were probably not welded.
For a stand-off distance of 2 mm, the workpieces performed at a high charging voltage (19 kV up to
20 kV) were found to be partly welded after cross-sectioning (test series 2.4). Although this indicates
157
that only a partial weld was formed, it is reasonable to assume that these welds would show
sufficient shear strength in the compressive test.
Figure 6.50: Force-displacement curve recorded during the compression test on workpiece SD-CuMs-1.112.
(test series 2.4)
Welds SD-CuMs-1.101 (test series 2.2), SD-CuMs-1.120, SD-CuMs-1.123 and SD-CuMs-1.124 (test
series 2.4) were performed at stand-off distance 1,5 mm and a charging voltage level of 19 kV. Weld
SD-CuMs-1.101 was one of the reproducibility experiments, and the other 3 workpieces had grooves
machined in the collar of the inner rod. As can be seen in Table 6.19, all 4 workpieces failed at a
compressive force of about 25 kN. The force-displacement curve of workpiece SD-CuMs-1.101 is
shown in Figure 6.51. At the onset of testing, the compressive force increased linearly with
displacement (elastic deformation). At a force of 25,1 kN the copper tube buckled, as shown in Figure
6.52. This was observed for all 4 experiments. There is no sudden force decrease after the maximum
force is reached (Figure 6.51). This sloped course of the chart can be explained by the fact that the
weld does not fail and a large force is still required to further deform the copper tube. So, the weld
can bear a larger axial force than the copper tube (25 kN).
Based on an average weld length of 4,85 mm calculated from all the reproducibility experiments (test
series 2.4), the weld shear strength can be estimated to exceed 86 MPa. This does not take into
account that interruptions of the weld zone occurred, caused by the field shaper damage. If it is
assumed that the weld extends over only 75% of the circumference, the shear strength of the weld is
minimum 115 MPa. These weld interruptions make an accurate calculation of the shear strength
rather difficult.
The fact that the copper tube fails before the weld when subjected to an axial force indicates that
weld strength exceeds the buckling resistance of tube. From the design point of view for future
applications, it is enough to know that the connection is stronger than the base material. The
allowable compressive axial force for the connection is based on the buckling resistance of the
copper tube (25 kN), which will fail before the weld.
158
Figure 6.51: Force-displacement curve recorded during the compression of workpiece SD-CuMs-1.101.
Figure 6.52: At an axial force of about 25 kN, the copper tube buckles. The welds can bear an axial force larger than the
buckling resistance of the tube.
It should be noted that for the workpieces with grooved collar, the copper tube also buckled before
the weld sheared. However, for the workpieces with grooved collar that were cross-sectioned, the
copper tube spontaneously separated from the inner rod at both sides. This indicates that the cutting
process destructs the welds easily.
It was planned to also evaluate welds SD-CuMs-1.96, SD-CuMs-1.98 and SD-CuMs-1.100 (from test
series 2.3) by compressive testing. These welds were not tested, as it is most likely that they would
also buckle.
159
If the welds’ shear strength should be accurately determined (for research or design purposes), a
hollow steel cylinder should be fabricated with an inner diameter of 25,6 mm (slightly larger than the
outer diameter of the flyer tubes), thickness of 20 mm (to ensure sufficient strength) and an axial
length of 50 mm. The cylinder is to be placed over the work piece before the push test, to prevent
the copper tube to buckle outwards. As a small clearance exists between the tube and the cylinder,
the tube will buckle slightly until it comes into contact with the cylinder. During further compression,
friction will occur between the copper tube and the steel cylinder. The applied compressive force
must push the inner rod out of the copper tube (shearing the weld), but also overcome this friction
force. The inside surface of the steel cylinder should be coated to guarantee an acceptable
coefficient of friction. Using this steel cylinder, forces higher than 25 kN could be introduced in the
weld. Note that an efficiency increase of the test can probably be obtained by a reduction of the
length of the tube. Material can be removed from the open end of the tube as only a small
displacement is necessary to break the weld or to measure the maximum compressive axial force.
160
6.10.1.3 Roundness measurements
Effect of field shaper slit
The field shaper concentrates the magnetic field to a narrow section where welding occurs. Due to
the fact that currents mainly flow near the surface of conductors, the field shaper is constructed with
a radial slit. The field shaper is illustrated in Figure 6.53, where the arrows on the side view mark the
currents. In reality the slit is only 2 mm wide.
Figure 6.53: Radial slit in field shaper (left: side view – right: section view)
The field shaper increases the amplitude of the magnetic field in a narrow axial zone, also shown in
Figure 6.53. A stronger magnetic field will lead to a larger Lorentz force and magnetic pressure. The
higher pressure finally leads to a greater acceleration of the flyer tube. Using a field shaper, the
performance of a MPW machine can thus be enhanced. As discussed in Chapter 3, no analytical
equations were found in literature on the magnitude of this effect. To quantify the magnetic field
increase, we performed a field measurement inside the field shaper.
The magnetic field does not increase with the same amount over the entire circumference of the
welding zone. This is a direct consequence of the fact that the field shaper is not axially symmetrical
due to the slit. No current flows at the slit region, so the magnetic pressure is lower in amplitude
(when compared to the rest of the circumference). The effect of the slit on the magnitude of the
magnetic field is very difficult to quantify analytically. Finite element simulations could provide an
estimate.
The result of such a simulation is shown in Figure 6.54, which illustrates the vectorial depiction of the
Lorentz force distribution over the field shaper. It can be seen that the Lorentz forces on the part of
the workpiece under the slit region decrease [77].
The accuracy of such simulations is evidently limited to the accuracy of the parameters introduced
into the model. Verification of the results is necessary to confirm the applicability of these
simulations. However, experiments to determine the magnetic field inside the field shaper are
complicated by the difficult access to the welding region (due to the construction of the coil,
161
insulating material and various clamping mechanisms). Measuring the magnetic field exactly at the
slit region is not feasible.
Figure 6.54: The vectorial depiction of the Lorentz force distribution over the field shaper. It can be seen that the Lorentz
forces on the part of the workpiece under the slit region decrease.
In [17], a buckling effect was suggested to occur in the flyer tube during electromagnetic
compression. The flyer tube deforms inwards due to the radial compression caused by the magnetic
pressure. The diameter of the tube will decrease and the radial stresses cause circumferential
compressive stresses and longitudinal tensile stresses. From the experiments discussed in [17], it was
concluded that these stresses cause both an increase of the thickness of the flyer tube, and buckling
along the circumference. Buckling is essentially a mechanical instability phenomenon that causes the
material to move outwards at several locations. The direction of movement at these locations is
opposite to the magnetic pressure, as shown in Figure 6.55.
Figure 6.55: Buckling of the flyer tube
It was assumed that the flyer tube does not buckle randomly. The positions along its circumference
where the material is pushed outwards are associated with the location of the field shaper slit. As
mentioned before, at the place of the slit the magnetic field (and thus magnetic pressure) is
significantly smaller than at the rest of the circumference. The tube is assumed to have a tendency to
buckle at this location. Due to symmetry, the tube is assumed to also buckle in positions 90˚ relative
to the slit. The buckling around the circumference of the outer tube (under these assumptions) is
schematically shown in Figure 6.55 [17].
162
Observations
Many copper-aluminium welds produced in this dissertation show various distortions in the
deformation of the flyer tube. These distortions, if present, are located at the field shaper slit, and at
positions field shaper slit + 90˚, FS slit + 180˚ and FS slit + 270˚. Figure 6.56 shows these four positions
for workpiece SD-CA-3.1. Although all four positions show a distortion, it can be seen that the
distortions are more significant at the FS slit and opposite the slit (FS slit + 180˚). The distortions were
observed in most copper-aluminium welds.
