Weld Fatigue Analysis

Ron Gagnon, PE
Senior FEA Consulting Engineer
SimuTech Group
1800 Brighton Henrietta Town Line Road
Rochester, New York 14623
www.simutechgroup.com
[email protected]

April 3, 2019 1
 Introduction

Fatigue assessments are critical to cyclically loaded

structures
Fatigue life prediction can be particularly challenging in
welded structures
Welded joints contain residual stresses, material
property changes, and other factors which add
significant complexity to the fatigue problem
This presentation describes how to approach weld
fatigue analysis using both traditional code-based
methods and finite element analysis methods

April 3, 2019 Weld Fatigue Analysis 2

 Interesting Examples…..

April 3, 2019 Weld Fatigue Analysis 3

 Fatigue Failures in Welds
Examples

Carbon steel structure

(ASTM A36)
Subjected to static
and dynamic load
cycling

April 3, 2019 Weld Fatigue Analysis 4

 Fatigue Failures in Welds
Examples

Titanium tank
structure (SB265
Grade 2)
Subjected to over-
the-road random
vibration

April 3, 2019 Weld Fatigue Analysis 5

 Fatigue Failures in Welds
Examples
Carbon steel spring plate Subjected to rotating
structure (ASTM A588) loads

April 3, 2019 Weld Fatigue Analysis 6

 Fatigue Overview

April 3, 2019 Weld Fatigue Analysis 7

 Fatigue Overview
Beginnings of Fatigue:
 In the mid-1800’s, August Woehler studied the relationship
between stress cycles and fatigue life using a rotating bending
test that could subject railway axles to alternating stresses
 Plotted the nominal stress to failure to develop original stress-life
(S-N) fatigue curves

April 3, 2019 Weld Fatigue Analysis 8

 Fatigue Overview

Fatigue damage is the initiation and/or growth of a crack

under fluctuating loading
 Progressive damage accumulation
 Almost all structural components are subjected to some form of cyclic
service

σy

Fatigue failure occurs after

Stress

repeated loadings, even though the

stress is low
Time

April 3, 2019 Weld Fatigue Analysis 9

 Fatigue Overview
Fatigue failure is typically viewed as a 3-stage phenomena:
 stage 1: crack initiation
 stage 2: stable crack growth
 stage 3: fast fracture
Fatigue life = crack initiation + crack growth
 no precise transition from crack initiation to crack growth
 up to 90% of fatigue life involves crack initiation!

April 3, 2019 Weld Fatigue Analysis 10

 Fatigue Overview
Crack Initiation
Surface flaws initiate at tiny dislocations in the material’s
microstructure
 very localized stress concentration
 practically undetectable and impossible to model using FEA
 these tiny surface flaws create persistent slip bands that propagate
along the maximum shear plane under alternating stress
 bands slip back and forth like a shifting deck of cards
Alternating Stress

Crystal surface

Slip bands form

along planes of
maximum shear
giving rise to
surface extrusions
and intrusions

April 3, 2019 Weld Fatigue Analysis 11

 Fatigue Overview
Stage I Cracks
Slip bands grow into micro-cracks (stage I cracks)
 small compared to material’s microstructure
 difficult to detect, invisible to the naked eye
 grow in direction of maximum shear stress (45°to applied load)

April 3, 2019 Weld Fatigue Analysis 12

 Fatigue Overview
Stage II Cracks
After traversing 2-3 grain boundaries, the micro-cracks grow into
fatigue cracks (stage II cracks)
 large compared to material’s microstructure
 crack itself physically interrupts the flow of stress
 stress concentration causes plastic stress at the crack tip
 local plastic stress causes the crack to change direction
 growth is now propagated by cyclic plastic stress at the crack tip
 oriented perpendicular to maximum principal stress

April 3, 2019 Weld Fatigue Analysis 13

 Fatigue Overview
Fatigue Failure
If sufficient surface energy exists, Stage II cracks continue to grow
until tensile failure occurs
 most lack sufficient energy to propagate across grain barriers
 the crack “arrests”

slip bands micro-cracks fatigue cracks tensile failure

initial crack

fatigue damage
producing striations
(aka “beach marks”)

fast final fracture due

to tensile failure – no
striations

April 3, 2019 Weld Fatigue Analysis 14

 Fatigue Overview
Cyclic Loading Terms

Fatigue damage is caused by a changing stress/strain state

 sine curve is a simple representation of cyclic loading
 stress ratio (R) = Smin/Smax
 amplitude ratio (A) = Sa/Sm
Smax = maximum stress
Smin = minimum stress
Smax
Sm = mean stress
Sa = alternating stress
Sa
ΔS = stress range

