Theoretical and FE Analysis of

Ultrasonic Welding of Aluminum

Alloy 3003
Ultrasonic welding (consolidation) process is a rapid manufacturing process that is used
to join thin layers of metal at low temperature and low energy consumption. Experimen-
tal results have shown that ultrasonic welding is a combination of both surface (friction)
A. Siddiq and volume (plasticity) softening effects. In the presented work, an attempt has been
made to simulate the ultrasonic welding of metals by taking into account these effects
E. Ghassemieh (surface and volume). A phenomenological material model has been proposed, which
incorporates these two effects (i.e., surface and volume). The thermal softening due to
Department of Mechanical Engineering, friction and ultrasonic (acoustic) softening has been included in the proposed material
University of Sheffield, model. For surface effects, a friction law with variable coefficient of friction that is
Sheffield S1 3JD, UK dependent on contact pressure, slip, temperature, and number of cycles has been derived
from experimental friction tests. The results of the thermomechanical analyses of ultra-
sonic welding of aluminum alloy have been presented. The goal of this work is to study
the effects of ultrasonic welding process parameters, such as applied load, amplitude of
ultrasonic oscillation, and velocity of welding sonotrode on the friction work at the weld
interface. The change in the friction work at the weld interface has been explained on the
basis of softening (thermal and acoustic) of the specimen during the ultrasonic welding
process. In the end, a comparison between experimental and simulated results has been
presented, showing a good agreement. 关DOI: 10.1115/1.3160583兴

Keywords: ultrasonic metal welding, ultrasonic consolidation, aluminum alloy 3003,

thermomechanical analysis, friction laws, ultrasonic softening

1 Introduction and other contaminants. Internal stresses set up elastoplastic de-

formations across the interface to create true metallurgical bonds
Ultrasonic consolidation 共UC兲 is a metal laminate process, de- between the layers of foil.
veloped by Solidica, Inc. UC combines ultrasonic welding 共USW兲 Kong et al. 关3–5兴 studied the ultrasonic consolidation of AA-
and computer numerical control 共CNC兲 milling technologies to 6061 and AA-3003. On aluminum alloy 6061, they performed
produce three-dimensional metal parts in a single process 关1兴. Fig- microhardness tests on welded and unwelded specimens. Their
ure 1 shows the schematic layout of the UC apparatus 关2兴. The UC results indicate that hardness near the weld interface is larger than
machine uses a standard ultrasonic welding system, with 3.3 kW the hardness away from the interface. This difference in hardness
input power, for large output ultrasonic power for metal welding. values depicted that both surface effect 共friction兲 at the weld in-
In contrast to other metal matrix composite 共MMC兲 fabrication terface and ultrasonic softening 共acoustic softening兲 in the mate-
techniques, UC does not apply heat to the work pieces during the rial were present during the welding. They did similar studies on
aluminum alloy 3003 and found that unlike AA-6061, AA-3003
process. The technology uses ultrasonic oscillation to produce
does not require cleaning prior to the welding.
friction and adhesion of surface asperities at the weld interface of Cheng and Li 关6兴 investigated the heat generation and tempera-
two metal foils, and applies relatively low pressures on the ture profile during ultrasonic metal welding, using microsensor
sonotrode to produce bonds. As a “cold” and low pressure pro- arrays. The materials used for welding were copper alloy and
cess, it would enable temperature sensitive and fragile functional nickel substrate. It was found that temperature ranges from 100° C
fibers/elements to be embedded within the metal structure, with- to 250° C for various loading cases.
out damage or distortion. Doumanidis and Gao 关7,8兴 performed the mechanical analysis
UC is now capable of building monolithic aluminum compo- of an ultrasonic spot welding process of a metal foil on a sub-
nents and composite structures through the repeated deposition strate. The mechanical analysis was based on the definition of
and welding of metal foils with fibers, placed within them. Suc- frictional boundary conditions at the foil/substrate interface. The
cessive layers of metal foil are ultrasonically welded to the layer friction boundary condition was defined by using the experimen-
below and the profile, as defined by a solid computer aided design tally measured strain on the substrate surface and adjacent to the
ultrasonic probe.
共CAD兲 model of the component, machined onto it to define the
So far, most of the research on ultrasonic consolidation has
complete solid object. relied on laboratory work. Trial and error has been the main
Bonding of the layers of foil occurs through the action of ultra- method in establishing the process parameters. To date, little the-
sonic oscillation energy that causes highly localized interfacial oretical understanding and explanation exists for the reasons be-
slip between two surfaces that breaks up the surface oxide layer hind the success of the process at a certain condition, and lack of
success in others. Despite the report on the surface and volume
effect during the consolidation process, these effects and the pa-
Contributed by the Manufacturing Engineering Division of ASME for publication
in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received
rameters affecting them are not quantified or fully understood.
November 27, 2007; final manuscript received November 16, 2008; published online In the present research, a material model based on cyclic plas-
July 13, 2009. Review conducted by Bin Wei. ticity theory has been proposed to take into account the volume

Journal of Manufacturing Science and Engineering AUGUST 2009, Vol. 131 / 041007-1
Copyright © 2009 by ASME

