You are on page 1of 32

Fatigue Assessment of

**

Weld Joints Using ANSYS,
Verity & FE-Safe
Zhichao Wang
Aditya Sakhalkar
5/8/2007
** Verity is a weld fatigue assessment software
developed by Battelle based on Structural
Stress & Fracture Mechanics theory proposed
by Dr. Dong et al, Battelle
Part I: Key Points on Equivalent SS Method
Part II: Example Application
ECT_wht_templ 2/28/2013 11:09 AM 2
Part I: Key Points on SS & Weld Assessments
Zhichao Wang
Sr. Lead Engineer
Emerson Climate Technology Inc.

 Background
 Why The Battelle’s Structure Stress Is Not Sensitive To FE Mesh
Size?
 How Could Multiple Joint S~N Curves Be Reduced to A Single S~N
Curve, the Master S~N Curve?
 Example Application (Part ii, Aditya)
ECT_wht_templ 2/28/2013 11:09 AM 3
Background
Fatigue Assessment – Unwelded Structures
ECT_wht_templ 2/28/2013 11:09 AM 4
Fatigue Assessment – Welded Structures
It’s been commonly recognized that the fatigue life of the polished
specimen is dominated by fatigue crack initiation, whereas that of
welded structures is dominated by small crack propagation from some
pre-existing discontinuity.
ECT_wht_templ 2/28/2013 11:09 AM 5
(i). Nominal Stress Method (BS,IIW)
The nominal stress range is used to develop the S~N curves using
samples with actual weld joint geometry. The life curves refer to
particular weld details, there is no need for the user to attempt to quantify
the local stress concentration effect of the weld detail itself.
(ii). Hotspot Stress Approach (structural stress, geometric stress,
BS, IIW, CEN,DNV)
This procedure uses hot-spot stress range as a parameter. The S–N
curves are obtained from tests of actual welded joints based on the hot-
spot stress range rather than the nominal stress range.
(iii). Local Notch Stress Method (ASME, BS, IIW).
The notch stress approach attempts to include all sources of stress
concentration in the stress used with the design S–N curve. Thus a single
S–N curve may be sufficient for a given type of material. The problem is
that the local geometry of the toe or root of a weld is highly variable. It
may be hard to achieve consistent results.
Fatigue Assessment – Welded Structures
ECT_wht_templ 2/28/2013 11:09 AM 6
Fatigue Assessment – Welded Structures
(iv). The Fracture Mechanics Approach (For crack propagation life)
The parameter widely used is SIF, K. The fatigue resistance is represented by
fatigue crack growth rate da/dN. Many crack propagation laws are available. The
simplest one is Paris Law in which the crack growth law approximates to a linear
relationship:
For a flaw size starts from a
0
to a critical fatigue crack size a
f
, the remaining
fatigue life N under stress range S is obtained by integrating Eq (1):
(v) Verity Equivalent Structural Stress Method (Battelle)
Equivalent Structural Stress + Single Master S~N Curve
n
da
C ( K) (1)
dN
= A
f
0
a
n
a
da
C N (2)
( K)
=
A
}
ECT_wht_templ 2/28/2013 11:09 AM 7
Structural Stress & Mesh Sensitivity
(b) Comparison of SCF predicted by various modeling
procedures and extrapolation based HSS at the weld toe [12].
(b) Comparison of SCF predicted by various modeling
procedures and extrapolation based HSS at the weld toe [12].
(c) Structural Stress Using Verity Method (c) Structural Stress Using Verity Method
Hot Spot Nomina
Nominal
l
FE
(Human Factor)
~
0 ·
o
o
< o <
o
x
o
X
~0.4 t
Hot Spot
o
FE
o
Nominal
o
(a) Normal Stress at the sharp corner
(Structural Stress)
t
σ
x
σ s
p
A
A
P
Y
X
0.4 t
A
A
t
Hot Spot Nomina
Nominal
l
FE
(Human Factor)
~
0 ·
o
o
< o <
o
x
o
X
~0.4 t
Hot Spot
o
FE
o
Nominal
o
(a) Normal Stress at the sharp corner
x
o
X
~0.4 t
Hot Spot
o
FE
o
Nominal
o
(a) Normal Stress at the sharp corner
(Structural Stress)
t
σ
x
σ s
p
A
A
(Structural Stress)
t
σ
x
σ s
p
(Structural Stress)
t
σ
x
σ s
p
A
A
P
Y
X
0.4 t
A
A
t
P
Y
X
0.4 t
A
A
A
A
t
ECT_wht_templ 2/28/2013 11:09 AM 8
Why Structure Stress
Is Not Sensitive To FE Mesh?
(a)
(b)
(c) Structural Stress Using Verity Method (c) Structural Stress Using Verity Method
N
P
1
P
2
Q
m
P
1
P
2
P
1
P
2
N
/
Q
/
m
/
P
1
P
2
t
X
Y
F
x
P
1
P
2
F
y
Neutral axis
t
σ
t
m
σ
t
b
σ
)'
t

