Mato Perić

Bestprojekt Ltd.
Petrovaradinska 7 Numerical Analysis of Residual
10000 Zagreb
Croatia Stresses in a T-Joint Fillet Weld Using
Zdenko Tonković a Submodeling Technique
Faculty of Mechanical Engineering and
Naval Architecture A submodeling technique is applied in the framework of this study on a T-
Ivana Lučića 5
10000 Zagreb joint fillet weld example in order to check finite element mesh sensitivity as
Croatia well as to obtain more accurate temperatures, displacements and residual
stress fields in the weld and its vicinity where the temperature and stress
Katarina S. Maksimović
gradients are very high. The submodeling procedure of the welding
Research Associate
City Administration of the City of Belgrade process is demonstrated step-by-step. The obtained results of the
11000 Belgrade temperature, residual stress and displacement distributions correspond
very well with the experimental measurements and analytical solutions
Dragi Stamenković from the literature.
Research Assistant
Visoka brodarska škola
akademskih studija Keywords Submodeling, T-joint fillet weld, finite element analysis,
Bulevar vojvode Mišića 37 residual stress, welding distortion, Abaqus, welding simulation
11000 Belgrade

1. INTRODUCTION to speed up the calculation process that significantly

reduces the number of freedom in a finite element model.
Welding is one of the most frequently used engineering Most of the suggested solutions above are related to the
methods of joining structural components in many reduction of calculation time primarily in the processing
industrial fields. Large localised heat input during phase. For time-reduction in the pre-processing phase,
welding and subsequent rapid cooling of melted there are only limited data in the literature provided. In
materials can have as a harmful consequence permanent that sense, Seleš et al. [16] and Elmesalamy et al. [17]
residual stresses and geometrical imperfection occur– used Abaqus Plug-in Software (AWI) to speed-up the
rences in the welded structures. Such geometrical modeling of thermal loads and boundary conditions that
imperfections can cause serious problems during greatly reduce the model preparation time. Furthermore,
structure assembly, while high tensile residual stresses it is well known that the size of finite element meshes has
in combination with the workload can have a a great influence on the result accuracy and compu–
detrimental impact on its integrity and life-time [1-4]. tational time, so that the finite element calculations with
The elimination of these consequences by using refined meshes have to be done several times in a row
post-weld thermal or mechanical treatments requires [18,19] before tackling the numerical simulation of the
extended production time and causes additional finan– welding process. This approach is often unsuitable for
cial expenses. For these reasons, it is highly desirable to engineering applications, especially for models with a
know the magnitude of residual stresses and distortions large number of finite elements, since it consumes too
in the structure as early as the design phase. Due to the much computational time. For this reason, the submo–
rapid progress of computer technology in recent deling technique in this study which is based on global-
decades, the numerical calculations of residual stresses local transition is applied to check the finite element
and distortions have become an unavoidable tool for mesh sensitivity as well as to obtain finer residual stress
predicting these phenomena by shifting expensive and displacement distribution in the area of interest.
experiments to computationally based procedures [5-9]. In the submodeling technique, the results from the
The main problem that exists here is the long-time global model are interpolated onto the nodes on the
duration of numerical simulations in models with a large appropriate parts of the submodel boundary. As the
number of finite elements. submodel region has a finer mesh, the submodel can
To overcome this problem, there are many suggested provide an accurate, detailed solution. Unlike a large
solutions in the literature. Shen and Chen [10], Rong et number of papers where the submodeling capability is
al. [11], Perić et al. [12,13] and Rezaei et al. [14] used in the mechanical analysis to obtain a distribution of
proposed various solutions based on coupling shell and high stresses caused by geometric discontinuities [20], a
three-dimensional finite elements that significantly very limited number of studies have been reported in the
reduce the calculation time. Huang and Murakawa [15] area dealing with submodeling techniques in the
developed in their study a dynamic mesh refining method thermomechanical analysis of welding processes [21, 22].
This paper is organized as follows: Section 2 provides
a brief description of the submodeling application on a
Received: September 2018, Accepted: November 2018
welding simulation process. The numerical model is
Correspondence to: Mato Perić
presented in Section 3. Section 4 gives a comparison of
Bestprojekt Ltd., Petrovaradinska 7
temperatures, deflections and residual stresses obtained
10000, Zagreb, Croatia
using the submodeling technique with the results from
E-mail: [email protected]
doi:10.5937/fmet1901183P
© Faculty of Mechanical Engineering, Belgrade. All rights reserved FME Transactions (2019) 47, 183-189 183
literature as well as with the results from the global parameters are as follows: welding voltage U = 29 V,
model. Finally, some concluding remarks are given in the welding current I = 270 A, welding speed v = 400
last section. mm/min and angle of torch α = 45°. The plates are made
of SM400A carbon steel which chemical compositions
2. WELDING SIMULATION USING SUBMODELING are given in Table 1, while its thermal and mechanical
TECHNIQUE properties are given in Figs. 3 and 4.
Table 1. Chemical composition of SM400A steel [23]
The principle of the submodeling technique for the
simulation of the welding process consists of four steps Elements C Si Mn P S
and it is given in Fig. 1. In the first step, a heat transfer Mass % 0.23 - 0.56 <0.0035 <0.0035
analysis is done to obtain temperature-time history at
each node on the global model. The obtained thermal 1.4
Thermal conductivity
field is then used as a thermal load in the mechanical 1.2 Specific heat

