You are on page 1of 21

International Journal of Fatigue 139 (2020) 105705

Contents lists available at ScienceDirect

International Journal of Fatigue


journal homepage: www.elsevier.com/locate/ijfatigue

State-of-the-art review of peak stress method for fatigue strength assessment T


of welded joints
Giovanni Meneghetti , Alberto Campagnolo

Department of Industrial Engineering, University of Padova, 35131 Padova, Italy

ARTICLE INFO ABSTRACT

Keywords: In the fatigue design framework of welded structures, the peak stress method (PSM) is a rapid engineering, FE-
Fatigue based technique for estimating notch stress intensity factors (NSIFs). This review provides a theoretical back-
Welded joints ground and applies the PSM to assess the fatigue strength of arc-welded joints made of aluminium alloys or
Peak stress method (PSM) structural steels, which are subjected to uniaxial and multiaxial fatigue loads. Because two fatigue design curves
Strain energy density (SED)
are available for each class of materials, a criterion for selecting the appropriate curve depending on the local
Coarse mesh
biaxiality of the stress state is established.

1. Introduction relevant to pure axial and torsion loading, respectively. More specifi-
cally, Eurocode 3 and 9 suggest the use of an equivalent stress range
In the fatigue design framework of welded structures made of alu- based on the Palmgren–Miner linear damage summation rule [1,2],
minium alloys and structural steels, current international standards and whereas the IIW recommendations [3] suggest the use of an equivalent
recommendations [1–3] suggest several methods, namely nominal stress range based on the Gough–Pollard multiaxial criterion.
stress, hot-spot stress, notch stress, and linear elastic fracture mechanics According to literature, criteria based on local quantities such stress,
(LEFM) approaches. All methods assume a linear elastic material be- strain, or strain energy are considered reliable for fatigue assessments of
haviour. More specifically, the nominal stress approach involves solid- welded structures, particularly for complex joint geometries or loading
mechanics-based stress calculations and does therefore not consider any conditions [4,6–8]. The most widely applied criteria in technical lit-
local stress concentrations. The fatigue assessment of the analysed joint erature are based on the notch SIFs (NSIFs) [9,10], averaged strain
is performed by comparing the nominal stress with the relevant design energy density (SED) [11,12], notch stress approach [4,5,13,14], cri-
category, which depends on the overall geometry and loading condi- tical plane models [6,7,15,16], and theory of critical distances (TCD)
tion. The hot-spot stress approach involves a linear stress extrapolation [7,16–21]. A recent comprehensive comparison of several multiaxial
at the weld toe, which can be realised by post-processing a finite-ele- criteria for the fatigue design of welded structures was presented in
ment (FE) model or by performing strain gauge measurements. It is [22].
worth noting that hot-spot stress includes structural stress concentra- Regarding local approaches based on NSIFs, the peak stress method
tion effects; therefore, this method requires fewer fatigue design curves (PSM) [23–28] is a rapid engineering FE-oriented tool for readily
than the nominal approach. Moreover, the notch stress approach [3] evaluating the NSIF parameters with 2D or 3D coarse FE analyses. The
involves the modification of the weld toe and weld root geometries by theoretical background of the NSIF and averaged SED local approaches
an introduction of a fictitious 1 mm notch radius in place of the actual will be summarised in the next section. The objectives of the present
weld toe and weld root profiles [4,5]. The LEFM-based approach [3] is review paper are as follows:
based on the range of the stress intensity factor (SIF) acting on the tip of
a propagating crack; it is used as the crack driving force in the Paris’ • summarising the PSM and its 2D and 3D applications for arc-welded
law. joints made of aluminium alloys and structural steels, which are
For fatigue assessments of welded structures subjected to multiaxial subjected to both uniaxial and in-phase as well as out-of-phase
loadings, design codes and recommendations [1–3] suggest interaction multiaxial fatigue loadings.
equations according to which the equivalent design stress composes the • introducing a criterion for selecting the proper PSM-based fatigue
normal and shear stress components related to the fatigue strength design curve for a certain class of materials based on the local


Corresponding author.
E-mail address: giovanni.meneghetti@unipd.it (G. Meneghetti).

https://doi.org/10.1016/j.ijfatigue.2020.105705
Received 4 April 2020; Received in revised form 9 May 2020; Accepted 12 May 2020
Available online 25 May 2020
0142-1123/ © 2020 Elsevier Ltd. All rights reserved.
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Nomenclature cylindrical coordinate system


σθθ,θ=0,peak linear elastic opening (mode I) peak stress calculated at
a reference dimension for selecting the maximal FE size d notch tip by FE analysis based on PSM
for PSM application τnom nominal shear stress
cwi coefficient accounting for mean stress effect of i-th loading τrθ,θ=0,peak linear elastic in-plane shear (mode II) peak stress cal-
mode (i = 1, 2, or 3) culated at notch tip by FE analysis based on PSM
d average FE size of adopted mesh pattern τrz, τθz out-of-plane shear stress components in cylindrical co-
e1, e2, e3 coefficients for calculating W̄ as function of 2α and ν ordinate system
E Young’s modulus of material τθz,θ=0,peak linear elastic anti-plane shear (mode III) peak stress
fw1, fw2, fw3 coefficients for calculating σeq,peak calculated at notch tip by FE analysis based on PSM
K1, K2, K3 NSIF parameters relevant to opening, sliding, and tearing
loading modes Abbreviations
K FE ,K FE ,K FE non-dimensional NSIF parameters based on PSM
R0 size of structural volume in which the SED is averaged EBW Electron beam welding
r, θ, z coordinates of cylindrical reference system FCAW Flux-cored arc welding
W̄ averaged SED FE Finite element
FEM Finite-element method
Symbols GMAW Gas metal arc welding
GTAW Gas tungsten arc welding
2α opening angle of considered sharp notch LEFM Linear elastic fracture mechanics
Δ range of cyclic quantity (maximum minus minimum) MAG Metal-arc active gas
ϕ phase angle between combined fatigue loadings MIG Metal-arc inert gas
λ1, λ2, λ3 stress singularity degrees relevant to opening, sliding, and NSIF Notch stress intensity factor
tearing loading modes PSM Peak stress method
Λ nominal biaxiality ratio, i.e. τnom/σnom SAW Submerged arc welding
λ local biaxiality ratio defined according to PSM SED Strain energy density
ν Poisson's ratio of material SIF Stress intensity factor
σeq,peak equivalent peak stress based on PSM SMAW Shielded metal arc welding
σnom nominal normal stress TCD Theory of critical distances
σrr, σθθ, τrθ normal and in-plane shear stress components in

biaxiality of the stress state. The criterion is validated against ap- of mode I, II, and III singular stress fields:
proximately 330 and 1300 experimental fatigue results relevant to 1
K1 = 2 ·lim[( ) = 0 ·r
1]
aluminium and steel joints, respectively. r 0 (1a)

K2 = 2 ·lim[( ) 1
= 0 ·r
2]
r 0
r
(1b)
2. Theoretical background
K3 = 2 ·lim[( 1
z ) = 0 ·r
3]
(1c)
2.1. Notch stress intensity factors (NSIFs) r 0

These equations are based on a cylindrical coordinate system


According to the local approach for fatigue design based on the NSIF centred at the V-notch tip; the z-direction is tangential to the notch tip
parameters [4,9,10], the worst-case geometry is assumed for the weld line, the θ-direction originates from the notch bisector line and r is the
toe and weld root of the joint; they consist of a sharp V-notch with null radial direction. The linear elastic local stress components σθθ, τrθ, and
tip radius (ρ = 0) and opening angles of 135° and 0°, respectively τθz close to the notch tip (r → 0) and along the notch bisector line
(Fig. 1). Owing to the assumption of a sharp V-notch geometry, external (θ = 0) are calculated, as shown in Fig. 1b. Moreover, parameters λ1,
loads generate singular stress fields at the weld toe and weld root. In the λ2, and λ3 are the stress singularity degrees [31,32] that depend on the
framework of linear elasticity, the NSIF parameters quantify the in- opening angle 2α of the V-notch. Values of λ1, λ2 and λ3 referred to
tensity of these stress distributions and are therefore useful parameters certain opening angles, namely 2α = 0°, 90°, 120°, and 135°, have been
for assessing the fatigue strength of sharp V-notches [8]; similarly, SIFs reported in Table 1. It is worth noting that mode II stresses are not
are useful for assessing the fatigue strength of sharp U-shaped notches singular at notch opening angles of 2α > 102° [31,34]; therefore, λ2 is
[29,30]. It is worth noting that NSIFs are local stress parameters. not listed in Table 1 for 2α = 120° and 135°.
Therefore, they can be used to estimate the lifetime spent to initiate and
propagate a fatigue crack inside a small material volume ahead of the 2.2. Averaged strain energy density (SED) approach
V-notch tip; the NSIFs are the leading terms that control the stress
fields. Lazzarin et al. [11,12,35] introduced a local approach for the fa-
Fig. 1a shows an example of a typical partial-penetration tube-to- tigue strength assessment of notched components, which assumes that
flange welded detail subjected to multiaxial bending and torsion fatigue the average value of the SED calculated within a material structural
loadings; Fig. 1b presents the mode I, II, and III stress components at the volume is the damage parameter. The following assumptions on the
weld toe, while the local stresses acting on the root side are not ex- structural volume were made:
plicitly presented in the figure.
Regarding mode I and II loadings, Williams [31] first derived ana- • its size depends on the material;
lytically the linear elastic local stress distributions ahead of the tip of • it encloses the stress singularity locations of the structure; dealing
sharp V-notches. Afterwards, Qian and Hasebe [32] derived the sin- with welded joints, weld roots, or weld toes (see Fig. 1c);
gular stress fields under mode III loading, and Gross and Mendelson • it has a circular shape with radius R [35], as shown in Fig. 1c.
0
[33] defined the NSIF parameters with Eq. (1) to quantify the intensities

2
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Fig. 1. (a) Assumptions of the NSIF-based approach in fatigue design of welded joints referring to a partial-penetration tube-to-flange welded joint under combined
bending and torsion fatigue loading. The sharp V-notch opening angle 2α is typically 0° at the weld root and 135° at the weld toe. (b) Cylindrical reference system
(r,θ,z) centred at the weld toe and local stress components. (c) The structural volume of radius R0 centred at the weld toe or at the weld root according to the averaged
SED approach.

