72nd IIW Annual Assembly and International Conference

7–12 July 2019 Conference Proceedings

Effect of the spot welding simulation strategy on

analysis of performance in automotive parts
Mahyar Asadi1,a, Majid Tanbakuei Kashani1,b, Chris Timbrell2,c, Ramesh Chandwani2,d, Mathew Smith1,e
1Advanced Welding Engineering Services, Applus Canada - SKC Engineering, North Vancouver, Canada
2
Research & Development, Zentech International Limited, London, United Kingdom
[Link]@[Link], [Link]@[Link], cchris@[Link], dramesh@[Link],
[Link]@[Link]

Abstract
Spot weld plays a critical role in the integrity of automotive and fatigue durability of the body over the course of service.
Numerical analysis is a common tool to model the behaviour of spot welds in automotive under both service and crash
loading. However, the modeling strategy can significantly change the result of analysis and therefore the final engineering
decision. Authors observed that modelers typically define the spot welds as a mechanical tie between two plates and ignore
the local effect from welding process that changes the material properties in weld nugget and the surrounding Heat Affected
Zone (HAZ). However, spot welding leads to property evolution to higher strength and toughness and lower ductility in the
weld nugget than parents metal’s property. This evolution is opposite in HAZ with lesser strength and toughness and higher
ductility than parents metal’s property. This local effect at the weld region changes the stress-strain behaviour of the structure
and it becomes more significant when dealing with fatigue life prediction because fatigue is a local phenomenon and the
analysis of fatigue uses the local stress and strain for life calculation. Therefore the simulation of welding process needs to be
part of the analysis for making a realistic decision. On the other hand, welding modeling could take longer CPU time for a
large number of welding involved in the automotive body such that it is not feasible to model every individual welding
process. Here, authors compared two modeling scenarios where one defines the weld as mechanical tie vs. the other one that
includes the modeling of welding process and evolution of material properties in the weld metal and HAZ. The main criteria
for the comparison are the fatigue durability and crack evolution in the weld nuggets and the structure. We also proposed a
methodology to avoid the repetitive weld modeling for every weld nuggets and using the result of characteristic welds to
cover all welding locations in the structure.
Keywords
simulation, HAZ, welding, automotive, crash
 72nd IIW Annual Assembly and International Conference
7–12 July 2019 Conference Proceedings

defined by levels of peak force, displacement and energy.