Figure 6.56: Buckling effect in workpiece SD-CA-3.1. These distortions in the tube deformation are located at the field
shaper slit, and at positions FS slit + 90˚, FS slit + 180˚ and FS slit + 270˚. Although all four positions show a distortion, it
can be seen that the distortions are more significant at the FS slit and opposite the slit (FS slit + 180˚). This is the case for
most copper-aluminium welds.
Based on these observations, the occurrence of the ‘buckling effect’ could possibly be confirmed.
However, buckling is probably not a suitable designation for these irregularities. Buckling implies
that the shapes of the distortions in the deformation pattern are caused by instability during the
compression of the flyer tube to a smaller diameter. At the FS slit, the flyer tube is compressed less
firmly against the inner rod due to the locally lower magnetic pressure. The same occurs at the
positions FS slit + i.90˚ (i = 1, 2, 3), probably for symmetry reasons. Nevertheless, the fact that the
tube material is not pressed very tightly against the inner rod at these positions is related to the
reduced magnetic pressure in the slit region, and not a consequence of a mechanical instability in the
deformation of the tube.
The distortions of the tube deformation were observed at the position of the collar of the inner rod
(Figure 6.56). In the copper-brass experiments this effect was not observed in the same severity,
although the flyer tube material is again copper. The occurrence of a distortion is only detrimental to
the quality of the weld, if it is located at the weld zone. Therefore, roundness measurements were
performed at the location of the weld zone.
163
Roundness measurement
To verify if the buckling effect occurs consistently at the field shaper slit, the roundness of several
workpieces was measured. The position of the workpiece relative to the FS slit was marked before
each experiment. The roundness measurement is illustrated in Figure 6.57. The workpiece is clamped
in the claw plate of a turning lathe, which can rotate around its axis. A measuring pin is placed
perpendicular to the work piece. The pin is attached to a spiral spring, which causes the pointer to
deflect if the pin moves up or down.
Figure 6.57: Roundness measurement
First, the roundness of the inner rod is measured, to ensure that the work piece is accurately
centered in the claw plate. The rod is assumed to be perfectly circular (turned workpiece), so the
claws are adjusted until the pin no longer deflects when in contact with the rod surface. In practice
the three claws are adjusted separately, so it is very difficult to center the workpiece perfectly. To
ensure an accurate measurement, the (relatively small) residual deflection caused by the non-perfect
alignment of the inner rod, is measured at each point around the circumference.
After this calibration, the claws are in the optimal position relative to each other and the roundness
of the welded zone can be measured. The buckling of the flyer tube (differences in tube thickness)
can be accurately determined using this method. The magnitude of the pointer deflection on the
welded zone is measured at intervals of 10˚.
At each measuring point, the difference in deflection between the welded zone and the inner rod
(reference) gives a relative measure for the out of roundness of the welded zone. As the slit position
is marked on each workpiece, the angle at which buckling occurs can easily be determined (the claw
plate has marks for the rotation angle).
This measurement was executed on seven workpieces with a copper flyer tube and brass inner rod:
SD-CuMs-1.82, SD-CuMs-1.83, SD-CuMs-1.91, SD-CuMs-1.93, SD-CuMs-1.94, SD-CuMs-1.95 and
SD-CuMs-1.96 (test series 2.1). The geometrical parameters and charging voltage of these
experiments can be found in § 6.9.1. A measurement was performed on these copper-brass
workpieces, as no distortion in roundness could be observed visually.
164
Of these workpieces, SD-CuMs-1.91, SD-CuMs-1.93 and SD-CuMs-1.95 were leak free. SD-CuMs-1.82
and SD-CuMs-1.83 failed the leak test and showed large leaks, while SD-CuMs-1.94 and SD-CuMs1.96 had very small leakage. Workpieces SD-CuMs-1.82 and SD-CuMs-1.83 were not welded (the
inner tube spontaneously separated from the flyer tube when cutting them in half for microscopic
evaluation). All the other workpieces were welded properly.
Figure 6.58 shows the measurement of the roundness profile of workpiece SD-CuMs-1.82. The
squares mark the position of the field shaper slit (in this case at 350˚) and the positions 90˚, 180˚ and
270˚ with reference to the slit. This allows verifying the hypothesis that the flyer tube “buckles” at
these positions relative to the FS slit. The triangles mark the positions where leaks were found,
during the leak test.
Similar graphs for the other workpieces can be found in Appendix D.
As could be expected, the measurements reveal that none of the welded zones are perfectly round.
However, the distortions in the roundness of the welded zone of the copper-brass welds do not
indicate a systematic relation with the position of the field shaper slit. At some positions, a larger
diameter can be found. The difference between the largest diameter and the smallest diameter of
the welded zone never exceeds 0,2 mm (see Appendix D). In addition, these small distortions do not
systematically occur at positions FS slit + i 90˚ (i = 0, 1, 2, 3).
From these observations, it can be concluded that no significant “buckling effect” is noticed at the
weld zone of the copper-brass welds. Also, the positions of the leaks found in the air leak test are
neither related to the measured distortions at the weld zone, nor to the position of the FS slit.
relative height
[x0.01mm]
10
Relative height
FS at 350 Deg
Leaks
9
8
7
6
5
4
3
2
1
0
0
60
120
180
240
300
360
angle [˚]
Figure 6.58: Measured roundness profile (SD-CuMs-1.82): the distortions in the roundness of the welded zone of the
CuMs welds are very small, and do not indicate a systematic relation with the position of the field shaper slit.
165
Microscopic investigation for heterogeneity
The roundness measurements on the copper-brass welds did not show proof to confirm the
“buckling effect”. Minor bulges were noticed, but not consistently at the location of the field shaper
slit (or at a 90˚ angles).
In the copper-aluminium welds, a bulge in the copper tube at the location of the rod collar could be
seen visually. The bulge, similar to those shown in Figure 6.56, did not consistently occur in every
workpiece. Nevertheless if the bulges occurred, they were found to be in one or more of these
locations (FS slit, FS slit+90˚, FS slit+18˚0, FS slit+270˚). For example, in workpiece SD-CA-3.4, bulges
were only observed at the FS slit and at positions 90˚ and 180˚relative to the slit (counterclockwise
when looking from the top at the outer tube on Figure 6.59).
A possible reason for this buckling could be due to the heterogeneities in the copper tube material,
resulting from the extruding process. To investigate this hypothesis, the flyer tube was cut as shown
in Figure 6.59. Two half ring-shaped pieces were examined microscopically. The arrows mark the
positions where the microscopic evaluation was carried out. The goal is to detect a possibly different
grain size in the different zones around the flyer tube circumference. If zones showing a different
grain size are located at the same position of the bulges, this could indicate that the buckling effect is
related to the extruding process of the copper tubes, rather than to the lower magnetic pressure at
the field shaper slit.
Figure 6.59: Work piece SD-CA-3.4
The two ring-shaped specimens were polished and etched before microscopic evaluation. The grains
were inspected at eight positions around the circumference (FS slit + i 45˚, i = 0..7). Figure 6.60 and
Figure 6.61 show the grains of the copper tube, respectively at the FS slit and at 225˚
counterclockwise from the FS slit.
166
Figure 6.60: Grains of the copper tube at FS slit (SD-CA-3.4)
Figure 6.61: Grains of the copper tube at FS slit + 225˚ (SD-CA-3.4)
No significant differences were found between the grains at the positions FS slit + i 90˚ (i = 0, 1, 2, 3)
and at positions FS slit + 45˚+ i 90˚ (i = 0, 1, 2, 3). It was concluded that the extruding process has no
important impact on the deformation behaviour of the copper flyer tube.
Conclusion
The distortions of the tube deformation were observed at the position of the collar of the inner rod
(Figure 6.56). The occurrence of a distortion can only detrimental to the quality of the weld, if it is
located at the weld zone. The roundness measurements of 7 copper-brass welds, performed at the
location of the weld zone, did not show any significant evidence of roundness distortions at the
position of the FS slit and at the 90˚-positions. Although the roundness was not distorted, weld
defects were found at these positions, as shown in Figure 6.62.