Table 1
Parameters depending on the notch opening angle 2α.
2α (°) λ1 (a) λ2 (a) λ3 (a) Aluminium ν = 0.33(b) Steel ν = 0.3(b)

e1 e2 e3 e1 e2 e3

0 0.500 0.500 0.500 0.125 0.337 0.423 0.134 0.341 0.414


90 0.545 0.909 0.667 0.138 0.168 0.318 0.146 0.168 0.310
120 0.616 – 0.750 0.124 – 0.282 0.130 – 0.276
135 0.674 – 0.800 0.113 – 0.265 0.117 – 0.259

a
Values from [12].
b
Values calculated under plain strain conditions.

3
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

For a partial-penetration tube-to-flange joint subjected to multiaxial e1 K1


2
e2 K2
2
e3 K3
2

bending and torsion loadings (Fig. 1a), a mixed mode I + II + III W̄ = c w1 + c w2 + c w3


E R 10 1 E R 10 2 E R 10 3 (2)
loading condition is generated at the weld root; at the weld toe, mixed
mode I + III loading occurs, and mode II is not singular for Eq. (2) assumes that the strain energy contribution of higher-order,
2α = 135° > 102°. At the weld root, the averaged SED can be written non-singular stress terms are negligible; E represents the material
as a function of the ranges of the NSIF terms, namely ΔK1, ΔK2, and ΔK3 modulus of elasticity, and the parameters e1, e2, and e3 depend on the
[12]: sharp notch geometry and material, through the opening angle 2α and
Poisson’s ratio ν. The values of e1, e2, and e3 [12] referred to some
opening angles, namely 2α = 0°, 90°, 120°, and 135°, and to the

Fig. 2. FE models to apply the PSM according to Eqs. (4a)-(4c) for a partial-penetration tube-to-flange welded joint under combined bending and torsion loading
using: (a) 2D 4-node plane elements, (b) 3D 8-node brick elements, (c) 3D 4-node or (d) 3D 10-node tetra elements. For the geometry and loading conditions under
consideration the most effective strategy to apply the PSM is reported in (a).

4
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Poisson’s ratios ν = 0.33 (aluminium alloys) and 0.3 (structural steels), subjected to mode I loading and made of
as shown in Table 1. Furthermore, R0 is the structural volume size,
which Lazzarin et al. determined by equating the averaged SEDs at 5 • aluminium alloys, R is 0.12 mm for joints that exhibit failures at the
0
million loading cycles evaluated from experimental tests of two master weld toe and weld root;
cases [11,36]: (i) ground arc butt-welded (i.e. unnotched welded ma- • structural steels, R is 0.28 or 0.36 mm for joints that exhibit failures
0
terial) and (ii) cruciform arc-welded joints exhibiting failures at the at the weld toe or weld root, respectively. To be safe, Lazzarin at al.
weld toe (2α = 135°) or weld root (2α = 0°) (i.e. notched welded [11] suggested to calculate the averaged SED by adopting a struc-
material). More in detail, R0 was calibrated for arc-welded joints tural volume size R0 = 0.28 mm, regardless of the notch opening

Fig. 2. (continued)

5
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Fig. 2. (continued)

angle and loading mode. In addition, Fischer et al. [37,38] proposed 1 + R2i
if stress relieved and 1 Ri 0,
a slightly different size (i.e. R0 = 0.32 mm) after evaluating mis- (1 R i) 2

alignment effects. c wi (R i) = 1 R2i


if stress relieved and 0 Ri < 1,
where i
(1 R i) 2

Finally, parameters cwi account for the mean stress effect and de- 1 if as welded for any R value
pend on the load ratio Ri of the i-th mode of the loading (i = 1, 2, or 3) = 1, 2, or 3. (3)
[10]:
Concerning as-welded joints, international standards and re-
commendations [1,3] suggest that their fatigue strength does not de-
pend on the load ratio R, at least for R ≥ -0.25 and residual stresses are

6
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

tensile and medium–high in comparison to the yield strength of the mesh generation algorithm of the employed numerical software.
base material. In this context, Lazzarin et al. [11,12] suggested to use The estimation of the NSIF terms is advantageous when using the
cw = 1 when as-welded joints are investigated, regardless of the re- PSM based on Eq. (4) because (i) coarse FE meshes can be employed
sidual stress state. In contrast to the case of as-welded joints, Eq. (3) and (ii) only the linear-elastic peak stresses calculated at the notch tip
must be used for stress-relieved joints; as an example of its significance: are necessary instead of the complete set of stress distance numerical
cw is 1 and 0.5 for pulsating (R = 0) and fully reversed (R = − 1) results required by Eq. (1). Moreover, the parameters K*FE, K**FE, and
loadings, respectively. K***FE in Eq. (4) depend on the (i) element type and formulation, (ii)
Eq. (2) does not consider the phase shift between the different FE mesh pattern, and (iii) procedure employed by the numerical soft-
loading modes. However, several multiaxial fatigue results generated ware to extrapolate stresses at nodes [44].
by shifting the torsional (mode III) load with respect to the axial/ The NSIF approach and the averaged SED concept and its relation to
bending (mode I) load components have been re-analysed the PSM have been reviewed thoroughly by Radaj [45,46].
[12,27,28,39]. According to the results, the influence of the phase shift
can be distinguished, although it is not remarkable. This result was also
observed in a recent comparison of the TCD and SED approaches for the 3.1. Calibration of parameters K*FE, K**FE, and K***FE
multiaxial fatigue strength assessment of notched components [40].
Therefore, the experimental data seem to support the use of the SED The calibration of the parameters K*FE, K**FE, and K***FE was
approach based on its original formulation [12], which does not con- performed with several 2D and 3D FEs and commercial FE codes.
sider the phase shift between the different applied loadings. For a In [23,25,26], the 2D PSM (see example in Fig. 2a) has been cali-
proper calculation of the range of the averaged SED as a function of the brated for 4-node quadrilateral plane elements (PLANE 182 with K-
time-history of the local stresses, the reader is referred to [41]. option 1 set to 3 of Ansys® element library) and for harmonic elements
More details about the averaged strain energy density approach can (PLANE 25); the resulting coefficients K*FE, K**FE, and K***FE are 1.38,
be found in the extensive reviews by Berto and Lazzarin [42,43]. 3.38, and 1.93, respectively. In a Round Robin Project [44], the para-
meters K*FE and K**FE were calibrated with 2D FEs available in six
3. Peak stress method (PSM) commercial numerical codes other than Ansys®: Abaqus®, Straus7®,
MSC® Patran/Nastran, LUSAS®, HyperMesh/OptiStruct/HyperView®,
The PSM is a numerical tool for a rapid estimation of the NSIF terms and HyperMesh/Ls-Dyna/HyperView®.
K1, K2, and K3. It considers the opening, in-plane shear, and out-of- Afterwards, the PSM was extended to 3D FE models discretised with
plane shear peak stresses calculated with a linear elastic FE analysis 8-node brick elements of the Ansys® element library (SOLID 185 with K-
with a coarse mesh, as shown in Fig. 2 (tube-to-flange joint). A non- option 2 set to 3), and it was demonstrated that the K*FE, K**FE, and
dimensional coefficient was defined for each loading mode [23,25,26]: K***FE values previously derived with 2D 4-node plane elements (i.e.
1.38, 3.38, and 1.93, respectively) are still valid [24]. It might be dif-
K1 ficult to apply the PSM based on 8-node brick elements to complex 3D
K FE =
, = 0, peak · d
1 1
(4a) structures because the regular mesh patterns required by brick elements
could not be applicable to complex 3D geometries. Therefore, a main
K2
K FE = model must be analysed first by employing a free mesh of tetra ele-
r , = 0, peak · d
1 2
(4b) ments; subsequently, a submodel with simple geometry can be meshed
K3 with brick elements and analysed with the PSM (see example in Fig. 2b
K FE = and next section for more details).
z, = 0, peak · d
1 3
(4c)
However, 3D modelling of large-scale structures is growing rapidly,
where σθθ,θ=0,peak, τrθ,θ=0,peak, and τθz,θ=0,peak are the peak stresses thereby pushing the need for more and more efficient, and time-saving
defined in a local cylindrical coordinate system centred at the node at fatigue strength assessment methods. In this context, the 3D PSM has
the V-notch tip; z-direction is tangential to the notch tip line; the θ- been recently accelerated: the coefficients K*FE, K**FE, and K***FE have
direction originates from the notch bisector line and r is the radial di- been calibrated with 4-node [47] and 10-node [48] tetra elements
rection. The subscript ‘θ = 0’ highlights the direction along which local (SOLID 285 and SOLID 187 of Ansys® element library, respectively),
stresses must be evaluated; as an example, σθθ,θ=0,peak means that the which are more efficient than brick elements for discretising complex
opening stress acts in normal direction with respect to the notch bi- 3D joint geometries with free meshing techniques. Therefore, the PSM
sector, as shown in Fig. 2a. Parameter d in Eq. (4) is the average size of can be directly applied to the free-meshed main model, and an addi-
the FEs; typically, the FE analyst has to input d before running the free tional submodel is unnecessary (see Fig. 2c and 2d in comparison to

The PSM based on TETRA elements: weld toe side Fig. 3. Details of the FE model to apply the
3D PSM based on 4-node or 10-node tetra
elements (see the full models in Fig. 2c and
(a) (b)
d). (a) Number of finite elements (# FE)
16 FE 14 FE 12 FE
sharing each node located along the weld
toe line. Figures (a) and (b) highlight also
the different sizes and shapes of finite ele-
10 FE ments sharing the nodes belonging to the
weld to line.

