Anda di halaman 1dari 70

Recent developments in fatigue of

welds

Professor Greg Glinka - University of Waterloo


Dr. Mohamad El-zein – John Deere
Jim Wong - John Deere
SAE FD&E - 15 Oct 2008

Run Smart, Run Fast, Run Lean 1


Speaker Contact Info:

• Jim Wong
– WongJamesD@JohnDeere.com
– 309-765-3891

John Deere - MTIC


One John Deere Place
Moline IL 61265

Run Smart, Run Fast, Run Lean 2


Where are they now?

Run Smart, Run Fast, Run Lean 3


Deere Weld Modeling History
• 2000 - Investigation into using Shell FE Models
for use obtain local peak stresses for e-N fatigue
calculations
– Evaluated both Nominal and Hot Spot Stress Methods
– Kt’s were based on traditional definition of Nominal
stress

Run Smart, Run Fast, Run Lean 4


Deere Weld Modeling History
• 2003 - Investigation to determine Through-the-
Thickness stress distributions using Shell FE
– Established method to calculate actual stress field
– Kt library for various weld types created

Run Smart, Run Fast, Run Lean 5


Deere Weld Modeling History
• 2005 – Establishment of GY2 FE Shell Model
Technique
– Correlation to 3D-FE fine mesh solutions
– Residual Stress Effects Included

Run Smart, Run Fast, Run Lean 6


Weld Life Check List

Global Loads
Material Properties
Local Loads, Stresses or Strains
Stress concentration Factors
Peak stress at weld toe/critical points
Through thickness stress distribution
Residual Stresses

Calculate Weld Life

Run Smart, Run Fast, Run Lean 7


Load
F

peak
n

© 2008 Grzegorz Glinka. All rights reserved. 8


peak
nom

nom

© 2008 Grzegorz Glinka. All rights reserved. 9


tp

peak
D
r E
hs

h


p

n
M
h P
B

t
A
C

A) remote (nominal) through thickness stress, B) the actual through-


thickness stress distribution in the weld toe cross section, C)
linearized through-thickness stress distribution in the weld toe cross
section, D) the actual stress distribution in the plate surface, E)
extrapolated (linearly) stress distribution in the plate surface
© 2008 Grzegorz Glinka. All rights reserved. 10
V
n n

H
t
C

tp

r
V
 n peak n

H
t
C

© 2008 Grzegorz Glinka. All rights reserved. 11


Shown are: the experimental definition and determination
of the hot spot stress, the actual through thickness stress
distribution and the hot spot stress resulting from the
y
linearization of the actual stress fields

peak The stress xx in the plate


surface is believed to be
linear in this region !
hs
xx(x)

x
t

0.4t Strain
Courtesy: E. Niemi
1.0t gauges

© 2008 Grzegorz Glinka. All rights reserved. 12


r n peak b
1
)
T

n2

The stress concentration


factors, , and
are not constant and not
the same!
They depend on the
geometry and on the
stress ratio: mn/ bn!

a) A body with an angular notch subjected to multiple loading modes


and resulting through-the-thickness stress distribution, b)
decomposition of the nominal (linear) stress distribution in the notch
cross section into the membrane and bending contribution
© 2008 Grzegorz Glinka. All rights reserved. 13
1)
y
a1= n1 1peak
x
t

b1

2)
y
b2 =n2
2pe
ak
x
t

b
2

© 2008 Grzegorz Glinka. All rights reserved. 14


y
a) Pure axial
hs m
m peak load
F
x
t

Stress concentration factors Kmt,hs and Kbt,hs


DO NOT DEPEND on the stress ratio mhs/ bhs
and they are constant for given geometry!!

b) y
b) Pure bending
hs b
bpeak load
M
x
t

© 2008 Grzegorz Glinka. All rights reserved. 15


a)
y
a pe b
ak )
x
T

b

c) d
)

The stress concentration factors and


depend only on the geometry and they
can be used for any stress ratio !!