167
As discussed in §6.5, damage on the inner surface of the field shaper was observed. Severe cracks
were observed exactly at the positions FS slit + i.90˚ (i = 1, 2, 3), the same positions as the weld
defects. It is assumed that the magnetic pressure is locally reduced at these positions (similar to the
reduction at the slit region), due to the cracks in the field shaper. So, the distortions in the tube
deformation are probably not caused by symmetry (as assumed in [17]), but by the damaged field
shaper.
The positions of the leaks found in the air leak test are neither related to the measured distortions at
the weld zone, nor to the position of the FS slit.
Figure 6.62: The roundness measurements of 7 copper-brass welds, performed at the location of the weld zone, did not
show any significant evidence of roundness distortions at the position of the FS slit and at the 90˚-positions. Although
the roundness was not distorted, weld defects were found at these positions.
168
6.10.2
Investigation of the wave interface
To investigate the influence of the process parameters on the wave pattern, some of the welds from
§6.8 were examined by means of microscopy. During this investigation, both wavelength and
amplitude were measured, as well as the absolute position of the wave in the weld.
The wave location was measured using a micrometer attached to the microscope table. The
micrometer measures the displacement in order to obtain the absolute position of a certain wave.
This method allows to investigate the magnitude of the amplitude and wavelength of the waves
throughout the entire weld zone. An example of such measurements is given in Figure 6.63.
Figure 6.63: Measurement of wavelength and amplitude
Numerical simulations of the MPW process will be conducted by OCAS, using identical parameters as
those which were used in the experiments. The outcome of these simulations should provide the
exact collision angle, collision speed and propagation speed at every location throughout the weld.
Once the course of these parameters throughout the weld is known, they can be linked to the values
of the amplitude and wavelength, and their influence can be investigated.
The following plots show the results of the measurements which were performed on several copperbrass welds. Note that the start of the plots, x=0, corresponds to the end of the weld and is
positioned at the shoulder of the workpiece.
It was observed that not only in the welded zone waves had been formed. Some workpieces show a
wavy pattern in zones which were not bonded. In the plots, the welded zones are depicted by the
orange line and the blue line shows the wavelengths of the waves in the non-welded zone. Also a
polynomial trend line has been added for both wavelength and amplitude plots.
When considering all the results of this investigation, no clear relations can be found. In most welds
the wavelength increases towards the end of the weld (x=0, and thus for a larger collision angle), but
169
at the weld start different phenomena occur. In SD-CuMs-1.29 (Figure 6.64), the wavelength will
start to increase starting from the beginning of the weld to the weld end. In SD-CuMs-1.43 (Figure
6.70) however, the wavelength will first decrease vastly before it starts to increase again. The same
conclusion can be suggested with respect to the amplitude of the waves. In some welds the
amplitude increases throughout the weld, in others it decreases. In SD-CuMs-1.29, the amplitude
increases in the first part of the weld and will decrease in the second part.
Considering the theory of the influence of the different parameters on the wave pattern (see §2.8),
weld SD-CuMs-1.29 appears to be corresponding closely with it. The course of the plots in Figure 6.64
and Figure 6.65 is analogue to the path of the plot in Figure 2.21. The other parts however do not
seem to follow the theory that was stated above. A last remark concerns weld SD-CuMs-1.36. This
workpiece shows two wave patterns separated by a zone where no bonding took place and where
no waves were created( Figure 6.68, Figure 6.69).
As mentioned earlier, the results of the microscopic investigation are a local evaluation of the weld
quality. The wavelength measurements will therefore also be local. To obtain a more precise view,
wavelength measurements should be performed at several locations of the circumference.
A more detailed comparison with the theory can be conducted when the results of the simulations
done by OCAS are available.
Figure 6.64: Wavelength measurement for SD-CuMs-1.29
170
Figure 6.65: Amplitude measurement for SD-CuMS-1.29
Figure 6.66: Wavelength measurement for SD-CuMs-1.35
171
Figure 6.67: Amplitude measurement for SD-CuMs-1.35
Figure 6.68: Wavelength measurement for SD-CuMs-1.36
172
Figure 6.69: Amplitude measurement for SD-CuMs-1.36
Figure 6.70: Wavelength measurement for SD-CuMs-1.43
173
Figure 6.71: Amplitude measurement for SD-CuMs-1.43
174
6.10.3
Weldability window
In this section the weldability window of the copper-brass connections is discussed (Figure 6.72). This
process window is derived from the experiments which were conducted in the framework of this
thesis. Note during the copper-brass experiments the flyer tube length was always 48 mm
(corresponding overlap length = 10 mm).
Partially welded joints will be depicted as orange squares. One must keep in mind that all the
experiments were performed with a damaged field shaper. Although it is not certain, it is probable
that welds which were partially-welded would have been high-quality welds if it were not for the
broken field shaper. Further experiments are necessary to provide confirmation.
Weld
No weld
Partial weld
wedability copper-brass: flyer tube length = 48 mm
Stand-off distance [mm]
2,5
2,0
1,5
1,0
13
14
15
16
17
18
19
20
21
Charging Voltage [kV]
Figure 6.72: A weldability window for copper-brass welds with a flyer tube length of 48 mm. The varying parameters in
this plot are voltage and stand-off distance.
Figure 6.72 shows that for the stand-off distance a value of 1,5 mm should be preferred over 2,0 mm.
This confirms the recommendation that was made by the weldability windows in § 6.8, stating that a
stand-off distance equal to or larger than 2,0 mm should be avoided. To investigate the viability of a
stand-off distance which is smaller than 1,5 mm, it is recommended to perform further experiments.
175
6.11 Conclusions of the experimental research
In this work trial welds were constructed with the material combinations: copper-aluminium and
copper-brass. The main purpose of the experiments was to develop weldability windows for the
magnetic pulse welding of these material combinations.
During the development of the weldability windows some of the occurring phenomena were also
examined. These phenomena include wave formation, deformation of the internal workpiece, etc…
The following will give a brief summary of the results of the experiments.
6.11.1
Copper-Aluminium
None of the copper-aluminium experiments resulted in high-quality welds, only some were partially
welded. It was observed that a stand-off distance value of 2,5 or 3 mm was too large. The inner rods
were severely deformed, which indicates that the impact velocity was very high. Also, the impact
angle was too large.
The experiments conducted with a smaller stand-off distance (2 and 1,5mm) showed smaller leakage.
At higher voltage levels (19 kV), several workpieces were partially welded.
No relevant weldability windows could be established for welding copper to aluminium, because of
the poor weld quality observed in the experiments. The preliminary conclusion for this material
combination is that further experiments should be performed at low stand-off distances and high
voltage levels.
A fourth series of experiments was planned but could not be performed due to the field shaper
damage. It is recommended that these experiments are conducted in future research.
6.11.2
Copper-Brass
The experiments in this thesis were performed with an overlap length of 10 mm, because the
previous experiments showed that this is the optimal value for the copper-brass combination.
It was observed that a stand-off distance of 1,5 mm produced higher quality welds than a stand-off
distance of 2 mm. At high voltage levels (18 up to 20 kV), high-quality copper-brass welds were
produced, without leakage and with sufficient shear strength.
By breaking the welds, it was observed that the weld length varied around the circumference. So, the
value measured by microscopic examination is not necessarily representative for the entire weld
zone.
A weldability window for the copper-brass experiments has been established. It should be noted
that many workpieces are labeled “partially-welded”. These welds were partially welded, but showed
irregularities in the weld zone, caused by the large cracks in damaged the field shaper. The damaged
field shaper probably influenced the reproducibility experiments, which showed significant variation
when compared to welds performed using the same parameters. The weld interruptions did not
cause an unacceptable reduction of the weld strength.
Future experiments are necessary to confirm the reproducibility of the MPW process with an
undamaged field shaper. Also, tests should be performed using a smaller stand-off distance to
further develop the weldability window.