Thermal properties
Density
analysis in the second step to obtain residual stresses 1
and displacements. The obtained temperature histories 0.8 (104 kgm-3)

on the global model boundary nodes are then transferred 0.6 (103 Jkg-1K-1)
as driven variables from the global model boundaries 0.4 (102 Wm -1K-1)
onto the submodel ones in the third step. Finally, in the 0.2
fourth step, the obtained submodel temperature histories 0
from the third step are then applied as a thermal load in 0 250 500 750 1000 1250 1500
the fourth step, while the displacement obtained from Temperature [°C]
the global model boundaries from the second step are Figure 3. Thermal properties of SM400A steel [24]
simultaneously used as driven variables. Since the
350
submodel dimensions are usually significantly less than
Yield stress (MPa)
the global ones, this allows for much denser finite 300
Mechanical properties

Modulus of elasticity (GPa)

element meshes (mesh-2) in the submodel analysis. 250 Thermal expansion coefficient (1/°C)
Poisson's ratio
200
3. NUMERICAL MODEL
150
(10-7)
100
To verify the submodel technique, a T-joint fillet weld
with experimentally measured plate deflections is taken 50 (10-2)

from the literature [23]. The dimensions of the welded 0

model with mechanical restraints are given in Fig. 2. A 0 250 500 750 1000 1250 1500
Temperature [°C]
two single pass MAG welding with no time-gap
between the weld passes is used. The welding Figure 4. Mechanical properties of SM400A steel [24]

1 Thermal analysis Driven variables 3 Thermal analysis

Globa model ( mesh-1) ( temperatures) Submodel ( mesh-2)
1st order DC3D8 elements Global.odb 1st order DC3D8 elements

Thermal field Thermal field

Global.odb Submodel.odb

2 Mechanical analysis Driven variables 4 Mechanical analysis

Global model ( mesh-1) ( displacements) Submodel ( mesh-2)
1st order C3D8I elements Global.odb 1st order C3D8I elements

Figure 1. Sheme of welding submodeling technique

*dimensions in milimeters 9
Submodel Section A-A

d
el
99

second

w
45°

300

weld

weld
first

20
6

6
217
y
G
x 25
0
C D
z nd D
C co E F
se eld 50
0
w
0
A t
firseld A 25
mechanical restraint in x , y and z direction
w
mechanical restraint in y and z direction
500
9