X Z
X

Y Y Z

7
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Fig. 2b). Plane-4, Brick-8, Tetra-4, and Tetra-10 for the sake of brevity, respec-
The deficiency of a tetra element mesh is its typical irregularity, i.e. tively.
each node belonging to the notch tip line can be shared by a different
number of elements with significantly different sizes and shapes (see 3.2. Range of PSM applicability
Fig. 3). Owing to the different number of elements sharing each node,
the evenness of stress averaging from one FE node to another is de- The PSM parameters K*FE, K**FE, and K***FE have been calibrated
stroyed. Consequently, the peak stress can vary along the notch tip line under the following conditions (Table 2).
although the NSIFs are constant there. This deficiency was overcome by
defining an average peak stress value [48], which smoothens the peak 3.2.1. Notch opening angles
stress distribution along the notch tip line. More specifically, the peak A range of notch opening angles has been considered:
stress was defined as the moving average of peak stresses calculated at
three adjacent vertex nodes; therefore, the peak stress at node n = k is • for mode I and III loadings: 2α has been varied between 0° and 135°,
defined as follows: which represent typical opening angles at the weld root and weld
toe [23,24,26,44,47,48].

ij,peak,n= k 1 + ij,peak,n=k + ij,peak,n= k+1
¯ij,peak,n=k = for mode II loading: only 2α = 0° [25,47,48] and 90° [49] have
3 (5)
n= node
been investigated because they are the typical cases of a weld root
Accordingly, the PSM-parameters K*FE, K**FE, and K***FE have without or with a gap (i.e. g in Fig. 4a); the notch opening angle
been calibrated for 4-node [47] and 10-node [48] tetra elements by 2α = 135° has not been considered because mode II stresses are not
inserting the average peak stresses according to Eq. (5), i.e. ¯ , = 0, peak , singular in this case.
¯r , = 0, peak , and ¯ z , = 0, peak instead of the peak stresses σθθ,θ=0,peak,
τrθ,θ=0,peak, and τθz,θ=0,peak into Eq. (4). Two additional remarks have to 3.2.2. FE mesh pattern
be made: the PSM based on tetra elements requires that peak stresses When 4-node plane or 8-node brick elements are adopted, the FE
acting on nodes at a free surface of the analysed structure are neglected mesh pattern in the proximity of the notch tip must satisfy the following
[47,48] because their values can be affected by the distorted mesh conditions (see also Fig. 2a and b): the number of elements sharing the
pattern (Fig. 2c and 2d). Furthermore, only peak stresses calculated at node located at the singularity point must be 4 when 2α ≤ 90° (e.g. at
vertex nodes of tetra elements must be used in Eq. (5); therefore, peak the weld root 2α ≅ 0°) and 2 when 2α > 90° (e.g. at the toe side
stresses at mid-side nodes must be neglected when adopting 10-node 2α ≅ 135°). The mesh patterns in Fig. 2a and b satisfy these conditions;
tetra elements (Fig. 2d). moreover, they were automatically generated by the free mesh gen-
Table 2 summarises the PSM coefficients K*FE, K**FE, and K***FE eration algorithm available in ANSYS® FE code, after only the ‘global
calibrated with 4-node plane, 8-node brick, 4-node tetra, and 10-node element size’ parameter d without additional input parameters or spe-
tetra elements of the Ansys® element library, which have been called cial settings was used as input. However, some cases have been

Table 2
Summary of parameters K*FE, K**FE and K***FE and mesh density a/d requirements to apply the PSM with Ansys® [23–26,47–49].
Loading FE analysis PSM parameters 2α = 0° 2α = 90° 2α = 120° 2α = 135° a – root side° a – toe side°

2D/3D FE type#

Mode I 2D Plane-4 K*FE 1.38 ± 3% min{l, z} t


Eq. (4a) 3D+ Brick-8 (a/d)min 3
FE at notch tip^ 4 4 2 2
Tetra-4 K*FE 1.75 ± 22%
(a/d)min 3
FE at notch tip^ not to be checked
Tetra-10 K*FE 1.05 ± 15% 1.21 ± 10%
(a/d)min 3 1
FE at notch tip^ not to be checked

Mode II 2D Plane-4 K**FE 3.38 ± 3% 2.62 ± 10% – – min{l, z} –


Eq. (4b) 3D+ Brick-8 (a/d)min 14 10 – –
FE at notch tip^ 4 4 – –
Tetra-4 K**FE 2.65 ± 15% 2.90 ± 10% – –
(a/d)min 3 1 – –
FE at notch tip^ not to be checked
Tetra-10 K**FE 1.63 ± 20% 2.65 ± 10% – –
(a/d)min 1 1 – –
FE at notch tip^ not to be checked

Mode III 2D Plane-4 K***FE 1.93 ± 3% min{l, z} t


Eq. (4c) 3D+ Brick-8 (a/d)min 12 – – 3
FE at notch tip^ 4 – – 2
Tetra-4 K***FE 2.50 ± 15%
(a/d)min 5
FE at notch tip^ not to be checked
Tetra-10 K***FE 1.37 ± 15% 1.70 ± 10%
(a/d)min 3 3
FE at notch tip^ not to be checked

+
‘Full graphics’ option of Ansys® code must be activated when calculating peak stresses according to 3D PSM.
#
FE of Ansys® code: Plane-4 = PLANE 182 (K-option 1 set to 3) or PLANE 25, Brick 8 = SOLID 185 (K-option 2 set to 3), Tetra 4 = SOLID 285, Tetra 10 = SOLID
187.
^
number of finite elements which share the node at the notch tip.
°
l, z, t are defined in Fig. 4.

8
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

(a1)
weld root weld toe (b) weld root weld toe
z
z Mb
g=0
2l 2t
t F
F 2l
Mb

(a2) (c) weld toe


weld toe
weld root
z Mb
g z t
F l F
t
2l weld root
Mb

(d) (e)
weld toe
weld root
z weld toe
Mb
F z
2l 2t
l t

weld Mb Mt
root
F

Fig. 4. Geometrical parameters (l, z, t) required to define the size a (see Table 2) for (a) non-load-carrying fillet-welded T-joint without (a1) or with (a2) root gap g,
(b) double-sided partial-penetration butt joint, (c) single-sided partial-penetration butt joint, (d) load carrying fillet-welded cruciform joint, (e) partial-penetration
tube-to-flange welded joint.

reported in which the automatic free-mesh generation algorithm • 4-node plane (Fig. 2a) or 8-node brick elements (Fig. 2b): the
available in ANSYS® failed to create the standard mesh patterns in minimal mesh density ratio a/d is determined by mode II at the weld
Fig. 2a and b. In that case, the problem was readily fixed by clearing the root, i.e. a/d = 14; therefore, d = a/14, where a is the depth of root
wrong mesh, changing slightly the global element size d, and re-starting face l.
the mesh generation [44]. • 4-node tetra elements (Fig. 2c): the most demanding mesh density
However, there are no standard mesh patterns to comply with for requirement is mode III at the weld root, i.e. a/d = 5; therefore
the PSM based on 4-node or 10-node tetra elements; therefore, the 3D d = a/5, where a is the depth of root face l.
PSM based on tetra elements is easier to apply than the 3D PSM based • 10-node tetra elements (Fig. 2d): the mesh density is determined by
on brick elements. In this case, slightly larger scattering of K*FE, K**FE, modes I and III at the root side, i.e. a/d = 3; therefore d = a/3,
and K***FE must be accepted, i.e. ± 10% to ± 22% for tetras ( ± 3% where a = l.
to ± 10% for bricks; Table 2).
Finally, Fig. 4 presents some typical welded details and the defini-
tion of the geometrical parameters t, z, and l, which are used to define
3.2.3. Mesh density ratio a/d the reference dimension a.
The K*FE, K**FE, and K***FE values in Table 2 are valid when the
minimal mesh density ratio a/d is satisfied, where a is the reference 3.3. Evaluation of peak stresses by FE analyses according to PSM
dimension in Table 2. The minimal value of a/d depends on the adopted
FE type, loading mode, and notch opening angle. As an example, Fig. 2 The PSM has been applied to re-analyse more than 30 different joint
shows the FE mesh of the PSM applied to a partial-penetration tube-to- geometries of arc-welded details made of aluminium alloys
flange joint, which is subjected to combined bending and torsion [23,27,50,51] or structural steels [23,25–28,39,50–57] and subjected
loadings; in this case, the mixed mode I + II + III stresses occur at the to pure axial, pure bending, pure torsion, and multiaxial fatigue load-
weld root, while only mixed mode I + III stresses act on the weld toe. ings. Table 3 details the joint geometries and FE analyses with the PSM
To analyse both locations with a single FE model, the maximal adop- for different types of FE models:
table FE size d must be determined with Table 2 and by considering the
most demanding mesh density requirement between the weld toe and • For plane geometries under axial or bending loadings (see geome-
weld root for the given loading modes. Regarding the example in Fig. 2, tries (1)–(8) in Table 3), a 2D free mesh pattern of 4-node quad-
the maximal FE sizes d for different FE types are as follows: rilateral plane elements (PLANE 182 with K-option 1 set to 3 of

9
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Table 3
Joint geometries and loading conditions reanalysed according to the PSM in Refs. [23,25–28,39,50–54,56,57].

(continued on next page)

10
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Table 3 (continued)

Ansys® element library) has been employed; the resulting mesh one main plate or tube thickness. The 3D mesh pattern of the sub-
pattern is similar to that in Fig. 2a. model (Fig. 2b) has been obtained by generating a 2D free mesh of
• For axis-symmetrical welded joints under bending, torsion, or 4-node quadrilateral PLANE 182 elements with the global size d;
combined loadings (geometries (30)–(33) in Table 3), a 2D free next, the 2D mesh pattern has been extruded along the weld toe or
mesh pattern of 4-node quadrilateral harmonic elements (PLANE 25 weld root line by setting the extrusion step size equal to the global
of the ANSYS® element library) has been adopted, as shown in element size d and employing 3D 8-node brick elements (SOLID185
Fig. 2a. with K-option 2 set to 3).
• For more complex welded details (geometries (9)–(29) and • As an alternative approach, 3D joint geometries can be analysed by
(34)–(36) in Table 3), 3D FE analyses according to the PSM and adopting the PSM based on either 4-node (SOLID 285 of Ansys®
based on brick elements have been conducted. First, an FE model of element library) or 10-node (SOLID 187) tetra elements, which do
the entire joint geometry has been analysed by employing a main not require submodels; in fact, the PSM is directly applicable to the
model, and a submodel of the critical locations of the joint (i.e. the main model in this case, as shown in the examples in Fig. 2c and 2d,
weld toe or weld root) has been created by exploiting the sub- respectively. The calibration of the PSM based on tetra elements is
modelling technique available in Ansys® code. The main model has recent [47,48]; this strategy has been applied to analyse only geo-
been discretised with a 3D free mesh pattern of 10-node tetra ele- metries (34)–(36) in Table 3.
ments (SOLID 187 of Ansys® element library), as shown in the ex-
ample in Fig. 2b. Subsequently, to define the submodel, the main Two remarks have to be made here:
model has been cut at a distance from the critical location equal to