)'
t
m
(σ )'
t
b

N
P
1
P
2
Q
m
N
P
1
P
2
Q
m
P
1
P
2
P
1
P
2
P
1
P
2
N
/
Q
/
m
/
P
1
P
2
N
/
Q
/
m
/
P
1
P
2
t
X
Y
P
1
P
2
P
1
P
2
t
X
Y
F
x
P
1
P
2
F
y
Neutral axis
F
x
P
1
P
2
F
y
F
x
P
1
P
2
F
y
Neutral axis
t
σ
t
m
σ
t
b
σ
t
σ
t
m
σ
t
b
σ
)'
t

)'
t
m
(σ )'
t
b

)'
t

)'
t
m
(σ )'
t
b

Calculation of Structure Stress
3-D Solid Element:
i. Calculate nodal force
ii. Transformation of nodal force to neutral
axis to obtain resultant forces (N, m).
iii.Calculation of structural stress
Shell Element:
i. Calculate nodal force ( N
/
, m
/,
Q
/
)
ii. Calculation of structural stress
Because equilibrium has to be satisfied
Thus: m
/
=m
Q
/
=Q And
N
/
=N
t
m
N
(σ )=
A
t
b
m
(σ)=
W
( ) ( )
,
/ /
t t t t
m m b b
σ = σ σ = σ
The equations in the calculation of
structural stress are the same. Hence, the
structural stress is mesh independent
Calculation of Structure Stress
3-D Solid Element:
i. Calculate nodal force
ii. Transformation of nodal force to neutral
axis to obtain resultant forces (N, m).
iii.Calculation of structural stress
Shell Element:
i. Calculate nodal force ( N
/
, m
/,
Q
/
)
ii. Calculation of structural stress
Because equilibrium has to be satisfied
Thus: m
/
=m
Q
/
=Q And
N
/
=N
t
m
N
(σ )=
A
t
b
m
(σ)=
W
( ) ( )
,
/ /
t t t t
m m b b
σ = σ σ = σ
The equations in the calculation of
structural stress are the same. Hence, the
structural stress is mesh independent
ECT_wht_templ 2/28/2013 11:09 AM 9
Master S~N Curve
Most codes divide weld joints into different types
Categorization of Weld Joints (BS 7608) Weld Joints (IIW)
Weld Joint Categorization
ECT_wht_templ 2/28/2013 11:09 AM 10
Multiple S~N Curves Due to Weld Joint Categorization
Example Design S~N curves for
welded joints: (a) Steel weld joint S~N
curves (BS 7608); (b) Weld joint type
and S~N curves (IIW
recommendations); (c) Aluminum weld
joint S~N curves (IIW
recommendations)
(a)
(b)
(c)
Multiple S~N curves provide flexibility
for the selection of life curves and
increase difficulty to select the proper
one due to the variation of actual
joints
ECT_wht_templ 2/28/2013 11:09 AM 11
How Could multiple joint S~N curves be
merged into one - Master S~N curve?
Nominal Stress Method (BS) Hot Spot Stress Method (IIW)
Equivalent SS Method
Weld Joints (IIW)
ECT_wht_templ 2/28/2013 11:09 AM 12
Small Crack Propagation Behavior
The data does not fall into one curve for the same specimen;
Multiple da/dN curves obtained for the same or different
specimens, which tells that the Paris Law does not hold for
small crack growth. This is called anomalous crack growth
behavior.
(g) AK-correlated small
crack vs. large crack
growth, 7075-T6 AI [13]
The curves show that
small crack growth faster
than corresponding large
cracks at the same value
of the driving force AK.
Specimens for notch induced short crack growth test. (a)
Compact tension (CT) with keyhole by Ramulu; (b) single edged
notched (SEN) by Shin and Smith; (c) double edged notched
(DEN) by Shin and Smith; (d) center notched (CB) by Tanaka and
Nakai; (e) for CT specimen; (f) for three different
specimens
da
K
dN
A
da
K
dN
A
Specimens for notch induced short crack growth test. (a)
Compact tension (CT) with keyhole by Ramulu; (b) single edged
notched (SEN) by Shin and Smith; (c) double edged notched
(DEN) by Shin and Smith; (d) center notched (CB) by Tanaka and
Nakai; (e) for CT specimen; (f) for three different
specimens
da
K
dN
A
da
K
dN
A
Different factors may attribute to short crack anomalous growth such as the effect of crack
closure, micro structure interaction and that of notch details.
A unified SIF formulation for both short and long crack and a two stage crack propagation model
were proposed by Dr. Dong & his coworkers, Battelle, to use Equivalent Structural Stress as a
parameter for weldments fatigue life assessment.
ECT_wht_templ 2/28/2013 11:09 AM 13
A Unified Stress Intensity Factor (SIF) Formulation
(a) Weld geometry with a hypothetical crack l ; (b) Actual normal stress distribution; (c)
Simplification; (d) Decomposition; (e) Equilibrium-equivalent structural stress or far-field
stress; (f) Self-equilibrating stress (notch stress) with respect to a reference depth t
1
.
( b )
t
A
A
x
σ
l 0
( b )
t
A
A
x
σ
l 0
( a )
A
A
t
l
( a )
A
A
t
l
( c )
t
1
t
1
2
3
1
σ
2
σ
3
σ
A
A
R
1
R
2
'
x
σ
( c )
t
1
t
1
2
3
1
σ
2
σ
3
σ
1
2
3
1
σ
2
σ
3
σ
A
A
R
1
R
2
'
x
σ
t
1
t
1
2
3
A
A