a) T-butt weldment and resulting through-the-thickness stress distribution,


b) decomposition of the nominal (linear) stress distribution in the weld toe
plate cross section, c) the hot spot stress as a sum of the hot spot membrane
and bending stress, d) the actual peak stress as a sum of the stress
concentration on the hot spot membrane and bending stress
© 2008 Grzegorz Glinka. All rights reserved. 16
a) b)
y
y
P xL x L
P P P
(x,
t y) t
peak
(x,y)

h
s

- depends on L and is constant along the weld toe


line

Independent of L but it
changes along the weld toe
line
a) Stress distribution in the critical cross section near the cover plate ending
and the nominal or the hot spot stress n (independent of length L ) and hs
(independent of length L),
b) Stress distribution in the critical plane near the ending of a vertical
attachment (gusset) and the nominal or the hot spot stress n (dependent on
length L ) or hs (independent of length L)
© 2008 Grzegorz Glinka. All rights reserved. 17
The advantage of using expression

lies in the fact that the membrane stress hsm and the bending
stress hsb can be determined by simple decomposition of the
linearized through-thickness stress field, (x=0,y), which can be
directly obtained from the coarse mesh 3-D or shell Finite
Element (GY2) analysis. Thus the equation above provides the
link between the FE stress analysis data, hsm and hsb, and the
peak stress, peak, at the weld toe, necessary for the fatigue
analysis

© 2008 Grzegorz Glinka. All rights reserved. 18


GY2 Approach for Fatigue Life
Prediction Using Shell Finite
Element Results

Run Smart, Run Fast, Run Lean 19


• Multiaxial state of
stress at weld toe
• One shear and two zz
xx
normal stresses
• Due to stress
concentration, xx
is the largest
xz
component zx
– Predominantly responsible xx
for fatigue damage zz

© 2008 Grzegorz Glinka. All rights reserved. 20


Middle plane of the attachment

a) c)
tp
hp physical common plane
middle plane for the attachment and
of the main the main plate
plate
h
The image cannot be displayed. Your computer may not have enough memory to open the image, or the

t image may have been corrupted. Restart your computer, and then open the file again. If the red x still
appears, you may have to delete the image and then insert it again.

y
x
y d) 0
b) h
z
x
0
z
h/2
(h+t/2)

h/2
h/2
(h/2+tp/2) h/2 h/2 h/2
(h/2+t/2)

(h+tp/2)

© 2008 Grzegorz Glinka. All rights reserved. 21


The GY-2 Model

Middle plane of the attachment


T

tp h
e
i
m
a
g
e
c
h physical common
middle plane of a
n
n plane for the
the main plate o
t
b
e attachment and the
d

h i
s
main plate
The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been

t corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and
then insert it again.

Element Nodes

h
Reference points where
stress is to be determined h/2
(h+t/2)

h/2
h/2
(h/2+t/2)

(h/2+tp/2) h/2 h/2 h/2

(h+tp/2)

Run Smart, Run Fast, Run Lean 22


doubler
middle plane physical common plane
h for the doubler and
the main plate
tp

The image cannot be displayed. Your computer may not have enough memory to open the
image, or the image may have been corrupted. Restart your computer, and then open the file
again. If the red x still appears, you may have to delete the image and then insert it again. t

main plate
middle plane

(tp/2 + t/2)



t/2
t/2
h/2 t/2 t/2

© 2008 Grzegorz Glinka. All rights reserved. 23


Welded joint A- B-
A B
y t
t
h
x

Shell FE model

© 2008 Grzegorz Glinka. All rights reserved. 24


Shell Element Model Details

• The FE formulation for shell elements gives top and bottom


stresses, top, and bottom
• The stress distribution through the thickness is considered to
be linear
• The membrane and bending stresses are obtained from

top
Shell element at midplane

bottom
Run Smart, Run Fast, Run Lean 25
6 mm

© 2008 Grzegorz Glinka. All rights reserved. 26


© 2008 Grzegorz Glinka. All rights reserved. 27
Weld Life Check List

Global loads
Material Properties
Local loads, stresses or strains
Stress concentration factors
Peak stress at weld toe/critical points
Through thickness stress distribution
Residual stresses