1. Introduction Another simple solution is a model that represents spot weld
as beams [6] where the beams connect shell elements. The
Spot resistance welding is the critical joining process in
diameter of the beams was formulated by dividing the area of
the automotive body-in-white assembly with a car
the cross-section of the representative spot weld with the
comprising thousands of spot welds that connect parts with a
number of beams. If the correlation between experiment and
wide-ranging variety of configuration involving different
model was built correctly, these approaches could investigate
thickness, materials, numbers of sheet metals, size of welds,
a general failure of spot welds (i.e. fail/no fail binary
and so on. The quality of spot welds, and subsequently, the
decision) under loading conditions.
performance of the joint, depends on many parameters, for
The mechanical properties of the weld and surrounding
example, Dipak and Ganesh [1] summarized many
subzones are not easy to obtain due to the small size of the
parameters that can affect the spot weld strength. The fusion
zones. Some techniques can experimentally simulate the
nature of spot resistance welding complicates the evaluation
thermal profile of weld and form similar microstructure but
of performance and failure for spot welded joints. Poranvari
uniform and at a larger size for measuring the properties.
and Marashi [2] correlated the fusion understanding of the
Yet, once these properties are available, defining the zones
spot welding process to the structural property of the
with these properties are time-consuming at the component
automotive component. They identified three microstructural
level with many spot welds. Therefore a typical approach is
zones namely, Fusion Zone (FZ), Heat Affected Zone
to model a single-spot-weld [7] and map or replace the other
(HAZ), and Base Metal (BM) that exhibits nonhomogeneous
welds with this single-spot-weld. This characteristic spot
microstructures and properties therefore different mechanical
weld can be one of all welds in the component. In this paper,
strength in each zone as well as strength mismatch between
we form the digital twin of a car component subjected to
the zones that form strain concentration at the weakest side.
crash where we utilized all weld zones in the analysis.
Additional to the heterogeneity, geometrical factors reduce
the load-bearing capacity of the joints for examples,
indentation, the micro-voids in the FZ and crack-like tip of
2. Automotive Component
the gap into the FZ [2]. Chabok et al. [3] separated the HAZ
region into the Coarse grain HAZ (CGHAZ), the Fine grain Fig. 1 shows the spot-welded automotive component used
HAZ (FGHAZ), and the Inter Critical HAZ (ICHAZ) and for crash analysis in this paper. The component consists of an
experimentally extracted mechanical and fracture properties omega-shaped and a plate, made of DP1000 dual-phase high-
of each region for DP1000 steel. strength steel profiles, showing a total length of 300 mm.
When evaluating a welded joint’s performance, the Both parts are bonded together at the flanges of the omega
complexity of dealing with these factors usually leads to section, through spot welds showing 8mm diameter, and
relying on physical welding trials to set the process equal spacing between them, 30 mm.
parameters and examine the performance. An alternative is
the digital twin of the process that captures the physics of
welding for simulating the performance of joint once loaded 3. Digital Spot Welding
[4]. However, constructing a proper digital twin of A full 3D model of each weld was created using in-house
automotive components, which include multiple welds, can subroutines. The subroutine uses the fying surfaces for fusion
be labour intensive and could take long CPU time to analyze. line defined by the user and assigns a heat flux by Joule’s
This disadvantage is mainly due to the local variation of law which in its purest form can be expressed by the Eq. 1.
microstructure and properties in the weld, and HAZ that
needs fine mesh and various materials definition when
building the model. Majority of modellers take the shortest
path toward the simplification of the problem and ignore any
where is the weld heat input in joules, I is the current in
local change around the weld and limit the analysis to a
amperes, R is the resistance in ohms and t is the time in
simple mechanical definition with spot weld as a mechanical
seconds. A series of cool down steps were added after the
tie between to surfaces. These models are efficient for load
welding was performed to model the returning thermal
carrying analysis at the component level where the objective
profile to ambient temperature.
is not an investigation of failures such as fatigue or crash. In
The initial temperature was 21 °C. A convection boundary
other words, failure is a local phenomenon and cannot be
condition generated a boundary flux on all external surfaces.
adequately investigated without adding local features to the
The total temperature-dependent convection coefficient is
model.
computed from Eq. 1 where T is the temperature in °C [8].
Ford Motors [5] improved the prediction by introducing
spring elements where the separation of the spot weld was

2
 72nd IIW Annual Assembly and International Conference
7–12 July 2019 Conference Proceedings

activates the Austenite grain growth by merging and

Eq.2 migrating grain boundaries; however, the existence of
Below are the properties are taken from [9] and [10] for Carbide and Nitrides precipitate blockades this migration
this welding analysis: until a critical temperature Ts (≈1100 °C) when the
• Temperature-dependent Thermal Conductivity precipitates dissolve into the Austenite matrix and allow the
• Temperature-dependent Thermal Expansion Austenite grain growth. During Cool down, the grain growth
• Temperature-dependent Heat Capacity continues below A3 when the Austenite decomposition starts
• Temperature-dependent Module of Elasticity mostly by supercooling nucleation on the boundary of large
• Temperature-dependent Yield Stress Austenite, the formation of Bainite below, Bs (≈500 °C) and
• Temperature-dependent Density Martensite below Ms (≈400 °C).
• Poisson's ratio

Fig. 2 shows the result of the simulation, including the

formation and evolution of weld nugget on the automotive
spot-welded component of interest in this paper.

Fig. 2 Formation and evolution of weld nugget (gray colour)

on the automotive spot-welded component of interest.