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 1 Ultrasonic welding specimen, „a… finite element model, and „b… geometry of ultrasonic welding specimen
„side view…

effects, while a kinematic friction model has been proposed to creasing plastic deformation. C is the kinematic shift of the yield
include the contribution of surface effects during the ultrasonic surface.
welding. It must be noted that exact bonding mechanism during In the present study, the thermomechanical term is adopted
ultrasonic welding process, including the relative interface kine- from the hardening model by Johnson and Cook 关18兴. The inclu-
matics, is one of the continued investigation, and friction based sion of such thermomechanical coupling term is quite straightfor-
model hypothesized in the presented work can be on such influ- ward. This was done by embedding the temperature term 共1
ential mechanism to better understand the welding behavior. The − ␪ˆ m兲 in the nonlinear isotropic and kinematic hardening model.
proposed material and friction models are inputted to a quasistatic
The modified nonlinear isotropic hardening law is given by
thermomechanical finite element model of the process of consoli-
dation. The effect of pressure, amplitude of vibration, sonotrode pl
Rth = Q共1 − e−b␧¯ 兲 · 共1 − ␪ˆ m兲 共3兲
velocity on the friction work, temperature, and plastic flow within
the bonding zone are analyzed and reported. where m is the material parameter, and ␪ˆ is the nondimensional
temperature given as
2 Material Model ␪ − ␪transition
␪ˆ = 共4兲
The thermomechanical cyclic plasticity model proposed in this ␪melt − ␪transition
work is based on the combined nonlinear isotropic/kinematic
hardening model for time independent cyclic plasticity, presented ␪transition is the transition temperature, at or below which there is
by Chaboche and co-worker 关9–12兴. no temperature dependence on yield stress, and ␪melt is the melting
The basic constitutive equations of the model can be found in temperature.
Refs. 关9–12兴. For brevity, only isotropic and kinematic hardening Similarly, the modified nonlinear kinematic hardening law is
relations are discussed below. given by
The nonlinear isotropic hardening rule adopted here was pre- ␣គ th = ␣គ isothermal共1 − ␪m兲 共5兲
sented by Lemaitre and Chaboche 关13兴, and Huber and Tsakmakis
关14兴. The isotropic hardening 共R兲, which describes the expansion ␣គ isothermal is obtained from the numerical integration of Eq. 共2兲
of the yield surface, is defined as an exponential function of ac- 共see Ref. 关19兴兲.
cumulated plastic strain 共i.e., isotropic hardening is dependent on In the present work, a simple phenomenological approach has
the plastic strain history兲, which is given by been used to include the acoustic 共ultrasonic兲 softening during
pl
deformation of the material. The main assumption for this ap-
R = Q共1 − e−b␧¯ 兲 共1兲 proach comes from Refs. 关20–23兴. The main assumptions are: The
decrease in plastic 共yield兲 limit is proportional to the ultrasonic
where ␧ is the equivalent plastic strain, while Q and b are ma-
¯ pl
intensity. Also, the decrease in plastic 共yield兲 limit is independent
terial parameters to be identified by inverse modeling. Q is the of the frequency 共in the range of 15–80 kHz兲, temperature 共in the
maximum change in the size of the yield surface, and b is the rate range of 30– 500° C兲, and preliminary deformation. It must be
at which the size of the yield surface changes with changing plas- noted that acoustic softening effect still requires thorough experi-
tic strains. mental investigation to better understand the process. In the recent
A nonlinear kinematic hardening proposed by Armstrong and years, Gillman 关24兴 made an effort to explain the activation of
Frederick 关15兴 has been used to capture nonlinear hardening be- dislocations without any thermal aid. It was proposed that there
havior and smooth transition from elastic to plastic deformation exists a critical speed at which an extended dislocation will have a
关16,17兴. The evolution of back stress 共␣兲 is given by higher total energy than a unit dislocation, minus the energy of its
C stacking fault energy. At this critical velocity, the extended dislo-
␣គ̇ = 共␴គ − ␣គ 兲␧
¯ pl − ␥␣គ ¯␧pl 共2兲 cations will contact into unit dislocations, causing screw disloca-
␥ tions to cross glide freely, without the aid of the thermal activa-
where C and ␥ are the material parameters, which can be identi- tion. However, authors still believe these phenomena require
fied from cyclic testing. The ␥ term determines the rate at which further experimental investigations, especially in the context of
the saturation value of kinematic hardening decreases with in- ultrasonic welding process.

041007-2 / Vol. 131, AUGUST 2009 Transactions of the ASME

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 A phenomenological softening term, dependent on the ultra- Table 1 Thermal and mechanical properties of AA-3003
sonic energy density per unit time, has been introduced in the
relations of isotropic and kinematic hardening terms 共Eqs. 共3兲 and Thermal properties
共5兲兲. The modified equations of isotropic and kinematic hardening
Thermal conductivity 162 W / m K
are given by
Thermal expansion coefficient 23.2⫻ 10−6 / ° C
Rultrasonic = Rth · 共1 − d · Eultrasnonic兲2 共6兲 Specific heat 896 J / kg K
Density 2730 kg/ m3
␣គ ultrasonic = ␣គ th · 共1 − d · Eultrasonic兲2 共7兲
Elastic properties
where Rth is the isotropic hardening rule with thermal softening, Young’s modulus 70 GPa
defined in Eq. 共3兲, d is the ultrasonic softening parameter, which Poisson’s ratio 0.33
has to be identified from the experiments of deformation behavior Yield stress 190 MPa
of the material in the presence of ultrasonic energy. Eultrasonic is the
ultrasonic energy density per unit time, transferred from the
sonotrode to the material. ␣គ th is the kinematic back stress term
with thermal softening, defined in Eq. 共5兲. The identified set of material parameters 共Q , b , C , ␥ , m兲 are
The above discussed hardening models, Eqs. 共1兲–共7兲, are imple- given in Table 2, and the comparison of the uniaxial experimental
mented in ABAQUS user subroutine user defined hardening subrou- and simulated stress-strain curves are shown in Figs. 2共a兲 and
tine 共UHARD兲. This user subroutine is called at all material inte- 2共b兲. The ultrasonic softening parameter 共d兲 has been identified by
gration points of elements for which the material definition comparing the experimental results of ultrasonic softening 关21兴 of
includes user-defined isotropic or cyclic hardening for metal plas- aluminum. The experimental and simulated initial yield stresses as
ticity 关19兴. This user subroutine can be used to define a material’s a function of ultrasonic energy density per unit time are plotted in
isotropic yield behavior, the size of the yield surface in a com- Figs. 2共a兲 and 2共b兲.
bined hardening model or material behavior, dependent on field or
state variables. The implementation is very straight forward. The 4 Friction Model
UHARD user subroutine requires three quantities to be defined, Friction and wear play an important role during the sliding of
i.e., hardening model 共Eqs. 共1兲–共7兲兲, rate of change in hardening two surfaces under cyclic loading, such as high cycle fatigue in
rule with respect to the accumulated plastic strain ¯␧pl, which is turbine blades and ultrasonic processes, etc. The significance of
given by the effect of the friction coefficient on the simulation results of the
dRultrasonic pl
ultrasonic process has been discussed in previous research, where
pl
= Q · b · e−b␧¯ 共1 − ␪ˆ m兲 · 共1 − d · Eultrasonic兲2 共8兲 it is mainly focused on constant friction coefficient or a test driven
d␧¯ friction model 关24兴. Here, building on the understanding of the
importance of a realistic friction model, a friction law has been
d␣ultrasonic pl pl
pl
= 共Ce−␥¯␧ − ␣1 · ␥ · e−␥¯␧ 兲共1 − ␪m兲 · 共1 − d · Eultrasonic兲2 proposed, which comprises of static and kinematic friction com-
d␧¯ ponents.
共9兲 The friction model proposed in this work is based on the de-
pendence of coefficient of friction ␮, on number of cycles N,
and rate of change in hardening rule with respect to the tempera- temperature T, and parameters a and b, which depend on magni-
ture, which is given by