Global model mechanical restraint in y direction

Figure 2. Geometry of T-joint welded plates

184 ▪ VOL. 47, No 1, 2019 FME Transactions

 As it can be seen in Fig. 1, a sequentially coupled field and deformations can be neglected in low-carbon
numerical simulation of welding process is applied in steel [26,27]. Creep material behavior is also neglected
this study. It means that a full analysis consists of two because the thermal cycles during the welding are of
independent analyses: thermal and mechanical ones. In very short time duration. In doing so, the total strain
the thermal analysis, the governing equation for increments can be decomposed into three components:
transient nonlinear heat transfer is given in the form of
∂  ∂T  ∂  ∂T  ∂  ∂T  ∂T
{ } { } { }
d ε total = d ε e + d ε p + d ε th (6)
 kx  +  ky  +  ky  + Q = ρC (1)
∂x  ∂x  ∂y  ∂y  ∂z  ∂z  ∂t where {dεe}, {dεp} and {dεth} are elastic, plastic and
thermal strain increments, respectively. Furthermore, in
where kx, ky, and kz are the thermal conductivities in the all numerical simulations it is assumed that the base
x, y and z directions, respectively; T is the temperature; metal and weld filler metal have the same thermal and
Q is the heat input; ρ is the material density; C is the mechanical properties.
specific heat capacity; and t is the time, respectively. A The mesh sensitivity of the full (global) model,
general solution of Eq. (1) can be obtained by intro– using the submodeling technique, is conducted on a
ducing the following initial and boundary conditions small volume, 217 mm × 99 mm × 20 mm, in the weld
T ( x, y, z , 0) = T0 ( x, y, z ) (2) and its vicinity where the temperature gradients are very
high. The location of the submodeling is shown in Figs.
 ∂T ∂T ∂T  2 and 5. It should be noted that the submodel mesh is
 kx N x +k y N y + kz N z  + qs +
 ∂ x ∂ y ∂ z  (3) created with a very high dense finite element mesh. It is
+ h c (T − T∞ ) + hr (T − Tr ) = 0 clear, that the use of such dense finite element meshes
in the full global model can lead to thermal and mecha–
where Nx, Ny, and Nz are the direction cosine of the nical simulations which are computationally unsolvable.
normal to the boundary; hc and hr are the convection and The dimensions of the submodel and the number of
radiation heat transfer coefficients, respectively; qs finite elements of the submodel are given in Table 2.
denotes the boundary heat flux; Tr denotes the tempe– Additionally, a comparison is given between the
rature of radiation; and T∞ represents the ambient finite element number of the submodel and the equi–
temperature. Radiation heat loses can be expressed by valent volume of the same dimensions in the full global
the following equation model. All numerical simulations in this study are
performed by using the Abaqus/Standard software.
hr = σε F (T 2 + Tr2 )(T + Tr ) (4)

where σ = 5.67 × 10-8 Jm-2K-4 denotes the Stefan–

Submodel
Boltzmann constant, ε is the surface emissivity factor,
and F is the configuration factor. The total heat input
applied to the weld is given by:
ηUI
Q= (5)
VH

where η represents the heat input efficiency, I is welding

current, U is the arc voltage, and VH is the volume of the
activated set of finite elements. In the thermal analysis
of this study, the filler metal addition is simulated by
element birth and death technique. The segmented heat
flux with uniformly distributed Q = 5.22 × 1010 Jm-3s-1
per weld volume is applied and it is calculated according Global model
to Eq. (5). Additionally, the following thermal boundary
conditions are assumed: convective heat transfer Figure 5. Typical finite element mesh
coefficient k = 10 Wm-2K-1, emissivity of plate surfaces
Table 2. Dimensions and number of finite elements of
ε = 0.9 and the weld process efficiency η = 80%. In the submodel and equivalent part of global model
thermal analysis, three-dimensional 8-node solid
DC3D8 elements are used. Name Dimensions Number of finite
To speed up-the simulation process, the mechanical elements
analysis is conduced simultaneously in one step, without Global model* 217 mm × 99 mm × 1092
using the element and birth technique [25]. In the me– 20 mm
Submodel 217 mm × 99 mm × 22,320
chanical analysis, an elastic-perflectly plastic behavior
20 mm
of the material is assumed. The same finite element (*equivalent volume to submodel volume)
mesh that consists of a 19,188 finite element mesh (Fig.
5) as in the thermal analysis is used, with only the finite 4. RESULTS AND DISCUSSION
element type being converted to C3D8I. The influence
of the metal phase transformation is not considered in The temperature time-history of node E which is located
this study because its influence on the residual stress at the bottom surface of the horizontal plate (Fig. 2) for

FME Transactions VOL. 47, No 1, 2019 ▪ 185

the first 150 s after the beginning of the welding process Due to a lack of experimental measurements in [23],
is given in Fig. 6. It is obvious that the temperature the calculated longitudinal stresses are compared with
time-history of the global model and submodel are the widely used analytical solution [28]. The peak
almost identical. More detailed peak temperature com– longitudinal residual stresses is about 5% higher than
parisons after the first and second weld pass are given in the analytical solution. The width of the numerically
Figs. 7 and 8, respectively. The peak temperatures of the calculated tensile longitudinal stress zone is 66.3 mm
global model are 1% higher than in the submodel. and is the same for the global model and submodel,
Further submodeling with meshes of higher density has while the analytical calculated tensile residual stress
no additional influence on the submodel temperatures, zone is 69.2 mm. The full field longitudinal residual
and it can be stated that the temperatures converge. stress distribution of the T-joint welded plates along C-
Keeping this in mind, it can be concluded that the global D line (Fig. 2), for the global model and submodel is
mesh for the thermal analysis is properly designed. given in Fig. 10.
The obtained longitudinal residual stress (stress in 350
the welding direction) at the middle plane of the ho– Global
300 model
rizontal plate along line C-D (Fig. 2) for the global Submodel
model and submodel is shown in Fig 9. Here, it is seen 250