Two parameters of a welding thermal profile can define

Fig. 1 Automotive component used for analysis in this paper.
the microstructure of HAZ, namely, the peak temperature
and cooling rate [11]. The peak temperature primarily
4. Digitization of Microstructure Evolution depends on the distance to the fusion line and the cooling
rate depends on the overall heat sink in the joint. Using our
During the heating cycle, the equilibrium transformation welding simulation tool for spot welding, we showed that the
temperature starts forming Austenite at A1 (≈800 °C) and heat sink is approximately constant for every peak
becomes fully Austenitic above A3 (≈900 °C). During temperature on a given joint and spot welding process.
welding, the available time between A1 and A3 is short for Therefore, we used the welding simulation for forming
equilibrium transformation and therefore the Austenization thermal profiles associated with every location of HAZ and
finishes above the A3 by massive recrystallization due to the pick the typical cooling rate at every pick temperature
superheating rather than diffusion. Higher temperature

3
 72nd IIW Annual Assembly and International Conference
7–12 July 2019 Conference Proceedings

without losing a substantial precision of predicting the final 6. Digital Crash Testing Set-up.
microstructure.
This assumption simplifies the microstructure models as a A finite element model representing the test configuration
function of peak temperature such that a relationship was is created using LS-Dyna. The DP1000 steel is modeled
created between the final microstructure and the trajectory of using shell elements with the corresponding plate thickness.
the peak temperature for the critical transformation It is defined using the MAT24 material model and enriched
temperature. Fig. 3 shows that simulation result for ICHAZ, through the methodology. It is essential to mention that the
FGHAZ, CGHAZ, FZ regions that formed during our permanent effects produced in the material due to a cold
welding process. Table 1 summarizes the mechanical forming procedure on the omega-shaped beam are introduced
properties of each region used in the analysis [12]. in the model.

Fig. 3 Simulation result that defines each region of HAZ.

Microstructure Elasticity Hardness Yield UTS Strain

Region (GPa) (HV) (MPa) (MPa) at UTS
FZ 234 510 1175 1350 0.07 Fig. 4 Axial impact test set-up on a DP1000 steel beam used as
CGHAZ 226 450 1514 1740 0.08
a validation method.
FGHAZ 239 420 1940 2000 0.1
ICHAZ 228 390 1159 1212 0.06
Tempered 219 280 662 1000 0.05 For this cold forming effect, an initial plastic strain is
Unaffected 219 290 662 1000 0.05 introduced in the regions defined by the profile edges,
Tab. 2 Mechanical properties of each region used in the obtained from the simulation of the manufacturing process.
analysis. The definition of these regions with respect to the rest of the
beam profile can be observed in Fig. 5. Spot welds are
modeled using a combination of solid elements representing
5. Experimental Crash Testing Set-up. the nugget, and a crown of shell elements representing the
HAZ. Material characterization of these elements has been
An axial impact test of a steel beam subjected to different
carried out using IDIADA HAZ methodology, which is
velocities is used in this work as a validation method for the
explained further on. Weld seams joining the beam ends to
presented spot weld modeling methodology. A schematic
the vertical faces of the testing device are modeled by shell
view of the test configuration can be found in Fig 4. Samples
elements, defined by the same material as the one used in
are tested under different speeds ranging from 10 to 40 km/h,
spot weld nugget.
driven by a mass of 283 kg. The objective of the test is to
Four impact tests using different velocities are performed,
study the correlation of the experimental force versus
comparing in every case, the performance of the two
displacement curves with those obtained by simulation, using
proposed HAZ discretization scales.
different approaches toward spot weld characterization and
modeling.

4
 72nd IIW Annual Assembly and International Conference
7–12 July 2019 Conference Proceedings