冋 册
tude of slip and contact pressure.
The friction model proposed in this work is based on the fric-
dRultrasonic pl 共␪ˆ m−1兲 tion experiments performed by Naidu and Raman 关26兴. These re-
= − Q · m · 共1 − e−b␧¯ 兲 ·
d␪ˆ 共␪melt − ␪transition兲 searchers found that the coefficient of friction increases with the
increasing number of cycles, and after a certain number of cycles,
· 共1 − d · Eultrasonic兲2 共10兲 coefficient of friction saturates to a steady value 共see Fig. 3兲.

冋冉 冊 册
A simple logarithmic correlation has been used to define the
d␣ultrasonic C pl pl 共␪ˆ m−1兲 experimental friction behavior. The correlation is given by
=− 共1 − e−␥¯␧ 兲 + ␣1 · e−␥¯␧ ·
d␪ˆ ␥ 共␪melt − ␪transition兲 ␮iso = ␮s + ␮s · 共a · log共N兲 + b兲 共12兲
· 共1 − d · Eultrasonic兲 2
共11兲 where a and b are friction parameters, which depend on the mag-
nitude of the slip amplitude and contact pressure, while ␮s is the
The above discussed nonlinear isotropic/kinematic hardening law
initial static coefficient of friction, and N is the number of cycles.
requires the identification of six parameters 共Q , b , C , ␥ , m , d兲 from
The values of a and b are determined from the experimental re-
cyclic stress-strain data, thermal softening data, and acoustic 共ul- sults presented by Naidu and Raman 关26兴. For the case of contact
trasonic兲 softening data. pressure of 50 MPa, and stress amplitude of 193 MPa, 共equivalent
to a slip magnitude of 8.4兲 a and b are found to be a = 0.323 and
b = −0.1. The comparison of the model with the experimental re-
3 Material Parameter Identification sult for this case is shown in Fig. 4.
The material model discussed above with nonlinear isotropic A simple correlation for a and b has been used based on the
and kinematic hardening rules, along with thermomechanical cou- experimental results and is given by
pling and ultrasonic softening, involves six material parameters a = a1 − a2 · P 共13兲
共Q , b , C , ␥ , m , d兲. The identification of these material parameters
is performed using inverse modeling method. In this method ex- and
perimental stress-strain curves are compared with the simulated
stress-strain curves for each set of parameters, and the difference
Table 2 Final set of parameters for aluminum alloy 3003
between the experimental and simulated stress-strain curves is
minimized by the variation in these parameters. The process is Q b C ␥ m d
repeated until a good agreement is achieved between experimental
and simulated stress-strain curves. The thermal and mechanical 150 MPa 25.0 12.0 GPa 500.0 0.7 1.3⫻ 10−6 m2 / W
properties, taken from Refs. 关14,25兴, are given in Table 1.

Journal of Manufacturing Science and Engineering AUGUST 2009, Vol. 131 / 041007-3

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 2 „a… Comparison between experimental and simulated uniaxial
stress-strain curves. „b… Comparison between experimental and simulated
initial yielding as a function of temperature. „c… Effect of ultrasonic energy
per unit time on initial yielding

b = b1 − b2 · P 共14兲 perature, and then decreases. This temperature dependence has

been taken into account by including an additional fourth order
The values of a1, a2, b1, and b2 are identified from the experi- polynomial as a function of temperature. The modified coefficient
ments for different contact pressures. The values are given in is given by
Table 3 for different displacement amplitude values.
The temperature dependence of the coefficient of friction is
introduced using the experimental observation by Zhang et al. ␮ = ␮iso · 共p · T4 + q · T3 + r · T2 + s · T + t兲 共15兲
关27兴. They studied the friction behavior of aluminum alloy at dif-
ferent temperatures. It was found that the coefficient of friction The additional friction parameters p , q , r , s , t are identified using
increases with the increasing temperature, until a specific tem- the experimental results 关27兴 for aluminum alloy and are given in

041007-4 / Vol. 131, AUGUST 2009 Transactions of the ASME

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 3 Ratio of coefficient of friction after each cycle to static Fig. 5 Coefficient of friction as a function of temperature „ex-
coefficient of friction as a function of number of cycles †26‡ periments and friction model…