Sm ∆S

Sa

Smin Common loadings:

fully reversed: R = -1, A = infinity
zero to maximum: R = 0, A = 1

April 3, 2019 Weld Fatigue Analysis 15

 Fatigue Overview

Stress-Life Fatigue

Stress-life (S-N) fatigue method assumes that fatigue damage is

produced by fluctuating stresses
 only applicable to elastic stress, so limited to low stress/high cycle
applications
usually more than 100,000 cycles to failure for ductile metals
 fatigue life is based on alternating stress range and Woehler fatigue
curves (S-N curves)
 original numerical fatigue approach
 vast amount of industry experience
 simple technique
often can be checked with hand calculations

April 3, 2019 Weld Fatigue Analysis 16

 Fatigue Overview

Strain-Life Fatigue

Strain-life (E-N) fatigue assesses fatigue damage using cyclic strain

ranges and strain-life equations
 applicable to both low cycle and high cycle applications
stresses less than or greater than yield
 uses local elastic-plastic strains
either directly calculated or adjusted from elastic results
 predicts crack initiation
 seven discrete Individual components
 relatively new fatigue analysis technique
usage began approximately 30 years ago
difficult to implement with hand calculations
limited to FEA

April 3, 2019 Weld Fatigue Analysis 17

 Fatigue Overview

Damage Accumulation
Multiple cyclic loading conditions have a cumulative fatigue effect
Rainflow counting groups the loading history into blocks
 each block causes a fraction of the total damage (“damage fraction”)

ni
damage fraction (Di ) = Di = damage fraction caused by loading block (i)
Ni ni = number of applied cycles of loading block (i)
Ni = available fatigue life for loading block (i)

April 3, 2019 Weld Fatigue Analysis 18

 Fatigue Overview

Miner’s Rule
Miner’s Rule assumes that the total damage is simply the linear
summation of the damage fractions
n
n1 n2 n3 nn ni
total damage (D ) = + + + ⋅⋅⋅ = 
N1 N 2 N 3 Nn i =1 N i

 first proposed by Palmgren in 1924 and further refined by Miner in 1945

 applicable to both S-N and E-N
 widely used
simple to implement and as accurate as more complicated methods
 load sequence is not considered
 failure occurs when sum of damage fractions equals the fatigue life
D ≥ 1.0

April 3, 2019 Weld Fatigue Analysis 19

 Fatigue Overview
Accuracy
Fatigue calculations are much less precise than strength
calculations
 statistical, not deterministic, phenomenon
 empirical rules
 order of magnitude errors in life estimates are common
 large Factor of Safety (FS) typically assumed to ensure safe design
FS of 10 is common
 FEA fatigue can potentially enable smaller FS values

April 3, 2019 Weld Fatigue Analysis 20

 Learning Questions & Answers (1)

List the three stages of fatigue failure

 Stage I: Crack Initiation

 Stage II: Stable Crack Growth
 Stage III: Fast Fracture

What mathematical rule is used to accumulate fatigue

damage?

 Miner’s Rule (or Palmgren-Miner)

April 3, 2019 Weld Fatigue Analysis 21

 Fatigue Failure in Welds

April 3, 2019 Weld Fatigue Analysis 22

 Fatigue Failure in Welds

Observations

Structure life generally limited by joint fatigue (welded or bolted)

Prudent to minimize stress concentration sites on structure
Handling features (lifting lugs, etc.) may become crack initiation sites
Manufacturing tolerances may induce initial cracking
Avoid joint eccentricity, welds at free edges
Involve a welding engineer for proper joint design
To improve fatigue life, increase base metal thickness and/or weld
size (however, fatigue penalty for thicker, stiffer joint!)
Surface finish is a major factor influencing the material endurance
limit. Others are loading type, size, temperature, and notch factor