176
6.11.3
Effect of the field shaper slit
Finite element simulations reported in literature show that the magnetic pressure is locally reduced
at the radial field shaper slit. The weld defects were consistently found at the positions of the cracks
in the damaged field shaper, but very few were observed at the slit region. If weld interruptions
occurred at that location, they were very small (< 2mm). In addition, roundness measurements did
not confirm the buckling effect, suggested in a previous master thesis. So, it is not recommended to
further investigate the phenomena associated with the slit.
6.11.4
Wave formation
The wave pattern which is often present at the interface of pulse welded workpieces was quantified
for several welding experiments. However, no clear pattern could be found in the results of the
wavelength and amplitude measurements. At the research centre of OCAS, simulations are being
conducted to calculate the impact velocity and impact angle at different positions throughout the
weld. A comparison of these results with the measurements of the waves can verify the theory on
wave formation which was found in literature.
6.11.5
Deformation of the inner workpiece
During the copper-aluminium experiments sometimes severe deformation of the aluminium inner
workpiece was observed. This can be explained by an excessive stand-off distance which allows the
impact velocity to become too high.
Also it is advised to perform copper-aluminium experiments with a copper internal workpiece.
Severe deformations can be prevented by using the material with the highest yield strength as the
base material for the inner workpiece.
6.11.6
Weld strength
Both compressive and torsion testing were applied to determine the shear strength of the weld zone.
Torsion testing on both copper-aluminium and copper-brass welds resulted in failure of the inner
rod. The weld strength thus exceeds the strength of the base material, even for the copperaluminium connection, which was only partly welded. From the compressive test, it was observed
that the shear strength increases with the voltage level. For high-quality welds, the shear strength
exceeds the buckling resistance of the tube. It was concluded that the compressive test is more
suitable to evaluate the weld strength of tubular workpieces.
177
Chapter 7
Single turn coil installation
7.1 Introduction
As mentioned earlier, the original field shaper was damaged during the experiments. The damage
was of such severity that the field shaper could no longer be used. The construction of a new field
shaper takes about six weeks. Due to a limitation of time, the construction of a new field shaper was
not possible in the scope of this work, so an alternative had to be found.
It was decided to install another coil (produced by Pulsar) on the machine. The coil is a steel single
turn coil which does not use a field shaper to concentrate the magnetic field. This coil is suited for
welding cylindrical parts with an outer diameter of 60 mm. An image of the coil is given in Figure 7.1.
The coil is to be connected to the transformer table using four tension bolts. In the bottom plate of
the coil two slits are made in which copper bars are placed to ensure a good conductivity between
the coil and the transformer (Figure 7.2).
Figure 7.1: Single turn steel coil is suited for welding tubular parts with an outer diameter of 60 mm.
Due to the high currents and energy levels generated during the electromagnetic welding process, it
is important that the installation of the coil is carried out precisely to prevent damage due to sparks.
An installation manual was provided with this coil and it was carefully followed as will be described in
the following section.
178
Figure 7.2: In the bottom plate of the coil two slits are made in which copper bars are placed. These copper bars enable a
better conductivity between the coil and the transformer
7.2 Installation of the coil
The installation guide prescribes that the tension bolts should be fastened with a torque of 200Nm.
After bolting the coil onto the transformer, a procedure has to be followed to burn in the new
contact surfaces. The installation manual prescribes the following steps:
a. Power up the system
b. Set the voltage to 2200 V
c. Perform several pulses and verify the absence of sparks ( use the help of another
colleague to inspect for sparks)
d. If sparks appear, tighten the screws of the new contact in opposing corners uniformly
e. Raise the voltage in steps of 500 V and repeat steps c-d, until all sparks disappear.
f. As all sparks disappear, raise the voltage to the desired voltage for the application.
Now perform 5 – 10 cycles and verify whether the system is working properly.
7.2.1 Extra insulation - step 1
Although the manual was followed carefully, sparks kept on appearing even at low voltages. As the
bolts were already tightened to their limit torque (300 Nm), they could not be fastened any further.
Hence, another approach had to be used to obtain spark-free operation of the machine. An example
of a spark observed during the installation procedure is showed in Figure 7.3.
Figure 7.3: A spark is visible between the copper bars at the bottom of the coil.
Some extra insulation was applied below the between of the coil and the copper surface of the
transformer bank. Although the sparks were now smaller, they still appeared. At an energy level of
13 kV, an electrostatic discharge took place within the coil. Figure 7.4 shows the path of this
discharge. This phenomenon was accompanied by a loud noise and a very bright flash of light. Due to
this dischargement, the coil was damaged: some melting of the surface occurred and a crack was
179
induced (Figure 7.5).The occurrence of this electrostatic discharge once more emphasizes the need
of proper ear protection for the operators. The loud explosion which was created can easily induce
hearing loss.
Figure 7.4: An electrostatic discharge took place in the coil at a charging voltage of 13 kV. This discharge followed the
yellow line.
Figure 7.5: As a result of the electrostatic discharge, the coil was damaged. A crack was induced and the surface of the
coil showed evidence of limited melting.
Also, both the copper bars and the steel bottom of the coil were damaged by small sparks which
occurred during the tests. Figure 7.6 shows cavities which were created by these sparks. Note that it
is important to prevent such damage as the material and the components’ surface will deteriorate
rapidly around these flaws. The high intensity current which flows through the materials during the
electromagnetic pulse process will induce spark erosion at these craters, which will severely
deteriorate the components even further.
Figure 7.6: Both materials show cavities which were created by spark erosion.
180
7.2.2 Extra insulation- step 2
As the outcome of the former section showed, further measures had to be taken. First of all, extra
insulation was applied inside the coil to prevent another electrostatic discharge and the
accompanied damage. Also, the insulating part which positions the flyer tube inside the coil was
installed to provide even more security (see Figure 7.7).
The bottom of the coil underwent some changes: the slits which hold the copper bars were widened.
This allowed the bars to be placed further apart. Also, the bars were no longer pressed into the slits
enabling a better contact surface which is free of tensional stresses. The cavities which were created
by this operation were filled with a polymer and silicone to provide proper insulation (Figure 7.8).
Figure 7.7: The part which positions the flyer tube was installed inside the coil to prevent electrostatic dischargement.
Figure 7.8: The slit which holds the copper bar was widened and filled with a (white) polymer and silicone. The sparks
now appeared on the outer side of the copper bars.
Before this modification, sparks appeared on the inner side of the copper bar (= the side which is
now filled with polymer). Although those sparks disappeared, sparks now appeared on the outer side
of the bar. The problem was thus not solved but simply moved to another location. As a
consequence, even more insulation was necessary.
181
7.2.3 Extra insulation - step 3
In this step, extra insulation was added in an attempt to solve the problem. Insulation was also
applied between the other side of the copper bar and the steel. The extra insulation exists out of
insulating tape which was secured onto the steel foot of the coil by silicone. This silicone fills all the
gaps which may be present, preventing spark formation (Figure 7.9).
Figure 7.9: The other side of the copper bars now was also insulated with insulating tape and silicone. This measure was
taken to prevent the formation of sparks on that side of the copper bars.
After this modification, sparks again appeared. These new sparks were created in the cavity shown
on Figure 7.10, somewhere along the width of the coil. These sparks are much larger than those
which were described in § 7.2.1 (even at 3 kV). Test were immediately cancelled to prevent further
damage of the machine.
Figure 7.10: New sparks are formed in this cavity, somewhere along the width of the coil.
182
7.2.4 Conclusion
Even after thorough insulation of both sides of the copper bars, sparks were still present during the
tests. The bolts could not be tightened any further as they could break. Except for adding extra
insulation, further measures should be taken.
The following method is advised (depicted in Figure 7.11):
•
•
•
Every contact surface should be reground to remove all flaws. These surfaces include
the bottom of the coil, the surfaces of the copper bars and the surface of the
transformer.
A few millimeters of steel should be removed from the bottom of the steel coil.