Table 4, while experimental and friction model results are plotted available friction subroutine 关28兴 has been modified by including
in Fig. 5. Eqs. 共19兲–共22兲. This has been done by replacing the constant co-
4.1 Implementation of the Friction Model. The proposed efficient of friction with a coefficient of friction that is dependent
friction model has been implemented by the ABAQUS user subrou- on temperature, contact pressure, amplitude of oscillation, and
tine 共FRIC兲 to define the user friction model. The implementation number of cycles.
is based on Coulomb’s friction law with stick-slip algorithm 关28兴
with 兩␶fric兩 ⱕ ␮ · P, i.e., if 兩␶fric兩 ⬍ ␮ · P, then it is a stick, and if 5 Ultrasonic Welding Model and Boundary
兩␶fric兩 = ␮ · P, then it is a slip. The numerical implementation of Conditions
simple coulomb’s friction law with the constant friction coeffi-
cient has been discussed in Ref. 关28兴. In this work, the already Typical ultrasonic metal welding setup is shown in Fig. 1. The
welding setup consists of three main components, a foil, a sub-
strate, and a sonotrode 共attached to the ultrasonic welding unit兲.
The substrate is fixed to an anvil at the bottom surface, a foil is
placed at the top surface of the substrate, and with the help of the
sonotrode load, is applied in the vertical direction 共as shown in
Fig. 1共a兲兲. The sonotrode oscillates at a frequency of 20 kHz, in
the direction perpendicular to the rolling 共welding兲 direction.
The process parameters during ultrasonic welding are applied
load 共Papplied兲, velocity of sonotrode 共V兲, frequency of the ultra-
sonic oscillation 共f兲, and the amplitude of oscillation, which are
normally varied in order to check the effect of each of these pa-
rameters on weld quality.
The geometric parameters for the present study are the width of
the specimen 共w兲 at 20 mm, thickness of the substrate 共ts兲 at 1
mm, and thickness of the foil 共t f 兲 at 100 ␮m. The radius of the
sonotrode used in the present study is 25 mm 关4兴. In the present
work, the effect of the ultrasonic welding process parameters 共ap-
plied load, velocity of sonotrode, and amplitude of ultrasonic os-
Fig. 4 Comparison between friction model and experiments cillation兲 is studied. Frequency of the ultrasonic oscillation 共f
†26‡ for a contact pressure= 50 MPa and stress amplitude = 20 kHz兲 is always kept constant during the present study. The
= 193 MPa applied load is varied from 25 MPa to 175 MPa. The amplitude of
oscillation is varied from 8.4 ␮m to 14.4 ␮m, while velocity of
Table 3 Identified values of a1, a2, b1, and b2 for different dis- the sonotrode is varied from 27.8 mm/s to 38.8 mm/s, which gives
placement amplitudes the weld time of 0.0258–0.036 s/mm.
Finite element analyses of the ultrasonic welding specimen
Displacement amplitude
共␮m兲 a1 a2 b1 b2
have been performed using coupled temperature-displacement
analysis. The finite element model has been constructed for a
8.4 0.362 0.000827 ⫺0.07303 0.000592 small region of approximately 2 mm, as shown in Fig. 1共a兲. The
10.4 0.25 ⫺0.00033 ⫺0.1447 0.000313 selected width of the model 共2 mm兲 has been chosen, based on the
12.4 0.2924 0.00054 ⫺0.18218 0.00015 width of the contact area during the specific weld cycle 共which
14.4 0.2192 0.0000854 ⫺0.2157 0.000015 ranges from 0.133 mm to 0.324 mm, depending on applied load-
ing兲. That is, the size of the model is large enough, so that the
weld area is not affected from the boundary. The geometry of the
Table 4 Friction parameters for temperature dependent fric-
model from the side view is shown in Fig. 1共b兲. As discussed
tion coefficient
above, foil has the dimensions of 2 ⫻ 20⫻ 0.1 mm3, while sub-
p q r s t strate is of dimensions 2 ⫻ 20⫻ 1 mm3 共20 mm is in the direction,
orthogonal to the side view, as shown in Fig. 1共b兲. The geometry
8.485⫻ 10−10 −8.842⫻ 10−7 1.969⫻ 10−4 −9.762⫻ 10−3 1.12 of the sonotrode, which is cylindrical in the real experiments 共di-
ameter of 50 mm兲 has been modeled as hollow. Also, only a

Journal of Manufacturing Science and Engineering AUGUST 2009, Vol. 131 / 041007-5

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Table 5 Thermal and mechanical properties of sonotrode
„steel…

Thermal properties

Thermal conductivity 80 W / m K
Thermal expansion coefficient 11⫻ 10−6 / ° C
Specific heat 440 J / kg K
Density 7800 kg/ m3