t
σ
A
A
t
σ
t
1
t
1
2
3
A
A

( d )
( e )
( f )
t
1
t
1
2
3
A
A

t
σ
t
1
t
1
2
3
A
A

t
σ
A
A
t
σ
A
A
t
σ
A
A
t
σ
t
1
t
1
2
3
A
A

t
1
t
1
2
3
A
A

( d )
( e )
( f )
/ t
x
= - o A o o
/ t
x
= o o + Ao
/ t
x
= - o A o o
/ t
x
= o o + Ao
t
1
/t=0.1
ECT_wht_templ 2/28/2013 11:09 AM 14
A Unified Stress Intensity Factor
t
( ) ( ) (0 l t) A o + < < o
s n
K K = K
Notch Stress
Structural Stress, Far field stress
Notch Stress
Structural Stress, Far field stress
The drive force for crack to start and grow is the crack tip stress, introduce crack
surface traction p
s
called self equilibrating surface traction due to Ao (notch effect)
x
t /
s
( ) (p , (0 l t ) ) ( ) Ao ~ = o o < <
s s s
K K K
t
l
t
l
( a ) ( b )
A
A
t
σ
A
A
t
σ
A
A
t
σ
t
1
t
1
2
3
A
A

t
1
t
1
2
3
A
A

( c )
t
s
p
l
( d )
t
s
p
l
tt
s
p
l
( d )
'
x
σ
t
1
t
3
A
A
l
/
s x
t
p = f( Δσ)= σ f( , σ )
( e )
'
x
σ
t
1
t
3
A
A
l
/
s x
t
p = f( Δσ)= σ f( , σ )
( e )
ECT_wht_templ 2/28/2013 11:09 AM 15
A Unified Stress Intensity Factor
t
s
(p ( ) (0 l t ) ) o + < <
n s
K = K K
Notch Effect
Structural Stress, Far field stress
t
s
(p ( ) (0 l t ) ) o + < <
n s
K = K K
Notch Effect
Structural Stress, Far field stress
Notch Effect
Structural Stress, Far field stress
t
l
( d )
( e )
A
A
t
σ
( f )
t
s
p
l
t
l
( d )
t
l
( d )
( e )
A
A
t
σ
( e )
A
A
t
σ
A
A
t
σ
A
A
t
σ
( f )
t
s
p
l
( f )
t
s
p
l
tt
s
p
l
Notch Effect
Notch Effect
Far Field SIF
Far Field SIF
Notch Effect
Notch Effect
Far Field SIF
Far Field SIF
Stress intensity solutions
using published weight
function results: (a) remote
tension; (b) remote
bending.
ECT_wht_templ 2/28/2013 11:09 AM 16
SIF Magnification Factor M
kn
t
s
(p ( ) (0 l t ) ) o + < <
n s
K = K K
Introduction of notch induced SIF magnification factor M
kn
:
t
t
s
(p ) ( )
1.0 (0 l t)
( )
o +
> < <
o
kn
s n
n
K
M =
K
K
t
( ) (0 l t) o < <
kn n
K = M K
a/t a/t a/t a/t
Comparisons of stress intensity magnification factor M
kn
at 135° sharp V notch for various
specimen geometries and loading conditions: (a) Edge crack solutions; (b) Elliptical crack
solutions for a/c = 0.4
ECT_wht_templ 2/28/2013 11:09 AM 17
M
kn
for 135
0
V Notch Specimen
a/t
a/t
Conclusions:
• M
kn
approaches unity as crack size a/t approaches 0.1 i.e., a/t=0.1 (short
crack correction factor) Thus a/t=0.1 can be taken as a characteristic
parameter beyond which the notch effect is negligible
• The difference between edge crack and elliptical crack solutions are not
significant.
ECT_wht_templ 2/28/2013 11:09 AM 18
Modified Paris Law
m
n
n m
kn n
Paris Law :
da
C K (a)
dN
Introduce t
da
CM K
dN
heUnified SIF:
(b) = A
= A
Figures showed the
significant improvement
for the application of
Paris Law and its
application
More important the
crack starts from zero
that covers crack
initiation life
ECT_wht_templ 2/28/2013 11:09 AM 19
Master S~N Curve
÷
=
=
= ·
=
÷
A
÷
A
÷
=
=
}
-m
s
Δσ
f
m
1
n m
2
n
kn
kn n
a a
n m
a 0
m b
n
m
kn
Using the Unified Paris Law,
da
N (b)
CM K
Introduction of
S ~ N curve can be obtained
1
N t I(r) (c)
C
d(a/ t)
Where,
generallized SIF ra
da
C
I(r)
M {f (a
M K (a)
dN
nge
/ t) r [f (a/ t) f
, leads
(a
to,
/ t)]
=
=
}
a / t 1
m
a / t 0
(d)
}
1
2 m 1 1
m
2m m m
2 m 1
2
1
1
m
m m
m