April 3, 2019 Weld Fatigue Analysis 23

 Weld Geometry & Terminology
Typical Double-Sided T-Joint Fillet Weld

April 3, 2019 Weld Fatigue Analysis 24

 Weld Joint Characteristics
Residual Stress

Large Tensile Residual Stresses Exist

 induced by non-uniform heating/cooling during the welding process
 conservatively assumed equal to the material tensile yield strength
 may be eliminated by post-weld heat treatment
Impractical in many situations
 tensile residual (mean) stress decreases fatigue life
 compressive residual (mean) stress increases fatigue life
usually ignored
 more restraint during welding = more residual stress

April 3, 2019 Weld Fatigue Analysis 25

 Weld Joint Characteristics
Residual Stress

Image courtesy of Jitendra Malav

April 3, 2019 Weld Fatigue Analysis 26

 Weld Joint Characteristics
Localized Discontinuities
Localized Discontinuities Exist
 cracks already initiated
cyclic loading may result in crack growth
a minimum acceptable flaw size is frequently specified; verified by NDT techniques
(dye-penetrant, radiographic, etc.)
 weld porosity (voids)
 slag or other inclusions
 regions of incomplete fusion
 impractical to model weld geometry with discontinuities in detail

Image courtesy of
Eddyfi.com

April 3, 2019 Weld Fatigue Analysis 27

 Weld Joint Characteristics
Stress Concentrations

Disruption in Stiffness due to Weld Section Geometry

 creates stress concentrations due to sudden change in stiffness
 at weld toes, irregular as-welded features

April 3, 2019 Weld Fatigue Analysis 28

 Weld Joint Characteristics
Material Properties

Low Fatigue Strength Compared to Un-Welded Base Metal

 all as-welded carbon steels exhibit similar fatigue life for a given class
therefore, the same S-N curve can be used!
 using a higher-strength carbon steel has no added fatigue life benefit
 material properties vary across the weld section

April 3, 2019 Weld Fatigue Analysis 29

 Weld Joint Characteristics
Material Properties

April 3, 2019 Weld Fatigue Analysis 30

 Learning Questions & Answers (2)

List general factors influencing material endurance limit

 surface finish
 loading type (axial, bending, shear)
 size
 temperature
 stress concentrations or notch factors

April 3, 2019 Weld Fatigue Analysis 31

 Learning Questions & Answers (3)

List several welded joint characteristics

 high residual tensile and compressive stresses

 localized discontinuities
 stress concentrations
 material properties vary across the weld section

April 3, 2019 Weld Fatigue Analysis 32

 Weld Fatigue
Calculation Methods

April 3, 2019 Weld Fatigue Analysis 33

 Weld Fatigue Calculation Methods

There are four primary weld fatigue calculation methods:

 nominal stress method

 structural hot spot stress method
 effective notch stress method
 crack tip stress intensity

Each will be discussed on the following slides

April 3, 2019 Weld Fatigue Analysis 34

 Weld Fatigue Calculation Methods
Nominal Stress Method

The Nominal Stress method is prescribed in the following standards,

to name a few:

 AWS D1.1 (carbon steels)

 AWS D1.6 (stainless steels)
 AWS D1.9 (titanium)
 IIW
 BS7608 (British Standards Institute)

April 3, 2019 Weld Fatigue Analysis 35

 Weld Fatigue Calculation Methods
Nominal Stress Method

April 3, 2019 Weld Fatigue Analysis 36

 Weld Fatigue Calculation Methods
Nominal Stress Method

Example welded joint stress classification table (BS7608) for standard

basic design S-N curves

April 3, 2019 Weld Fatigue Analysis 37

 Weld Fatigue Calculation Methods
Nominal Stress Method

Probability of Failure = 2.3 % (Survival = 97.7%)

April 3, 2019 Weld Fatigue Analysis 38
 Weld Fatigue Calculation Methods
Structural Hot Spot Stress Method

Structural Hot Spot Stress method:

 normally used in situations where it is difficult to determine the
nominal stress due to unusual loading or geometric factors
 applies to problems with a crack initiating at the weld toe
 highly sensitive to FEA model mesh size

April 3, 2019 Weld Fatigue Analysis 39

 Weld Fatigue Calculation Methods
Structural Hot Spot Stress Method

σHS = 1.67σ0.4t – 0.67σ1.0t

April 3, 2019 Weld Fatigue Analysis 40

 Weld Fatigue Calculation Methods
Structural Hot Spot Stress Method

April 3, 2019 Weld Fatigue Analysis 41

 Weld Fatigue Calculation Methods
Effective Notch Stress Method
Effective Notch Stress method:
 linear elastic method, weld toe radius is modeled explicitly
 1 mm radius toe notch is included in FEA model, a smaller radius
is used for thin structures
 single S-N curve used and is dependent on probability of failure