Hence, the copper bars would reach further outside the steel base (). This would
guarantee that the entire current will flow through the copper bars and no current
would jump from the base plate of the transformer to the steel bottom of the coil.
Finally, the space that was created by removal of some steel of the coil can be filled
with insulation. A proposal for this insulation is a flexible polymer which can be
pressed in between the steel coil and the copper base plate of the transformer.
Hence the polymer will be pressed in every cavity, preventing the formation of
sparks.
Figure 7.11: A recommendation for further measures which can be taken to prevent the formation of sparks. Material
should be removed from the steel bottom of the coil. This operation enables enough room for proper insulation. Also it
will increase the distance between the steel bottom of the coil and the copper transformer bank.
183
Chapter 8
Conclusions and recommendations
for further research
8.1 Summary and conclusions of this work
Magnetic pulse welding is a complex combination of electromagnetism, mechanics and impact
welding. The process is capable of welding dissimilar metals in microseconds without the use of filler
materials, shielding gases or much preparation of the workpieces. The effect of the many parameters
is often difficult to interpret, as they often influence the process through several physical
phenomena. The MPW process has many similarities with explosion welding, especially the aspect of
the creation of the bond. The impact velocity and impact angle of the tube are the most important
parameters that influence weld formation.
Process analysis
Due to the complex nature of the process, it is difficult to describe it analytically. An analytical model
developed by the manufacturer of the MPW machine was investigated. Several simplified
assumptions were exposed, which result in inaccuracy of the model. The most important
simplification is neglecting the time-dependency of key process parameters, such as magnetic
pressure and acceleration.
The essential components required to develop an analytical model were discussed. Using electrical
circuit analysis, combined with several current measurements (to characterise the MPW machine), it
is possible to predict the current waveform. It was attempted to quantify the relation between the
magnetic field in the air gap and the discharge current, by measurement of the magnetic field
waveform. Several difficulties were encountered, some possibly related to the field shaper damage.
An alternative method to determine the B(I)-relation is the use of multi-physics finite element
models, which combine both the electromagnetic and the mechanical aspect of the MPW process.
Computations are currently being performed at OCAS to determine the B(I)-relation. Using this
relation, the magnetic field and consequently the magnetic pressure exerted on the flyer tube, can
be predicted.
The magnetic pressure results in both deformation and acceleration of the tube. The complicated
deformation behaviour of the flyer tube is the most difficult to take into account. Neglecting the
tube’s resistance against deformation, time-functions of acceleration and velocity can be estimated
by integration of the time-dependant magnetic pressure. Again, FE simulations can be applied to
study the deformation behaviour of the tube. For example, the Johnson-Cook model can be used to
model the influence of strain hardening and the strain rate hardening on the flow stress during the
process. If these simulations could generate a set of analytical equations relating the pressure to the
impact velocity and angle, the analytical model would be complete. It should be noted that in that
case, the model would not be valid in general, but only for the specific coil/field shaper and
workpiece geometry applied in the experiments.
184
The wave pattern which is often present at the interface of pulse welded workpieces was quantified
for several welding experiments. However, no clear pattern could be found in the results of the
wavelength and amplitude measurements. At the research centre of OCAS, simulations are being
conducted to calculate the impact velocity and impact angle at different positions throughout the
weld. A comparison of these results with the measurements of the waves can verify the theory on
wave formation which was found in literature.
On-line process measurements
Several on-line measurements techniques were investigated, which could improve insight in the
process. A probe was constructed to measure the magnetic field strength in the gap between tube
and field shaper.
The results of the measurements were not completely satisfying. First, the calibration of the probe
area was difficult. In addition, at high voltages, consecutive measurements at the same voltage level
showed different signals. At low voltage levels, the measurement of the magnetic field as a function
of time resulted in sinusoidal damped waveforms with the same frequency as the current. Also, the
linearity between the current and the magnetic field was only found consistently at voltage levels
lower than 12 kV. The magnitude of the magnetic field differed slightly from that calculated with
finite element analysis.
The difficult calibration of the probe area is related to the high field strength in the MPW process, as
the same problems were reported in literature. The irregularities found at higher voltage levels are
possibly caused by the field shaper damage. Therefore, it is recommended to perform the
measurements again with an undamaged field shaper.
The time-measurement using an electrical circuit is easy to implement, and can give valuable
information for the development of the analytical model.
For crimping experiments, a deformation measurement using a light source and a detector is
recommended, provided that the clamping mechanism can be properly adjusted.
Experimental research
In the MPW process, a large number of parameters are important. Given the MPW machine
(capacitance, coil inductance and field shaper geometry) and the choice of tube and rod materials
(electrical conductivity, magnetic permeability, mechanical properties), geometrical parameters of
the workpieces can be varied. The most important geometrical parameters are the overlap length
and the stand-off distance. The only machine parameter that can be changed is the charging voltage.
In this thesis, the influence of these variables on weld formation (and weld quality) was investigated
experimentally. Experiments were performed with the material combinations copper-aluminium and
copper-brass.
Because it is difficult to find the optimal range for each parameter using analytical calculations, the
objective of the experiments was to determine these values experimentally. So, by conducting a
large number of experiments with different parameter combinations, weldability windows were
established for each material combination. The experimental weldability windows can be found in
Chapter 6.
185
Weld evaluation methods
Both destructive and non-destructive testing methods were applied to evaluate the quality of the
welds.
Non-destructive tests
•
•
•
Leak test
Computerised tomography
Ultrasonic inspection
Because for some applications it is an important quality criterion for the welds to be leak free, a
simple leak test setup using pressurised air was established. The leak test is quick and easy, and gives
a straightforward measure of weld quality. Advanced non-destructive inspection techniques as
computerised tomography and ultrasonic inspection were performed. Although the tested
workpieces were not welded, no flaws could be detected. The testing equipment used was not suited
for the workpiece geometry.
Destructive tests
•
•
•
Microscopic examination
Compressive test
Torsion test
Microscopic examination is used to evaluate the weld length and wave formation. Both compressive
and torsion testing were applied to determine the shear strength of the weld zone. Torsion testing
on both copper-aluminium and copper-brass welds resulted in failure of the inner rod. The weld
strength thus exceeds the strength of the base material, even for the copper-aluminium connection,
which was only partly welded. From the compressive test, it was observed that the shear strength
increases with the voltage level. For high-quality welds, the shear strength exceeds the buckling
resistance of the tube. It was concluded that the compressive test is more suitable to evaluate the
weld strength of tubular workpieces.
It is important to emphasize the need for applying multiple testing methods when evaluating weld
quality. For example, several welds did not show leakage, but the tube separated from the rod during
or after cross-sectioning. Other welds leaked, notwithstanding a weld with wave pattern was
observed during microscopic examination.
Metallographic examination showed that the specimens were only partially welded, or even lead to
spontaneous separation of flyer tube and inner workpiece. It was however observed that most welds
have sufficient shear strength. It is possible that the cross-sectioning operation breaks welds,
because of the release of large residual stresses. So, using only microscopic examination, workpieces
might be evaluated as not-welded when they in reality have sufficient weld strength.
186
Weldability windows
Copper-Aluminium
None of the copper-aluminium experiments resulted in high-quality welds, only some were partially
welded. It was observed that a stand-off distance value of 2,5 or 3 mm was too large. The inner rods
were severely deformed, which indicates that the impact velocity was very high. Also, the impact
angle was too large.
The experiments conducted with a smaller stand-off distance (2 and 1,5mm) showed smaller leakage.
At higher voltage levels (19 kV), several workpieces were partially welded.
No relevant weldability windows could be established for welding copper to aluminium, because of
the poor weld quality observed in the experiments. The preliminary conclusion for this material
combination is that further experiments should be performed at low stand-off distances and high
voltage levels.
A fourth series of experiments was planned but could not be performed due to the field shaper
damage. It is recommended that these experiments are conducted in future research.
