8 finite element modelling

Upload: kakay123

Post on 07-Apr-2018

229 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 8 Finite Element Modelling

    1/36

    1

    FE-MODELING

    Bertil Jonsson

    Expert weld strength

    Nov 2011

    2

    Content General

    Program overview

    Boundary conditions

    Accuracy

    Non linearity

    Sub modeling

    3D-models

    Some notes about software

    Summary

    3

    General

  • 8/3/2019 8 Finite Element Modelling

    2/36

    4

    Method used determines often the tool

    5

    6

  • 8/3/2019 8 Finite Element Modelling

    3/36

    7

    8

    9

    Program overview

  • 8/3/2019 8 Finite Element Modelling

    4/36

    10

    Some software tools

    11

    Boundary conditions

    12

    A HAULER AT WORK

  • 8/3/2019 8 Finite Element Modelling

    5/36

    13

    Boggi beam

    14

    GEOMETRY REAR FRAME

    15

    GEOMETRY BOGGI BEAM

  • 8/3/2019 8 Finite Element Modelling

    6/36

    16

    Example bearing in boggi

    17

    Example bearing in boggi

    18

    Project LOST example of HSS Light Optimized STructures

    Duration 2006-2009

    Budget appr 18 mSek

    End-conference 24-25 march 2010 in Sweden

    13 Work-packages, 3 of them are

    New weld class system

    Analysis course on local based methods

    Demonstrator boggi beam

    Education &

    seminars planned

    Continuation

    project: WIQ

  • 8/3/2019 8 Finite Element Modelling

    7/36

    19

    LOST Workpackage no 10GOAL : decrease weight by at least 20%

    Demonstrator boggi beam

    20

    Old BB compare to the new

    Old New

    t = 15 / 8 mm t = 12 / 6 mm

    183 kg 143 kg (-22%)

    Mtrl grade: 350 Mpa Usage of high strength steel (HSS)

    21

    Max static load material grade

    Yield 350 Mpa at bearing

    Yield 460 Mpa in flanges

    Yield 600 Mpa in webs

    Control of buckling

  • 8/3/2019 8 Finite Element Modelling

    8/36

    22

    Fatigue analysis using local based methods

    Equivalent load from test track

    Main principal stress range + FAT 225 Mpa

    Fatigue life > 1000 hours on test track

    allowed stress level 370 Mpa

    Radius = 1 mm

    Notch method

    Loads on global model Transferring displacements

    local sub models and study

    the stresses

    23

    Test set up

    Rig setup Spectrum load (range pair)

    Range Pai r

    0

    100

    200

    300

    400

    500

    600

    10 10 0 10 00 10 00 0 100 00 0

    Antal ranger

    24

    Test of boggi no 1, failure after 120 h

    Expected > 1000 h

    Failure after 120 hours Lack of fusion

    (demand penetration i = 2 mm)

    However :

    Chocking !!! this could not be the only reason

  • 8/3/2019 8 Finite Element Modelling

    9/36

    25

    Multi axial stresses !

    Normally This case is in shear loading

    max Principal Stress range use von Mises (diff signs !) & FAT 200

    PS = 516 Mpa vM = 745 Mpa

    N = 420 hours N = 100 h [agrees well with test 120 !]

    26

    Test of boggi no 3 failure after 920 h

    Fully penetrated on root side

    Now failure from toside !