11
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

• Peak stresses σ θθ,θ=0,peak, τrθ,θ=0,peak, and τθz,θ=0,peak in Eq. (4) are possible because they combine a rapid solution and post-processing
defined in a local cylindrical coordinate system with its origin at an with very good result accuracy; in fact, the scattering of the PSM
FE node located along the V-notch tip profile; the z-direction is parameters K*FE, K**FE, and K***FE is reduced, i.e. of the order of ±
tangential to the notch tip line, the θ-direction originates at the 3% in most cases; there is only one exception: at 2α = 90° and under
notch bisector line, and r is the radial direction. When a 2D or 3D mode II, the deviation is ± 10%. When 3D FE models are necessary,
PSM is adopted for rectilinear weld toes and weld root lines, the tetra elements seem to be preferable because they accelerate the mesh
definition of a single cylindrical coordinate system is sufficient for preparation and avoid submodelling, although they introduce a slightly
calculating peak stresses at all nodes of the weld toe and weld root. larger deviation of the PSM parameters than brick elements. Finally,
However, in the most general case of 3D joint geometries with among tetra elements, 10-node tetra are more convenient than 4-node
curvilinear weld toes and weld root lines, a dedicated local cylind- tetra because the former allows the analyst to adopt coarser FE meshes.
rical coordinate system must be defined at each node because the z- In fact, the mesh density ratio a/d is 1–3 for 10-node tetra and 3–5 for
direction must change from node to node to ensure tangency to the 4-node tetra (see Table 2). Typically, more refined FE meshes of 4-node
notch tip line. tetra result in a longer computation time.
• To apply the 3D PSM with the Ansys® FE code, the ‘FULL graphics’
option must be activated to evaluate the peak stress in the post-
3.4. Equivalent peak stress for fatigue design
processing environment.
According to Eq. (2), the averaged SED can be expressed as a
Consequently, the adoption of 2D FE models is preferable when
function of the NSIF terms K1, K2, and K3, which the PSM can rapidly

Fig. 5. Master curves for fatigue design of welded joints made of aluminium alloys according to the PSM and subjected to (a) pure axial or (b) multiaxial loadings.
The scatter bands were calibrated in Refs. [23,27] on weld toe failures of joints tested in the as-welded conditions: (a) cruciform or T joints (see (1)-(3) in Table 3)
under pulsating (R = 0) axial loading, (b) tube-to-flange and tube-to-tube (see (30) and (31) in Table 3) under alternating (R = −1) bending-torsion loading.

12
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

estimate with Eqs. (4a)–(4c). Therefore, the averaged SED can be re- model; however, the equivalent peak stress defined in Eqs. (6a) and
written as a function of the relevant peak stresses. Moreover, by in- (6b) is independent of the FE size d owing to the multiplication of the
troducing the SED value of an equivalent uniaxial plane strain state (i.e. peak stresses by the relevant fwi parameters. Remarkably, when the toe
eq, peak /2E ), the equivalent peak stress can be defined as in side is analysed, mode II stresses are always non-singular because
W = (1 2) 2

Eqs. (6a) and (6b) [27,28,58]: 2α ≅ 135° > 102°; therefore, the relevant contribution in Eqs. (6a) and
(6b) vanishes.
eq, peak

= c w1· fw21 · 2
+ c w2·fw2 2 · 2
+ c w3· fw23 · 2
, = 0, peak r , = 0, peak z, = 0, peak 4. Fatigue design curves according to PSM
Plane 4 or Brick 8, (6a)
The equivalent peak stress (Eq. (6)) has been adopted in
eq, peak [23,25–28,39,50–57] to re-analyse experimental fatigue data presented
= c w1· fw21 · ¯ 2 , + c w2·f w22 · ¯r2 , + c w3· fw23 · ¯ 2z , in [6,25,52–54,59–111], which are relevant to arc-welded details made
= 0, peak = 0, peak = 0, peak
of aluminium alloys or structural steels subjected to pure axial/pure
Tetra 4 or Tetra 10. (6b) bending, pure torsion, or multiaxial in-phase and out-of-phase fatigue
Eq. (6a) is valid for 4-node plane or 8-node brick elements, while loadings.
Eq. (6b) is valid for 4-node or 10-node tetra elements, which require the In the next sections, the master curves for the fatigue design for
average peak stresses defined in Eq. (5). Parameters fwi (where i = 1, 2, aluminium and steel arc-welded joints will be presented; afterwards, a
3 represent the loading mode) of Eqs. (6a) and (6b) account for the criterion for selecting the appropriate fatigue design curve for a given
stress averaging inside the material-structural volume with size R0 class of materials as a function of the local biaxiality of the stress state
(Fig. 1c), which is R0 = 0.12 mm for aluminium joints and will be introduced.
R0 = 0.28 mm for steel joints [27,28]:
1 1 4.1. Arc-welded joints made of aluminium alloys
2e1 d
fw1 = KFE · ·
1 2 R0 (7a)
Fig. 5a presents the master fatigue design scatter band, which was
1 2 fitted on approximately 90 fatigue results from literature and concerns
2e2 d
fw2 = KFE · · only weld toe failures (2α ≅ 135°) [23]. The test data were generated by
1 2 R0 (7b)
testing welded details under as-welded conditions and pulsating (R ≈
2e3 d
1 3 0) pure axial loadings. The analysed joints were T- and load-carrying or
fw3 = KFE · · non-load-carrying fillet-welded joints (i.e. geometries (1)–(3) in
1 2 R0 (7c)
Table 3). The main plate thickness T was between 3 and 24 mm, and the
According to Eqs. (4a)–(4c) and (7a)–(7c), the peak stresses and base material was a 5000- or 6000-series aluminium alloy with a yield
coefficients fwi are functions of the average FE size d employed in the FE stress of 250–304 MPa (see [99,108–110] in Table 4). It is worth

Table 4
Materials, welding processes and testing conditions of aluminium welded joints, which have been previously analysed using the PSM in Refs. [23,27,50,51].
Ref. Joint Main Material Yield Welding Testing Load type° Λ = τnom/σnom Load Phase Failure λ Eq. (11)
geometry* thickness stress process condition# Eq. (8) ratio shift ϕ location^
[mm] [MPa] R [°]

Maddox [99] (1) 3–24 6061-T6 277–298 GMAW AW A 0 0 – WT 0


Meneghetti [108] (2) 12 5083-H3 255 MIG AW A 0 0 – WT 0
Ribeiro et al. (2), (3) 12 6061-T651 250 MIG AW A 0 0 – WT 0
[109]
Jacoby [110] (3) 12 Al Zn Mg 1 304 MIG AW A 0 0 – WT 0
Pirsic [111] (1) 6 5083 250 MIG AW B 0 0 – WT 0
Da Cruz et al. (4) 3 AlMgSi1, T6 245 TIG AW A 0 0 – WT 0
[59]
Meneghetti et al. (2) 12 5083 H321 250 – AW A 0 0.1 – WT 0
[60]
Voutaz et al. [61] (9) 11 6082 260 TIG AW B 0 0.1 – WT 0
Macdonald (10) 3 6082-T5 290 – AW B 0 0.1 – WT 0
Haagensen
[62]
Haagensen Oma (4) 6 5083-H22 250 – AW A 0 0.1 – WT 0
[63] 8
(11) 8
Sonsino et al. (5) 5–25 AlMg4.5Mn 175 GMAW SR A 0 0,-1 – WR 0
[64]
Brandt et al. [65] (6) 5 AlMg4.5Mn 175 GMAW SR A 0 0,-1 – WT 0

Kueppers Sonsino (30) 10 6082 T6 315 TIG AW B 0 −1 - WT 0


[66] MIG T ∞ - ∞
B+T 0.58 0, 90 0.42
Costa et al. [67] (31) 3 6060 T6 215 – SR B 0 0,-1 - WT 0
T ∞ - ∞
B+T 0.33–6 0 0.10 ÷ 32

* See Table 3. All joints have un-machined welds.


#
AW = as-welded, SR = stress-relieved.
°
A = axial, B = bending, T = torsion.
^
WR = weld root, WT = weld toe.

13
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

mentioning that the original scatter band reported in [23] was provided peak stress range at 2 · 106 cycles valid for aluminium joints under pure
in terms of Δσθθ,θ=0,peak, which was calculated at the weld toe with FE axial or bending loadings is presented in Fig. 5a: the resulting
analyses, where 4-node plane elements having a global element size of Δσeq,peak,A,50% is 123 MPa (50% survival probability); the inverse slope
d = 1 mm were adopted. Conversely, the scatter band in Fig. 5a has of the design scatter band of aluminium welded joints subjected to pure
been converted to the equivalent peak stress with Eq. (6a) and fw1 torsion loading was experimentally derived by Lazzarin et al.; the result
calculated with Eq. (7a) by using K*FE = 1.38, ν = 0.33, e1 = 0.113, was k = 6.5 [10]. This scatter band agrees well with approximately 30
λ1 = 0.674, R0 = 0.12 mm, and d = 1 mm; consequently, fw1 is 1.387. fatigue results from literature and regards only weld toe failures
According to Fig. 5a, the design curve can be applied to any mode I (2α ≅ 135°) generated by testing welded details under as-welded con-
loading case (axial or bending). ditions and alternating (R = −1) bending-torsion loadings. The welded
Fig. 5b shows the master scatter band, which was fitted on the fa- details were tube-to-flange and tube-to-tube welded joints (i.e. geo-
tigue data of aluminium arc-welded joints subjected to in-phase and metries (30) and (31) in Table 3); the tube thicknesses T were 10 and
out-of-phase combined bending and torsion loadings [27]. More spe- 3 mm, respectively, while the base material was 6000-series aluminium
cifically, the scatter band was defined by combining the endurable alloy with a yield stress between 215 and 315 MPa (see [66,67] in
stress range at 2 · 106 cycles derived under pure axial loading and the Table 4).
inverse slope k obtained under pure torsion loading. This definition of Finally, Fig. 5a and b present the scatter index Tσ = 1.80 for sur-
the design scatter band for multiaxial loadings is consistent with that of vival probabilities of 2.3% to 97.7% (i.e. the mean value ± two
Lazzarin et al. and the averaged SED approach [12]. The equivalent standard deviations), which is close to Tσ = 1.50 for the survival

Fig. 6. Master curves for fatigue design of welded joints made of structural steels according to the PSM and subjected to (a) pure axial or (b) pure torsion loadings.
The scatter bands were calibrated in Refs. [26,56] on weld toe failures of joints: (a) as-welded cruciform or T-joints (see (1)–(3) in Table 3) under pulsating (R = 0)
axial or bending loading, (b) stress-relieved tube-to-flange joints (see (30),(32) in Table 3) under alternating (R = −1) torsion loading.