Residual stress [MPa]

Analytical
that the longitudinal residual stress of the global model 200 [28]
and submodel coincide very well.
150
600
Global model 100
500 Submodel 50
Temperature [°C]

0
400
-50150 200 250 300 350
300
-100
200
x - Coordinate [mm]

Figure 9. Longitudinal residual stress distribution of global

100 model and submodel at middle surface of horizontal plate
along C-D line shown in Fig. 2
0
0 25 50 75 100 125 150
Time [s]

Figure 6. Temperature time-history at node E (Fig 2.) of

global model and submodel
460
Global model
440 Submodel
420
Temperature [°C]

400

380

360

340

320

300
35 40 45 50 55
Time [s]

Figure 7. Peak temperature of global model and submodel

after first weld pass
560
Global model
550 Submodel
540
Temperature [°C]

530

520

510

500 Figure 10. Full field longitudinal residual stress distribution

490 of T-joint welded plates along C-D line shown in Fig. 2: a)
global model, b) submodel
480
115 120 125 130 The calculated transversal residual stress (stress
Time [s] parpendicular to the welding direction) at the middle
Figure 8. Peak temperature of global model and submodel plane of the horizontal plate along line C-D (Fig. 2) for
after second weld pass both, global model and submodel is shown in Fig 11. As

186 ▪ VOL. 47, No 1, 2019 FME Transactions

in the longitudinal stress comparison, it is noticeable given in Fig. 14, while their full field comparison is
that the longitudinal residual stress of the global model given in Fig. 15. As it can be seen, the calculated
and submodel coincide very well. The full field vertical displacement of the global model and submodel
transversal residual stress distribution of the T-joint correspond very well.
welded plates along C-D line (Fig. 2), for the global 90
model and submodel is given in Fig. 12. Additionaly, 80 Submodel
the thickness of the transversal residual stress profile 70 Global model

Residual stress [MPa]

along F-G line (Fig. 2) is shown in Fig. 13. Here, node F 60
is located on the lower surface of the horizontal plate, 50
and node G on the lower surface of the horizontal plate. 40
The obtained average through the thickness of the 30
transverse residual stress difference between the global 20
model and submodel is under 10%. 10
80 0
Global model
70 -10
Submodel 0 1 2 3 4 5 6 7 8 9
Residual stress [MPa]

60 y - Coordinate [mm]
50 Figure 13. Through thickness transversal residual stress
40 profile along F-G line shown Fig. 2
-3.0
30
150 200 250 300 350
20 -3.5
Experimental [23]
10
Deflection [mm]
Global model
-4.0
Submodel
0
150 200 250 300 350 -4.5
x - Coordinate [mm]
-5.0
Figure 11. Transversal residual stress distribution of global
model and submodel at middle surface of horizontal plate
along C-D line shown in Fig. 2 -5.5

-6.0
x - Coordinate [mm]

Figure 14. Deflection distributions of horizontal plate at

middle surface along C-D line shown in Fig. 2

Figure 12. Full field transversal residual stress distribution

of T-joint welded plates along C-D line shown in Fig. 2: a)
global model, b) submodel

A comparison between the vertical deflections of

Figure 15. Full vertical deflection field of T-joint welded
the horizontal plate at the middle surface along C-D plates along C-D line shown in Fig. 2: a) global model, b)
(Fig. 2) between the global model and submodel is submodel