Copper-Brass
This series of experiments was a continuation of experiments performed at the Belgian Welding
Institute. The experiments in this thesis were performed with an overlap length of 10 mm, because
the previous experiments showed that this is the optimal value for the copper-brass combination.
It was observed that a stand-off distance of 1,5 mm produced higher quality welds than a stand-off
distance of 2 mm. At high voltage levels (18 up to 20 kV), high-quality copper-brass welds were
produced, without leakage and with sufficient shear strength.
By breaking the welds, it was observed that the weld length varied around the circumference. So, the
value measured by microscopic examination is not necessarily representative for the entire weld
zone.
A weldability window for the copper-brass experiments has been established. It should be noted
that many workpieces are labeled “partially-welded”. These welds were partially welded, but showed
irregularities in the weld zone, caused by the large cracks in damaged the field shaper. The damaged
field shaper probably influenced the reproducibility experiments, which showed significant variation
when compared to welds performed using the same parameters. The weld interruptions did not
cause an unacceptable reduction of the weld strength.
Finite element simulations reported in literature show that the magnetic pressure is locally reduced
at the radial field shaper slit. The weld defects were consistently found at the positions of the cracks
in the damaged field shaper, but very few were observed at the slit region. If weld interruptions
occurred at that location, they were very small (< 2mm). In addition, roundness measurements did
not confirm the buckling effect, suggested in a previous master thesis. So, it is not recommended to
further investigate the phenomena associated with the slit.
187
Future experiments are necessary to confirm the reproducibility of the MPW process with an
undamaged field shaper. Also, tests should be performed using a smaller stand-off distance to
further develop the weldability window.
Single turn coil
The field shaper used for the experiments with tubes with an outer tube diameter of 25 mm, was
damaged. Therefore, a single-turn coil was mounted on the equipment. This coil is used for tube
diameters of 60 mm.
Experiments with this coil could not be performed due to problems during installation. Severe sparks
occurred between the coil and the base plate of the transformer. These sparks should be prevented
at any cost to protect both the coil and transformer from damage.
Several attempts were undertaken to prevent the sparks: multiple isolation layers, silicones to seal
small air gaps, etc. In Chapter 7, further measures to prevent sparks are suggested. The sparks are a
direct result of the high energy, transferred during the MPW process.
188
8.2 Future research
Several recommendations for future research were already suggested in the conclusions. The most
important recommendations are elaborated in this section.
•
The magnetic field measurements should be performed again, when an undamaged field
shaper is available. The measurements should confirm the functionality of the probe and can
determine the B-I relation, which is necessary to continue the development of the analytical
model. They would also allow a comparison with the FE simulation results by OCAS.
•
The deformation behaviour of the flyer tube should be investigated to complete the
analytical model. Finite element simulations should give insight in the impact velocity and
impact angle, resulting from a given magnetic pressure acting on a given overlap length.
•
Another possibility to gain more insight in the process parameters is to perform experiments
with a simple geometry. For example, by reducing the tube length to less than 15 mm (which
is the field shaper length at the inside surface), the entire tube length is subjected to almost
uniform magnetic pressure. If the inner rod has no collar, the tube deformation can be
predicted more easily, and the deformation pressure could be estimated.
•
The suggested geometrical simplifications in combination with additional measurements
could provide a lot of information (especially to investigate time-dependency). The time
measurement suggested in Chapter 4 is quite easy to implement. The deformation
measurement using a light source and a detector is only possible if the clamping mechanism
can be properly adjusted.
•
Further experiments are necessary to develop a weldability window for copper-aluminium
welds. The experiments planned in the fourth series could be performed (Chapter 6).
•
Copper-brass experiments are necessary to confirm the quality of the partially-welded
workpieces. Also, additional experiments with a smaller stand-off distance would complete
the weldability window for welds with a flyer tube length of 48 mm.
•
The weld quality evaluation methods should be further optimised. None of the evaluation
methods which were used in this work can confirm the weld quality with 100% certainty.
Further research on this subject is strongly recommended.
•
The measurements of the wave pattern should be compared to the finite element
simulations performed at OCAS once they become available. Hence, the influence of
different process parameters on the wavelength and amplitude can be further investigated.
•
The inner surface of the field shaper and single-turn coil should be inspected on a regular
basis to detect the initiation of crack formation.
189
Finally, some words of advice for future thesis students. Both the MPW process and the welds
produced by it, have many aspects to investigate. You should develop a general understanding of all
aspects, but it will not be possible to perform research (literature or experimental) on all of them
within one academic year. In our experience, it is important to focus on the ones you choose, or you
will be at risk to get ‘lost’ in the multitude of parameters.
An example is the analytical modelling. In literature, many formulas are suggested and many FE
simulations are discussed, which apply only to one specific field shaper or coil geometry. It is
recommended that you clearly define what it is that you are looking for, by breaking the problem
into pieces.
As for the experiments, it should be taken into account that a certain period is required to deliver the
ordered materials. You should start with performing experiments early, even if you do not fully
understand all the effects of every parameter at that time. It is necessary to perform a large number
of experiments, to determine the statistical relevance of results and to complete relevant weldability
windows.
The only parameter that can be changed from the MPW machine is the charging voltage. The tube
length determines the overlap length. The tube thickness and the rod diameter determine the standoff distance. These three are the most important parameters to choose in the experiments. (*)
Perform the experiments, take photographs, and plot the weldability windows in Excel. From the
weldability windows, it will be clear which parameters you should adapt to higher or lower values.
After a while, when you have a better understanding of the parameters, you can revisit the
photographs and draw conclusions.
(*) It is also possible to change the capacitance of the capacitor bank, and the transformer (both
change current frequency and amplitude). Changing these parameters could also be interesting.
Perhaps if the current amplitude is lower and the frequency higher (by reducing the capacitance), the
magnetic field measurement will provide better results.
190
Appendix
Appendix A: Drawing of the coupling for the leak test
191
Appendix B : Copper-Brass, test series 1
Workpiece
series 1
SD-CuMs-1.13
SD-CuMs-1.14
SD-CuMs-1.15
Voltage Diameter
(kV)
(mm)
12
20,5
15
20,5
18
20,5
Stand-off Tube Length
(mm)
(mm)
0,75
50
0,75
50
0,75
50
Weld?
No
No
No
SD-CuMs-1.16
SD-CuMs-1.17
SD-CuMs-1.18
12
15
18
20,5
20,5
20,5
0,75
0,75
0,75
48
48
48
No
No
No
SD-CuMs-1.19
SD-CuMs-1.20
SD-CuMs-1.21
12
15
18
20,5
20,5
20,5
0,75
0,75
0,75
46
46
46
No
No
No
SD-CuMs-1.50
SD-CuMs-1.51
18
19
Partially
No
12
15
18
20
1,0
1,0
1,0
1,0
1,0
1,0
1,0
51
51
SD-CuMs-1.22
SD-CuMs-1.23
SD-CuMs-1.24
SD-CuMs-1.25
20,0
20,0
20,0
20,0
20,0
20,0
20,0
50
50
50
50
No
Partially
Partially
No
SD-CuMs-1.58
SD-CuMs-1.59
18
20
20,0
20,0
1,0
1,0
50
50
Partially
No
SD-CuMs-1.66
SD-CuMs-1.67
18
19
20,0
20,0
1,0
1,0
49
49
Yes
No
SD-CuMs-1.26
SD-CuMs-1.27
SD-CuMs-1.28
SD-CuMs-1.74
SD-CuMs-1.75
SD-CuMs-1.29
12
15
18
18
19
20
20,0
20,0
20,0
20,0
20,0
20,0
1,0
1,0
1,0
1,0
1,0
1,0
48
48
48
48
48
48
No
No
Partially
Yes
Yes
Yes
SD-CuMs-1.30
SD-CuMs-1.31
SD-CuMs-1.32
SD-CuMs-1.33
12
15
18
20
20,0
20,0
20,0
20,0
1,0
1,0
1,0
1,0
46
46
46
46
No
No
No
No
series 2
192
Workpiece
Voltage
Diameter
Stand-off
Tube Length
(kV)
(mm)
(mm)
(mm)
Weld?