April 3, 2019 Weld Fatigue Analysis 42

 Weld Fatigue Calculation Methods
Crack Tip Stress Intensity
Crack Tip Stress Intensity:
 fracture mechanics approach
 calculate crack tip stress intensity (K) and compare to material
plane-strain fracture toughness (KIC)
 use the Paris law to determine the crack growth rate given
material constants (C, m) and stress intensity range (ΔK)

April 3, 2019 Weld Fatigue Analysis 43

 Weld Fatigue Calculation Methods
Crack Tip Stress Intensity

Fatigue Crack Loading Modes

April 3, 2019 Weld Fatigue Analysis 44

 Learning Questions & Answers (4)

List some weld fatigue calculation methods:

 nominal stress method

 structural hot spot stress method
 effective notch stress method
 fracture mechanics (crack tip stress intensity)

April 3, 2019 Weld Fatigue Analysis 45

 ANSYS FEA Weld Models

April 3, 2019 Weld Fatigue Analysis 46

 T-Joint Fillet Weld FEA Model
Nominal Stress & Structural Hot Spot Stress Methods

Plate thickness = 10 mm
infinitely rigid Weld leg length = 10 mm
boundary condition

200 kN applied load

April 3, 2019 Weld Fatigue Analysis 47

 T-Joint Fillet Weld FEA Model
Nominal Stress Method Results

The nominal stress (range for R = 0) is calculated from the sum of

the membrane and bending components stresses at the surface of
the base metal away from the weld toe:

 200,000 

    100 
 0.01  0.2 

Fatigue life from BS7608 Figure 10

for class F2 weld ≈ 450,000 cycles

April 3, 2019 Weld Fatigue Analysis 48

 T-Joint Fillet Weld FEA Model
Structural Hot Spot Stress Method Results

The structural hot spot stress (range for R = 0) is calculated from the
stress results at the 0.4t and 1.0t surface locations adjacent to the
weld toe:

.  100.2 

.  100.0 

  1.67 100.2   0.67 100.0   100.3 MPa

Fatigue life from BS7608

Figure 10 for class D weld
≈ 1.4E6 cycles

April 3, 2019 Weld Fatigue Analysis 49

 T-Joint Fillet Weld FEA Model
Effective Notch Stress Method

loads and boundary conditions

identical to nominal stress model,
but 1 mm radii added at weld
toes with fine mesh

April 3, 2019 Weld Fatigue Analysis 50

 T-Joint Fillet Weld FEA Model
Effective Notch Stress Method Results

The effective notch stress (range for R = 0), taken directly from the
maximum principal stress at the weld toe radius, is approximately
170 MPa. This represents a notch factor of 1.7 compared to the
nominal stress method.

Fatigue life from IIW for

FAT199 [3] class ≈ 3.0E6
cycles (see next slide)

April 3, 2019 Weld Fatigue Analysis 51

 T-Joint Fillet Weld FEA Model
Effective Notch Stress Method Results

April 3, 2019 Weld Fatigue Analysis 52

 T-Joint Fillet Weld FEA Model
Results Comparison

Weld Stress Method Element Fatigue

Type Life
Nominal Stress (Baseline) Solid 4.5E5
Structural Hot Spot Stress Solid 1.4E6
Effective Notch Stress Solid 3.0E6
nCode “Volvo” (Not Shown) Shell 5.0E4

Significant variation in fatigue life depending upon the methodology used!

April 3, 2019 Weld Fatigue Analysis 53

 References

1) BS7608:2014+A1:2015. Guide to fatigue design and assessment of

steel products. British Standards Institute (bsi) 2014
2) Doerk O, Fricke W, Weissenborn C. Comparison of different
calculation methods for structural stresses at welded joints.
International Journal of Fatigue 2003;25: 359-369
3) Pedersen M, Mouritsen O, Hansen M, Andersen J. Experience with
the notch stress approach for fatigue assessment of welded joints.
Proceedings of the Swedish Conference on Light Weight Optimized
Welded Structures 2010
4) AWS D1.1/D1.1M:2006. Structural Welding Code – Steel. American
Welding Society

April 3, 2019 Weld Fatigue Analysis 54

 THANK YOU!

April 3, 2019 Weld Fatigue Analysis 55