Then
C t I(r) N (e)
Introduce Equivalent Structrual Stress Range,
t I(r)
C N (f )
÷
÷
=
=
= ÷
s
s
s s
ΔS
Δσ
Δσ
ΔS
ECT_wht_templ 2/28/2013 11:09 AM 20
Master S~N Curve
Nominal Stress Method Hot Spot Stress Method
Equivalent SS Method
Weld Joints (IIW)
ECT_wht_templ 2/28/2013 11:09 AM 21
Summary
• The Structural Stress Method developed at Battelle is
mesh insensitive that removes the uncertainty in the
calculation of structural stress for weld joint fatigue
assessment
• The nature of weld joint fatigue is considered through the
introduction of Equivalent Structural Stress based on
fracture mechanics, which enable most fatigue curves of
weld joints merged into a narrow band, the Master S~N
curve. Thus one S~N curve can be used for majority of
weld joints
Note: I believe that the weld joints have to be stress
concentration dominant to achieve consistent
results with test data.
ECT_wht_templ 2/28/2013 11:09 AM 22
References
[1]. BS PD 5500: , BSI Standards, London, 2000.
[2]. European Standard for Unfired Pressure Vessels, EN 13445: 2002, BS EN 13445:2002, BSI, London, 2002
[3]. ASME Boiler and Pressure Vessel Code, Section VIII, Rules for construction of pressure vessels, Division 2-
Alternative rules, ASME, 2003.
[4]. Carl E. Jaske, FSRF for WPVP, Journal of Pressure Vessel Technology, AUGUST 2000, Vol. 122, 297-304
[5]. S.J.Maddox, Review of fatigue assessment procedures…, Int. J. of Fatigue, Vol. 25, 12, 2003, 1359-1378.
[6]. Maddox S J: 'Fatigue aspects of pressure vessel design…, Spence J and Tooth A S, E & F N Spon, London, 1994.
[7]. Harrison J D and Maddox S J: 'A critical examination of rules for the design of pressure vessels subject to fatigue
loading' in Proc. 4th Int. Conf. on 'Pressure Vessel Technology', Mech E, London, 1980.
[8]. Taylor N (Ed): 'Current practices for design against fatigue in pressure equipment', EPERC Bulletin No.6, European
Commission, NL-1755ZG, Petten, The Netherlands, 2001.
[9]. Dong, P., 2005, ‘‘A Robust Structural Stress Method for Fatigue Analysis of Offshore/Marine Structures’’, Journal of
Offshore Mechanics and Arctic Engineering , Vol. 127, pp. 68-74.
[10]. Dong, P., 2001, ‘‘A Structural Stress Definition and Numerical Implementation for Fatigue Evaluation of Welded
Joints,’’ Int. J. Fatigue, 23/10, pp. 865–876.
[11]. Dong, P., Hong, J. K, Osage, D., and Prager, M., ‘‘Assessment of ASME’s FSRF Rules for Pipe and Vessel Welds
Using A New Structural Stress Method,’’ Welding In the World, Vol. 47, No. 1/2, 2003, pp. 31–43.
[12] Dong, P., Hong, J. K., Osage, D., Prager, M., 2002, ‘‘Master S-N Curve Method for Fatigue Evaluation of Welded
Components,’’ WRC Bulletin, No. 474, August.
[13]. Lankford, J. , Fatigue of Eng Mater and Structures 5 (1982), pp233-248
[14]. R. Craig McClung, et al, Behavior of Small Fatigue Cracks, ASME, Vol. 19, Fatigue & Fracture, P153
[15]. Fricke W., 2001, ‘‘Recommended Hot-Spot Analysis Procedure for Structural Details of FPSO’s and Ships Based on
Round-Robin FE Analysis,’’ ISOPE Proceedings, Stavanger, Norway, June.
ECT_wht_templ 2/28/2013 11:09 AM 23
Part II: Example Application
Aditya Sakhalkar
Sr. Applied Mechanics Engineer
Emerson Climate Technology Inc.