Elastic properties
Young’s modulus 200 GPa
Poisson’s ratio 0.27

portion of the sonotrode that covers the 2 mm wide specimen has Fig. 6 Coefficient of friction „measured… between sonotrode
been modeled, as shown in Fig. 1共b兲. This has been done in order „steel… and foil „aluminum alloy…
to save computation time. The radius of the sonotrode is 25 mm
共see Fig. 1共b兲兲. Also, the thickness of the sonotrode sheet is cho-
sen to be 0.2 mm; this size is chosen so that heat flux does not foil is plotted in Fig. 6. Throughout this work, it is assumed that
reach the inner surface of the sonotrode for the weld time used in all of the friction energy generated between different contacting
the simulations. Process parameters used during the simulations, surfaces is converted into heat energy. This has been achieved by
i.e., velocity of sonotrode, applied load, and oscillation ampli- using the “gap heat generation” option, already available in
tudes, are applied through a reference point associated with the ABAQUS Inc. 关19兴. With the help of this option, the user can
sonotrode, as shown in Fig. 1共b兲. Normal load is applied in the define what fraction of the friction energy is converted into heat.
vertical direction; rolling velocity of the sonotrode is applied us- For the present study, this fraction is set to 1, i.e., all of the friction
ing the velocity boundary condition in the rolling direction 共as energy is converted into heat energy. It is also assumed that the
shown in Fig. 1共b兲兲, while ultrasonic oscillation is applied or- heat energy is equally transferred to both the contacting surfaces
thogonal to the side view in Fig. 1共b兲, using displacement bound- in every interface. Results of the thermomechanical ultrasonic
ary conditions with a periodic amplitude curve 共A = Amax sin ␻t兲, welding analyses have been discussed in the Sec. 6.
where A is the amplitude of the oscillation dependent on time,
Amax is the maximum amplitude of the oscillation, ␻ is the angular 6 Results and Discussion
velocity, and t is the total time. The values of the process param-
eters used are already given in the previous paragraph. The bottom 6.1 Effect of Applied Load and Displacement Amplitude
surface of the substrate is fixed in all directions, the same as on Friction Work Between the Foil/Substrate Interface. The
during the real experiments 关4兴. Initial temperature of 20° C has friction work at the interface of the foil and substrate is plotted in
been prescribed to the complete geometry. For the current analy- Fig. 7. These values are obtained when friction work saturates
ses, all the free edges 共other than the contact surfaces and left and 共approximately after 100–160 cycles for the applied loading be-
right edges in Fig. 1共b兲兲 in the model are under free convection tween 125 MPa and 175 MPa兲. The results plotted in Fig. 7 are for
with the convection coefficient of 30 W/m K and ambient tem- the case when velocity of sonotrode was 27.8 mm/s. It is found
perature of 20° C. The left and right edge boundaries are assumed that as the applied load increases, the friction work at the interface
of foil and substrate also increases. This increase in friction work
to be at 20° C, by prescribing the temperature boundary condition.
is related to the ultrasonic energy being transferred to the foil by
These temperature boundary conditions have been used, based on
the sonotrode during the welding process. It is found that ultra-
the experimental findings in Ref. 关6兴, which showed that for dif-
sonic energy transferred to the foil increases with the increasing
ferent weld times, the temperature away from the weld regions
load. The ultrasonic energy per second that is transferred to the
decrease, as one moves away from the weld area. Also, heat flux
foil material at different points in the foil along the thickness
has been tracked during all the simulations, and it has also been
direction is plotted in Fig. 8. It can also be seen that ultrasonic
found that heat flux never reached these boundaries, as the weld
time is very small, i.e., 4 – 9 ␮s, which also makes the assumption
of temperature 共20° C兲 valid for the left and right edges. The
bottom surface of substrate and top surface of the sonotrode are
also kept at 20° C. Throughout this work, foil and substrate have
been modeled using the material properties of aluminum alloy
3003 共see Sec. 3兲, while the sonotrode is modeled as steel 共elastic兲
with properties given in Table 5.
Finite element mesh consisted of 31842 coupled temperature-
displacement elements 共C3D8RT兲. Thermomechanical interaction
properties between foil/sonotrode interface and foil/substrate in-
terface are discussed in Sec. 6. The normal contact properties
between the foil/sonotrode interface and the foil/substrate inter-
face are defined using hard contact formulation, available in Ref.
关19兴.
The friction properties between foil and substrate contact sur-
face is defined using the thermomechanical friction model pro-
posed in this work 共Sec. 4兲. Friction properties used are already
discussed in Sec. 4. The friction coefficient for sonotrode/foil in-
terface 共steel/aluminum兲 is defined using a pressure dependent
isothermal coefficient friction in Coulomb’s friction model. The Fig. 7 Friction work at foil/substrate interface, velocity
pressure dependent coefficient of friction between sonotrode and = 27.8 mm/ s

041007-6 / Vol. 131, AUGUST 2009 Transactions of the ASME

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 8 Ultrasonic power transferred to the foil along the thick-
ness direction
Fig. 10 Friction work between foil/substrate interface as a
energy transferred in the foil along the thickness direction de- function of velocity of sonotrode „applied load= 155 MPa…
creases, as one moves from the sonotrode surface towards the
substrate surface.
It can also be inferred from Fig. 7 that as the amplitude of 6.3 Plastic Deformation. Plastic deformation in the specimen
ultrasonic oscillation increases, the friction work at the interface is caused by two different phenomena, namely surface 共friction
of foil and substrate also increases. This shows that the amplitude dissipation兲 and volume 共ultrasonic softening兲 effects. It is found
of ultrasonic oscillation has a similar effect to that of the load. The that as the applied load increases, the plastic deformation in the
reason for this increase could be understood by considering the foil at foil/sonotrode interface also increases 共Fig. 11兲. This is due
fact that as the amount of applied loading is always lower than the to the increase in friction dissipation 共surface effect兲 at foil/
static yield stress of the material, the friction energy dissipation sonotrode interface, along with the increasing ultrasonic softening
starts well, before yielding can occur as a result of applied load 共volume effect兲 due to ultrasonic energy transferred from the
and ultrasonic oscillation. Therefore, the friction work for the case sonotrode to the foil 共Fig. 8兲.
of AA-3003 共order of 1 – 55 mJ/ mm2兲 increases with increasing The contour plots of equivalent plastic strain are shown in Fig.
applied loading and amplitude of oscillation. These results depict 12. It is found that as the amount of applied load increases, the
a completely opposite behavior than the AA-6061, where it is plastic deformation in the foil and substrate at the foil/substrate
found that friction work decreases with the increasing applied interface decreases. This could be explained by the fact that the
loading and amplitude of oscillation due to the plastic dissipation, time and rate at which plastic deformation occurs and saturates
which starts in less number of ultrasonic cycles. affect the amount of induced plastic deformation. Our results on
the plastic strain variation with number of cycles are reported in
6.2 Effect of Sonotrode Velocity on Friction Work Be- Fig. 13. These indicate that as the loading rate increases, the plas-
tween Foil/Substrate Interface. Friction work as a function of tic strains saturate in less number of ultrasonic cycles. In other
velocity of the sonotrode for different applied loads and constant words, friction dissipation saturates in less number of cycles for
oscillation amplitude 共8.4 ␮m兲 has been plotted in Fig. 9. It can higher applied loading. Similar effects on plastic deformation
be inferred from the plot that as the velocity of the sonotrode have been observed with increasing displacement amplitude of
increases, the friction works decreases. This is due to the higher ultrasonic vibration.
friction dissipation in the elastic range with higher loading rate. The results plotted in Figs. 11 and 12 show that the regions of
Similar trend of friction work has been found for different am- the foil near foil/sonotrode interface undergo severe plastic defor-
plitudes of the ultrasonic oscillation. Figure 10 shows the plot of
friction work between foil/substrate interface as a function of ve-
locity for different displacement amplitudes and constant applied
load 共155 MPa兲. It can be seen that as the amplitude of the ultra-
sonic oscillation increases, the friction work also increases.