14
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Table 5
Materials, welding processes and testing conditions of steel welded joints, which have been previously analysed using the PSM in Refs.
[23,25,26,28,39,51–54,56,57].
Ref. Joint Main Material Yield Welding Testing Load type° Λ = τnom/σnom Load Phase Failure λ Eq. (11)
geometry* thickness stress process condition# Eq. (8) ratio shift ϕ location^
[mm] [MPa] R [°]

Maddox [68] (1) 13–100 BS4360:50 360–398 MAW AW A 0 0 – WT 0


Gurney [69] (1) 13–100 BS4360:50 290–405 MAW AW A 0 0 – WT 0
25–100 B
Kihl Sarkani (1) 6–25 HSLA-80 598–671 GMAW AW A 0 0 – WT 0
[70,71]
Gurney [72] (1), (3) 6 Steel UTS 412 MIG AW A 0 0 – WT 0
(2) 515 MPa B
Balasubramanian (3) 8 ASTM 517F 690 SMAW AW A 0 0 – WR 0
Guha [73]
Infante et al. [74] (3) 10 Duplex 2205 527 TIG AW A 0 0 – WR 0
Singh et al. [75] (3) 6 AISI 304 L 297 GTAW AW A 0 0 – WR 0
Guha [76] (3) 12.5 C-Mn steel 280 MIG AW A 0 0 – WR 0
Yamaguchi et al. (3) 10 SM41 294 – AW A 0 0 – WR 0
[77] HT60 490
Ouchida and (3) 16–32 SM41 263 – AW A 0 0 – WR 0
Nishioka [78]
Macfarlane (3) 12.7 BS15 252 – AW A 0 0 – WR 0
Harrison [79]
Lassen et al. [80,81] (1) 25–32 C steel 250–416 SMAW SR A 0 0.3 – WT 0
SAW
FCAW
Meneghetti [25] (7) 10 S355 355 MAG AW A 0 0.1 – WT 0
(8) WT 0
WR 5876
Fricke Feltz [82] (7) 12 S355 355 MAG AW A 0 0 – WT 0
(8) WT 0
WR 18–79
Meneghetti et al. (12) 12.5 – – – AW B 0 0.1 – WT 0
[83]
Chiew et al. [84] (13) 10–16 S355J2 355 – AW A,B 0 0 – WT 0
Chiew et al. [85] (14) 25.4 API 5L 302 – AW A,B 0 0 – WT 0
Gandhi Berge [86] (15) 4–12.5 API 2H 355 MAW AW A 0 0 – WT 0
Schumacher (16) 8–20 S355J2 355 FCAW AW A,B 0 0 – WT 0
Nussbaumer
[87]
Fricke Doerk [88] (17) 12 – – – AW A 0 0 – WR 0.17–0.20
Fricke Doerk [88] (18) 12 – – – AW A 0 0 – WT 0
WR 2.70
Fricke et al. [89,90] (19) 10 S235JRG2 235 MAW AW A 0 0.5 – WR 0
MAG B
Kang Kim [91] (20)–(22) 10 Grade A 235 FCAW AW A 0 −1 – WT 0
Meneghetti et al. (23)–(27) 8 S355 355 MAG AW A 0 0.1 – WR, WT 0
[52]
Petershaghen [92] (6) 20 St 52–3 – SAW AW A 0 0,-1 – WT 0
Hentschel et al. (6) 10–20 St 38 – MAG AW A 0 0.2 – WT 0
[93] MAW
Yakubovskii (6) 16 St 3 – MAW AW A 0 0 – WT 0
Valteris [94] 15 HSND EBW
15G2AFD MAW
Takahashi et al. (28) 12 JIS SM400B 283 MAG AW A 0 0 - WT 0
[95,96] (29) bi-A 0, 180

Amstutz et al. [97] (30) 7.7 StE 460 520 – SR T ∞ 0,-1 – WT ∞


Seeger Olivier (32) 8 St 35.29 240 MAG SR T ∞ −1 – WT ∞
[98,100] StE 770 770
WR
Meneghetti et al. (30) 10 S355JR 355 MIG SR, AW T ∞ −1 – WR ∞
[53]
Razmjoo [101] (33) 7 BS 4360-50E 415 SMAW AW A 0 0 - WT 0
T ∞ - WR, WT ∞
3.2 A 0 - WT 0
A+T 0.5–2.94 0, 90 0.26–8.95
Siljander et al. (32) 9.5 A519 tube 414 tube MIG SR B 0 0,-1 - WT 0
[102] A36 flange 250 T ∞ - ∞
flange B+T 0.14–1 0 0.02–1
Yousefi et al. [103] (30) 8 StE 460 520 MIG SR B 0 0,-1 - WT 0
T ∞ - ∞
B+T 1 0, 90 1.14
Sonsino [6,104] (30) 10 StE 460 520 MAG SR B 0 −1 - WT 0
T ∞ - ∞
B+T 0.58 0, 90 0.38
(continued on next page)

15
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Table 5 (continued)

Ref. Joint Main Material Yield Welding Testing Load type° Λ = τnom/σnom Load Phase Failure λ Eq. (11)
geometry* thickness stress process condition# Eq. (8) ratio shift ϕ location^
[mm] [MPa] R [°]

Bertini et al. [105] (32) 10 S355 JR 360 – AW T ∞ 0,-1 - WR ∞


B+T 0.88–3.25 0 WT 0.06–0.88
Frendo Bertini (32) 10 S355 JR 360 – AW B+T 0.88–3.25 0,-1 90 WT 0.05–0.73
[106]
Meneghetti et al. (34) 8 S355 JR 355 MIG SR, AW B+T 0.38 0.1 0 WT 0.41
[54] (35) 11.25 S355 J2 MIG SR 1.15 −1 WR 11.4
Shams Vormwald (36) 2 E355 + N, 355 GMAW SR A 0 0,-1 - WR 35.25
[107] S340 + N 340 T ∞ - 2.34
A+T 0.99–1.55 0, 90 3.32–4.64

* See Table 3. All joints have un-machined welds.


#
AW = as-welded, SR = stress-relieved.
°
A = axial, B = bending, T = torsion.
^
WR = weld root, WT = weld toe.

probabilities of 10% to 90% and single test series reported by Haibach moment, respectively; A, Wf, and Wt are the section area and section
[112]. moduli, respectively.
To be consistent within the adopted local approach for the fatigue
4.2. Arc-welded joints made of structural steels design, the local biaxiality ratio λ was defined as the ratio between the
SED contributions due to mode II/III shear and mode I stresses [54,55];
Fig. 6a shows the master fatigue design scatter band, which was λ can be expressed as a function of the peak stresses:
fitted on approximately 180 fatigue results from literature and regard
c w2· fw2 2 · 2
+ c w3·f w23 · 2
only weld toe failures (2α ≅ 135°) [56]. These data were generated by =
r , = 0, peak z, = 0, peak
for Plane 4 or Brick 8,
testing as-welded joints under pulsating (R ≈ 0) pure axial or bending c w1·f w21 · 2
, = 0, peak
loadings. The considered details were T- or cruciform fillet-welded (11a)
joints ((1)–(3) in Table 3); the main plate thickness T was between 6
and 100 mm, and the base material was structural steel with a yield c w2· fw2 2 · ¯r2 , = 0, peak + c w3·f w23 · ¯ 2z, = 0, peak
= for
stress of 250–690 MPa (see [68–72] in Table 5). c w1·fw21 · ¯2 , = 0, peak
Fig. 6b shows the master fatigue design scatter band fitted on ap-
Tetra 10 or Tetra 4. (11b)
proximately 20 experimental data from literature, which regard only
weld toe failures (2α ≅ 135°) [26]. The results originate from welded The case λ = 0 corresponds to a pure local mode I stress state,
joints tested under stress-relieved conditions and pure torsion loading whereas λ → ∞ represents a local mode II + III shear stress state; λ
with the nominal load ratio R = −1. The joints were full-penetration or values between 0 and ∞ imply mixed mode opening-shear stress con-
fillet-welded tube-to-flange specimens ((30) and (32) in Table 3); the ditions.
tube thickness T was 8–10 mm, and the employed structural steels had The PSM-based master curves of Figs. 5 and 6 are compared in
yield stresses between 240 and 520 MPa (see [6,98,100,102–104] in Figs. 7 and 8 for approximately 330 and 1300 experimental fatigue
Table 5). results of aluminium welded joints (Fig. 7) and steel welded joints
It is worth mentioning that the 2.3–97.7% scatter index Tσ of the (Fig. 8), respectively. The original experimental results of arc-welded
scatter bands in Fig. 6a and b is equal to 1.90, which is only slightly joints made of aluminium alloys originate from
higher than that of the scatter bands of the aluminium welded joints [59,60,108–111,61–67,99], while those of steel joints originate from
(Fig. 5a and b). [6,25,73–82,52,83–92,53,93–98,100–103,54,104–107,68–72] (Tables
4 and 5). In addition, the tables include details about the geometries,
4.3. Criterion for selecting master fatigue design curve materials, welding processes, and testing and loading conditions. The
experimental results were generated by fatigue testing on as-welded
According to [54,55], the proper master curve for the fatigue design and stress-relieved joints under pure axial/pure bending, pure torsion,
of arc-welded joints should be selected based on the relative contribu- and multiaxial in-phase and out-of-phase loading conditions. Nominal
tions from the local shear and local normal stresses [54,55]. For nom- load ratios R of −1 to 0.5 and several nominal biaxiality ratios Λ ac-
inal stresses, the biaxiality ratio Λ between the shear and normal cording to Eq. (8) were considered. Moreover, the weld root and weld
stresses can be defined as follows: toe failures were included. The original experimental results were
nom converted in terms of equivalent peak stress according to the PSM in
= [23,27,50,51] for aluminium joints and according to that in
nom (8)
[23,25,26,28,39,51–54,56,57] for steel joints.
where the axial, bending, and torsion nominal stresses are calculated In this study, the experimental results were classified as a function
for the main loaded plate or tube: of the local biaxiality ratio λ defined in Eq. (11). In particular, the
F experimental results of each test series were compared with the PSM-
= for axial loading,
nom
A (9a) based master curves in Figs. 5 and 6 for the respective material. The
classification in Figs. 7 and 8 guarantees the best agreement between
Mf
nom = for bending loading, the experimental results and theoretical estimations and provides a
Wf (9b) criterion for the selection of an appropriate reference fatigue design
curve. The results are listed in Table 6, which does not cover welded
Mt
nom = for torsion loading, joints with main plate or tube thicknesses below 3 mm for aluminium
Wt (10)
joints and below 2 mm for steel joints because they have not been
where F, Mf, and Mt are the applied axial load, bending, and torsional analysed. By considering technical literature, the effect of the joint