• Background
• ANSYS, Verity, Fe-Safe Weld Assessment Procedure
• ANSYS Preprocessing
• Verity Analysis
• Fe-Safe Analysis
• ANSYS Results
• Comparison of Analysis Results with the Test Results
• Conclusions
ECT_wht_templ 2/28/2013 11:09 AM 24
• Resistance welded suction fitting
• Three failures in the suction fitting weld
during reliability testing
• Crack initiated at the weld toe (9 o’ clock)
and propagated through the shell.
• Failure due to reverse bending fatigue.
Background
ECT_wht_templ 2/28/2013 11:09 AM 25
ANSYS, Verity, Fe-Safe Weld Analysis Procedure
Verity, Fe-Safe Weld analysis
Ansys Preprocessing

Ansys Post processing
- Meshing per Verity requirements
- Linear material properties
- Apply load
- Solution
Verity Analysis (Eq. Structural Stress
calculations along the weld line)
- Fe-Safe analysis (Weld life calculations
using Master S-N curve)

- Import .rst file
- Post processing
Verity result validation
and comparison
Reliability Testing
Calibration
Quantify loading
relative to the test
for FEA
ECT_wht_templ 2/28/2013 11:09 AM 26
Suction Fitting

Shell

Weld Line

Element Type: Solid 186
Ansys Preprocessing- FE Model
Mesh Requirements
• Shell Elements:
• 3D mid-surface element with no through-
thickness dimension
• Solid Element:
• Rectangular faces required along the through-
thickness cut from the weld line
• Regular mesh along the through thickness cut
from weld line
• Recommended elements: Hexahedral (brick) and
Pentahedral (wedge)
• Tetrahedral elements can be used. Requires
special handling.

Hex elements along weld line
Linear material properties
(E = 29,000 Ksi, ν = 0.29)
Symmetry BC
On sides
Top & Bottom
fixed
ECT_wht_templ 2/28/2013 11:09 AM 27
ANSYS Results
o
eq

o
1

Maximum Stress
Equivalent Stress
ECT_wht_templ 2/28/2013 11:09 AM 28
Verity Procedures
• Structural Stresses calculated along the
weld line at each node
• Fe-Safe builds a connectivity table and
maps the stresses to elemental stresses.
These stresses are inserted into the
stress matrix to be used for fatigue
evaluation

ECT_wht_templ 2/28/2013 11:09 AM 29
Fe-Safe Analysis – Weld life predictions
using Master SN Curve
Fully reverse loading defined
Fe-Safe calculates weld fatigue life
based on the Battelle’s Master SN
curve
The weld life (log N) data at each
node is written in the ANSYS .rst
format

ECT_wht_templ 2/28/2013 11:09 AM 30
ANSYS Post processing – Weld Life Results
Minimum life at element # 4436, node # 4212
ECT_wht_templ 2/28/2013 11:09 AM 31
Comparison of Verity Predictions
with Reliability Test Results
Conservative life estimate!
Predicted Failure Location correlated well with the test failures!!
ECT_wht_templ 2/28/2013 11:09 AM 32
• ANSYS could be used for the fatigue assessment of welded
joints in conjunction with Verity & Fe-Safe
• Verity provides conservative estimate of the weld life
• Good correlation with the predicted and test failure locations
• Verity & Fe-Safe provides a excellent weld design tool in
conjunction with ANSYS



Conclusions