Fig. 9 Friction work between foil/substrate interface as a func- Fig. 11 Equivalent plastic strain in the foil at foil/sonotrode
tion of velocity of sonotrode „oscillation amplitude= 8.4 ␮m… interface „amplitude= 8.4 ␮m; velocity= 27.8 mm/ s…

Journal of Manufacturing Science and Engineering AUGUST 2009, Vol. 131 / 041007-7

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 12 Equivalent plastic strain in substrate and foil at foil/substrate interface „amplitude= 8.4 ␮m; velocity
= 27.8 mm/ s…

mation. This high plastic deformation is due to high friction dis- surface effects, i.e., friction dissipation at the foil/substrate inter-
sipation and ultrasonic energy transferred to the foil near foil/ face, and very small amount of volume 共ultrasonic softening兲 ef-
sonotrode interface 共Fig. 8兲. It can also be inferred from Fig. 12 fects to cause plastic deformation.
that the amount of plastic deformation in the foil near the foil/
substrate interface is higher than the plastic deformation in the 6.4 Temperature at the Interfaces. The temperature of the
substrate. As explained before, the reason for having larger plastic specimen, due to the friction dissipation at the weld interface dur-
deformation in foil is due to the dual effect, i.e., surface 共friction ing ultrasonic metal welding, has been found to be 31–75% of the
dissipation at foil/substrate interface兲 and volume 共ultrasonic soft- melting temperature 共Tmelt = 660° C兲 for the cases when amplitude
ening兲 effects. On the other hand the substrate has the dominating of ultrasonic vibration was 12.4 ␮m or below. In some cases,
when amplitude of vibration was 14.4 ␮m, the temperature of the
material in the foil reached upto 97% of the melting temperature.
That is why, it was found during the experiments that at higher
amplitudes of ultrasonic vibration, some of the materials of the
foil stick to the sonotrode, which then requires additional process-
ing of the sonotrode before further welding 关25兴. Figure 14 shows
the contour plot of the maximum temperature in the weld speci-
men. The plot in Fig. 14 shows that as the amount of applied load
increases, the temperature of the foil surface at foil/sonotrode also
increases due to the higher friction dissipation.
It is not easy to measure the temperature directly at the weld
surface or contact area as this will damage the thermocouples.
Therefore, temperature adjacent to the contact area 共⬃500 ␮m兲
was measured 关29兴. The experimentally measured temperature and
simulated temperature at the corresponding distance from the con-
tact area are plotted in Fig. 15, which shows a similar trend.
The temperature results are also in agreement with the experi-
mental results of Cheng and Li 关6兴, where temperature for the case
of copper and nickel were found to be in the range of
100– 250° C, at a distance of 200 ␮m 共at point A in Fig. 14兲 from
the weld interface. Similar values of the temperature are found for
lower vibration amplitudes 共see Fig. 14兲, where temperature away
Fig. 13 Equivalent plastic strains v/s number of ultrasonic from the interface decreases and is at a distance of 200 ␮m 共re-
cycles „amplitude= 8.4 ␮m; velocity= 27.8 mm/ s… gion A兲 from the weld interface; it is in the range of 174– 200° C.

041007-8 / Vol. 131, AUGUST 2009 Transactions of the ASME

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 14 Temperature in the weld specimen for three different applied load-
ings „amplitude= 8.4 ␮m; velocity= 27.8 mm/ s…

One of the most important conclusions, which can be inferred equals to 34.5 mm/s 关4兴. Using the peel test force-displacement
from Fig. 14 is, there exists a temperature gradient from the curves, fracture energies have been computed using the procedure
sonotrode to the foil, and then to the substrate. Figure 14 shows described by Korn et al. 关30兴, and Siddiq and co-workers 关31,32兴.
that the highest temperature is at the foil/sonotrode interface, and Friction work between the foil/substrate interface and fracture
decreases as one moves toward the foil/substrate interface. This energies computed from peel test curves are plotted in Fig. 16 as
important conclusion justifies the argument given by many experi- a function of amplitude of ultrasonic oscillation, for an applied
mentalists, that friction only plays a secondary role during ultra- load of 175 MPa. It can be inferred from Fig. 16 that as the
sonic welding, i.e., it is only required to break and disperse the amplitude of ultrasonic oscillation increases, the experimental
surface impurities and oxide layers. The main bonding mechanism fracture energy obtained from peel tests also increases. Similar
is mechanical interlocking of atoms when two surfaces are in trend can be seen in Fig. 16 for friction work at the foil/substrate
contact at atomic scale, i.e., the clearance between the two sur- interface, i.e., as the amplitude of ultrasonic oscillation increases,
faces is in atomic distances. This small atomic distance is then friction work at the foil/substrate interface also increases. The
enough for the surface atoms to form a bond. reason for the increase in experimental fracture energy can be
explained on the basis of increasing friction work.
6.5 Comparison With Experimental Results. In order to As explained in the previous sections, the friction energy serves
compare the simulated results with the experiments, peel tests are
performed on the AA-3003 specimen, welded at an applied load
of 155 MPa and 175 MPa, and a velocity of sonotrode that is