    Same cause: Multiaxial stress state with different signs

    in principal stresses

    PS = 241 Mpa vM = 393 Mpa

    N = 4100 hours N = 660 h (appr in agree with test 920 h)

    27

    Multi axial proportional cases

    - effective notch method -

    Use Mises stress range & FAT 200 Mpa

    in cases when Mises > Max principal stress range

    Comparison of life calculation in notch method

    ( multi axial cases with diff. signs in principal stresses, proportional loads )

    10

    100

    1000

    10000

    10 100 1000 10000

    Tested life (hours BII)

    Calculatedlife(hoursBII)

    Mises

    Princ stress

    Equality

    Non c

    onser

    vative

    Conse

    rvative

  • 8/3/2019 8 Finite Element Modelling

    10/36

    28

    Cost estimations

    Lowered weight & production cost

    win-win situation for customer & producer

    COST ESTIMATION (% of total) OF NEW BOGGI BEAM

    -10

    -8

    -6

    -4

    -2

    0

    2

    4

    6

    slit r emo ved mtrl+cut plate a way fixture

    improved

    extra weld weld

    prearation

    TIG

    treatment

    Totally

    Percentage

    (%)

    -8%

    29

    Demonstrator Boggi Beam results:

    Lowered weight 22%

    Using HSS & improved weld quality

    Life target almost reached

    Lower production cost 8 %

    Lessons learned:

    Multiaxial problem found and verified

    Use Mises when > principal stresses

    A high production quality is needed !

    Process variation & control

    30

    Accuracy & non linearities

  • 8/3/2019 8 Finite Element Modelling

    11/36

    31

    A-stay

    32

    Assembly of an A-stay

    with studied welds

    33

    FE-models

    SimplifiedSimplified linearlinear

    elasticelastic modelmodel

    ComplexComplex modelmodel

    includingincluding screwsscrews,,

    axleaxle casingcasing,,

    contactcontact elementselements

  • 8/3/2019 8 Finite Element Modelling

    12/36

  • 8/3/2019 8 Finite Element Modelling

    13/36

    37

    Global model

    38

    Sub model

    39

    Example of the NOTCH-method

    Rear frame of a loader

    (without a weld prepared plate)

  • 8/3/2019 8 Finite Element Modelling

    14/36

    40

    GEOMETRY OF REAR FRAME

    41

    Section through the sub model

    42

    Result symetric load

    Global model w/o notch

  • 8/3/2019 8 Finite Element Modelling

    15/36

    43

    Zoomed picture

    44

    45

    Submodell incl notch

  • 8/3/2019 8 Finite Element Modelling

    16/36

    46

    Submodell incl notch

    47

    Submodell incl notch

    48

    Submodell incl notch

  • 8/3/2019 8 Finite Element Modelling

    17/36

    49

    Sectio through the sub model

    50

    Zoom around the weld / notch

    51

    Result w/o deformation

  • 8/3/2019 8 Finite Element Modelling

    18/36

    52

    Result w deformation

    53

    Max principal stress in the notch

    54

    Zoom, max stress = 1932 MPa

  • 8/3/2019 8 Finite Element Modelling

    19/36

    55

    Hand calculation on white board !

    56

    Brottmekanisk ansats

    Global modell

    Submodell 1

    Submodell 2

    57

    Fracture mechanicsFracture mechanics

    > @)4()4(21

    21122

    CBCBI vvvvl

    GK

    S

    N

    mKCdN

    da)(* '

    )/( WafaK SV''

    '

    fa

    a

    mda

    KCN

    0

    1

    ParisParis lawlaw DDisplacementisplacementcorrelation techniquecorrelation technique

  • 8/3/2019 8 Finite Element Modelling

    20/36

    58

    Automatic crack propagation in

    3D

    59

    Tvrsnitt genom submodellen

    60

  • 8/3/2019 8 Finite Element Modelling

    21/36

    61

    XCRACK

    62

    Submodel 2Section through weld with crack

    63

    Submodel 2 swept to 3D

  • 8/3/2019 8 Finite Element Modelling

    22/36

    64

    Submod 2 merged into submod 1

    65

    Result in section

    66

    Zoom at the root crack

    Opening of the crack determinesthe stress intensity

  • 8/3/2019 8 Finite Element Modelling

    23/36

    67

    Stress intensity along the sub model

    Bakrams-spricka

    SPG-INT lngs axelgenomfringen

    0,0

    2,0

    4,0

    6,0

    8,0

    10,0

    12,0

    14,0

    0 200 400 600 800 1000

    s (mm)

    dK[(MPa_

    rot(m)

