Scribd
Upload
441 views23 pages

ALTAIR - EHTC - OCT - 2010 - PJH - Final - Fatigue Analysis of Welded Structures Using Ncode DesignLife

- nCode DesignLife is a fatigue analysis software that efficiently analyzes large problems using a variety of fatigue analysis methods for welded structures. - It uses the "Volvo" method to calculate structural hot-spot stresses around welds based on cross-sectional forces and moments, allowing for a reasonable level of mesh insensitivity. - The software can minimize mesh sensitivity by using appropriate stress recovery methods like CUBIC interpolation, displacement-based stresses, or grid point forces, along with modeling guidelines.

Uploaded by

svk_nt
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
441 views23 pages

ALTAIR - EHTC - OCT - 2010 - PJH - Final - Fatigue Analysis of Welded Structures Using Ncode DesignLife

- nCode DesignLife is a fatigue analysis software that efficiently analyzes large problems using a variety of fatigue analysis methods for welded structures. - It uses the "Volvo" method to calculate structural hot-spot stresses around welds based on cross-sectional forces and moments, allowing for a reasonable level of mesh insensitivity. - The software can minimize mesh sensitivity by using appropriate stress recovery methods like CUBIC interpolation, displacement-based stresses, or grid point forces, along with modeling guidelines.

Uploaded by

svk_nt
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 23

Fatigue Analysis of Welded Structures Using nCode DesignLife

Dr Peter Heyes - HBM UK Ltd Altair EHTC, Versailles, Oct 27-29 2010

Agenda

What is DesignLife? Fatigue analysis methods for welded structures Volvo method Stress recovery methods for welds and mesh sensitivity Validation and application Concluding remarks

nCode DesignLife summary

General-purpose FE-based fatigue analysis system Efficiently analyse large problems Wide range of fatigue analysis methods Highly configurable Integration with test Automate analysis and reporting Intuitive flow-based GUI or process in batch Available through HyperWorks Partner Alliance

Weld fatigue calculation strategies LEFM (fracture mechanics) Local approach

Structural hot spot stress

Examples using structural hot-spot stress approach


Standards: BS7608 Eurocode 3 ASME Pressure Vessel Code Volvo method

T2

s = m + b = P/A + Mc/I
r

T1

Background to Volvo method


Rupp method for spot weld fatigue calculate structural hot-spot stress around spotweld based on cross-sectional forces and moments Ref: SAE 950711 Fermer et al: Analogous method for seam welds simple mesh to transfer loads and uses forces and moments to calculate membrane and bending stresses normal to weld toe. Reasonable level of mesh insensitivity. Distinction between stiff and flexible behaviour. Size effect included. Ref: SAE 982311

Weld configurations and modelling guidelines Fillet Overlap

Calculation process in nCode DesignLife

eff

t = Max tref

n ,1

Size effect

Mean stress effect

Typical analysis process

Mesh and FE code sensitivity

How do we get usable stress results from a coarse model

How to get useful weld toe stresses from relatively coarse models In linear shells, stresses extrapolated from integration points give rather inconsistent results:. Sensitivity to mesh quality and mesh refinement Especially close to geometric features and weld ends Somewhat better for higher order elements 3 other options in DesignLife 1. Use CUBIC stress recovery in NASTRAN (or RADIOSS) 2. Calculate stress tensor in a similar way to the CUBIC method based on nodal displacements and rotations 3. Use Grid Point Forces to calculate membrane and bending stress normal to weld toe

How does it work? Using nodal displacements and rotations.