A comparison of the geometrical definition of a generic

spot weld on the mesoscale and automotive scale is shown in
Fig. 6. As can be seen, the most noticeable difference
between the two exposed geometries lies in the discretization
of the HAZ. One of the main drawbacks that the analyst
faces at the time of setting up a numerical model is the
definition of an appropriate time step for the simulation,
limited by the Courant stability criterion. In explicit
simulations, mass scaling of individual elements is a
Fig. 5 Schematic representation (left) and FEM model (right) common practice, carried out to increase the time step
of the axial impact sample. sufficiently. The added mass is often assumable when it is
relatively small and distributed over the entire geometry, or
large but concentrated in non-critical areas. Also, it can be
dismissed when it takes place in quasi-static simulations in
7. Digitization of Spot-Welds for Crash Simulations which kinetic energy is minimal compared to internal energy.
Each of the microstructure regions (i.e. FZ, CGHAZ, However, in many cases, mass scaling can negatively affect
FGHAZ, ICHAZ) shows different mechanical properties; the physical behaviour of the model. It is, therefore, a task of
therefore, spot weld modeling should represent such the analyst to evaluate the effects that this measure can have
properties accurately without conditioning the original mesh on the simulation results.
size and time step.
Spot weld modeling of components for automotive crash
simulations is performed using LS-Dyna software. This
modeling is done using two different approaches – namely,
mesoscale and automotive scale – depending on the detail in
the definition of the HAZ. In both scales, an aggregation of
solid elements is used for the representation of the spot weld
nugget. This part is attached to the surrounding base material Fig. 6 Detail view of the FEM definition of a spot weld using
through a crown formed by shell elements representing the the two proposed scales: mesoscale (left) using element size from
different parts of the HAZ, as can be seen in Fig. 6. The microstructural analysis of spot welding process, and automotive
generation of these elements is done over the contact scale (right) using an element size according to the discretization
surfaces of the component, with the help of the connection of surrounding base material.
manager tool of ANSA pre-processor.
Both in the case of mesoscale and automotive scale In this sense, the localized mass increase in the tested
models, spot weld nugget is represented by four hexahedra, components was checked during this work. The mesoscale
whose bases together form an octagon in which the distance model serves as a subsequent step after the micro-definition
between opposite vertices is equivalent to the nominal of spot welds described in previous sections. This approach
diameter of the spot weld. The solids from the nugget are is used to characterize the different zones of the HAZ, but it
defined by a thickness equal to the separation between the is not useful in complex cases such as Full Vehicle crash
middle planes of the sheets since these are modeled by shell simulations. In these cases, it is fundamental to minimize the
elements. This fact must be taken into account when defining addition of mass to the whole model, although spot welds are
the elasticity of the nugget. In the mesoscale model, the generally dispersed throughout a vehicle body-in-white. The
width of the shell crowns that define both zones of thermal definition of a macroscale model for spot welds is then
affectation is directly taken from the results of the performed mandatory. It gathers the properties of the different zones of
microanalysis, explained above. ICHAZ has been neglected the HAZ in a single crown of shell elements of similar size to
in the numerical models, due to its relatively small size and those that define the base material. This approach allows
insignificant influence in the obtained results. The whole keeping a minimum time step of 0.5˟10-4 ms without an
spot weld, consisting of two sub-zones of thermal affectation excessive introduction of added mass.
(FCHAZ, CGHAZ) plus the nugget, is defined using the LS- The starting point of the characterization process of the
Dyna material model MAT24 spot weld regions is the material card of the unaffected steel,
MAT_PIECEWISE_LINEAR_PLASTICITY, which is also applying a series of transformations based on the results
used to model the physical behaviour of the unaffected base arising from the microstructural analysis. LS-Dyna material
material. model MAT24 represents an elasto-plastic material in which
the plasticity is dependent on the strain-rate. A subsequent

5
 72nd IIW Annual Assembly and International Conference
7–12 July 2019 Conference Proceedings

enrichment allows a detailed description of the physical definition of weld seams elements can lead to a better
behaviour of the steel. As the first step, the characterization correlation of this part of the curve.
of the plastic regime, based on an analytical approach, is
capable of capturing the post-necking behaviour during the
realization of tensile tests on any steel sample. On the other
hand, the damage initiation envelope is obtained through
analytical methods enriched by detailed simulations of such
tensile test. Finally, the evolution of failure is defined to
obtain mesh-independent behaviour.
IDIADA HAZ plasticity prediction model has obtained
plasticity curves corresponding to each zone of the spot weld.
This model is based in the Ultimate Tensile Strength (UTS)
values provided by the microstructural analysis, which
allows predicting the modified properties from the Base
Material.
Additionally, the damage initialization and the failure
envelopes at each zone are estimated by IDIADA HAZ
damage prediction model. This model generates an estimated
analytical damage initialization and the failure envelopes
based on Energy dissipation criteria.
In the automotive scale, mechanical properties are
gathered into a single crown by applying the mechanical
properties of the most fragile part of the spot weld to the
whole HAZ.