16
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Fig. 7. Fatigue assessment of toe and root failures in welded joints made of aluminium alloys tested in as-welded or stress-relieved conditions (see details in Tables 3
and 4) according to the PSM: (a) λ = 0 and T ≥ 5 mm, (b) λ > 0 or 3 mm ≤ T < 5 mm. The design scatter bands were calibrated in Refs. [23,27] and are reported
in Fig. 5a and b, respectively.

thickness T on the fatigue curves was analysed in detail by Sonsino et al. ratios are equal. Hence, results of joint (3) must be included in Fig. 8a,
[113]; they reported that the inverse slopes k of fatigue curves of axial/ while those of joint (8) must be included in Fig. 8b. The agreement
bending and shear/torsion loadings are typically in the ranges 3–5 and between the experimental results and proposed fatigue design curves
5.4–6.5 for aluminium thin joints with T < 3 mm and in the ranges (Fig. 8a and b) support the criterion based on the local biaxiality ratio λ
3–6 and 5–8 for steel thin joints with T < 5 mm, respectively. (Eq. (11)) for the selection of the proper master design curve.
Figs. 7 and 8 and Tables 4 and 5 indicate that the governing para-
meter for selecting the proper PSM-based master curve is the local
5. Conclusions
biaxiality ratio λ (Eq. (11)) and not the nominal one (Λ; Eq. (8)). As an
example, steel welded details with geometries (3) [78] and (8) [25] in
In fatigue design of welded joints, the PSM is an engineering FE-
Table 3 were compared; such joints have main plate thicknesses of at
oriented technique for evaluating readily the NSIFs at the weld toe and
least 10 mm and fail at the weld root. According to Table 5, both joint
weld root, which are idealised as sharp V-notches with a null tip radius.
geometries were subjected to a nominal biaxiality ratio Λ = 0 because a
In more detail, the linear elastic, opening, sliding, and tearing peak
pure axial load was applied. However, a different biaxiality of the stress
stresses calculated at the V-notch tip with coarse FE analyses proved to
state occurred at the crack initiation point: the weld root of geometry
be linked to mode I, II, and III NSIFs, respectively, provided that certain
(3) was subjected to pure mode I loading (i.e. λ = 0), whereas the weld
conditions of the applicability of the PSM are satisfied, as reported in
root of geometry (8) was subjected to practically pure mode II loading
the present paper. As long as the fatigue strength criterion based on the
(i.e. λ → ∞). Therefore, the criterion based on λ suggests the use of
NSIF-dependent averaged SED is adopted, the PSM design stress (i.e.
different PSM-based design curves although the nominal biaxiality
the equivalent peak stress) can be expressed as a combination of the

17
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

Fig. 8. Fatigue assessment according to the PSM of weld toe and weld root failures in welded joints made of structural steels tested in as-welded or stress-relieved
conditions (see details in Tables 3 and 5) according to the PSM: (a) λ = 0 and T ≥ 2 mm, (b) λ > 0 and T ≥ 2 mm. The design scatter bands were calibrated in Refs.
[26,56] and are reported in Fig. 6a and b, respectively.

previously presented opening, sliding, and tearing peak stresses. distribution of the equivalent peak stress; a remarkable application
Developed in the past ten years, the resulting PSM exhibits the fol- is when there is competition between the weld toe and weld root
lowing main capabilities and features: fatigue failures;
• The fatigue life of 2D and 3D arc-welded joints made of aluminium
• 2D and 3D linear elastic FE analyses are supported by the method alloys or structural steels subjected to uniaxial or combined in-phase
for a range of FE types and commercial FE software packages, the and out-of-phase multiaxial loadings can be estimated; the PSM was
reference one being Ansys®; validated against a database of fatigue data consisting of approxi-
• The crack initiation point can be identified by analysing the mately 330 and 1300 experimental fatigue results for aluminium

Table 6
Criterion for selecting the reference PSM-based fatigue design curve for arc-welded joints.
Class of materials T [mm] λ Eq. (11) PSM- master curve Δσeq,peak,A,50% [MPa] k Tσ

Aluminium alloys T ≥ 5 mm λ =0 Fig. 5a 123 3.8 1.80


T ≥ 5 mm λ > 0 Fig. 5b 123 6.5 1.80
3 mm ≤ T < 5 mm λ ≥0
Structural steels T ≥ 2 mm λ =0 Fig. 6a 214 3 1.90
T ≥ 2 mm λ > 0 Fig. 6b 354 5 1.90

18
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

welded joints and steel welded joints, respectively, including as- fatigue behaviour of magnesium welds. Int J Fatigue 2017;101:115–26. https://
welded and stress-relieved joints with nominal load ratios of −1 to doi.org/10.1016/j.ijfatigue.2016.10.023.
[22] Pedersen MM. Multiaxial fatigue assessment of welded joints using the notch stress
0.5;