Relative displacements between node pairs used to calculate strains Strains from G1-G2, G1-G3, & G1G4 used as strain-gauge rosette Bending stresses calculated from assumed cubic displacement function Structural stresses calculated from resulting strain tensor Full stress tensor available Damage normally calculated using abs max principal stress or critical plane approach Good level of mesh insensitivity Rather conservative at singular locations or where stress is not normal to weld toe

How does it work? Using grid point forces and moments


Membrane and bending stresses normal to weld are calculated for weld toe and weld root elements

GPForces are collected at weld toe nodes and shared in proportion to element edge length Line forces and moments are averaged to mid point of edge and translated to local co-ordinate system

top ,normal

f x' 6 my' = +6 2 t t
6

Mesh sensitivity typical test specimens Quite uniform stress field Very little sensitivity to mesh density or quality Using CUBIC, GPF or Displacement based stresses gives almost identical results Similar results from different FE codes
Skewed mesh Refined mesh double lap-shear lap-shear

start-stop

Durability of Advanced High Strength Steel Gas Metal Arc Welds Bonnen et al. SAE 2009-01-0257

Mesh density sensitivity - CUBIC stresses vs GPF derived stresses


Volvo T-tube specimen

SAE 982311 Fermer et al

Stress recovery at a singular location

Perch mount specimen Highly singular stresses Results are sensitive to mesh density and weld modelling strategy Stresses from mid edge of weld toe elements give reasonable correlation

Durability of Advanced High Strength Steel Gas Metal Arc Welds Bonnen et al. SAE 2009-01-0257

Correlation of test data from steel partnership project


Predicted vs Experimental life
100000000

10000000

1000000

Experimental life

100000

Double lap shear 3.4 mm Double lap shear 1.6 mm 10000 Single lap shear 3.4 mm Single lap shear 1.6 mm Start-stop 1000 Perchmount tensile Perchmount shear

100 100

1000

10000

100000 Predicted life

1000000

10000000

100000000

Data from Steel Partnership project: Durability of Advanced High Strength Steel Gas Metal Arc Welds Bonnen et al. SAE 2009-01-0257

* dubious boundary conditions

Effect of FE code and mesh quality


Original mesh

Stress at Weld Toe


4500 4000 3500 3000 Stress 2500 2000

NASTRAN CUBIC Original


1500 1000 500 0 0 5 10 15 20 Distance 25 30 35 40

NASTRAN Displacements Original NASTRAN GPFORCES Original

Effect of FE code and mesh quality


Original mesh

Stress at Weld Toe


4500 4000 3500 3000 Stress 2500 2000 1500 1000

ABAQUS Stresses Original Mesh ABAQUS Displacements Original Mesh ABAQUS NFORC Original Mesh

NASTRAN

500 0 0 5 10 15 20 Distance 25 30 35 40

Significantly different performance between CQUAD4 and S4 for example!


Bad mesh

Stress at Weld Toe


4500 4000 3500 3000 Stress 2500 2000 1500 1000

NASTRAN Displacements Bad Mesh NASTRAN GPF Bad Mesh ABAQUS Displacements Bad Mesh ABAQUS NFORC Bad Mesh

NASTRAN (original mesh)

500 0 0 5

10

15

20 Distance

25

30

35

40

Application to spot weld fatigue analysis

Rupp method well-established ACM (HEX/MPC) spotwelds adequate for identification of critical areas ACM method too sensitive to mesh to establish trends Refined approach local remesh of critical welds and reassessment using seam weld method

Refined analysis of spotwelds using seam weld method

Most damaged ACMs automatically replaced by spider Additional shell elements inserted to permit recognition as seam welds Stable predictions permit reliable evaluation of design alternatives

Test 501 repeats

Prediction 600 repeats

Concluding remarks

DesignLife is readily configurable to predict fatigue life of welds using a variety of methods, including well-known standards The Volvo method implemented in DesignLife provides an effective, highly configurable, cost-effective and easy to use solution for seamweld fatigue life prediction in welded sheet structures All FE results are mesh sensitive, but mesh sensitivity can be minimised by:
4 Sensible

application of modelling guidelines 4 Using appropriate stress recovery methods (CUBIC, displacements, GPF) 4 Averaging and translating stresses to mid point of element edges

Not all linear quad elements are the same! In what are rather nonconverged meshes, results may differ using different FE codes

You might also like