FME Transactions VOL. 47, No 1, 2019 ▪ 187

 The calculated peak deflection for both model and analysis of welded joints, Metals, Vol. 7, No. 4, pp.
submodel is 5.6 mm.Keeping in mind that the calculated 136, 2017.
residual stresses and deflections of the global model and [7] Maksimovic M., Vasovic I., Maksimovic K.,
submodel are in good correlation and further Trišovic N., Maksimovic S., Residual life
submodeling has no influence on the results, it can be estimation of cracked aircraft structural
concluded that the global mesh for the mechanical components, FME Transactions, Vol. 46, No. 1,
analysis is properly designed. Some aspects of practical pp. 124-128, 2018.
design with respects residual life estima–tions of
[8] Sadeghi, S., et al.: Residual stress evaluation in friction
structural components, using FEM or XFEM for stress stir welding of aluminum plates by means of acoustic
analysis, are given in references [29-35]. emission and ultrasonic waves, FME Transactions, Vol.
46, No. 2, 230-237, 2018.
5. CONCLUSIONS
[9] Nguyen, K., et al.: Thermomechanical modeling of
welding and galvanizing a steel beam connection detail to
In this study, the submodeling technique is applied on a examine susceptibility to cracking, Materials
T-joint fillet welded model to check the finite element Performance and Characterization, Vol. 7, No. 2, pp. 1-
mesh sensibility and to obtain a more accurate solution 23, 2018.
of temperature, displacement and residual stress field in [10] Shen, J., Chen, Z.: Welding simulation of fillet welded
the area of interest. Special attention is given to the joint using shell elements with section integration,
investigation of the weld zone and its vicinity, where the Journal of Materials Processing Technology, Vol. 214,
thermal and residual stress gradients are very high. No.11, pp. 2529-2536, 2014
Considering the submodel dimensions to be very small [11] Rong, Y., et al.: Study of welding distortion and residual
at times, the finite element mesh can be very dense and stress considering nonlinear yield stress curves and multi-
the convergence of temperature, deformations and constraint equations, Journal of Materials Engineering
residual stress fields are much faster than on the full and Performance, Vol. 25, No. 10, pp. 4484-4494, 2016.
model in the numerical simulations. It can be concluded [12] Perić, M., et al.: Numerical analysis and experimental
that the submodeling technique is a reasonable solution investigation in a T-joint fillet weld, Materials and
for this purpose. Design, Vol. 53, pp. 1052–1063, 2014.
[13] Perić, M., et al.: Comparison of residual stresses in butt-
ACKNOWLEDGMENT welded plates using software packages Abaqus and
Ansys, Scientific Technical Review (ISSN 1820-0206),
The herein investigations described are part of the Vol. 60 No. 3–4, pp. 22–26, 2010.
project “Centre of Excellence for Structural Health” [14] Rezaei, A., et al.: Comparison of full 3D, shell/3D and
(CEEStructHealth) supported by the EU under contract inherent deformation numerical methods for prediction of
IPA2007/HR/16IPO/001-040513. out-of-plane welding-induced distortion, Journal of the
Brazilian Society of Mechanical Sciences and
REFERENCES Engineering, Vol. 40, No. 6, pp. 287, 2018.
[1] Oh, S.H., et al.: Evaluation of J-groove weld [15] Huang, H., Murakawa, H.: Thermo-mechanical analysis
of line heating process by an efficient and accurate multi-
residual stress and crack growth rate of PWSCC in level mesh refining method, Marine Structures, Vol. 49,
reactor pressure vessel closure head, Journal of pp. 239-255, 2016.
Mechanical Science and Technology, Vol. 29, No.
[16] Seleš, K., et al.; Numerical simulation of a welding
3, pp. 1225-1230, 2015. process using a prescribed temperature approach, Journal
[2] Sedmak, S., et al.: Influence of multiple defects in of Constructional Steel Research, Vol. 45, pp. 49-57,
welded joints subjected to fatigue loading 2018.
according to SIST EN ISO 5817:2014, Structural [17] Elmesalamy, A.S., et al.: Measurement and modelling of
Integrity and Life (eISSN 1820-7863), Vol. 18, No. the residual stresses in autogenous and narrow gap laser
1. pp. 77-81, 2018. welded AISI grade 316L stainless steel plates,
International Journal of Pressure Vessels and Piping, Vol.
[3] Predan, J., et al.: Fatigue crack propagation in 147, pp. 64–78, 2016.
threshold regime under residual stresses,
[18] Teng, T.L., et al.: Effect of welding sequences on
International Journal of Fatigue, Vol. 32, No. 7, pp.
residual stresses, Computers and Structures, Vol. 81, pp.
1050-1056, 2010. 273-286, 2003.
[4] Chen, Z., et al.: Stress intensity factor-based [19] Vacev, T., et al.: Plasticization conditions and strength of
prediction of solidification crack growth during a welded beam-to-column connection in case of various
welding of high strength steel, Journal of Materials detailing, Engineering Structures, Vol. 165, pp. 11-26,
Processing Technolology, Vol. 252, pp. 270-278, 2018.
2018. [20] Marenić, E., et al.: On the calculation of stress intensity
[5] Lostado R.L., et al.: Using genetic algorithms with factors and J-integrals using the submodeling technique,
multi-objective optimization to adjust finite element Journal of Pressure Vessel Technology, Vol. 132, pp.
041203-1, 2010.
models of welded joints, Metals, Vol. 8, No. 4, pp.
230, 2018. [21] Perić, M., et al.: An engineering approach for a T-joint
fillet welding simulation using simplified material
[6] Lostado, R.L., et al.: Residual stresses with time- properties, Ocean Engineering, Vol. 128, pp. 13-21,
indenpendent cyclic plasticity in finite element 2016.