approach. Int J Fatigue 2016;83:269–79. https://doi.org/10.1016/j.ijfatigue.
For a given class of welded materials and joint geometry, the fatigue 2015.10.021.
resistance depends primarily on the relative contributions from the [23] Meneghetti G, Lazzarin P. Significance of the elastic peak stress evaluated by FE
analyses at the point of singularity of sharp V-notched components. Fatigue Fract
local shear and local normal stresses; therefore, a criterion for se- Eng Mater Struct 2007;30:95–106. https://doi.org/10.1111/j.1460-2695.2006.
lecting the proper fatigue design curve was provided based on the 01084.x.
biaxiality of the local stress state. In addition, the relevant endurable [24] Meneghetti G, Guzzella C. The peak stress method to estimate the mode I notch
stress intensity factor in welded joints using three-dimensional finite element
stresses and slopes of the master curves for adoption in the fatigue models. Eng Fract Mech 2014;115:154–71. https://doi.org/10.1016/j.
design of welded joints were presented (see Table 6). engfracmech.2013.11.002.
[25] Meneghetti G. The use of peak stresses for fatigue strength assessments of welded
lap joints and cover plates with toe and root failures. Eng Fract Mech
Declaration of Competing Interest
2012;89:40–51. https://doi.org/10.1016/j.engfracmech.2012.04.007.
[26] Meneghetti G. The peak stress method for fatigue strength assessment of tube-to-
The authors declare that they have no known competing financial flange welded joints under torsion loading. Weld World 2013;57:265–75. https://
doi.org/10.1007/s40194-013-0022-x.
interests or personal relationships that could have appeared to influ- [27] Meneghetti G, Campagnolo A, Rigon D. Multiaxial fatigue strength assessment of
ence the work reported in this paper. welded joints using the Peak Stress Method – Part I: Approach and application to
aluminium joints. Int J Fatigue 2017;101:328–42. https://doi.org/10.1016/j.
ijfatigue.2017.03.038.
References [28] Meneghetti G, Campagnolo A, Rigon D. Multiaxial fatigue strength assessment of
welded joints using the Peak Stress Method – Part II: Application to structural steel
[1] Eurocode 3: Design of steel structures – part 1–9: Fatigue. CEN; 2005. joints. Int J Fatigue 2017;101:343–62. https://doi.org/10.1016/j.ijfatigue.2017.
[2] Eurocode 9: Design of aluminium structures - Part 1-3: Structures susceptible to 03.039.
fatigue. CEN; 2011.uns. [29] Smith RA, Miller KJ. Prediction of fatigue regimes in notched components. Int J
[3] Hobbacher AF. Recommendations for Fatigue Design of Welded Joints and Mech Sci 1978;20:201–6. https://doi.org/10.1016/0020-7403(78)90082-6.
Components. IIW Collection. Springer International Publishing; 2016. doi:10. [30] Atzori B, Lazzarin P, Meneghetti G. Fracture mechanics and notch sensitivity.
1007/978-3-319-23757-2. Fatigue Fract Eng Mater Struct 2003;26:257–67. https://doi.org/10.1046/j.1460-
[4] Radaj D, Sonsino CM, Fricke W. Fatigue assessment of welded joints by local ap- 2695.2003.00633.x.
proaches. 2nd ed. Cambridge: Woodhead Publishing; 2006. [31] Williams ML. Stress singularities resulting from various boundary conditions in
[5] Fricke W. International Institute of Welding. IIW recommendations for the fatigue angular corners of plates in tension. J Appl Mech 1952;19:526–8.
assessment of welded structures by notch stress analysis: IIW-2006-09. Woodhead [32] Qian J, Hasebe N. Property of eigenvalues and eigenfunctions for an interface V-
Pub; 2012. notch in antiplane elasticity. Eng Fract Mech 1997;56:729–34. https://doi.org/10.
[6] Sonsino C. Multiaxial fatigue of welded joints under in-phase and out-of-phase 1016/S0013-7944(97)00004-0.
local strains and stresses. Int J Fatigue 1995;17:55–70. https://doi.org/10.1016/ [33] Gross B, Mendelson A. Plane elastostatic analysis of V-notched plates. Int J Fract
0142-1123(95)93051-3. Mech 1972;8:267–76. https://doi.org/10.1007/BF00186126.
[7] Susmel L. Multiaxial notch fatigue. Cambridge, UK: Woodhead Publishing; 2009. [34] Lazzarin P, Tovo R. A unified approach to the evaluation of linear elastic stress
[8] Radaj D, Vormwald M. Advanced methods of fatigue assessment. Berlin, fields in the neighborhood of cracks and notches. Int J Fract 1996;78:3–19.
Heidelberg: Springer Berlin Heidelberg; 2013. doi:10.1007/978-3-642-30740-9. https://doi.org/10.1007/BF00018497.
[9] Lazzarin P, Tovo R. A notch intensity factor approach to the stress analysis of [35] Lazzarin P, Zambardi R. A finite-volume-energy based approach to predict the
welds. Fatigue Fract Eng Mater Struct 1998;21:1089–103. https://doi.org/10. static and fatigue behavior of components with sharp V-shaped notches. Int J Fract
1046/j.1460-2695.1998.00097.x. 2001;112:275–98. https://doi.org/10.1023/A:1013595930617.
[10] Lazzarin P, Sonsino CM, Zambardi R. A notch stress intensity approach to assess [36] Lazzarin P, Lassen T, Livieri P. A notch stress intensity approach applied to fatigue
the multiaxial fatigue strength of welded tube-to-flange joints subjected to com- life predictions of welded joints with different local toe geometry. Fatigue Fract
bined loadings. Fatigue Fract Eng Mater Struct 2004;27:127–40. https://doi.org/ Eng Mater Struct 2003;26:49–58. https://doi.org/10.1046/j.1460-2695.2003.
10.1111/j.1460-2695.2004.00733.x. 00586.x.
[11] Livieri P, Lazzarin P. Fatigue strength of steel and aluminium welded joints based [37] Fischer C, Fricke W, Rizzo CM. Review of the fatigue strength of welded joints
on generalised stress intensity factors and local strain energy values. Int J Fract based on the notch stress intensity factor and SED approaches. Int J Fatigue
2005;133:247–76. https://doi.org/10.1007/s10704-005-4043-3. 2016;84:59–66. https://doi.org/10.1016/J.IJFATIGUE.2015.11.015.
[12] Lazzarin P, Livieri P, Berto F, Zappalorto M. Local strain energy density and fatigue [38] Fischer C, Fricke W, Rizzo CM. Fatigue tests of notched specimens made from butt
strength of welded joints under uniaxial and multiaxial loading. Eng Fract Mech joints at steel. Fatigue Fract Eng Mater Struct 2016;39:1526–41. https://doi.org/
2008;75:1875–89. https://doi.org/10.1016/j.engfracmech.2006.10.019. 10.1111/ffe.12473.
[13] Karakaş Ö, Morgenstern C, Sonsino C. Fatigue design of welded joints from the [39] Campagnolo A, Vormwald M, Shams E, Meneghetti G. Multiaxial fatigue assess-
wrought magnesium alloy AZ31 by the local stress concept with the fictitious ment of tube-tube steel joints with weld ends using the peak stress method. Int J
notch radii of rf=1.0 and 0.05mm. Int J Fatigue 2008;30:2210–9. https://doi.org/ Fatigue 2020:105495https://doi.org/10.1016/j.ijfatigue.2020.105495.
10.1016/j.ijfatigue.2008.05.017. [40] Hu Z, Berto F, Hong Y, Susmel L. Comparison of TCD and SED methods in fatigue
[14] Karakaş Ö. Consideration of mean-stress effects on fatigue life of welded magne- lifetime assessment. Int J Fatigue 2019;123:105–34. https://doi.org/10.1016/J.
sium joints by the application of the Smith–Watson–Topper and reference radius IJFATIGUE.2019.02.009.
concepts. Int J Fatigue 2013;49:1–17. https://doi.org/10.1016/j.ijfatigue.2012. [41] Sonsino CM, Łagoda T. Assessment of multiaxial fatigue behaviour of welded joints
11.007. under combined bending and torsion by application of a fictitious notch radius. Int
[15] Carpinteri A, Spagnoli A, Vantadori S. Multiaxial fatigue life estimation in welded J Fatigue 2004;26:265–79. https://doi.org/10.1016/S0142-1123(03)00143-9.
joints using the critical plane approach. Int J Fatigue 2009;31:188–96. https://doi. [42] Berto F, Lazzarin P. A review of the volume-based strain energy density approach
org/10.1016/j.ijfatigue.2008.03.024. applied to V-notches and welded structures. Theor Appl Fract Mech
[16] Susmel L. Three different ways of using the Modified Wöhler Curve Method to 2009;52:183–94. https://doi.org/10.1016/j.tafmec.2009.10.001.
perform the multiaxial fatigue assessment of steel and aluminium welded joints. [43] Berto F, Lazzarin P. Recent developments in brittle and quasi-brittle failure as-
Eng Fail Anal 2009;16:1074–89. https://doi.org/10.1016/j.engfailanal.2008.05. sessment of engineering materials by means of local approaches. Mater Sci Eng R
016. Reports 2014;75:1–48. https://doi.org/10.1016/j.mser.2013.11.001.
[17] Taylor D, Barrett N, Lucano G. Some new methods for predicting fatigue in welded [44] Meneghetti G, Campagnolo A, Avalle M, Castagnetti D, Colussi M, Corigliano P,
joints. Int J Fatigue 2002;24:509–18. https://doi.org/10.1016/S0142-1123(01) et al. Rapid evaluation of notch stress intensity factors using the peak stress
00174-8. method: Comparison of commercial finite element codes for a range of mesh
[18] Susmel L. Modified Wöhler curve method, theory of critical distances and patterns. Fatigue Fract Eng Mater Struct 2018;41. https://doi.org/10.1111/ffe.
Eurocode 3: a novel engineering procedure to predict the lifetime of steel welded 12751.
joints subjected to both uniaxial and multiaxial fatigue loading. Int J Fatigue [45] Radaj D. State-of-the-art review on extended stress intensity factor concepts.
2008;30:888–907. https://doi.org/10.1016/j.ijfatigue.2007.06.005. Fatigue Fract Eng Mater Struct 2014;37:1–28. https://doi.org/10.1111/ffe.12120.
[19] Susmel L. The Modified Wöhler Curve Method calibrated by using standard fatigue [46] Radaj D. State-of-the-art review on the local strain energy density concept and its
curves and applied in conjunction with the Theory of Critical Distances to estimate relation to the J -integral and peak stress method. Fatigue Fract Eng Mater Struct
fatigue lifetime of aluminium weldments. Int J Fatigue 2009;31:197–212. https:// 2015;38:2–28. https://doi.org/10.1111/ffe.12231.
doi.org/10.1016/j.ijfatigue.2008.04.004. [47] Campagnolo A, Roveda I, Meneghetti G. The Peak Stress Method combined with
[20] Karakaş Ö, Zhang G, Sonsino CM. Critical distance approach for the fatigue 3D finite element models to assess the fatigue strength of complex welded struc-
strength assessment of magnesium welded joints in contrast to Neuber’s effective tures. Procedia Struct Integr 2019;19:617–26. https://doi.org/10.1016/j.prostr.
stress method. Int J Fatigue 2018;112:21–35. https://doi.org/10.1016/j.ijfatigue. 2019.12.067.
2018.03.004. [48] Campagnolo A, Meneghetti G. Rapid estimation of notch stress intensity factors in
[21] Karakaş Ö. Application of Neuber’s effective stress method for the evaluation of the 3D large-scale welded structures using the peak stress method. MATEC Web Conf