Calculate Weld Life

Run Smart, Run Fast, Run Lean 28


l = hp
r

g=h


t
P P

Range of application - reasonably designed weldments, (K.Iida and T. Uemura, ref. 11)

© 2008 Grzegorz Glinka. All rights reserved. 29


l = hp
r

g=h


t
M M

Range of application - reasonably well designed weldments, (K.Iida and T. Uemura, ref. 11)

© 2008 Grzegorz Glinka. All rights reserved. 30


t1= tp hp
Validated for : 0.02  r/t 0.16 and
30o    60o, source [11]

h

x

P P

t
y

© 2008 Grzegorz Glinka. All rights reserved. 31


Validated for : 0.02  r/t 0.16 t1= tp hp
and 30o    60o, source [11]

h

x
M
M

t
y

where:

© 2008 Grzegorz Glinka. All rights reserved. 32


 r

© 2008 Grzegorz Glinka. All rights reserved. 33


Probability Density Function
0.05
 r

0.04

MF-5 MF-1

0.03
f()

MF-4 MF-3

0.02

MF-2
MF-1: μ=4.0789, =0.1710
MF-2: μ=3.7854, =0.4668
MF-3: μ=3.9496, =0.2665
0.01
MF-4: μ=3.9669, =0.4135
MF-5: μ=4.0084, =0.2832

0.0
0.0 40.0 80.0 120.0 160.0 200.0
Angle,  [deg]

© 2008 Grzegorz Glinka. All rights reserved. 34


 r
f (r)

© 2008 Grzegorz Glinka. All rights reserved. 35


Probability Density Function – Log-normal
1.20

0.96

MF-3

0.72 MF-5
MF-4
f(Ktt)

MF-1

0.48

MF-2 MF-1: μ=0.9224, =0.1978


MF-2: μ=1.0180, =0.2834
MF-3: μ=0.7998, =0.1635
MF-4: μ=0.9602, =0.1835
MF-5: μ=0.8481, =0.1770
0.24

0.00
0.0 1.8 3.6 5.4 7.2 9.0
SCF, (Ktt)

© 2008 Grzegorz Glinka. All rights reserved. 36


Probability Density Function - Bending
0.80

MF-3

0.64
MF-5
MF-4

MF-1
0.48
f(Ktb)

MF-2
0.32

MF-1:
MF-2:
MF-3:
0.16 MF-4:
MF-5:

0.00
0.0 2.0 4.0 6.0 8.0
SCF, (Ktb)

© 2008 Grzegorz Glinka. All rights reserved. 37


Statistical distribution of weld toe radii measured on
similar weldments produced by three manufacturers
0.99

0.90
MQ2 – production line 0.80


MQ1 – R&D workshop 0.70

MQ5 – University 0.60


0.50
laboratory
0.40

Probability
0.30
0.25
0.20
0.15

0.10
0.08
0.06

0.04

0.02

0.01
0.1 0.2 0.4 0.7 1 2 4 7 1
0
Weld toe radius, r
[mm]
© 2008 Grzegorz Glinka. All rights reserved. 38
Weld Life Check List

Global loads
Material properties
Local loads, stresses or strains
Stress concentration factors
Peak stress at weld toe/critical points
Through thickness stress distribution
Residual stresses

Calculate Weld Life

Run Smart, Run Fast, Run Lean 39


Where:

Derived
for:
© 2008 Grzegorz Glinka. All rights reserved. 40
Where:

Derived for:

© 2008 Grzegorz Glinka. All rights reserved. 41


© 2008 Grzegorz Glinka. All rights reserved. 42
© 2008 Grzegorz Glinka. All rights reserved. 43
Weld Life Check List

Global Loads
Material Properties
Local Loads, Stresses or Strains
Stress concentration Factors
Peak stress at weld toe/critical points
Through thickness stress distribution
Residual Stresses

Calculate Weld Life

Run Smart, Run Fast, Run Lean 44


Residual stresses

Run Smart, Run Fast, Run Lean 45


How to account for RS

The residual stress can not be added linearly to the actual stresses at the
notch tip. However, the residual stress effect can be accounted for by adding
it to the pseudo-elastic stress in the Neuber formula.