    ]

    Chan

    TH=2

    TH=5

    68

    3 sub models are analysed:

    Spricklngd Spg-intensitet Lutning

    a=8 mm K = 12 MPam (16-12)/(12-8) = 1

    a=12 K = 16 (40-16)/(16-12) = 6

    a=16 K = 40

    Approx. bi-linjr fkn : K = 1*a+4 resp 6*a-56

    69

    Stress-intensity as fcn of crack growth

    Antag linjr fkn : K = a+4 resp 6a-56

    Spnnings-intensitet sfa sprickvg

    0

    5

    10

    15

    20

    25

    30

    35

    40

    45

    5 7 9 11 13 15 17

    Sprickstorlek (mm)

    Spg.int(Mparot(m))

  • 8/3/2019 8 Finite Element Modelling

    24/36

    70

    Hand calculation on white board !

    71

    Comparison of calculated lifes

    Konservativ berkning kan vara fr skert

    Notchmetod stmmer bra med brottmekanik

    Livslngd med olika metoder

    0

    50000

    100000

    150000

    200000

    250000

    300000

    350000

    400000

    450000

    1 rt linje 2 rta linjer krkt linje notch-metoden

    Livslngd(cykler)

    N

    72

    More about accuracy

  • 8/3/2019 8 Finite Element Modelling

    25/36

    73

    Use the right stiffness in the

    FE-model...

    74

    Arm to excevator

    75

    B/BRG

    Initial root

    crack

    4mm~6mm

    Field Crack Section View

    CRACK PATTEN

    Section view

  • 8/3/2019 8 Finite Element Modelling

    26/36

    76

    LEFM Analysis of arm

    [Case A]

    16t

    2mm

    2mm

    2.8mm

    5.6mm

    4mm

    Sub1

    Sub2

    Sub3

    Sub4

    Model Cases

    Digging Force Side Force

    Load Cases

    77

    Definition of Crack Direction in Sub1 Model

    - Principal Max Stress Direction Check

    Crack Growth Direction

    Principal Max Drection

    2mm

    78

    L

    Sub2 Model Analysis Result

    - Sub2 Model

    2.8mm

    Principal Max Directio

    Crack Growth

    Direction

    Keff = 28.75

    Digging Force

    Keff = 1.57

    Side Force

  • 8/3/2019 8 Finite Element Modelling

    27/36

    79

    LEFM Analysis of arm

    2mm

    2.8mm

    5.6mm

    4mm

    Sub1

    Sub2

    Sub3

    Sub4

    Model Cases

    Digging Force Side Force

    Load Cases

    [Case B]

    18t

    2mm

    80

    Arm FE Model ( Case B )

    [ Analysis Result ] [ Coupled Model ]

    [ Sub Model ]

    81

    Conclusion

    Bench Test Loading

    Failure Life=99677(Cycle)

    Failure Life=79095(Cycle)

    Initial Crack length : 2mm

    Failure Crack length : 25mm

    (Median 50%)

  • 8/3/2019 8 Finite Element Modelling

    28/36

    82

    Cold lap

    Spatter induced cold lap

    Compared with

    line cold lap

    83

    What is the difference between a line

    cold lap and a spatter induced cold lap

    ?

    84

    Geometry spatter induced cold lap

  • 8/3/2019 8 Finite Element Modelling

    29/36

    85

    Cold lap 2D vs 3D

    Question: what is the expected increase in life for a spatter induced

    cold lap compared to a line cold lap ?

    Spatter induced: has a/c=1 at start, this requires a 3D-model &

    analysis.

    Line cold lap: has a/c=0 at start, can be calculated in a 2D-

    model.