19
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

2018;165. https://doi.org/10.1051/matecconf/201816517004. [76] Guha B. A new fracture mechanics method to predict the fatigue life of welded
[49] Visentin A. Automated fatigue strength assessment of V - notch structures ac- cruciform joints. Eng Fract Mech 1995;52:215–29. https://doi.org/10.1016/0013-
cording to the peak stress method. University of Padova; 2020. 7944(95)00004-F.
[50] Meneghetti G. The peak stress method applied to fatigue assessments of steel and [77] Yamaguchi I, Terada Y, Nitta A. On the fatigue strength of steels for ship struc-
aluminium fillet-welded joints subjected to mode I loading. Fatigue Fract Eng tures. IIW Annu. Assem. IIW Doc. XIII-425-66; 1966.
Mater Struct 2008;31:346–69. https://doi.org/10.1111/j.1460-2695.2008. [78] Ouchida H, Nishioka A. A study of fatigue strength of fillet welded joints. IIW Doc.
01230.x. XIII-338-64; 1964.
[51] Meneghetti G, Campagnolo A, Berto F. Fatigue strength assessment of partial and [79] Macfarlane DS, Harrison JD. Some fatigue tests of load carrying transverse fillet
full-penetration steel and aluminium butt-welded joints according to the peak welds. Br Weld J 1965;12:613–23.
stress method. Fatigue Fract Eng Mater Struct 2015;38:1419–31. https://doi.org/ [80] Lassen T. The effect of the welding process on the fatigue crack growth. Weld J
10.1111/ffe.12342. 1990;69:75S–81S.
[52] Meneghetti G, Marini D, Babini V. Fatigue assessment of weld toe and weld root [81] Engesvik K, Lassen T. The effect of weld geometry on fatigue life. In: Proc. Third
failures in steel welded joints according to the peak stress method. Weld World Int. OMAE Conf., Houston, Texas: 1988. p. 440–46.
2016;60:559–72. https://doi.org/10.1007/s40194-016-0308-x. [82] Fricke W, Feltz O. Fatigue tests and numerical analyses of partial-load and full-load
[53] Meneghetti G, De Marchi A, Campagnolo A. Assessment of root failures in tube-to- carrying fillet welds at cover plates and lap joints. Weld World 2010;54:R225–33.
flange steel welded joints under torsional loading according to the Peak Stress doi:10.1007/BF03263508.
Method. Theor Appl Fract Mech 2016;83:19–30. https://doi.org/10.1016/j. [83] Meneghetti G, Marigo M, Righetto F. The local approach combined with a FE
tafmec.2016.01.013. modelling technique for fatigue analysis of a steel tubular fillet-welded joint. In:
[54] Meneghetti G, Campagnolo A, Babini V, Riboli M, Spagnoli A. Multiaxial fatigue Proc. Int. Conf. CAE Comput. Technol. Ind. TCN CAE, Lecce (Italy): On CD Rom;
assessment of welded steel details according to the peak stress method: Industrial 2005.
case studies. Int J Fatigue 2019;125:362–80. https://doi.org/10.1016/j.ijfatigue. [84] Chiew S-P, Lee C-K, Lie S-T, Ji H-L. Fatigue behaviors of square-to-square hollow
2019.04.014. section T-joint with corner crack. I: Experimental studies. Eng Fract Mech
[55] Campagnolo A, Meneghetti G, Babini V, Riboli M, Spagnoli A. Multiaxial fatigue 2007;74:703–20. https://doi.org/10.1016/j.engfracmech.2006.06.022.
assessment of welded steel details according to the peak stress method based on [85] Chiew S-PP, Lie S-TT, Lee C-KK, Huang Z-WW. Fatigue performance of cracked
tetra elements. MATEC Web Conf 2019;300:19002. https://doi.org/10.1051/ tubular T joints under combined loads. I: experimental. J Struct Eng
matecconf/201930019002. 2004;130:562–71. https://doi.org/10.1061/(ASCE)0733-9445(2004)130: 4(562).
[56] Meneghetti G, Lazzarin P. The Peak Stress Method for Fatigue Strength Assessment [86] Gandhi P, Berge S. Fatigue behaviour of T-joints: square chords and circular
of welded joints with weld toe or weld root failures. Weld World 2011;55:22–9. braces. J Struct Eng 1998;124:399–404.
https://doi.org/10.1007/BF03321304. [87] Schumacher A, Nussbaumer A. Experimental study on the fatigue behaviour of
[57] Meneghetti G, Guzzella C, Atzori B. The peak stress method combined with 3D welded tubular K-joints for bridges. Eng Struct 2006;28:745–55. https://doi.org/
finite element models for fatigue assessment of toe and root cracking in steel 10.1016/j.engstruct.2005.10.003.
welded joints subjected to axial or bending loading. Fatigue Fract Eng Mater Struct [88] Fricke W, Doerk O. Simplified approach to fatigue strength assessment of fillet-
2014;37:722–39. https://doi.org/10.1111/ffe.12171. welded attachment ends. Int J Fatigue 2006;28:141–50. https://doi.org/10.1016/
[58] Campagnolo A, Meneghetti G, Babini V, Riboli M, Spagnoli A. Multiaxial fatigue j.ijfatigue.2005.04.008.
assessment of welded steel details according to the peak stress method based on [89] Fricke W, Kahl A, Paetzold H. Fatigue assessment of root cracking of fillet welds
tetra elements. MATEC Web Conf 2019;300. https://doi.org/10.1051/matecconf/ subject to throat bending using the structural stress approach. Weld World
201930019002. 2006;50:64–74. https://doi.org/10.1007/BF03266538.
[59] Pinho da Cruz JA, Costa JD, Borrego LF, Ferreira JA. Fatigue life prediction in [90] Fricke W, Kahl A. Fatigue assessment of weld root failure of hollow section joints
AlMgSi1 lap joint weldments. Int J Fatigue 2000;22:601–10. https://doi.org/10. by structural and notch stress approaches. In: Packer JA, Willibald S, editors.
1016/S0142-1123(00)00023-2. Tubul. Struct. XI, London: Taylor & Francis Group; 2006. p. 593–600.
[60] Meneghetti G, Tovo R, Patricolo M, Volpone M. Experimental analysis of the fa- [91] Kang S, Kim W. A proposed S-N curve for welded ship structures. Weld J
tigue strength of welded structural details in aluminium alloy. In: Proc. 26th AIAS 2003;82:161–9.
Natl. Conf., Catania (Italy); 1997. p. 393–400 [in Italian]. [92] Petershagen H. Fatigue tests with hyperbaric dry butt welded specimens. IIW
[61] Voutaz B, Smith IF, Hirt MA. Fatigue behaviour of aluminium beams with welded Annu. Assem. IIW Doc. XIII-1445-92; 1992.
attachments. In: Proc. 3rd Int. Conf. Steel Alum. Struct. ICSAS, Istanbul, Turkey; [93] Hentschel K, Berger P, Rossler K, Schmidt M. Weld geometry as a factor controlling
1995. p. 213–20. the fatigue strength of butt welded joints. Weld Int 1990;4:494.
[62] Macdonald KA, Haagensen PJ. Fatigue design of welded aluminum rectangular [94] Yakubovskii VV, Valteris IJI. Geometrical parameters of butt and fillet welds and
hollow section joints. Eng Fail Anal 1999;6:113–30. https://doi.org/10.1016/ their influence on the welded joints fatigue life; 1989.
S1350-6307(98)00025-9. [95] Takahashi I, Ushijima M, Takada A, Akiyama S, Maenaka H. Fatigue behaviour of a
[63] Haagensen PJ, Oma S. Fatigue life improvement of welded aluminium joints by box-welded joint under biaxial cyclic loads. Fatigue Fract Eng Mater Struct
peening techniques. In: Proc. 8th Int. Conf. Joints Alum. Ina., Munich, Germany; 1999;22:869–77. https://doi.org/10.1046/j.1460-2695.1999.00224.x.
2001. p. 1–14. [96] Takahashi I, Takada A, Ushijima M, Akiyama S. Fatigue behaviour of a box-welded
[64] Sonsino CM, Radaj D, Brandt U, Lehrke HP. Fatigue assessment of welded joints in joint under biaxial cyclic loading: effects of biaxial load range ratio and cyclic
AlMg 4.5Mn aluminum alloy (AA 5083) by local approaches. Int J Fatigue compressive loads in the lateral direction. Fatigue Fract Eng Mater Struct
1999;21:985–99. https://doi.org/10.1016/S0142-1123(99)00049-3. 2003;26:439–48. https://doi.org/10.1046/j.1460-2695.2003.00645.x.
[65] Brandt U, Lawrence FV, Sonsino CM. Fatigue crack initiation and growth in [97] Amstutz H, Storzel K, Seeger T. Fatigue crack growth of a welded tube-flange
AlMg4.5Mn butt weldments. Fatigue Fract Eng Mater Struct 2001;24:117–26. connection under bending and torsional loading. Fatigue Fract Eng Mater Struct
https://doi.org/10.1046/j.1460-2695.2001.00372.x. 2001;24:357–68. https://doi.org/10.1046/j.1460-2695.2001.00408.x.
[66] Kueppers M, Sonsino CM. Critical plane approach for the assessment of the fatigue [98] Seeger T, Olivier R. Tolerable and allowable shear stresses at fatigue loaded
behaviour of welded aluminium under multiaxial loading. Fatigue Fract Eng Mater welded joints. Stahlbau (in Ger) 1987;8:231–8.
Struct 2003;26:507–13. https://doi.org/10.1046/j.1460-2695.2003.00674.x. [99] Maddox S. Scale effect in fatigue of fillet welded aluminium alloys. In: Proc. Sixth
[67] Costa JDM, Abreu LMP, Pinho ACM, Ferreira JAM. Fatigue behaviour of tubular Int. Conf. Alum. Weldments, Cleveland, Ohio: 1995. p. 77–93.
AlMgSi welded specimens subjected to bending-torsion loading. Fatigue Fract Eng [100] Seeger T, Olivier R. Slope and knee-point of the S-N curve of shear loaded fillet
Mater Struct 2005;28:399–407. https://doi.org/10.1111/j.1460-2695.2005. welds. Stahlbau (in Ger) 1992;61:137–42.
00875.x. [101] Razmjoo G. Fatigue of load-carrying fillet welded joints under multiaxial loadings.
[68] Maddox SJ. The effect of plate thickness on the fatigue strength of fillet welded In: Abington, editor. TWI REF. 7309.02/96/909, Cambridge, UK; 1996.
joints. Abington, Cambridge: Abington Publishing; 1987. [102] Siljander A, Kurath P, Lawrence F. Non proportional fatigue of welded structures.
[69] Gurney TR. The fatigue strength of transverse fillet welded joints. Abington, In: Mitchell M, Landgraf R, editors. Adv. fatigue lifetime Predict. Tech. ASTMSTP
Cambridge: Abington Publishing; 1991. 1122, Philadelphia, USA: American Society for Testing and Materials; 1992. p.
[70] Kihl D, Sarkani S. Thickness effects on the fatigue strength of welded steel cruci- 319–38.
forms. Int J Fatigue 1997;19:311–6. https://doi.org/10.1016/S0142-1123(97) [103] Yousefi F, Witt M, Zenner H. Fatigue strength of welded joints under multiaxial
00041-8. loading: experiments and calculations. Fatigue Fract Eng Mater Struct
[71] Kihl DP, Sarkani S. Mean stress effects in fatigue of welded steel joints. 2001;24:339–55. https://doi.org/10.1046/j.1460-2695.2001.00397.x.
Probabilistic Eng Mech 1999;14:97–104. https://doi.org/10.1016/S0266- [104] Sonsino CM, Kueppers M, Gaeth N, Maddox SJ, Razmjoo GR. Fatigue behaviour of
8920(98)00019-8. welded high-strength components under combined multiaxial variable amplitude
[72] Gurney TR. Fatigue of thin walled joints under complex loading. Abington, loading. Darmstadt (Germany): 1999. doi:LBF-FB-218(1999).
Cambridge: Abington Publishing; 1997. [105] Bertini L, Cera A, Frendo F. Experimental investigation of the fatigue resistance of
[73] Balasubramanian V, Guha B. Fatigue life prediction of shielded metal arc welded pipe-to-plate welded connections under bending, torsion and mixed mode loading.
cruciform joints containing LOP defects by a mathematical model. Int J Press Int J Fatigue 2014;68:178–85. https://doi.org/10.1016/j.ijfatigue.2014.05.005.
Vessel Pip 1999;76:283–90. https://doi.org/10.1016/S0308-0161(98)00137-9. [106] Frendo F, Bertini L. Fatigue resistance of pipe-to-plate welded joint under in-phase
[74] Infante V, Branco CM, Martins R. A fracture mechanics analysis on the fatigue and out-of-phase combined bending and torsion. Int J Fatigue 2015;79:46–53.
behaviour of cruciform joints of duplex stainless steel. Fatigue Fract Eng Mater https://doi.org/10.1016/j.ijfatigue.2015.04.020.
Struct 2003;26:791–810. https://doi.org/10.1046/j.1460-2695.2003.00681.x. [107] Shams E, Vormwald M. Fatigue of weld ends under combined loading. Int J
[75] Singh PJJ, Achar DRRG, Guha B, Nordberg H. Fatigue life prediction of gas Fatigue 2017;100:627–38. https://doi.org/10.1016/J.IJFATIGUE.2016.12.020.
tungsten arc welded AISI 304L cruciform joints with different LOP sizes. Int J [108] Meneghetti G. Metodologie di analisi e progettazione dei giunti saldati. University
Fatigue 2003;25:1–7. https://doi.org/10.1016/S0142-1123(02)00067-1. of Padua; 1998.

20
G. Meneghetti and A. Campagnolo International Journal of Fatigue 139 (2020) 105705

[109] Ribeiro A, Goncalves J, Oliveira F, Castro P, Fernandes A. A comparative study on [112] Haibach E. Service fatigue-strength – methods and data for structural analysis.
the fatigue behaviour of aluminium alloy welded and bonded joints. In: Proc. Sixth Dusseldorf: VDI; 1989.
Int. Conf. Alum. Weldments, Cleveland, Ohio; 1995. p. 65–76. [113] Sonsino CM, Bruder T, Baumgartner J. S-N lines for welded thin joints — sug-
[110] Jacoby G. Uber das verhalten von schweissverbindungen aus aluminiumlegier- gested slopes and FAT values for applying the notch stress concept with various
ungen bei schwingbeanspruchung. Hannover: Technische Hochschule; 1961. reference radii. Weld World 2010;54:R375–92. https://doi.org/10.1007/
[111] Pirsic T. Experimentally based method for fatigue life prediction of aluminium BF03266752.
welded joints. In: Proc. 7th Int. Fatigue Conf.; 1999. p. 1309–14.

21

You might also like