In general residual stresses change the


notch tip mean stress rather than the amplitude.

Run Smart, Run Fast, Run Lean 46


Residual stress effect on mean stress

Run Smart, Run Fast, Run Lean 47


Weld Life Check List

Global Loads
Material Properties
Local Loads, Stresses or Strains
Stress concentration Factors
Peak stress at weld toe/critical points
Through thickness stress distribution
Residual Stresses

Calculate Weld Life

Run Smart, Run Fast, Run Lean 48


EXAMPLES

Run Smart, Run Fast, Run Lean 49


The T-Joint Weldment

Run Smart, Run Fast, Run Lean 50


Weld Geometry (all locations)

=tp hp

t
y

Run Smart, Run Fast, Run Lean 51


GY2 - Shell Element Model Details

19,197 nodes
18,858 elements
114,069 dof

Material:
A22H Steel (ASTM A500 Cold Formed
Steel for Structural Tubing)

Run Smart, Run Fast, Run Lean 52


Shell Element Model Summary

top (MPa) bottom m (MPa) b (MPa) hs (MPa)


(MPa)
Location 1 -7.14 2.19 -2.48 4.67 15.7
Location 2 -7.43 2.04 -2.69 4.74 16.3
Location 3 -5.20 -1.59 -3.39 1.81 10.9

Run Smart, Run Fast, Run Lean 53


Solid Element Model Details

885,069 nodes
613,891 elements
2,700,000 dof

0.008” element size at


weld toe
Run Smart, Run Fast, Run Lean 54
Comparison GY2 shell vs. 3D fine mesh FEA

peak via Kt peak from 3-D FEA


(GY-2 with 1.5t weld)
Location 1 17.1 MPa 16.9 MPa
Location 2 17.4 MPa 17.4 MPa
Location 3 11.4 MPa 10.6 MPa

Run Smart, Run Fast, Run Lean 55


T-Joint stress distribution

© 2008 Grzegorz Glinka. All rights reserved. 56


T-joint

© 2008 Grzegorz Glinka. All rights reserved. 57


T-joint

© 2008 Grzegorz Glinka. All rights reserved. 58


Fatigue Analysis Results (-N Method)

Run Smart, Run Fast, Run Lean 59


T-Joint experimental vs predicted life

60
T-Joint experimental vs predicted life

61
Gusset Weld

62
Gusset – Out of Plane Loading

63
Gusset –In Plane Loading

64
Square tube on plate

Overview of the experimental set-up

65
Square tube on plate

Fig. 41. Calculated total fatigue live based on the GY2 hot spot stress approach and the experimental fatigue data
using the JD material data; Welded tube-on-plate specimen (fully reversed lateral load of 21350 N)

66
Weld Life Check List

Global Loads
Material Properties
Local Loads, Stresses or Strains
Stress concentration Factors
Peak stress at weld toe/critical points
Through thickness stress distribution
Residual Stresses

Calculate Weld Life

Run Smart, Run Fast, Run Lean 67


Conclusions

• The GY-2 method provided an accurate


and moderately simple way to obtain the
peak stress for fatigue evaluations
• The GY-2 method with a single set of
membrane and bending stress
concentration factors provided a good
representation for the through-thickness
stress field.

Run Smart, Run Fast, Run Lean 68


Other Areas of Investigation
Manufacturing Process Simulation:

• Prediction of local mechanical properties and microstructures, residual


stresses and distortions to decrease manufacturing costs and improve
durability calculations.
• Welding Simulations
• Heat Treat Simulation
• Castings

Structural Analysis

• Consolidation of Fatigue calculations : Initiation and Propagation

Run Smart, Run Fast, Run Lean 69


Thank You

Run Smart, Run Fast, Run Lean 70

Anda mungkin juga menyukai