    86

    Studied case

    Korsprovstav:

    t = 12 mm

    = 100 MPa

    a = 5 mm = 70 grader

    R = 0

    i = 2 mm

    87

    2D analysis in FRANC2D

    of

    a line cold lap

  • 8/3/2019 8 Finite Element Modelling

    30/36

    88

    2D-modellen (a/c=0)

    89

    3D analysis in FRANC3D

    of

    a spatter induced cold lap

    90

    Initial flaw: cold lap of crack size a=0.2 [mm] and a/c=1

    (fringe values are crack opening displacement in [mm])

    (deformation magnified 5)

  • 8/3/2019 8 Finite Element Modelling

    31/36

    91

    Crack trajectory at step 3(fringe values are crack opening displacement in [mm])

    (deformation magnified 5)

    92

    Crack trajectory at step 8(fringe values are crack opening displacement in [mm])

    (deformation magnified 50)

    93

    Crack trajectory at step 13(fringe values are crack opening displacement in [mm])(deformation magnified 5)

  • 8/3/2019 8 Finite Element Modelling

    32/36

    94

    Resultat 3D

    95

    Resultat 3D

    a/c goes to 0 relatively slow Is appr 0,1-0,2 at full life (N=2,7E6 cycles)

    N is calculated along the edge with K=K(c) (not area) and

    with extrapolated.

    Mixed mode in the beginning, there after mode I KI+KII at bottom gives a great kink-angle KI+KIII at edge gives a small kink-angle

    Crack grows deepest at bottom (comp kink) Similar conditions as in the 2D-analysis

    96

    Result in lifeLife for cold lap = 0,2 mm - comparison 2D (a/c=0) with 3D (a/c=1)

    ( crusiform joint, t12, a5, 70 deg, R0, 100 MPa )

    0

    1

    2

    3

    4

    5

    6

    0,0E+00 5,0E+05 1,0E+06 1,5E+06 2,0E+06 2,5E+06 3,0E+06

    LIFE (cycles)

    Crac

    kgrowth(mm)

    3D (a/c=1)

    2D (a/c=0)

  • 8/3/2019 8 Finite Element Modelling

    33/36

    97

    Comparison line CL vs spatter CL

    Life N cold lap from spatter = 2,7E6

    Life N lilne cold lap = 1,0E6

    Thus is N appr 3 times bigger

    Edge crack has K(a/c=1) 0,63*K(a/c=0)

    Theoretical longer N is then (0,63) -3 4

    Fits well with 3 above

    98

    Some notes on softwares

    Ansys Classic fracture mechanics 2D

    Ansys WorkBench notch method 3D

    Afgrow fracture mechanics 2D

    (load plane)

    Franc2D fracture mechanics 2D

    (load//plane)

    99

    Ansys Classic

    fracture mechanics, 2D

  • 8/3/2019 8 Finite Element Modelling

    34/36

    100

    Ansys Classic

    fracture mechanics, 2D

    Result:

    KI, KII, KIII for a point

    Repeat of growth points

    Integration gives life

    101

    Ansys WorkBench (+DesignModeler)

    notchmetoden 3D

    Result:

    Stresses in the notch

    SN-curve gives life

    102

    A new layout is seen in WB.12

  • 8/3/2019 8 Finite Element Modelling

    35/36

    103

    Afgrow

    fracture mechanics in 2D

    Result

    Cycles = life

    104

    Afgrow, result

    Result

    Cycles = life

    105

    Franc2D

    Automatic crack growth in 2D

    Result

    KI, KII, KIII as fcn

    of crack path.

    (Integration of

    Paris law

    gives life)

  • 8/3/2019 8 Finite Element Modelling

    36/36

    106

    Summary

    107

    Some short conclusions

    Daily work: Use notch-method

    Failure analysis: Use linear fracture mechanics

    Free of charge ! 2D: Agfrow (?) and Franc2D

    Be careful with: Contacts, stiffnes

    Do not forget: Root side of welds

    Useful technique: Sub-modeling