SD-CuMs-1.52
SD-CuMs-1.53
18
19
19,0
19,0
1,5
1,5
51
51
Partially
Yes
SD-CuMs-1.34
SD-CuMs-1.35
SD-CuMs-1.60
SD-CuMs-1.36
SD-CuMs-1.61
15
18
18
20
20
19,0
19,0
19,0
19,0
19,0
1,5
1,5
1,5
1,5
1,5
50
50
50
50
50
No
Yes
Yes
Yes
Yes
SD-CuMs-1.68
SD-CuMs-1.69
18
19
19,0
19,0
1,5
1,5
49
49
Partially
Yes
SD-CuMs-1.37
SD-CuMs-1.38
SD-CuMs-1.76
SD-CuMs-1.77
SD-CuMs-1.39
15
18
18
19
20
19,0
19,0
19,0
19,0
19,0
1,5
1,5
1,5
1,5
1,5
48
48
48
48
48
No
No
Yes
Yes
Yes
SD-CuMs-1.40
SD-CuMs-1.41
SD-CuMs-1.42
15
18
20
19,0
19,0
19,0
1,5
1,5
1,5
46
46
48
No
No
No
series 3
193
Workpiece
Voltage
Diameter
Stand-off
Tube Length
(kV)
(mm)
(mm)
(mm)
Weld?
SD-CuMs-1.54
SD-CuMs-1.55
18
19
18,0
18,0
2,0
2,0
51
51
No
No
SD-CuMs-1.62
SD-CuMs-1.63
18
19
18,0
18,0
2,0
2,0
50
50
No
No
SD-CuMs-1.1
SD-CuMs-1.2
SD-CuMs-1.3
SD-CuMs-1.4
12
15
18
19
18,0
18,0
18,0
18,0
2,0
2,0
2,0
2,0
50
50
50
50
No
No
No
No
SD-CuMs-1.70
SD-CuMs-1.71
18
19
18,0
18,0
2,0
2,0
49
49
No
No
SD-CuMs-1.5
SD-CuMs-1.6
SD-CuMs-1.7
SD-CuMs-1.43
SD-CuMs-1.44
SD-CuMs-1.78
SD-CuMs-1.79
SD-CuMs-1.45
10
12
15
15
18
18
19
20
18,0
18,0
18,0
18,0
18,0
18,0
18,0
18,0
2,0
2,0
2,0
2,0
2,0
2,0
2,0
2,0
48
48
48
48
48
48
48
48
No
No
SD-CuMs-1.9
SD-CuMs-1.10
SD-CuMs-1.11
SD-CuMs-1.12
10
12
15
18
18,0
18,0
18,0
18,0
2,0
2,0
2,0
2,0
46
46
46
46
No
No
No
No
SD-CuMs-1.56
SD-CuMs-1.57
18
19
17,0
17,0
2,5
2,5
51
51
Partially
No
SD-CuMs-1.64
SD-CuMs-1.65
18
19
17,0
17,0
2,5
2,5
50
50
Partially
Partially
SD-CuMs-1.72
SD-CuMs-1.73
18
19
17,0
17,0
2,5
2,5
49
49
No
No
SD-CuMs-1.80
SD-CuMs-1.81
18
19
17,0
17,0
2,5
2,5
48
48
Partially
No
series 4
Partially
Yes
No
Partially
No
Yes
series 5
194
Appendix C: Force-displacement curves (compressive test)
195
196
197
Appendix D: Roundness Measurements
Relative height
FS at 350 Deg
Leaks
SD-CuMs-1.82
relative height
[x0.01mm]
30
25
20
15
10
5
0
0
60
120
relative height
[x0.01mm]
30
180
240
300
360
angle [˚]
Relative height
SD-CuMs-1.83
FS at 105 Deg
Leaks
25
20
15
10
5
0
0
60
120
180
240
300
360
Relative height
SD-CuMs-1.91
relative height
[x0.01mm]
30
angle [˚]
FS at 30 Deg
25
20
15
10
5
0
0
60
120
180
240
300
360
angle [˚]
198
Relative height
FS at 80 Deg
SD-CuMs-1.93
relative height
[x0.01mm]
30
25
20
15
10
5
0
0
60
120
180
240
300
360
Relative height
FS at 120 Deg
Leak
SD-CuMs-1.94
relative height
[x0.01mm]
30
angle [˚]
25
20
15
10
5
0
0
60
120
180
240
300
SD-CuMs-1.95
relative height
[x0.01mm]
30
360
angle [˚]
Relative
height
25
20
15
10
5
0
0
60
120
180
240
300
360
angle [˚]
199
SD-CuMs-1.96
relative height
[x0.01mm]
30
Relative
height
25
20
15
10
5
0
0
60
120
180
240
300
360
angle [˚]
200
References
[1]
V. Shribman. Magnetic Pulse welding. PULSAR Ltd. Magnetic Pulse Solutions, 2007.
[2]
V. Shribman. Magnetic pulse welding of automotive HVAC parts, PULSAR Ltd. Magnetic Pulse
Solutions, July 2006.
[3]
Figures: BWI
[4]
Haiping Yu, Chunfeng Li, Zhiheng Zhoa, Zhong Li. Effect of field shaper on magnetic pressure
in electromagnetic forming, China, 2005
[5]
Skin Effect. http://en.wikipedia.org/wiki/Skin_effect.
[6]
Chromerics. EMI Shielding theory
[7]
K. Faes. Tube welding. Belgium, 2010
[8]
J. Colotti. Telephonics – Command System Division presentation. EMC design fundamentals,
2005
[9]
P.V. Ramanathan, A.R. Vaidyanathan. Design for quality explosive welding. Journal of
processing technology, 32, 2007
[10]
Figures: PST
[11]
B.S. Zlobin. Explosion Welding of Steel With Aluminum, 2001
[12]
I. Masumoto, K. Tamaki, M. Kojima. Electromagnetic welding of aluminum tube to aluminum
or dissimilar metal cores., Japan, 1985
[13]
F. Gringon, D. Benson, K.S. Vecchio, M.A. Meyers, Explosive welding of aluminum to
aluminum: analysis computation and experiments, San Diego, 2004
[14]
A. Nobili, Explosion Bonding Process, France
[15]
V. Couplez, Perfect metallische binding in fractie van seconde. Belgium
[16]
P. Goes, A.M. Valle, Modelling of the Magnetic Pulse Forming Process, Zelzate, 2009
[17]
K. Loncke, An exploratory study into the feasibility of magnetic pulse welding, Master thesis,
Ghent University, 2009
[18]
Pressure vessel, http://en.wikipedia.org/wiki/Pressure_vessel
[19]
M.S. Peihui Zhang. Joining enabled by high velocity deformation, Ohio, 2003
[20]
A. Ben-Artzy, A. Stern, N. Frage, V. Shribman, O. Sadot. Wave formation mechanism in
magnetic pulse welding, Israel, 2009
[21]
AH. Holtzman, GR. Cowan, Bonding metals with explosives, 1965
[22]
A.A. Akbari Mousavi, S.T.S. Al-Hassani, Numerical and experimental studies of the mechanism
of the wavy intervace formations in explosive/impact welding, Teheran, Iran, 2005
201
[23]
Y. Zhang, S.S. Babu, C. Porthe, M. Blakely, J. Kwasegroch, M.LaHa, G.S. Daehn, Application of
high velocity impact welding at varied different length scales. USA, 2010
[24]
O.B. Drennov, About the state of two-metal contact boundary ar high velocity oblique impact,
Russia, 1999
[25]
Ahmet Durgutlu, Hasan Okuyucu, Behcet Gulenc, Investigation of effect of the stand-off
distance on interface characteristics of explosively welded copper and stainless steel, Turkey,
2007
[26]
Pulsar. MP Application Development, Internal Report
[27]
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
[28]
MS. Manish Kamal, A uniform pressure electromagnetic actuator for forming flat sheets,
Ohio, 2005
[29]
A. Henselek, M. Beerwald, C. Beerwald, Design and Adaptation of EMF Equipment- From
Direct Acting Multi-turn Coils to Seperable Tool Coils for Electromagnetic Tube Compression,
2004
[30]