188 ▪ VOL. 47, No 1, 2019 FME Transactions

[22] Bonifaz, E.A., Submodeling simulations in fusion welds, duralumin spar fatigue life estimation, Technical Gazette,
Journal of Multiscale Modelling, Vol. 4, No. 4, pp. 1-14, Vol. 19(3), pp. 557-562, 2012.
2012. [33] Rasuo, B., Grbovic, A., Petrasinovic, D.: Investigation of
[23] Deng, D., et al.: Determination of welding deformation in fatigue life of 2024-T3 aluminum spar using Extended
fillet-welded joint by means of numerical simulation and Finite Element Method (XFEM), SAE International
comparison with experimental measurements, Journal of Journal of Aerospace, 6 (2013-01-2143), pp 408-416,
Materials Processing Technology, Vol. 183, No. 2-3, pp. 2013.
219–225, 2007. [34] Grbovic A., Rasuo B.: Use of modern numerical methods
[24] Gannon, L., et al.: Effect of welding sequence on residual for fatigue life predictions, in: Recent Trends in Fatigue
stress and distortion in flat-bar stiffened plates, Marine Design, Branco R. (Ed.), Nova Science Publishers, New
Structures, Vol. 23, No. 3, pp. 385–404, 2010. York. Chapter 2., pp.31-75, 2015.
[25] Perić. M., et al.: A simplified engineering method for a [35] Kastratović, G., Vidanović, N., Grbović, A., Rašuo, B.:
T-joint welding simulation, Thermal Science, Vol 22, No. Approximate determination of stress intensity factor for
3, pp. S867-S873, 2018. multiple surface cracks, FME Transactions, Vol. 46, No.
[26] Deng, D., FEM prediction of welding residual stress and 1, pp 41-47, 2018.
distortion in carbon steel considering phase
transformation effects, Materials and Design, Vol. 30,
No. 2, pp. 359-366, 2009.
[27] Knoedel, P., et al.: Practical aspects of welding residual НУМЕРИЧКА АНАЛИЗА ЗАОСТАЛИХ
stress simulation, Journal of Constructional Steel НАПОНА У ЗАВАРЕНИМ ПЛОЧАМА У
Research, Vol. 132, pp. 83-96, 2017. ОБЛИКУ Т-СПОЈА КОРИШЋЕЊЕМ
[28] Chen, B.Q., Soares G.C.: Effects of plate configurations ТЕХНИКЕ ПОДМОДЕЛИРАЊА
on the weld induced deformations and strength of fillet-
welded plates, Marine Structures, Vol. 50, pp. 243-259, М. Перић, З. Тонковић, К.С. Максимовић,
2016. Д. Стаменковић
[29] Grbovic, A., Rasuo, B.: FEM based fatigue crack growth
predictions for spar of light aircraft under variable У оквиру овoг рада примењена је техника
amplitude loading, Engineering Failure Analysis, Vol. подмоделирања на примеру заварених плоча у
26, pp. 50–64, 2012. облику Т-споја, како би се проверила осетљивост
[30] Boljanovic S., Maksimovic S., Computational mixed мреже са коначним елементима, као и да се добију
mode failure analysis under fatigue loadings with прецизније расподеле температура, померања и
constant amplitude and overload, ENGINEERING поља заосталих напона у завареном споју и у
FRACTURE MECHANICS, (2017), vol. 174, pp. 168- његовој околини где су градијенти температуре и
179. напона веома високи. Процедура субмоделирања
[31] Petrašinović, N., Petrašinović, D., Rašuo B., Milković D.: процеса заваривања приказана је корак по корак.
Aircraft Duraluminum Wing Spar Fatigue Testing, FME Добијени резултати расподеле температуре,
Transactions, Vol. 45, No. 4, pp 531-536, 2017.
заосталих напона и померања веома добро се слажу
[32] Petrašinović, D., Rašuo, B., Petrašinović, N.: Extended са експерименталним мерењима и аналитичким
finite element method (XFEM) applied to aircraft решењима из литературе.

FME Transactions VOL. 47, No 1, 2019 ▪ 189