Fig. 16 Friction work and experimental fracture energy as a

Fig. 15 Comparison of temperature adjacent to the weld area function of amplitude of ultrasonic oscillation „velocity
„experimental and simulations… = 34.5 mm/ s; load= 175 MPa…

Journal of Manufacturing Science and Engineering AUGUST 2009, Vol. 131 / 041007-9

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 • The results of fracture energy estimated from experiments
shows similar trends as the simulation prediction for the
friction work.

Acknowledgment
The authors thankfully acknowledge the financial support of
EPSRC 共Engineering and Physical Sciences Research Council兲
and MOD 共Ministry of Defense兲 through the Grant No. GR/
T19988, and the collaborative support of the Solidica Ltd.

References
关1兴 White, D. R., 2003, “Ultrasonic Consolidation of Aluminium Tooling,” Ad-
vanced Materials and Processes, 161, pp. 64–65.
关2兴 Feng, Z., and Diamond, Z., 2003, “Basic Studies of Ultrasonic Welding for
Advanced Transportation Systems,” Oak Ridge National Laboratory, Oak
Fig. 17 Friction work and experimental fracture energy as a Ridge, TN, Tennessee FY 2003.
function of amplitude of ultrasonic oscillation „velocity 关3兴 Kong, C. Y., Soar, R. C., and Dickens, P. M., 2003, “Characterisation of
= 34.5 mm/ s; load= 155 MPa… Aluminium Alloy 6061 for the Ultrasonic Consolidation Process,” Mater. Sci.
Eng., A, 363, pp. 99–106.
关4兴 Kong, C. Y., Soar, R. C., and Dickens, P. M., 2004, “Optimum Process Pa-
rameters for Ultrasonic Consolidation of 3003 Aluminium,” J. Mater. Process.
as a secondary purpose for welding, i.e., it is required to break and Technol., 146, pp. 181–187.
disperse the oxide layers and other impurities to bring the two 关5兴 Kong, C. Y., Soar, R. C., and Dickens, P. M., 2004, “A Model for Weld
contacting surfaces closer to each other 共approximate atomistic Strength in Ultrasonically Consolidated Components,” Proc. Inst. Mech. Eng.,
distances兲. Therefore, as the friction work at the foil/substrate in- Part C: J. Mech. Eng. Sci., 219, pp. 83–91.
关6兴 Cheng, X., and Li, X., 2007, “Investigation of Heat Generation in Ultrasonic
terface increases, the dispersion of the oxide layer and other im- Metal Welding Using Micro Sensor Arrays,” J. Micromech. Microeng., 17, pp.
purities also increase, causing larger number of bonds at the foil/ 273–282.
substrate interface. This higher density of bonds ultimately results 关7兴 Doumanidis, C., and Gao, Y., 2004, “Mechanical Modeling of Ultrasonic
into a higher fracture energy as exhibited by the welded specimen Welding,” Welding Journal, 83共4兲, pp. 140S–146S.
关8兴 Gao, Y., and Doumanidis, C., 2002, “Mechanical Analysis of Ultrasonic Bond-
of the AA-3003 specimen during experiments 共Fig. 16兲. ing for Rapid Prototyping,” ASME J. Manuf. Sci. Eng., 124, pp. 426–434.
Similar trends can be seen for applied loading of 155 MPa 共see 关9兴 Chaboche, J. L., 1986, “Time-Independent Constitutive Theories for Cyclic
Fig. 17兲. It is found that as the applied loading or amplitude of Plasticity,” Int. J. Plast., 2, pp. 149–188.
ultrasonic oscillation increases, the experimental fracture energy 关10兴 Chaboche, J. L., 1989, “Constitutive Equations for Cyclic Plasticity and Cyclic
Viscoplasticity,” Int. J. Plast., 5, pp. 247–302.
also increases; similar trends have been shown by the friction 关11兴 Chaboche, J. L., and Rousselier, G., 1981, “On the Plastic and Viscoplastic
work at the weld interface during thermomechanical analyses. The Constitutive Equations Based on the Internal Variables Concept,” Proceedings
reason for this increase in friction work has been explained above. of the Post SMIRT-6 Conference, ONERA, Paris, France.
关12兴 Chaboche, J. L., and Rousselier, G., 1983, “On the Plastic and Viscoplastic
Constitutive Equations,” ASME J. Pressure Vessel Technol., 105, pp. 153–
7 Conclusion 164.
A procedure to simulate the ultrasonic welding process of alu- 关13兴 Lemaitre, J., and Chaboche, J. L., 1990, Mechanics of Solid Materials, Cam-
bridge University Press, Cambridge, England.
minum alloy 3003 has been presented in this work. Both volume 关14兴 Huber, N., and Tsakmakis, C., 1999, “Determination of Constitutive Properties
and surface effects have been incorporated in the proposed cyclic From Spherical Indentation Data Using Neural Networks—Part II: Plasticity
plasticity material model with thermal and acoustic softening. A With Nonlinear Isotropic and Kinematic Hardening,” J. Mech. Phys. Solids,
friction law, based on the dependence of the coefficient of friction 47, pp. 1589–1607.
关15兴 Armstrong, P. J., and Frederick, C. O., 1966, “A Mathematical Representation
upon the contact pressure, amount of slip, temperature, and num- of the Multiaxial Bauschinger Effect,” CEGB Report No. RD/B/N 731.