8. Crash Simulation Results

A comparison of simulation results with experimental data
is shown in Fig. 7. It can be observed that the elastic regime
of the curve is adequately captured in faster velocity
scenarios, whereas a certain lack of accuracy arises in the
case of impacts at lower velocities. A similar behaviour
observed at the peak force. For instance, simulation results
for testing at 23.4 km/h accomplish a high level of
correlation of the first part of the curve either using the
mesoscale or the automotive scale mesh. Maximum force
value of 339 kN is comparable to values of 341 kN with the
mesoscale and 333 kN with the automotive scale approach.
At lower velocities, Unless using mesoscale, the stiffness and
peak force are smaller than expected from experimental
results. For instance, at 14.1 km/h the experimental
maximum force is 358 kN, while simulations reach values of
273 kN and 270 kN for the mesoscale and the automotive
scale, respectively. Simulations also show a lack of precision
in the first energy release as the beam collapses. Since the
choice of one discretization scale or another does not have
any effect in that sense, it is feasible to consider a direct
relation between the energy released by the model and the
deformation of the weld seams at each beam end. This Fig. 7 Comparison of experimental and simulation results
relationship is confirmed by the fact that a finer under different velocities (i.e. 12.9, 14.1, 18.6 and 23.4km/h from
discretization of the mesh does not lead to an improvement top to bottom), model using two discretization levels in the HAZ.
in the quality of the results in this area. Thus, a more accurate

6
 72nd IIW Annual Assembly and International Conference
7–12 July 2019 Conference Proceedings

Fundamental phenomena to be taken into account in crash

simulations such as propagation of failure, intrusion, beam
folding and deterioration of spot welds are properly captured
by the proposed numerical models, both at meso and
automotive scales. The accuracy is even higher in the case of
lower velocities.
Also, as happens in reality, a different number of folds
appears depending on the velocity at which the sample
impacts with the rigid wall. The review of the deformed
geometry shows, compared to the experimental results, an
excellent accuracy of the model predicting complex failure
phenomena at spot welds such as unbuttoning, base material
tearing or spot weld cracking, depicted at Fig. 8 and Fig. 9.
Finally, it is essential to highlight the fact that certain aspects
of the simulation, such as the peak values or area of collapse, Fig. 9 Details of the different failure modes on spot welds that
show high sensitivity to minimum changes performed in the arise from the simulation at d = 38 mm of an axial impact test.
geometry of the model, such as the distribution of the spot
welds or weld seams length or thickness.
9. Microstructure Dependency vs. Mechanical Tie.
This section compares the results obtained using the
modeling approach described in this paper and the classical
approach where the spot weld nugget is represented using an
aggregation of solid elements, and the HAZ properties are
not taken into account. The size of the nugget in the classic
approach is identical to that used in the proposed
methodology. The material used to define the nugget, in this
case, is the LS-Dyna MAT100
MAT_SPOTWELD_DAMAGE-FAILURE. The bond
between these solids and the beam flanges is made through
tied-type contacts, as can be observed in Fig. 10.
Definition of the base material, weld seams as well as
boundary conditions is identical to that used in the precious
models presented. A comparison of the results is shown in
Fig. 11. It can be seen that the section of the curves
corresponding to the initial deformation and collapse of the
beam are similar because, as mentioned, modeling of the
base material and weld seams is equivalent in both cases.
The most noticeable difference between the compared
approaches takes place in the damaged regime of the curve.
When comparing the damage propagation across the
Fig. 8 General view of an axial impact test, showing its
component, it is observed that in general, the curves obtained
deformed geometry in different stages of the simulation,
by classical approach overestimate the force supported, and
further away from the experimental values. As a
consequence, classical modeling predicts lower intrusion
values at all velocities, which is, as mentioned above, a key
factor to be taken into account when performing automotive
crash simulations.

7
 72nd IIW Annual Assembly and International Conference
7–12 July 2019 Conference Proceedings

Fig. 10 Detail view of a spot weld using a traditional approach.

Compared to the traditional way of modeling spot welds –

using beams or solid elements attached to the base material
through tied contacts – the approach presented in this work
allows a more precise definition of the different material
properties from each of the entities that form the spot weld.
Thus, more accurate simulations compared to reality are
obtained. In this regard, the deformed geometry can represent
the complex failure modes that are susceptible to appear
during crash testing of components joined with spot welds,
including phenomena like nugget cracking, unbuttoning or
base material tearing.