T. Hammers, M. Marré, J. Rautenberg, P. Barreiro, V. Schulze, D. Löhne, A. Brosius and A.E.
Tekkaya, Influence of Mandrel’s surface on the mechanical properties of joints produced by
electromagnetic compression., 2008
[31]
Wei Xu, Hongyuan Fang, Wenli Xu, Analysis of the variation regularity of the parameters of
the discharge circuit with the distance between workpiece and inductor for electromagnetic
forming processes, Journal of Materials Processing Technology, 2007
[32]
M.Kleiner, C. Beerwald, M. Homberg, Analysis of process parameters and forming
mechanisms within the electromagnetic forming process, 2007
[33]
P. Zhang, et al., Analysis of the electromagnetical impulse joining process with a field
concentrator, 2004
[34]
Kashani, Tomokatsu Aizawa and Mehrdad, Magnetic pulse welding (MPW) method for
dissimilar sheet metal joints, Tokyo Metropolitan college of Technology, Japan.
[35]
http://www.physics.sjsu.edu/becker/physics51/mag_field.htm, 2010
[36]
University of Pennsylvania. The Magnetic Field Produced by Coils and Solenoids
[37]
A.G. Mamalis, D.E. Manolakos, AG. Kladas, AK. Koumoutos, Physical principles of
electromagnetic forming process: a constitutive finite element. Greece
[38]
P. Barreiro, V. Schulze, D. Löhe, M. Marré, C. Beerwald, W.Homberg, M. Kleiner. Strength of
Tubular joints made by electromagnetic compression at quasi-static and cyclic loading, 2006
[39]
Pulsar, Basic Principles of field shaper design, 2008
202
[40]
A. G. Mamalis, D. E. Manolakos, A. G. Kladas, and A. K. Koumoutsos. , Electromagnetic
Forming Tools and Processing Conditions: Numerical Simulations., Materials and
Manufacturing Processes, 2006
[41]
O. Gafri, A. Izhar, Y. Livshitz, V. Shribman., Magnetic Pulse Acceleration, 2006
[42]
Banerjee, Biswajit, An evaluation of plastic flow stress models for the simulation of hightemperature and high-strain-rate deformation of metals, 2005
[43]
Y.U. Haiping, L.I. Chunfeng, Effects of current frequency on electromagnetic tube
compression, Journal of Materials and Processing Technology, 2009
[44]
A. Brosius, M. Kleiner, Determination of Material Characteristics using Electromagnetic
Forming and Weak Coupled Finite Element Simulations, Germany, 2004
[45]
T. Aizawa, M. Kashani and K. Okagawa, Application nof Magnetic Pulse Welding for Aluminum
Alloys and SPCC Steel Sheet Joints, 2007
[46]
G. S. Daehn, Y. Zhang, S. Golowin, K. Banik, A. Vivek, J. R. Johnson, G. Taber, G. K. Fenton, I.
Henchi, P. L’Eplattenier, Coupling Experiment and Simulation in Electro- magetic Forming
Using Photon Doppler Velocimetry, 2008
[47]
Yuan Zhanf, P. L'Eplattenier, A. Vivek, G. Taber, G. Daehn, S. Babu, Numerical Simulation and
Experimental Study for magnetic Pulse Welding Process on AA6061-T6 and Cu101 Sheet, 10th
International LS-DYNA User Conference
[48]
R. Merte, D. Peier, J. Teunissen, Optical Postition and Time Resolved Measurement of
Magnetic Field Distribution in High Speed Metal Forming, 2004
[49]
Filip Broekaert, Michaël De Ketele, An exploratory study into the feasibility of magnetic pulse
forming, Master thesis, Ghent University, 2009
[50]
Hewson, W.F. Ray and C.R, High Performance Rogowski Current Transducers, Italy, 2000
[51]
Internal Communication with P. Goes, OCAS, 2009
[52]
L. Dupré, http://www.ugent.be/ir/eesa/en/research.
[53]
ASM Committee. Explosion welding metals Handbook 6
[54]
Ultrasonic
Testing.
http://www.ndt-ed.org/EducationResources/CommunityCollege/
Ultrasonics /cc_ut_index.htm
[55]
Robert E. Shaw Jr., P.E., M. ASCE, Ultrasonic Testing Procedures, Technician Skills and
Qualifications, Journal of materials in Civil Engineering, 2002
[56]
Versaevel Bart, Anckaert Tom, Optimalisatie van een super-heterodyne ontvanger van
ultrageluid, 2006
[57]
Ondacorp, Acoustic properties of solids, Sunnyvale CA, 2003
203
[58]
GE Sensing & inspection Technologies, Ultrasonic transducers for flaw detection and sizing.,
Germany, 2010
[59]
Computerised tomography; http://www.ndt-ed.org/EducationResources /CommunityCollege
/Radiography/AdvancedTechniques/computedtomography.htm
[60]
Phoenix x-ray: http://www.phoenix-xray.com/en/applications/3d_computed_tomography
/index.html
[61]
http://aluminium.matter.org.uk/content/html/eng/default.asp?catid=175&pageid=
2144416592
[62]
N.K. Naik, A. Asmelash et al, Interlaminar shear properties of polymer matrix composites:
Strain rate effect, Mechanics od materials, India, 2007
[63]
Ramin Hosseinzadeh, Farid Taheri, Non-linear investigation of overlap length effect on
torsional capacity of tubular adhesively bonded joints,Composite structures, Canada, 2009
[64]
Tinius Olsen, Torsion Test catalogue [www.TiniusOlsen.com]
[65]
M. Ouddane, R. Boukhili, Proposal for a torsion shear ficture for bonded joints, Journal of
Material Science, Canada, 1999
[66]
L. Dorn, TALAT Lecture 4705 (Quality Assurance), Germany, 1994
[67]
S. Kalpakjian, The metallurgy of welding: welding design and process selection,
Manufacturing engineering and technology
[68]
V. Shribman, Magnetic Pulse Joining of light metal castings, Germany, 2008
[69]
RoyMech,
Shear
http://www.roymech.co.uk/Usefum_Tables/Matter/shear_tensile.htm.
[70]
J. McGinley, Electromagnetic Pulse Technology as a means of joining generation IV cladding
materials, Brussels, Belgium, 2009
[71]
B. Gülenc, Investigation of interface properties and weldability of aluminum and copper
platers by explosive welding method, Materials and Design, 2008
[72]
ASNT: www.asnt.org/publications/materialseval/basics/june97basics/june97basics.htm.
[73]
ATC, Inc, Bubble test: www.atcinc.net/bubble-test.asp.
[74]
TQC, Leak test: www.tqc.co.uk/leaktesting/leak_basic_principles.htm.
[75]
Magneform: http://www.magneform.com/pres.html.
[76]
I. Eguia, P. Zhang, G. S. Daehn, Improved Crimp-Joining of Aluminum Tubes onto Mandrels
with Undulating Surfaces, Ohio, USA, 2004
[77]
A.G. Mamalis, D.E. Manolakos, A.G. Kladas, A.K. Koumoutsos, Electro-magnetic Tooling for
Metal Forming and Powder Compation: Numerical Simulation, Greece, 2004
strength:
204
[78]
M. Gerloch, Magnetism and Ligand-field Analysis, 1983
[79]
Magnetic Flux Density: http://en.wikipedia.org/wiki/magnetic_flux_density
[80]
E. Purcell, Electricity and Magnetism, 1963
205