ber of cycles has been used. Results show the following. 关16兴 Chun, B. K., Kim, H. Y., and Lee, J. K., 2002, “Modeling the Bauschingers
Effect for Sheet Metals—Part II: Applications,” Int. J. Plast., 18, pp. 597–616.
• As the ultrasonic welding parameters change, the friction 关17兴 Wang, H., and Barkley, M. E., 1998, “Strain Space Formulation of the
work required to break and disperse the oxides at the inter- Armstrong-Frederick Gamily of Plasticity Models,” ASME J. Eng. Mater.
face also changes. It has been found that as the amount of Technol., 120, pp. 230–235.
关18兴 Johnson, G. R., and Cook, W. H., 1985, “Fracture Characteristics of Three
applied loading or amplitude of oscillation increases, the Metals Subjected to Various Strains, Strain Rates, Temperatures and Pres-
friction work at the weld interface also increases. sures,” Eng. Fract. Mech., 21, pp. 31–48.
• It is also found that as the rolling velocity of sonotrode 关19兴 ABAQUS, 2006, ABAQUS version 6.5, Online Documentation, Hibbitt, Karlsson &
increases, the friction work decreases, due to the less num- Sorenson Inc.
关20兴 Green, R. E., Jr., 1975, “Non-Linear Effects of High-Power Ultrasonics in
ber of cycles of oscillation transmitted to the foil before the Crystalline Solids,” Ultrasonics, 13, pp. 117–127.
plastic yielding starts. 关21兴 Langenecker, B., 1966, “Effects of Ultrasounds on Deformation Characteris-
• The maximum temperature reached during the ultrasonic tics of Metals,” IEEE Trans. Sonics Ultrason., SU-13, pp. 1–8.
welding is below melting temperatures of the joining mate- 关22兴 Mordyuk, N. S., 1975, “Influence of Ultrasonic Oscillations on the Physical
Properties of Metals and Alloys,” Naukova Dumka, Kiev.
rials for lower amplitudes of ultrasonic vibration 关23兴 Severdenko, V. P., Klubovich, V. V., and Stepanenko, A. V., 1973, “Metal
共8.4– 12.4 ␮m兲, while for higher amplitudes of ultrasonic Working Under Pressure With Ultrasound,” Nauka i Tekhnika, Minsk.
vibration 共14.4 ␮m兲, the temperature near the sonotrode 关24兴 Gilman, J. J., 2001, “Contraction of Extended Dislocations at High Speeds,”
Mater. Sci. Eng., A, 319–321, pp. 84–86.
reaches up to 97% of the melting temperature. 关25兴 Li, D., and Soar, R. C., 2007, “Report for Machine Performance Validation
• Maximum plastic strains are observed on the top of the foil Through Comparison of Peel Strength and Linear Weld Density of Samples
at the interface of the foil and substrate 共sonotrode??兲. Or- Made Before and After Machine Maintenance,” Rapid Manufacturing Re-
ders of magnitude smaller plastic strain takes place in the search Group, Loughborough University, Loughborough, LeicestershireUK.
关26兴 Naidu, N. K. R., and Raman, S. G. S., 2005, “Effect of Contact Pressure on
substrate, indicating that less of the vibration energy is Fretting Fatigue Behaviour of Al–Mg–Si Alloy AA-6061,” Int. J. Fatigue, 27,
transferred to the substrate. Specially increasing the applied pp. 283–291.
load will amplify this effect. 关27兴 Zhang, C. B., Zhu, X. J., and Li, L. J., 2006, “A Study of Friction Behaviour
• Maximum temperature occurs on the top of the foil at the in Ultrasonic Welding 共Consolidation兲 of Aluminium,” Proceedings of the
AWS Conference: Session 7: Friction and Resistance Welding/Materials Bond-
interface of the foil and sonotrode. The temperatures at the ing Processes.
interface of the foil and substrate while smaller, are still 关28兴 ABAQUS, 2001, “Coulomb Friction in User Subroutine FRIC,” Hibbitt, Karlsson
comparable with the top foil temperature. & Sorenson Inc., Rhode Island.

041007-10 / Vol. 131, AUGUST 2009 Transactions of the ASME

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 关29兴 Li, D., and Soar, R. C., 2007, “Temperature Measurements of Monolithic 3003 关31兴 Siddiq, A., and Schmauder, S., 2006, “Interface Fracture Analyses of a Bic-
Samples During Ultrasonic Consolidation Process,” Project report, Loughbor- rystal Specimen Using Cohesive Modelling Approach,” Modell. Simul. Mater.
ough University, pp. 1–11. Sci. Eng., 14, pp. 1015–1030.
关30兴 Korn, D., Elssner, G., Cannon, R. M., and Ruhle, M., 2002, “Fracture Proper- 关32兴 Siddiq, A., Schmauder, S., and Huang, Y., 2007, “Fracture Of Bicrystal Metal/
ties of Interfacially Doped Nb-A12O3 Bicrystals: I, Fracture Characteristics,” Ceramic Interfaces: A Study Via the Mechanism-Based Strain Gradient Crystal
Acta Mater., 50, pp. 3881–3901. Plasticity Theory,” Int. J. Plast., 23, pp. 665–689.

Journal of Manufacturing Science and Engineering AUGUST 2009, Vol. 131 / 041007-11

Downloaded From: https://manufacturingscience.asmedigitalcollection.asme.org on 06/19/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use