Fig. 11 Comparison of experimental and simulation results of

the axial impact test performed on a DP1000 steel beam under
different velocities, modeled using the proposed HAZ definition
from microstructural analysis and a traditional approach.

10. Conclusion.
Crash performance evaluation of the welded joint relies on
the level of welding details considered. Defining the weld as
a mechanical tie between two sheet metals is the fastest and
most straightforward approach but remains blind to any local
variation of microstructure and properties in the weld, and
HAZ. These models can give insight toward the load
carrying capacity of welded components; however, failure
analysis of crash cannot be investigated adequately without
adding local weld properties to the model. On the other hand,
constructing a full digital twin of automotive components,
which include multiple welds, can be less practically feasible
because of fine mesh and various materials definition in
welds and HAZs. With the current power of computation and
automation in meshing, evidence recommends modeling a
single-spot-weld that can characterize many welds then use it
for other welds. The thicknesses, material combination, and
heat sink around the weld can be useful criteria for defining
the characteristic weld. While this approach saves time in the
evaluation, it builds weld details in the evaluation for
capturing local performances, including failure and fractures.

8
 72nd IIW Annual Assembly and International Conference
7–12 July 2019 Conference Proceedings

Acknowledgements Science, vol. 54, pp. 1703-1715, 2019.

The authors sincerely acknowledge the support by the team

of Simulation at Applus IDIADA (Santa Oliva, Spain) with
special thanks to Eduardo Martin, Alfredo Alameda, Sudip
Chandratre, Monica Cartanya, and Ines Lama for their
expertise and assistance throughout all aspects of this work
and for their help in conducting high quality crash
simulations.

References

[1] V. P. Dipak and A. S. Ganesh, "A Review on Effect of Spot

Weld Parameters on Spot Weld Strength Weld Strength,"
International Journal of Engineering Development and
Research, vol. 3, no. 1, 2014.
[2] M. Pouranvari and P. Marashi, "Critical review of automotive
steels spot welding: process, structure and properties,"
Science and Technology of Welding and Joining, vol. 18,
no. 5, pp. 361-403, 2013.
[3] A. Chabok, E. GaliGalinmoghaddam, J. De Hosson and Y. T.
Pei, "Micromechanical evaluation of DP1000-GI dual-
phase high-strength steel resistance spot weld," J Mater
Sci, vol. 54, p. 1703–1715, 2019.
[4] O. Andersson, D. Semere, A. Melander, M. Arvidsson and B.
Lindberg, "Digitalization of Process Planning of Spot
Welding in Body-in-white," Procedia CIRP, vol. 50, pp.
618-623, 2016.
[5] O. Faruque, N. Saha, K. Mallela, T. Tyan and C. Madasamy,
"Modeling of Spot Weld under Impact Loading and Its
Effect on Crash Simulation," in SAE International
published, 2006.
[6] A. S. AJALAKSHMI and T. SHAFIQ, Modeling of Spot
Weld with Failure, Gothenburg, Sweden: Master’s thesis
in Automotive Engineering, Department of Applied
Mechanics, CHALMERS UNIVERSITY OF
TECHNOLOGY, 2017.
[7] H.-Y. Hwang and N. Q. Nghiem, "Spot Weld Property
Testing and Simulation Modeling with Failure Criteria,"
Applied Mechanics and Materials, Vols. 284-287, pp.
198-203, 2013.
[8] J. Goldak and M. Akhlaghi, Computational Welding
Mechanics, ISBN-10: 0-387-23287-7: Springer, 2005.
[9] N. J. Den and Ujil, "Thermal and Electrical Resistance in
Resistance Spot Welding," in 17th International
Conference Computer Technology in Welding and
Manufacturing, Cambridge, United Kingdom, 2008.
[10] "Characteristics of multi-pass narrow-gap laser welding of
D406A ultra-high strength steel," J of Materials
Processing Technology, vol. 270, pp. 168-181, 2019.
[11] J. C. Ion, Modeling the Microstructural Changes in Steels due
to Fusion Welding, PhD Thesis: Lulea University of
Technology, 1984.
[12] A. Chabok, E. Galinmoghadam, J. T. M. De Hosson and Y. T.
Pei, "Micromechanical evaluation of DP1000-GI dual-
phase high-strength steel resistance spot weld," J Mater